trends in recorded capillary blood glucose and hypoglycaemia in hospitalised patients with diabetes

Trends in recorded capillary blood glucose andhypoglycaemia in hospitalised patients withdiabetes

G.C. Jones a,*, H. Casey a, C.G. Perry a, B. Kennon b, C.A.R. Sainsbury a

aDepartment of Diabetes, Gartnavel General Hospital, Glasgow G11 0YN, United KingdombDepartment of Diabetes, Southern General Hospital, Glasgow G51 4TF, United Kingdom

d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e 1 0 4 ( 2 0 1 4 ) 7 9 – 8 3

a r t i c l e i n f o

Article history:

Received 29 August 2013

Received in revised form

15 October 2013

Accepted 17 January 2014

Available online 25 January 2014

Keywords:

Capillary blood glucose

Near patient testing

Hypoglycaemia

Inpatient

Diabetes

a b s t r a c t

Aims: To utilise whole-system analysis of capillary glucose measurement results to exam-

ine trends in timing of glucose monitoring, and to investigate whether these timings are

appropriate based on observed patterns of hypoglycaemia.

Methods: Near-patient capillary blood glucose results from eight acute hospitals collected

over 57 months were analysed. Analysis of frequency of measurement, and measurements

in the hypoglycaemic (<4 mmol/l) and severe hypoglycaemic (<2.5 mol/l) range per time of

day was made.

Results: 3 345 241 capillary glucose measurements were analysed. 1 657 594 capillary blood

glucose values were associated with 106 624 admissions in those categorised as having

diabetes. Large peaks in frequency of glucose measurements occurred before meals, with

the highest frequency of capillary glucose measurement activity being seen pre-breakfast.

Overnight, an increase in measurement activity was seen each hour. This pattern was

mirrored by frequency of measured hypoglycaemia. 27 968 admissions (26.2%) were asso-

ciated with at least one hypoglycaemic measurement. A greater proportion of measure-

ments were within the hypoglycaemic range overnight with 61.7% of all hypoglycaemia

between 2100 and 0900 h, with peak risk of measured capillary glucose being hypoglycaemic

between 0300 and 0400 h.

Conclusions: Hypoglycaemic is common with the greatest risk of hypoglycaemia overnight

and a peak percentage of all readings taken being in the hypoglycaemic range between 0300

and 0400 h. Measurement activity overnight was driven by routine, with patterns of

proportion of measurements in the hypoglycaemic range indicating that there may be a

significant burden of undiscovered hypoglycaemia in the patients not routinely checked

overnight.

Contents available at ScienceDirect

Diabetes Researchand Clinical Practice

journal homepage: www.elsevier.com/locate/diabres

1. Introduction

Capillary blood glucose monitoring is performed routinely in

hospital to manage patients at risk of hyperglycaemia and

hypoglycaemia and point of care capillary blood glucose

* Corresponding author. Tel.: +44 141 211 3259.E-mail address: g.jones3@nhs.net (G.C. Jones).

0168-8227/$ – see front matter # 2014 Elsevier Ireland Ltd. All rights

http://dx.doi.org/10.1016/j.diabres.2014.01.021

measurement can be considered as analogous to an additional

‘‘vital sign’’ for hospitalised patients with diabetes.

Diabetes is common amongst hospital inpatients in the

United Kingdom, Europe and North America being thought to

be present in between 1 in 5 and 1 in 10 hospital stays [1–3].

reserved.

Diabetes has been unequivocally shown to negatively impact

both mortality and length of stay in hospital in a variety of

medical and surgical settings prompting an increasing focus

on improving inpatient glycaemic control [2,4–7].

In hospitalised patients with diabetes, hypoglycaemia is

common, and in patients both with and without diabetes its

frequency and severity has been associated with an increase

in pre and post discharge mortality and length of stay [8–10].

The ACE/ADA Task Force on Inpatient Diabetes in 2006

stressed the importance of glycaemic control and raised

concerns regarding under and over treatment of glycaemia

citing insufficient frequency of glucose monitoring as a

potential safety issue [11]. The 2013 ADA standards of medical

care for patients with diabetes suggest glucose monitoring in

all people with, and at risk of, diabetes citing the importance of

glycaemic control and hypoglycaemia avoidance. Guidance on

frequency of glucose monitoring by point of care testing is only

given for patients treated with insulin with between 4 and 6

hourly to ½ hourly to 2 hourly for those on intravenous insulin

suggested [12]. In the UK the National Diabetes Inpatient Audit

has recommended that glucose is checked weekly for stable

diet controlled patients and 4 times a day for unwell patients,

unstable diabetes or those treated by basal bolus insulin

therapy [13].

