patient safety in the acute healthcare chain · patient safety in the acute healthcare chain: is it...
Post on 26-Feb-2019
218 Views
Preview:
TRANSCRIPT
PATIENT SAFETY IN THE ACUTE HEALTHCARE CHAIN
PATIENT SAFETY IN THE ACUTE HEALTHCARE CHAIN
I s i t s a f e r @ h o m e ?
Is it safer@hom
e?
L o u i s e S . v a n G a l e n
Louise S. van Galen
PATIENT SAFETY IN THE ACUTE HEALTHCARE CHAIN
PATIENT SAFETY IN THE ACUTE HEALTHCARE CHAIN
I s i t s a f e r @ h o m e ?
Is it safer@hom
e?
L o u i s e S . v a n G a l e n
Louise S. van Galen
Cover/layout: Off Page / P.P.P. Bos
Printed by: Off Page, Amsterdam
ISBN: 978-94-6182-799-9
Copyright © L.S. van Galen
VRIJE UNIVERSITEIT
Patient Safety in the Acute Healthcare Chain: Is it Safer@home?
ACADEMISCH PROEFSCHRIFT
ter verkrijging van de graad Doctor aan
de Vrije Universiteit Amsterdam,
op gezag van de rector magnificus
prof.dr. V. Subramaniam,
in het openbaar te verdedigen
ten overstaan van de promotiecommissie
van de Faculteit der Geneeskunde
op dinsdag 27 juni 2017 om 15.45 uur
in de aula van de universiteit,
De Boelelaan 1105
door
Louise Sandra van Galen
geboren te Hengelo
promotor: prof.dr. M.H.H. Kramer
copromotor: dr. P.W.B. Nanayakkara
Leescommissie:
dr. J. Car
prof.dr. M.E. Holland
prof.dr. H.A.H. Kaasjager
prof.dr. J. Klein
prof.dr. D.A. Legemate
prof.dr. C. Wagner
TABLE OF CONTENTS
Chapter 1 Introduction and general outline of the thesis 9
The starting point of the hospital care chain: novel ways of organizing care
Chapter 2 Acute medical units: The way to go? A literature review 27Eur J Intern Med 2017;39:24-31
Chapter 3 Measurement of generic patient reported outcome 47 measures (PROMs) in an acute admission unit: A feasibility study
Acute Med 2016;15(1):13-9
Prevention of serious adverse events on the clinical wards
Chapter 4 Delayed recognition of deterioration of patients in general wards is 61 mostly caused by human related monitoring failures: A root cause analysis of unplanned ICU admissions
PLoS One 2016;11(8):e0161393
Chapter 5 A protocolised once a day modified early warning score (MEWS) 79 measurement is an appropriate screening tool for major adverse events in a general hospital population
PLoS One 2016;11(8):e0160811
The use of quality indicators to assess patient safety
Chapter 6 Exploring the preventable causes of unplanned readmissions using 93 root cause analysis: Coordination of care is the weakest link
Eur J Intern Med 2016;30:18-24
Chapter 7 Hospital readmissions: A reliable quality indicator? 111Ned Tijdschr Geneeskd 2016;160:A9885
Chapter 8 Physician consensus on preventability and predictability of 117 readmissions based on standard case scenarios
Neth J Med 2016;74(10):434-42
Chapter 9 Patients’ and providers’ perceptions of the preventability of 131 hospital readmission: A prospective, observational study in 4 European countries
BMJ Qual Saf. 2017; In press
Chapter 10 Hospital Standardized Mortality Ratio: 153 A reliable indicator of quality of care?
Submitted
Chapter 11 Summary of main results, general discussion and future perspectives 165
Chapter 12 Dutch summary 181
Appendix List of publications 193List of scientific presentations 194Author affiliations 195Word of thanks 198Curriculum vitae / biography 200
CHAPTER 1
INTRODUCTION AND GENERAL OUTLINE OF THE THESIS
‘Primus non nocere (in the first place, do not harm)’ Hippocrates
11
INTRODUCTIONThe (acute) healthcare chains in Europe are increasingly under pressure due to the rise in demand
for care caused by the change in demographics (aging populations) and policy changes in
the healthcare system. This growing demand for healthcare is also reflected by the expenditure
in healthcare in the Netherlands. Currently, 90 billion euros are spend yearly on healthcare,
this is 40% more than in 2005.1 While emergency presentations have decreased, emergency
department (ED) admissions have risen greatly in the last decades, and the adjusted mean
length of hospital stay in the Netherlands has also increased by 8.5%.2,3 The ageing patient
often has more comorbidities and uses more medicine when compared to a younger patient.
In addition, due to overloaded general practitioners, budget cuts for nursing homes and
the introduction of obligatory deductible excess these patients tend to stay at home longer
without receiving the care they need. Because of this, they eventually present themselves
to the hospital with more complicated problems which are often not straightforward and
therefore require more diagnostics and therapy.4,5 This leads to congestion in the whole care
chain with resultant negative implications for the patients. As a result, the 1,1 million people
employed in the healthcare field in the Netherlands face a daily burden trying to make patient
flow as effective as possible.6 Healthcare workers in hospitals are constantly struggling to
find the balance between increasing number of ED admissions and a relative shortage of
beds. Furthermore, general practitioners (GPs) and primary care facilities such as home nurses
are also overloaded trying to organize appropriate care at home using the limited resources
available. This accumulation in medical teams’ workload in and outside the hospital results in
problems in communication and coordination within the acute healthcare chain. Inadequate
care coordination and poor teamwork take place both across sites of care as between providers7
and can result in poor handovers and inadequate teamwork.8,9 These issues are relevant to
focus on since they may lead to (serious) adverse events and complications, endangering
patient safety in hospitals.8,10 It is known, that in every ten patients, one is still harmed while
receiving hospital care.11
THE HEALTHCARE SYSTEM IN THE NETHERLANDS The Dutch healthcare system often receives positive appraisal. Its overall ranking on healthcare
quality, access and efficiency compared to other countries is high.12,13 The philosophy of
the system is based on some universal principles: ‘access to care for all, solidarity through
medical insurance (which is compulsory and available to all) and high-quality health services.’6
It is divided into three compartments: long-term care for chronic conditions, supplementary
care (i.e. dental care and physiotherapy) and basic and essential medical care (general
practitioner (GP), visits to hospital admissions and specialist appointments or procedures). This
last compartment consists of ‘planned’ care and ‘unplanned acute’ care.
Managed competition in healthcare market
The system has undergone some changes in the last decades. Since 2000, the government has
attempted to promote efficiency through ‘managed competition’, by introducing healthcare
insurers that consumers can choose from. These insurance companies are monitored by
12
the government but operate independently. The Netherlands has 9 insurance companies, of
which the four largest share 90% of the total.14 In the last decade, the increased amount for
obligatory deductible excess has led to a restriction to care.14,15 This ‘own risk’ is the amount
paid for covered health services before the insurance plan starts to pay. As a result, this
increase has unleashed a cascade effect on the insured, since the amount of supplementary
insurance has steadily decreased in the last 10 years.14 This financial burden for patients results
in less visits to a physician when needed, as well as not undergoing recommended diagnostics
or treatment. This causes a patient population that is more severely ill than needed if these
patients had timely consulted a doctor, since reasonably ‘simple’ problems could potentially
evolve in complicated syndromes.16
Patient flow through the healthcare chain (see infographic inside front cover)
Most of the times the patient flows start with the GP, the GP is seen as the gate-keeper of
the Dutch healthcare system. The Dutch GP-system consists of about 5000 GP offices and is
originally a well-designed system. Patients requiring hospital care mostly need a GP-referral
before they are seen by specialists. An exception is the patients who live in nursing homes
and are cared for by a nursing home physician or geriatric specialist on location. On average,
a patient contacts his GP 4.4 times a year, children from 5-17 years visit their GP 2.3 times,
patients above 85 visit 13.2 times.17 When a patient falls ill he usually visits the GP first, if
the GP assesses the patient as needing acute care he will refer the patient to the ED, mostly
after contacting emergency/acute physicians in the hospital for referral. If the problem is not
acute, the patient might be send home with a therapy initiated by the GP, or when the patient
needs non-acute specialized care he is referred to a specialist outpatient clinic.
If a patient becomes acutely ill without having time or the physical capability to visit the GP,
an ambulance may be called upon. This could for example be the case in an out-of-hospital
cardiac arrest on the streets or a severely septic patient who is in immediate need of life support
at home. Patients in the ED are mostly seen by emergency doctors or residents of the specialty
they were referred for. From the ED, patients can be sent home for problems which do not
require admission. Sometimes it is judged that better care can be provided in another hospital,
or in case of a social problem transfer to a nursing home might be most suitable. If a patient
does need in-hospital diagnostics and/or treatment he/she will be admitted. This can be either
to an acute medical unit (AMU) or directly on to a (specialist) clinical ward. On these wards
the patient is cared for by medical specialists, nurses, and other supportive healthcare providers
such as occupational therapists and geriatric nurses. If patients are severely ill, they might
also directly be admitted to an intensive care unit (ICU) or a unit likewise. During admission,
patients may revive rapidly and can be discharged soon, either from the AMU or from another
clinical ward. However, if patients deteriorate during their admission on a regular ward, an
unplanned ICU admission or even death might be inevitable.
At discharge from hospital patients are often given a follow-up appointment and their GP has
to receive a discharge letter from a hospital physician in order to be updated on the situation
13
of his patient and the course of events during hospital admission. Not all patients can be
discharged to their pre-existent (home) situation. Sometimes a transfer to another hospital,
a nursing home or a rehabilitation centre might be required temporarily, or in some cases
transfer to a nursing home permanently. Also, extra care such as professional ADL-support
might sometimes be needed when sending the patient back to their previous home situation.
Unfortunately, some discharged patients are readmitted unexpectedly to hospital, with
readmission rates varying from 10-30% internationally.18 Reasons for these readmissions can
for example be recurrence of or unresolved earlier complaints, or lack of social support in
the (new) home situation.
The text below describes an example of a patient moving through hospital and the difficulties
faced within the chain, which was used to illustrate the problem in a recent Dutch article.19
An 80-year old woman fell at home during the evening and was not able to get up by herself. The next morning a homecare nurse found her on the floor. In the past years, this patient hardly visited her general practitioner. She had suffered from some cognitive deterioration, but with the support of her family (two sons), current home situation was still acceptable. At the time of her fall, her sons were on vacation.
The home-care nurse called the GP immediately. He visited the patient and after seeing her he wanted to refer her to the hospital for further diagnostics and treatment. The ambulance services were busily engaged and after waiting for hours an ambulance picked her up at 5.15 pm. The GP had already left then and could not provide a face-to-face handover. After patient arrived at the overcrowded ED she was seen firstly by a medical student (an intern), followed by an emergency physician, and lastly by an internal medicine resident. The patient was diagnosed with delirium and had a low sodium, probably caused by a so called ‘tea and toast’-diet. Patient was unable to recall the exact medication she currently used. Initially, there were no hospital beds available to admit patient to hospital. Finally, at 11.30 pm a bed became available and patient was admitted. After a few days in hospital, patient recovered quickly and wished to go back home. She reported to the ward nurses that things at home were going well before the admission, and she did not need any additional help. This was a relief for the (transfer) nurses, since the originally intended nursing home did not have a place at that moment for her anyway. She was discharged to her initial home situation. It was however agreed that extra professional home care would be needed and provided twice a week. In addition, the GP’s practice nurse was advised to visit her occasionally.
A few days later, patient was readmitted to hospital, again in a confused state with low sodium. This time it was decided, in agreement with the GP among others, to move patient to a nursing home. Patients was discharged and she was advised not to continue with her diuretics because of the recurring hyponatremia. A week after discharge patient presented once again to the ED. Patient was found in bed, unresponsive. The letter from the ward doctor with advice on fluid restriction and discontinuation of diuretics was sent to the GP in time, but had not reached the nursing home doctor.
CHANGING DEMOGRAPHY The proportion of patients of 65 year and older is rising, in 2017 18% of the Dutch population is
over 65 years. In 2050, it is estimated this percentage will rise to 26%.8,20 The healthcare system
is forced to adapt to this development since preventable adverse events occur significantly
more often in older patients.21 In addition, older patients tend to have multiple comorbidities
and polypharmacy and often have atypical clinical presentations leading to more diagnostics
and as a result 75% of these patients spend too long in the emergency department (ED).5,22
14
The high workload caused by this increased patient flow and a decrease of human resources
due to budget cuts leads to the expense of time for patients. This is important especially in
this older population where communication plays an important role in the treatment. Patient’s
expectations and wishes should be spoken of thoroughly in order to prevent overtreatment
and unnecessary diagnostics.16 Lack of proper communication can have an effect on patient
safety sometimes even resulting in an unwanted ICU-admission or CPR in the elderly.23,24
Another problem faced by the Dutch healthcare system in the care for elderly is the recent
governmental budget cuts on chronic care. In 2012 the Netherlands spend three times more
on chronic care than Germany, and two times more than France, and the peak of the aging
population was yet to be reached.25 These cuts resulted in a decrease in numbers of nursing
homes available. To date, this gap in needed care has not been adequately taken over by
homecare nurses since the organization of home care by the local authorities to compensate
has fallen behind. Because of this, professional help is getting more scarce and the home-
caregivers are put under pressure to provide the necessary care and are forced to take
responsibility for the care of the patient. Ideally, general practitioners should provide care
for these patients. However, it is often difficult for a GP, who takes care of 2000 patient on
average, to manage this complex care at home for probably more than one patient. Since there
is no professional support for these patients, many patients remain at home until the situation
is unsustainable and are ultimately admitted to the ED, often quite severely ill. Older patients
are admitted more often than their younger equals and when admitted they often stay longer
in hospital than required. This results in stagnation of patient flow within the chain.19
PATIENT SAFETY As previously stated, the increasing burden on the acute healthcare chain can potentially lead to
serious adverse events within the chain.26 Iatrogenic harm was already described by Hippocrates
2400 years ago (‘iatros’=doctor, ‘genesis’=origin), but in the last decades the importance of
patient safety has been increasingly emphasized.8 Adverse events bring unintended, potentially
preventable care-related harm, and can compromise patient safety. Patient safety is defined as:
‘preventing errors and adverse effects to patients associated with health care’.11 These serious
advents such as in-hospital mortality, unplanned readmissions, and ICU admissions are also
often the result of a lack in adequate communication and coordination between professionals,
poor handover and inefficient teamwork.8,9,27,28 It is reported that patients can be handed
over 15 times in a 5-day hospitalization, and individual doctors participate in more than 3000
handovers per month.29 In addition, medical teams that are overloaded tend to perform poor
on patient safety outcomes.30
Since the first report ‘To err is human’ of the Institute of Medicine in 1999 which revealed
that 7-13% of all health-related harm contributes to the death of a patient, many countries
have started similar investigations.8,31 In 2004 the WHO started the Patient Safety Program,
and today 140 countries have worked to address challenges of unsafe care. In developed
countries it is estimated that 1 in ten patients (8-12%) is harmed while receiving hospital
15
care.11 This rate has stagnated and has not changed for more than 50 years.32 Statistics have
exposed that solely reducing the rate of adverse events in the European Union would prevent
more than 750.000 harm-inflicting errors per year leading to over 3.2 million fewer days of
hospitalization, 260.000 fewer incidents of permanent disability and 95000 fewer deaths per
year, and thus providing a relief of the current pressure on the healthcare system.11 It must
be remarked that most failures do not arise from lack of knowledge or inadequate skills of
the physicians, but rather from problems in the systems of care in place at time.33
Patient safety in the Netherlands
In the Netherlands, patient safety has also been an important subject of attention. EMGO-NIVEL
has investigated triggers for care-related harm since 2004. Due to multiple interventions and
improvement strategies such as implementing quality indicators34 (unplanned readmissions,
unintended long length of stay, HSMR), checklists35 (SBAR), and new monitoring instruments36
(early warning scores) the incidence of potentially preventable death was reduced from 5.5%
in 2008 to 2.6% in 2012.37
However, not all these new implementations are found beneficial within the healthcare
chain. An unintended result of these implementations is the resistance it causes among
healthcare workers on the frontline, who find most measures time-consuming, exhaustive, too
administrative, and unconnected to their actual work on the floor.32 They claim the government
is battling the complexity of hospital organizations with even more complex quality and safety
tools. As a result attention has shifted from care for the patient to care for the system.38
The Dutch Hospital Association has revealed that the number of health indicators is currently
up to more than 3400.8,38 In other European countries some of these indicators, such as
unplanned 30-day readmissions, are already used as a quality tool to rate and reimburse
hospitals, and financial penalties may also be attached.18,39 The intensive bureaucracy also has
resulted in nurses spending 60-70% of their time performing administrative actions, of which
they consider more than half as useless.38,40
Current situation in patient safety
In order to avoid harm to patients current caveats in improving patient safety must
be addressed:
→ Specialization and sub specialization within hospitals result in multiple
healthcare providers taking care of an individual patient, who only feel responsible
for certain aspects of the patient, rather than the patient as a whole. Due to this,
patients sometimes see from 50-200 healthcare workers during their treatment.8
Since these groups of professionals are more divided than ever, but still deal with
the same patient, more handovers are needed to get all required information to
other healthcare workers. More handovers weakens appropriate communication
and coordination.9 Handovers between shifts and disciplines should therefore be
a key element for appropriate improvement.28
16
→ Fragmentation in healthcare processes (the GP’s office, the ED, the AMU,
the ward, the ICU, transfer locations, nursing home etc.) results in inefficient and
discontinuity in patient flow since the transfer between these fragments is not
appropriately synchronized. Interdisciplinary interventions to streamline the patient
flow between all these partners, may reduce the incidence of adverse events.28,41
→ Hospitals are often trying to assess patient safety issues independently, and are
learning too little from each other’s work. Hospital partnerships could for example
play a critical role for technical-knowledge exchange between healthcare workers.
Also, co-development of solutions might result in rapidly evolving efficient global
health systems.42
→ The rise of electronic help devices and E-health to perform more care outside
the hospitals warrants adequate and smooth implementation, without impairing
patient safety.
→ Since most impulses improving patient safety have been external impulses, initiated
by supra-professional and regulatory bodies for example by the government or
the inspection of healthcare, they have caused an insufficient bureaucratic system,
which removes/takes away the attention from the real problems in the primary
process on the workfloor.8,28 These politicians, regulators, policy makers among
others, are too far from the clinical frontline, basing their efforts and solutions on
what they imagine everyday clinical work to be. However, these insights are all
based on second or third-hand accounts of how work on the frontline is actually
done, relying on data often analysed with delays.32 Efficient patient flow providing
a safer environment should be investigated from a practical and healthcare worker
point of view in order to find connection with the safety issues which are faced in
the field on a daily basis. Research on healthcare systems should focus on systems
‘that are real rather than ideal’.8,32
→ A culture change providing more patient participation, attitude changes in
professionals and more value-based leadership is needed to shift the attention
on patient safety from extrinsic to intrinsic motivation. Intrinsic motivation entails
activities undertaken because of internal motivation and provide immediate effect
after the action itself whereas extrinsic motivation makes people undertake actions
to get a reward or to dodge a penalty from an external party, without actually
being interested its immediate potential effect. High levels of ward’s shared values,
beliefs and behaviours plus an individual’s perception of culture are associated with
organization-wide reductions in adverse events.43
→ The focus in patient safety is mainly on the things that go wrong, instead of
the things that do go right. This fixation on errors is reactive and encourages a ‘find
and fix’ approach, without changing the mindset and culture on the workfloor.32
It is essential to learn from the far more frequent cases where things go right
and develop ways to support, augment and encourage these. We need to shift
our perspectives: from ‘as few things as possible go wrong’ to ‘as many things as
possible go right’.26
17
→ The current quality indicators used often do not measure real ‘quality’ or ‘safety’.
Of the 3400 quality indicators in the Netherlands, hardly any are developed
that actually measure outcomes of quality of care. The added value of all these
benchmarks and registration which should be performed to report them does not
measure up to the burden they cause in daily clinical life.38,44
→ Most importantly, even though we are aiming to improve patient care, we often
forget to include them in this development.
Patient involvement
In an age where patients are increasingly active as partners and demanders of care and in
which patient feedback is a compulsory part it seems surprising that patients’ (and their
carers’) views are not more routinely requested. In order to retrieve insights in how to do
this properly we must frequently converse with patients, especially those who have suffered
through unintended harm.8 The World Health Organization has also emphasized this by stating
that patient and community engagement and empowerment are the key. Their experiences
and perspectives are valuable resources for identifying needs, measuring progress and
evaluating outcomes.11
GENERAL OUTLINE OF THE THESISThe aim of this thesis is to provide insight and potential improvement strategies in safety issues
within the chain using novel ways to investigate these matters.
The fragments of the chain focused on in this thesis:
The starting point of the hospital care chain: novel ways of organizing care
Originally, patients enter the hospital through the ED, and if needed are admitted to the regular
clinical wards in the hospital. Over the past few decades shifts in this model have occurred.
Acute medical units (AMUs) have been introduced as a novel way to potentially improve
the efficiency of patient flow within the healthcare chain. AMUs are defined as hospital wards
specifically staffed and equipped to receive medical inpatients presenting with acute medical
illness from the ED, outpatient clinics and/or the community for expedited multidisciplinary
and medical specialist assessment, care and treatment for up to a designated period (typically
between 24 and 72 h) prior to discharge or transfer to other wards.45 In the Netherlands,
AMUs do not limit their case-mix to medical patients only, other specialties such as surgery and
neurology also admit patients to these units. AMUs form a link between the ED and the regular
wards in the hospital. Implementation of an AMU could lead to better and a more efficient
use of available beds. In addition, an AMU potentially reduces complications and unintended
errors by stimulating multidisciplinary teamwork, simplifying logistics and acceleration of
diagnostics and therapy.46-48
18
Over the past years their implementation has rapidly evolved, especially in the UK and Australia.
Since 2000, AMUs (or similar units) have also started up in the Netherlands. However, since
no national Dutch guidelines or implementation standards have been created, it is still hard to
conclude whether patient safety benefits from this (re)organization. In addition, no national
registration of centres with such a unit are easily available, making it hard for hospitals to find
and learn from each other and exchange experiences and pitfalls faced during implementation.
In chapter 2 of this thesis a systematic literature overview is provided describing the effects
of an acute medical unit on patient outcomes. In addition, it gives an overview of the current
situation in the Netherlands and makes recommendations that could be used to formulate
national AMU guidelines.
Even though research on the effectiveness of these units has been performed extensively,
the perspectives of patients on acute medical units have been underreported. The wards are
often busy and operate 24 hours a day, generating a constant buzz and a potential turmoil
for patients. It might be worthwhile talking to the patients in order to gather information
about their experienced quality of care and the feeling of safety in such a ward. Patients are
predominantly the initiator of their own healthcare pathway and their motives and experiences
should therefore be investigated in order to construct more streamlined care. In addition, since
the AMU caters to a heterogeneous patient population with a variety of pathology, it might
be interesting to test whether the planned patient-experience orientated research is feasible
in such a diverse group.
In chapter 3, we present the results of a feasibility study using Patient Reported Outcome
Measures (PROMs) to measure the perceived quality of care and feeling of safety in an AMU.
Prevention of serious adverse events on the clinical wards
Serious adverse events and adverse outcomes on clinical wards such as cardiac arrest, death
and unplanned ICU admissions endanger patient safety if they are the result of healthcare
processes. They are often preceded by signs of clinical deterioration, such as changes in vital
parameters as pulse rate, respiratory rate and level of consciousness.49,50
Unplanned ICU admissions from the ward
An unplanned ICU admission of an inpatient to the Intensive Care Unit from a general ward
is considered as a SAE, and is associated with poor long term survival, especially in older
patients.51 Therefore, it is important to investigate the root causes of these admissions, in order
to improve detection of these deteriorating patients and to potentially avoid unplanned ICU
admissions to enhance patient safety. A useful tool to analyse root causes of adverse events is
the PRISMA-tool (Prevention and Recovery Information System for Monitoring and Analysis).52
The main goal of this method is to build a database of incidents and process deviations by
creating causal trees, from which conclusions can be drawn to suggest countermeasures. It is
19
accepted by the World Alliance of Safety of the World Health Organization and has shown to
provide effective starting point for improvement in quality and safety of care.52-54
In chapter 4 we used a record review study to analyse the healthcare worker-, organizational-,
technical-, disease- and patient-related causes that contribute to late detection and treatment
of deteriorating patients in the general words using PRISMA-analysis.
Timely recognition of deterioration
To increase patient safety on nursing wards by timely detecting and treating the deteriorating
patients to prevent adverse events, rapid response systems (RRs) have been introduced. These
RRSs consist of an afferent limb residing a track and trigger system (TTS) such as the Modified
Early Warning Score (MEWS), and an efferent limb residing a rapid intervention team (RIT)
consisting of trained ICU personnel who will deliver treatment at the bedside when required.55
The effectiveness of these systems depends on timely recognition of these patients, and
requires proper implementation of both limbs. The TTS such as the MEWS, mostly used by
nurses to systematically monitor the patients in the wards, should be adequately implemented
in order to prevent potential adverse events.
In addition to the PRISMA-analysis, chapter 4 also assessed the adherence and effectiveness
of an already implemented track and trigger system (TTs) to investigate if patient deteriorations
were identified timely.
In chapter 5 we determined protocol adherence of a recently reimplemented rapid response
system in a prospective real-life setting, with a main focus on the effectuation of the afferent
limb, assessed by the use of MEWS on the wards. We also investigated the ability of this once
a day protocolised MEWS measurement to predict patient outcomes associated with poor
patient safety: in-hospital mortality, hospital length of stay, cardiac arrests, ICU-admissions and
unplanned 30-day readmissions.
The use of quality indicators to assess patient safety
Readmissions
Since 2015 the percentage of unplanned readmissions within 30 days after discharge of
a clinical admission in the same hospital was added as an official quality indicator to already
existing indicators, unintended long length of stay and in-hospital mortality.34
Currently, these indicators are regarded as a ‘major adverse events’ and are supposed to
reflect the quality of healthcare in Dutch hospitals. Policy makers in the USA and the UK are
already using the percentage of readmissions as a marker of quality and hospital safety; in
these countries a ‘high’ readmission rate can result in financial penalties.18,39 Empirical Dutch
research into readmissions is lacking, and policy makers are still struggling with the realization
of a reliable indicator – to date it does not distinguish between planned and unplanned
20
readmissions, and does not integrate ‘preventability’. The problem is that preventability is not
easily incorporated in this quality indicator, but should be the focus of improvement.
Readmissions are not easy to investigate, since they are often multi-causal, and their causes
do not solely lie in inadequate hospital care. It is however important to address this issue since
the increasing burden on healthcare might pressure doctors to discharge patients too soon,
potentially causing more unplanned readmissions since the actual problems may not have
been resolved during initial admission.30 In addition, it is known that in an older population,
which will make up an increasing amount of our patient-mix in the next years, mortality
increases if these patients are readmitted frequently.56 When studying a readmission the whole
process and decision-making in the chain should be looked upon, instead of concentrating
solely on the patient and his comorbidities.
Chapter 6 aimed to identify organizational-, technical-, healthcare worker- and patient-
related causes that contributed to readmission using previously described PRISMA-analysis. In
addition, we assessed how many of the investigated readmissions were deemed preventable.
The commentary presented in chapter 7 argues why readmissions cannot be regarded as
a reliable way of assessing quality of healthcare within a hospital in its current form.
Uniform factors related to preventable readmissions have not yet been identified.57,58 In
addition, consensus definition of the preventability of a readmission has not been established.
The current gold standard predominantly used is the opinions of physicians determining if
a readmission is preventable. Subsequently, factors that could predict preventable readmissions
are extracted from these findings.58-60
In chapter 8 we performed an international study to assess if there is any consensus between
physicians regarding the predictability and preventability of medical readmissions
As previously stated, in comparison to extensive US data, no large-scale European studies
have been performed about the predictability and preventability of readmissions. Furthermore,
patient and healthcare worker opinions, the stakeholders that are really confronted with this
problem on the frontline, are not frequently addressed to elucidate their vision.
In chapter 9, we describe the first prospective observational study of 1398 unplanned medical
readmissions to 15 centres in 4 European countries aiming to investigate the opinions of
readmitted patients, their carers, nurses and physicians on the predictability and preventability
of the readmissions. We also investigated the contributing factors that could potentially
predict (preventable) readmissions using their opinions.
21
Hospital Standardized Mortality Ratio (HSMR)
In order to be transparent about the quality of care, Dutch hospitals are compulsory to
publish their HSMR since 2014.61 It is one of the numerous healthcare indicators used in
the Netherlands, but together with unplanned readmissions and unexpected long length
of stay, these three elementary quality indicators ought to rate and represent the quality
of care within Dutch hospitals. The HSMR represents the ratio between the observed and
expected deaths. The expected deaths are calculated with the use of a statistical model
that corrects for certain factors such as age, socio-economic status and comorbidity.62 It is
questionable if the indicator in its current form really portrays the quality of a hospital, since
it is hard to extract a complete dataset which also incorporates the clinical illness in these
deceased patients.
Therefore, in a cohort of deceased patients with the diagnosis pneumonia, in chapter 10,
we examined whether the HSMR model under- or overestimates clinical disease severity. In
addition, we rated the completeness of the data collected by administrators to calculate HSMR.
Chapter 11 of this thesis provides a summary of the main results, a general discussion and
elucidates on future perspectives. A Dutch summary of this thesis can be found in chapter 12.
22
REFERENCES1. Centraal bureau voor de statistiek. Statline zorguitgaven; kerncijfers. http://statline.cbs.nl/Statweb.
Published December 2016. Accessed January 2, 2017.
2. Gaakeer MI, van den Brand CL, Gips E, et al. [National developments in Emergency Departments in the Netherlands: numbers and origins of patients in the period from 2012 to 2015]. Ned Tijdschr Geneeskd. 2016;160(0):D970.
3. Consultancy C. Ligduurmonitor Nederlandse ziekenhuizen 2013 & 2014. Trendbreuk in de ligduur. http://www.coppa.nl/wp-content/uploads/2016/01/Ligduurmonitor-2013-2014.pdf. Published December 2015. Accessed August, 2016.
4. Schrijver EJ, Toppinga Q, de Vries OJ, Kramer MH, Nanayakkara PW. An observational cohort study on geriatric patient profile in an emergency department in the Netherlands. Neth J Med. 2013;71(6):324-330.
5. Samaras N, Chevalley T, Samaras D, Gold G. Older patients in the emergency department: a review. Ann Emerg Med. 2010;56(3):261-269.
6. Ministry of Health Welfare and Sport. Healthcare in the Netherlands. http://investinholland.com/nfia_media/2015/05/healthcare-in-the-netherlands.pdf. Published January 2015. Accessed January 2, 2017.
7. Commonwelth Fund Commission. Why not the best? Results from the National Scorecard on U.S. Health System Performance. http://www.commonwealthfund.org/publications/fund-reports/2011/oct/why-not-the-best-2011. Published 2011. Accessed January 2, 2017.
8. Dillman R, Wagner C, Schellekens W, Klein J, Jager M, van der Grinten T. Patientveiligheid. Utrecht: de Tijdstroom uitgeverij. 2016.
9. Greenstein EA, Arora VM, Staisiunas PG, Banerjee SS, Farnan JM. Characterising physician listening behaviour during hospitalist handoffs using the HEAR checklist. BMJ Qual Saf. 2013;22(3):203-209.
10. Ong M, Bostrom A, Vidyarthi A, McCulloch C, Auerbach A. House staff team workload and organization effects on patient outcomes in an academic general internal medicine inpatient service. Arch Intern Med. 2007;167(1):47-52.
11. World Health Organization. Patient Safety. http://www.euro.who.int/en/health-topics/Health-systems/patient-safety/patient-safety. Accessed January 3, 2017.
12. David K, Stremikis K, Squires D, Schoen C. Mirror, Mirror on the Wall. How the Performance of the U.S. Health Care System Compares Internationally. http://www.commonwealthfund.org/~/media/files/publications/fund-report/2014/jun/1755_davis_mirror_mirror_2014.pdf. Published 2014. Accessed January 2, 2017.
13. Bjornberg A. Euro Health Consumer Index. www.healthpowerhouse.com/files/EHCI_2015_report.pdf Published January 2016. Accessed January 24, 2017.
14. Romp M, Merkx P. Zorgthermometer Vektis. Verzekerden in beeld 2015. http://www.vektis.nl/downloads/Publicaties/2015/Zorgthermometer%20nr14/#1/z. Published April 2015. Accessed January 3, 2017.
15. Heijne S. Waarom de zorgpremie in 2017 met gemiddeld 9% stijgt. NtvG. 01-2017.
16. van den Berg M, de Boer D, Gijsen R, et al. RIVM Zorgbalans 2014. De prestaties van de Nederlandse gezondheidszorg. http://www.gezondheidszorgbalans.nl/ dsresource? type=pdf&disposition= inline&objectid= rivmp:259835&versionid= &subobjectname=. Published September 2014. Accessed January 2, 2017.
17. de Hoon S, Verberne L, Davids R, et al. Zorg door de huisarts. NIVEL Jaarcijfers en trendcijfers 2011-2015. http://www.nivel.nl/sites/default/files/jaarrapport_huisarts_2015.pdf. Published October 2016. Accessed January 2, 2017.
18. Greysen SR, Stijacic Cenzer I, Auerbach AD, Covinsky KE. Functional impairment and hospital readmission in Medicare seniors. JAMA Intern Med. 2015;175(4):559-565.
19. van der Horst H. Medisch contact. http://www.medischcontact.nl/archief-6/Tijdschriftartikel/153800/Spoedhulp-voorouderen-in-het-gedrang.html. Published May 2016. Accessed January 2, 2017.
23
20. VUmc Netwerk Acute Zorg. Ontwikkelingen aanbod acute patiënten SEH’s RsHsQQ, 2014 en 2015 ROAZ-regio VUmc. https://www.vumc.nl/afdelingen-themas/719134/8513847/8514115/20150929_RapportAcuteZorgQ11.pdf. Published September 2015. Accessed August, 2016.
21. Merten H, Zegers M, de Bruijne MC, Wagner C. Scale, nature, preventability and causes of adverse events in hospitalised older patients. Age Ageing. 2013;42(1):87-93.
22. Vegting IL, Alam N, Ghanes K, et al. What are we waiting for? Factors influencing completion times in an academic and peripheral emergency department. Neth J Med. 2015;73(7):331-340.
23. Singal RK, Sibbald R, Morgan B, et al. A prospective determination of the incidence of perceived inappropriate care in critically ill patients. Can Respir J. 2014;21(3):165-170.
24. Cardona-Morrell M, Kim J, Turner RM, Anstey M, Mitchell IA, Hillman K. Non-beneficial treatments in hospital at the end of life: a systematic review on extent of the problem. Int J Qual Health Care. 2016;28(4):456-469.
25. Stellinga M, Weeda F. Interview E. Schippers in NRC Handelsblad 08-09-2012.
26. Hollnagel E, Wears RL, Braithwaite J. From Safety-I to Safety-II: A White Paper. https://www.england.nhs.uk/signuptosafety/wp-content/uploads/sites/16/2015/10/safety-1-safety-2-whte-papr.pdf. Published 2015, Accessed January 10, 2017.
27. Arora VM, Farnan JM. Inpatient Service Change: Safety or Selection? JAMA. 2016;316(21):2193-2194.
28. Pannick S, Beveridge I, Wachter RM, Sevdalis N. Improving the quality and safety of care on the medical ward: A review and synthesis of the evidence base. Eur J Intern Med. 2014;25(10):874-887.
29. Vidyarthi AR, Arora V, Schnipper JL, Wall SD, Wachter RM. Managing discontinuity in academic medical centers: strategies for a safe and effective resident sign-out. J Hosp Med. 2006;1(4):257-266.
30. Averbukh Y, Southern W. The impact of the number of admissions to the inpatient medical teaching team on patient safety outcomes. J Grad Med Educ. 2012;4(3):307-311.
31. Institute of Medicine. To err is human: building a safer health system. http://www.nationalacademies.org/hmd/~/media/Files/Report%20Files/1999/To-Err-is-Human/To%20Err%20is%20Human%201999%20%20report%20brief.pdf. Published November 1999. Accessed December 28, 2016.
32. Braithwaite J, Wears RL, Hollnagel E. Resilient health care: turning patient safety on its head. Int J Qual Health Care. 2015;27(5):418-420.
33. Francis R. Report of the Mid Staffordshire NHS Foundation Trust Public Inquiry. London. Stationery Office. 2013.
34. Basisset Kwaliteitsindicatoren 2014. Utrecht: Inspectie voor de gezondheidszorg, 2015. Accessed January, 2016.
35. Veiligheidsmanagementsysteem. Communicatie tussen hulpverleners volgens het SBAR-proces. VMS Veiligheidsprogramma. 02-2009.
36. Subbe CP, Gao H, Harrison DA. Reproducibility of physiological track-and-trigger warning systems for identifying at-risk patients on the ward. Intensive care medicine. 2007;33(4):619-624.
37. Langelaan M, de Bruijne M, Baines R, et al. NIVEL Monitor Zorggerelateerde Schade 2011/2012. Dossieronderzoek in Nederlandse ziekenhuizen. http://www.nivel.nl/sites/default/files/bestanden/monitor_zorggerelateerde_schade_2011_2012.pdf. Published 2013, Accessed December 28, 2016.
38. Poortmans J. Onderzoek - Zorgkwaliteit in ziekenhuizen. Driemaaldaags een pijnschaal. De Groene Amsterdammer. 2016(49).
39. Balla U, Malnick S, Schattner A. Early readmissions to the department of medicine as a screening tool for monitoring quality of care problems. Medicine (Baltimore). 2008;87(5):294-300.
40. Actal Eindrapportage. Onderzoen naar de ervaren regeldruk onder verpleegkundigen in het ziekenhuis. 114-bijlage-Eindrapportage-EY-ervaren-regeldruk-verpleegkundigen.pdf. Published September 2012. Accessed January 3, 2017.
24
41. O’Leary KJ, Buck R, Fligiel HM, et al. Structured interdisciplinary rounds in a medical teaching unit: improving patient safety. Arch Intern Med. 2011;171(7):678-684.
42. Giesen P, Smits M, Huibers L, Grol R, Wensing M. Quality of after-hours primary care in the Netherlands: a narrative review. Ann Intern Med. 2011;155(2):108-113.
43. Mardon RE, Khanna K, Sorra J, Dyer N, Famolaro T. Exploring relationships between hospital patient safety culture and adverse events. J Patient Saf. 2010;6(4):226-232.
44. Beaver K, Tysver-Robinson D, Campbell M, et al. Comparing hospital and telephone follow-up after treatment for breast cancer: randomised equivalence trial. BMJ. 2009;338:a3147.
45. Scott I, Vaughan L, Bell D. Effectiveness of acute medical units in hospitals: a systematic review. Int J Qual Health Care. 2009;21(6):397-407.
46. Cooke MW, Higgins J, Kidd P. Use of emergency observation and assessment wards: a systematic literature review. Emerg Med J. 2003;20(2):138-142.
47. Group V. De Acute Opname Afdeling ‘Het wat, waarom en hoe’. 2011.
48. Reid LE, Dinesen LC, Jones MC, Morrison ZJ, Weir CJ, Lone NI. The effectiveness and variation of acute medical units: a systematic review. Int J Qual Health Care. 2016.
49. Ludikhuize J, Smorenburg SM, de Rooij SE, de Jonge E. Identification of deteriorating patients on general wards; measurement of vital parameters and potential effectiveness of the Modified Early Warning Score. J Crit Care. 2012;27(4):424 e427-413.
50. Hillman K, Parr M, Flabouris A, Bishop G, Stewart A. Redefining in-hospital resuscitation: the concept of the medical emergency team. Resuscitation. 2001;48(2):105-110.
51. Ridley S, Jackson R, Findlay J, Wallace P. Long term survival after intensive care. BMJ. 1990;301(6761):1127-1130.
52. van der Schaaf TW HM. PRISM-Medical. A Brief Description. Eindhoven University of Technology, Faculty of Technology Management, Patient Safety Systems: Eindhoven. 2005.
53. van Vuuren W SC, van der Schaaf TW. The Development of an Incident Analysis Tool For the Medical Field. Eindhoven University of Technology: Eindhoven. 1997.
54. van Wagtendonk I, Smits M, Merten H, Heetveld MJ, Wagner C. Nature, causes and consequences of unintended events in surgical units. Br J Surg. 2010;97(11):1730-1740.
55. Alam N, Hobbelink EL, van Tienhoven AJ, van de Ven PM, Jansma EP, Nanayakkara PW. The impact of the use of the Early Warning Score (EWS) on patient outcomes: a systematic review. Resuscitation. 2014;85(5):587-594.
56. Zanocchi M, Maero B, Martinelli E, et al. Early re-hospitalization of elderly people discharged from a geriatric ward. Aging clinical and experimental research. 2006;18(1):63-69.
57. Jackson AH, Fireman E, Feigenbaum P, Neuwirth E, Kipnis P, Bellows J. Manual and automated methods for identifying potentially preventable readmissions: a comparison in a large healthcare system. BMC medical informatics and decision making. 2014;14:28.
58. van Walraven C, Jennings A, Forster AJ. A meta-analysis of hospital 30-day avoidable readmission rates. Journal of evaluation in clinical practice. 2012;18(6):1211-1218.
59. Cakir B, Gammon G. Evaluating readmission rates: how can we improve? Southern medical journal. 2010;103(11):1079-1083.
60. Meisenberg BR, Hahn E, Binner M, et al. ReCAP: Insights Into the Potential Preventability of Oncology Readmissions. J Oncol Pract. 2016;12(2):153-154.
61. Keesman A. Publicaties HSMR en SMR’s verplicht. https://www.hsmr.nl/publicatie-hsmr-en-smrs-verplicht/ Published 2013. Accessed January 5, 2017.
62. van der Laan. J, de Bruin. A, van den Akker-Ploemacher J, Penning C, Pijpers F. HSMR 2014 methodological report. http://www.hsmr.nl/wp-content/uploads/2016/01/2015hsmrmethodologicalreport2014.pdf. Published 2015. Accessed March, 2016.
THE STARTING POINT OF THE HOSPITAL CARE CHAIN: NOVEL WAYS OF ORGANIZING CARE
CHAPTER 2
ACUTE MEDICAL UNITS: THE WAY TO GO? A LITERATURE REVIEW
Louise S. van Galen & Eline M. Lammers | Linda J. Schoonmade | Nadia Alam | Mark H. Kramer | Prabath W. Nanayakkara
Eur J Intern Med 2017;39:24-31
‘Focus on systems that are real rather than ideal’ Jeffrey Braithwaite
28
ABSTRACT Background
The acute healthcare chains in the Netherlands are increasingly under pressure because of
rising emergency department (ED) admissions, relative bed shortages and government policy
changes. In order to improve acute patient flow and quality of care through hospitals, an
acute medical unit (AMU) might be a solution, as demonstrated in the UK. However, limited
information is available concerning AMUs in the Netherlands.
Therefore the aims of this study were to:
1. Systematically provide an overview of current international literature regarding
the effectiveness of AMUs
2. Give an overview of the current situation in the Netherlands
3. Make recommendations that could be used for future Dutch AMU guidelines
Methods
A systematic literature search was performed searching 3 electronic databases: PubMed,
Cochrane and EMBASE. All 106 hospitals in the Netherlands were contacted, inquiring about
the status of an ED, the AMU or future plans to start one.
Results
The literature search resulted in 31 studies that met inclusion criteria. In general, these studies
reported significant benefits on number of admissions, hospital length of stay (LOS), mortality,
other wards and readmissions. Among the Dutch hospitals with an Ed, 33 out of 93 Dutch
hospitals implemented an AMU or similar ward, these are however organized heterogeneously.
Following current trends, more AMUs are expected to be realized in the future.
Conclusion
In order to improve the current strain on the Dutch acute healthcare system, an AMU could
potentially provide benefits. However, a uniform guideline is warranted to optimize and
compare quality of care throughout the Netherlands.
Keywords
Acute medical unit, Healthcare quality, Implementation guidelines
Highlights
1. An Acute Medical Unit (AMU) could improve patient flow and hospital quality
of care.
2. Current literature shows an overall beneficial effect after implementation.
3. In the Netherlands one-third of the hospitals have implemented an AMU.
4. Existing AMUs are designed heterogeneously, this calls for a uniform guideline.
29
INTRODUCTIONAcute healthcare chains in the Netherlands have been increasingly under pressure in the last
decade. Due to the aging population and the policy changes in the Dutch healthcare system
the hospitals have seen a steady rise in emergency admissions. In addition, these patients
present themselves with a higher severity of illness because this population is increasingly
older and often have multiple comorbidities and polypharmacy.1 This strain on the emergency
departments (ED) increases ED length of stay (LOS) which in turn is related to a higher
morbidity and mortality.2,3 Older patients tend to have atypical clinical presentations leading
to more diagnostics and 75% spend too long in the ED.4 A recent report has stated that,
for the first time in years, the adjusted mean hospital length of stay in the Netherlands has
increased by 8,5 %.5 In addition to increasing admissions, a relative shortage of hospital beds
in combination with a high percentage of beds not being used efficiently have also contributed
to this increase.5 Therefore, the introduction of novel ways to improve the efficiency of
the whole acute care chain is needed.
Acute medical units (AMU), which have widely been implemented in the United Kingdom
(UK) and Australia, might offer a solution to the current burden on acute healthcare chains
in the Netherlands. AMUs or units with similar names [i.e. medical assessment units (MAU),
emergency assessment units (EAU)] are defined as hospital wards specifically staffed and
equipped to receive medical inpatients presenting with acute medical illness from the ED,
outpatient clinics and/or the community for expedited multidisciplinary and medical specialist
assessment, care and treatment for up to a designated period (typically between 24 and 72
hours) prior to discharge or transfer to other wards.6 Frequently admitted medical conditions
are for example: sepsis, acute coronary syndrome, pneumonia, and COPD.7 An implementation
of an AMU by reorganizing already available hospital beds could potentially lead to a better
and a more efficient use of available beds. Other possible advantages of implementing an
AMU is to improve the quality of care by stimulating multidisciplinary (team)work, simplifying
logistics, and accelerating diagnostics and therapy. In addition, an AMU potentially reduces
complications and unintended errors.8
Since 2000, an increasing number of hospitals have started implementing AMUs in
the Netherlands. However, there are no national guidelines setting uniform standards for
creating such a ward and therefore these AMUs have been established heterogeneously.
Also, no value-based indicators have been developed to evaluate the quality of healthcare in
these AMUs.
Previous work has already shown the effectiveness of an AMU on patient outcomes and
certain quality indicators.6,9,10 Since, in the coming years more AMUs are expected to start up
in the Netherlands in order to cope with the increasing acute patient flow, a clear guideline
could provide a gold standard from which the quality of these units could be standardized
and measured.
30
Therefore, our primary aim is to [1] Provide a systematic overview of current knowledge in
international literature about the effectiveness of AMUs, [2] give an overview of the current
situation in the Netherlands, and [3] make recommendations that could be used to formulate
national AMU guidelines.