The Greater Glasgow and Clyde Health Board in Scotland

performs around 700 000 near patient capillary blood glucose

tests per year. The purpose of this study is to utilise whole

system analysis of these results to examine current trends in

timing of monitoring and assess if these timings are appro-

priate based on observed patterns of hypoglycaemia.

2. Materials and methods

Capillary blood glucose data were extracted from the Abbott

Precision Webb system [Abbott, UK]. Whole-hospital data

were included from eight acute hospitals and associated units

within Greater Glasgow and Clyde Health Board (Gartnavel

General Hospital, Western Infirmary, Glasgow Royal Infirmary,

Victoria Infirmary, Southern General Hospital, Royal Alexan-

dria Hospital, Inverclyde Hospital and Vale of Leven Hospital)

between dates 01/01/2008 and 01/05/2013. Data collected from

intensive care environments were excluded as it was felt likely

that the capillary glucose monitoring in these areas would be

very different from general medical and surgical wards. Data

available from the system included patient identifier, patient

location, test date and time and capillary glucose value.

All data manipulation and analysis was performed using

bespoke code written in statistical language R [14].

Prior to analysis of the complete data set it became

apparent that a number of capillary blood glucose measure-

ment values were not attributable to an identifiable patient

and were removed.

We wished to limit our analysis as much as possible to data

from patients with diabetes and we were aware that a

proportion of capillary blood glucose measurements are

performed as a routine admission test in people without

diabetes. Based on the World Health Organisation diagnostic

criteria for diabetes we assumed that either morning (between

0500 and 0700 h) capillary blood glucose of �7 mmol/l or 2

random capillary blood glucose of �11.1 mmol/l indicated

diabetes. This was felt to give the best possible dataset for

further analysis, consisting of subjects likely to have diabetes

without unduly excluding individuals with diabetes from

analysis.

Dates of admission and discharge from hospital were

implied following examination of the comparison with case

note review dataset. Any gap of five or more days between

capillary blood glucose measurements was taken as the

threshold for determining a new admission episode.

Analysis of frequency of glucose measurement per time of

day was made. A similar frequency analysis was made of

hypoglycaemic measurements (<4 mmol/l), and severe hypo-

glycaemic capillary glucose measurements (<2.5 mmol/l). We

used the National Health Service Diabetes guideline treatment

cut-off value (blood glucose values <4 mmol/l) to categorise

hypoglycaemia. Severe hypoglycaemia is best categorised by

the need for third-party assistance in treating the episode. As

this information was not available to us a value of <2.5 mmol/l

was used to describe severe hypoglycaemia [15]. Comparison

was made between the frequency of hypoglycaemic measure-

ments during daytime (0900–2100) and night time (2100–0900)

hours.

The number of capillary glucose measurement in the

hypoglycaemic range (<4 mmol/l) as a proportion of all

capillary glucose measurements was calculated for the

dataset as a whole and for each hour of the day.

We observed peaks and troughs of glucose measurement

activity overnight with peaks occurring at 0100, 0200, 0300, and

0400 h. We calculated the proportion of capillary blood glucose

measurements in the hypoglycaemic range for a 30 min

window centred on these peaks in activity, and also in the

corresponding 30 min activity troughs bordering the peaks.

We repeated the analysis of the four peaks in measurement

activity at 0100, 0200, 0300 and 0400 h excluding any measured

capillary blood glucose occurring within 1 h of any initial

measurement <4 mmol/l.

3. Results

The initial raw dataset consisted of 3 345 241 capillary glucose

measurements. 1 836 258 capillary glucose values were asso-

ciated with 60 256 valid patient identifiers, corresponding to

172 771 admission episodes over 57 months of data acquisi-

tion. 1 657 594 capillary blood glucose values were associated

with 106 624 admissions during which a diagnosis of diabetes

was made using glycaemic parameters. The remainder of the

discussion will consider those subjects in whom a glycaemic

diagnosis of diabetes was made.