METHODSSearch strategy
In order to find the correct key terms used to identify units similar to ‘acute medical units’, we
scanned previously written literature overviews and documentations on the matter. To identify
all relevant publications search terms included controlled terms from MeSH in PubMed, EMtree
in EMBASE.com plus free text terms for The Cochrane Library. Terms expressing ‘acute medical
units’ and their synonyms identified were used. The full search strategies and synonyms
used for all databases can be found in the Supplementary Information (Appendix A: search
strategy). Afterwards, a systematic search was performed in the bibliographic databases
PubMed, EMBASE.com and The Cochrane Library (via Wiley) from inception to 08-08-2016 in
collaboration with a medical librarian (LS). The search was performed without date, language
or publication status restriction. The references of the identified articles were searched for
relevant publications.
Selection process
A total of 1909 records were excluded from the analysis primarily, because these articles
did not address the effect of implementation of an AMU as their main subject (for example,
articles about HIV testing or the use of antibiotics in the AMU obtained during the search were
excluded). Both reviewers (LG, EL) independently screened all remaining articles identified
with the search on title and abstract. If considered necessary, the whole article was checked
for eligibility. After individual screening, included articles were compared and differences in
opinions between the reviewers were resolved through consensus. Inclusion criteria were (i)
the study should report the outcome measures on an AMU (ii) the AMUs or similar wards
described should have a minimum LOS of 24 h and should conform to the definition given
in the introduction [designated hospital wards specifically staffed and equipped to receive
medical inpatient presenting with acute medical illness from the ED, outpatient clinics and/
or the community for expedited multidisciplinary and medical specialist assessment, care and
treatment for up to a designated period (typically between 24-72 h) prior to discharge or
transfer to other wards] (iii) the AMU must admit medical patients because the AMUs were
originally designed for medical patients, and most available literature was on medical patients
(iv) the study should include patients from 14 years and above (v) the unit in the study must
be functional 7 days a week, 24 hours a day. Exclusion criteria were (i) no full text available
(in English) (ii) certain publication types: editorials, letters, legal cases, annual hospital reports,
etc. (iii) AMUs which solely admitted surgical patients, (iv) psychiatric AMUs.
Literature assessment
After identifying suitable studies both reviewers independently assessed the papers for their
quality using the NICE guidelines for qualitative studies and cohort studies.11 No controlled
31
trials were found. An item ‘clinical relevance’ was added to individually assess the relevance of
the paper in answering the primary objective (varying in four categories from ‘poor’ to ‘highly’
relevant). A standardized data worksheet was used for data extraction with the following
variables: study design, ward description, study population, time period, aim, methods,
primary endpoint, secondary endpoint, clinical relevance and overall study quality (NICE). This
worksheet was filled out after reaching consensus by both reviewers during their meetings.
Search results
In total the literature search generated 3386 references, excluding duplicates. Flowchart 1
(Fig. 1) illustrates our search and selection process. After selection 100 articles remained and
2 extra articles were added after reference checks. After assessment by both reviewers, a total
of 31 studies were suitable for inclusion. Appendix B shows an overview of all included articles
in a table with the variables selected for data extraction.
Situation in the Netherlands
To evaluate the current situation in the Netherlands, all 106 Dutch hospital locations were
telephoned and inquired 1) whether they had an ED 2) if yes, whether they had an AMU 3)
if yes, since when and 4) how their AMU was organized. If hospitals did not have an AMU,
they were asked if they had a similar construction such as a short stay unit, an observation
unit or an acute ambulatory unit, all being located on or near the ED. Subsequently, they were
asked if they had plans to implement an AMU in the future. The staff members involved with
the management were inquired about the organization of the AMU if available. The results
were processed anonymously. To assure no hospitals were missed a registered list of all
hospitals was used. For this analysis we only included hospitals that delivered (unplanned)
acute care, and thus had an ED. Fused hospitals were counted and assessed separately if they
both had a separate ED.
RESULTSStudy design
The studies included in this review were published in different countries; twelve originate from
the UK7,12-22, seven from Australia23-29, seven from Ireland30-36, one from Australia and New-
Zealand37, one from Canada38, one from Denmark39, one from Hong Kong40 and one from
the Netherlands41. The data of all studies were collected between 1993 and August 2016. 27
of the 30 studies had a quantitative design, three had a qualitative design.16,18,26
The most common types of quantitative study designs were comparisons of the situation
before and after establishment of the AMU (thirteen in total) and comparison of the AMU
to other medical wards/clinical teaching units (four in total). Furthermore, three studies used
questionnaires to assess the outcomes in a cross-sectional design.19,21,35 Brand et al. (2010)
and Suthers et al. (2012) made two comparisons: pre-AMU vs. AMU and AMU vs. ward.23,27
Elder et al. (2015) compared standard care to physician at triage and the combination of
physician at triage with the medical assessment unit.24 Relihan et al. (2009) compared
32
Figure 1 | Flowchart search and selection procedure of studies
the acute medical assessment unit to international benchmarking data.35 Sullivan et al. (2013)
compared different AMUs to emergency admissions in other specialties and to scheduled
admissions.21 The qualitative studies all used either focus groups, semi-structured interviews
or questionnaires. Two had a cross-sectional design and one compared the situation before
and after implementation of the AMU.18
All three articles published by Boyle et al. described the situation before and after reorganization
of the ED and the AMU into one emergency assessment unit.12-14. Seven articles described
the same 59-bedded acute medical assessment unit in an Irish hospital.30-36 Pradhan et al.
(2015) and McNeill et al. (2011) were multicentre analyses that includes different AMUs,
55 in the UK and 32 in Australia and New-Zealand.19,37 Sullivan et al. (2013) performed
a national study of different AMUs within the UK (the exact amount of AMUs is unknown).21
In order to determine factors that could potentially affect outcome measures, three studies
used a logistic regression model.30,34,36 In one study univariate and loglinear analyses were
33
performed.32 Multivariate logistic regression and multiple linear regression were used in two
different studies.21,27 One longitudinal cohort study performed by Van Galen et al. (2016) used
a Bonferroni correction to compare between different time periods.41
Study population
All AMUs included in this review organized care for ‘medical’ patients. Medical patients are
patients admitted for the following specialties: cardiology, gastroenterology, internal medicine,
nephrology, neurology, oncology, pulmonary medicine, geriatrics, and rheumatology. In
addition, the emergency assessment unit described in the three articles by Boyle et al. also
consist of a surgical and a paediatric unit.12-14 Furthermore, Hadden et al. (1996) described
an AMU dedicated to both medical and surgical patients.17 For these four articles we only
report on medical patients (Appendix 2). Van Galen et al. described an acute admission unit
with medical and surgical patients, but no results for the separate group of medical patients
were reported and therefore could not be displayed separately.41 Figure 2 shows the Dutch
AMU model, compared to the traditional model of emergency admissions and the patient flow
model mostly used on the AMU in the United Kingdom.
Becket et al. and Vork et al. described patients in the AMU who were referred from their
general physician (GP) or the ED.7,39 The acute admission unit described by van Galen et
al. (2016) received patients mainly from the ED but also from the outpatients clinic.41 In
all other included studies, patients were referred from the ED only or the process was not
described clearly.
Figure 2 | Models of acute admissions (Traditional model, AMU in the Netherlands, AMU in the UK)
34
Ward description
The AMUs included in this paper have heterogeneous designs. The units differ in name, location
(next to ED or not), number of beds, number of nurses and physicians and the maximum LOS on
the unit before discharge home or to another medical ward. When reported, the number and
type of physicians on rotation differed significantly. In addition, the frequency of supervision
and number of ward rounds varied.
Assessed NICE quality and clinical relevance
Cohort studies were assessed for selection, attrition, performance and detection bias
according to the NICE-criteria.11 Boyle et al. (2012), Coary et al. (2014), McNeill et al. (2011),
Moloney et al. (2005), Slatyer et al. (2008), Suthers et al. (2012) and Rooney et al. (2008)
were rated the highest quality cohort studies included in this review.13,26,30,34,36,37 Hadden et
al. (1996) and Yong et al. (2011) were rated the lowest quality cohort studies included.17,29
Qualitative studies were scored for the quality of their study methodology, data collection and
analysis. The qualitative study with the best rating was Hanlon et al. (1997) and the study
with the lowest rating was McErlain Burns et al. (1997).16,18 Basey et al. (2016), Hadden et al.
(1997) and Yong et al. (2011) were assessed as poorly relevant in contribution to answering
the primary objective of this paper,17,22,29 while Coary et al. (2014), Conway et al. (2014),
McNeill et al. (2011), Moloney et al. (2005) , Rooney et al. (2008) and Suthers et al. (2012)
were assessed as highly relevant.27,30,31,34,36,37
Patient outcomes*(only significant outcomes are reported, other outcomes can be found in table 2/appendix B)
Number of emergency admissions
Two articles from the Irish acute medical assessment unit describe an increase of 9,7% (5038
vs. 5528 admissions) and 4.4%(5476 vs. 5715 admissions) in the number of acute medical
admissions after establishment of an AMU.32,34 Man Lo et al. (2014) in Hong Kong found
a 14.4% decrease in the average emergency admissions rate after reorganization into an
emergency medicine ward (absolute figures unknown).40
In the UK, Boyle et al. (2008) found a 16.3% decrease in the number of acute medical admissions
after reorganization of a separate AMU and ED into one emergency assessment unit, from
3667 admissions during a 3-month period before reorganization to 3068 afterwards.14 No
other reorganization structure as such was described elsewhere.
Length of stay (LOS)
St. Noble t al. (2008), Downing et al. (2008) and Williams et al. (2000) found a decrease
in mean hospital LOS after reorganization into an AMU from 9.3 to 7.8 days, from 5.5 to
4.6 days and from 3.8 to 2.6 days respectively.15,20,28 Moloney et al. (2005, 2006 and 2007),
Conway et al. (2014), Vork et al. (2011), Boyle et al. (2008) and Suthers et al. (2012) reported
a decrease in median hospital LOS after establishment of an AMU: from 6 days to 5 days, from
7.1 to 6.9 days, from 4.1 to 3.8 days, from 7 to 5 days and from 6.8 days to 5.2 days.14,27,31-34,39
35
Hanlon et al. (1997) reported no significant difference in median hospital LOS.18 Man Lo et
al. (2014) reported a significant decrease in hospital LOS (excluding the time spent in the ED)
after establishment of an AMU: from 5.1 days to 4.1 days.40 In addition, a study reporting on
a medical assessment and coordination Unit in Australia showed that patients that solely stay
in this unit have the shortest LOS (1.9 days), followed by patients that only stay in the medical
ward (4.9 days). Patients that stay in the Medical Assessment and Coordination Unit first and in
the medical ward afterwards have the longest LOS (9.2 days). St. Noble et al. (2008) reported
an increased trend of patients discharged in less than 24 h and 48 h after implementation
of an AMU.20 In contrast, Yong et al. (2011) showed a decrease in the number of patients
discharged within 72 h.29
Mortality
The acute medical assessment unit used in the analysis from Ireland showed a significant
decrease in 30-day mortality and annual mortality after establishment this unit from 8.8% to
5.6% and from 12.6% to 7%, respectively.36 Coary et al. (2014) also calculated the multivariate
estimate odds ratio (OR) of in-hospital 30-day mortality on this Irish acute medical assessment
unit and found an OR of 0.67 (95% CI; 0.71-0.82).30 In addition, Conway et al. (2014)
calculated the relative risk ratio (RRR) of mortality by episode and by patient in this cohort
and saw a decrease of both.31 An Australian study performed by Brand et al. (2010) obtained
a significantly lower in-hospital mortality rate in their medical assessment and planning Unit
(3.2%) compared to the rate in the medical ward (7.6%), however there was no significant
difference between overall in-hospital mortality before and after implementation.23 Boyle et
al. (2011) saw a lower actual all-cause mortality rate than the expected all-cause mortality,
whereas Yong et al. (2011) measured no significant difference, the all-cause mortality persisted
at 2%.12,29 Boyle et al. (2012) also calculated the hospital standardized mortality ratio (HSMR)
and had the lowest ratio of the included UK hospitals in 3 out of 4 years.13 Suthers et al. (2012)
reported no significant change in mortality either (measurement of mortality undefined).27
Effect on other wards
Beneficial effects on other wards have been described by Downing et al. (2008) who reported
a significant decrease in the number of medical patients in non-medical beds after reorganization
(38 vs. 11 patients).15 Man Lo et al. in Hong Kong found an average monthly decrease of 187
patients (15%) admitted to other medical wards than the AMU.40 Hadden et al. (1997) showed
a lower percentage of patients that needed transfer to an outpatient clinic from the short stay
ward in comparison to the medical ward.18 Another beneficial effect, solely described by Yong
et al. (2011), is the decrease of patients that needed transfer to the intensive care unit (ICU)
(3% before vs. 2% after reorganization).29. Besides positive effects, Abenhaim et al. (2000)
reported a percentage of 20% of patients who required transfer from the medical short stay
unit to ‘ordinary wards’ because of the severity of their disease.38 Also, in an Australian study
comparing 32 AMUs, McNeill et al. (2011) stated that 10%-19% of potential AMU patients
required transfer to another medical ward because of limited capacity.37
36
Emergency Department LOS / outpatient waiting times
Two studies reported a significant decrease in median ED LOS after establishment of an AMU,
3.1 h to 2.9 h and 8.7 h to 8.0 h respectively.24,27 Ong et al. (2012) found a significantly shorter
ED LOS for medical assessment unit patients than for medical ward patients (4.9 h vs. 6.5
h).25 Hanlon et al. (1997) saw no significant change in outpatient waiting times.18 Elder et al.
(2014) showed a significantly increased compliance to the 4-h target in the ED from 53.3%
before the medical assessment unit to 73% after implementation.24 However, McNeill et al.
(2011) reported that in more than 50% of the included units patients had to wait more than
4 hours on a daily basis because of limited AMU capacity.37 Basey et al. (2016) argued that
68% of patients were treated within the 4-h target.22 In addition, Moloney et. al (2005 and
2006) showed a decrease in the number of patients waiting in the ED after establishment of
the acute medical assessment unit, 9 patients before implementation, vs. 8 afterwards.32,34
Readmission rate
Five studies describe the effect of an AMU on the readmission rate. Two studies recorded no
change in 28-day readmission rate after reorganization into an AMU.15,33 Brand et al. (2010)
reported a decrease in the number of readmissions within 28 days in the medical hospital
population after establishment of the medical assessment and planning unit (MAPU) (6 %
readmitted vs. 5.4% readmitted), yet there was no significant difference between the MAPU
and the non-MAPU group.23Abenhaim et al. (2000) obtained a lower 30-day readmission rate
in the AMU than in the ward (9.6% vs. 13.9%), however this was not statistically significant.38
The only study performed in Denmark included in this literature review, states a significant
decrease in 30-day readmission rate from 19.8% before implementation of the Medical
Admission Unit to 14.6% afterwards.39
Staff and patient perspectives
The qualitative studies in this review mostly reported on the staff and patient experiences.
One study from the UK presented positive reactions to the AMU from both GPs and patients.
Furthermore, they also reported 64% of patients who were in the hospital before and
after the establishment of the AMU indicated improvement.16 Another study from the UK,
performed by Hanlon et al. (1997) also presented positive feedback from patients and staff.
However, medical staff were significantly more concerned about blocked beds in other clinical
wards than before implementation and nurses expressed that they felt significantly more
stressed in the AMU.18 A cross-sectional study performed in Ireland showed that the staff rated
their working environment in the AMU higher than the international benchmarking data.35
Van Galen et al. (2016) on a Dutch AMU found that patients felt safe in the AMU.41 Slatyer
et al. (2013) yet reported that patients and family members often leave the hospital with
limited understanding of their health problems.26 Sullivan et al. (2013) showed that the AMU
scored significantly lower in patient reviews than emergency admissions in other specialties or
scheduled admissions.21
37
Situation in the Netherlands
Of the 106 locations of hospitals in the Netherlands, 93 had an emergency department. Out
of these, more than one third had an acute medical unit or a comparable unit such as an
acute admission unit (AAU) or acute admission and diagnostic unit (AADU). Characteristics
of the Dutch hospitals with and without an AMU are described in Table 1. Eleven hospitals
reported to have another type of ward which was comparable to an AAU but did not meet
the definition of an AMU as previously mentioned. These were often wards functioning as
short stay units (<24 hours) which were located on or very near the ED entitled ‘short-stay unit’
or ‘observatorium’. Emergency physicians were mostly responsible for these wards. However,
the concept of the AMUs are fundamentally different from a short-stay unit. The short stay
units do not change existing system of patient flow, and can therefore be considered as
‘temporary normal care beds’ since they are simply a place where patients admitted through
the ED are waiting for a clinical bed and are temporarily placed for no longer than 12 to 24 h.
Therefore, these units do not share the essential principles of an AMU which are intensive and
active supervision, multi-disciplinary teamwork, and rapid diagnostics and therapy.
The first AMU in the Netherlands was implemented in 2000. Fig. 3 shows the trend in AMUs
implementation since 2000. These AMUs had no uniform organization. Firstly, the location
and number of beds differed. Most AMUs were placed on a separate ward near the ED and
the diagnostic facilities. The number of beds differed from 4 to 54 beds. Secondly, the AMUs
reported different specialties which were able to admit to the ward, diverging from all
medical and surgical specialties to only internal medicine patients. Most AMUs do not admit
cardiology patients, pregnant patients and children. Thirdly, different daily schedules were
described, varying from one doctor being available only in the morning, to intensive consultant
supervision during the day. Two other hospitals reported the start of an AMU in the next year,
five are currently talking about implementation but do not have any concrete plans. Three
reported to have already investigated this reorganization but decided not to go ahead, because
they expected that AMU would not be an effective addition to the current organization of
their hospitals.
Table 1 | Characteristics of Dutch hospitals with and without AMU
Hospitals with an ED 93
Hospitals with an AMU 33
Hospitals without an AMU but with SSU* 11
Beds AMU 887
Beds SSU 101
Mean beds AMU (SD) 28 (13)
Mean beds SSU (SD) 11 (6)
Mean Maximum Expected LOS (in hours) goal AMU (SD) 50,9 (10,4)
Mean Maximum Expected LOS (in hours) goal SSU (SD) 26,2 (11,8)
Mean clinical beds hospital (SD) 424 (255)
*SSU: Short Stay Unit or similar wards with a LOS <24 hours, usually on the ED
38
33
Table 1 | Characteristics of Dutch hospitals with and without AMU
Hospitals with an ED 93
Hospitals with an AMU 33
Hospitals without an AMU but with SSU* 11
Beds AMU 887
Beds SSU 101
Mean beds AMU (SD) 28 (13)
Mean beds SSU (SD) 11 (6)
Mean Maximum Expected LOS (in hours) goal AMU (SD) 50,9 (10,4)
Mean Maximum Expected LOS (in hours) goal SSU (SD) 26,2 (11,8)
Mean clinical beds hospital (SD) 424 (255)
*SSU: Short Stay Unit or similar wards with a LOS <24 hours, usually on the ED
Figure 3 | Trend AMUs Netherlands 2000‐2016
DISCUSSION The primary aim of this review was to systematically provide an overview of current literature about the effectiveness of AMUs in terms of different patient outcomes. Most literature included in this review originated from three countries (the UK, Ireland and Australia), in which AMUs have already been implemented on a large scale. Overall, beneficial effects of implementing an AMU were reported. According to the current international literature, implementation of an AMU with reorganization of the existing beds mostly results in higher numbers of emergency medical admissions.32,34 In contrast, one study by Man Lo et al. describe a decrease in the amount of acute medical admissions. One of the aims of the reorganization to an emergency medicine ward described in this article was to reduce the number the of unnecessary admissions which could be an explanation for this decrease.40 Another important finding emerging from our study was that not only hospital LOS was significantly reduced when introducing the AMU, but also in‐hospital mortality and 28‐day readmission rates decreased. In most studies, ED LOS also decreased significantly after implementation of an AMU. In addition, several positive effects of the implementation of an AMU on other wards were reported: a
0
5
10
15
20
25
30
35
2000 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
Num
ber o
f AMUs
Figure 3 | Trend AMUs Netherlands 2000-2016
DISCUSSION The primary aim of this review was to systematically provide an overview of current literature
about the effectiveness of AMUs in terms of different patient outcomes. Most literature
included in this review originated from three countries (the UK, Ireland and Australia), in
which AMUs have already been implemented on a large scale. Overall, beneficial effects
of implementing an AMU were reported. According to the current international literature,
implementation of an AMU with reorganization of the existing beds mostly results in higher
numbers of emergency medical admissions.32,34 In contrast, one study by Man Lo et al. describe
a decrease in the amount of acute medical admissions. One of the aims of the reorganization
to an emergency medicine ward described in this article was to reduce the number the of
unnecessary admissions which could be an explanation for this decrease.40 Another important
finding emerging from our study was that not only hospital LOS was significantly reduced
when introducing the AMU, but also in-hospital mortality and 28-day readmission rates
decreased. In most studies, ED LOS also decreased significantly after implementation of an
AMU. In addition, several positive effects of the implementation of an AMU on other wards
were reported: a decrease of medical patients who were placed on non-medical beds, and
a decrease of transfers to other medical wards, outpatient clinics and the ICU.15,17,40 Patient
and staff’s perspectives were positive in most cases, for example improvements in admission
procedures resulting in more time for personal care and better cooperation between and
within the wards were mentioned.16,18 However, some studies reported coordination failures
such as insufficient communication to patients regarding their underlying health problems
at discharge from the AMU.26,37 In general, conclusions derived from studies performed in
the UK and Ireland were uniformly positive, but results from Australia were more mixed.
There is limited evidence available from the Netherlands. Although data from three university
hospitals in the Netherlands have demonstrated up to 14% (from 7000 to 8000 admissions)
increase in the number of emergency admissions after implementation of an AMU (personal
communication), these data have not been published yet.
39
The secondary aim of this study was to provide an overview of the current situation concerning
AMUs in the Netherlands. Through phone interviews we found that in one-third of the Dutch
hospitals where acute care is provided an AMU was present. We expect that if this trend
persists, almost half of the Dutch hospitals will have an AMU in approximately 10 years.
The AMUs in the above mentioned hospitals did have very different organizations, varying in
number of beds and organization of the ward. They had a goal of expected mean length of
stay of 50.9 h on average (SD 10,4 h) with a maximum length of stay between 48 and 72 h.
Our investigation has also shown that probably due to changing patient flow in acute care
chains (more elderly and sicker patients), many hospitals are working towards implementation
of an AMU, with the goal of improving patient flow.
Our study demonstrates that both internationally and in the Netherlands the AMUs are
very heterogeneously organized. Dutch hospitals started their AMUs independently without
proper (national) guidelines or support from national institutions, resulting in dissimilar
ward structures. In order to compare the efficiency and clinical outcomes among the AMUs,
a uniform organization is warranted. An AMU must be designated with the aim of improving
patient flow and efficiency, stimulate multidisciplinary healthcare, increase supervision and
take away borders between different specialties in the hospital.
In the Netherlands, however, we have limited data to conclude whether the currently
implemented AMUs are optimally organized. If we are aiming to tackle the current rise in
emergency admissions, and make patient flow more efficient through introduction of AMUs,
introducing uniform national guidelines may help hospitals to introduce the AMUs in a uniform
and efficient way. An important issue already stated by the Vreeland Group, a commercial
group involved in the implementation of AMUs in some hospitals, in their acute admission unit
booklet, is that when implementing an AMU, one must not forget to restructure the wards
around it.8 Other wards will be spared of acutely sick patients being admitted constantly, but
adequate communication between the AMU and these wards and enough beds in these wards
is required in order to guarantee efficient patient flow from the AMU. When this flow is not
adequate, patients may need to wait longer in the AMU and may even lead to increased LOS in
the ED, as already described on a MAU in Australia: half of the 32 interviewed AMUs patients
had to wait more than 4 hours in the ED because of limited MAU capacity.37 Also, a longer
LOS could potentially harm patient safety, since these units are not designed to treat/care for/
cater to patients longer than 48-72 h. The establishment of an AMU is an integral process
within the hospital. The reorganization required to create an AMU leads to a redistribution of
patient groups, and when this process is not performed with absolute care, this may negatively
affect certain patient groups. This is also demonstrated in the article from Denmark, where
cardiology patients were negatively affected by reorganization, with an increased mortality
rate as a result.39
40
Recommendations
On the basis of these findings we will attempt to make a few recommendations to improve
the implementation of an AMU in the Netherlands.
Before realizing an AMU one should have a clear vision about the purpose of an AMU. An AMU
should provide timely high quality of care for the acutely ill patient, in a suitable environment
to achieve this. This requires several pre-implementation decisions: the location of the AMU,
the number of beds, the facilities it should have, the amount and type of staffing, a plan for
proper training of the staff, the type of protocols required, the specialties that can admit to
the AMU and details about the medical and logistical responsibilities. These matters ought to
be profoundly thought of and talked about with every specialty and other partners involved
before reorganization. In addition, the hospital management must strongly support the idea.
The location of an AMU should be near the ED, with easy access to diagnostic facilities such
as radiology. Bell et al. (2008) already reported that in some hospitals in the UK, because
of organizational issues, the AMU was not close to ‘the front door’ of the hospital, which
resulted in less productive use of the unit.42
In order to arrange the ward as effectively as possible, an adequate number of beds is
needed. To decide how many beds should be in the ward, one must have an idea of the acute
healthcare flow through the hospital. Mostly, acute health-care is more predictable than
regular (chronic) hospital care, and the number of acute admissions is fairly constant per
hospital throughout the year. Also, approximately 50% of the patients admitted to the AMU
go home after two days of admission. In addition to introducing an AMU with new care
concepts, the planned admission schemes (for example for chemotherapy) can be adapted in
such a way that the congestion of beds in the regular wards can be reduced or prevented so
that the flow of patients needing transfers can be achieved without delay. This withholds that
these regular wards should at least adapt some of the concepts such as intensifying supervision
and multidisciplinary care so that flow in these wards can also be improved. Congestion in
the AMU leading to longer LOS is undesirable and can affect patient safety because the AMUs
are not designed or equipped to provide chronic care to patients. Therefore, Scott et al. (2011)
proposed rigorous business rules around patient entry clearly defining processes of admission
and discharge of eligible patients, as success factors that could influence AMU operations – not
specifically describing in which the effect of these factors could be measured in.6 In order to
promote patient flow and increase bed occupancy, post-discharge services such as professional
homecare or ambulatory treatment should be easily accessible.
The teams staffing the AMU should consist of dedicated professionals who are able to work
in multidisciplinary teams and deal with a high turnover of patients. The staffing should be
continuous seven days a week with a clear rotation schedule with ward rounds twice daily. It
is highly recommended for the supervisors to work closely with their juniors in order to speed
up the decision making process and thereby patient flow. Enough staff is needed to ensure
41
an adequate patient; nurse ratio, which could prevent delay of diagnostics and inadequate
monitoring. Allied support teams such as physiotherapists, geriatric nurses, should be easily
accessible and see the patient within 24 hours when necessary to prevent unnecessary delay
in patients’ recuperation. The acute medical unit is very suitable to consult other specialties
easily, because patients from multiple specialties are admitted to these units and therefore
the physicians from these specialties are often present during ward rounds. New concepts such
as a short (maximum 15 minutes) multidisciplinary meeting every morning with al involved
specialties (all supervisors, their juniors, nurses and allied health support teams) to shortly
discuss the medical and logistical bottlenecks of all the patients present at the AMU, will be
of help not only to enhance teamwork but also to improve the flow of patients. In the VU
University Medical Centre (VUmc) this is called the ‘standing session’ in which every morning
the diagnosis, therapeutic plan and predicted LOS for each patient on the AMU are briefly
presented. In this way every specialty knows about the patients they might be consulted
for, and the doctors and nurses are informed about the availability of beds on the wards.
Also, the team can easily identify with whom they will/can work together to take care of
their patient which improves communication and coordination around the patient. Because
of the high turnover, handovers should take place systematically at every change of shift from
team to the next team, in order to provide continuous care. Handover systems such as SBAR
(situation, background, assessment, recommendation) should be introduced and propagated
vigorously to make sure that the handovers are performed uniformly.43
The original design of the AMU was used for medical patients only, however to improve
the cooperation of specialty colleagues in early review and maximum use of available beds one
can debate whether surgical specialties should also admit patients to the AMU. Many hospitals
in the Netherlands have already chosen this construction. This construction may be beneficial,
because when all medical beds are occupied, one can use a surgical bed for a medical patient
and the other way around. One must however ensure that nurses are adequately trained in
caring for both surgical and medical patients.
Before implementation adequate training for the staff on the ward is required. Previous
literature has reported that AMUs in the past have found difficulties recruiting nurses and
allied health staff with appropriate levels of acute assessment skills.6 Nurses are not specially
trained for working on an AMU and the resources and time to train them should be allocated.
A recent study performed in Ireland has done this by training their nurses with a program
called ‘Nursing the Acute Medical Patient’ and it has made significant contributions to both
staff development and the delivery of evidence-based care in the AMU.44 Also, since the AMU
is a clinical setting in which many diagnostics are performed on divergent types of patients,
it is a useful place to train young doctors and medical trainees. A location for them such as
a ‘skills lab’ or a library on or near the AMU should be thought of beforehand.
As previously mentioned, in order to improve quality and compare different AMUs throughout
the Netherlands a uniform guideline for an AMU is needed. In the UK, the Society of Acute
42
Medicine has guidelines to help hospitals starting up and instruct them about minimal
requirements.45 These have been proven very effective, and at present hospitals in the UK
without an AMU can hardly be found. Also, once implemented, standardized evidence-
based protocols have been described as improving patient flow and quality of care.6,42
These guidelines also offer an adequate way to assess quality of care and advise on
improvements constructively.
Limitations
The limitations of our literature review must be acknowledged. Firstly, one-third of the articles
included, extracted their data from one hospital in Ireland and another hospital in the UK.
These results therefore may not be extrapolated to the all AMUs in these countries. Secondly,
the designs of the AMUs were rather heterogeneous, which makes it hard to deduce which
elements of the AMU correlate strongly to better patient outcomes. Thirdly, little is known
about the precise effect of an AMU on bed occupancy on other medical wards , this information
is essential as sparing these wards from a continuous stream of acutely ill patients and thereby
improving patient care in these units is also one of the underlying motives to reorganize into
an AMU.
CONCLUSIONThis study has shown that current literature proves the AMU to be an effective model to
provide acute care and improve efficiency in the care chain thereby improving the patient
flow in acute care chain. However, the current situation in the Netherlands shows that
AMUs are individually implemented and are therefore heterogeneously organized. To
optimize the effectiveness and compare the quality of care in our AMUs, a clear national
guideline is needed to provide a gold standard, especially since the trend shows that
more and more AMUs will be developed in the coming years. Very few studies have been
performed on the effectivity of AMUs in the Netherlands and more research is needed
to investigate the efficacy of AMUs in Dutch setting so that national guidelines can
be formulated.
CONFLICT OF INTEREST STATEMENTThe authors have nothing to disclose.
FUNDINGThis research did not receive any specific grant from funding agencies in the public, commercial,
or not-for-profit sectors.
SUPPLEMENTARY INFORMATION Appendix A. ‘Search strategy’
Appendix B. ‘Systemic overview included studies’
43
REFERENCES1. VUmc Netwerk Acute Zorg. Ontwikkelingen aanbod acute patiënten SEH’s RsHsQQ, 2014 en 2015
ROAZ-regio VUmc. https://www.vumc.nl/afdelingen-themas/719134/8513847/8514115/20150929_RapportAcuteZorgQ11.pdf. Published September 2015. Accessed August, 2016.
2. Plunkett PK, Byrne DG, Breslin T, Bennett K, Silke B. Increasing wait times predict increasing mortality for emergency medical admissions. Eur J Emerg Med. 2011;18(4):192-196.
3. Schrijver EJ, Toppinga Q, de Vries OJ, Kramer MH, Nanayakkara PW. An observational cohort study on geriatric patient profile in an emergency department in the Netherlands. Neth J Med. 2013;71(6):324-330.
4. Samaras N, Chevalley T, Samaras D, Gold G. Older patients in the emergency department: a review. Ann Emerg Med. 2010;56(3):261-269.
5. Consultancy C. Ligduurmonitor Nederlandse ziekenhuizen 2013 & 2014 ‘Trendbreuk in de ligduur’. December 2015.
6. Scott I, Vaughan L, Bell D. Effectiveness of acute medical units in hospitals: a systematic review. Int J Qual Health Care. 2009;21(6):397-407.
7. Beckett DJ, Raby E, Pal S, Jamdar R, Selby C. Improvement in time to treatment following establishment of a dedicated medical admissions unit. Emerg Med J. 2009;26(12):878-880.
8. Vreeland Group. De Acute Opname Afdeling ‘Het wat, waarom en hoe’. Published 2011. Accessed August 2016.
9. Cooke MW, Higgins J, Kidd P. Use of emergency observation and assessment wards: a systematic literature review. Emerg Med J. 2003;20(2):138-142.
10. Reid LE, Dinesen LC, Jones MC, Morrison ZJ, Weir CJ, Lone NI. The effectiveness and variation of acute medical units: a systematic review. Int J Qual Health Care. 2016.
11. National Institute for Health and Care Excellence Guidelines. www.nice.org.uk Accessed August, 2016.
12. Boyle A, Fuld J, Ahmed V, Bennett T, Robinson S. Does integrated emergency care reduce mortality and non-elective admissions? A retrospective analysis. Emerg Med J. 2012;29(3):208-212.
13. Boyle AA, Ahmed V, Palmer CR, Bennett TJ, Robinson SM. Reductions in hospital admissions and mortality rates observed after integrating emergency care: a natural experiment. BMJ Open. 2012;2(4).
14. Boyle AA, Robinson SM, Whitwell D, et al. Integrated hospital emergency care improves efficiency. Emerg Med J. 2008;25(2):78-82.
15. Downing H, Scott C, Kelly C. Evaluation of a dedicated short-stay unit for acute medical admissions. Clin Med (Lond). 2008;8(1):18-20.
16. McErlain-Burns T, Byrne C, Smith C. Admissions wards. Admitting a need for change. Health Serv J. 1997;107(5542):33.
17. Hadden DS, Dearden CH, Rocke LG. Short stay observation patients: general wards are inappropriate. J Accid Emerg Med. 1996;13(3):163-165.
18. Hanlon P, Beck S, Robertson G, et al. Coping with the inexorable rise in medical admissions: evaluating a radical reorganisation of acute medical care in a Scottish district general hospital. Health Bull (Edinb). 1997;55(3):176-184.
19. Pradhan S, Ratnasingham D, Subbe CP, Stevenson S, Ward D, Cooksley T. Society for Acute Medicine Benchmarking Audit 2015: The Patient Perspective. Acute Med. 2015;14(4):147-150.
20. St Noble VJ, Davies G, Bell D. Improving continuity of care in an acute medical unit: initial outcomes. QJM. 2008;101(7):529-533.
21. Sullivan P, Harris ML, Bell D. The quality of patient experience of short-stay acute medical admissions: findings of the Adult Inpatient Survey in England. Clin Med (Lond). 2013;13(6):553-556.
22. Basey AJ, Kennedy TD, Mackridge AJ, Krska J. Delays and interruptions in the acute medical unit clerking process: an observational study. JRSM Open. 2016;7(2):2054270415619323.
44
23. Brand CA, Kennedy MP, King-Kallimanis BL, Williams G, Bain CA, Russell DM. Evaluation of the impact of implementation of a Medical Assessment and Planning Unit on length of stay. Aust Health Rev. 2010;34(3):334-339.
24. Elder E, Johnston AN, Crilly J. Improving emergency department throughput: An outcomes evaluation of two additional models of care. Int Emerg Nurs. 2016;25:19-26.
25. Ong BS, Van Nguyen H, Ilyas M, Boyatzis I, Ngian VJ. Medical Assessment Units and the older patient: a retrospective case-control study. Aust Health Rev. 2012;36(3):331-335.
26. Slatyer S, Toye C, Popescu A, et al. Early re-presentation to hospital after discharge from an acute medical unit: perspectives of older patients, their family caregivers and health professionals. J Clin Nurs. 2013;22(3-4):445-455.
27. Suthers B, Pickles R, Boyle M, Nair K, Cook J, Attia J. The effect of context on performance of an acute medical unit: experience from an Australian tertiary hospital. Aust Health Rev. 2012;36(3):320-324.
28. Williams AG, Jelinek GA, Rogers IR, Wenban JA, Jacobs IG. The effect on hospital admission profiles of establishing an emergency department observation ward. Med J Aust. 2000;173(8):411-414.
29. Yong TY, Li JY, Roberts S, Hakendorf P, Ben-Tovim DI, Thompson CH. The selection of acute medical admissions for a short-stay unit. Intern Emerg Med. 2011;6(4):321-327.
30. Coary R, Byrne D, O’Riordan D, Conway R, Cournane S, Silke B. Does admission via an acute medical unit influence hospital mortality? 12 years’ experience in a large Dublin hospital. Acute Med. 2014;13(4):152-158.
31. Conway R, O’Riordan D, Silke B. Long-term outcome of an AMAU--a decade’s experience. QJM. 2014;107(1):43-49.
32. Moloney ED, Bennett K, O’Riordan D, Silke B. Emergency department census of patients awaiting admission following reorganisation of an admissions process. Emerg Med J. 2006;23(5):363-367.
33. Moloney ED, Bennett K, Silke B. Effect of an acute medical admission unit on key quality indicators assessed by funnel plots. Postgrad Med J. 2007;83(984):659-663.
34. Moloney ED, Smith D, Bennett K, O’Riordan D, Silke B. Impact of an acute medical admission unit on length of hospital stay, and emergency department ‘wait times’. QJM. 2005;98(4):283-289.
35. Relihan E, Glynn S, Daly D, Silke B, Ryder S. Measuring and benchmarking safety culture: application of the safety attitudes questionnaire to an acute medical admissions unit. Ir J Med Sci. 2009;178(4):433-439.
36. Rooney T, Moloney ED, Bennett K, O’Riordan D, Silke B. Impact of an acute medical admission unit on hospital mortality: a 5-year prospective study. QJM. 2008;101(6):457-465.
37. McNeill GB, Brand C, Clark K, et al. Optimizing care for acute medical patients: the Australasian Medical Assessment Unit Survey. Intern Med J. 2011;41(1a):19-26.
38. Abenhaim HA, Kahn SR, Raffoul J, Becker MR. Program description: a hospitalist-run, medical short-stay unit in a teaching hospital. CMAJ. 2000;163(11):1477-1480.
39. Vork JC, Brabrand M, Folkestad L, Thomsen KK, Knudsen T, Christiansen C. A medical admission unit reduces duration of hospital stay and number of readmissions. Dan Med Bull. 2011;58(8):A4298.
40. Lo SM, Choi KT, Wong EM, et al. Effectiveness of Emergency Medicine Wards in reducing length of stay and overcrowding in emergency departments. Int Emerg Nurs. 2014;22(2):116-120.
41. Galen LS, der Schors W, Damen NL, Kramer M, Wagner C, Nanayakkara P. Measurement of generic patient reported outcome measures (PROMs) in an acute admission unit: A feasibility study. Acute Med. 2016;15(1):13-19.
42. Bell D, Skene H, Jones M, Vaughan L. A guide to the acute medical unit. Br J Hosp Med (Lond). 2008;69(7):M107-109.
43. Veiligheidsmanagementsysteem. Communicatie tussen hulpverleners volgens het SBAR-proces. VMS Veiligheidsprogramma. 02-2009.
45
44. Mitchell L, Bury E, Leonard O. Meeting the learning needs of AMU nurses through collaborative working. Br J Nurs. 2016;25(6):314-318.
45. The Society of Acute Medicine. Quality Standarts and Toolkits. http://www.acutemedicine.org.uk/resources/quality-standards/ Accessed August, 2016.
CHAPTER 3
MEASUREMENT OF GENERIC PATIENT REPORTED OUTCOME MEASURES (PROMS) IN AN ACUTE
ADMISSION UNIT: A FEASIBILITY STUDY
Louise S. van Galen | Wouter van der Schors | Nikkie L. Damen | Mark H. Kramer | C. Wagner | P.W.B. Nanayakkara
Acute Med 2016;15(1):13-9
‘There is far too little thought about what it means to be a patient’ Edith Schippers
48
ABSTRACTObjective
Measuring patient-reported outcome measures (PROMs) is a challenge in Acute Admission
Units (AAUs), where patients present with a variety of pathologies. Generic PROMs may be
used to measure the quality of care in this population. The main objective of this study was to
assess the feasibility of measuring generic PROMs in a Dutch AAU.
Design
Longitudinal cohort study
Setting
An AAU of a tertiary hospital in Amsterdam, the Netherlands
Participants
123 patients admitted to the AAU during 5 weeks in May and June 2015
Methods
Patients admitted to the AAU were asked to fill out a questionnaire relating to three time points:
7 days before, during, and within 2 weeks after admission. Additionally, patients were asked
to report on their experienced level of safety on the AAU and the contribution of the AAU to
their recovery.
Results
There were significant trends in generic PROMs for all three domains. Physical functioning
decreased during hospital admission and almost fully returned to the previous level after
discharge. Satisfaction with social role and anxiety significantly decreased over time.
Conclusions
Measuring generic PROMs in the AAU is feasible. The analysis of the PROMs took little effort
and results could be reported back to the healthcare workers on the AAU quickly. Patients
appreciated being asked about their own perceived health and the quality of care. Given that
this is the first study focusing on PROMs in AAU patients in the Netherlands, future studies with
larger sample sizes, and from other nations are needed to further investigate PROMs in this
patient group to establish International reference values.
Keywords
PROMs, Acute Admission Unit, patient experience, measurement of quality, patient safety
Keypoints
1. Measuring and analysis of generic PROMs in a Dutch AAU is feasible.
2. Results could be reported back to the healthcare workers on the AAU quickly.
3. Patients appreciated being asked about their own perceived health and the quality
of care.
4. There were significant trends in generic PROMs for all domains.
49
INTRODUCTIONAlthough widely recognised in the UK, Acute Admission Units (AAU) are a relatively new
concept in the Netherlands. The AAU in VUMC is a designated ward where a multidisciplinary
medical and surgical team manages patients for up to 72 hours, following emergency
admission via the Emergency Department (ED) or outpatient clinic. The AAU is staffed and
equipped to receive medical and surgical inpatients needing treatment for acute illnesses,
along similar lines to models described in the UK.1 In the Netherlands, where about two
million patients rely on acute care at the emergency department every year, a large number
of (academic) hospitals have implemented an AAU.2 In contrast to other countries in Europe
such as the United Kingdom, where typically only medical patients are treated, the AAU in
the Netherlands also houses beds for a broader set of specialities, including surgery, trauma,
orthopaedics, and urology.
Because of this heterogeneity it can be a challenge to measure patient experiences and quality
of care in an AAU universally. In an age where growing attention is paid to the patients’
perspective on quality of care and treatment effectiveness, evaluating ‘patient-related
outcome measures’ (PROMs) has become increasingly important. To assess PROMs, standard
measurement tools have been developed to capture patients’ perceptions of their general health
and/or specific diseases and conditions.3,4 PROMs can provide information on how a patient
appreciates his or her own health status, as well as the experienced impact of a (surgical)
procedure or a prescribed treatment.5 PROMs are increasingly utilised alongside clinical quality
indicators to enhance shared decision-making between patients and clinicians.2,6,7
The National Institution of Health (NIH) in the United States has developed the Patient-
Reported Outcomes Measurement Information System. (PROMIS®)8-10 to enable validated and
uniform implementation PROMs in clinical practice. Disease-specific PROMs 11 are designed
to evaluate the impact of a specific condition on patients’ functioning.8,11-16 Generic PROMs,
enable various aspects of functioning, such as self-care, physical functioning, and anxiety to
be measured.3,10
Until now, the use of PROMs in the Netherlands has been largely restricted to orthopaedic
patients, and implementation in other departments is still in early stages.17,18 Several
International studies have been performed using and validating PROMs in a more heterogeneous
population.5,19-21 In the Netherlands, however, no study has investigated the feasibility of
the generic PROMs in a general Dutch hospital population (regardless of specialty or diagnosis
at admission).3
It seems logical that, given the heterogeneity of AAU patients, generic PROMs are used to
assess quality of care based on patients’ experiences.
50
The main objectives of this longitudinal cohort study were:
1. To assess the feasibility of measuring generic PROMs in a Dutch Acute
Admission Unit;
2. To investigate whether generic PROMs could be used to measure physical functioning,
anxiety and satisfaction with social role in a general Dutch AAU population.
Secondary aims were to evaluate aspects of other key elements of quality: the feeling of safety
and overall experiences in an AAU through the patient’s perspective.
METHODS Study design
Between May and June 2015, in a period of 5 weeks (day shifts Monday to Friday), patients
admitted to the AAU were invited to participate in this study by the researcher. 123 patients
aged 18 or over were recruited as soon as possible upon their admission and included in
the study. Following the morning handover one of the two professionally trained researchers
(WS or LG) recruited the patients. Patients were not asked to participate if they were unable
to give informed consent as judged by researcher, for example because of severity of illness,
impaired cognition, language barriers, or strict isolation measures. After obtaining informed
consent, participants were requested to complete the questionnaire. First, they were asked
to judge in retrospect (T0) the situation 7 days prior to receiving care on AAU and second,
the situation at the moment of completing the questionnaire (T1) (Table I). Depending on
the capability and preference of the patient, the coordinating researchers either asked
questions himself or let the patient fill out the questionnaire on his or her own. During and
after completing the questionnaire, the researcher documented the time taken to finish
the questionnaire and monitored any extra comments or observations made by the patient.
In this way, practical feasibility was observed. Additionally, participants were asked if they had
any remarks on the questionnaire or advice for improvement.
Ten days after the discharge from AAU the patient was called (T2) to obtain a phone interview
provided the patient granted permission to contact him through the phone during the index
interview. Patients were excluded from telephone follow-up if they were still in hospital at T2,
having been transferred to another clinical area on discharge from AAU. During this telephone
call, participants were once again asked to assess their health at that moment and the level of
satisfaction regarding the care received during their admission (T2). The local Ethics Committee
approved the study protocol.
Questionnaire development
The questionnaire comprised questions from three domains of the Dutch version of
the PROMIS Profile 29.0 (approved and translated by the Dutch-Flemish PROMs group in 2014),
supplemented by a few short questions on the evaluation of patient safety, general functioning
and overall experience on the ward (see table I). PROMIS 29.0 items are rated on a 5-point
Likert scale.22 The following domains of the PROMIS 29.0 were selected: anxiety, physical
51
functioning, and satisfaction with social role. These domains were selected on the basis of
their appropriateness and suitability within the patient population, after consulting researchers
specialised in working with and translating PROMs, epidemiologists, and supervising clinicians.8
Completion of the questionnaire took around 5-10 minutes per patient.
As is shown in figure 2, during the telephone follow-up 10 days after discharge from the AAU
(T2), the same 12 items from the PROMIS 29.0 were asked. In addition, 4 other questions
about the perception of safety, contribution of AAU towards recovery, a question about
general functioning and an item from the Net Promotor Score (NPS) were asked.23 The NPS is
based on the question ‘Would you recommend the hospital to friends and family?’ The score
is calculated by subtracting the percent of patients rating the hospital 0-6 from the percent
recommending the hospital with a score of 9 or 10. In total, the questionnaire at T2 contained
16 questions. Appendix I provides the complete questionnaire.
Data analyses
Categorical outcome measures were presented as frequencies and percentages, whereas
continuous variables were presented by mean and standard deviation and/or median and
interquartile range. Since the PROMs variables were not normally distributed, we chose a non-
parametric test over the T-test for paired samples. We used Wilcoxon signed-rank test to
compare the PROMs scores on T0, T1 and T2. To control for the family wise error rate of
the three comparisons (T1 vs T0, T2 vs T1 and T2 vs T1) separately within each domain we
used a Bonferroni correction and set the two-sided significance level at p=0,05/3=0,017.
Satisfaction with social role was compared between T0 and T2 using a Wilcoxon rank-sum test
using a two-sided significance level of 0,05 since there was no measurement of satisfaction
with social role on T1. All statistical analyses were performed with Statistical Package for Social
Sciences (SPSS) version 20.
Figure 1 | Patient characteristics
52
Figure 2 | Patient inclusion
53
RESULTSPatient characteristics
The total study population consisted of 79 men (61.7%) and 49 women (38.3%) with a mean
age of 58.8 years (SD 18.7, range 46-72). The mean length of stay in the AAU was 45 hours
and 48 minutes (SD 32 hours). A total of 99 patients entered the AAU through the ED (80.5%).
Other patient flow characteristics are shown in Figure I. Admitting speciality varied widely, thus
creating a cross-section of an overall hospital patient population. Internal medicine (n=19,
15.5%), gastroenterology (n=17, 13.8%), and traumatology (n=15, 12.2%) were the largest
specialty groups, followed by pulmonary medicine (n=13, 10.6%), orthopaedics (n=11, 8.9%),
nephrology (11, 8.9%), urology (n=8, 6.5%), general surgery (n=7, 5.7%), haematology (n=7,
5.7%), neurology (n=6, 4.9%), and vascular surgery (n=2, 1.6%). A total of 33 patients had
been admitted into a hospital within the 30 days prior to the AAU admission, indicating
a readmission rate of 26.8% in this population. For 6 patients this AAU admission was their
first time in hospital (4.9%), whereas for 87 patients this admission was the first time on an
AAU (72.5%).
Feasibility
Figure II shows patient inclusion. A total of 123 patients completed the questionnaire at T1.
During the study period, a total of 195 patients were admitted to the AAU. Seventy-two
patients were not approached for inclusion on T1 since they were not eligible. The main
reasons for not approaching for inclusion were severity of illness (n=29, 41.4%), logistic
Table 1 | Content questionnaire at T0, T1, T2
54
11,0 (5,7)
5,4 (2,6)
12,4 (6,4)
6,7 (3,3)
8,5 (4,8)
11,8 (6,1)
7,1 (3,6)
12,9 (6,1)*
0 2 4 6 8 10 12 14
Satisfaction with Social Role
Anxiety
Physical Function
T0 (7 days prior to admission) T1 (on AMU) T2 (after discharge)
reasons (being away for surgery etc.) (n=23, 32.9%) and language barriers (n=9, 12.9%). At
T1, 125 patients were approached for inclusion by coordinating researchers. The vast majority
of patients (n=123, 98.4%) participated, only 2 (1.6%) refused to participate. More than half
of the included patients (N=65, 52.8%) also completed the telephone questionnaire within 10
days after discharge from the Acute Admission Unit.
A few patients (5.7%, n=7) did not complete the total questionnaire at T1. Finishing
the questionnaire took 6 (SD 2.9) minutes on average. However, patients who took
the opportunity to share their worries and feelings with the coordinating researcher needed
longer time to finish their questionnaire which explains the wide range in time. A group of
105 patients understood all the questions asked (85.4%), whereas 14 participants (11.4%)
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
To what extent has the AMU contributed to your recovery?(T2, N=64)
Did you feel safe during your stay on the AMU? (T2, N=65)
Do you feel safe on the AMU? (T1, N=123)
Not at all A little bit Somewhat Quite a bit Very much
Figure 3 | Physical functioning, anxiety and satisfaction with social role*Mean score (standard deviation)
Figure 4 | Feelings of safety and contribution to recovery on the AAU
55
needed additional explanation. For various reasons, the majority of patients (n=106, 86.2%)
chose to obtain help in completing the questionnaire, resulting in the researcher reading
out the questions. However, the researcher judged that about half of them were physically
able to fill out the questionnaire themselves (n=51, 48.1%). During telephone conversations
patients firstly expressed their appreciation for someone calling them after discharge and
they valued the fact that their unanswered questions after hospital admission were answered.
In general, all patients understood the follow-up questions appropriately and were able to
answer the questions.
Eighteen patients (14.6%) refused telephone follow-up. Thirty-three (31.4%) other patients
did give permission for telephone follow-up but were not included at T2. This is because at T2
a third of the patients included at T1 were not immediately discharged home (n=33, 31.4%),
because they either received additional care on other wards or health institutions, and therefore
not eligible to be included (figure II). At T2, three patients could not be included because of
they were not reachable at given numbers (7.5%). In addition, 4 patients in this group died
during hospital admission (10%). None of the participants had remarks or suggestions for
improvement of the questionnaire at either T1 or T2.
Patient reported outcome measures
At three time points (T0, T1, and T2) patients were asked to report on the PROMs domains
physical functioning, anxiety and satisfaction with social role. The mean scores and standard
deviations on these scales are given in Figure III. There was a significant decrease in physical
functioning between T0 and T1 (p<0.001), followed by a significant increase between T1 and
T2 (p<0,001). Patients almost fully recovered to their pre-admission level, however physical
functioning was still significantly lower at T2 when compared to T0 (p = 0,008). Satisfaction
with social role also significantly decreased between T0 and T2 (p=0,01). Experienced anxiety
also declined between T1 and T2 (p=0.007) and between T0 and T2, (p=0.000).
At T2, 96.9% (n=63) of the patients felt they were in a safe environment on the AAU (‘quite
a bit’ to very much’) – figure III. In addition 60.0% (n=39) of the patients indicated that their
stay at the ward contributed to their recovery (‘quite a bit’ to ‘ very much’). The average Net
Promotor Score at T2 (a 0-10 score) was 39,7%. This score implies that around 40.0% of
patients at T2 would recommend the hospital with a 9 or 10. This is in line with the overall
hospital Net Promotor Score of 40.0 in this hospital.
DISCUSSIONSummary and interpretations
The results of this study suggest that measuring generic PROMs on the AAU is feasible. Overall,
patients were positive about the questionnaire, resulting in a high response rate. The questions
asked were easily interpretable and applicable to the diversity of patients on the AAU. Because
the questionnaire could be filled out on paper or with assistance of coordinating researcher
56
it is applicable to the diverse population on the AAU. Since filling out the questionnaire was
not time consuming (approximately 6 minutes), this also can be seen as an indication of its
feasibility. For telephone follow-up (approximately 5 minutes) most patients also showed their
readiness to be approached again. Patients were positive about the questionnaire since it gave
them the feeling the hospital was really trying to take their perspectives on perceived care into
account. Also it gave them the opportunity to share their experiences and were told where
to look for answers to certain questions. The fact that PROMs took little effort and results
could be reported back to the healthcare workers on the AAU quickly was also an important
finding. This study also shows that generic PROMs were suitable for all patients on the AAU;
no adaptations were necessary for specific patient groups. Results indicated significant trends
in PROMs for all three domains. Physical functioning decreased during stay and almost fully
returned to old level after discharge, and anxiety decreased significantly over time. These
findings suggest that generic PROMs are able to illustrate changes in patient perspectives on
own perceived wellbeing before, during and after hospital admission.
This study also aimed to evaluate other aspects of key elements of quality: feelings of safety
and overall patient experiences on the AAU through patients’ perspective. In general, most
patients showed a positive response regarding their feeling of safety during their stay on
the AAU and regarding the contribution of the AAU to their recovery.
Relation to other evidence
To our knowledge, this was the first study performed in the Netherlands to investigate generic
PROMs in an Acute Admission Unit. Black (2013) noted that emergency admissions present
a challenge in assessing patients’ experiences of care since PROMs are often only available
after the acute event.3 The current study addresses this issue, by assessing PROMs on an
AAU. In comparison to most PROMs research, where focus is mostly on effectiveness of
care using PROMIS®, this is the first study that adds safety and experience. Grosse (2012)
reported that poor experience of safety has an adverse impact on the effectiveness of care
and therefore these key dimensions should not be ignored.24 Previous work has demonstrated
that generic PROMs can highlight physical or psychological problems which might otherwise
be overlooked.5 Generic PROMs could therefore function as a starting point to discuss more
appropriate care for the individual patient, thus enhancing shared-decision making.
A problem arising whilst assessing the PROMS in the Netherlands is that, in contrast to other
countries, no Dutch reference values are available to enable comparison.19 International data,
such as the Net Promotor Score could be obtained25, although comparisons across different
Nations can be problematic.26 The UK ‘Friends and Family test’, for example, which assesses
the likelihood of the patient recommending an NHS hospital, uses a different scoring system.23,25
Limitations
As already stated by McKenna (2011) in a well-developed measure, patients will only be asked
questions that are relevant. In a heterogeneous group of patients, addressing these relevant
57
areas can be a challenge because these issues may minimize respondent acceptability and
increase missing data. In this study we decided to take three specific domains since it was
difficult to capture other areas of concern applicable to all patient populations. This may have
led to our missing data on areas that could be of concern to a specific patient on the AAU.27
Another limitation of this study is the direct approach of the coordinating researcher who in
many cases assisted with the completion of the questionnaire. In practice there would not
be a researcher available on every AAU. Since listening to the questionnaire being read out
might lead to different scores than personal viewing of the questionnaires (one is probably
more inclined to give extreme answers once one hears 5 answers rather than seeing them on
paper) this might lead to less nuance/other results. In addition, we excluded the patients who
were unable to give informed consent, for example because of severity of illness. This may
have influenced the case-mix of the group. Finally, a general caveat of this study is the loss
in follow-up at T2. Since a substantial group of the patients were still in hospital after their
initial admission on the AAU, it can be challenging to create an appropriate time frame to
ask questions for T2. Future studies should investigate measuring T2 at a later time point to
achieve more complete follow-up information.
CONCLUSION AND IMPLICATION FOR PRACTICE The current study is the first study performed in the Netherlands to assess PROMs on the AAU.
The most important finding emerging from this study is that the measurement of generic
PROMS in the Acute Admission Unit is feasible. This study has shown that, in a heterogeneous
patient population PROMs may be used to measure the wellbeing and feeling of safety through
the patient’s perspective. Patients appreciated the fact that their view on these matters was
taken into account contributing to the high percentage of overall patient participation.
The analyses of the PROMs took little effort and results could be reported back to the healthcare
workers on the AAU quickly.
Future studies with larger sample sizes are warranted to establish reference values and ideal
follow-up time to compare PROMs throughout the Netherlands.
58
REFERENCES1. Scott I, Vaughan L, Bell D. Effectiveness of acute medical units in hospitals: a systematic review. Int J
Qual Health Care. 2009;21(6):397-407.
2. Dutch national Institute for Public Health and Environment. Acute Care. Accessed August 2015.
3. Black N. Patient reported outcome measures could help transform healthcare. BMJ. 2013;346:f167.
4. Marshall S, Haywood K, Fitzpatrick R. Impact of patient-reported outcome measures on routine practice: a structured review. J Eval Clin Pract. 2006;12(5):559-568.
5. Valderas JM, Kotzeva A, Espallargues M, et al. The impact of measuring patient-reported outcomes in clinical practice: a systematic review of the literature. Qual Life Res. 2008;17(2):179-193.
6. Stiggelbout AM, Van der Weijden T, De Wit MP, et al. Shared decision making: really putting patients at the centre of healthcare. BMJ. 2012;344:e256.
7. Lamberts MP, Drenth JP, van Laarhoven CJ, Westert GP. [Outcome of treatment reported by patients: instrument to reduce variations in clinical practice]. Ned Tijdschr Geneeskd. 2013;157(7):A5369.
8. Terwee CB, Roorda LD, de Vet HC, et al. Dutch-Flemish translation of 17 item banks from the patient-reported outcomes measurement information system (PROMIS). Qual Life Res. 2014;23(6):1733-1741.
9. Cella D, Yount S, Rothrock N, et al. The Patient-Reported Outcomes Measurement Information System (PROMIS): progress of an NIH Roadmap cooperative group during its first two years. Med Care. 2007;45(5 Suppl 1):S3-S11.
10. PROMIS ADULT PROFILE INSTRUMENTS, A brief guide to the PROMIS Profile instruments for adult respondents 2014: http://www.assessmentcenter.net/documents/PROMIS%20Profile%20Scoring%20Manual.pdf Accessed August 2015.
11. Spiegel BM, Hays RD, Bolus R, et al. Development of the NIH Patient-Reported Outcomes Measurement Information System (PROMIS) gastrointestinal symptom scales. Am J Gastroenterol. 2014;109(11):1804-1814.
12. Zorg Instituut Nederland https://www.zorginstituutnederland.nl/kwaliteit/toetsingskader+en+register/proms Accessed July, 2015.
13. Orbai AM, Bingham CO, 3rd. Patient reported outcomes in rheumatoid arthritis clinical trials. Curr Rheumatol Rep. 2015;17(4):28.
14. Rolfson O, Karrholm J, Dahlberg LE, Garellick G. Patient-reported outcomes in the Swedish Hip Arthroplasty Register: results of a nationwide prospective observational study. J Bone Joint Surg Br. 2011;93(7):867-875.
15. Flynn KE, Dew MA, Lin L, et al. Reliability and construct validity of PROMIS measures for patients with heart failure who undergo heart transplant. Qual Life Res. 2015.
16. Irwin DE, Gross HE, Stucky BD, et al. Development of six PROMIS pediatrics proxy-report item banks. Health Qual Life Outcomes. 2012;10:22.
17. Working on quality with PROMS: http://www.netqhealthcare.nl/patient-reported-outcome-measures/ Accessed August, 2015.
18. Zwaap J, Derksen J, Enzing J. Analysis for consultation care in knee and hip arthritis. Dutch Care Institute ( In Dutch). 2014.
19. Jensen RE, Potosky AL, Reeve BB, et al. Validation of the PROMIS physical function measures in a diverse US population-based cohort of cancer patients. Qual Life Res. 2015.
20. Sullivan P, Harris ML, Bell D. The quality of patient experience of short-stay acute medical admissions: findings of the Adult Inpatient Survey in England. Clin Med. 2013;13(6):553-556.
21. Cleary PD, Edgman-Levitan S, Roberts M, et al. Patients evaluate their hospital care: a national survey. Health Aff (Millwood). 1991;10(4):254-267.
22. Elaine IE, Seaman CA. Likert scales and data analyses. Quality Progress 40.7 2007: 64-65.
59
23. The Net Promoter Score and System: http://www.netpromoter.com/why-net-promoter/know. Accessed July, 2015.
24. Grosse Frie K, van der Meulen J, Black N. Relationship between patients’ reports of complications and symptoms, disability and quality of life after surgery. Br J Surg. 2012;99(8):1156-1163.
25. Friends and Family Test data http://www.england.nhs.uk/ourwork/pe/fft/friends-and-family-test-data/ [Consulted 20-10-2015].
26. Hamilton DF, Lane JV, Gaston P, et al. Assessing treatment outcomes using a single question: the net promoter score. Bone Joint J. 2014;96-B(5):622-628.
27. McKenna SP. Measuring patient-reported outcomes: moving beyond misplaced common sense to hard science. BMC Med. 2011;9:86.
PREVENTION OF SERIOUS ADVERSE EVENTS ON THE CLINICAL WARDS
CHAPTER 4
DELAYED RECOGNITION OF DETERIORATION OF PATIENTS IN GENERAL WARDS IS MOSTLY CAUSED BY
HUMAN RELATED MONITORING FAILURES: A ROOT CAUSE ANALYSIS OF UNPLANNED ICU ADMISSIONS
“UNPLANNED ICU ADMISSIONS: BEWARE OF THE MONITORING FAILURES”
Louise S. van Galen & Patricia W. Struik | Babiche E. Driesen | Hanneke Merten | Jeroen Ludikhuize | Johannes I. van der Spoel | Mark H. Kramer | Prabath W. Nanayakkara
PLoS One 2016;11(8):e0161393
‘Never let a good crisis go to waste’ Winston Churchill
62
ABSTRACTBackground
An unplanned ICU admission of an inpatient is a serious adverse event (SAE). So far, no in
depth-study has been performed to systematically analyse the root causes of unplanned
ICU-admissions. The primary aim of this study was to identify the healthcare worker-,
organisational-, technical,- disease- and patient- related causes that contribute to acute
unplanned ICU admissions from general wards using a Root-Cause Analysis Tool called
PRISMA-medical. Although a Track and Trigger System (MEWS) was introduced in our hospital
a few years ago, it was implemented without a clear protocol. Therefore, the secondary aim
was to assess the adherence to a Track and Trigger system to identify deterioration on general
hospital wards in patients eventually transferred to the ICU.
Methods
Retrospective observational study in 49 consecutive adult patients acutely admitted to
the Intensive Care Unit from a general nursing ward. 1. PRISMA-analysis on root causes
of unplanned ICU admissions 2. Assessment of protocol adherence to the early warning
score system.
Results
Out of 49 cases, 156 root causes were identified. The most frequent root causes were
healthcare worker related (46%), which were mainly failures in monitoring the patient. They
were followed by disease-related (45%), patient-related causes (5%), and organisational root
causes (3%). In only 40% of the patients vital parameters were monitored as was instructed
by the doctor. 477 vital parameter sets were found in the 48 hours before ICU admission, in
only 1% a correct MEWS was explicitly documented in the record.
Conclusions
This in-depth analysis demonstrates that almost half of the unplanned ICU admissions from
the general ward had healthcare worker related root causes, mostly due to monitoring failures
in clinically deteriorating patients. In order to reduce unplanned ICU admissions, improving
the monitoring of patients is therefore warranted.
Keywords
Root cause analysis, Unplanned ICU admission, Early warning score, Patient safety,
Healthcare quality
63
INTRODUCTIONDespite steady improvement in hospital care and programs to reduce harm in hospitalised
patients, serious adverse events (SAEs) are still common.1-4 In one out of ten patients,
the occurrence of an SAE during their admission contributes to permanent disability or to
death.5 Previous studies have documented that some adverse events are probably preventable
and that they are often related to errors in management 4. An unplanned admission to
the Intensive Care Unit (ICU) of an inpatient from a general ward is considered as a SAE.
According to the NICE (National Intensive Care Evaluation) criteria, an unplanned ICU admission
is defined as an admission that could not have been deferred without risk for at least 12 hours.6
It is known that unexpected ICU admission leads to a poor long term survival, especially in
older patients.7
Hence, it is important to detect deteriorating patients timely and treat them early in order to
prevent an eventual ICU admission or to transfer them to the ICU on time to improve clinical
outcomes. One of the ways to achieve this is by implementing a Rapid Response System
(RRS).8,9 These systems, composed of an afferent and efferent limb, are specifically designed
to enable early recognition and management of deteriorating patients on general wards.
The efferent limb consists of trained ICU personnel forming a Rapid Intervention Team (RIT)
who deliver immediate treatment to deteriorating patients at the bedside after being called in
by ward clinical staff. The clinical staff can detect patients deterioration early by using a Track
and Trigger Systems (TTS) such as Modified Early Warning Score (MEWS).10,11 This afferent limb
plays a key-role in limiting unplanned ICU admission, since early detection of deteriorating
patients could potentially prevent this.
However, the effectiveness of these TTSs in preventing unplanned ICU admissions remains
unclear.8,12 In addition, (root) causes leading to delayed detection of these deteriorating
patients in the wards are mostly unknown. The circumstances leading to delayed detection of
deteriorating patients in the wards are probably multi-causal. Current literature has identified
certain patient characteristics associated with unplanned admissions into the ICU, such as
age, and having a surgical procedure.13 Also, certain iatrogenic events such as disease induced
by a drug prescribed or environmental events (falls) have shown to cause ICU admissions.14,15
These studies, however, tell comparatively little about the healthcare worker related and
organisational/system related root causes, which, if known, can be used to improve early
detection of deteriorating patients and potentially avoid unplanned ICU-admissions.
A useful tool analysing these types of root causes is the PRISMA-tool (Prevention and Recovery
Information System for Monitoring and Analysis). The main goal of the PRISMA method is
to build a database of incidents and process deviations by creating causal trees, from which
conclusions may be drawn to suggest optimal countermeasures. This in-depth analysis method
has been accepted by the World Alliance for Patient Safety of the World Health Organisation
and has shown to provide effective starting points for improvement in quality of care.16-19
64
No study has yet been performed to systematically analyse and identify the root causes
leading to late detection and treatment of deteriorating patients in wards. To formulate
possible improvement and prevention strategies with the aim of reducing the number of ICU
admissions insight into these causes is essential. Also, an optimally implemented RRS could
contribute to increased patient safety by recognising those in need of extra care without any
unnecessary delay.
Therefore, the main aims of this retrospective record review study were to (1.) analyse
the healthcare worker-, organisational-, technical-, disease- and patient- related causes that
contribute to late detection and treatment of deteriorating patients in the general wards
using PRISMA-medical analysis, and (2.) assess the adherence to and effectiveness of an
already implemented TTS on general wards in the early recognition of deteriorating patients
transferred to the ICU.
MATERIAL AND METHODSStudy design
This retrospective, record review study included unplanned ICU admissions from general
wards in the VU University Medical Centre in Amsterdam (VUmc), The Netherlands. This is
an academic 733-bedded medical centre with approximately 50,000 admissions per year.
The Adult Intensive Care Unit consists of 22 beds.
We decided to include the first 50 consecutive patient records of 2015 meeting the inclusion
criteria to explore the causes of unplanned ICU-admissions, since previous studies have shown
that around 50 PRISMA-analyses are credible and provide a well-founded causal-profile.16,19
The following criteria were used for inclusion: all patients on general wards aged 18 years and
older who were admitted to the ICU unplanned according to the NICE criteria (“an admission
that could not have been deferred without risk for at least 12 hours”).6 Only the first ICU
admission of the patient was included.
Excluded from the study sample were: patients admitted on the ICU immediately from
the emergency department, the operation room, medium care, high care, or coronary care;
patients transferred from other hospitals; patients with a planned ICU admission (i.e. after
surgery). The Ethics committee of VU University Medical Centre, Amsterdam, approved
the study and necessity for informed consent was waived. Information from the clinical records
was anonymized and de-identified prior to analysis.
The RRS used in the VUmc is the MEWS, validated by Subbe et al. in 2007, which was
implemented in the VUmc in 2014.11 The MEWS used at the VUmc consists of an easy-to-use
algorithm of seven parameters: respiratory rate, saturation rate, heart frequency, systolic blood
pressure, temperature, consciousness, urine production (Fig. 1). Also, 1 point can be added if
a nurse is worried about the patient. If a vital parameter was not documented in the system,
65
this parameter was considered to be normal, therefore 0 points were given. When a patient
on a general ward has a score of 3 or higher, the nurse is expected to call the ward doctor
on duty or can immediately call the RIT team (Rapid Intervention Team). However, the nurses
were not required to perform MEWS on daily basis on set time points and the MEWS was
mostly performed on indication which probably led to late detection of the deteriorating
patient. It was decided to reintroduce the MEWS protocol in 2015, this study was performed
before reimplementation.
Assessment
For analysis doctor’s charts, nurse’s charts and electronic patient files including all test results
were available. Using a standardised abstraction form for each individual ICU-admission,
patient characteristics such as the age, mortality, APACHE scores and circumstances under
which ICU-admission took place (such as length of stay in the wards and admission speciality)
were collected (S1 File).20-22 Vital parameters measured and the use of the MEWS-protocol in
the 48 hours before the acutely unplanned ICU admission were also systematically registered
into this form.
Two medically and PRISMA-trained investigators (BD, PS) reviewed each case extensively
and filled out these chart abstraction forms. This analysis was a labour intensive process
(approximately 90 minutes average per investigator per chart). Subsequently they composed
individual causal trees; thereafter consensus was reached in a structured meeting with a third
independent experienced medical and PRISMA reviewer (LG), from which a composite root
causal tree was constructed. Finally, all cases and their causal trees were discussed with two
senior physicians (PN, JL) and a psychologist with a special interest in PRISMA-analysis (HM),
which resulted in the final root causal trees.
PRISMA-analysis
Fig. 2 shows three examples of causal trees from this study. In general the PRISMA-method
can be used to examine latent factors (technical and organisational), active failures (human)
and other factors (patient-related and other). A PRISMA-analysis starts with an incident, in this
study the unplanned ICU-admission was seen as the incident. This incident was then placed
at the top of the causal tree and was the starting point for the analysis. The next step was to
identify the direct causes underlying the unplanned ICU-admission. Direct causes are revealed
by asking ‘why’ this incident has occurred. Information from the records was used to identify
these direct causes. Below every direct cause the indirect causes were inserted. By constantly
asking ‘why’ an event had taken place, relevant indirect causes were systematically exposed.
If no more objective information was acknowledged as a cause, the last noted indirect cause
was labelled as a root cause and was placed at the bottom of the causal tree. The analysis
also stopped when the underlying cause lied outside of the hospital. The root causes were
then classified using the Eindhoven Classification Model (ECM), which is the corresponding
taxonomy of PRISMA-medical to classify root causes. The ECM is based on the skill-rules-
knowledge-based behavioural model of Rasmussen and on the system approach to human
66
error of Reason.23-25 The ECM was also used as a foundational component in the conceptual
framework for the International Classification for Patient Safety.26,27 Table 1 shows
the classification of the root causes according to the Eindhoven Classification Model.28 Since
previous studies have shown that progression of disease is a valuable addition to this model
applied in medical settings, this was added to the root causes for the current study.16
Statistical analysis
Using IBM SPSS Statistics, Chicago, USA, Version 22.0 descriptive characteristics and
frequencies were calculated. Categorical outcome measures are presented as frequencies and
percentages. For continuous variables we chose to use median and interquartile ranges since
none of them were normally distributed.
RESULTSPatient characteristics
PRISMA-analysis was performed on 49 available cases, since one chart was unavailable for
review during the study period. Table 2 displays baseline patient characteristics. The median
age was 69 years (range 34-90), and both sexes were represented almost equally (47% vs
53%). Nineteen patients died during their hospital admission, resulting in 39% in-hospital
mortality. The three scores for ICU-mortality (SAPS II, APACHE II, APACHE IV) had a median of
51, 24, and 95 resulting in a predicted mortality of 50%, 55%, 36% respectively. The median
time between hospital admission and ICU admission was 88 hours and 34 minutes (range
1h38 min – 733h). Most patients were transferred to the ICU between midnight and six
o’clock in the morning. Sixty-nine percent of the patients (n=34) had a ‘Do not resuscitate’-
policy (DNR), for three patients the DNR-policy was not clear (6%).
Figure 1 | MEWS protocol in VUmc
67
Figure 2 | Three examples of root causal trees**HKK: Human-related knowledge behaviour, HRI: Human-related intervention, PRF: Patient-related factor, HRM: Human-related monitoring, HRV: Human-related verification, DRF: Disease-related factor, HSS: Human-related skills-based.
68
Tab
le 1
| D
escr
iptio
n of
cat
egor
ies
of t
he E
indh
oven
Cla
ssifi
catio
n m
odel
: PRI
SMA
- m
edic
al V
ersi
on 17
,18
Mai
n c
ateg
ory
Sub
cate
go
ryC
od
eD
escr
ipti
on
Exam
ple
s (i
f av
aila
ble
) *
Tech
nica
lEx
tern
alT-
exTe
chni
cal f
ailu
res
beyo
nd t
he c
ontr
ol o
f th
e or
gani
zati
on.
Not
ava
ilabl
e
Des
ign
TDFa
ilure
s to
poo
r de
sign
of
equi
pmen
t et
c.N
ot a
vaila
ble
Con
stru
ctio
nTC
Cor
rect
des
ign
inap
prop
riat
ely
cons
truc
ted
or p
lace
d.N
ot a
vaila
ble
Mat
eria
lsTM
Mat
eria
l def
ects
not
cla
ssifi
ed u
nder
TD
or
TC.
Not
ava
ilabl
e
Org
aniz
atio
nal
Exte
rnal
O-e
xFa
ilure
s at
an
orga
niza
tion
al le
vel b
eyon
d th
e co
ntro
l and
re
spon
sibi
lity
of t
he in
vest
igat
ing
team
. N
ot a
vaila
ble
Tran
sfer
of
know
ledg
eO
KFa
ilure
res
ulti
ng f
rom
inad
equa
te m
easu
res
to t
rain
or
supe
rvis
e ne
w o
r in
expe
rien
ced
staf
f.
Not
ava
ilabl
e
Prot
ocol
sO
PFa
ilure
s re
lati
ng t
o th
e qu
alit
y or
ava
ilabi
lity
of a
ppro
pria
te
prot
ocol
s.N
ot f
ollo
win
g pa
in t
reat
men
t pr
otoc
ol a
fter
sur
gery
Man
agem
ent
prio
riti
esO
MIn
tern
al m
anag
emen
t de
cisi
ons
whi
ch r
educ
e fo
cus
on
pati
ent
safe
ty w
hen
face
d w
ith
confl
icti
ng p
rior
itie
s.N
o be
ds a
vaila
ble
at IC
U
Cul
ture
OC
Failu
re d
ue t
o at
titu
de a
nd a
ppro
ach
of t
he t
reat
ing
orga
niza
tion
.W
ard
whe
re v
ital
par
amet
ers
are
not
freq
uent
ly
take
n si
nce
‘no
one
does
it’
Hum
anEx
tern
alH
-ex
Hum
an f
ailu
res
beyo
nd t
he c
ontr
ol o
f th
e or
gani
zati
on/
depa
rtm
ent
Into
xica
tion
of
too
high
dos
age
med
icat
ion
pres
crib
ed o
utsi
de h
ospi
tal c
are
(by
GP)
Kno
wle
dge-
base
d be
havi
our
HK
KFa
ilure
of
an in
divi
dual
to
appl
y th
eir
know
ledg
e to
a n
ew
clin
ical
sit
uati
onN
o ad
equa
te d
iagn
osti
cs
No
phys
ical
exa
min
atio
n do
ne
Qua
lifica
tion
sH
RQA
n in
appr
opri
atel
y tr
aine
d in
divi
dual
per
form
ing
the
clin
ical
tas
kN
ot a
vaila
ble
Co-
ordi
nati
onH
RCA
lack
of
task
co-
ordi
nati
on w
ithi
n th
e he
alth
care
tea
mN
o co
ordi
nati
on o
f hy
pert
ensi
on t
reat
men
t
Veri
ficat
ion
HRV
Failu
re t
o co
rrec
tly
chec
k an
d as
sess
the
sit
uati
on b
efor
e pe
rfor
min
g in
terv
enti
ons
DN
R po
licy
not
adeq
uate
ly d
iscu
ssed
Inte
rven
tion
HRI
Failu
re r
esul
ting
fro
m f
ault
y ta
sk p
lann
ing
or p
erfo
rman
ceN
o di
agno
stic
s an
d ad
equa
te t
reat
men
t de
liriu
m
Mon
itor
ing
HRM
Failu
re t
o m
onit
or t
he p
atie
nt’s
pro
gres
s or
con
diti
onN
o ev
alua
tion
of
vita
ls a
fter
cha
ngin
g tr
eatm
ent
Vit
als
not
mon
itor
ed a
nd a
ctio
n un
dert
aken
aft
er
repo
rted
det
erio
rati
on
69
Tab
le 1
| (c
ontin
ued)
Mai
n c
ateg
ory
Sub
cate
go
ryC
od
eD
escr
ipti
on
Exam
ple
s (i
f av
aila
ble
) *
Skill
s-ba
sed
HSS
Failu
re in
per
form
ance
of
high
ly d
evel
oped
ski
llsO
bstr
ucti
ve le
sion
tra
chea
not
rec
ogni
zed/
mis
sed
by
radi
olog
ist
on C
T
Pati
ent
Pati
ent-
rela
ted
PRF
Failu
res
rela
ted
to p
atie
nt c
hara
cter
isti
cs o
r co
ndit
ions
, w
hich
are
bey
ond
the
cont
rol o
f st
aff
and
influ
ence
clin
ical
pr
ogre
ss
Mon
itor
ing
not
adeq
uate
bec
ause
pat
ient
ref
used
C
AD
Dis
ease
-rel
ated
DRF
Failu
res
rela
ted
to t
he n
atur
al p
rogr
ess
of d
isea
se w
hich
are
be
yond
con
trol
of
pati
ent,
its
care
rs a
nd s
taff
Tum
our
prog
ress
ion
in v
ena
cava
infe
rior
Bilia
ry p
ancr
eati
tis
XU
ncla
ssifi
able
XM
edic
atio
n w
as s
till
bein
g do
sed
prop
erly
Toxi
c re
acti
on c
hem
othe
rapy
*A t
able
wit
h ov
ervi
ew o
f al
l roo
t ca
uses
is p
rovi
ded
in S
1 Ta
ble.
70
Root causes
In total, from 49 unplanned ICU causal trees, 155 root causes were identified after PRISMA-
analysis. See Fig. 2 for three of the composed causal trees. Twelve unplanned ICU admissions
had one root cause (22%), 7 (14%) had two root causes, 11 (22%) three root causes, 12 four
root causes (25%) and 2 cases had 5 root causes (4%). Twelve percent of the cases had six or
more root causes (n=6). Almost half of the root causes identified were disease-related (n=70,
45%). Forty-four cases had at least one disease-related root cause. These disease-related root
causes comprise failures which are related to the natural progression of a disease which are
beyond control of patient, its carers and staff, for example neutropenic sepsis in a patient
with chemotherapy for an hematologic disease. Table 1. illustrates the subcategories used
in PRISMA-analysis with examples from our study. The distribution of all the root causes is
illustrated in Fig. 3a.
Figure 3 | Distribution root causes3a Main categories root causes. 3b Healthcare worker (HCW) root causes.
71
Almost half of the root causes classified were human-related (n=71, 46%), followed by
patient-related causes (n=7, 5%), and organisational root causes (n=5, 3%). Human related
causes are causes which are healthcare worker related. Most of the healthcare worker related
causes (34%) were monitoring failures (HRM), these are. failures in monitoring the patient’s
progress or condition, e.g. not monitoring vital parameters in a patient who is clinically
deteriorating. Fig. 3b shows the subcategorization of healthcare worker root causes. Human
intervention causes were accountable for almost one-third (27%) of the causes. They represent
failures resulting from faulty task planning or performance, for example the case where no
intervention was started after the nurse repeatedly mentioned patient’s vital signs worsening.
The remaining human-related root causes were verification (HRV: 11,3%), coordination (HRC:
7%), and skills-based (HSS: 4,2%) failures. Finally, the organisation root causes were those
caused by management issues, such as lack of beds on the ICU, or not handling certain
treatment according to the institutional protocol. The two unclassifiable root causes are shown
in table 1. No technical related root causes (T) were identified in this PRISMA-analysis.
Use of Track and Trigger System/Early Warning Score
Table 3 shows that in 86% (n=42 patients) of the cases orders were given by the doctors for
vital parameter monitoring. This was done as agreed in only half of the cases (n=20, 49%).
In the 48 hours before ICU admission, a total of 477 vital parameter sets were measured in
the 49 patients (i.e. 1 vital set could include: a blood pressure, pulse, breathing frequency
and temperature), with a median of 6 sets measured per patient. Of these 477 sets, an
explicit MEWS was calculated and documented correctly, according to protocol, in only 6
of these measurements in 4 patients (1%). After recalculation by the researcher based upon
the available vital parameters, 207 (43%) of the vital sets gave a critical MEWS score of 3 or
higher. 46 patients had at least one critical MEWS of 3 or higher during the 48 hours before
ICU admission. According to MEWS protocol, this implied that a doctor had to be called. In
125 of the measured sets, in 42 patients, this was done (although the MEWS score was often
not calculated and documented), and after this the doctor started an action upon this in 117
of these 125 phone calls. In total, 67 of all measurements, in 92% of the patients (n=45), led
to a call to the RIT at least once.
DISCUSSIONThe current study is the first to systematically investigate the root causes of unplanned ICU
admissions. The most important finding from this study was that almost half of the root causes
contributing to unplanned ICU admission were human (healthcare worker) related. These
causes predominantly included human monitoring and intervention failures, indicating flaws
in monitoring the patients progress or condition and faulty task planning or performance. This
illustrates a potential for improvement. The other half of the root causes were disease-related,
comprising the root causes related to the natural progression of the disease, which was to be
expected in this overall severely ill patient population, as reflected in their high mortality rates.
The secondary aim was to further investigate monitoring by analysing the use of a TTS on
a general ward in the early recognition of deteriorating patients. Results showed that in the 48
72
Table 2 | Patient characteristics
N (%) 49 (100%)
Age - median(range) 69 (34-90)
Male 23 (47%)
Deceased during admission 19 (39%)
Admission specialty :
Cardiology
Gastroenterology
Haematology
Internal medicine
ENT-diseases
Pulmonary medicine
Neurology
Nephrology
Oncology
Traumatology
Vascular surgery
1 (2%)
5 (10%)
7 (14%)
12 (24%)
1 (2%)
9 (18%)
3 (6%)
3 (6%)
3 (6%)
4 (8%)
1 (2%)
Polypharmacy* 36 (73)
Length of stay before unplanned transfer ICU in hours – median (range) 88 h 34 m (1h38m-733h)
Time unplanned ICU admission
24.00 – 06.00
06.00 – 12.00
12.00 – 18.00
18.00 – 24.00
17 (35%)
12 (24%)
11 (22%)
9 (18%)
DNR-policy before ICU admission
No restriction
Do not resuscitate, do ventilate
Do not resuscitate, do not ventilate
No ICU admission
Not clear
34 (69%)
9 (18%)
3 (6%)
0 (0%)
3 (6%)
SAPS II – median(range)21 51 (18-110)
APACHE II – median (range)20 24 (6-45)
APACHE IV – median (range)22 95 (36-186)
hours before deterioration only in 1% of the measured vital sets an explicit correct ‘MEWS’
was reported, although in 43% of the measurements patients had a critical score. This is an
important clinical finding, since it seems essential that recognition of critical scores by hospital
staff is present in order to improve monitoring.29
Monitoring failures emerging from the PRISMA-analysis were diverse. One example is a nurse
who documented in the chart that a patient repeatedly complains about shortness of breath,
did not undertake an action to measure vital parameters or request a physician for assessment.
73
Another example was unclear handovers resulted in confusion about how intensively the patient
needed monitoring, resulting in inadequate and insufficient monitoring measurements.
Interventional problems are exemplified by not performing adequate diagnostics and treatment
in a severely immunosuppressed patient with a suspected pneumonia. In our study, an unclear
DNR-policy was also found to be responsible for an unnecessary unplanned ICU admission
in a few cases, since these patients did not want to be admitted to the ICU but this was not
known or discussed on the ward. These are human-related verification failures (HRV).
The above results seem to be consistent with earlier research indicating that unplanned ICU
admissions are not only caused by the underlying disease, but are also potentially caused by
suboptimal care on the ward, and by inadequate assessment and monitoring of a patients
status.30,31 Other studies have shown that certain patient characteristics such as older age, being
male and having a higher co-morbidity increase the chance of unplanned ICU admissions.32 In
addition to above mentioned non-modifiable factors, our study results provides implications
for practice to potentially reduce unplanned ICU admission by improving patient monitoring.
One method for improving the recognition of these patients, is the implementation of TTSs.
Although the conclusions on effectiveness of TTSs in reducing clinical endpoints are still not
uniform, when properly followed they are effective in identifying deteriorating patients.33,34
The effectiveness depends on appropriate implementation, compliance and an effective clinical
Table 3 | Use of Track and Trigger System
Vital parameters documentation Frequency (%), N=49 (100%)
Orders were given for vital monitoring* 42 (86%)
Vital monitoring performed as agreed 20 (41%)
Registration of ICU admission in nurse’ charts 32 (65%)
Registration of ICU admission in doctors’ charts 38 (78%)
MEWS DocumentationTotal vital set measurements done in 48 hours before ICU admission in 49 patients
N = 477 (100%)
Number of vital set measurement done per patient in 48 hours before ICU admission – median(range)
6 (1-22)
Doctor called after vital parameters measured 174 (36%) 42 patients
Doctor started an action after being called 164 (34%) 42 patients
Evaluation after 60 minutes of started action 96 (20%) 41 patients
RIT-call 69 (14%) 45 patients
MEWS calculated and documented correctly in charts 6 (1%) 4 patients
Critical MEWS (after recalculation by researcher using vital set measurements ≥3)
207 (43%) 46 patients
Doctor called at a critical-MEWS according to the recalculated score by the researcher
125 (26%) 42 patients
74
response.35 The current study has shown that the MEWS score in the studied population was
registered according to protocol in very few patients. For the other patients, a score was
calculated by the researchers based on the available vital parameters. After recalculation 43%
(207 out of 477) of the measurements were critical, and in these cases in 60% (125 out of
207 measurements) a doctor was called in although an explicit MEWS was not documented
in the charts. If the doctor was notified, in most cases they started an action, which shows
the relevance of these critical measurements. These results show that the implementation
of MEWS at the time of the study was insufficient. When we implemented the protocol
a few years back the nurses and doctors were requested to perform MEWS when they
found patients to be sick and in need of critical care. The protocol did not require the MEWS
to be taken daily on set times to identify deterioration early. The added value of the daily
measurements has been demonstrated.35 We therefore implemented a new protocol in which
it was compulsory to measure the MEWS in all patients at least once a day in the morning.
If the MEWS is >3 the doctor has to be called within 30 minutes. The clinical staff was (re)
trained aiming to chance their mind-set about the importance of MEWS. A few weeks after
this implementation the protocol adherence improved to 89%, underlining the importance of
robust implementation.36
This in-depth study has provided analysis of the root causes of unplanned ICU admissions to
the full extent. To our knowledge, this is the first study performed that explores the underlying
causes of unplanned ICU admissions in such a way that it provides clear insight in the areas for
potential improvement in the healthcare system. Hence, it equips us with valuable information
about areas in need of attention. From this, tools for implementing process improvement in
hospitals can be initialised. A limitation of the study was its retrospective nature, which means
that researchers had to rely on written information in the patient records, instead of actively
interviewing the persons involved. Therefore valuable information might have been missed
and this could potentially have led to an underestimation of the factors leading to delayed
recognition. For future studies it is therefore recommended to combine this research method
with other methods, such as collecting qualitative data using interviews. Also, by assessing
these cases only on the basis of information available on paper, analysis might not elucidate all
the organisational factors which are not typically written down and are latent.
CONCLUSIONThis study is the first one that has systematically provided insight in the root causes of unplanned
ICU admissions. Half of the root causes identified were healthcare worker related failures,
mainly resulting from monitoring and interventional failures. The monitoring of patients might
be improved by a properly functioning Track and Trigger System. Results in this study however,
have shown that this warrants improvement.
75
CONFLICTS OF INTERESTS None
SUPPORTING INFORMATION CAPTIONSS1 File. Data collection sheet unplanned ICU admissions.S1 Table. Overview root causes.S1 Dataset. ICU admissions.
76
REFERENCES1. de Vries EN, Ramrattan MA, Smorenburg SM, Gouma DJ, Boermeester MA. The incidence and
nature of in-hospital adverse events: a systematic review. Qual Saf Health Care. 2008;17(3):216-223.
2. Wilson RM, Runciman WB, Gibberd RW, Harrison BT, Newby L, Hamilton JD. The Quality in Australian Health Care Study. Med J Aust. 1995;163(9):458-471.
3. Vlayen A, Verelst S, Bekkering GE, Schrooten W, Hellings J, Claes N. Incidence and preventability of adverse events requiring intensive care admission: a systematic review. J Eval Clin Pract. 2012;18(2):485-497.
4. Leape LL, Brennan TA, Laird N, et al. The nature of adverse events in hospitalized patients. Results of the Harvard Medical Practice Study II. N Engl J Med. 1991;324(6):377-384.
5. Zegers M, de Bruijne MC, Wagner C, et al. Adverse events and potentially preventable deaths in Dutch hospitals: results of a retrospective patient record review study. Qual Saf Health Care. 2009;18(4):297-302.
6. Arts D, de Keizer N, Scheffer GJ, de Jonge E. Quality of data collected for severity of illness scores in the Dutch National Intensive Care Evaluation (NICE) registry. Intensive Care Med. 2002;28(5):656-659.
7. Ridley S, Jackson R, Findlay J, Wallace P. Long term survival after intensive care. BMJ. 1990;301(6761):1127-1130.
8. Alam N, Vegting IL, Houben E, et al. Exploring the performance of the National Early Warning Score (NEWS) in a European emergency department. Resuscitation. 2015;90:111-115.
9. Hillman K, Parr M, Flabouris A, Bishop G, Stewart A. Redefining in-hospital resuscitation: the concept of the medical emergency team. Resuscitation. 2001;48(2):105-110.
10. Moon A, Cosgrove JF, Lea D, Fairs A, Cressey DM. An eight year audit before and after the introduction of modified early warning score (MEWS) charts, of patients admitted to a tertiary referral intensive care unit after CPR. Resuscitation. 2011;82(2):150-154.
11. Subbe CP, Gao H, Harrison DA. Reproducibility of physiological track-and-trigger warning systems for identifying at-risk patients on the ward. Intensive Care Med. 2007;33(4):619-624.
12. McGaughey J, Alderdice F, Fowler R, Kapila A, Mayhew A, Moutray M. Outreach and Early Warning Systems (EWS) for the prevention of intensive care admission and death of critically ill adult patients on general hospital wards. Cochrane Database Syst Rev. 2007(3):CD005529.
13. Quinn TD, Gabriel RA, Dutton RP, Urman RD. Analysis of Unplanned Postoperative Admissions to the Intensive Care Unit. J Intensive Care Med. 2015.
14. Mercier E, Giraudeau B, Ginies G, Perrotin D, Dequin PF. Iatrogenic events contributing to ICU admission: a prospective study. Intensive Care Med. 2010;36(6):1033-1037.
15. Darchy B, Le Miere E, Figueredo B, Bavoux E, Domart Y. Iatrogenic diseases as a reason for admission to the intensive care unit: incidence, causes, and consequences. Arch Intern Med. 1999;159(1):71-78.
16. Fluitman KS, van Galen LS, Merten H, et al. Exploring the preventable causes of unplanned readmissions using root cause analysis: Coordination of care is the weakest link. Eur J Intern Med. 2016.
17. van der Schaaf TW HM. PRISM-Medical. A Brief Description. Eindhoven University of Technology, Faculty of Technology Management, Patient Safety Systems: Eindhoven. 2005.
18. van Vuuren W SC, van der Schaaf TW. The Development of an Incident Analysis Tool For the Medical Field. Eindhoven University of Technology: Eindhoven. 1997.
19. van Wagtendonk I, Smits M, Merten H, Heetveld MJ, Wagner C. Nature, causes and consequences of unintended events in surgical units. Br J Surg. 2010;97(11):1730-1740.
20. Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a severity of disease classification system. Crit Care Med. 1985;13(10):818-829.
77
21. Le Gall JR, Lemeshow S, Saulnier F. A new Simplified Acute Physiology Score (SAPS II) based on a European/North American multicenter study. JAMA. 1993;270(24):2957-2963.
22. Zimmerman JE, Kramer AA, McNair DS, Malila FM. Acute Physiology and Chronic Health Evaluation (APACHE) IV: hospital mortality assessment for today’s critically ill patients. Crit Care Med. 2006;34(5):1297-1310.
23. Rasmussen J. Skills, rules and knowledge: signals, signs and symbols and other distinctions in human performance models. IEEE Trans Systems Man Cybernetics. 1983;13:257-266.
24. Reason JT. Human Error. Cambridge University Press. 1990.
25. Reason JT. Managing the Risk of Organisational Accidents. Asgate Aldershot. 1997.
26. Runciman W, Hibbert P, Thomson R, Van Der Schaaf T, Sherman H, Lewalle P. Towards an International Classification for Patient Safety: key concepts and terms. Int J Qual Health Care. 2009;21(1):18-26.
27. Sherman H, Castro G, Fletcher M, et al. Towards an International Classification for Patient Safety: the conceptual framework. Int J Qual Health Care. 2009;21(1):2-8.
28. Runciman WB, Williamson JA, Deakin A, Benveniste KA, Bannon K, Hibbert PD. An integrated framework for safety, quality and risk management: an information and incident management system based on a universal patient safety classification. Qual Saf Health Care. 2006;15 Suppl 1:i82-90.
29. Storm-Versloot MN, Verweij L, Lucas C, et al. Clinical relevance of routinely measured vital signs in hospitalized patients: a systematic review. J Nurs Scholarsh. 2014;46(1):39-49.
30. Bapoje SR, Gaudiani JL, Narayanan V, Albert RK. Unplanned transfers to a medical intensive care unit: causes and relationship to preventable errors in care. J Hosp Med. 2011;6(2):68-72.
31. McGloin H, Adam SK, Singer M. Unexpected deaths and referrals to intensive care of patients on general wards. Are some cases potentially avoidable? J R Coll Physicians Lond. 1999;33(3):255-259.
32. Frost SA, Alexandrou E, Bogdanovski T, Salamonson Y, Parr MJ, Hillman KM. Unplanned admission to intensive care after emergency hospitalisation: risk factors and development of a nomogram for individualising risk. Resuscitation. 2009;80(2):224-230.
33. Ranji SR, Auerbach AD, Hurd CJ, O’Rourke K, Shojania KG. Effects of rapid response systems on clinical outcomes: systematic review and meta-analysis. J Hosp Med. 2007;2(6):422-432.
34. Chan PS, Khalid A, Longmore LS, Berg RA, Kosiborod M, Spertus JA. Hospital-wide code rates and mortality before and after implementation of a rapid response team. JAMA. 2008;300(21):2506-2513.
35. Ludikhuize J, Borgert M, Binnekade J, Subbe C, Dongelmans D, Goossens A. Standardized measurement of the Modified Early Warning Score results in enhanced implementation of a Rapid Response System: a quasi-experimental study. Resuscitation. 2014;85(5):676-682.
36. van Galen LS, Dijkstra CC, Ludikhuize J, Kramer MH, Nanayakkara PW. A Protocolised Once a Day Modified Early Warning Score (MEWS) Measurement Is an Appropriate Screening Tool for Major Adverse Events in a General Hospital Population. PLoS One. 2016;11(8):e0160811.
CHAPTER 5
A PROTOCOLISED ONCE A DAY MODIFIED EARLY WARNING SCORE (MEWS) MEASUREMENT IS AN
APPROPRIATE SCREENING TOOL FOR MAJOR ADVERSE EVENTS IN A GENERAL HOSPITAL POPULATION
“A MEWS IN THE MORNING, A VERY GOOD WARNING’’
Louise S. van Galen & Casper C. Dijkstra | Jeroen Ludikhuize | Mark H.H. Kramer | Prabath W.B. Nanayakkara
PLoS One 2016;11(8):e0160811
‘Sometimes the protocol is a too tight harness which traps medical specialists’ Dana Ploeger
80
ABSTRACTBackground
The Modified Early Warning Score (MEWS) was developed to timely recognise clinically
deteriorating hospitalised patients. However, the ability of the MEWS in predicting serious
adverse events (SAEs) in a general hospital population has not been examined prospectively.
The aims were to (1) analyse protocol adherence to a MEWS protocol in a real-life setting
and (2) to determine the predictive value of protocolised daily MEWS measurement on SAEs:
death, cardiac arrests, ICU-admissions and readmissions.
Methods
All adult patients admitted to 6 hospital wards in October and November 2015 were included.
MEWS were checked each morning by the research team. For each critical score (MEWS ≥ 3),
the clinical staff was inquired about the actions performed. 30-day follow-up for SAEs was
performed to compare between patients with and without a critical score.
Results
1053 patients with 3673 vital parameter measurements were included, 200 (19.0%) had
a critical score. The protocol adherence was 89.0%. 18.2% of MEWS were calculated wrongly.
Patients with critical scores had significant higher rates of unplanned ICU admissions [7.0%
vs 1.3%, p < 0.001], in-hospital mortality [6.0% vs 0.8%, p < 0.001], 30-day readmission
rates [18.6% vs 10.8%, p < 0.05], and a longer length of stay [15.65 (SD: 15.7 days) vs 6.09
(SD: 6.9), p < 0.001]. Specificity of MEWS related to composite adverse events was 83% with
a negative predicting value of 98.1%.
Conclusions
Protocol adherence was high, even though one-third of the critical scores were calculated
wrongly. Patients with a MEWS ≥ 3 experienced significantly more adverse events. The negative
predictive value of early morning MEWS < 3 was 98.1%, indicating the reliability of this score
as a screening tool.
Keywords
Early warning score, Protocol adherence, Patient safety, Serious adverse events, Rapid
intervention teams, Track and trigger systems
81
INTRODUCTIONSerious adverse events (SAEs) in hospitalised patients are preceded by signs of clinical
deterioration in up to 80% of the patients.1 Therefore, changes in vital parameters such as
pulse rate, respiratory rate, and level of consciousness are often considered as predictors of
SAEs such as cardiac arrest, death and unplanned intensive care unit (ICU) admissions.1,2 To
improve timely detection and treatment of deteriorating patients on nursing wards, rapid
response systems (RRSs) have been introduced.3-5 RRSs consist of two different components:
an afferent limb consisting of track and trigger systems (TTS) such as Modified Early Warning
Score (MEWS) and an efferent limb, a rapid intervention team (RIT) consisting of trained ICU
personnel who will deliver immediate treatment to deteriorating patient at the bedside.
Some studies have demonstrated positive effects of implementing TTSs such as MEWS on
patient outcomes.6 On the basis of these results TTSs have been introduced in many hospitals
to increase patient safety.7,8 Firstly introduced in 1997 by Morgan et al. the TTS functions as
the afferent limb and is designed to detect deterioration early.9 Since this first introduction
multiple early warning bedside monitoring tools have been developed and implemented
internationally.10,11 These TTSs are used to detect deterioration and call upon a team to monitor
and treat patients to prevent further deterioration.12 In the VU university medical centre
(VUmc), RRS with an afferent limb consisting of a TTS (MEWS) and an efferent limb consisting
of a rapid intervention team (RIT) was introduced a few years ago. Because the afferent
limb of the system (RIT) did not function optimally, it was decided to reintroduce the MEWS
protocol in 2015 and (re)train the clinical staff aiming to change their mindset and improve
protocol adherence.
The effectiveness of a RRS is not only decided by the quality of the RIT but also by an
appropriate implementation and use of the TTS such as the MEWS.8,13 Unfortunately, very
few prospective studies have yet been performed investigating the compliance to any TTS
protocol in a real-life setting. In addition, although Smith et al. (2008) demonstrated MEWS
as a predictor for clinical outcomes retrospectively, prospective studies investigating the ability
of the MEWS to predict relevant clinical outcomes in a general in-hospital population are
lacking.14 In addition, no previous studies have investigated the association between MEWS
and the chance of 30-day readmissions. Positive association of MEWS with these endpoints
can be used to convince doctors and nurses about the value of MEWS as a prediction tool and
thereby increase their protocol adherence.
Therefore, the main aim of this study was to determine the protocol adherence mainly to
the afferent limb but also to the efferent limb in a real-life setting. The secondary aims were
to investigate the ability of once a day MEWS measurement to predict patient outcomes:
in-hospital mortality, hospital length of stay, cardiac arrests, ICU-admissions and 30-day
readmissions. Ultimate goal was to provide the hospital staff more insights into the value
of the MEWS in predicting outcomes in their own patient population and thereby increase
the awareness and protocol adherence.
82
MATERIALS AND METHODSThis prospective study was conducted in a large urban university medical centre (VUmc) with
approximately 50,000 admissions per annum in the Netherlands.
Patient selection
In the 7-week inclusion period from the 8th October until the 30th of November, all adult
patients who were in hospital at 08.00 at the date of inclusion on five wards (acute admission
unit, general surgery, internal medicine, trauma surgery, vascular surgery/urology/nephrology
ward) were included. Due to logistical reasons patients from the pulmonary ward were
included from the 1st of November. Patients 18 years and older with at least one overnight stay
were included. The Ethics committee of VU University Medical Centre Amsterdam, approved
the study and necessity for informed consent was waived.
MEWS protocol in our institution
In our hospital all vital parameter measurements are stored in an automatic electronic system.
According to the hospital wide protocol, every morning at the end of the nightshift or at
the beginning of the dayshift, nurses were requested to determine the MEWS using vital
parameter measurements recorded in this electronic system. Although MEWS measurements
could be repeated any time during the day on indication by the nurses and doctors, only
these early morning scores were used for analysis. The MEWS consists of an easy-to-use
algorithm of seven parameters (Fig. 1).15 The range for the MEWS is between 0 and 19.
During the implementation of the protocol staff was trained extensively and the protocol card
containing the protocol was distributed. MEWS was calculated by hand and electronically
documented in patients’ charts. A total score of 3 or higher was considered as a critical score.
Once a patient reaches a critical MEWS (≥ 3) nurses were requested to contact the doctor in
charge immediately. The doctor must then assess the patient within 30 minutes and draft
a plan for treatment, evaluate this after 60 minutes or call a RIT team. The RIT may also directly
be called by the nurses or the doctor at the outset.
Data collection
Charts of all included patients were checked by the coordinating investigator (CD) to obtain
the patients’ MEWS and to determine whether scores were documented and calculated
correctly. The MEWS were perceived as documented if MEWS was explicitly reported in
nurses charts’. If a vital parameter was not documented in the system, this parameter was
considered to be normal. Scores were recalculated by CD using available data in the charts.
If a patient had a critical score, charts were examined to find out what actions has been
taken, subsequently the nurses and doctors were asked about their actions. If no action
was undertaken the investigator inquired the staff about the reasons. If during recalculation
a patient had a MEWS of ≥ 3 and this was not explicitly documented by the nurse, nurses were
still asked about their actions. However, if a patient had a MEWS of ≥ 3 during recalculation
by the CD and this was wrongly calculated and documented by the nurse as a MEWS < 3,
no questions were asked. At the end of the inclusion period all answers were categorised.
83
Figure 1 | Mews and protocol in VUmc
If more than one action was taken, the most serious action was used in the categorisation.
For patients who were in hospital for multiple days the highest reached MEWS, labelled as
‘MaxScore’, was taken for predictive analysis.
Follow-up
All patients admitted during the study period were followed up for 30 days after inclusion. In
addition, patients were followed up for 30 days after discharge to obtain information about
the 30-day unplanned hospital readmission rate. MaxScore per patient was used to perform
the predictive analysis of MEWS.
Statistical analysis
Descriptive characteristics and frequencies were calculated in SPSS version 22.0 (SPSS, Chicago,
IL, USA). Categorical outcome measures are presented as frequencies and percentages.
Continuous variables are summarised by mean and standard deviation since data was
distributed normally. To illustrate the comparison in adverse events between patients who
had a MEWS < 3, versus MEWS ≥ 3 a chi-squared test was used. P-values below 0.05 were
considered significant.
RESULTSPatient characteristics
A total of 1053 patients were included during the 8-week inclusion period. Table 1 shows
patient characteristics. Most patients were admitted to the Acute Medical Unit (n=408,
38.8%), the least to the general surgery ward (n=113, 10.7%). The mean age of patients in
this cohort was 61.1 (SD 17.6).
84
Figure 2 | Protocol adherenceMeasurement and documentation. Horizontal section I representing all MEWS measurements, regardless of score, Horizontal section II representing MEWS ≥ 3, as recalculated by the coordinating researcher.
Figure 3 | Actions undertaken on patients by clinical staff after critical score reachedN= number of MEWS measurements ≥ 3. *Expectative since this high score is expected as a result of the (known) disease process or the treatment.
85
Table 1 | Patient characteristics (N=1053)
WardPatients Number (%) Male (%) Mean age (SD)
MaxScore*** Median [range]
Acute medical Unit 408 (38.8) 220 (53.9) 61.4 (18.9) 1.0 [0-9]Non-critical score* 365 (89.5) 1.0 [0-2]
Critical score** 43 (10.5) 4.0 [3-9]
Internal medicine 120 (11.4) 60 (50.0) 66.4 (16.8) 2.0 [0-8]Non-critical score* 80 (66.7) 2.0 [0-2]
Critical score** 40 (33.3) 3.5 [3-8]
General surgery 113 (10.7) 69 (61.1) 65.2 (14.5) 2.0 [0-8]Non-critical score* 70 (61.9) 1.0 [0-2]
Critical score** 43 (38.1) 4.0 [3-8]
Vascular/urology/nephrology 140 (13.3) 92 (65.7) 60.1 (14.8) 1.0 [0-6]Non-critical score* 119 (85.0) 1.0 [0-2]
Critical score** 21 (15.0) 3.0 [3-6]
Trauma surgery 151 (14.3) 72 (47.7) 53.8 (18.6) 1.0 [0-5]Non-critical score* 122 (80.2) 1.0 [0-2]
Critical score** 29 (19.2) 3.0 [3-5]
Pulmonary diseases 121 (11.5) 56 (46.3) 61.6 (14.4) 1.0 [0-6]Non-critical score* 97 (80.2) 1.0 [0-2]
Critical score** 24 (19.8) 3.5 [3-6]
Total cohort 1053 (100.0) 569 (54.0) 61.1 (17.6) 1.0 [0-9]Non-critical score* 853 (81.0) 450 (52.8) 60.5 (17.4) 1.0 [0-2]
Critical score** 200 (19.0) 119 (59.5) 63.8 (18.0) 3.0 [3-9]
* MEWS < 3**MEWS ≥ 3***MaxScore: Highest reached MEWS for patients who were in hospital for multiple days.
Table 2 | Patient Outcomes
MEWS < 3 n=853 (81%)
MEWS ≥ 3 n = 200 (19% Significance
Odds Ratio (95% CI)
Composite endpoint reached (%) 16 (1.9) 25 (12.5) p < 0.001¹ 7.5 (3.9 - 14.3)
ICU-admissions 11 (1.3) 14 (7.0) p < 0.001² 5.8 (2.6 - 12.9)
In-hospital mortality 7 (0.8) 12 (6.0) p < 0.001² 7.7 (3.0 - 19.9)
Resuscitation 0 (0.0) 1 (0.5) p = 0.190² -
Readmission (%) 91 (10.8) 35 (18.6) p < 0.05¹ 1.9 (1.2 - 2.9)
Length of Stay (SD) 6.09 (6.9) 15.7 (15.7) p < 0.001³ -
RIT-call (%) - 21 (10.5) - -
1: Pearson Chi-squared 2: Fisher’s exact test 3: Independent samples t-test
86
Measuring and documentation
There were 4041 patient days where vital parameter measurements could have taken place
according to protocol. Fig. 2 displays a flowchart of the measurement and documentation.
368 potential measurement moments were missed because these patients were not present on
the ward during the time of assessment or because they were in palliative care. This resulted in
a total of 3673 morning round measurements in 1053 patients. Of these 3673 vital parameter
measurements, 3270 were explicitly documented in nurses’ charts, resulting in a protocol
adherence of 89.0%. The investigator recalculated all MEWS using the vital parameters
measurements in the charts. The determined MEWS were referred to as recalculated MEWS.
We observed a correct calculation in 2600/3673 (70.8%) of the scores in nurses’ charts, 670
(18.2%) scores were calculated incorrectly. The recalculated MEWS were < 3 in 3316 (90.3%)
and were ≥ 3 in 357 (9.7%) measurements.
Actions performed by clinical staff
In 257 (72.0%) instances in which MEWS ≥ 3 the investigator inquired clinical staff what
action they undertook. Fig. 3 shows the actions undertaken by hospital staff. In 10 (3.5%)
cases no actions could be found in charts and no staff members could answer the questions.
Of the remaining 247 cases a doctor was contacted 169 (68.4%) times and 78 (31.6%) times
no doctor was contacted. The categorised actions performed are displayed in S1 Table and
S2 Table. Of the 169 times a doctor was contacted the doctor intervened 70 (41%) times.
The main reason for not intervening was that clinical staff did not feel the urge to perform an
action since they judged the situation as not alarming enough.
Patient outcomes
The vital parameters to calculate a MEWS were measured in 1053 patients. Two-hundred
patients (19.0%) had a critical score during their hospital stay. The remaining 853 (81%)
patients did not have a critical score. Table 2 shows the relation between a critical MEWS and
patient outcome. Having a critical score was associated with a higher percentage of unplanned
ICU admission [7.0% vs. 1.3%, OR 5.8 (2.6 – 12.9), p < 0.001], and a higher in-hospital
mortality [6.0% vs. 0.8, OR 7.7 (3.0 – 19.9), p < 0.001]. Also, results show that patients
with a critical score had a longer length of stay [15.7 days (SD: 15.7) vs. 6.09 days (SD: 6.9)
p < 0.001] and the 30-day readmission rate was higher [18.6% vs. 10.8%, OR 1.9 (1.2 – 2.9),
p < 0.05] than patients without a critical score. Sensitivity for MEWS related to composite
adverse events was 61%, specificity 83%, positive predicting value 12.5% and the negative
predicting value was 98.1%. MEWS of 3 to 5 show significant more adverse events compared
to MEWS below 3. MEWS above 5 show significant more adverse events than MEWS < 3
(p < 0.001) but compared to MEWS 3-5 no significance was reached (p = 0.196). Fig. 4 shows
patient outcomes compared between different scores.
DISCUSSIONIn this prospective study conducted in a real-life setting, we have demonstrated that adherence
to the MEWS protocol in our hospital was good (89%). However in some cases (18%)
87
the MEWS was calculated incorrectly because values were not added up properly, influencing
the total score. Although, in the majority of the cases the nurse informed the doctor about
the critical score an intervention only occurred in one-third of the cases mostly because
the situation was judged as not alarming. The MEWS of 3 or higher was a strong predictor of
clinical endpoints such as in-hospital mortality, 30-day readmissions, hospital length of stay. In
addition, the negative predictive value of MEWS < 3 in this general hospital population was
98.1% indicating the reliability of this score as a screening tool.
The afferent limb is an important component of a RRS, since an effective clinical response
depends on early recognition of deterioration.8,13 When we implemented the RRS in our hospital
a few years ago the afferent limb was implemented without a clear protocol. Therefore the TTS
did not function properly. We re-trained the clinical staff and a clear protocol was implemented
in 2015. In this protocol nurses were requested to always take a MEWS score in the morning.
The main aim of this study was to analyse protocol adherence after this reimplementation. In
addition, we aimed to analyse the value of the morning MEWS measurement in predicting
clinical outcomes in this general hospital population because this has not been evaluated
in a prospective study in a real-life setting. The results of this study showed a high protocol
adherence with nurses completing MEWS documentation in 89% of the measurements.
However, a percentage (18%) of wrongly documented scores were also seen, likely due to
wrong calculations in adding up separate MEWS parameters. An important finding was that
due to these wrong calculations, a relatively high percentage of critical scores were missed
by nurses. Twenty-eight percent of the critical scores, where a doctor was supposed to be
alarmed, were not recognised by the nurses. Our study has also shown, that doctors were
not contacted in one-third (32%) of the critical scores. When physicians were contacted,
they only undertook an action in 28% of the cases. The main reason for not taking action
was that staff judged the situation as not alarming. These findings are comparable to Jones
Fig 4 | Adverse events compared between MEWS groupsSignificant with MEWS < 3 with a p-level of p < 0.001. OR = Odds ratio
88
et al. (2011) who also found a percentage of 29%.16 Reasons for these findings, as explained
in previous work, are that clinical staff feel the parameter is too rigorous in its cut-offs or
the nursing staff estimate the situation as being under control.17,18 However, previous work
has already demonstrated that changing the critical cut-off to 4 devaluates MEWS as a reliable
screening tool.19
Since creating awareness and emphasising the importance of the MEWS can increase protocol
adherence a secondary aim was to validate the MEWS as a predictor for adverse events in our
own hospital population. We demonstrate prospectively for the first time in a real-life setting
that patients with a MEWS ≥ 3 in one of the morning measurements had an increased risk
for an unplanned adverse event than the patients with a MEWS < 3. No significant increase
was observed for unplanned resuscitations, likely due to the very low incidence of events. To
our knowledge, this is the first study validating this MEWS protocol prospectively in a general
in-hospital population in real-life setting. One other study has prospectively validated the value
of MEWS in predicting adverse events in a European surgical population. This study was also
performed in a real-life setting. Their results are consistent with our findings.20 In addition,
a recent publication in Africa validated the MEWS prospectively in a research setting in low-
resource circumstances.21 They too found that the MEWS was a useful triage tool to identify
patients at the greatest risk of experiencing an adverse event. We also demonstrate for the first
time that MEWS ≥ 3 is associated with a significantly higher readmission rate within 30 days
for a critical score (10.8% vs. 18.6%). Since readmissions are known to increase mortality and
are associated with functional decline, it again underlines the importance of the MEWS as
a screening tool.22-24
MEWS as part of the RRT system, was implemented in many Dutch hospitals to potentially
increase patient safety.25 MEWS is a relatively low-cost and convenient bedside monitoring
tool, however critical scores can lead to a higher workload for clinical staff. This study, however,
again emphasised the clinical importance of recognising patients with a MEWS higher than
3 since these patients are at high risk of developing adverse events. In addition, the negative
predictive value of MEWS < 3 was 98.1 underscoring the importance of MEWS as a screening
tool. Nevertheless, it is worth mentioning only 7% of the patients in our population with
a MEWS ≥ 3 were transferred to the ICU. We do not know how many patients were prevented
from ICU admission by early recognition and prompt treatment on the wards.
The strength of this study is its prospective study design in a real world general hospital sample
in which 3290 MEWS values were analysed. In addition we personally contacted every nurse
who was involved with the MEWS or vital parameter measurements to collect information
daily. This is the largest prospective study conducted so far validating MEWS as a screening
tool in a general in-hospital (medical and surgical) population.26,27 The study was conducted
in a single-centre which uses one specific MEWS protocol. Therefore, results might not be
generalised to hospitals using another EWS protocol. Also, since our aim was to determine
clinical relevance of MEWS in daily practice, a real-life hospital situation was studied. As
89
a result, the determined MEWS and not the completeness of the vital parameter set was taken
into account. This could possibly under- or overestimate the relation between MEWS and
patient outcomes.
CONCLUSIONIn this prospective study performed in a real-life setting we demonstrated that adherence to
the MEWS protocol in our hospital is good (89%). A morning Modified Early Warning Score
of 3 or higher was a strong predictor of clinical endpoints such as in-hospital mortality, 30-day
readmissions, hospital length of stay. In addition, the negative predictive value of MEWS < 3 in
this general hospital population was 98.1% indicating the reliability of this score as a screening
tool. Therefore, it is important to keep emphasising the clinical relevance of the MEWS among
clinical staff.
CONFLICTS OF INTERESTSNone
ACKNOWLEDGEMENTSOur thanks to all the nurses and the doctors of the units for their help during the study. Special
thanks to Edwin Pompe (manager care) and Sascha Spoor (emergency nurse) for their support.
SUPPORTING INFORMATION CAPTIONSS1 Table. Actions undertaken on patients by clinical staff after critical score reached.
*Since it is part of disease/treatment or patient is familiar with abnormalities
S2 Table. Categorisation of actions of clinical staff.
*Since it is part of disease/treatment or patient is familiar with abnormalities
S1 Dataset. Vital parameters and measured MEWS.
S2 Dataset. Categorization actions undertaken at MEWS ≥ 3 by clinical staff.
S3 Dataset. Patient outcomes.
90
REFERENCES1. Ludikhuize J, Smorenburg SM, de Rooij SE, de Jonge E. Identification of deteriorating patients on
general wards; measurement of vital parameters and potential effectiveness of the Modified Early Warning Score. J Crit Care 2012; 2012/02/22:424 e427-413.
2. Kim WY, Shin YJ, Lee JM, et al. Modified Early Warning Score Changes Prior to Cardiac Arrest in General Wards. PLoS One. 2015;10(6):e0130523.
3. Hillman K, Parr M, Flabouris A, Bishop G, Stewart A. Redefining in-hospital resuscitation: the concept of the medical emergency team. Resuscitation. 2001;48(2):105-110.
4. Mathukia C, Fan W, Vadyak K, Biege C, Krishnamurthy M. Modified Early Warning System improves patient safety and clinical outcomes in an academic community hospital. J Community Hosp Intern Med Perspect. 2015;5(2):26716.
5. Alam N, Hobbelink EL, van Tienhoven AJ, van de Ven PM, Jansma EP, Nanayakkara PW. The impact of the use of the Early Warning Score (EWS) on patient outcomes: a systematic review. Resuscitation. 2014;85(5):587-594.
6. Moon A, Cosgrove JF, Lea D, Fairs A, Cressey DM. An eight year audit before and after the introduction of modified early warning score (MEWS) charts, of patients admitted to a tertiary referral intensive care unit after CPR. Resuscitation. 2011;82(2):150-154.
7. Kyriacos U, Jelsma J, Jordan S. Monitoring vital signs using early warning scoring systems: a review of the literature. J Nurs Manag. 2011;19(3):311-330.
8. Ludikhuize J, Borgert M, Binnekade J, Subbe C, Dongelmans D, Goossens A. Standardized measurement of the Modified Early Warning Score results in enhanced implementation of a Rapid Response System: a quasi-experimental study. Resuscitation. 2014;85(5):676-682.
9. Morgan R, Williams F, Wright M. An early warning scoring system for detecting developing critical illness. Clin Intensive Care. 1997;8(2):100.
10. Gao H, McDonnell A, Harrison DA, et al. Systematic review and evaluation of physiological track and trigger warning systems for identifying at-risk patients on the ward. Intensive Care Med. 2007;33(4):667-679.
11. Subbe CP, Gao H, Harrison DA. Reproducibility of physiological track-and-trigger warning systems for identifying at-risk patients on the ward. Intensive Care Med. 2007;33(4):619-624.
12. DeVita MA, Bellomo R, Hillman K, et al. Findings of the first consensus conference on medical emergency teams*. Critical care medicine. 2006;34(9):2463-2478.
13. Kolic I, Crane S, McCartney S, Perkins Z, Taylor A. Factors affecting response to national early warning score (NEWS). Resuscitation. 2015;90:85-90.
14. Smith GB, Prytherch DR, Schmidt PE, Featherstone PI. Review and performance evaluation of aggregate weighted ‘track and trigger’ systems. Resuscitation. 2008;77(2):170-179.
15. Subbe C, Kruger M, Rutherford P, Gemmel L. Validation of a modified Early Warning Score in medical admissions. Qjm. 2001;94(10):521-526.
16. Jones S, Mullally M, Ingleby S, Buist M, Bailey M, Eddleston JM. Bedside electronic capture of clinical observations and automated clinical alerts to improve compliance with an Early Warning Score protocol. Crit Care Resusc. 2011;13(2):83-88.
17. Shearer B, Marshall S, Buist MD, et al. What stops hospital clinical staff from following protocols? An analysis of the incidence and factors behind the failure of bedside clinical staff to activate the rapid response system in a multi-campus Australian metropolitan healthcare service. Bmj Quality & Safety. 2012;21(7):569-575.
18. Davies O, DeVita MA, Ayinla R, Perez X. Barriers to activation of the rapid response system. Resuscitation. 2014;85(11):1557-1561.
91
19. van Rooijen CR, de Ruijter W, van Dam B. Evaluation of the threshold value for the Early Warning Score on general wards. Neth J Med. 2013;71(1):38-43.
20. Smith T, Den Hartog D, Moerman T, Patka P, Van Lieshout EM, Schep NW. Accuracy of an expanded early warning score for patients in general and trauma surgery wards. Br J Surg. 2012;99(2):192-197.
21. Kruisselbrink R, Kwizera A, Crowther M, et al. Modified Early Warning Score (MEWS) Identifies Critical Illness among Ward Patients in a Resource Restricted Setting in Kampala, Uganda: A Prospective Observational Study. PLoS One. 2016;11(3):e0151408.
22. Greysen SR, Stijacic Cenzer I, Auerbach AD, Covinsky KE. Functional impairment and hospital readmission in medicare seniors. JAMA Intern Med. 2015;175(4):559-565.
23. Zanocchi M, Maero B, Martinelli E, et al. Early re-hospitalization of elderly people discharged from a geriatric ward. Aging Clin Exp Res. 2006;18(1):63-69.
24. Balla U, Malnick S, Schattner A. Early readmissions to the department of medicine as a screening tool for monitoring quality of care problems. Medicine (Baltimore). 2008;87(5):294-300.
25. Ludikhuize J, Hamming A, de Jonge E, Fikkers BG. Rapid response systems in The Netherlands. Jt Comm J Qual Patient Saf. 2011;37(3):138-144, 197.
26. Armagan E, Yilmaz Y, Olmez OF, Simsek G, Gul CB. Predictive value of the modified Early Warning Score in a Turkish emergency department. Eur J Emerg Med. 2008;15(6):338-340.
27. Ho le O, Li H, Shahidah N, Koh ZX, Sultana P, Hock Ong ME. Poor performance of the modified early warning score for predicting mortality in critically ill patients presenting to an emergency department. World J Emerg Med. 2013;4(4):273-278.
THE USE OF QUALITY INDICATORS TO ASSESS PATIENT SAFETY
CHAPTER 6
EXPLORING THE PREVENTABLE CAUSES OF UNPLANNED READMISSIONS USING ROOT CAUSE ANALYSIS: COORDINATION OF CARE IS THE WEAKEST LINK
Louise van Galen & Kristien Fluitman | Hanneke Merten | Saskia M. Rombach | Mikkel Brabrand | Tim Cooksley | Christian H. Nickel | Christian P. Subbe |
Mark H. Kramer | Prabath W. Nanayakkara |
*On behalf of the safer@home consortium
Eur J Intern Med 2016;30:18-24
‘The things that count aren’t always the things that can be counted’ Kiki Lombarts
94
ABSTRACTImportance
Unplanned readmissions within 30 days are a common phenomenon in everyday practice and
lead to increasing costs. Although many studies aiming to analyse the probable causes leading
to unplanned readmissions have been performed, an in depth-study analysing the human
(healthcare worker)-, organizational-, technical-, disease- and patient-related causes leading
to readmission is still missing.
Objective
The primary objective of this study was to identify human-, organizational-, technical-,
disease- and patient-related causes which contribute to acute readmission within 30 days
after discharge using a Root-Cause Analysis Tool called PRISMA-medical.
The secondary objective was to evaluate how many of these readmissions were deemed
potentially preventable, and to assess which factors contributed to these preventable
readmissions in comparison to non-preventable readmissions.
Design
Cross-sectional retrospective record study.
Setting
An academic medical centre in Amsterdam, the Netherlands.
Participants
Fifty patients aged 18 years and older discharged from an internal medicine department and
acutely readmitted within 30 days after discharge.
Main Outcome Measures
Root causes of preventable and unpreventable readmissions.
Results
Most root causes for readmission were disease-related (46%), followed by human (healthcare
worker)- (33%) and patient- (15%) related root causes. Half of the readmissions studied were
considered to be potentially preventable. Preventable readmissions predominantly had human-
related (coordination) failures.
Conclusion and Relevance
Our study suggests that improving human-related (coordinating) factors contributing to
a readmission can potentially decrease the number of preventable readmissions.
95
Keywords
Patient readmission, root cause analysis, quality improvement
Highlights
1. PRISMA-analysis is a suitable method to perform in-depth analysis of readmissions.
2. This study provides a structured analysis of root causes for unplanned readmissions.
3. Half of the readmissions studied were considered to be potentially preventable.
4. Healthcare worker coordination failures were mostly responsible.
96
INTRODUCTIONUnexpected hospital readmissions within 30 days after an index admission are highly prevalent
and costly. The proportion of patients readmitted within 30 days varies widely and is estimated
to be between 7 and 24% across borders.1-4 In some countries readmission rates are used as
a quality and safety criteria to rate and reimburse units.5,6 In 2016 the Dutch government will
introduce readmission rates as an official quality indicator in hospitals.7 To identify the patients
at high risk of readmission, attempts have been made to develop a prediction score based
on clinical variables.8,9 However, these scores have not been validated to predict readmissions
in different populations worldwide and show low to moderate discriminatory power in
predicting readmissions.10-12
More importantly, little is known about the preventability of the unexpected readmissions. It
would seem logical that hospitals only have to accept ramifications for preventable readmissions.
Reported preventable readmission rates vary from 5 % to 79 %.13 However, current literature
has not been able to define clear risk factors that predict a preventable readmission, which
may be targeted to reduce the rate of unexpected readmissions and improve the quality of
care.14 In addition, preventability has not yet been defined uniformly.15
The problem is that readmissions appear to be multi-causal. They are potentially related to
multimorbidity and recent studies have found a progressive increase in readmission risk as
the degree of functional impairment increases.3,16,17 Some research groups have investigated
factors contributing to preventable readmissions in more detail and have classified causes
according to not only patient but also system and social causes.6,18,19 These causes, however,
are often non-modifiable and beyond the reach of implementing process improvements in
hospitals. Currently, no studies have been published on other potentially relevant factors such
as the healthcare worker-, organizational-, technical- and patient-related causes that could
contribute to acute readmissions within 30 days.
A useful tool to analyse these types of root causes is the PRISMA-tool (Prevention and Recovery
Information System For Monitoring and Analysis). The main goal of the PRISMA method is to
build a quantitative database of incidents and process deviations, from which conclusions may
be drawn to suggest optimal countermeasures. This method has been accepted by the World
Alliance for Patient Safety of the World Health Organization.20-22
Insight into these characteristics and potential preventability of acute readmissions may be of
help to ultimately reduce the number of readmissions and the costs attached. Understanding
factors contributing to (preventable) readmissions would help physicians increase the safety
surrounding discharge for patients and their caregivers.
In this retrospective record review study, the main aim was to identify the organizational-,
technical-, healthcare worker- and patient-related causes that contribute to readmissions
97
using “PRISMA-medical” analysis. The secondary aim was to evaluate how many of these
readmissions were deemed potentially preventable, and to assess what factors contributed to
these preventable readmissions in comparison to non-preventable readmissions.
SUBJECTS AND METHODSStudy Design
This retrospective, cross-sectional record review study included readmissions at the VU University
Medical Centre in Amsterdam, the Netherlands. The VUmc is an academic medical centre with
approximately 3400 admissions per annum to its Acute Medical Unit (AMU) and a hospital
wide readmission rate of around 10%. Previous studies have shown that 50 PRISMA-analyses
are valid and sufficient basis for a reliable causal-profile.[23] In the current study, we therefore
decided that 50 records would be included for a review to explore the causes and potential
preventability of unplanned readmissions. All readmission records of the year 2013 meeting
the inclusion criteria were selected. To get the most recent results, the reviewers started with
the last record from December 2013 and included each consecutive record backwards until
the number of 50 records was reached, with the last record coming from June 2013.
The following criteria were used for inclusion into the study sample: patients aged 18 years
and older; initial discharge from the internal medicine ward (which included the following
specialties: general internal medicine, nephrology, oncology, geriatrics, haematology,
pulmonary medicine); admission and readmission through the emergency department in 2013
and readmission to any department/ward, regardless of the medical specialty within 30 days
after initial discharge. All records not meeting the above mentioned inclusion criteria were
excluded from the study.
If a patient was readmitted more than once during the study period only the first readmission
was analysed. The local Medical Ethics Committee approved this study.
Assessment
Doctor’s charts, nurse’s charts and electronic patient files including all test results were
available for analysis. For each individual readmission information on patient characteristics
(such as age, co-morbidity23, living situation) and circumstances under which initial discharge
and readmission took place (such as length of stay and specialty for admission) was collected
according to a standardized chart abstraction form (appendix 1). Two medically and PRISMA-
trained investigators (LG, KF) reviewed each case separately and filled out these chart
abstraction forms. Subsequently consensus was reached.
PRISMA Analysis
In order to perform PRISMA-analysis on all readmissions, the above mentioned data collection
forms were used. These consisted of a free text description of the circumstances contributing
to the readmission and identification of direct, indirect and root causes. This information
98
was used to construct a causal tree. At the top of each tree the readmission was placed as
an unforeseen adverse event. The direct causes of the readmissions were noted hereunder.
Under each of the direct causes the indirect causes were stated. By constantly asking ‘why’ an
event and each subsequent event had taken place relevant indirect causes were revealed. This
continued until no more objective information was available to reveal an underlying cause.
The last noted indirect cause was labelled as root cause and was located at the bottom of
the causal tree (Figure 1).
Root causes were classified as technical-, organizational-, human- (healthcare worker) and
patient-related factors according to the Eindhoven Classification Model, see Table 1.20,21
Disease-related causes were added as fifth category to this model. We anticipated that
progression of disease would be identified as root cause in many readmissions without any
technical-, organizational-, healthcare worker- or patient-related factors contributing to
the readmission. Table 1 shows the subcategories of PRISMA-root causes with case examples
from our study. Finally, both reviewers studied the cases independently and extensively. They
concluded whether the readmission related to its index admission and if it was potentially
preventable or not. The readmission was judged to be related to the index admission if both
admissions were based on the same medical issue or if cause of the readmission originated
during index admission. A readmission was judged as potentially preventable if it could have
been reasonably foreseen by discharging physician and could reasonably have been prevented
by any action undertaken by hospital staff or the patient. If consensus could not be reached
satisfactorily, a third independent party was consulted. These cases were then re-analysed and
discussed with a senior physician (PN) and a psychologist with a special interest in PRISMA-
analysis (HM). The mean time to assess a case by a single reviewer was 72 minutes (range
25-180, SD=71), reaching consensus thereafter took up approximately 30 minutes per case.
Statistical Analysis
Descriptive characteristics and frequencies were calculated in SPSS version 22.0. Categorical
outcome measures are presented as frequencies and percentages. Continuous variables are
summarized by median and interquartile ranges since none of them were normally distributed.
To illustrate the comparison in patient characteristics between preventable and non-preventable
readmissions we used the Mann-Whitney U test for continuous and ordinal variables. Pearson’s
chi-square test and Fisher’s exact were used for dichotomous and categorical data. P-values
below 0,05 were considered significant.
RESULTSPatient characteristics
For 46 out of the 50 reviewed cases, consensus on the root-cause analysis and preventability
was reached by the two reviewers. Independent observers were consulted to assess the root
causes and preventability in 4 cases, after which agreement was reached. During the study
period, six cases had more than one readmission within 30 days. Table 2 shows baseline
99
Figure 1 | Two examples of causal trees IA: index admission, DRF: disease-related factor, HRC: human-related coordination, PRF: patient-related factor, H-ex: human-related external.
characteristics of the study population. The mortality one year after index admission was 40%
(n=20). Most readmissions (n=28, 56%) occurred more than one week after discharge, about
one third occurred within 72 hours after discharge (n=17, 34%). In 13 (26%) readmissions
no discharge letter was provided. For 35 (70%) of the readmissions the admitting specialty
for the index admission was the same as for the readmission. The majority (n=45, 90%) lived
independently before the index admission. A large proportion of the population (n=36, 72%)
used five or more drugs at the time of the index admission.
Preventability
Reviewers judged more than half (n=26, 52%) of readmissions as potentially preventable.
All preventable readmissions were found to be related to the index admission. However,
16 (32%) readmissions that were related were considered not preventable. There were no
significant differences in patient characteristics between the (potentially) preventable versus
non-preventable readmission-groups although this should be interpreted with caution because
of the small sample size (Table 2). An example of a related and preventable human-related
caused readmission within one week after discharge concerned a patient who was readmitted
with malaise and self-neglect. During the index admission both patient and family members
had repeatedly expressed their concerns about the discharge. They felt the patient was unable
100
Table 1 | Description of categories of the Eindhoven Classification Model: PRISMA-medical Version.20,21
Main category Subcategory Code Description Example (if available)
Technical External T-ex Technical failures beyond the control of the organization
Design TD Failures to poor design of equipment etc.
Construction TC Correct design inappropriately constructed or placed
Materials TM Material defects not classified under TD or TC
Organizational External O-ex Failures at an organizational level beyond the control and responsibility of the investigating team
Transfer of knowledge
OK Failure resulting from inadequate measures to train or supervise new or inexperienced staff
Protocols OP Failures relating to the quality or availability of appropriate protocols
Management priorities
OM Internal management decisions which reduce focus on patient safety when faced with conflicting priorities
Culture OC Failure due to attitude and approach of the treating organization.
Human External H-ex Human failures beyond the control of the organization/department
Nursing home did not monitor fluid restriction
Knowledge-based behaviour
HKK Failure of an individual to apply their knowledge to a new clinical situation
Not adjusting tramadol dosage to poor kidney function
Qualifications HRQ An inappropriately trained individual performing the clinical task
Co-ordination HRC A lack of task co-ordination within the healthcare team.
Inadequate medication handover to the general practitioner
Verification HRV Failure to correctly check and assess the situation before performing interventions
Urine culture not verified before patient was discharged
Intervention HRI Failure resulting from faulty task planning or performance
Monitoring HRM Failure to monitor the patient’s progress or condition
Response to oral antibiotics not monitored after switching from intravenous antibiotics
101
to deal with her own medication and to take care of herself. Hospital staff did report this in
the medical records, but they did not take extra discharge precautions or arrange extra care
at home. An example of a readmission judged as non-related and non-preventable is the case
of a patient initially being admitted for a community-acquired pneumonia. Two weeks later
the patient was readmitted with angioedema after the start of a new antihypertensive drug
prescribed by the general practitioner, which had nothing to do with the initial admission.
Root causes
After PRISMA-analysis, a total of 100 root causes were identified. Twenty-one (42%) of
the readmissions had a single root cause, 15 (30%) had two root causes, 8 (16%) three root
causes, 5 (10%) four root causes and one readmission (2%) had five root causes. The mean
number of root causes per readmission was 2.0 (SD = 1.1). Figure 1 shows two examples of
causal trees. Table 1 presents the subcategories of PRISMA-root causes with case examples
from our study.
The results in Figure 2 show the distribution over the root causes plotted against preventability. It
illustrates which root causes were identified for preventable and non-preventable readmissions.
Regardless of preventability, most root causes were disease-related (n=46, 46%), followed by
human- (n=33, 33%) and patient- (n=15, 15%) related root causes. Human-related causes
were causes which were healthcare worker related. Disease-related causes (DRF) consisted of
natural progression of disease, which means that they could not be influenced by the physician
or patient. This is, for example, evident in the case of a patient being readmitted due to
symptoms caused by brain metastases from melanoma. Patient-related causes (PRF) originated
from failures related to patient characteristics or conditions which were beyond the control
of hospital staff, for example the refusal of further diagnostics or treatment by the patient.
Finally, 6% of the root causes were unclassifiable. We did not identify any technical (T) or
organizational (O) causes. Organizational causes would include failures that are related to
Table 1 | (continued)
Main category Subcategory Code Description Example (if available)
Skills-based HSS Failure in performance of highly developed skills
Patient Patient-related PRF Failures related to patient characteristics or conditions, which are beyond the control of staff and influence clinical progress
Non-compliance in taking prescribed medication
Disease-related DRF Failures related to the natural progress of disease which are beyond control of patient, its carers and staff
Subjective dyspnoea without disease- or patient-related substrate
102
Table 2 | Baseline characteristics stratified by preventability
Patient CharacteristicsTotal (N=50)
Considered Preventable (N=26)
Considered Non-preventable (N=24) P-value*
Age – median(range) 71.0(32-97) 75.0(41-97) 70.0(32-90) 0,0771
Gender (%(N))
male
female
38.0(19) 62.0(31)
34.6(9) 65.4(17)
41.7(10) 58.3(14)
0,6082
Deceased (%(N))
< 6 months after IA*
6-12 months after IA
> 12 months after IA
Unknown
26.0(13)
14.0(7) 4.0(2)
56.0(28)
26.9(7) 15,4(4) 7,7(2)
50.0(13)
25.0(6) 12,5(3) 0.0(0)
62.5(15)
0,4622
Length of stay IA (days) - median(range) 6.0(2-25) 7.5(2-24) 5.0(2-25) 0,1461
Length of stay RA (days)- median(range) 6.0(2-45) 6.5(2-44) 5.0(2-45) 0,2581
Length between IA – RA (%(N))
< 24 h
24 - 72h
72 h-1 week
More than 1 week
2.0(1)
32.0(16)
10.0(5) 56.0(28)
3.8(1) 26.9(7) 11.5(3) 57.7(15)
0.0(0) 37.5(9) 8.3(2) 54.1(13)
0.9041
Medical Speciality IA (%(N)
Geriatrics
Haematology General internal Nephrology
Oncology
Pulmonary medicine
2.0(1) 2.0(1) 54.0(27)
26.0(13) 14.0(7)
2.0(1)
0.0(0) 0.0(0) 61.5(16) 26.9(7) 11.5(3) 0.0(0)
4.2(1) 4.2(1) 45.8(11) 25.0(6) 16.7(4) 4.2(1)
0,6753
Speciality IA same as RA (%(N))
yes no
70.0(35) 30.0(15)
76.9(20) 23.1(6)
62.5(15) 37.5 (9)
0,2662
Living situation (%(N))
independent dependent
90.0(45) 10.0(5)
88.5(23) 11.5(3)
91.7(22) 8.3(2)
1,03
Living with partner (%(N))
yes
no
Unknown
56.0(28) 42.0(21)
2.0(1)
53.8(14) 42.3(11)
3.8(1)
58.3(14) 41.7(10)
0.0(0)
0.8692
Home care prior to IA (%(N))
None
Formal
Informal
Not applicable
Unknown
32.0(16) 18.0(9) 38.0(19)
10.0(5)
2.0(1)
30.8(8) 30.8(8) 23.1(6) 11.5(3)
3.8(1)
33.3(8) 4.2(1) 54.2(13) 8.3(2)
0.0(0)
1,0003
KATZ*24-score IA – median (range) 0.0(0-6) 1.0(0-6) 0.0(0-6) 0,0941
SNAQ*25-score IA – median (range) 0.5 (0-6) 0.0(0-6) 1.0(0-3) 0,7811
CCI*23 IA- median (range) 6.0 (1-14) 6.0(2-11) 6.5(1-14) 0,6241
103
Table 2 | (continued)
Patient CharacteristicsTotal (N=50)
Considered Preventable (N=26)
Considered Non-preventable (N=24) P-value*
Polypharmacy* (%(N))
yes
no
72.0(36) 28.0(14)
80.8(21) 19.2(5)
62.5(15) 37.5(9)
0,1512
Fall in last 6 months (%(N))
yes
no
Unknown
28.0(14) 70.0(35)
2.0(1)
34.6(9) 65.4(17)
0.0(0)
20.8(5) 75.0(18)
4.2(1)
0,3192
Use of mobility device (%(N))
yes
no
36.0(18) 64.0(32)
38.5(10) 61.5(16)
33.3(8) 66.7(16)
0,7062
Complications during IA (%(N))
yes
no
18.0(9) 82.0(41)
19.2(5) 80.8(21)
16.7(4) 83.3(20)
1.0003
Location of discharge (%(N))
Home
Other institution
84.0(42)
16.0(8)
80.8(21) 19.2(5)
87.5(21) 12.5(3)
0,7043
Living situation after discharge IA dependentindependent (%(N))
yes no
6.0(3) 94.0(47)
7.7(2) 92.3(24)
4.2(1) 95.8(23)
1.0003
Extra care after IA (%(N))
yes
no
22.0(11) 78.0(39)
26.9(7) 73.1(19)
16.7(4) 83.3(20)
0,7223
Outpatient follow-up (%(N))
yes
no
68.0(34) 32.0(16)
61.5(16) 38.5(10)
75.0(18) 25.0(6)
0,3082
Discharge letter send to GP (%(N)
yes
no
74.0(37)
26.0(13)
76.9(20)
23.1(6)
70.8(17)
29.2(7)
0,6242
*IA: Index Admission, RA: Readmission, Polypharmacy: the concomitant use of five or more drugs, CCI: Charlson Comorbidity Index corrected by age[24], SNAQ: assessment of nutritional status[32]:, KATZ: score tool for assessing a patient’s ability to perform activities of daily living [31]**1: Mann-Whitney U, 2: Pearson chi-squared, 3: Fisher’s exact test.
quality and availability of protocols within the department. A technical cause could originate
from failures resulting from poor design of equipment such as software.
Of 100 root causes, 69 were identified in patients with readmissions regarded as preventable, in
contrast to 31 that were identified in a non-preventable readmission. Results reveal that a non-
preventable readmission was never caused by a human-related factor; these were only found in
readmissions regarded as preventable. Nearly half (46.3%) of the root causes in readmissions
104
identified as preventable consisted of human-related factors. Further subcategorized, these
preventable human-related factors originated mostly from coordination (13.0%) and monitoring
(11.6%) failures. The first represent failures in task coordination within a healthcare team, i.e.
no clear medication handover towards the home situation. The latter comprises incompetence
in monitoring a process or patient status. This is exemplified in our study by a patient admitted
with urosepsis where the response to oral antibiotics was not monitored after switching from
intravenous to oral antibiotics. The patient was readmitted a few days after discharge with
high fever, and additional diagnostics at readmission revealed that bacteria found in the urine
culture were not sensitive to the oral antibiotics given. In addition, patient-related causes
explained 17.4% of potentially preventable readmissions. External factors were also present
(7.2%). These are causes originating in areas beyond the control of the hospital. For instance,
causes resulting from failures by a general practitioner or within a nursing home (Figure 1).
Non-preventable readmissions were mostly caused by disease related factors (71.0%).
DISCUSSIONThe purpose of the present retrospective record review study was to identify organizational-,
technical-, human-, disease and patient-related causes which contributed to readmissions.
The most important finding from this study is that half of the readmissions were considered
potentially preventable. The second major finding was that human (healthcare worker)-
related root causes were exclusively found in preventable readmissions. This was mostly due
Figure 2 | Root causes readmissions
105
to human-related coordination failures. Non-preventable readmissions were mostly the result
of disease-related factors.
To our knowledge, the current study is the first on causes and preventability of unplanned
readmissions performed in the Netherlands. However, using large databases, efforts have been
made globally to identify those patients at a higher risk of readmission and to develop risk
models.8,9,12,19 These risk models attempt to predict chances of readmission based on several
patient characteristics (for example the Charlson Comorbidity Index) which can be derived
from big hospital data systems. The problem with these models is that they have not yet been
validated in different populations. In addition, as highlighted by Van Walraven (2015) there is
no risk model incorporating aspects of preventability in predicting readmissions. These aspects
are of particular interest, as they could be the key to improvement measures in preventing
readmissions and facilitating a safer discharge home.10,11,26
Many studies have tried to assess the proportion of preventable readmissions and patient
characteristics related to these readmissions. In a review of 34 studies Van Walraven et al.
(2011) found that the percentage of preventable readmissions varied from 5% to 79%, with
a median of 27.1%.13 Our results of 52% preventable readmissions are in agreement with
results found in more recent studies where percentages are around 45%.6,27,28
As mentioned above, the majority of studies performed focused mainly on patient and disease
characteristics associated with preventable readmissions such as age and comorbidity.3,5,29
However, as already stated by Joynt (2012), there is a growing body of evidence that suggests
that primary drivers of variability in readmission rates reach beyond patient characteristics
since they are multi-factorial.30 Our results are consistent with recent studies which suggest
that in order to predict potentially preventable readmissions one should look into patient- and
human-related factors.6,28 Feigenbaum et al. (2012) suggested that potentially preventable
readmissions were due to patient- and human-related factors such as suboptimal medical
treatment and a lack of coordination in the discharge process.27 Bianco et al. (2012) suggested
that preventable readmissions are often associated with diagnostic and therapeutic errors, but
also with patient-related factors such as non-adherence.31 In contrast to our study, they have
not subdivided root causes into specific areas for improvement. Furthermore, despite the fact
that some previous studies did have independent reviewers, they do not elaborate precisely on
how they interpret and evaluate the reasons for readmission as being potentially preventable.
The outcomes of this study are of interest from a policy perspective. As mentioned before,
readmission rate will be included as an official quality indicator for Dutch hospitals in 2016.7
However, this careful in-depth analysis of 50 readmissions in our study showed that half of
the readmissions is non-preventable and the majority of their causes are either patient- or
disease-related, which mostly lay outside the influence of hospital personal. Therefore, it is
questionable to what extent the general readmission rate can be reliably linked to the quality
provided by the hospital and the hospital care professionals. Our study shows that careful
106
consideration is needed to judge whether a readmission may have been preventable, and
therefor of interest from a policy perspective. A large scale multicentre study analysing
readmissions could be a first step to show whether the results of our study also apply to other
hospitals. The PRISMA-analysis would be a useful tool to use, because it systematically addresses
the causes that may have contributed to the readmission. It is able to provide the reviewer with
the information needed to judge whether a readmission is potentially preventable. Analysing
these causes implicitly provides valuable information for improvement.
Furthermore, the strength of our study lies in case by case comprehensive investigation,
performed by two independent researchers to find the root causes of readmissions using
the full extent of clinical data available for each case. To our knowledge, this is the first study
that explore the underlying causes of readmissions and determine their eventual preventability.
This retrospective nature of this study limited reviewers to patient records which meant that
they had to rely solely on information that was written down in the patient records when
investigating root causes of readmissions. In addition, no technical or organizational factors
could be discovered to be instrumental in readmissions. As illustrated by Van der Schaaf et al.
these often entail latent causes. An organizational factor could for instance be a shortage of
hospital staff.21 However, such causes are not typically written down in patient charts and were
therefore inaccessible to the reviewers. This also applies to cultural factors such as hospital-wide
accepted faulty behaviour (e.g. the hospital-wide acceptance of sending discharge letters to
the general practitioner with substantial delay). This is unfortunate since these organizational
factors could provide very suitable grounds for improvements. Finally, our sample size was
relatively limited. However, when using PRISMA-tool a population size of 50 is thought to offer
a sufficient variability to perform a reliable root cause analysis.32
CONCLUSIONDespite its exploratory nature, this study offers insight into the healthcare worker- and patient-
related causes of readmissions. The relevance of human-related factors is clearly supported
by the fact that human-related causes are exclusively found in preventable readmissions.
The most significant finding of this study was that the most prevalent human-related cause of
preventable readmission was the lack of coordination within the healthcare system. The most
frequent findings were inadequate assessment of a patient’s home situation and the lack of
a proper information handover to the general practitioner, the patient and its carers at home.
This flaw in communication between caregivers is also highlighted by the large number of
missing discharge letters (n=13, 26%). In our opinion these results could function as starting
points for improvement. Perhaps a more standardized method of discharge could help reduce
the amount of potentially preventable readmissions. This should include requirements of
a clear handover of care to the primary physician and a clear assessment of extra care needed
at home during index admission.
To study the possible usefulness of these interventions, prospective study designs are needed
where more relevant and reliable information on factors that may have contributed to
107
the readmission can be collected. Perspectives of patients their informal carers and involved
healthcare professionals can also provide valuable additional information. This will provide
more detailed information on communication breakdowns and gaps in information transfer.
In addition, causal factors responsible for readmissions in different healthcare systems can be
diverse. Shortly a prospective pan-European study named ‘ the CURIOS@-study (CaptUring
Readmission InternatiOnally to prevent readmission by safer@home consortium) will start
recruiting patients with the aim of investigating the predictability and preventability of
readmissions (clinicaltrials.gov: NCT02621723).
ACKNOWLEDGEMENTSAll members of safer@home consortium: M. Brabrand, T. Cooksley, K.S. Fluitman, L.S. van
Galen, R. Kidney, J. Kellett, H. Merten, P.W. Nanayakkara, C.H. Nickel, J. Soong, C.P. Subbe,
L. Vaughan, I. Weichert
SUPPORTING INFORMATION Appendix 1. Acute Readmissions: Data collection sheet chart review.
108
REFERENCES1. Blunt I, Bardsley M, Grove A, Clarke A. Classifying emergency 30-day readmissions in England using
routine hospital data 2004-2010: what is the scope for reduction? Emergency medicine journal : EMJ. 2015;32(1):44-50.
2. Graham H, Livesley B. Can readmissions to a geriatric medical unit be prevented? Lancet. 1983;1(8321):404-406.
3. Donze J, Lipsitz S, Bates DW, Schnipper JL. Causes and patterns of readmissions in patients with common comorbidities: retrospective cohort study. BMJ (Clinical research ed.). 2013;347:f7171.
4. Shimizu E, Glaspy K, Witt MD, et al. Readmissions at a public safety net hospital. PloS one. 2014;9(3):e91244.
5. Balla U, Malnick S, Schattner A. Early readmissions to the department of medicine as a screening tool for monitoring quality of care problems. Medicine. 2008;87(5):294-300.
6. Yam CH, Wong EL, Chan FW, et al. Avoidable readmission in Hong Kong--system, clinician, patient or social factor? BMC health services research. 2010;10:311.
7. Inspectorate for Public Health: Quality indicators hospitals 2016. Utrecht. http://www.igz.nl/Images/IGZ%20Basisset%20kwaliteitsindicatoren%20ziekenhuizen%202016_tcm294-367407.pdf. Updated September 2015. Accessed December, 2015.
8. Donze J, Aujesky D, Williams D, Schnipper JL. Potentially avoidable 30-day hospital readmissions in medical patients: derivation and validation of a prediction model. JAMA internal medicine. 2013;173(8):632-638.
9. van Walraven C, Wong J, Forster AJ. LACE+ index: extension of a validated index to predict early death or urgent readmission after hospital discharge using administrative data. Open medicine : a peer-reviewed, independent, open-access journal. 2012;6(3):e80-90.
10. Cotter PE, Bhalla VK, Wallis SJ, Biram RW. Predicting readmissions: poor performance of the LACE index in an older UK population. Age and ageing. 2012;41(6):784-789.
11. Cooksley T, Nanayakkara PW, Nickel CH, et al. Readmissions of medical patients: an external validation of two existing prediction scores. QJM : monthly journal of the Association of Physicians. 2015.
12. Spiva L, Hand M, VanBrackle L, McVay F. Validation of a Predictive Model to Identify Patients at High Risk for Hospital Readmission. Journal for healthcare quality : official publication of the National Association for Healthcare Quality. 2015.
13. van Walraven C, Bennett C, Jennings A, Austin PC, Forster AJ. Proportion of hospital readmissions deemed avoidable: a systematic review. CMAJ : Canadian Medical Association journal = journal de l’Association medicale canadienne. 2011;183(7):E391-402.
14. Vest JR, Gamm LD, Oxford BA, Gonzalez MI, Slawson KM. Determinants of preventable readmissions in the United States: a systematic review. Implementation science : IS. 2010;5:88.
15. Maurer PP, Ballmer PE. Hospital readmissions--are they predictable and avoidable? Swiss medical weekly. 2004;134(41-42):606-611.
16. Kwok T, Lau E, Woo J, et al. Hospital readmission among older medical patients in Hong Kong. Journal of the Royal College of Physicians of London. 1999;33(2):153-156.
17. Greysen SR, Stijacic Cenzer I, Auerbach AD, Covinsky KE. Functional impairment and hospital readmission in medicare seniors. JAMA internal medicine. 2015;175(4):559-565.
18. Lindquist LA, Baker DW. Understanding preventable hospital readmissions: masqueraders, markers, and true causal factors. Journal of hospital medicine. 2011;6(2):51-53.
19. Barnett ML, Hsu J, McWilliams JM. Patient Characteristics and Differences in Hospital Readmission Rates. JAMA internal medicine. 2015.
109
20. van Vuuren W SC, Schaaf TW. The development of an incident analysis tool for the medical fiels Eindhoven website. [http://alexandria.tue.nl/repository/books/493452.pdf]. 1997. Accessed March, 2015.
21. Team helps hospital avoid readmission penalties. Hospital case management : the monthly update on hospital-based care planning and critical paths. 2013;21(4):50-51.
22. Antibiotic therapy for acute appendicitis in adults. Fewer immediate complications than with surgery, but more subsequent failures. Prescrire international. 2014;23(150):158-160.
23. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. Journal of chronic diseases. 1987;40(5):373-383.
24. Katz S, Ford AB, Moskowitz RW, Jackson BA, Jaffe MW. STUDIES OF ILLNESS IN THE AGED. THE INDEX OF ADL: A STANDARDIZED MEASURE OF BIOLOGICAL AND PSYCHOSOCIAL FUNCTION. Jama. 1963;185:914-919.
25. Kruizenga HM, de Jonge P, Seidell JC, et al. Are malnourished patients complex patients? Health status and care complexity of malnourished patients detected by the Short Nutritional Assessment Questionnaire (SNAQ). European journal of internal medicine. 2006;17(3):189-194.
26. van Walraven C. The Utility of Unplanned Early Hospital Readmissions as a Health Care Quality Indicator. JAMA internal medicine. 2015:1-2.
27. Feigenbaum P, Neuwirth E, Trowbridge L, et al. Factors contributing to all-cause 30-day readmissions: a structured case series across 18 hospitals. Medical care. 2012;50(7):599-605.
28. Cakir B, Gammon G. Evaluating readmission rates: how can we improve? Southern medical journal. 2010;103(11):1079-1083.
29. van Walraven C, Dhalla IA, Bell C, et al. Derivation and validation of an index to predict early death or unplanned readmission after discharge from hospital to the community. CMAJ : Canadian Medical Association journal = journal de l’Association medicale canadienne. 2010;182(6):551-557.
30. Joynt KE, Jha AK. Thirty-day readmissions--truth and consequences. The New England journal of medicine. 2012;366(15):1366-1369.
31. Bianco A, Mole A, Nobile CG, Di Giuseppe G, Pileggi C, Angelillo IF. Hospital readmission prevalence and analysis of those potentially avoidable in southern Italy. PloS one. 2012;7(11):e48263.
32. van Wagtendonk I, Smits M, Merten H, Heetveld MJ, Wagner C. Nature, causes and consequences of unintended events in surgical units. The British journal of surgery. 2010;97(11):1730-1740.
CHAPTER 7
HOSPITAL READMISSIONS: A RELIABLE QUALITY INDICATOR?
Louise S. van Galen | Prabath W.B. Nanayakkara
Ned Tijdschr Geneeskd 2016;160:A9885
‘Culture eats policy for breakfast‘ Peter Drucker
112
ABSTRACTThe percentage of readmissions within 30 days after discharge is an official quality indicator
for Dutch hospitals in 2016. In this commentary the authors argue why readmissions cannot
be regarded as a reliable way of assessing quality of healthcare in a hospital. To date, policy
makers have been struggling with its precise definition and the indicator has not been properly
formulated yet. It does not distinguish between planned and unplanned readmissions and
does not take into account the ‘preventability’. Therefore the authors believe that the indicator
in its current form might falsely interpret the quality of care of a hospital and it is questionable
to use readmissions as a quality indicator.
113
MAIN TEXT (TRANSLATED FROM DUTCH)In the 2015 cooperation agreement between the Dutch Inspectorate of Healthcare and Dutch
hospitals it was decided that the percentage of unplanned readmissions within 30 days after
discharge of a clinical admission in the same hospital was added as an official quality indicator.1
Readmission, together with already existing quality indicators: unexpected long length of stay
and in-hospital mortality, are regarded as a ‘major adverse events’ and these deem to indicate
the negative results of clinical healthcare. It is known that in an older population mortality
increases if these patients are readmitted frequently.2 For several years, healthcare policy
makers in the USA and the UK have already used the percentage of readmissions as a marker
of quality and hospital safety. A ‘high’ readmission rate often results in financial penalties in
these countries.3,4 In the Netherlands there is hardly any literature concerning readmissions
and policy makers struggle with the realization of a reliable indicator.
Internationally, there seems to be a lack of insight in the predictability and preventability of
readmissions. As a reaction to this problem extensive research has been performed using
retrospective databases to create prediction models which could potentially indicate the risk of
readmission. In these models demographics and severity of disease characteristics are used to
assess if there is a significant difference in the patients who are, and who are not readmitted.
The question rises if this research and composed models truly add to increasing the quality
of care. Knowing a patient with heart-failure will have a higher chance of getting readmitted
does not tell us what the direct cause of this readmission entails, and can therefore not be
used to improve the handling of these patients.
Current literature has shown that none of the developed prediction models are validated
adequately. These models are not yet capable to predict the chance of readmission in an average
hospital population.7 Also, in daily practice, these ‘static’ predictors, are often not useful as
efficient targets aiming to prevent readmission. Recently published work using a PRISMA-
analysis has shown that causes of potentially preventable readmissions are mostly human-
related coordination and communication failures.8 An example is a patient who is readmitted
due to a medication-related complication which occurred because no clear handover was
available for the general physician, another example from this work is a readmission in which
the home situation of the patient was not assessed thoroughly enough during index admission.
In the elaborate international literature, often no distinction is made between preventable
and non-preventable readmissions when describing potential predictors of readmission. Many
readmissions are the cause of the natural course of disease, for example a readmission because
of an epileptic insult caused by brain metastasis of a melanoma.9,10 This seems important
since preventable readmissions significantly increase mortality within six months as compared
to non-preventable readmission. 3 Available research, however, has not yet defined clear risk
114
factors that could predict or prevent such readmissions. Therefore, policy makers still have
minimal tools to reduce preventable readmissions.
Given the before mentioned one must critically ask if ‘readmission’ is a reliable quality indicator
in its current form. The indicator does not distinguish between predictable and non-predictable
readmissions, and more importantly, no clear definition has been formulated yet. This might
result in systemic distortion. If a hospital has a relatively high percentage of non-preventable
readmissions, this does not automatically imply that their quality of care is not up to standards.
This hospital might strive for admissions as short as possible in patients who are in a terminal
stage of their disease. This could improve their quality of life since these patients are able to
spend as much time as home as possible with their loved ones, and therefore does not indicate
improper quality of care but rather the opposite. In addition, some healthcare institutions
struggle with the balance between high burden on available beds and need for readmissions.
In the current health system doctors are under constant pressure to discharge patient rapidly.
Also, the increase of acute medical units in the Netherlands, which increases patient flow and
decrease hospital length of stay might unintentionally increase readmission rate.
In the Netherlands empirical research concerning readmissions that could be prevented in our
healthcare system is lacking. In addition, patients are also increasingly seen as active partners
in healthcare: it seems logical to ask patients and their carers to participate in the judgement of
quality of care in hospitals. In early 2016 CURIOS@HOLLAND, the first prospective readmission
project will commence in several Dutch hospitals. The aim of this study is to assess risk factors
and causes that could potentially predict preventable readmissions by using the perspectives
of the most important participants in the healthcare chain.
CONCLUSIONThe percentage of readmissions within 30 days after discharge in the same hospital is not
a reliable quality indicator to measure quality of care and patient safety of a hospital. Current
research has mainly focused on retrospective analysis of databases to derive predictors,
however readmissions do not seem to be ‘static’ and are not easily framed into figures. In
order to improve quality of care in hospitals problem-orientated interventions are needed
which mainly focus on unplanned potentially preventable readmissions. A quality indicator
should not use the total amount of readmissions, but should solely address the ones which
are preventable. This could provide us with a more honest and goal-directed way to assess
the quality in hospitals. However, the question remains if it wouldn’t be more beneficial to
drop this indicator in total. In our vision, in every hospital, regardless the preventability, it
should be natural to investigate the causes for each individual readmission that has returned
back to hospital unintentionally within 30 days after discharge.
115
REFERENCES1. Basisset Kwaliteitsindicatoren 2016. Utrecht: Inspectie voor de Gezondheidszorg, 2015. Accessed
November, 2015.
2. Zanocchi M, Maero B, Martinelli E, Cerrato F, Corsinovi L, Gonella M, et al. Early re-hospitalization of elderly people discharged from a geriatric ward. Aging Clin Exp Res 2006;18: 63-9. Medline
3. Balla U, Malnick S, Schattner A. Early readmissions to the department of medicine as a screening tool for monitoring quality of care problems. Medicine (Baltimore) 2008;87: 294-300. Medline
4. Greysen SR, Stijacic Cenzer I, Auerbach AD, Covinsky KE. Functional impairment and hospital readmission in Medicare seniors. JAMA Intern Med 2015;175:559-65. Medline
5. van der Ven MJ, Schoon Y, Olde Rikkert MG. Ongeplande heropnames bij kwetsbare ouderen: Retrospectieve analyse van opnames in een academisch ziekenhuis. Ned Tijdschr Geneeskd 2015;159:A9211. Medline
6. Donze J, Aujesky D, Williams D, Schnipper JL. Potentially avoidable 30-day hospital readmissions in medical patients: derivation and validation of a prediction model. JAMA Intern Med 2013;173:632-8. Medline
7. Cooksley T, Nanayakkara PW, Nickel CH, Subbe CP, Kellett J, Kidney R, et al. Readmissions of medical patients: an external validation of two existing prediction scores. QJM 2015 Jul 10. Medline
8. Fluitman K, van Galen L, Merten H, Rombach S, Brabrand M, Cooksley M, et al. Exploring the preventable causes of unplanned readmissions using root cause analysis: Coordination of care is the weakest link. Eur J Int Med IN PRESS
9. Halfon P, Eggli Y, Pretre-Rohrbach I, Meylan D, Marazzi A, Burnand B. Validation of the potentially avoidable hospital readmission rate as a routine indicator of the quality of hospital care. Med Care 2006;44:972-81. Medline
10. Bianco A, Mole A, Nobile CG, Di Giuseppe G, Pileggi C, Angelillo IF. Hospital readmission prevalence and analysis of those potentially avoidable in southern Italy. PLoS One 2012;7:e48263. Medline
CHAPTER 8
PHYSICIAN CONSENSUS ON PREVENTABILITY AND PREDICTABILITY OF READMISSIONS BASED ON
STANDARD CASE SCENARIOS
Louise S. van Galen | Tim Cooksley | Hanneke Merten | Mikkel Brabrand | Caroline B. Terwee | Christian H. Nickel | Christian P. Subbe | Rachel M. Kidney |
John Soong | Louella Vaughan | Immo Weichert | Mark H. Kramer | Prabath W. Nanayakkara |
*On behalf of the safer@home consortium
Neth J Med 2016;74(10):434-42
‘Don’t get too comfortable too change’ Graham Dixon
118
ABSTRACTBackground
Policy makers struggle with unplanned readmissions as a quality indicator since integrating
preventability in such indicators is difficult. Most studies on the preventability of readmissions
questioned physicians whether they consider a given readmission to be preventable, from
which conclusions on factors predicting preventable readmissions were derived. There is no
literature on the interobserver agreement of physician judgement.
Aim
To assess the degree of agreement among physicians regarding predictability and preventability
of medical readmissions.
Design
An online survey based on eight real-life case scenarios was distributed to European physicians.
Methods
Physicians were requested to rate from the first four (index admission) scenarios whether they
expected these patients to be readmitted within 30 days (the predictability). The remaining
four cases, describing a readmission, were used to assess the preventability. The main outcome
was the degree of agreement among physicians determined using intra class correlation
coefficient (ICC).
Results
526 European medical physicians completed the survey. Most physicians had internal medicine
as primary specialism. The median years of clinical experience was 11 year. ICC for predictability
of readmission was 0.67 (moderate to good) and ICC for preventability of readmission was
0.13 (poor).
Conclusion
There was moderate to good agreement among physicians on the predictability of readmissions
while agreement on preventability was poor. This study indicates that assessing preventability of
readmissions based solely on the judgement of physicians is far from perfect. Current literature
on the preventability of readmissions and conclusions derived on the basis of physician opinion
should be interpreted with caution.
Keywords
Patient safety, Quality improvement, Readmissions
119
INTRODUCTIONHospital readmissions within 30 days are of interest to many policy makers internationally.1
They are used as a quality and safety indicator with financial penalties levied in many countries
including the United States and United Kingdom.2 The main problem in using readmissions as
a quality indicator is that the preventability of these readmissions is not properly defined and
integrated in this indicator making it difficult to use as a genuine measure of quality of care.3,4
By not distinguishing between preventable and non-preventable readmissions this indicator
might therefore result in distorted evaluation of hospital care. Furthermore, there is increasing
evidence that the causes of mostly medical readmissions are often multifactorial and usually
the result of natural disease progression, underlying comorbidity or socio-environmental
factors beyond the control of hospital and not solely caused by inadequate hospital care.5-9
The use of readmissions as a quality indicator necessitates that they reflect poor care, are
preventable and that a consensus definition for these two aspects is agreed.
Previous research has not yet been able to determine uniform factors related to preventable
readmissions.10 To date, consensus definition of preventability has not been established.
Many studies use the opinion of physicians as the gold standard to determine if readmissions
are preventable, and derive factors that would predict preventable readmissions from these
findings.11-13 However, to our knowledge, no study has yet been performed to examine
the interobserver reliability of the physicians’ judgement on preventability.
Therefore, we performed an international study to assess if there is any consensus between
physicians regarding the predictability and preventability of medical readmissions.
MATERIAL AND METHODSThis study is an initiative of the safer@home consortium, an international group founded in
2013 consisting of 13 acute medical physicians, emergency physicians and epidemiologists
from Europe that focus on readmissions and safer discharge processes.
During the 3-month study period (1 September to 1 December 2015) a survey on eight
cases based on common clinical scenarios (see appendix 1) was distributed to physicians
throughout Europe.
Survey
The survey consisted of eight case-based medical scenarios (Table 1 shows summary of case
vignettes). The scenarios were generated using a Delphi-type methodology, whereby multiple
scenarios were generated and then represented to the safer@home consortium in two rounds.
In the first round underlying assumptions and information leading to different judgements was
explored using current readmission literature. This round took place in a face-to-face half-yearly
consortium meeting with all 13 members. After this seven clinically active medical physicians
in the group were asked to provide examples from their daily work in order to compose cases.
120
In the second round, these cases were discussed in a conference call during which the cases
that would be representable for all countries were selected through consensus. In addition we
assessed if, in our ‘expert’-opinion, the cases could potentially be used to fulfil the purpose
of our research question. Subsequently, a pilot was performed on a small group of physicians
from all countries to ensure cases were understandable and varied sufficiently. Final case
selection ensured that: a) the cases would be representative of patients requiring unselected
medical admission in northern Europe; b) the scenarios covered the range of factors suggested
by the literature to impact on readmissions; c) cases were not traceable to real-life patients.
The online survey consisted of two parts: 1) Physicians were asked about their opinion on
predictability of medical readmissions; from four cases describing an index admission, physicians
were asked to rate the chance of readmission within 30 days. 2) Physicians were asked to
assess the preventability of four described medical readmission cases. From the physicians’
assessment of predictability and preventability, the degree of consensus could be derived.
For both parts of the survey a five point Likert Scale was used as an answering model ((part
1: Definitely not predictable (1) – Definitely predictable (5); part 2: Definitely not preventable
(1) - Definitely preventable (5)).14
Data on the country and primary specialty of the responding physician filling out the survey
and the number of years of clinical experience were collected in order to explore agreement
within these subgroups. The survey was anonymized to ensure the researchers could not trace
which physician filled out which survey. Finally, general comments concerning readmission
could be made after completing the survey.
Distribution
The survey was distributed among physicians throughout Europe, they all worked solely in
a medical specialty and not in any surgical specialty. Invitations were sent to the members
of the Society for Acute Medicine the UK, the Dutch Acute Medicine Society, the Danish
Society for Emergency Medicine, physicians from Switzerland and Ireland using a common
web-based platform SurveyMonkey®. In order to calculate an accurate response rate, each
physician communicated the number of requests sent to one research member (LG), who was
responsible for data processing and statistical analysis. The ethics committee of VU University
Medical Centre, Amsterdam approved the study. No funding was received for this study.
Statistics
Descriptive characteristics and frequencies were calculated in SPSS version 22.0. Ratings of
physicians are presented as frequencies and percentages. Using the intraclass correlation
(ICC, a reliability coefficient) we assessed the agreement among physicians regarding
the predictability and subsequently, the preventability of the assessed medical readmissions.
This coefficient (ICC) is used to assess the agreement of ratings made by multiple observers
(in our study ‘physicians’) measuring the same outcome (in our study ‘the predictability and
121
preventability of readmissions both based on four-real-life readmission scenarios’). The ICC is
a ratio ranging in value between 0 (representing no agreement) and 1 (implying agreement).
Calculating the variance components we constructed the ICC formulas from which the ICC
could be calculated. For dependent variables we used the outcome ‘Likert scores’ and for
random factors ‘physicians’ and ‘case numbers (1.1, 1.2, 1.3, 1.4 and 2.1, 2.2, 2.3, 2.4)’ were
used. The variance among cases (case numbers1.1, 1.2, 1.3, 1.4 and 2.1, 2.2, 2.3, 2.4) were
analysed separately, among physicians, and the random error were calculated in SPSS using
the VARCOMP procedure. From the variance components we calculated the ICC for absolute
agreement as the variance among cases divided by the total variance of the cases, physicians
and random error.15
RESULTSPhysician characteristics
During the three-month study period (1 September to 1 December 2015) the survey was
distributed to physicians in Europe. In total 526 medical physicians filled out the survey.
The overall response rate was 24.2%. Seventy-seven (14.6%) physicians did not complete
all the questions in the survey. Table 2 shows physician characteristics. Dutch physicians were
the largest group of respondents (46.2%), followed by Danish (25.1%) and physicians from
the United Kingdom (23.6%). Most physicians had internal medicine (33.3%) as their primary
specialty followed by acute medicine (24.5%) and geriatrics (12.5%). The median years of
clinical experience was 10.75 (interquartile range: 5-20).
Agreement on predictability of readmission
For the first part of the survey physicians were asked if they could predict a readmission based
on the four case descriptions of medical index admissions. Responses are shown in figure 1.
The results show that there was substantial variation in the degree of predictability between
the physicians’ judgements in all four cases. The cases were assessed with different degrees
of predictability. To illustrate, for case number 1.2, about half of the physicians assessed
the likelihood of readmission as ‘definitely not’ (score 1), while in case number 2.1 over 60%
of the physicians predicted that the patient will definitely be readmitted (score 5). The ICC for
agreement of predictability was 0.67 (Var(Casenumber) 1,444, Var(Observer) 0.054, Var(error)
0.649) which indicates a moderate to strong interobserver agreement between the raters
(physicians). These findings suggest that the surveyed doctors had a moderate to good degree
of agreement about the patients that were prone to come back, they predicted the same
patients as having a higher chance of a readmission occurring.
Agreement on preventability of readmission
In the second part of the survey the respondents were asked to rate the preventability of
four medical readmission cases. The results in figure 2 show the distribution of answers by
the physicians. It shows that the physicians rated the cases differently, there was a wide variety
in assessment. In all four cases no clear majority seemed to rate the same readmissions with
122
Table 1 | Summary of case vignettes
A. Predictability: Please assess on a scale from 1-5 if you think the following four admissions are followed by a readmission within 30 days: Definitely not (1) - Definitely (5).
Demographics Presenting complaint Diagnosis Investigations Management
Other information
Case 1.1
83 year old female
Collapse Atrial Fibrillation, Hypertension,
Urinary tract infection
Raised inflammatory markers, Positive urine culture
Antibiotics, Aspirin
Cardioverts with sepsis treatment
Case 1.2
20 year old female
Headache Migraine CT brain: normal, Lumbar-puncture: normal
Intra-venous fluids, Paracetamol,
Non-steroidal anti-inflammatory
Case 1.3
60 year old female
Dyspnoea
Productive cough
Infective exacerbation of COPD
Sputum culture negative
Oxygen, Bronchodilators, Steroids, Antibiotics
Has home nebulizers
Case 1.4
94 year old female
Dyspnoea Pneumonia Persistently raised inflammatory markers two days before discharge
Antibiotics Chest pain, dyspnoea, vomiting prior to discharge
B. Preventability: Please assess on a scale from 1 to 5 if you think the following four readmissions within 30 days are: Definitely not preventable (1) - Definitely preventable (5).
DemographicsPresenting complaint Diagnosis Investigation Management
Readmission diagnosis
Case 2.1
63 year old lady
Fever Gemcitabine induced fever
None of note Supportive treatment
Neutropenic sepsis 10 days later
Case 2.2
40 year old male
Ascites Childs B cirrhosis
Alcohol Dependency
None of note Abdominal paracentesis,
Diuretics, vitamins, lactulose,
Alcohol support declined
Upper Gastro-intestinal bleed 3 weeks later
Case 2.3
55 year old male
Chest pain Anterior ST-elevation Myocardial infarction
Angiogram, Echo-cardiogram with moderate left ventricular dysfunction
Angioplasty of the left anterior descending artery, Secondary prevention
Pulmonary oedema 3 weeks later
Case 2.4
32 year old female
Loin pain Pyelonephritis, Hydronephrosis due to ureter stenosis
Ultra-sound abdomen
Urine culture
Intra-venous antibiotics as outpatient
Pyelonephritis one month later
123
similar scores. These findings were also reflected in the ICC for this part of the survey. The ICC
was calculated at 0.13 (Var(Casenumber) 0,194, Var(Observer) 0.168, Var(error) 1.076), which
implies poor agreement. Doctors do not seem to agree on the preventability of readmissions.
However, one must note that the variance among case numbers was relatively low which
may indicate that the cases assessed were not sufficiently distinct enough to obtain a high
reliability coefficient.
Subgroup analysis
To assess if there was any difference in agreement between subgroups of physicians
we subdivided the doctors into years of clinical experience. They were grouped based on
clinical experience up to 5 years (n=151, 28,7%), from 5-15 years (n=208, 39,5%), and 15
years and higher (n=167, 31,7%). Results suggest that medical physicians with less clinical
experience had a trend towards greater agreement than those with more clinical experience
as to the likelihood of readmission but these differences are minimal (ICC 0.70, 0.69, 0.63,
respectively). Physicians with more clinical experience seemed to have more agreement
about the preventability of a readmission compared with those with less clinical experience
(ICC 0.08, 0.01, 0.19, respectively).
Table 2 | Physician characteristics
Country Percentage 100% (n=526) Primary specialty
Percentage 100% (n=526)
The Netherlands 46,2 (243) Internal Medicine 33,3 (175)
Denmark 25,1 (132) Acute/Emergency Medicine 24,5 (129)
United Kingdom 23,6 (124) Geriatrics 12,5 (66)
Switzerland 1,9 (10) Other 5,3 (28)
Other 3,2 (17) Nephrology 4,2 (22)
Intensive Care 3,6 (19)
Endocrinology 3,2 (17)
Gastroenterology 2,9 (15)
Pulmonary medicine 2,3 (12)
Haematology 1,7 (9)
Medical Oncology 1,5 (8)
Rheumatology 1,3 (7)
Hepatology 0,8 (4)
Cardiology 0,4 (2)
124
Figure 1 | Part I: The predictability of readmissionSpread of Likert-Scores (Definitely not predictable (1) – Definitely predictable (5)) given per case (1,2,3,4) by 526 physicians (in percentage)
Figure 2 | Part II: The preventability of readmissionSpread of Likert-Scores (Definitely not preventable (1) – Definitely preventable (5)) given per case (5,6,7,8) by 526 physicians (in percentage)
125
DISCUSSIONIn this survey among 526 European physicians, there was moderate agreement as to
the predictability of medical readmissions but poor agreement about their preventability. These
results suggest that doctors agree on the patients who have a higher risk of being readmitted,
but the physicians differ on how preventable these readmissions are. To our knowledge,
the current study is the first to investigate the interobserver reliability on the evaluation of
unplanned readmissions in such a large group of observers.
Unplanned readmissions are a complex phenomenon, which are influenced not only by medical
factors but also by a range of social and political issues.5,16,17 Readmission risk is difficult to
define and is less predictable than mortality.18 Nevertheless, there are a number of risk factors
which are recognized as increasing the risk of readmission and multiple predictive scoring
systems based on these factors have been designed.5,6,19
Although the risk factors for predominantly medical readmission are increasingly well
recognized, the dynamic of how they interact and whether they can be influenced remains
controversial.20 The poor consensus among physicians found in our study as to whether
the readmissions were preventable underlines this issue. A US study of 17 hospitalists reviewing
300 consecutive readmissions also found wide variation in their scoring of preventability,
however comparability might be limited since these were real-life readmissions.21 We can
concur their findings of interobserver variability in a (European) setting.
The above findings illustrate the problem faced by policy makers trying to integrate preventability
in the readmission indicator since doctors, who are supposed to be experts in the field, cannot
even agree on the readmissions that are potentially preventable. Current literature, however,
often uses the opinion of one or more physicians as the gold standard to get insight into
preventability and draw conclusions on factors predicting preventability. The results in this
study, however, demonstrate that the assumptions derived from these studies might lead to
misperception since physicians do not share similar ideas on the potential preventability of
readmissions.22-24 Hence, it can be questioned whether conclusions drawn from these studies
might not provide reliable conclusions to create an appropriate quality indicator.
Readmitted medical patients are a heterogeneous group; there is a wide variation in the age,
comorbidities and social support of these patients. It remains unclear as to whether the factors
which drive unplanned readmission, including medical, social, cultural and environmental, are
modifiable.20 This is reflected by an increasing body of evidence that suggests readmissions do
not always reflect poor care and preventability of these readmissions is poorly defined.3,4,21,25
More research studying ‘the preventability’ in a structured manner might help to improve
the difficult task in creating a reliable indicator.
We used adapted real-life case scenarios in our study, which may be a limitation. This was also
reflected in the comments section, where physicians mentioned they were missing information
126
that would allow them to thoroughly assess the case, for example more details on the patients’
social situation. It would however be difficult to incorporate all the potentially relevant social
and environmental factors into scenarios particularly in a pan-European study where there
is a wide variety of political and health policies that influence readmissions. Furthermore,
in calculating the ICC for the preventability part of the survey one could suggest that there
was little variation in the preventability of the cases. This may reflect either that there was
insufficient variation with regards to preventability within the scenarios, potentially caused
by balancing between uniformity in the cases in a way they could be representable for all
countries participating in the study and enough variation in the cases in order to create
different opinions per case. It may also reflect an uncertainty among physicians regarding
what comprises a preventable admission.
On a final note, our respondents were of high seniority with a median of 11 years of clinical
experience. If clinicians with this level of experience cannot agree on the predictability of
readmission, is it wise to use it as a marker of quality of care?
CONCLUSIONThis study demonstrates that there is moderate agreement among experienced medical
physicians about the predictability of readmissions but poor agreement about their
preventability. Therefore, the conclusions derived from earlier studies on preventability, on
the basis of physician consensus as the gold standard, are questionable. Hence, a good way of
defining and integrating preventability into this quality indicator remains elusive.
ACKNOWLEDGEMENTSA full list of membership of the safer@home Consortium is as follows: M. Braband, T. Cooksley,
L. van Galen, H. Haak, R. Kidney, J. Kellet, H. Merten, P. Nanayakkara, C. Nickel, J. Soong,
C. Subbe, L. Vaughan, I. Weichert.
CONFLICT OF INTEREST STATEMENTThe authors have nothing to disclose.
127
REFERENCES1. Fischer C, Lingsma HF, Marang-van de Mheen PJ, Kringos DS, Klazinga NS, Steyerberg EW. Is
the readmission rate a valid quality indicator? A review of the evidence. PLoS One. 2014;9(11):e112282.
2. Inspectie van de volksgezondheid: Basisset Kwailiteitsindicatoren 2016. Utrecht. http://www.igz.nl/Images/IGZ%20Basisset%20kwaliteitsindicatoren%20ziekenhuizen%202016_tcm294-367407.pdf. Laatste update september 2015. Accessed July, 2016.
3. van Galen LS, Nanayakkara PW. [Hospital readmissions: A reliable quality indicator?]. Ned Tijdschr Geneeskd. 2015;160:A9885.
4. van Walraven C. The Utility of Unplanned Early Hospital Readmissions as a Health Care Quality Indicator. JAMA Intern Med. 2015;175(11):1812-1814.
5. Donze J, Aujesky D, Williams D, Schnipper JL. Potentially avoidable 30-day hospital readmissions in medical patients: derivation and validation of a prediction model. JAMA Intern Med. 2013;173(8):632-638.
6. Cooksley T, Merten H, Kellett J, et al. PRISMA Analysis of 30 Day Readmissions to a Tertiary Cancer Hospital. Acute Med. 2015;14(2):53-56.
7. Rico F, Liu Y, Martinez DA, Huang S, Zayas-Castro JL, Fabri PJ. Preventable Readmission Risk Factors for Patients With Chronic Conditions. J Healthc Qual. 2015.
8. Vest JR, Gamm LD, Oxford BA, Gonzalez MI, Slawson KM. Determinants of preventable readmissions in the United States: a systematic review. Implement Sci. 2010;5:88.
9. Anderson MA, Helms LB, Hanson KS, DeVilder NW. Unplanned hospital readmissions: a home care perspective. Nurs Res. 1999;48(6):299-307.
10. Jackson AH, Fireman E, Feigenbaum P, Neuwirth E, Kipnis P, Bellows J. Manual and automated methods for identifying potentially preventable readmissions: a comparison in a large healthcare system. BMC Med Inform Decis Mak. 2014;14:28.
11. van Walraven C, Jennings A, Forster AJ. A meta-analysis of hospital 30-day avoidable readmission rates. J Eval Clin Pract. 2012;18(6):1211-1218.
12. Cakir B, Gammon G. Evaluating readmission rates: how can we improve? South Med J. 2010;103(11):1079-1083.
13. Meisenberg BR, Hahn E, Binner M, et al. ReCAP: Insights Into the Potential Preventability of Oncology Readmissions. J Oncol Pract. 2016;12(2):153-154.
14. Elaine IE, Seaman CA. Likert scales and data analyses. Quality Progress 40.7 2007: 64-65.
15. de Vet HC, Terwee CB, Knol DL, Bouter LM. When to use agreement versus reliability measures. J Clin Epidemiol. 2006;59(10):1033-1039.
16. Billings J, Blunt I, Steventon A, Georghiou T, Lewis G, Bardsley M. Development of a predictive model to identify inpatients at risk of re-admission within 30 days of discharge (PARR-30). BMJ Open. 2012;2(4).
17. Fluitman KS, van Galen LS, Merten H, et al. Exploring the preventable causes of unplanned readmissions using root cause analysis: Coordination of care is the weakest link. Eur J Intern Med. 2016.
18. Kansagara D, Englander H, Salanitro A, et al. Risk prediction models for hospital readmission: a systematic review. Jama. 2011;306(15):1688-1698.
19. van Walraven C, Bennett C, Jennings A, Austin PC, Forster AJ. Proportion of hospital readmissions deemed avoidable: a systematic review. CMAJ. 2011;183(7):E391-402.
20. Brown EG, Burgess D, Li CS, Canter RJ, Bold RJ. Hospital readmissions: necessary evil or preventable target for quality improvement. Ann Surg. 2014;260(4):583-589; discussion 589-591.
21. Koekkoek D, Bayley KB, Brown A, Rustvold DL. Hospitalists assess the causes of early hospital readmissions. J Hosp Med. 2011;6(7):383-388.
22. Bianco A, Mole A, Nobile CG, Di Giuseppe G, Pileggi C, Angelillo IF. Hospital readmission prevalence and analysis of those potentially avoidable in southern Italy. PLoS One. 2012;7(11):e48263.
128
23. Shimizu E, Glaspy K, Witt MD, et al. Readmissions at a public safety net hospital. PLoS One. 2014;9(3):e91244.
24. Donze J, Lipsitz S, Bates DW, Schnipper JL. Causes and patterns of readmissions in patients with common comorbidities: retrospective cohort study. Bmj. 2013;347:f7171.
25. Cooksley T, Nanayakkara PW, Nickel CH, et al. Readmissions of medical patients: an external validation of two existing prediction scores. QJM. 2016;109(4):245-248.
CHAPTER 9
PATIENTS’ AND PROVIDERS’ PERCEPTIONS OF THE PREVENTABILITY OF HOSPITAL READMISSION:
A PROSPECTIVE, OBSERVATIONAL STUDY IN 4 EUROPEAN COUNTRIES
“CAPTURING READMISSIONS INTERNATIONALLY TO PREVENT READMISSION BY THE SAFER@HOME
GROUP (CURIOS@)”
Louise S. van Galen | Mikkel Brabrand | Tim Cooksley | Peter M. van de Ven | Hanneke Merten | Ralph K. So | Loes van Hooff | Harm Haak | Rachel M. Kidney |
Christian H. Nickel | John Soong | Immo Weichert | Mark H. Kramer | Christian P. Subbe | Prabath W.B. Nanayakkara |
*On behalf of the safer@home consortium
BMJ Qual Saf. 2017; In press
‘Patient safety is a dissatisfier, safe care is normal, non-safe care makes unsatisfied’ Jan Klein
132
ABSTRACT Background
Unscheduled readmissions are increasingly being used as a quality indicator. In comparison to
extensive US data, no large-scale European data are available.
Objectives
To investigate opinions of readmitted patients, their carers, nurses and physicians on
predictability and preventability and using majority opinion to determine contributing factors
that could potentially foresee (preventable) readmissions.
Methods
This prospective observational study included medical patients readmitted unscheduled within
30 days in 15 European hospitals. Readmitted patients, carers and treating professionals
were surveyed to assess the discharge process and predictability and preventability of
the readmission. Variables associated with risk of readmission were also collected. Cohen’s
Kappa measured pairwise agreement of considering readmission as predictable/preventable
by patients, carers and professionals. Subsequently, multivariable logistic regression identified
risk factors associated with predictability and preventability.
Results
In this cohort (1398 readmissions), the majority deemed 27.8% readmissions potentially
predictable and 14.4% potentially preventable. Consensus on predictability and preventability
was poor, especially between patients and professionals (kappa’s ranged from 0.105-
0.173). The interviewed selected different factors potentially associated with predictability
and preventability. When a patient reported that he was ready for discharge during index
admission, readmission was deemed less likely by majority (predictability OR 0.55;CI 0.40-
0.75, preventability OR 0.35;CI 0.24-0.49).
Conclusions
There is no consensus between readmitted patients, their carers, treating nurses and physicians
about predictability and preventability of readmissions, nor associated risk factors. The patient
reporting not feeling ready for discharge was strongly associated with preventability and
predictability. Therefore, healthcare workers should question patient’s readiness to go home
timely before discharge.
Keywords
Readmission, Healthcare quality indicator, Patient involvement, Patient
discharge, Communication
133
INTRODUCTIONUnplanned readmissions within 30 days are perceived as a major adverse event after hospital
discharge. Current readmission rates vary from 10-30% internationally and the related costs
are increasing.1 Internationally, readmission rates are often used to audit and reimburse units
and used as a quality of care monitoring tool.2,3 In the Netherlands, readmission rates were
introduced as an official quality indicator in 2016.4
Due to an aging population and healthcare policy changes, Emergency Department (ED)
admissions have risen in the last few years.5,6 As a result, hospital beds are becoming scarce
and physicians are under constant pressure to discharge patients rapidly.
The problem with the use of readmissions as a quality indicator is that not all readmissions
are preventable and causes for readmission might find their origin in natural progression or
unavoidable recurrence of underlying diseases.7,8 In addition, despite many efforts to reduce
readmissions such as the Hospital Readmission Reduction Plan and telephonic follow-up
interventions, it is still questionable whether these interventions truly result in lower
readmission rates.5-8 Moreover, since preventability has not been defined uniformly, it remains
uncertain whether this quality indicator, in its current form, is reliable.9,10 Several studies have
used the opinion of physicians as the gold standard to determine whether readmissions are
preventable, and have defined factors that would predict preventability from these findings.11-13
However, recent work has shown that even among doctors there is poor consensus about
the preventability of a readmission.14
Although many attempts have been made to create readmission prediction models, there
are still no widely utilized and internationally validated tools predicting the chance of
readmission.15-17 Most risk models were developed in the USA and Canada, and due to
differences in healthcare systems and case-mix they probably may not be suitable to be used
in European populations.18-20 In addition, they do not measure preventability. A recent analysis
demonstrated that causes of potentially preventable readmissions are mostly human-related
coordination and communication failures.21 A few studies have been performed investigating
opinions of patients and health-care workers on the preventability of readmissions and
discharge planning.22,23 Yet, most of these studies had small sample sizes, and questioned
patients retrospectively after discharge.
Therefore, we performed the first prospective observational study of 1398 unscheduled
medical readmissions to 15 centres in 4 European countries aiming to investigate:
(1) the opinions of readmitted patients, their carers, nurses and physicians on the predictability
and preventability of the readmissions;
134
(2) the contributing factors that could potentially predict (preventable) readmissions using
the majority opinion as the gold standard.
METHODSSites and Participants
This research project ‘CURIOS@’ (CaptUring Readmissions InternatiOnally to prevent
readmissions by safer@home consortium) took place in academic and non-academic centres
using the network of the safer@home (‘Scoring Acute admissions For Estimating Readmission’)
consortium. The study was an initiative of the collaboration founded in 2014 consisting of 12
acute medical physicians, emergency physicians and epidemiologists from Europe which focus
on readmissions and safer discharge processes. Fifteen centres participated (nine centres in
the Netherlands, three in the United Kingdom, one in Ireland, and two in Denmark). The data
collected was derived from two clinical episodes: readmission (RA) and index admission (IA)
between 1 January 2016 and 1 November 2016. Data collection took place on the unit of
readmission at any time within 72 hours after the readmission. Ethics Committees of all
participating sites individually approved the study, after primary approval was obtained by
the coordinating centre (VUmc, Amsterdam).
Eligible patients were those who had unscheduled admissions through ED, AMU or other
clinical areas, were aged above 18 years and readmitted to hospital following a previous
inpatient episode to any clinical specialty for a minimum of one night in the previous 30 days.
If a patient was readmitted more than once, only first readmission was included. In order to
be included, readmissions had to be to a medical (cardiology, geriatrics, gastroenterology,
haematology, internal medicine, nephrology, neurology, oncology, pulmonary medicine,
rheumatology) ward. Excluded were patients: readmitted electively for procedures, surgery or
chemotherapy; with an IA for psychiatry or gynaecology; who stayed shorter than one night
during IA or RA; transferred to the ward from an initial admission to the ICU; admitted to
another institution in their IA. All patients gave written informed consent.
Data Collection
After instructions and training from the coordinating investigator (LG), centres were allowed to
participate, aiming to include at least 50 patients consecutively. Investigating site researchers
were deemed competent once they followed adequate training and enough time for them
was allocated to perform interviews, data collection and root cause analysis. They were all
medically trained but not involved in patientcare during inclusion period. After obtaining
written informed consent, site researchers surveyed readmitted patients: the questionnaire
consisted of seven questions about their readiness for discharge during IA and predictability
and preventability of their readmission (Supplement 1). This questionnaire was constructed
using available literature and after reaching consensus among safer@home group members.
In addition, we ensured it consisted of questions that could easily be asked by physicians
and answered by patients in daily practice, and are reproducible to be used in other settings.
135
The questionnaires were tested in a pilot group before agreeing on a final amended version.
Subsequently, after obtaining patient’s permission, a carer (defined as a person providing
unpaid intensive and long-term care because of a personal relationship) if available, was
approached in person or by telephone, to answer two questions about predictability and
preventability of the readmission. Lastly, a doctor and a nurse responsible for the patient during
readmission were interviewed. To reduce bias, all interviews were performed separately by site
researcher. This researcher also answered the same two questions after assessing answers
given by the interviewees. In addition, data-items pre-identified through literature as being
potentially predictive of a (preventable) readmission were collected.15,16 These variables (i.e.
Charlson Comorbidity Index24 and Clinical Frailty Scale25) were extracted during readmission,
directly from patient, using clinical notes of IA and discharge communication. The dataset
contained no patient identifiable variables.
PRISMA-analysis
To assess predictability and preventability, we asked all interviewed about the reasons for
readmission. Subsequently, site researcher qualified these into one or more root causes
categorized in disease-, patient-, healthcare worker-, organizational- or other causes, originally
identified by PRISMA-analysis supplying us with more information about probable root causes
for the readmission. PRISMA-analysis has previously shown to provide objective and structured
insight into causes for adverse events, by composing root causal trees for adverse events.21,26
This method has been accepted by the World Alliance for Patient Safety (see Supplement 2 for
more detailed description).27
Measurement of Predictability and Preventability
Since a gold standard defining predictability and preventability is not available, after reaching
consensus in our half-yearly consortium meetings a new variable was composed. It was
decided that if a majority (50 percent or more) of interviewed groups (patients, carers, doctors,
nurses, researchers), assessed the readmission as predictable or preventable (options yes, no,
unknown), the readmission was decided as predictable or preventable. For example, if a carer
was not available, if 2 out of 4 remaining interviewed assessed readmission as preventable, it
was deemed preventable. If all 5 interviewees answered the questions, 3 ‘yes’ answers were
needed for this conclusion. In a separate analysis we used the answers given by the members
of the 5 interviewed groups separately and regarded option ‘unknown’ as missing.
Statistical Analysis
Statistical analysis was performed in SPSS version 22.0. Categorical variables are summarized
as frequencies and percentages. Continuous variables are summarized by mean and standard
deviation (SD) in case of a normal distribution or median and ranges otherwise.
Cohen’s kappa (κ) was used to measure agreement of predictability and preventability
assessments (yes, no, unknown) separately for each pair of five interviewed groups. McNemar
136
Bowker test was performed to assess whether different pairs of interviewed groups varied in
proportions of ‘yes’, ‘no’ and ‘unknown’ answers.
Logistic regression analyses were used to find variables that were associated with assessments
of preventability and predictability of readmissions. For all categorical variables where
‘unknown’ or ‘don’t know’ was an option, these were considered as “missing”. However,
these constituted less than 10% of the total answers. For predictability and preventability
assessed by the individual groups, the option ‘don’t know’ was considered as missing in
multivariable models, regardless of its percentage. (Table 1 and 2). Separate models were built
for each interviewed group and for majority judgement and separate for preventability and
predictability. Only variables with a two-sided P ≤ .10 in univariable logistic regression analyses
were included in a multivariable logistic regression analyses where backward elimination
was used to find a minimum set of variables that were independently associated with
predictability and preventability. To account for differences between countries, country was
included as a predictor in all models. The magnitude of the association between predictors and
outcome was quantified using Odds Ratios (OR) together with their 95% confidence interval.
The discriminative ability was quantified by means of the Area under the Receiver Operator
Curve (ROC) curve.
RESULTSPatient and hospital and readmission characteristics
During the study period 1961 patients were eligible of which 1398 patients participated,
resulting in an inclusion rate of 71.3%. The reasons for exclusion were: patient was already
sent home on day of intended inclusion, 41.0% (231 of 563); unwilling to participate, 20.4%
(115 of 563); being too ill, 19.2% (108 of 563); language barrier, 8.6%, (48 of 563); patient
deceased at RA, 2.1% (12 of 563); and other reasons i.e. being in quarantine, 8.7% (49 of
563). The median age was 70 (range 18-96), the median number of included patients per
hospital was 71 (range 48-226), other characteristics are given in Table 1.
The majority of interviewed deemed 27.8% (390 of 1398) of the readmissions as a potentially
predictable, and 14.4% (202 of 1398) potentially preventable. The assessment per interviewed
is listed in Table 2.
Consensus on readmission (κ)
Table 3 shows Cohen’s kappa for the two questions put forward to all individuals
interviewed: 1. ‘Do you feel the current readmission was expected?’, and 2. ‘Do you feel
the current readmission was preventable?’ (yes, no, don’t know). For predictability, none of
the kappa’s were satisfactory, they were all below κ = 0.7. The poorest consensus was found
between patient and physician, and patient and nurse, with a rate of κ = 0.173 and κ =
0.153, respectively. The highest kappa was found for physician and researcher (κ = 0.607).
The consensus on preventability of readmission was also unsatisfactory, the highest score
137
in this analysis was κ = 0.473, measuring agreement between researcher and physician.
The poorest scores in this group, κ = 0.105 and κ = 0.135, were found comparing patient to
physician, and nurse, respectively. McNemar Bowker test was P < .05 for all, indicating that
proportions of respondents answering ‘yes’, ‘no’ and ‘don’t know’ differed between all pairs of
interviewed groups.
Contributing factors to assessing the predictability and preventability of readmission
Using multivariate models, factors potentially contributing to predictability and preventability
of readmission were identified according to the opinion of the majority. Subsequently, models
were composed per interviewed group. The variables used in the models (Table 1) associated
with assessing readmission as potentially predictable and preventable, were different for every
group interviewed (Table 4 and 5). Using the opinion of the majority as the gold standard
(Table 4), factors significantly associated with a higher predictability of readmission included:
Having a non-elective IA (OR 2.55; 95% CI 1.59-4.08), having more than five admissions in
the year before IA (OR 2.69; 95% CI 1.72-4.20), lower age (OR 0.98; 95% CI 0.97-0.99),
higher CFS (OR 1.29; 95% CI 1.18-1.42), and a higher CCI (OR 1.08; 95% CI 1.02-1.15). When
a patient reported having felt ready at discharge during IA, the readmission was deemed less
likely (OR 0.55; 95% CI 0.40-0.75). Having a follow-up planned (OR 0.52; 95% CI 0.35-0.78),
and feeling ready for discharge (OR 0.35; 95% CI 0.24-0.49) were significantly associated with
a readmissions being deemed less preventable. As is illustrated in table 5, with exception of
the physician, all interviewed considered readmissions more predictable and preventable when
a patient reported not feeling ready for discharge. Table 4 and 5 demonstrate the discriminative
ability of the twelve models by means of the AUC varied from moderate to good (0.65–0.74).
Root causes
All interviewed were asked to qualify reasons for readmission into one or more of five available
root causes. Each readmission could have more than one root cause per interviewed and
the maximum causes per patient was 25. In total 6895 root causes were identified, with
a mean of 5.3 per patient (SD 1.9). Most root causes were disease-related 67.9% (4686 of
6895), followed by healthcare worker-related 17.5% (1208 of 6895), patient-related causes
10.9% (749 of 6895), organizational 2.2% (153 of 6895) and non-classifiable 1.4% (99 of
6895). In univariate analysis, when the composed variable for majority opinion was used,
disease-related root causes were negatively associated with a readmissions being considered
predictable [83.7% vs. 71.8%] (OR 0.5; 95% CI 0.37-0.66) and preventable [87.6% vs.
38.0%] (OR 0.09; 95% CI 0.06-0.12), healthcare worker-related root causes were positively
associated with readmissions being more predictable [11.9% vs. 20.6%] (OR 1.92; 95%
CI 1.39-2.65) and preventable [5.6% vs. 66.3%] (OR 33.43; 95% CI 22.48-49.727).
138
Tab
le 1
| ‘P
atie
nt c
hara
cter
istic
s’ab
Ch
arac
teri
stic
Tota
l (n
=13
98)
Pred
icta
ble
maj
ori
ty
Tota
l (n
=13
98)
P V
alu
e fo
r m
ajo
rity
no
n-
Pred
icta
ble
vs
Pred
icta
ble
Prev
enta
ble
maj
ori
ty
Tota
l (n
=13
98)
P V
alu
e fo
r m
ajo
rity
no
n-
Prev
enta
ble
vs
Prev
enta
ble
No
n-P
red
icta
ble
(n=
1008
)Pr
edic
tab
le
(n=
390)
No
n-P
reve
nta
ble
(n=
1196
)Pr
even
tab
le
(n=
202)
Cou
ntry
, N
o. (
%)
Th
e N
ethe
rlan
ds
911
(65.
2)67
9 (6
7.4)
232
(59.
5)P
< 0
.001
785
(65.
6)12
6 (6
2.4)
P <
0.0
01
D
enm
ark
262
(18.
7)19
1 (1
8.9)
71 (
18.2
)24
2 (2
0.2)
20 (
9.9)
U
nite
d K
ingd
om a
nd Ir
elan
d22
5 (1
6.1)
225
(16.
1)87
(22
.3)
169
(14.
1)56
(27
.7)
Pati
ent
age,
med
ian
(ran
ge),
y
70 (
18-9
6)70
(18
-96)
69 (
18-9
6)P
= 0
.03
70 (
18-9
6)70
.5 (
18-9
6)P
= 0
.62
Mal
e se
x, N
o. (
%)
737
(52.
7)53
8 (5
3.4)
199
(51.
0)P
= 0
.43
627
(52.
4)11
0 (5
4.5)
P =
0.5
9
Part
nere
d/M
arri
ed,
No.
(%
)82
8 (5
9.2)
61.3
(61
.1)
53.
8 (4
8.0)
P =
0.0
1173
1 (6
1.1)
97 (
48.0
)P
< 0
.001
IA t
ypec ,
Non
-ele
ctiv
e, N
o. (
%)
1195
(85
.2)
830
(82.
3)36
1 (9
2.6)
P <
0.0
0183
0 (8
2.3)
361
(92.
6)P
= 0
.21
LOS
IAc ,
med
ian
(ran
ge),
d4
(0-1
91)
4 (0
-136
)4
(0-1
91)
P <
0.0
014
(1-1
91)
4 (1
-136
)P
= 0
.94
Leng
th b
etw
een
IA a
nd R
Ac ,
med
ian
(ran
ge),
d9
(0-3
0)9
(0-3
0)10
(0-
30)
P =
0.0
329
(0-3
0)7
(0-3
0)P
= 0
.007
Clin
ical
Fra
ilty
Scal
e, m
ean
(SD
)25
4.4
(1.6
)4.
2 (1
.6)
4.9
(1.7
)P
< 0
.001
4.4
(1.6
)4.
5 (1
.6)
P =
0.3
3
Cha
rlso
n C
omor
bidi
ty In
dex,
m
edia
n (r
ange
)24
2 (0
-11)
4 (0
-136
)4
(0-1
91)
P <
0.0
012
(0-1
1)2.
5 (0
-11)
P =
0.6
8
HO
SPIT
AL
Scor
e16,d,
mea
n (S
D)
4.4
(2.1
)4.
2 (2
.0)
5.0
(2.2
)P
< 0
.001
4.4
(2.1
)4.
6 (2
.2)
P =
0.0
8
Num
ber
of m
edic
ine
at
disc
harg
e (c
hron
ic a
nd
non-
chro
nic
use)
, N
o. (
%)
0-
425
7 (1
8.4)
201
(20.
0)56
(14
.4)
P =
0.0
0122
2 (1
8.6)
35 (
17.3
)P
= 0
.062
5-
8 39
5 (2
8.3)
299
(29.
7)96
(24
.7)
342
(28.
7)53
(26
.2)
>
874
3 (5
3.3)
506
(50.
3)23
7 (6
0.9)
629
(52.
7)11
4 (5
6.4)
DN
R po
licy
at d
isch
arge
, D
o re
susc
itat
e, N
o. (
%)
961
(73.
1)71
2 (7
4.9)
249
(68.
8)P
= 0
.022
821
(73.
1)14
0 (7
3.3)
P =
0.9
6
ICU
/CC
U/H
DU
adm
issi
on d
urin
g IA
, Ye
s, N
o (%
)11
5 (8
.2)
94 (
9.4)
21 (
5.4)
P =
0.0
1510
2 (8
.5)
13 (
6.5)
P =
0.3
2
139
Tab
le 1
| (c
ontin
ued)
Ch
arac
teri
stic
Tota
l (n
=13
98)
Pred
icta
ble
maj
ori
ty
Tota
l (n
=13
98)
P V
alu
e fo
r m
ajo
rity
no
n-
Pred
icta
ble
vs
Pred
icta
ble
Prev
enta
ble
maj
ori
ty
Tota
l (n
=13
98)
P V
alu
e fo
r m
ajo
rity
no
n-
Prev
enta
ble
vs
Prev
enta
ble
No
n-P
red
icta
ble
(n=
1008
)Pr
edic
tab
le
(n=
390)
No
n-P
reve
nta
ble
(n=
1196
)Pr
even
tab
le
(n=
202)
Cog
niti
ve im
pair
men
t (o
ffici
al
diag
nosi
s)e ,
Yes
, N
o. (
%)
41 (
3.0)
23 (
2.3)
18 (
4.7)
P =
0.0
230
(2.
5)11
(5.
6)P
= 0
.02
Falle
n at
leas
t on
ce in
last
hal
f ye
ar,
Yes,
No.
(%
)38
0 (2
8.9)
279
(29.
6)10
1 (2
7.3)
P =
0.4
230
9 (2
7.6)
71 (
36.2
)P
= 0
.014
Tota
l adm
issi
ons
year
pri
or t
o IA
, el
ecti
ve a
nd n
on-e
lect
ive,
N
o. (
%)
0
507
(36.
3)40
4 (4
0.1)
103
(26.
4)P
< 0
.001
440
(36.
8)67
(33
.2)
P =
0.6
1
1-
574
5 (5
3.3)
534
(53.
0)21
1 (5
4.1)
632
(52.
8)11
3 (5
5.9)
>
514
6 (1
0.4)
70 (
6.9)
76 (
19.5
) 12
4 (1
0.4)
22 (
10.9
)
Dis
char
ge le
tter
issu
ed a
t ti
me
RA,
Yes,
No.
(%
)10
45 (
75.8
)75
9 (7
6.5)
286
(74.
1)
P =
0.3
576
.9 (
905)
140
(69.
7)P
= 0
.027
Follo
w-u
p pl
anne
d at
dis
char
ge
IA,
Yes,
No.
(%
)11
34 (
84.1
)82
4 (8
5.0)
310
(81.
6)P
= 0
.12
992
(86.
1)14
2 (7
2.1)
P <
0.0
01
Has
pat
ient
bee
n se
en b
y a
doct
or b
etw
een
IA a
nd R
A?,
Ye
s, N
o. (
%)
867
(66.
8)61
5 (6
6.1)
252
(68.
9)P
= 0
.34
735
(66.
6)13
2 (6
8.4)
P =
0.6
2
Was
a s
umm
ary
prov
ided
at
disc
harg
e? Y
es,
No.
(%
)99
4 (7
2.6)
720
(78.
5)27
4 (7
9.0)
P =
0.8
686
0 (
80.2
)13
4 (6
9.8)
P =
0.0
01
Did
you
(th
e pa
tien
t) s
ugge
st
doct
ors
to s
tay
long
er a
t IA
? Ye
s, N
o. (
%)
131
(9.4
)74
(7.
4)57
(14
.9)
P <
0.0
0110
3 (8
.7)
28 (
13.9
)P
< 0
.001
140
Tab
le 1
| (c
ontin
ued)
Ch
arac
teri
stic
Tota
l (n
=13
98)
Pred
icta
ble
maj
ori
ty
Tota
l (n
=13
98)
P V
alu
e fo
r m
ajo
rity
no
n-
Pred
icta
ble
vs
Pred
icta
ble
Prev
enta
ble
maj
ori
ty
Tota
l (n
=13
98)
P V
alu
e fo
r m
ajo
rity
no
n-
Prev
enta
ble
vs
Prev
enta
ble
No
n-P
red
icta
ble
(n=
1008
)Pr
edic
tab
le
(n=
390)
No
n-P
reve
nta
ble
(n=
1196
)Pr
even
tab
le
(n=
202)
Did
you
(th
e pa
tien
t) f
eel b
ette
r at
dis
char
ge (
IA)?
Yes
, N
o. (
%)
950
(69.
3)70
7 (7
1.9)
243
(63.
3)P
= 0
.002
839
(71.
9)11
1 (5
5.5)
P <
0.0
01
Did
you
(th
e pa
tien
t) f
eel r
eady
at
dis
char
ge (
IA)?
Yes
, N
o. (
%)
1014
(75
.1)
772
(78.
9)24
2 (6
3.9)
P <
0.0
0190
0 (7
7.7)
114
(57.
3)
P =
0.0
02
a Mis
sing
dat
a in
clud
e th
e fo
llow
ing:
did
you
fee
l bet
ter
for
disc
harg
e (n
=3)
, did
you
fee
l rea
dy f
or d
isch
arge
(n=
1), w
as a
sum
mar
y pr
ovid
ed a
t di
scha
rge?
( n=
12).
b Va
riab
les
whe
re ‘
unkn
own’
or
‘don
’t k
now
’ w
as a
nsw
ered
, th
ese
wer
e re
gard
ed a
s m
issi
ng s
ince
the
y w
ere
belo
w 1
0% (
see
met
hods
): N
umbe
r of
med
icin
e at
di
scha
rge
(n=
3), D
NR
polic
y at
dis
char
ge (n
=84
), IC
U/C
CU
/HD
U a
dmis
sion
dur
ing
IA (n
=4)
, Cog
niti
ve im
pair
men
t (o
ffici
al d
iagn
osis
) (n=
16),
Fal
len
at le
ast
once
in
last
hal
f ye
ar (n
=84
), D
isch
arge
lett
er is
sued
at
tim
e RA
(n=
20),
Fol
low
-up
plan
ned
at d
isch
arge
IA (n
=49
), H
as p
atie
nt b
een
seen
by
a do
ctor
bet
wee
n IA
and
RA
(n
=10
1), W
as a
sum
mar
y pr
ovid
ed a
t di
scha
rge
(n=
122)
wer
e an
swer
ed w
ith
‘unk
now
n’ b
y th
e ex
ecut
ive
rese
arch
er. D
id y
ou s
ugge
st d
octo
rs t
o st
ay lo
nger
at
IA?
(n=
9),
Did
you
fee
l bet
ter
at d
isch
arge
? (n
=28
), D
id y
ou f
eel r
eady
at
disc
harg
e? (
n=40
) w
as a
nsw
ered
wit
h ‘d
on’t
kno
w’
by t
he p
atie
nt s
urve
yed.
c
IA:
Inde
x A
dmis
sion
, RA
: Re
adm
issi
on,
LOSI
A:
Leng
th o
f st
ay I
A.
d Fo
r P
(pro
cedu
re)
1 po
int
was
giv
en t
o al
l no
n-m
edic
al i
ndex
adm
issi
ons,
0 t
o m
edic
al I
As,
m
issi
ng d
ata
was
con
side
red
as n
orm
al. e O
ffici
al d
iagn
osis
of
cogn
itiv
e im
pair
men
t in
med
ical
his
tory
, but
ass
esse
d as
men
tally
com
pete
nt t
o an
swer
the
que
stio
ns
(i.e.
usi
ng M
MSE
>20
)
141
Table 2 | ‘Predictability and preventability according to the interviewed’ ab
Patients: Predictable, No. (%) Patients: Preventable, No. (%)No 1037 (74.4) No 875 (62.9)
Yes 325 (23.3) Yes 337 (24.2)
Don’t know 31 (2.2) Don’t know 180 (12.9)
Physicians: Predictable, No. (%) Physicians: Preventable, No. (%)No 844 (64.7) No 1000 (76.6)
Yes 412 (31.6) Yes 222 (17.0)
Don’t know 49 (3.8) Don’t know 83 (6.4)
Nurses: Predictable, No. (%) Nurses: Preventable, No. (%)No 741 (55.4) No 856 (64.2)
Yes 427 (31.9) Yes 250 (18.2)
Don’t know 170 (12.7) Don’t know 237 (17.7)
Carers: Predictable, No. (%) Carers: Preventable, No. (%)No 566 (61.3) No 556 (59.3)
Yes 359 (38.3) Yes 298 (31.8)
Don’t know 13 (1.4) Don’t know 83 (8.9)
Researchers: Predictable, No. (%) Researchers: Preventable, No. (%)No 841 (60.3) No 988 (70.8)
Yes 508 (36.4) Yes 281 (20.1)
Don’t know 46 (3.3) Don’t know 126 (9.0)
Majority: Predictable, No. (%) Majority: Preventable, No. (%)No 1008 (72.1) No 1196 (85.6)
Yes 390 (27.8) Yes 202 (14.4)
a Questions: 1. Do you feel the current readmission was expected?’ (yes, no, don’t know), and 2. ‘Do you feel current readmission was preventable?’ (yes, no, don’t know)b Missing data include the following: patient predictable (n=5), patient preventable (n=6), physician predictable (n=93), physician preventable (n=93), carer predictable (n=460), carer preventable (n=461), nurse predictable (n=60), nurse preventable (n=62), researcher predictable (n=3), researcher preventable (n=3).
142
Table 3 | ‘Consensus on readmission’
Kappa Preventability (κ)
Kappa Predictability (κ)
- Patient Physician Nurse Carer Researcher
Patient - 0.105 0.135 0.360 0.174
Physician 0.173 - 0.273 0.225 0.473
Nurse 0.153 0.338 - 0.194 0.356
Carer 0.289 0.230 0.243 - 0.312
Researcher 0.231 0.607 0.408 0.380 -
Table 4 | ‘Factors positively associated with predictability and preventability of readmissions assessed by the majority’a
Predictability Final model (n=1240), AUROC: 0.71 Odds Ratio (95% CI) P value
Patient age, y 0.98 (0.97-0.99) P < 0.001
IAb type, Non-elective 2.55 (1.59-4.08) P <0.001
Clinical Frailty Scale25 1.29 (1.18-1.42) P < 0.001
Charlson Comorbidity Index24 1.08 (1.02-1.15) P = 0.013
Length between IA and RAb, d 1.02 (1.00-1.03) P = 0.053
Total admissions year prior to IA, elective and non-elective Overall P < 0.001
0
1-5 1.12 (0.82-1.52) P = 0.47
> 5 2.69 (1.72-4.20) P < 0.001
Did you (the patient) suggest to doctors to stay longer at IA? Yes
1.54 (0.99-2.40) P = 0.055
Did you (the patient) feel ready at discharge (IA)? Yes 0.55 (0.40-0.75) P < 0.001
Preventability Final model (n=1155), AUROC: 0.68 Odds Ratio (95% CI) P value
Discharge letter issued at time of readmission, Yes 0.68 (0. 47-1.00) P = 0.051
Follow-up planned at discharge IA, Yes 0.52 (0.35-0.78) P = 0.002
Did you (the patient) feel ready at discharge (IA)? Yes 0.35 (0.24-0.49) P < 0.001
a Corrected for country for all. b IA: Index Admission, RA: Readmission.
143
Tab
le 5
| ‘F
acto
rs p
ositi
vely
ass
ocia
ted
with
rea
dmis
sion
bei
ng a
sses
sed
as p
redi
ctab
le a
nd p
reve
ntab
le p
er in
terv
iew
ed g
roup
’ ab
Fin
al m
od
els
Pati
ent
Pred
icta
ble
(n
=12
40)
Pati
ent
Prev
enta
ble
(n
=10
67)
Phys
icia
nPr
edic
tab
le
(n=
1216
)
Phys
icia
n
Prev
enta
ble
(n=
978)
Nu
rse
Pred
icta
ble
(n=
945)
Nu
rse
Prev
enta
ble
(n=
961)
Car
er
Pred
icta
ble
(n=
860)
Car
er
Prev
enta
ble
(n=
816)
Res
earc
her
Pr
edic
tab
le(n
=11
71)
Res
earc
her
Pr
even
tab
le(n
=10
11)
AU
ROC
Fin
al m
odel
0.72
0.74
0.71
0.65
0.74
0.70
0.68
0.71
0.71
0.68
Ch
arac
teri
stic
Od
ds
Rat
io
(95%
CI)
, P-
valu
e
Od
ds
Rat
io
(95%
CI)
, P-
valu
e
Od
ds
Rat
io
(95%
CI)
, P-
valu
e
Od
ds
Rat
io
(95%
CI)
, P-
valu
e
Od
ds
Rat
io
(95%
CI)
, P-
valu
e
Od
ds
Rat
io
(95%
CI)
, P-
valu
e
Od
ds
Rat
io
(95%
CI)
, P-
valu
e
Od
ds
Rat
io
(95%
CI)
, P-
valu
e
Od
ds
Rat
io
(95%
CI)
, P-
valu
e
Od
ds
Rat
io
(95%
CI)
, P-
valu
ePa
tien
t ag
e, y
2.34
(1
.42-
4.02
),
P <
0.0
01
--
1.01
(1
.00-
1.02
), P
= 0
.058
0.98
(0
.96-
0.99
), P
< 0
.001
--
--
-
Mal
e se
x-
--
--
--
--
-
Part
nere
d/M
arri
ed-
--
0.67
(0
.48-
0.94
), P
= 0
.02
-0.
53
(0.3
8-0.
75),
P <
0.0
01
-0.
58
( 0.
42-0
.81)
, P
= 0
.001
-0.
73
(0.5
3-1.
01),
P =
0.0
54
IA t
ype,
Non
-ele
ctiv
e2.
34
(1.4
2-4.
02),
P <
0.0
01
-2.
21
(1.4
4-3.
38),
P <
0.0
01
-2.
03
(1.2
5-3.
23),
P =
0.0
04
1.81
(1
.03-
3.18
), P
< 0
.039
1.63
(1
.06-
2.51
), P
= 0
.026
-1.
95
(1.2
9-2.
93),
P =
0.0
01
-
Leng
th o
f st
ay IA
, d
--
--
--
--
--
Leng
th b
etw
een
IA
and
RA,
d-
--
--
--
0.96
(0
.94-
0.98
), P
< 0
.001
0.96
(0
.94-
0.98
), P
< 0
.001
Clin
ical
Fra
ilty
Scal
e25
--
1.31
(1
.20-
1.43
), P
< 0
.001
-1.
38
(0.9
6-1.
98),
P =
0.0
87
-1.
18
(1.0
7-1.
30),
P =
0.0
01
-1.
30
(1.1
9-1.
42),
P <
0.0
01
Cha
rlso
n C
omor
bidi
ty
Inde
x24
--
1.07
(1
.01-
1.13
), P
= 0
.029
--
0.92
(0
.85-
0.99
), P
= 0
.035
--
1.09
(1
.03-
1.16
), P
= 0
.004
0.92
(0
.86-
1.00
), P
= 0
.036
HO
SPIT
AL
Scor
e16-
--
--
--
--
1.12
(1
.03-
1.21
2),
P =
0.0
07
144
Tab
le 5
| (c
ontin
ued)
Fin
al m
od
els
Pati
ent
Pred
icta
ble
(n
=12
40)
Pati
ent
Prev
enta
ble
(n
=10
67)
Phys
icia
nPr
edic
tab
le
(n=
1216
)
Phys
icia
n
Prev
enta
ble
(n=
978)
Nu
rse
Pred
icta
ble
(n=
945)
Nu
rse
Prev
enta
ble
(n=
961)
Car
er
Pred
icta
ble
(n=
860)
Car
er
Prev
enta
ble
(n=
816)
Res
earc
her
Pr
edic
tab
le(n
=11
71)
Res
earc
her
Pr
even
tab
le(n
=10
11)
Num
ber
of m
edic
ine
at
disc
harg
e (c
hron
ic a
nd
non-
chro
nic
use)
--
--
P =
0.0
82-
P =
0.0
01P
= 0
.094
--
0-4
--
--
--
-
5-8
--
--
1.70
(1
.06-
2.72
), P
= 0
.028
-0.
74
(0.4
8-1.
15),
P =
0.1
78
1.00
(0
.63-
1.59
), P
= 0
.97
--
Mor
e th
an 8
--
--
1.55
(0
.98-
2.45
), P
= 0
.059
-1.
45
(0.9
6-2.
19),
P =
0.7
5
1.41
(0
.93-
2.15
), P
= 0
.11
--
DN
R po
licy
at
disc
harg
e, D
o re
susc
itat
e
-1.
44
(1.0
0-2.
07),
P =
0.0
48
--
0.6
2
(0.4
2-0.
92),
P =
0.0
18
--
--
-
ICU
/CC
U/H
DU
ad
mis
sion
dur
ing
IA,
Yes
--
--
--
--
--
Cog
niti
ve im
pair
men
t (o
ffici
al d
iagn
osis
), Y
es-
--
--
--
-2.
03
(0.8
9-4.
63),
P =
0.0
91
Falle
n at
leas
t on
ce in
la
st h
alf
year
, Ye
s-
--
--
1.37
(0
.95-
1.95
), P
= 0
.089
--
--
145
Tab
le 5
| (c
ontin
ued)
Fin
al m
od
els
Pati
ent
Pred
icta
ble
(n
=12
40)
Pati
ent
Prev
enta
ble
(n
=10
67)
Phys
icia
nPr
edic
tab
le
(n=
1216
)
Phys
icia
n
Prev
enta
ble
(n=
978)
Nu
rse
Pred
icta
ble
(n=
945)
Nu
rse
Prev
enta
ble
(n=
961)
Car
er
Pred
icta
ble
(n=
860)
Car
er
Prev
enta
ble
(n=
816)
Res
earc
her
Pr
edic
tab
le(n
=11
71)
Res
earc
her
Pr
even
tab
le(n
=10
11)
Tota
l adm
issi
ons
year
pr
ior
to IA
, el
ecti
ve
and
non-
elec
tive
--
P <
0.0
01
-P
= 0
.002
P =
0.0
96P
= 0
.090
-P
< 0
.001
-
0-
--
--
1-5
--
1.24
(0
.93-
1.67
), P
= 0
.149
-1.
45
(1.0
3-2.
04),
P =
0.0
31
1.39
(0
.94-
2.04
), P
= 0
.096
0.98
(0
.71-
1.34
), P
= 0
.876
1.40
(1
.04-
1.88
), P
= 0
.026
-
> 5
--
2.97
(1
.92-
4.59
), P
< 0
.001
-2.
37
(1.4
4-3.
88),
P =
0.0
01
1.79
(1
.03-
3.11
), P
= 0
.039
1.68
(1
.00-
2.80
), P
= 0
.049
-3.
40
(2.1
7-5.
34),
P <
0.0
01
-
Dis
char
ge le
tter
issu
ed
at t
ime
RA,
Yes
--
--
--
--
--
Follo
w-u
p pl
anne
d at
di
scha
rge
IA,
Yes
--
-0.
62
(0.4
1-0.
94),
P =
0.0
26
-0.
56
(95%
CI
0.37
-0.8
5),
P =
0.0
06
-0.
66
(95%
CI
0.43
-1.0
1),
P =
0.0
57
-0.
42
(95%
CI
0.22
-0.6
1),
P <
0.0
01
Has
pat
ient
bee
n se
en
by a
doc
tor
betw
een
IA a
nd R
A?
Yes
2.08
(1
.47-
2.93
),
P <
0.0
01
--
--
--
--
-
Was
a s
umm
ary
prov
ided
at
disc
harg
e?
Yes
--
-0.
67
(0.4
6-0.
98),
P =
0.0
4
1.38
(0
.96-
1.98
), P
= 0
.087
--
--
-
Did
you
(th
e pa
tien
t)
sugg
est
doct
ors
to s
tay
long
er a
t IA
? Ye
s
-2.
18
(1.3
5-3.
53),
P =
0.0
02
1.56
(0
.99-
2.46
), P
= 0
.056
--
--
--
-
Did
you
(th
e pa
tien
t)
feel
bet
ter
at d
isch
arge
(IA
)? Y
es
0.68
(0
.47-
0.98
),
P =
0.0
39
0.69
(0
.48-
0.98
), P
= 0
.039
-0.
62
(0.4
4-0.
87),
P =
0.0
06
--
--
--
146
Tab
le 5
| (c
ontin
ued)
Fin
al m
od
els
Pati
ent
Pred
icta
ble
(n
=12
40)
Pati
ent
Prev
enta
ble
(n
=10
67)
Phys
icia
nPr
edic
tab
le
(n=
1216
)
Phys
icia
n
Prev
enta
ble
(n=
978)
Nu
rse
Pred
icta
ble
(n=
945)
Nu
rse
Prev
enta
ble
(n=
961)
Car
er
Pred
icta
ble
(n=
860)
Car
er
Prev
enta
ble
(n=
816)
Res
earc
her
Pr
edic
tab
le(n
=11
71)
Res
earc
her
Pr
even
tab
le(n
=10
11)
Did
you
(th
e pa
tien
t)
feel
rea
dy a
t di
scha
rge
(IA)?
Yes
0.29
(0
.20-
0.42
),
P <
0.0
01
0.22
(0
.15-
0.32
), P
< 0
.001
0.71
(0
.52-
0.97
), P
= 0
.033
- 0
.46
(0
.33-
0.65
), P
< 0
.001
0.47
(0
.33-
0.68
), P
< 0
.001
0.42
(0
.30-
0.60
), P
< 0
.001
0.24
(0
.17-
0.34
), P
< 0
.001
0.48
(0
.36-
0.65
), P
< 0
.001
0.46
(0
.33-
0.65
) P
< 0
.001
a C
orre
cted
for
cou
ntry
for
all
b IA
: In
dex
Adm
issi
on,
RA:
Read
mis
sion
.
147
DISCUSSIONThis prospective European multicentre study of 1398 unscheduled medical readmissions
revealed that there was poor consensus on predictability and preventability among readmitted
patients, their carers, nurses, physicians. Especially, there was little consensus between patients
and their physicians. In addition, factors that could potentially contribute to (preventable)
readmission were notably different according to every group interviewed. Not feeling ready
for discharge was strongly associated with predictability and preventability when opinion
of the majority interviewed was used as the gold standard. This was also underlined when
opinions of the interviewees were taken separately. In addition, according to the opinion of
the majority, 27.8% of the readmissions were deemed predictable and 14.4% preventable.
Readmissions thought to have been caused by healthcare worker failures were more often
deemed preventable.
Our study is the largest multicentre study performed to date, and the first investigating
European readmissions prospectively. Previous large scale studies were mostly performed in
the USA and due to the differences in healthcare systems, results of these studies are probably
not applicable in a European setting.20 In addition, our study is one of the first studies in Europe
where opinions of most important stakeholders in the care-chain were taken into account.
Most previous studies were small retrospective studies, or did not involve all stakeholders.7,21,28
The percentages of readmissions which were deemed preventable are in line with previous
research, but their numbers do distinctly differ per interviewed group.11 An important finding
is the discrepancy in assessment between all interviewed, especially between patient and their
treating physician, the latter assessing less readmissions as likely preventable. This implies that
these healthcare professionals do not agree with their patients about the predictability and
preventability and associated factors.
Although, multiple risk models have been composed trying to create readmission prediction
models, most models do not perform satisfactorily in a European population.20 This was also
demonstrated in our cohort, 53.8% had a low risk of readmission according to the HOSPITAL-
score (Supplement 3).16 In addition, many variables in these models are not modifiable, and
therefore not suitable as interventions to improve the healthcare chain.18-20
According to our results, if a patient reported not feeling ready for discharge this was a risk
factor associated with a higher chance of (preventable) readmission. Auerbach et al. (2016)
also reported this by stating, that a proportion of US readmissions may be prevented with
better attention to patients’ readiness for discharge.23 In addition, we inquired all interviewed
for root causes that could better qualify the lack of readiness. Readmissions were more often
deemed preventable if attributed to healthcare worker-related causes, which are probably
modifiable and therefore a potential focus for improvement. Commencing improvement by
simply asking patient at the bedside whether he feels ready before discharge may be the first
step in understanding each other’s perspectives and could make other prediction models less
148
relevant. A recent study has underlined this by showing that early preparation for discharge
resulted in significant reductions in patients reporting they were unaware of problems to watch
out for post-discharge, and patients who did not understand their recovery plan.29 In addition,
Greysen et al. (2017) reported most readmitted patients understood their post-discharge
plan but were not explicitly asked about anticipated difficulties carrying out the aspects of
this plan.30
The interviewed groups did not agree on predictability, let alone on preventability. A high
readmission rate should be a signal for a hospital to further look into the causes, but looking
solely at the rate and penalizing the hospitals, without correcting for case-mix and looking
more closely into the type of care provided on the work floor is questionable. Our work
demonstrates that defining a gold standard integrating preventability in a quality indicator
is difficult.
The limitations of this study must be acknowledged. We collected data from four different
countries in 15 hospitals all using different healthcare systems, which may make results less
generalizable. However, European healthcare systems are more alike when compared to US
system and we corrected for country in our analysis. Another limitation is asking opinions
during readmission about the index admission which could potentially lead to (recall)bias and
subjectivity. The majority opinion was used as the gold standard to decide on preventability
and predictability which may be questionable. Therefore, we also reported on perspectives per
interviewed group separately. 8
CONCLUSIONThe European acute healthcare chain is under increasing pressure, potentially resulting in more
unscheduled readmissions. This international multicentre study performed in 2016 is the first
to prospectively assess this problem in Europe. The 1398 readmitted patients, their carers,
treating nurses and their physicians do not agree on the predictability and preventability
of readmissions, let alone associated risk factors. This raises the question of the validity of
readmissions as a quality monitor. Extensive research has been performed on risk models, but
healthcare professionals simply asking the patient whether they are feeling ready for discharge
may be a key question to target in preventing unnecessary readmissions.
ACKNOWLEDGEMENTSThe authors have nothing to disclose. No funding was received for this study.
A full list of membership of the safer@home consortium is as follows: Mikkel Braband, Tim
Cooksley, Louise S. van Galen, Harm R. Haak, Rachel M. Kidney, John Kellet, Hanneke Merten,
Prabath W. Nanayakkara, Christian. H. Nickel, John Soong, Christian P. Subbe, Immo Weichert.
Collaborators (alphabetically ordered): Lisette Ackermans1, MD, Alex B. Arntzenius14, MD,
PhD, Dennis Barten7, MD, Maarten H. van der Bie14, MD, Tom Boeije15, MD, Ella Chaudhuri16,
149
MD, Marieke Diepenbroek-Meekes17, Eva Durinck6, Bsc, Jelle A. van Erven1, BSc, MD, Tabitha
A. Graaff-de Kooter6, MD, Frits Holleman18, MD, PhD, Job Huussen17, MD, W. Jansen6, Jonas
Junghans Jensen19, BSc, John Kellet20, MB, FRCPI, Tessa Knol21, BSc, Irene Lee13, MD, Darleen A.
Leung13, MD, Bas van Lieshout18, BSc, Arend-Jan Meinders21, MD, Nieke E. Mullaart-Jansen15,
MD, Sietske van Nassau8, BSc, Jonas Larsen Pedersen19, BSc, Rune Overgaard Jensen19, BSc,
Anouschka Pronk21, MD, Mette Rahbek Kristensen2, Amy M. Ridge9, MD, Timo Roeleveld14,
MD, Marije C. Schipper18, BSc, Daisy Vedder15, MD, Joris van der Vorst18, BSc, MD, Joyce
Wachelder8, MD
1Department of Internal Medicine, VU University Medical Centre, Amsterdam, The Netherlands 2Department of Emergency Medicine, Hospital of South West Jutland, Denmark3Department of Acute Medicine, University Hospital of South Manchester, Manchester, United Kingdom4Department of Epidemiology and Biostatistics, VU University Medical Centre Amsterdam,
Amsterdam, The Netherlands5Department of Public and Occupational Health, EMGO Institute for Health and Care Research,
VU University Medical Centre, Amsterdam, The Netherlands 6Department of Quality, Safety & Innovation, Albert Schweitzer Hospital,
Dordrecht, The Netherlands 7Department of Emergency Medicine, VieCuri Hospital, Venlo, The Netherlands 8aDepartment of Internal Medicine, Maxima Medisch Centre, Eindhoven/Veldhoven, The Netherlands8bDepartment of Internal Medicine, Division of General Internal Medicine, Maastricht University
Medical Centre+. Maastricht University. 9Department of Acute Medicine, St. James Hospital, Dublin, Ireland10Department of Emergency Medicine, University Hospital Basel, Switzerland11Imperial College London, United Kingdom12Department of Acute Medicine, The Ipswich Hospital NHS Trust, Ipswich, United Kingdom13Department of Acute Medicine, Ysbyty Gwynedd Hospital, Wales, United Kingdom14Department of Internal Medicine, Spaarne Gasthuis, Hoofddorp, The Netherlands 15Department of Emergency Medicine, Westfries Gasthuis, Hoorn, The Netherlands 16Department of Acute Medicine, North Bristol NHS Trust, Bristol, United Kingdom17Department of Internal Medicine, Slingeland Hospital, Doetinchem, The Netherlands18Department of Internal Medicine, Academic Medical Centre, Amsterdam, The Netherlands19Department of Emergency Medicine, Odense University Hospital, Denmark20Thunder Bay Regional Health Sciences Centre, Thunder Bay, Ontario, Canada 21Department of Internal Medicine, Antonius Hospital, Nieuwegein, The Netherlands
SUPPLEMENTAL MATERIAL
Supplemental Material 1 ‘Questionnaire’
Supplemental Material 2 ‘PRISMA model’
Supplement Material 3 ‘Patient characteristics (additional)’
150
REFERENCES1. Greysen SR, Stijacic Cenzer I, Auerbach AD, Covinsky KE. Functional impairment and hospital
readmission in Medicare seniors. JAMA Intern Med. 2015;175(4):559-565.
2. Blunt I, Bardsley M, Grove A, Clarke A. Classifying emergency 30-day readmissions in England using routine hospital data 2004-2010: what is the scope for reduction? Emerg Med J. 2015;32(1):44-50.
3. Balla U, Malnick S, Schattner A. Early readmissions to the department of medicine as a screening tool for monitoring quality of care problems. Medicine (Baltimore). 2008;87(5):294-300.
4. Basisset Kwaliteitsindicatoren 2014. Utrecht: Inspectie voor de gezondheidszorg, 2015. Accessed January, 2016.
5. Pines JM, Hilton JA, Weber EJ, et al. International perspectives on emergency department crowding. Acad Emerg Med. 2011;18(12):1358-1370.
6. VUmc Netwerk Acute Zorg. Ontwikkelingen aanbod acute patiënten SEH’s RsHsQQ, 2014 en 2015 ROAZ-regio VUmc. https://www.vumc.nl/afdelingen-themas/719134/8513847/8514115/20150929_RapportAcuteZorgQ11.pdf. Published September 2015. Accessed August, 2016.
7. Bianco A, Mole A, Nobile CG, Di Giuseppe G, Pileggi C, Angelillo IF. Hospital readmission prevalence and analysis of those potentially avoidable in southern Italy. PLoS One. 2012;7(11):e48263.
8. Halfon P, Eggli Y, Pretre-Rohrbach I, Meylan D, Marazzi A, Burnand B. Validation of the potentially avoidable hospital readmission rate as a routine indicator of the quality of hospital care. Med Care. 2006;44(11):972-981.
9. van Galen LS, Nanayakkara PW. [Hospital readmissions: A reliable quality indicator?]. Ned Tijdschr Geneeskd. 2016;160:A9885.
10. van Walraven C. The Utility of Unplanned Early Hospital Readmissions as a Health Care Quality Indicator. JAMA Intern Med. 2015;175(11):1812-1814.
11. van Walraven C, Jennings A, Forster AJ. A meta-analysis of hospital 30-day avoidable readmission rates. J Eval Clin Pract. 2012;18(6):1211-1218.
12. Cakir B, Gammon G. Evaluating readmission rates: how can we improve? Southern medical journal. 2010;103(11):1079-1083.
13. Meisenberg BR, Hahn E, Binner M, et al. ReCAP: Insights Into the Potential Preventability of Oncology Readmissions. J Oncol Pract. 2016;12(2):153-154.
14. L.S. van Galen TC, H. Merten et al.*On behalf of the safer@home consortium. (10-2016) Physician Consensus on preventability and predictability of Readmissions based on standard case scenarios. Running title: Physician consensus on readmissions. Neth J Med. 2016 Dec;74(10):434-442.
15. van Walraven C, Dhalla IA, Bell C, et al. Derivation and validation of an index to predict early death or unplanned readmission after discharge from hospital to the community. CMAJ. 2010;182(6):551-557.
16. Donze J, Aujesky D, Williams D, Schnipper JL. Potentially avoidable 30-day hospital readmissions in medical patients: derivation and validation of a prediction model. JAMA Intern Med. 2013;173(8):632-638.
17. Donze JD, Williams MV, Robinson EJ, et al. International Validity of the HOSPITAL Score to Predict 30-Day Potentially Avoidable Hospital Readmissions. JAMA Intern Med. 2016;176(4):496-502.
18. Kansagara D, Englander H, Salanitro A, et al. Risk prediction models for hospital readmission: a systematic review. JAMA. 2011;306(15):1688-1698.
19. Zhou H, Della PR, Roberts P, Goh L, Dhaliwal SS. Utility of models to predict 28-day or 30-day unplanned hospital readmissions: an updated systematic review. BMJ Open. 2016;6(6):e011060.
20. Cooksley T, Nanayakkara PW, Nickel CH, et al. Readmissions of medical patients: an external validation of two existing prediction scores. QJM. 2016;109(4):245-248.
151
21. Fluitman KS, van Galen LS, Merten H, et al. Exploring the preventable causes of unplanned readmissions using root cause analysis: Coordination of care is the weakest link. Eur J Intern Med. 2016;30:18-24.
22. Mixon AS, Goggins K, Bell SP, et al. Preparedness for hospital discharge and prediction of readmission. J Hosp Med. 2016;11(9):603-609.
23. Auerbach AD, Kripalani S, Vasilevskis EE, et al. Preventability and Causes of Readmissions in a National Cohort of General Medicine Patients. JAMA Intern Med. 2016;176(4):484-493.
24. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373-383.
25. Rockwood K, Song X, MacKnight C, et al. A global clinical measure of fitness and frailty in elderly people. CMAJ. 2005;173(5):489-495.
26. van Wagtendonk I, Smits M, Merten H, Heetveld MJ, Wagner C. Nature, causes and consequences of unintended events in surgical units. Br J Surg. 2010;97(11):1730-1740.
27. Vuuren Wv, Shea CE, Schaaf TWvd. The development of an incident analysis tool for the medical fields. Eindhoven University of Technology.
28. Magdelijns FJ, Schepers L, Pijpers E, Stehouwer CD, Stassen PM. Unplanned readmissions in younger and older adult patients: the role of healthcare-related adverse events. Eur J Med Res. 2016;21(1):35.
29. Harrison JD, Greysen RS, Jacolbia R, Nguyen A, Auerbach AD. Not ready, not set...discharge: Patient-reported barriers to discharge readiness at an academic medical center. J Hosp Med. 2016;11(9):610-614.
30. Greysen SR, Harrison JD, Kripalani S, et al. Understanding patient-centred readmission factors: a multi-site, mixed-methods study. BMJ Qual Saf. 2016.
CHAPTER 10
HOSPITAL STANDARDIZED MORTALITY RATIO: A RELIABLE INDICATOR OF QUALITY OF CARE?
Jelle A. van Erven | Louise S. van Galen | Asselina A. Hettinga-Roest | Ellen P. Claessens | Jan C. Roos | Mark H. Kramer | Prabath W. Nanayakkara
Submitted
‘They die one by one’ Carol Haraden
154
ABSTRACTObjective
This study investigates (1) whether the HSMR (Hospital Standardized Mortality
Ratio) model under- or overestimates disease severity and (2) the completeness of
the data collected by administrators to calculate HSMR in a cohort of deceased patients with
the diagnosis pneumonia.
Design, Participants and Main Outcome Measures
In this retrospective cohort study PSI scores, abbMEDS scores and associated mortality
probabilities were obtained from 32 deceased pneumonia-patients over the year 2014 in
the VU University Medical Centre. These were compared to mortality probabilities of the Central
Bureau for Statistics (CBS) calculated for every patient using the HSMR model. Clinical charts
were examined to extract relevant comorbidities to determine the reliability of data sent to
the national registration of hospital care.
Results
Risk categories determined by using the PSI and the abbMEDS were significantly higher
compared to the risk categories according to HSMR (p = 0.001, respectively p = 0.000).
The mean difference between the number of comorbidities in our registration and the coders’
registration was 1.97 (p = 0.00). The mean difference was 0.97 (p = 0.000) for the number
of comorbidities of influence on the Charlson Comorbidity Index (CCI) and 1.25 (p=0.001) for
the calculated CCI.
Conclusion
The results of this study suggest that the mortality probabilities as calculated by the CBS
are an underestimation of the risk of dying for each patient. Our study also showed that
the registration of data sent to the CBS underestimated the actual comorbidities of the patient,
and could possibly influence the HSMR.
Key words
Data registration, disease severity, HSMR, quality indicator, patient outcomes
155
INTRODUCTIONSince March 2014, Dutch hospitals are compulsory to be transparent about their mortality
rates.1 To be able to compare the quality of hospital care using their mortality rate, these
rates have to be standardized in order to correct for the differences in the case-mix.2 This
standardized ratio is represented in the ‘Hospital Standardized Mortality Ratio’ and is the ratio
of the observed to the expected deaths, derived from data from the LBZ (LBZ: National
registration of hospital care).3 The expected deaths are calculated with the use of a statistical
model that corrects for certain factors such as age, socio-economic status and comorbidity.3
In 2014, this model contained standardized mortality ratios of 50 diagnosis groups (SMRs),
which account for 80% of in-hospital death. This has been extended to 157 diagnosis groups
(SMRs) in 2015.
Over the year 2014, the VUmc (VU Medical Centre) had a relatively high HSMR, in part caused
by a high SMR for the diagnosis group ‘pneumonia’. The SMR of a diagnosis group can be
used to investigate the cause of unexpected high mortality in a hospital more specifically
than by solely using the HSMR.4 For this reason a commission of independent external
investigators in VUmc were asked to investigate this high SMR. The aim was to investigate
whether preventable/avoidable factors contributed to these deaths. Their report showed no
avoidable causes of death in this cohort. These findings suggest that the cause of the high
SMR for pneumonia is probably due to other (unknown) factors. It could, for example, be
caused by insufficient registration of comorbidities or wide variations in the disease severity.
In the clinical setting, physicians and nurses use several different scoring systems to determine
the severity and to predict the mortality of pneumonia using patient characteristics such as
age, blood urea and respiratory rate. Two of the best-validated and most used scoring systems
are the PSI (Pneumonia Severity Index) and the abbMEDS (Abbreviated Mortality in Emergency
Department Sepsis).5,6
The HSMR is calculated by the CBS (Central bureau of statistics) and the data used for
this calculation is registered by DHD (Dutch Hospital Data) within the context of the LBZ.
The Medical Administration Office of each hospital provides the information that is used.
The HSMR is, among other covariates, derived from the primary diagnosis and the Charlson
Comorbidity Index (CCI),7which are obtained from patients’ charts and documented by coders.
This underlines the importance of a complete administration, as deficient or faulty data might
directly influence the HSMR. Van der Laan et al. (2013) showed that the effect of registering
10 percent more comorbidities could result in a decrease of 5 points of the HSMR.8 Although
the administration of data has improved significantly since the implementation of the HSMR
as an indicator of quality of care, there still might be inconsistencies in the (comorbidity) data
extracted by coders and registered by DHD, when compared with the actual data extracted by
doctors from the patients’ charts.8,9
Therefore, the main aim of this study is to examine whether the HSMR model under- or
overestimates the disease severity of pneumonia patients when compared to routinely used
156
clinical severity scores. Our secondary aim is to investigate the completeness of the data sent
to DHD to calculate the HSMR.
METHODSIn 2014, 32 deceased patients were registered in the ‘pneumonia’-group at the VUmc. In order
to obtain PSI scores and abbMEDS scores for these patients, patients’ charts were examined for
information needed to calculate these scores from which corresponding mortality probabilities
could be calculated. Missing information was considered as not contributing to the score.
The HSMR is calculated by logistic regression using below mentioned covariates with data
provided by hospital coders. With this information, regression coefficients for these covariates
are estimated and are used to calculate mortality probabilities for each individual admission.3
The results of the calculations is sent to each hospital in the annual HSMR report.
The HSMR is calculated using the following covariates3
– Age at admission
– Sex
– SES (socio-economic status) of the postal area of the patient’s address. The SES
classification per postal code is compiled by the Netherlands Institute for Social
Research (SCP)
– Severity of main diagnosis. Instead of CCS diagnosis subgroups (Clinical
Classifications Software: a tool to cluster patient diagnoses into a manageable
number of clinically meaningful categories, based on the International Classification
of Diseases. The CCS makes little distinction in regard to disease severity when
categorizing diagnosis codes), a classification of severity of the main diagnosis in
terms of mortality rates is used, as suggested by Van den Bosch et al. (2011)10
– Urgency of admission (elective, acute)
– Comorbidity (17 comorbidity groups of the Charlson comorbidity index7)
– Source of admission (home, nursing home or other institution, hospital)
– Year of discharge
– Month of admission
In order to compare the mortality probabilities derived from the PSI scores and abbMEDS scores
(which correspond with ordinal risk categories) and the mortality probabilities calculated by
the CBS (which can be considered a continuous variable), new categories needed to be formed
for the latter. It was decided to form three sets of categories from the CBS data, one for each
of the scores. Table 1 shows the risk categories and corresponding mortality probabilities of
the two scoring systems. The consensus was that the best way to establish limits for new
categories, was by using the median between each of the mortality probabilities, as those
are the mean of that risk category. As can be seen in table 1, the lowest risk categories
of the PSI predict a risk of 0.1% and of 0.6%. The median between these risks is 0.35,
therefore, the limits of the PSI categories used are 0 – 0.35; 0.35 – 0.75; 0.75 – 5.2; 5.2 – 18.1;
18.1 – 100 and the limits for the abbMEDS are 0 – 11.55; 11.55 – 32.85; 32.85 – 100.
157
The newly formed categories of the CBS calculated mortality probabilities were compared to
the categories of the PSI scores and abbMEDS scores. A Wilcoxon Sign-Rank Test was used for
statistical analysis to test for conformity.
To investigate whether data sent to DHD significantly differed from what is found in patients’
charts, data was gathered on the total amount of comorbidities that were present in charts,
which of these were directly of influence on the CCI (excluding the comorbidities that are not
in the Charlson comorbidity index) and finally the estimated CCI by hospitals itself. The coders
in VUmc primarily look at the discharge letter and only broaden their scope when they presume
this to be insufficient. In this study one researcher (JVE) thoroughly checked every patient’s
chart which included the discharge letter. If there was any uncertainty concerning a possible
comorbidity or diagnosis, a second researcher (PN) was consulted and consensus was reached.
The data that the CBS used was obtained from the Medical Administration Office. A paired
t-test was used to analyse the difference between our registration and the coder’s registration.
For all analyses, a two-tailed p-value of less than 0.05 was considered statistically significant.
RESULTSTable 2 gives an overview of the patient characteristics of our population. 10 patients had
a cause of death other than respiratory failure or sepsis.
Mortality probabilities
Table 3 illustrates the dispersion of mortality probabilities calculated by the CBS using the HSMR
model and those of the two clinical scoring systems. It can be seen that for the majority of
patients the estimated risk of dying within 28-30 days is much higher according to the clinical
scoring systems than the estimated risk of dying as calculated using the SMR-model. Especially
the abbMEDS assesses the risk to be significantly higher than the CBS does.
Descriptive statistics of conformity were performed and this showed that for the PSI 18 patients
were in a higher risk category than according to the CBS (SMR), 3 were in a lower category
and 11 were in the same category. When looking at the abbMEDS, all patients were either in
the same risk category (10) or in a higher risk category (22) compared to SMR.
Table 1 | Risk categories and corresponding mortality probabilities of the scoring systems.
PSI5 abbMEDS6,11
Low risk I 0.1% Low risk 3.6%
Low risk II 0.6% Intermediate risk 19.5%
Low risk III 0.9% High risk 46.2%
Medium risk 9.5%
High risk 26.7%
The mortality probabilities of the risk categories for the abbMEDS score are derived from a study by Roest et al. (2015)11
158
Table 2 | Demographic and clinical characteristics of the deceased patients in the ‘pneumonia’-group over the year 2014
Characteristics
Deceased patients (n=32)
Number Percentage
Demographic factor Age > 65 year Female sex
Nursing home resident
25
12
6
78.13
37.5
18.75
Admissions > 2 last 12 months ICU admissions last 12 months
Unexpected long admission*
16
14
6
50
43.75
18.75
Cause of death
Respiratory failure
Sepsis
Myocardial infarction
Heart failure
Other
14
12
5
4
1
43.75
37.5
15.63
12.5
3.13
Other clinical characteristics Immunocompromised§ Do not resuscitate
Polypharmacy±
Limited mobility&
Delirium
Malnutrition$
10
26
28
24
8
12
31.25
81.25
87.5
75
25
37.5
* = An admission minimally 50 percent longer than expected for a specific primary diagnosis. The calculation of the expected length of admission takes into account the age of the patient, primary diagnosis and any possible interventions.§ = Immunodeficiency by the use of immuno-suppressive drugs, by neutro- or leukopenia or other causes.± = The chronic use of ≥ 5 medications.& = Patient uses devices for mobility or was bedridden.$ = Patient has a SNAQ (Short Nutritional Assessment Questionnaire) of ≥ 2 or when the patient was described as cachexic.
Further analysis showed a significant increase in assigned risk categories for the PSI (p<0,001)
and for the abbMEDS (p<0,001) compared to the SMR. This indicates that risks of dying of
these patients according to clinical scoring systems were significantly higher than the risks of
dying according to SMR calculated by the CBS.
Registration of data
Figure 1 shows the number of comorbidities, the number of comorbidities influencing the CCI
and the calculated CCI itself from our own registration and those same outcome measures
which medical coders registered. For each of the outcome measures the mean of our registered
number is higher than the mean of what the coders registered.
159
As table 4 shows, the mean difference between the number of comorbidities in our registration
and the coder’s registration is 1,97. The mean difference between our registration and
the coders’ registration for the number of comorbidities of influence of the CCI and of the CCI
itself are 0,97 and 1,25 respectively. All of these results are statistically significant.
Table 3 | Mortality probabilities calculated by the CBS and those derived from the scoring systems
Patient nr. Mortality probabilities CBS (SMR) PSI abbMEDS
1 0.87% 0.60% 3.60%
2 6.29% 26.70% 19.50%
3 14.70% 26.70% 46.20%
4 6.43% 9.50% 46.20%
5 10.01% 9.50% 19.50%
6 3.64% 26.70% 19.50%
7 18.02% 26.70% 19.50%
8 15.33% 26.70% 46.20%
9 10.62% 9.50% 19.50%
10 12.19% 26.70% 19.50%
11 8.76% 9.50% 3.60%
12 18.05% 26.70% 46.20%
13 18.60% 9.50% 19.50%
14 3.45% 0.90% 19.50%
15 5.39% 26.70% 46.20%
16 5.76% 26.70% 19.50%
17 11.02% 26.70% 19.50%
18 9.34% 26.70% 46.20%
19 6.68% 26.70% 19.50%
20 7.74% 9.50% 19.50%
21 5.92% 9.50% 19.50%
22 6.18% 9.50% 19.50%
23 24.59% 26.70% 19.50%
24 10.54% 26.70% 19.50%
25 13.70% 9.50% 19.50%
26 5.01% 9.50% 19.50%
27 1.54% 9.50% 19.50%
28 2.89% 26.70% 3.60%
29 6.32% 26.70% 46.20%
30 12.87% 9.50% 19.50%
31 13.37% 26.70% 46.20%
32 21.24% 9.50% 19.50%
Green represents higher probability than calculated by the CBS (using the SMR-model), red represents a lower probability.
160
Figure 1
Comorbiditi
es study
Comorbiditi
es coders
Comorbiditi
es Charlson stu
dy
Comorbiditi
es Charlson co
ders
Charlson-sc
orestu
dy
Charlson-sc
oreco
ders0
2
4
6
8
10
Study vs. Coder's registration
Coun
t
Figure 1 | Compared means and SD of the outcome measures extracted during our study versus what coders registered and the CBS used to calculate the HSMR.
Table 4 | Statistical analysis of the outcome measures
Mean SD* SE* 95% CI p-value
Comorbidities study – comorbidities CBS
1.96875 2.53345 .44785 1.05535-
2.88215
.000
Comorbidities Charlson study – Comorbidities Charlson CBS
0.96875 1.37921 .24381 .47149 -1.46601
.000
Charlson index study – Charlson index CBS
1.25000 1.95101 .34489 .54658 -1.95342
.001
Outcome measures in the table are the number of comorbidities and the number of comorbidities influencing the Charlson Comorbidity Index and the Charlson Comorbidity index.* SD: Standard deviation, SE: Standard error of the mean.
DISCUSSIONThe findings in this paper indicate that (1) the risk of dying for an individual patient is
higher when calculated using validated clinical scoring systems than the risk of dying
calculated with the HSMR model. (2) The total number of comorbidities and the number of
comorbidities influencing the CCI is higher according to our registration than according to
the coders’ registration.
The results in this study further support the suggestion that has been made by Pleizier et al.
that the SMR for more diagnosis groups besides cerebrovascular diseases will also decrease
when adjusted for the severity of disease.12 They concluded that within the SMR group
‘cerebrovascular diseases’ there is no distinction between ‘stroke’, ‘cerebral haemorrhage’
161
and ‘subarachnoid haemorrhage’ while their mortality rates differ greatly.12 The mortality rates
were 18, 43 and 35 percent respectively, and when these differences were not taken into
account, the influence on the SMR could be considerable.13 They recalculated the SMR for
‘cerebrovascular diseases’ after correcting for the above mentioned sub-diagnoses and found
that this gave a reduction from 119 (95%-CI 105-133) to 102 (95%-CI 91-115).12 Beside this
diagnosis group, this possibly is also true for other SMR groups such as the ‘pneumonia-group’.
A subdivision for ‘cerebrovascular diseases’ was easily made by just looking at the mortality
rates for several sub diagnoses within that group. This however, is a lot harder for a diagnosis
group such as ‘pneumonia’, where there are no known distinct sub diagnoses. To make
a subdivision for ‘pneumonia’, two different scoring systems were used that indicate severity
of disease. The best way to prove that a subdivision by each of these scores has a direct
effect on the SMR, is by adjusting the SMR model in the same way Pleizier et al. did.12 They
incorporated a division in risk categories into the logistic regression model, just like the other
covariates. In our study, it was decided to compare the mortality probabilities of the validated
scores to the mortality probabilities calculated by the CBS with the use of the SMR-model.
The results show that for the large majority of patients the expected mortality within 28-30
days is much higher according to the two scoring systems than to the score calculated by
the CBS. Probably partly caused by underscoring the number of co-morbidity, but also a lack
of proper adjustment for the severity of the disease pneumonia in the individual patient.
These two scoring systems are widely used in clinical settings when dealing with pneumonia
patients and have been validated.6,13 They are specifically designed to assess the severity of
pneumonia/sepsis and should therefore be taken seriously as predictors of death. This suggests
that the mortality probabilities according to the HSMR model of CBS are an underestimation
of the real risk of dying for each patient. Naturally, estimating disease severity with the use of
nine variables results in a simplification of reality. In addition, it is known that university medical
centres predominantly provide tertiary care for a case-mix of patients with a higher severity of
disease than peripheral hospitals. Therefore, they might falsely have a ‘higher’ HSMR.
Our results indicate that the mortality probabilities calculated by the PSI and the abbMEDS
are higher than what the CBS calculated. It could be argued that the steps between the risk
categories of these scoring systems are fairly big. Therefore, when a patient is placed in
the highest risk category of for example the abbMEDS, their risk of dying could be even
higher than 46.2 percent. However, table 2 does compare categorical variables (the mortality
probabilities calculated by the scoring systems) with a continuous variable (the mortality
probabilities calculated by the CBS), which implies that these risks will almost always differ
from the risks as calculated by the CBS.
The secondary aim of this study was to assess the registration of comorbidities from the patients’
charts by medical coders. An unanticipated finding was that the source of admission was in
every case ‘home’. It seemed as though no distinction was made between ‘home’ and ‘nursing
home’. Nevertheless, table 1 shows that 6 out of 32 patients were admitted from a nursing
162
home. This could potentially have an impact on the HSMR as a whole, however, this influence
is probably rather small. It must be acknowledged that the source of admission is not primarily
registered by coders, but they are responsible for checking this registration.
Van der Laan et al. already underlined the influence of the registered number of comorbidities
on the HSMR.8 With this in mind, an average difference of two registered comorbidities seems
significant enough to be of influence of the HSMR. For a comorbidity to be of influence of
the HSMR it needs to add to the CCI, so to make the previous assumption more likely, the CCI
of every patient was also taken in consideration. It was found that there was a statistically
significant difference of 1.25 points, between the calculated CCI based on our registration and
the calculated CCI based on the coders’ registration. This strongly suggests that the apparent
insufficiently registered number of comorbidities does directly influence the HSMR. As stated
earlier, coders are dependent on proper documentation by others, including doctors. They
primarily look at the discharge letter and surgery reports, and are not expected to go through
the entire patient chart, mainly since this would be too time-extensive. This lack of time
might be one of the causes of the apparent under registration of comorbidities. One other
cause explaining the under registration is that according to coding protocol, sometimes an
ICD-10 (International Classification of Diseases) code which has less impact on the HSMR
than the actual diagnosis has to be selected. Although the precise impact cannot be judged
by the results of this study, these findings do raise the question whether the HSMR is reliable
enough to estimate what it is supposed to do or to be published for everyone to see.
The limitations of this work must be acknowledged. The self-formed categories composed
to compare categorical and continuous variables are merely based on what was thought to
be the most logical way to do this. Hence, a note of caution is due here when interpreting
these results. Also, in this study no control group was investigated. This withholds us
the opportunity to compare the mortality probabilities of the living patients with the deceased
patients’ and therefore we weren’t able to investigate if the severity of disease was greater in
the deceased group.
CONCLUSIONHospitals are compulsory to publish their HSMRs, which gives patients and healthcare
institutions the opportunity to judge and compare hospitals on the basis of this number.
However, we demonstrated that differences in case-mix and the incompleteness of the data
used to calculate the HSMR could negatively influence the HSMR. Although it seems quite
logical to look at the number of deaths in each hospital as an indicator of quality of care, there
are numerous pitfalls hidden in using the HSMR as a quality indicator. Therefore, HSMR should
always be interpreted with caution and openly publishing HSMRs may have unfair negative
consequences for some hospitals.
163
REFERENCES1. Hospital Standardised Mortality Ratio, de Praktijk Index, www.hsmr.nl. Accessed January 3, 2017.
2. Understanding HSMRs. A Toolkit on Hospital Standised Mortality Ratios. http://www.drfoster.com/wp-content/uploads/2014/09/HSMR_Toolkit_Version_9_July_2014.pdf. Published 2014. Accessed December, 2016.
3. van der Laan JdB, A., van den Akker-Ploemacher J, Penning C, Pijpers F. HSMR 2014 methodological report, november 2015. http://www.hsmr.nl/wp-content/uploads/2016/01/2015hsmrmethodologicalreport2014.pdf. Accessed November, 2016.
4. Jarman B, Pieter D, van der Veen AA, et al. The hospital standardised mortality ratio: a powerful tool for Dutch hospitals to assess their quality of care? Qual Saf Health Care. 2010;19(1):9-13.
5. Fine MJ, Auble TE, Yealy DM, et al. A prediction rule to identify low-risk patients with community-acquired pneumonia. N Engl J Med. 1997;336(4):243-250.
6. Vorwerk C, Loryman B, Coats TJ, et al. Prediction of mortality in adult emergency department patients with sepsis. Emerg Med J. 2009;26(4):254-258.
7. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373-383.
8. Central bureau of statistics (Van der Laan J). Quality of the Dutch Medical Registration (LMR) for the calculation of the Hospital Standardised Mortality Ratio. ISSN 2013; 1572-0314.
9. Tol J, Broekman M, Brauers M, van Gulik T, Busch OR, Gouma DJ. [Reliability of the registration of data on complex patients: effects on the hospital standardised mortality ratio (HSMR) in the Netherlands]. Ned Tijdschr Geneeskd. 2012;156(49):A4918.
10. van den Bosch WF, Spreeuwenberg P, Wagner C. [Hospital standardised mortality ratio (HSMR): adjustment for severity of primary diagnosis can be improved]. Ned Tijdschr Geneeskd. 2011;155(27):A3299.
11. Roest AA, Tegtmeier J, Heyligen JJ, et al. Risk stratification by abbMEDS and CURB-65 in relation to treatment and clinical disposition of the septic patient at the emergency department: a cohort study. BMC Emerg Med. 2015;15:29.
12. Pleizier CM, Geerlings W, Pieter D, Boiten J. Patientmix influences HSMR. Medisch Contact 2010. 36: 1777-1779.
13. Bots ML, Jager-Geurts H, Berger-van Sijl M. Risk of dying after first hospital admission for a cerebrovascular accident in the Netherlands. Cardiovascular disease in the Netherlands. The Hague, 2006.
CHAPTER 11
SUMMARY OF MAIN RESULTS, GENERAL DISCUSSION AND FUTURE PERSPECTIVES
‘We cannot solve our problems with the same thinking we used when we created them’ Albert Einstein
167
This final chapter summarizes and discusses the findings of the work presented in this thesis. In
addition, it presents some future perspectives on patient safety in the acute healthcare chain.
The main aim of this thesis was to obtain insight and provide potential improvement strategies
in safety issues within the healthcare chain using novel ways to investigate these matters.
The acute healthcare chain is increasingly under pressure due to changing demography
(ageing population) and healthcare policy changes. This has led to an increased work
burden for healthcare workers and has negatively affected communication and coordination
within the chain resulting in poor handovers and inadequate teamwork.1-4 It is important
to address these issues since they can eventually result in serious adverse events such as
unplanned readmission, ICU admission, and death potentially jeopardizing the patient safety
and adversely affecting the clinical outcome. In every ten patients, one is still harmed while
receiving hospital care.5 To enhance patient safety, we must change the mindset and culture in
hospitals. In order to do so, creating awareness for patient safety issues in healthcare workers
is needed. In addition, the healthcare workers should be stimulated to take an active role in
these matters with the aim of increasing their intrinsic motivation so that they can actually
contribute to improving the patient safety on the workfloor.6 This thesis aimed to focus on this
awareness by addressing bottlenecks within the acute healthcare chain by working on research
questions originating from daily issues on the floor. The ideas for these projects originated on
the workfloor using healthcare worker point of views aiming to concentrate on ‘real’ systems
rather than ‘ideal’ systems so that our results would be applicable in the real-life settings. In
addition, we hypothesized that the recognition and tackling of these well-known (clinical)
problems would attract the attention of the healthcare workers and thereby help change their
mindset.7 Therefore, the research presented in this thesis was performed on the frontline using
healthcare worker perspectives. In addition, since the patient is in the centre of the healthcare
chain, and the only one who sees and experiences all aspects of it, we deemed it valuable
to incorporate his/her opinion when formulating research and improvement strategies. We
hypothesized that using the perspectives of all stakeholders in the acute healthcare chain to
assess daily safety issues faced in the field would create new insights, awareness and potential
strategies to improve patient safety.
SUMMARY OF MAIN FINDINGSIn chapter 2 we provided a systematic literature overview describing the effects of an
acute medical unit on patient outcomes.8 In addition, we reviewed the current situation in
the Netherlands and made recommendations that could be used when implementing an AMU.
The review demonstrated that current literature proves the AMU to be an effective model to
improve patient flow in the acute care chain. However, the current situation in the Netherlands
showed that existing AMUs were started up individually and are therefore heterogeneously
organized. In order to optimize the effectiveness and compare the quality of care in our AMUs,
a clear national guideline is needed providing a gold standard, especially since the results of
this paper demonstrated that more and more AMUs will be developed in the coming years in
the Netherlands. In addition, we found that to date, very few studies have been performed
168
on the effectivity of the AMUs in the Netherlands. More evidence based research might prove
useful to learn from each other.
In chapter 3, we performed a feasibility study using Patient Reported Outcome Measures
(PROMs) to measure the perceived quality of care and feeling of safety in an AAU.9 The results
of this study showed that the measurement of generic PROMS in the acute admission unit,
with a heterogeneous patient population, is feasible. In general, most patients were positive
about their feeling of safety during their stay on the AAU and regarding the contribution
of the AAU to their recovery. Patients were positive about the questionnaire since they felt
the hospital really wanted to know about their wellbeing, and their perspectives on received
quality and safety of care. They also reported that the follow-up telephone call gave them
the opportunity to share their experiences and reflect on their recent admission. Healthcare
workers benefited from this study too since the PROMs data collection and analysis could
be performed easily making it possible to report the results to the nurses and physicians on
the ward rapidly. This feedback motivated them to keep up their good work and improve
patient care. However, since this was the first (small) study done assessing PROMs in an acute
admission unit future research warrants larger sample sizes to investigate the use of PROMs to
compare hospitals, establish reference values and determine ideal follow-up times.
In chapter 4 and 5 we assessed whether deteriorating patients on clinical wards were
recognized timely by healthcare workers and use of screening instruments such as early
warning scores could predict and prevent serious adverse events in these patients.
PRISMA-medical is an analysis tool which investigates the root causes of adverse events and
incidents in healthcare.10,11 It has shown to be an effective starting point for improvement
strategies in patient safety. In chapter 4 we identified healthcare worker-, organizational-,
technical-, disease- and patient-related causes that contributed to 50 acute unplanned ICU
admissions.12 We found that almost half of the root causes were human- (healthcare worker)
related, predominantly including human monitoring and intervention failures, indicating flaws
in monitoring the patients progress or condition and faulty task planning and performance.
The other half of the root causes were disease-related, comprising the root causes related to
the natural progression of the disease, as could be expected in this severely ill population with
a high mortality rate. We also investigated the recognition of these deteriorating patients on
the clinical wards and found that only in 1% of the vital sets an explicit correct ‘MEWS’ was
reported, although in 43% of the measurements patients had a critical score. This withholds
clinical implications, since it seems essential that the hospital staff in the chain recognize these
critical scores so that the deteriorating patients are detected and treated on time. A proper
implementation of the early warning score described in chapter 5 could prove beneficial. This
study was performed before adequate reimplementation of this protocol, the study described
in chapter 5 was performed afterwards.
In chapter 5 we determined protocol adherence and predictive value of a rapid response
system, with a main focus on the effectuation of the afferent limb, assessed by the use of
169
MEWS on the wards.13 This study exposed that protocol adherence to the MEWS protocol was
good (89%), however one-third of the critical scores were calculated wrongly. Results of this
study indicate that a critical MEWS at 8 a.m. in the morning was a strong predictor of adverse
events associated with poor patient safety: in-hospital mortality, hospital length of stay, cardiac
arrests, ICU-admissions and unplanned 30-day readmissions. This indicates the reliability of
a once a day MEWS measurement as a screening tool: ‘A MEWS in the morning, a very
good warning’.
In chapter 6-10 we evaluated the clinical use and reliability of quality indicators to assess
patient safety. The two quality indicators chosen (unplanned readmissions and Standardized
Hospital Mortality Ratio) are regarded as ‘major adverse events’ and are thought to indicate
the quality of care in Dutch hospitals.14 A great deal of resistance exists in healthcare workers
who find these measures unconnected to their actual work on the floor, too time consuming
and bureaucratic.1,14,15
PRISMA-analysis was used in chapter 6 to create a root causal profile of 50 unplanned
readmissions.16 After composing the root causal trees the two researchers judged if
the readmission could have been prevented. After analysis, half of the readmissions were
deemed potentially preventable. Non-preventable readmissions were mostly the result of
disease-related factors. Healthcare worker-related root causes were solely found in preventable
readmissions, mostly due to human related coordination failures representing failures in task
coordination within a healthcare team, for example no clear medication handover towards
the home situation. This study underlines the importance of communication and coordination
within the chain to potentially prevent adverse events.
In the commentary in chapter 7 we elaborate our perspectives regarding readmissions
and argue why readmission as a quality indicator in its current form cannot be regarded
as a reliable way of assessing quality of care.17 The indicator which is presently used does
not integrate the aspect of preventability and overgeneralizes readmissions not correcting for
hospitals’ case-mix. It remains a struggle for policy makers to compose a reliable indicator since
readmissions are multi-causal events. In addition, the reasons for readmissions are often not
solely related to the course of events in hospital during index admission but to the bottlenecks
in the whole acute care chain.
In chapter 8 we performed an international study to assess the degree of agreement among
physicians regarding predictability and preventability of medical readmissions using a survey
of eight real-life scenarios.18
The results from the 526 participating physicians illustrated that there was moderate to
good agreement among physicians on the predictability of readmissions while agreement on
preventability was poor. This study therefore questions the conclusions derived from current
literature on the basis of physician opinion and advices to interpret these with caution.
170
Chapter 9 describes a multicentre study which included 1398 patients in Europe. During this
study we made an inventory of the opinions of readmitted patients, their carers, nurses, and
physicians regarding the predictability and preventability of readmissions and subsequently used
their opinions to find which factors were associated with their judgements on predictability and
preventability. We demonstrated that (1) consensus among the interviewed on predictability
and preventability was poor (2) factors associated with deemed predictability and preventability
also distinctly differed. However, if a patient reported not feeling ready for discharge, this
was significantly associated with a higher predictability and preventability of readmissions by
the majority. This study has implications for the practice since it emphasizes the importance
of communication during the discharge process. Healthcare workers should question patients
timely about their readiness to go home before discharge in order to potentially reduce
(preventable) readmissions.
In chapter 10 of this thesis we compared the HSMR to clinical disease severity scores in
a cohort of deceased patients with the diagnosis pneumonia. Furthermore, we assessed
the completeness of the data collected by administrators to calculate the HSMR. Results
reveal that risk categories calculated using clinical disease severity scores for pneumonia were
significantly higher than the mortality risk calculated by the HSMR. In addition, we found
the completeness of the data sent to the Central Bureau for Statistics underestimated the actual
comorbidities, and could possibly influence HSMR. Therefore, HSMR should be interpreted
with caution and openly publishing HSMRs may have unfair negative consequences for
some hospitals.
GENERAL DISCUSSION: WHAT CAN WE LEARN FROM THESE RESULTS This thesis has assessed patient safety from a clinical point of view. We aimed to provide an
insight in the whole healthcare chain and have investigated the bottlenecks in the places
where care is transferred from one link to the next in the chain.
System approach
Patient safety is defined as: ‘preventing errors and adverse effects to patients associated with
health care’.5 Results of this thesis have revealed that most adverse events have multifactorial
causes and an easy remedy to prevent them does not exist. The hospital is a high risk
environment and adverse events or incidents occur if when individual barrier weaknesses align
and the system as a whole fails, as was illustrated by the Swiss Cheese model of James Reason
(Figure 1).19,20 This causes “a trajectory of accident opportunity”, in which the hazard passes
through all of the holes in all of the defences, leading to a failure. The complex and risky
environment in healthcare in which no clear cause-effects exists, pleads for a system approach
to improve patient safety. Root cause analysis tools such as the PRISMA-analysis (used in
this thesis) have been found to be useful for such an approach since it aims to investigate
whole system failures of adverse events comprising patient safety by simply asking the ‘why’
question.10,11
171
Soft skills and behaviour
Another lesson learned in this thesis is that even though healthcare is rapidly evolving in this
golden age of technology and digital evolution, this does not automatically guarantee safer
environments. Our work has shown that most adverse events are still the result of human-
(healthcare worker) related failures, namely poor communication and coordination within
the chain, which are often caused by overloaded medical teams and a culture of silence in
hospitals. As the pressure on the healthcare system is not expected to decrease in the coming years, and the ongoing fragmentation in care will result in more handovers, more awareness
for these ‘soft skills’ and behaviour is warranted. Clear handover methods are required to
transfer information adequately and completely.3,21,22 In addition, to change the culture in
the wards all healthcare workers within the chain should feel that they can have a say about
the course of events of a patient’s journey. The major feature of mistakes is that they repeat
themselves and in order to prevent repetition all people involved should be able to ‘stop-the-
chain’ and call out the (potential) harm on the lurk.23
Beneficial safety implementations
Since the first report of EMGO-NIVEL in 2004, patient safety has been an important subject
of attention in the Dutch healthcare system. Despite the fact that not all interventions and
improvement strategies to enhance patient safety have been found useful in clinical practice
and resistance exists among healthcare workers, this thesis has shown that an adequately
implemented monitoring instrument can actually predict and possibly prevent adverse
events.14,22,24 One of our PRISMA-studies (chapter 4) revealed that monitoring patients
adequately is essential to prevent unplanned ICU-admission. Nevertheless, to implement
Track and Trigger Systems sufficiently, proper training of the medical staff is required, as was
Figure 1 | The Swiss cheese model by James Reason19,20
172
illustrated in chapter 4. Nurses and physicians working with these systems must be aware
of the effectiveness of these tools in preventing adverse events so that the adherence to
the guidelines can be improved. Another focal point to improve patient safety is the stimulation
of multidisciplinary work in hospitals to create a more efficient patient flow. The research in
this thesis has shown that the implementation of measures such as acute medical units, which
are designed to improve multidisciplinary teamwork and effectuate rapid diagnostics and
therapy, are an effective and safe method to streamline the healthcare chain. Also, patients on
these units were positive about these wards and felt safe on them.
The reliability of quality indicators
Despite positive effects of the increased attention to safety, this thesis has also exposed not
all of these safety implementations are thought to be beneficial by the healthcare workers in
the field and therefore, lack the support of these essential stakeholders. Due to the increased
focus on safe hospital environments by the government, the Inspectorate of Health and
accreditation bodies (i.e. JCI25), clinicians feel under constant pressure to perform perfectly
and feel they are held personally responsible for the adverse events. This results in a reluctance
to talk openly about potential improvement strategies to prevent these incidents in the future,
and does not set the right example for other healthcare workers (in training) around them.
In addition, they feel that the quality indicators they ought to report on, which are supposed
to measure quality and safety in hospitals, do not actually do so.1,15,26,27 Many clinicians feel
the quality indicators do not focus on clinical actions undertaken during their daily work and
are actually far from real patient outcomes since they often do not integrate the ‘end-users’
during the formulation of these indicators. This thesis has underlined this by demonstrating
that the current quality indicators ‘readmission rate’ and ‘HSMR’ in its current form are
probably unreliable since they do not integrate clinical severity of disease of an individual
patient and often do not correct for case-mix. These quality indicators tend to overgeneralize
and shift the attention from the patient to ‘the system’. This has a negative effect as with
this approach an individual assessment of a patient’s journey becomes less important. All
patients have a unique story which should be addressed in its own way - one size does not
fit all. The research performed on the quality indicators presented in this thesis was mostly
qualitative and small-scale, with the focus on these individual stories, since adverse events and
unsafe situations are fortunately still rare. Therefore, we plead for a cautionary interpretation
of the quality reports on these indicators which are currently open to public.
The added value of stakeholder perspectives
Studies in this thesis have also taught us that integrating the perspectives of patient and
healthcare worker can contribute to new insights. As was hypothesized, their experiences
were found to be valuable resources for identifying their needs and evaluating patient
safety outcomes.5 During our work we gathered the opinions of patients who had and had
not suffered harm, to not solely focus on the caveats, but to also gain knowledge about
the protective factors.1 The added value of incorporating patients’ perspectives is exemplified
by the CURIOS@-study in which we asked the opinions of 1398 patients and their direct
173
carers about their readmission. It has been shown that the extensive amount of research on
readmissions performed to date has mainly focused on measurable ‘risk factors’ while the only
issue which was found to be significantly relevant for the patient was whether he or she was
feeling ready for discharge. This shifts back attention to our social skills and to the soft side of
medicine. Communication in the medical world still seems to be the key, but do we ever really
ask all the relevant questions on the bedside?
Methodologic considerations
Performing research in the acute healthcare chain is not straightforward. During the course of
this trajectory we have faced some methodologic challenges. As we dealt almost exclusively
with a heterogeneous hospital population in our studies and we investigated safety procedures
in only one centre for some of our studies, this might make results less generalizable.
Thinking of a proper study design to answer our research questions was not always easy since
the research performed in this thesis aimed to capture new perspectives on patient safety
and not many studies with validated methodologies on this theme have been performed.
To overcome this, we sought help of the many experts at an early stage to assist us with
drafting an appropriate research protocol. In addition, the recruitment of patients for research
in an acute setting was sometimes challenging. Patients who have just been admitted are
often thrown out by new information after having spoken to multiple therapists. Yet another
person (researcher) at their bedside requesting to participate in a study can sometimes just be
too overwhelming, and therefore these patients might not be willing to grant consent. Also,
admitted patients can be severely ill or cognitively impaired and not suitable for inclusion.
Excluding these groups may result in selection bias. Another consideration when interpreting
of our work is the fact that part of our conclusions are based on retrospective data analysis
of the patient records. This means researchers had to rely solely on information that was
written down in the patient records when investigating causes for adverse events. This might
have underestimated the failures caused by cultural or organizational behaviours which are
often not presented in written charts. Furthermore, the limitations in our multicentre studies
should be noted. The physicians answering our survey of adapted real-life cases reflected that
they were missing information that would allow them to assess the case more thoroughly,
such as the patients’ social situation. A major limitation of the multicentre readmission study
performed was the potentially (recall)bias and subjectivity. In addition, a newly composed
variable ‘majority opinion’ was used as the gold standard to decide on preventability and
predictability. One might argue that this is questionable.
FUTURE PERSPECTIVESStart with facing the facts, valuable leadership from bottom-up
In order to enhance patient safety we must start with the end-users in the field. The hospital
remains an environment in which ‘humans’ work, and unfortunately, humans make errors.
Nevertheless, these humans almost always do work with best intentions. Yet, they sometimes
do not get the best results, mainly because the system they work in often does not allow them
174
to do so.28 The Berwick-committee has underlined this in their report ‘crossing the quality’
chasm which was a reaction to the report “To err is human”. In this report they stated that
simply trying harder does not work: a redesign of the healthcare system is needed.29,30 This
report focused on chief executives in healthcare to motivate them to incorporate new best
practices. This implementation is not done overnight however, it takes an average of 17 years
to broadly spread best practices in healthcare. These findings are especially important for
hospital directors, who carry the final responsibility for a proper safety system in hospitals and
should disseminate it adequately. It is known that poor leadership is one of the most significant
factors contributing to the occurrence of incidents and adverse events in healthcare.31
Therefore, these medical leaders should know and understand the field to design the most
suitable safety system and to address current bottlenecks. A way to do this is by making
the leaders ‘walk the talk’, an initiative by dr. Kaplan, an internist and the CEO of the Virginia
Medical Centre. Directors should literally walk through the hospital to see its daily functioning,
in order to show concern and be visible on the work floor. This initiative has proven to be
effective in the University Medical Centre of Utrecht.1 The concept could bridge the hierarchic
gap between the medical staff on the work floor and the hospital directors.23 In addition,
the dialogue created is the start of a bottom-up approach in contrast with the top-down
approach most healthcare workers presently endure. Conversing on a micro level will create
new insights into current obstacles and potential improvement strategies.1
Adapt approach to safety, from safety 1 to safety 2
Safety implementations have improved patient safety in Dutch hospitals.32 However, due to
increased attention to this subject, patients and healthcare workers actually feel less safe –
this is called the safety paradox.33 Even though we have improved in patient safety matters,
we are not ‘done’ and cannot sit still, as was underlined by the work in this thesis. Firstly, it
Figure 2 | Our perspectives on improvement strategies in patient safety
175
must be emphasized that the improvement of patient safety is an ongoing cycle, which needs
constant attention and evaluation (Figure 3).5 Secondly, the coming years will probably be
even more challenging considering patients will only become older and more frail, technology
more advanced and working together more complicated.32 Considering increasing demands
and growing system complexity, we must therefore adjust our approach to patient safety.34
The current approach to patient safety, labelled safety-1, is established on the ‘find and fix’
model. This causes a linear model of cause and effect instead of the preferred system approach
we have discussed earlier. It is essential to learn from the far more frequent cases where
things go right and develop ways to support, augment and encourage these. In order to
find out what does work, a shift to a proactive safety-2 management could prove helpful in
the future.1,7,34 It concentrates on how everyday performance usually succeeds rather than
on why it occasionally fails, and actively strives to improve the former rather than simply
preventing the latter. In this way we will prevent degrading the resources and procedures
needed to make things go right.7,34
Face the hierarchic culture, from extrinsic to intrinsic motivation
An integrated, positive and transparent culture on wards is an essential factor to improve
patient safety. The culture on hospital wards is still mostly created by the medical specialists
working on these wards. Even though hierarchical gaps have closed over the past decades,
medical specialists are still at the top of the medical hierarchy and are seen as central role
players. Yet, specialists often do not feel they are part of the hospitals organization since
a mismatch/gap between hospital directors and medical specialists exits.1 In most specialisms
a culture of silence and not addressing colleagues exists. This prevents medical staff from
learning from each other and speaking openly about their doubts and (almost) failures. When
an incident does occur and everyone starts talking about this, the potentially responsible
Figure 3 | Patient safety cycle, World Health Organization 20035
176
healthcare worker may feel that he is the only one who is doing something wrong, resulting
in a ‘blame and shame’ culture. A culture change providing attitude changes towards patient
safety in professionals is needed to increase self-reflection and shift the attention on patient
safety from extrinsic to intrinsic motivation. Intrinsic motivation entails activities which are
undertaken because of internal motivation and provide immediate effect after the action
itself whereas extrinsic motivation makes people undertake actions to get a reward or to
dodge a penalty from an external party without being interested in the immediate potential
effect. An organization-wide reduction of adverse events can be achieved by high levels of
ward’s shared values, beliefs and behaviours plus an individual’s perception of contributing
to the culture.6 Soft skills such as communication and team training for non-technical skills
within the healthcare chain are key focal points. To improve these behavioural skills and to
change the culture, education should be provided from the early onset of a medical career.
Unfortunately, the time and resources reserved for such trainings is increasingly under pressure
in the last decades because of budget cuts. The tendency is to choose efficiency and high
production over investment in quality for the future.
E-health: Match supply and demand. Which way to go?
Due to new technologies more care can be delivered at home in the future. E-health is the use
of technology to support or improve a patient’s health and the healthcare system. It offers
patients the ability to direct and self-manage their own health. The use of e-consultation,
virtual monitoring systems and therapy at home (i.e. chemotherapy) might be beneficial
to relieve the pressure in hospitals and to make patients more aware of their own health
situation. A massive amount of novel technologies is released daily which state they could
potentially enhance the quality of care. A problem with these new inventions is that they are
often created by an external company with commercial intentions and look good on paper,
but it is known that they have not all been designed using insights from the frontline. It is
necessary for healthcare institutions to work together to match these inventions to the actual
needs on the floor. In addition, since these strategies might decrease the hours in hospital,
insurance companies and hospitals should think of novel ways to pay for this care. Finally, we
must not forget that even though these new flashy methods seem beneficial, they do not
substitute the need for proper communicational skills which are still essential for safe care.
Also, less modern inventions such as a consultation on the telephone and proper discharge
letters should not be left out in seeking for alternative care methods.35
Rely on patient’s journeys: Talk with them, not about them
Research has indicated that better patient experiences are associated with better clinical
outcomes, less healthcare utilization and better patient safety within hospitals.36 However,
current patient experience has shown little association with hospital quality management
strategies.37 To match supply and demand, more patient perspectives could be helpful. As
was highlighted in this thesis, patients are often willing to participate, however, barriers do
still exist. The traditional hierarchy between patient and doctors still makes some patients
feel subordinate to their clinician. This is an important barrier to their involvement in error
177
reduction.38 In addition, patients might feel they are being labelled ‘difficult’ and therefore
choose to take a more passive role in protecting their own personal safety. To enhance their
active role, frameworks have been developed by error victims to describe general courses of
action by which patients can contribute to their safety and increase resilience. These include:
1) informing the management plan by sharing information with clinicians and asking questions
about treatment decisions, 2) monitoring and ensuring safe delivery of treatment for example
by self-administration of medication and 3) informing systems improvement for example by
providing feedback on care quality.39
Another helpful tool that could be constructive to investigate patient experiences is
the usage of ‘patient tracers’, which are originally used by consultants and auditors from
the JCI (Joint Commission International).1,25 In this method the travelled path of a patient is
followed to stepwise investigate and reconstruct what would happen in an ideal situation,
who would carry responsibility and what would be needed (structure) to provide optimal
quality of care. The principal ‘tell me, show me, and take a good look’ is used to evaluate
the course through the healthcare chain. A tracer looks into the whole system and the process
(the healthcare chain) in contrast to individual acting and enhances the ‘safety 2’ approach.7
In the Netherlands, the nationwide network for acute care has recently reported and tracked
11 patient journeys in the acute chain. An executive from an insurance company and a patient
(representative) have successfully worked together on this assessment, and have found areas for
potential improvement.40
Personal reflection
On a final note, I would like to elucidate my personal reflection on the results of this thesis.
During the course of this PhD trajectory I have learned a lot about research practices, patient
safety and scientific writing. In my opinion, the thesis presented assesses safety issues from
a new perspective, addressing it in a more practical way as it was assessed by ‘insiders’. Some
might argue that results from our studies are logical findings, and causes for adverse events
are widely known by all healthcare workers. However, we noticed that our new and ‘fresh’
research strategies in combination with the quick feedback of results to healthcare workers
created an unexpected Hawthorne effect.42 This effect, which is also referred to as the observer
effect, is a type of reactivity in which individuals modify an aspect of their behaviour in response
of their awareness of being observed (participating in our research). During the research
performed, awareness was created in healthcare workers to actually actively think and reflect
with us about patient safety matters and the effectiveness of current safety implementations
and quality indicators. The interest in this research field was also underlined by the great
number of young medical students and trainees that were motivated to help us to perform our
studies. The attention and time they spent on participating patients was greatly appreciated,
and during their work they not only learned research basics but also gained knowledge and
awareness of the relevance of patient safety. Hopefully, this thesis will function as a starting
point to improve patient safety in the overloaded acute healthcare chain by initiating the shift
of extrinsic motivation to the required intrinsic motivation in the healthcare system.
178
REFERENCES1. Dillman R, Wagner C, Schellekens W, Klein J, Jager M, van der Grinten T. Patientveiligheid. Utrecht:
de Tijdstroom uitgeverij. 2016
2. Arora VM, Farnan JM. Inpatient Service Change: Safety or Selection? JAMA. 2016;316(21):2193-2194.
3. Greenstein EA, Arora VM, Staisiunas PG, Banerjee SS, Farnan JM. Characterising physician listening behaviour during hospitalist handoffs using the HEAR checklist. BMJ Qual Saf. 2013;22(3):203-209.
4. Pannick S, Beveridge I, Wachter RM, Sevdalis N. Improving the quality and safety of care on the medical ward: A review and synthesis of the evidence base. Eur J Intern Med. 2014;25(10):874-887.
5. World Health Organization. Patient Safety. http://www.euro.who.int/en/health-topics/Health-systems/patient-safety/patient-safety. Accessed January 3, 2017.
6. Mardon RE, Khanna K, Sorra J, Dyer N, Famolaro T. Exploring relationships between hospital patient safety culture and adverse events. J Patient Saf. 2010;6(4):226-232.
7. Braithwaite J, Wears RL, Hollnagel E. Resilient health care: turning patient safety on its head. Int J Qual Health Care. 2015;27(5):418-420.
8. van Galen LS, Lammers EM, Schoonmade LJ, Alam N, Kramer MH, Nanayakkara PW. Acute medical units: The way to go? A literature review. Eur J Intern Med. 2016.
9. Galen LS, der Schors W, Damen NL, Kramer M, Wagner C, Nanayakkara P. Measurement of generic patient reported outcome measures (PROMs) in an acute admission unit: A feasibility study. Acute medicine. 2016;15(1):13-19.
10. van Vuuren W SC, van der Schaaf TW. The Development of an Incident Analysis Tool For the Medical Field. Eindhoven University of Technology: Eindhoven. 1997.
11. van der Schaaf TW HM. PRISM-Medical. A Brief Description. Eindhoven University of Technology, Faculty of Technology Management, Patient Safety Systems: Eindhoven. 2005.
12. van Galen LS, Struik PW, Driesen BE, et al. Delayed Recognition of Deterioration of Patients in General Wards Is Mostly Caused by Human Related Monitoring Failures: A Root Cause Analysis of Unplanned ICU Admissions. PLoS One. 2016;11(8):e0161393.
13. van Galen LS, Dijkstra CC, Ludikhuize J, Kramer MH, Nanayakkara PW. A Protocolised Once a Day Modified Early Warning Score (MEWS) Measurement Is an Appropriate Screening Tool for Major Adverse Events in a General Hospital Population. PLoS One. 2016;11(8):e0160811.
14. Basisset Kwaliteitsindicatoren 2014. Utrecht: Inspectie voor de gezondheidszorg, 2015. Accessed January, 2016.
15. Poortmans J. Onderzoek - Zorgkwaliteit in ziekenhuizen. Driemaaldaags een pijnschaal. De Groene Amsterdammer. 2016(49).
16. Fluitman KS, van Galen LS, Merten H, et al. Exploring the preventable causes of unplanned readmissions using root cause analysis: Coordination of care is the weakest link. Eur J Intern Med. 2016;30:18-24.
17. van Galen LS, Nanayakkara PW. [Hospital readmissions: A reliable quality indicator?]. Ned Tijdschr Geneeskd. 2016;160:A9885.
18. van Galen LS, Cooksley T, Merten H, et al. Physician consensus on preventability and predictability of readmissions based on standard case scenarios. Neth J Med. 2016;74(10):434-442.
19. Reason J. Human error: models and management. BMJ. 2000;320(7237):768-770.
20. Reason J. James Reason: patient safety, human error, and Swiss cheese. Interview by Karolina Peltomaa and Duncan Neuhauser. Qual Manag Health Care. 2012;21(1):59-63.
21. O’Leary KJ, Buck R, Fligiel HM, et al. Structured interdisciplinary rounds in a medical teaching unit: improving patient safety. Arch Intern Med. 2011;171(7):678-684.
179
22. Veiligheidsmanagementsysteem. Communicatie tussen hulpverleners volgens het SBAR-proces. VMS Veiligheidsprogramma. 02-2009.
23. Interview with Dr. Gary S. Kaplan: Determined Steps to Transformation. http://ethix.org/2011/01/11/dr-gary-s-kaplan-determined-steps-to-transformation. Published 2011. Accessed February 16, 2017.
24. Langelaan M, de Bruijne M, Baines R, et al. NIVEL Monitor Zorggerelateerde Schade 2011/2012. Dossieronderzoek in Nederlandse ziekenhuizen. http://www.nivel.nl/sites/default/files/bestanden/monitor_zorggerelateerde_schade_2011_2012.pdf. Published 2013, Accessed December 28, 2016.
25. The Joint Comission. https://www.jointcommission.org/performance_measurement.aspx. Accessed February 2, 2017.
26. Ikkersheim D, van der Avoort J. KPMG Plexus. Onderzoek kosten kwaliteitsmetingen Nederlandse Vereniging van Ziekenhuizen (NVZ). https://www.nvz-ziekenhuizen.nl/_library/31906 Published 2015. Accessed January 10, 2017.
27. Botje D, Ten Asbroek G, Plochg T, et al. Are performance indicators used for hospital quality management: a qualitative interview study amongst health professionals and quality managers in The Netherlands. BMC Health Serv Res. 2016;16(1):574.
28. Holusha J. W. Edwards Deming, Expert on Business Management, Dies at 93. http://www.nytimes.com/1993/12/21/obituaries/w-edwards-deming-expert-on-business-management-dies-at-93.html?pagewanted=all. Accessed February 16, 2017.
29. Berwick DM. A user’s manual for the IOM’s ‘Quality Chasm’ report. Health Aff (Millwood). 2002;21(3):80-90.
30. Institute of Medicine. To err is human: building a safer health system. http://www.nationalacademies.org/hmd/~/media/Files/Report%20Files/1999/To-Err-is-Human/To%20Err%20is%20Human%201999%20%20report%20brief.pdf. Published November 1999. Accessed December 28, 2016.
31. Leisikov I. Patientveiligheid, de rol van de bestuurder (PhD thesis TU Delft). 2010. Accessed February 15, 2017.
32. Veiligheid Nederlandse ziekenhuizen aanzienlijk verbeterd. NFU. http://www.nfu.nl/actueel/veiligheid-nederlandse-ziekenhuizen-aanzienlijk-verbeterd. Published 2013. Accessed February 17, 2017.
33. Boutelier H. Boom Lemma Uitgevers. De veiligheidsupopie. Hedendaags onbehagen en verlangen rond misdaad en straf. 2005.
34. Hollnagel E, Wears RL, Braithwaite J. From Safety-I to Safety-II: A White Paper. https://www.england.nhs.uk/signuptosafety/wp-content/uploads/sites/16/2015/10/safety-1-safety-2-whte-papr.pdf. Published 2015, Accessed January 10, 2017.
35. Car J, Sheikh A. Telephone consultations. BMJ. 2003;326(7396):966-969.
36. Anhang Price R, Elliott MN, Zaslavsky AM, et al. Examining the role of patient experience surveys in measuring health care quality. Med Care Res Rev. 2014;71(5):522-554.
37. Groene O, Arah OA, Klazinga NS, et al. Patient experience shows little relationship with hospital quality management strategies. PloS one. 2015;10(7):e0131805.
38. Doherty C, Stavropoulou C. Patients’ willingness and ability to participate actively in the reduction of clinical errors: a systematic literature review. Soc Sci Med. 2012;75(2):257-263.
39. Peat M, Entwistle V, Hall J, Birks Y, Golder S. Scoping review and approach to appraisal of interventions intended to involve patients in patient safety. J Health Serv Res Policy. 2010;15 Suppl 1:17-25.
40. Landelijk Netwerk Acute Zorg. Uitkomsten bespreking patient journeys. Samenvattend overzicht. Published December 2016. Accessed February 17, 2017.
CHAPTER 12
DUTCH SUMMARY – NEDERLANDSE SAMENVATTING PATIËNTVEILIGHEID IN DE ACUTE ZORGKETEN:
“IS HET THUIS VEILIGER?”
‘The art of medicine consists of amusing the patient while nature cures the disease’ Voltaire
183
PATIËNTVEILIGHEID IN DE ZORG Dit proefschrift gaat over het inzichtelijk maken en verbeteren van patiëntveiligheid
in Nederlandse en Europese ziekenhuizen. Het onderzoek in dit proefschrift beschrijft
nieuwe inzichten en potentiele verbeterstrategieën ten behoeve van veiligheidskwesties in
de acute zorgketen.
Door toenemende vergrijzing , de marktwerking en aanhoudende bezuinigingen in de zorg
neemt de werkdruk op ziekenhuizen en zorgverleners toe. Ondanks een afname van
spoedeisende hulp presentaties in de laatste jaren, is er een toename van het aantal opnames.
Hierdoor zijn ziekenhuizen vaak vol en loopt de patiëntenstroom op de spoedeisende hulp
dagelijks vast. Met name de toename van stroom van oudere patiënten zorgt voor opstopping
omdat deze meer en vaak langere toegepaste zorg vergt door hun multimorbiditeit, polyfarmacie
en atypische presentatie van klachten. Huisartsen trachten acute patiëntenstromen indien
mogelijk af te wenden uit het ziekenhuis door patiënt zelf te behandelen, op te nemen in
een eerstelijns verblijf of te verwijzen voor een poliklinisch consult, zij functioneren als een
poortwachter. Helaas zijn door de toename van sociale hulpvragen van met name oudere
patiënten (‘het gaat niet meer thuis’) in combinatie met het terugschroeven van budget voor
verpleegtehuizen ook de huisartsen in de zorgketen vaak overbelast.
Communicatie in en coördinatie binnen de zorgketen zijn essentieel voor de stroomlijning
ervan, echter komen deze factoren door de huidige belasting op het zorgstelsel in het geding.
Hierdoor komt patiëntveiligheid potentieel in gevaar en kunnen ‘Serious Adverse Events’
(ernstige ongewenste voorvallen) zoals ongeplande heropname, onverwachte intensive care
opname of in het ergste geval overlijden het gevolg zijn. Patiëntveiligheid wordt gedefinieerd
als: het vermijden, voorkomen, en verminderen van ongewenste uitkomsten of lichamelijk
letsel ten gevolge van het zorgproces. Ondanks toegenomen aandacht voor en projecten ter
bevordering van patiëntveiligheid komen deze ongewenste uitkomsten nog steeds te vaak
voor. Zo krijgt 1 op elke 10 patiënten te maken met ongewenste schade tijdens zijn traject
door de zorgketen.
Om de patiëntveiligheid te bevorderen is er een cultuurverandering nodig in ziekenhuizen.
Zorgverleners en bestuurders dienen zich bewust te zijn van hun rol in het systeem, en
verantwoordelijkheid te nemen ten behoeve van de veiligheid van de patiënt. Door de grote
administratieve last van kwaliteitsindicatoren van externe partijen hebben zorgverleners het
gevoel dat het daadwerkelijke probleem op de vloer niet goed in kaart wordt gebracht, dan
wel gemeten. De motivatie voor het goed uitvoeren van zorg lijkt nu vooral ‘het aankruisen
van extern opgelegde hokjes’ in plaats van intrinsiek willen dat de patiënt daadwerkelijk in een
veiligere omgeving zorg krijgt.
De belangrijkste speler binnen patiëntveiligheid blijft uiteraard de patiënt. Te vaak echter,
wordt er in onderzoek en hervorming van de zorg, niet de opinie gevraagd van deze essentiële
stakeholder. De patiënt blijft de enige persoon die het traject van begin tot einde in diepgang
184
meemaakt, en hij zal dan ook waardevolle inzichten hebben ten behoeve van de verbetering
van zorg.
In dit proefschrift zijn de inzichten gebruikt van de stakeholders die op de frontlinie betrokken
zijn bij (bijna) incidenten in het ziekenhuis, met een grote focus op de patiënt en zijn
zorgverleners. Onze hypothese was dat het gebruik van deze perspectieven zou leiden tot
nieuwe inzichten, verbeterstrategieën, en tevens meer bewustwording onder zorgverleners
zou creëren.
DE START VAN DE ACUTE ZORGKETEN: NIEUWE MANIEREN VAN ZORG ORGANISEREN In het eerste deel van dit proefschrift bestudeerden wij nieuwe manieren om het eerste deel van
de acute zorgketen voor patiënten effectiever en veiliger te maken. In de afgelopen decennia
zijn er verschuivingen geweest in het originele model van opname via de spoedeisende hulp
naar (gewone) klinische afdelingen. Acute opname afdelingen (AOA), bestaan in Nederland
sinds 2000. Het zijn afdelingen die bemand en uitgerust zijn om patiënten op te nemen met
een acuut medische aandoening voor een snelle multidisciplinaire en medisch specialistische
evaluatie. De zorg op deze afdeling is voor een bepaalde periode (meestal tussen 24 en 72
uur) waarna patiënten naar huis worden ontslagen, dan wel naar gewone klinische afdelingen
worden overgeplaatst. In hoofdstuk 2 beschreven wij in een systematisch literatuuroverzicht
de positieve effecten van het implementeren van een AOA op patiënt uitkomsten zoals
opnameduur en patiënttevredenheid. Ook gaven wij de huidige stand van zaken in Nederland
weer en deden wij een aanbeveling voor een nationale richtlijn. In hoofdstuk 3 werden
patiënten gedurende hun opname op de AOA middels gevalideerde vragenlijsten (PROMS:
Patient Reported Outcome Measures) gevraagd naar hun ervaringen en gevoel van veiligheid.
Patiënten op deze afdeling voelden zich veilig en de resultaten van het onderzoek konden
direct teruggekoppeld worden naar de verpleegkundigen en artsen op de werkvloer, wat hen
motiveerden deze goed gewaardeerde zorg te blijven leveren.
VERMIJDEN VAN ONGEWENSTE VOORVALLEN OP DE KLINISCHE AFDELINGEN Ondanks het feit dat niet alle geïmplementeerde interventies en strategieën voor het verbeteren
van patiëntveiligheid gebruiksvriendelijk en bijdragend worden bevonden in de klinische
praktijk, toont dit proefschrift dat een adequaat ingevoerd instrument daadwerkelijk kan
bijdragen aan het voorspellen en mogelijk voorkomen van ongewenste uitkomsten. In
hoofdstuk 4 identificeerden wij factoren die bijdroegen aan 50 ongeplande Intensive Care
opnames. Middels een PRISMA-analyse werd gevonden dat voornamelijk menselijk gerelateerde
monitoringsfouten een oorzaak waren voor deze ongewenste uitkomsten. Doordat patiënten
op de klinische afdelingen niet adequaat werden bewaakt, was er vaak te laat herkenning
van hun achteruitgang, waardoor overplaatsing naar de Intensive Care noodzakelijk was.
In hoofdstuk 5 toonden wij het effect aan van een adequaat geïmplementeerde ‘Early
185
Warning Score’, een score die is ontwikkeld om deze patiënten tijdig op te sporen om verdere
achteruitgang te voorkomen. Deze tool zorgt dat er bij een afwijkende score tijdig een arts
wordt gewaarschuwd, waardoor eerder de juiste zorg wordt geleverd. Een afwijkende score
was in de heterogene ziekenhuispopulatie een goede voorspeller van ongewenste uitkomsten,
wat de betrouwbaarheid van dit instrument benadrukt. 89% van de meldingen werd volgens
protocol uitgevoerd, desalniettemin werd nog steeds een aanzienlijk deel van kritieke scores
gemist, hier is nog ruimte voor verbetering.
HET GEBRUIK VAN KWALITEITSINDICATOREN OM PATIËNTVEILIGHEID TE METEN EN WAARBORGEN In Nederland zijn er momenteel ongeveer 3400 kwaliteitsindicatoren in de gezondheidszorg.
Er is veel weerstand onder zorgverleners over deze indicatoren: ze staan niet in verbinding met
het echte werk op de vloer, zijn tijdsintensief, en te administratief. Het percentage heropnames
binnen 30 dagen na een eerdere klinische opname in hetzelfde ziekenhuis is samen met een
onverwacht lange opnameduur en overlijden in het ziekenhuis, één van de ‘major adverse
events’ die gelden als indicator voor een negatieve uitkomst van klinische ziekenhuiszorg in
Nederland. In de Verenigde Staten en het Verenigd Koninkrijk is het percentage heropnames al
enige jaren één van de indicatoren voor de kwaliteit en veiligheid van ziekenhuizen. Een hoog
percentage heropnames kan in deze landen resulteren in een geldboete.
Dit proefschrift trekt het huidige gebruik van kwaliteitsindicatoren ‘heropnames’ en ‘overlijden
in het ziekenhuis’ in twijfel, en beargumenteert dat de gepresenteerde cijfers van deze
indicatoren vaak geen betrouwbare afspiegeling zijn van de daadwerkelijke veiligheid en
kwaliteit van geleverde zorg in ziekenhuizen.
Middels een PRISMA-analyse in hoofdstuk 6 maakten wij van 50 heropnames een oorzaken-
risicoprofiel. Dit onderzoek toonde aan dat vermijdbare heropnames met name veroorzaakt
worden door menselijke coördinatie-fouten, bijvoorbeeld geen goede medicatieoverdracht naar
de thuissituatie. Deze studie benadrukte het belang van adequate en efficiënte communicatie
en coördinatie in de zorgketen om zo mogelijk vermijdbare heropnames te voorkomen. Het
opiniestuk in hoofdstuk 7 becommentarieert waarom de heropname in zijn huidige vorm geen
betrouwbare indicator is van kwaliteit van zorg. Het waarheidsgetrouw vormgeven van deze
indicator is een opgave, want heropnames hebben vaak meerdere oorzaken, binnen en buiten
(de macht van) het ziekenhuis. De huidige indicator houdt geen rekening met vermijdbaarheid,
en generaliseert ziekenhuizen door niet te corrigeren voor hun ziektezwaarte. In hoofdstuk 8
hebben wij gekeken of medisch specialisten (in opleiding) het met elkaar eens zijn over de
vermijdbaarheid van heropnames. Dit is gedaan omdat in de huidige literatuur conclusies
met name berust op de opinie van een klein aantal artsen. Dit onderzoek liet zien dat de
526 artsen die deelnamen het niet eens waren over de vermijdbaarheid van de heropname.
Er was wel consensus over de patiënten die een hogere kans hadden om heropgenomen te
worden. Dit onderzoek geeft aan dat huidige literatuur met een kritisch oog bekeken moet
worden en dat vermijdbaarheid opnemen in de kwaliteitsindicator ‘heropname’ een lastige
186
taak is. Relatief de meeste onderzoekuren in dit proefschrift werd verricht in onderzoek dat
wordt gepresenteerd in hoofdstuk 9. In dit onderzoek vroegen wij 1398 patiënten in 15
Europese ziekenhuizen gedurende hun heropname naar de reden van hun heropname. Ook
vroegen wij een mantelzorger, verpleegkundige en arts van het behandelteam naar hun
perspectieven betreffende dit ongewenste incident. Het onderzoek concludeerde dat met name
de communicatie tussen arts en patiënt tijdens ontslag erg belangrijk was. Als patiënten zich
niet klaar voelden voor ontslag gedurende hun eerste opname, was de kans significant groter
dat zij terugkwamen en dat dit vermijdbaar was dan als zij wel klaar waren. Dit onderzoek
plaats een kritische noot bij de veelvuldige heropname-predictiemodellen. Deze modellen
nemen statische factoren mee om een heropname te voorspellen, dit onderzoek toonde aan
dat met name ‘zachte factoren’ zoals communicatie en empathie belangrijk zijn.
De ‘Hospital Standardized Mortality Ratio’, is een ratio die de werkelijke sterfte in een
ziekenhuis afzet tegen de sterfte die op basis van de patiëntkenmerken wordt verwacht. Het
fungeert als indicator voor potentieel vermijdbare sterfte. In hoofdstuk 10 zetten wij deze
openbaar gepubliceerde cijfers in een ander perspectief. De studie, uitgevoerd in 32 patiënten
overleden aan een longontsteking, demonstreerde dat de risico scoren die klinisch gebruikt
worden om ziekte-ernst van de patiënten met longontsteking te beoordelen significant hogere
sterftecijfers gaven dan de sterftecijfers die werden berekend door de HSMR. Daarnaast
onthulde deze studie dat de volledigheid van gegevens die naar het Centraal Bureau voor
Statistiek gaan de werkelijke comorbiditeit onderschatte. Deze bevindingen kunnen de HSMR
negatief beïnvloeden, en de vraag is dan ook of deze openbaar gepubliceerde cijfers een
werkelijke weergave zijn van de kwaliteit van zorg.
WAT KUNNEN WIJ MET DEZE RESULTATEN? Dit proefschrift heeft vanuit een klinisch perspectief de knelpunten in de acute zorgketen
benaderd. De resultaten hebben ons het volgende geleerd:
→ Het onderzoeken van patiëntveiligheid vereist een systeembenadering in plaats
van een persoon- of ‘schakel’-benadering. Een ongewenste uitkomst vindt vaak
pas plaats als een aaneenschakeling van fouten in hetzelfde traject, in dit geval
de keten, plaatsvindt.
→ De meeste ongewenste uitkomsten vinden nog steeds plaats door foutief menselijk
handelen, met name inadequate en inefficiënte coördinatie en communicatie in
de zorgketen zijn belangrijke factoren voor incidenten. Veilige en gestructureerde
overdracht is nodig om kwaliteit van zorg te waarborgen.
→ Iedereen in de zorgketen moet zich kunnen uitspreken over een eventuele ‘bijna’
fout en de mogelijke consequenties. Dit betekent dat de hiërarchische structuur
plaats moet maken voor gelijkwaardig hoor en wederhoor onder zorgverleners en
ondersteunende functies in de zorgketen.
→ Om veiligheidsstrategieën in een zorgsysteem goed te implementeren is
goede training en besef van het belang van deze instrumenten in de dagelijkse
praktijk nodig.
187
→ Huidige kwaliteitsindicatoren kunnen een goed middel zijn om verder intern
onderzoek te verrichten als cijfers afwijken van de norm. Zij moeten echter niet
dienen als openbaar vergelijkingsmiddel, met name omdat de meeste indicatoren
geen betrouwbare weerspiegeling van de daadwerkelijke kwaliteit en veiligheid is.
→ Het meenemen van de perspectieven van alle stakeholders in de zorgketen
zorgt voor nieuwe inzichten en potentiele verbeterslagen ten behoeve van
patiëntveiligheid. Daarnaast creëert het betrekken van zorgverleners awareness en
intrinsieke motivatie voor manieren om veiligheid daadwerkelijk te verbeteren.
→ De ‘zachte’ kanten (zoals communicatie en empathie) van menselijk handen zijn
vaak lastig te meten, deze eigenschappen zijn echter vaak doorslaggevend voor
veilige en patiëntvriendelijke zorg.
HET BEVORDEREN VAN PATIËNTVEILIGHEID IN DE ACUTE ZORGKETEN IN DE TOEKOMST In de afsluitende woorden van dit proefschrift doen wij een aanbeveling voor het bevorderen
van patiëntveiligheid in de toekomst. Enkele kernpunten:
→ De medische leiders (bestuurders, managers) in het ziekenhuis moeten de feiten
onder ogen zien, zij zijn immers eindverantwoordelijk in een ziekenhuis. Dit betekent
dat zij de werkvloer op moeten om in gesprek te gaan met hun werknemers en
patiënten om zo te kunnen zien waar verbetering nodig is. Ziekenhuizen moeten
een bottom-up benadering kiezen, in plaats van een top-down.
→ Door toenemende huidige focus op patiëntveiligheid is er sprake van een
veiligheidsparadox: wij voelen ons minder veilig. Daarom zou veiligheid meer
bestudeerd moeten worden vanuit de momenten dat het wel goed gaat, in plaats
van die ene keer dat het (bijna) mis ging. Dit zorgt voor een positievere benadering
van het probleem en omarmt beschermende factoren.
→ Door de ouderwetse hiërarchische structuur in ziekenhuizen wordt cultuur op
afdelingen vaak nog bepaalt door medisch specialisten. Hierdoor worden mensen
vaak niet aangesproken op normen, waarden en gedrag die patiëntveiligheid
potentieel in gevaar kunnen brengen. Communicatie binnen het team en niet-
technische skills moeten getraind worden in teamverband om deze ouderwetse
cultuur te doorbreken.
→ Zorgverleners moeten intrinsiek gemotiveerd zijn om bij te dragen aan
patiëntveiligheid zodat zij meteen het effect van hun handeling kunnen zien. Dit in
tegenstelling tot de nu vaak aanwezige extrinsiek motivatie: mogelijke beloningen
en boetes door overkoepelende instanties die bepaalde indicatoren behartigen.
→ Technologische innovaties om zorg meer buiten het ziekenhuis te laten
plaatsvinden zullen in de komende jaren zijn opmars maken. Deze zien er op
papier vaak veelbelovend uit, echter zijn ze vaak niet ontworpen met inzichten van
de zorgverleners op de vloer. Daarnaast kunnen deze E-health mogelijkheden niet
de zachte kanten, zoals communicatie en empathie overnemen. Het gebruik zal in
de praktijk door de eindgebruikers getest en gevalideerd moeten worden.
188
→ Positieve patiëntervaringen leiden tot betere uitkomsten. Om kwaliteit
daadwerkelijk te meten, moeten wij de input van patiënten gebruiken om in te zien
wat zij belangrijk vinden. Een manier om dit te doen is bijvoorbeeld door middel
van ‘patiënt tracers’ waarin samen met de patiënt het gehele pad (patient journey)
van ervaringen door de zorgketen wordt doorlopen.
APPENDIX
LIST OF PUBLICATIONS LIST OF SCIENTIFIC PRESENTATIONS
AUTHOR AFFILIATIONS WORD OF THANKS
CURRICULUM VITAE/BIOGRAPHY
193
LIST OF PUBLICATIONS 2015
– Cooksley T, Nanayakkara PW, Nickel CH, Subbe CP, Kellett J, Kidney R, van Galen LS et al. (7-2015) Readmissions of medical patients: an external validation of two existing prediction scores. QJM
– Arntzenius AB, van Galen LS. (11-2015) Budesonide related adrenal insufficiency. BMJ Case reports
2016 – Fluitman KS & van Galen LS, Merten H, Cooksley T, Nanayakkara PW, Kramer MH et al. (1-2016)
Exploring the preventable causes of unplanned readmissions using root cause analysis: Coordination of care is the weakest link. Eur J Intern Med.
– van Galen LS, Nanayakkara PW. (1-2016) Hospital readmissions: A reliable quality indicator? Ned Tijdschr Geneeskd.
– van Galen LS, van der Schors W, Damen NL, Kramer MH, Wagner C, Nanayakkara PW. (3-2016) Measurement of generic patient reported outcome measures (PROMS) in an acute admission unit: A feasibility study. Acute Med.
– van Galen LS & Dijkstra CC, Ludikhuize J, Kramer MH, Nanayakkara PW. (8-2016) A Protocolised Once a Day Modified Early Warning Score (MEWS) Measurement Is an Appropriate Screening Tool for Major Adverse Events in a General Hospital Population. PLoS One
– van Galen LS & Struik PW, Driesen BE, Merten H, Ludikhuize J, van der Spoel JI, Kramer MH, Nanayakkara PW. (8-2016) Delayed recognition of deterioration of patients in general wards is mostly caused by human related monitoring failures: A root cause analysis of unplanned ICU admissions. PLoS One
– van Galen LS, Cooksley T, Merten H, Brabrand M, Terwee CB, Nickel CH et al. (10-2016) Physician Consensus on preventability and predictability of Readmissions based on standard case scenarios. Neth J med.
– van Galen LS, Lammers EM, Schoonmade LJ, Alam N, Kramer MH, Nanayakkara PW (11-2016) Acute Medical Units: The way to go? A literature review. Eur J Intern Med.
2017 – van Galen LS & Wachelder J. (4-2017) Cruising through the journey without getting drowned:
The saga of a PhD student in the Netherlands. Acute Med.
– Kidney R, Sexton E, van Galen LS, Silke B, Nanayakkara P, Kellet J. (4-2017) Hospital Readmissions - Independent Predictors of 30-day Readmissions derived from a 10 year Database. Acute Med.
– van Galen LS, Brabrand M, Cooksley T, van de Ven PM, Merten H, So RK et al. (5-2017) Patients’ and providers’ perceptions of the preventability of hospital readmission: a prospective, observational study in 4 European countries. BMJ Qual Saf.
LIST OF PUBLICATIONS
194
LIST OF SCIENTIFIC PRESENTATIONS – Cooksley T, Nanayakkara P, Nickel CH, Subbe CP, van Galen LS, Kellett J, Kidney RM, Merten H,
Vaughan L, Brabrand M. *On behalf of the safer@home consortium. Predicting readmissions – Existing scoring models of little use to acute physicians? Poster presentation at the Society of Acute Medicine Annual Spring Conference, 2015, Bristol, United Kingdom.
– van Galen LS, Subbe CP, Brabrand M, Cooksley T, Kellett J, Kidney RM, Merten H, Nickel CH, Vaughan L, Nanayakkara P.*On behalf of the safer@home consortium. Study proposal: CURIOS@. Poster presentation at the Society of Acute Medicine Annual Spring Conference, 2015, Bristol, United Kingdom.
– van Galen LS & Nanayakkara PWB. Study protocol CURIOS@: CaptUring Readmissions InternatiOnally to prevent readmissions by safer@home consortium. Oral presentation at the 4th annual Dutch Acute Medicine congress, 2015, Rotterdam, the Netherlands.
– van Galen LS & Dijkstra CC, Ludikhuize J, Kramer MH, Nanayakkara PW. A Protocolised Once a Day Modified Early Warning Score (MEWS) Measurement Is an Appropriate Screening Tool for Major Adverse Events in a General Hospital Population. Oral presentation at the 5th annual Dutch Acute Medicine congress, 2016, Enschede, the Netherlands.
– van Galen LS, Cooksley T, Merten H, et al. *On behalf of the safer@home consortium. Physician consensus on preventability and predictability of readmissions based on standard case scenarios. Oral and poster presentation at Society of Acute Medicine Annual autumn Conference, 2016, Edinburgh, Schotland.
– van Galen LS & Dijkstra CC, Ludikhuize J, Kramer MH, Nanayakkara PW. A Protocolised Once a Day Modified Early Warning Score (MEWS) Measurement Is an Appropriate Screening Tool for Major Adverse Events in a General Hospital Population. Poster presentation at the Society of Acute Medicine Annual autumn Conference, 2016, Edinburgh, Schotland.
– van Galen LS & Fluitman KS, Merten H, et al. *On behalf of the safer@home consortium. Exploring the preventable causes of unplanned readmissions using root cause analysis: Coordination of care is the weakest link. Poster presentation at Society of Acute Medicine Annual autumn Conference, 2016, Edinburgh, Schotland.* Prize for best poster.
– van Galen LS, Brabrand M, Cooksley T, van der Ven PM, Merten H, So, RK, van Hoof L, Haak HR, Kidney RM, Nickel CH, Soong J, Weichert I, Kramer MH, Subbe CP, Nanayakkara PWB. *On behalf of the safer@home consortium. Preliminary results from a prospective European study on predictability and preventability of readmissions using patient/healthcare worker perspectives (CURIOS@). Oral presentation at the Society of Acute Medicine Annual autumn Conference, 2016, Edinburgh, Schotland.
– van Galen LS & Dijkstra CC, Ludikhuize J, Kramer MH, Nanayakkara PW. A Protocolised Once a Day Modified Early Warning Score (MEWS) Measurement Is an Appropriate Screening Tool for Major Adverse Events in a General Hospital Population. Selected as ‘paper of the month’ by the director of the Swiss Safety Foundation, October 2016.
LIST OF SCIENTIFIC PRESENTATIONS
195
AUTHOR AFFILIATIONSNadia Alam, Section acute medicine, Department of Internal Medicine, VU Medical Centre,
Amsterdam, the Netherlands
Mikkel Brabrand, Department of Emergency Medicine, Hospital of South West Jutland,
Esbjerg, Denmark
Nikki L. Damen, NIVEL Netherlands Institute for health services research, Utrecht,
The Netherlands
Ellen P. Claessens, Section acute medicine, Department of Internal Medicine, VU Medical
Centre, Amsterdam, the Netherlands
Tim Cooksley, Department of Acute Medicine, University Hospital of South Manchester,
Manchester, United KingdomCasper C. Dijkstra, Section acute medicine, Department of Internal Medicine, VU Medical
Centre, Amsterdam, the Netherlands
Babiche E. Driesen, Section acute medicine, Department of Emergency Medicine, VU Medical
Centre, Amsterdam, the Netherlands
Jelle A. van Erven, Section acute medicine, Department of Internal Medicine, VU Medical
Centre, Amsterdam, the Netherlands
Kristien Fluitman, Department of Internal Medicine, VU Medical Centre, Amsterdam,
the Netherlands; Sahlgrenska Academy, Institute of Medicine, Gothenburg University,
Gothenburg, Switzerland.
Louise S. van Galen, Section acute medicine, Department of Internal Medicine, VU Medical
Centre, Amsterdam, the Netherlands
Harm Haak, Department of Internal Medicine, Maxima Medisch Centre, Eindhoven/Veldhoven,
The Netherlands; Department of Internal Medicine, Division of General Internal Medicine,
Maastricht University Medical Centre Maastricht University, Maastricht, The Netherlands
Asselina A. Hettinga-Roest, Section acute medicine, Department of Internal Medicine, VU
Medical Centre, Amsterdam, the Netherlands
Loes van Hooff, Department of Emergency Medicine, VieCuri Hospital, Venlo, The Netherlands
Rachel M. Kidney, St. James Hospital, Dublin, Ireland
Mark H. Kramer, Department of Internal Medicine, VU Medical Centre, Amsterdam,
the Netherlands
Eline M. Lammers, Section acute medicine, Department of Internal Medicine, VU Medical
Centre, Amsterdam, the Netherlands
Jeroen Ludikhuize, Department of Anaesthesiology, Academic Medical Centre, Amsterdam,
the Netherlands
Hanneke Merten, Department of Public and Occupational Health, EMGO Institute for Health
and Care Research, VU University Medical Centre, Amsterdam, The Netherlands
Prabath W. Nanayakkara, Section acute medicine, Department of Internal Medicine, VU
Medical Centre, Amsterdam, the Netherlands
Christian H. Nickel, Department of Emergency Medicine, University Hospital Basel,
Basel, Switzerland
Saskia M. Rombach, Department of Internal Medicine, VU University Medical Centre,
Amsterdam, The Netherlands
AUTHOR AFFILIATIONS
196
Jan C. Roos, Bureau of medical affairs, VU University Medical Centre, Amsterdam,
the Netherlands
Linda J. Schoonmade, Medical library, VU University, Amsterdam, the Netherlands
Wouter van der Schors, NIVEL Netherlands Institute for health services research, Utrecht,
The Netherlands
John Soong, Imperial College London, United Kingdom
Ralph K. So, Department of Quality, Safety & Innovation, Albert Schweitzer Hospital,
Dordrecht, The Netherlands
Johannes I. van der Spoel, Department of Intensive Care, VU University Medical Centre,
Amsterdam, The Netherlands
Patricia W. Struik, Section acute medicine, Department of Internal Medicine, VU Medical
Centre, Amsterdam, the Netherlands
Christian P. Subbe, Ysbyty Gwynedd Hospital, Wales, United Kingdom
Caroline B. Terwee, Department of Public and Occupational Health, EMGO Institute for
Health and Care Research, VU University Medical Centre, Amsterdam, The Netherlands
Louella Vaughan, Department of Medicine and Therapeutics, Chelsea and Westminster
Hospital, London, United Kingdom
Peter M. van de Ven, Department of Epidemiology and Biostatistics, VU University Medical
Centre Amsterdam, Amsterdam, The Netherlands
Cordula Wagner, VU University Medical Centre, Amsterdam, The Netherlands; NIVEL
Netherlands Institute for health services research, Utrecht, The Netherlands; Department of
Public and Occupational Health, EMGO Institute for Health and Care Research, VU University
Medical Centre, Amsterdam, The Netherlands
Immo Weichert, The Ipswich Hospital NHS Trust, Ipswich, United Kingdom
198
WORD OF THANKS
WORD OF THANKS
199
ALEXARNTZENIUSALINAOFFRINGAAMYVANGALENANNAEMANUEL
ANNEFLEURKOOPENANNELIESBROUWER
ANNELIESKABALTANOUSCHKAPRONK
BARBARABOSBASVANLIESHOUT
BOBOSCASPERDIJKSTRACHANTALWIEPJES
CHRISSUBBECHRISTELDEBLOKCHRISTIANNICKEL
CONNECTCARCORDULAWAGNER
DAISYVEDDERDANIELVANRAALTEDAVIDBRINKMANDEMUSONTWERPT
DENISEHOOGEVEENDENNISBARTEN
DIDIBRINKEDINKLEGEMATE
DUTCHACUTEMEDICINEEDMÉESCHRIJVERELINELAMMERSELLEMIJNKUIPER
ERIKVANBOMMELEVADURINCK
FRITSHOLLEMANHANNEKEMERTEN
ICARVUIRISDENIE
JANINESTOLWIJKJANKLEIN
JELLETICHELAARJELLEVANERVEN
JEROENLUDIKHUIZEJESBOS
JESKEVANDIEMENJESSICABEKEMA
JOERITIJDINKJONNESIKKENS
JORINDEKUIPERSJORISVANDERVORSTJORNWOERDEMAN
JOSIPCAR
JOYCEWACHELDERKARINKAASJAGER
KEWAJAMJAARNULACHTKEWAJENNES
KOENVANBEERSKRISTIENFLUITMAN
LENNARTTONNEIJCKLIEKEGLAUDEMANSLISETTEACKERMANS
LKCMEDICINELYNNVANDIJK
MAARTENVANDERBIEMAARTJEKLAVER
MAMAMARCELMUSKIET
MARIEKEDIEPENBROEKMARIEKEVANNOORTWIJK
MARIJESCHIPPERMARKHOLLANDMARKKRAMERMARTINECARIS
MCDONALDFAMILYMICHAELREUMERMANMICHIELVANAGTMAEL
MIKKELBRABRANDNADIAALAMNANALORNA
NICKDEJONGENIENKENOTA
PAPAPHILIPBOS
PRABATHNANAYAKKARARACHELKIDNEY
RISHINANNANPANDAYRSGVRIENDINNETJES
SAFERATHOMECONSORTIUMSECTIEFARMACOTHERAPIE
SHANEVANGALENSIETSKEVANNASSAU
SOCIETYACUTEMEDICINESTIENEKEDOORNWEERD
TABITHADEGRAAFFTIMCOOKSLEYTIMSCHUTTETOMBOEIJE
VANGALENFAMILIEWESSELFUIJKSCHOT
WILMAJANSENWOUTERVANDERSCHORS
YVOSMULDERS
200
CURRICULUM VITAE / BIOGRAPHYLouise Sandra van Galen was born on 27th of June
1989 in Hengelo (OV), the Netherlands. She is half-
Irish. Louise finished her secondary school in Harderwijk
(RSG ‘t Slingerbos), where she graduated cum laude in
2007. In secondary school she also received her diploma
‘International Baccalaureate English A2 higher level’. She
moved to Amsterdam at the age of 18 to start her medical
training at the VU University Medical Centre. During her
study she worked for the BIS Foundation in Leiden as
a transplantation employee and went abroad to Aruba for
her clinical internship in surgery. Louise received her medical
degree in the summer of 2014. Subsequently, she worked
as a junior doctor in internal medicine at the Westfries Gasthuis in Hoorn for eight months
after which she started her PhD on patient safety in the acute healthcare chain at the VUmc
under supervision of dr. Prabath Nanayakkara and professor Mark Kramer. The research
projects in her PhD attempt to work on research questions originating from daily clinical issues
on the workfloor. She was the coordinating researcher for the first European prospective
readmission study ‘CURIOS@’ (“CaptUring Readmission data InternatiOnally by Safer@home
group”). This is a research project that originated from the safer@home consortium, an
international group founded in 2013 consisting of 13 acute medical physicians, emergency
physicians and epidemiologists from Europe that focus on readmissions and safer discharge
processes. Results of her PhD research are presented in this thesis. During her PhD, Louise
also worked as a staff member for the section pharmacotherapy and developed and taught
pharmacotherapy education to medical students. Beside her medical duties, Louise has several
functions in voluntary work such as ambassador for the Hubrecht Institute in Utrecht, as
advisor in the Amsterdam city centre panel, and as a board member of the Medical Business
Education Committee. After her PhD Louise hopes to become a specialist in Acute Internal
Medicine, and to prolong her research in the acute healthcare chain field. In September 2017
Louise will start a post-doc fellowship at Nanyang University in Singapore under supervision of
dr. Josip Car in his health services outcomes research group.
Louise lives in Amsterdam with her partner, Philip Bos.
CURRICULUM VITAE / BIOGRAPHY
201
SPONSORS
top related