4. Timing of capillary blood glucosemeasurements

Increases in measurement activity were identified with large

peaks in frequency of capillary blood glucose measurements

at pre-meal times. The highest frequency of capillary blood

glucose measurement activity was seen pre-breakfast, with

less pronounced peaks identified before lunchtime and

time (hour)

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 0

Fig. 1 – Frequency of capillary blood glucose measurement

per minute over 24 h day. Data from 28 697 individuals

determined to have diabetes (1 657 594 capillary blood

glucose measurements).

time (hour)0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 0nu

Fig. 2 – Frequency of capillary blood glucose measurements

<4 mmol/l per minute. Data from 11 174 individuals

(75 529 capillary blood glucose measurements).

d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e 1 0 4 ( 2 0 1 4 ) 7 9 – 8 3 81

evening meals. A diffuse increase in activity was observed

during the late evening. Overnight, frequency of testing fell.

Peaks of activity were observed at 0000, 0100, 0200, 0300, 0400

and 0500 h (Fig. 1).

5. Hypoglycaemia

Peaks of ascertainment of hypoglycaemiccapillary blood glucose

measurement (<4 mmol/l) were observed prior to mealtimes,

with the pre-breakfast peak being dominant, mirroring the

patterns identified in capillary blood glucose measurement

activity. Overnight, increases in hypoglycaemia identification

were observed at 0000, 0100, 0200, 0300, 0400 and 0500 h (Fig. 2).

27 968 admissions were associated with at least one

hypoglycaemic capillary blood glucose measurement, repre-

senting a proportion of 26.2%. A majority (46 654, 61.7% of all

time (hour)

Fig. 3 – Proportion of capillary blood glucose measurements <4

hypoglycaemic measurements) were obtained nocturnally

between the hours of 2100 and 0900 h. The large majority of

these were measured between the hours of 0600 and 0800

(23 423, 31.0% of all hypoglycaemic measurements). Of 11 792

severe hypoglycaemic measurements (<2.5 mmol/l), 6492

(55.1%) occurred between 2100 and 0900, again with the large

majority occurring in the early morning. 75 529 capillary blood

glucose measurements were in the hypoglycaemic range

(<4 mmol/l) representing 4.1% of the total number of capillary

blood glucose values measured. This proportion varied

throughout the day, with a greater proportion of capillary

blood glucose values being hypoglycaemic between 0000 and

0700 than between 0700 and 0000 h.

11 792 capillary blood glucose measurements were in the

severe hypoglycaemic range (<2.5 mmol/l) representing 0.6%

of the total number of capillary blood glucose values

measured. Again a greater proportion of measurements were

time (hour)

mmol/l (left panel) and <2.5 mmol/l (right panel) by hour of

in the severe hypoglycaemic range between 0000 and 0700 h

(Fig. 3).

Overnight capillary blood glucose encapsulating the four

peaks in measurement activity at 0100, 0200, 0300 and 0400 h

were examined. 7.1% of measurements obtained during 30 min

windows centred on the peaks were in the hypoglycaemic range

(<4 mmol/l) compared to 8.6% of the capillary glucose results

during the intervening trough periods. 1.4% of measurements

obtained during 30 min windows centred on the peaks were in

the severe hypoglycaemic range (<2.5 mmol/l) against 1.7% of

the capillary glucose results during the intervening trough

periods.

In the repeated analysis of the four peaks in measurement

activity at 0100, 0200, 0300 and 0400 h excluding any measured

capillary blood glucose occurring within 1 h of any initial

measurement <4 mmol/l. 5.6% of measurements obtained

during 30 min windows centred on the peaks were in the

hypoglycaemic range (<4 mmol/l) compared to 6.5% of the

capillary glucose results during the intervening trough periods.

6. Conclusions

Hypoglycaemia is a common problem in diabetic inpatients

and confers a significant risk in terms of morbidity and

mortality [2,4–7]. Monitoring of capillary blood glucose plays a

key role in identifying and preventing hypoglycaemia. Our

data set is the largest to be examined to describe trends in

timing of capillary blood glucose monitoring and frequency of

ascertainment of hypoglycaemia. Our data show that mon-

itoring appears to be done mainly before breakfast with

increased testing pre-meals and testing on the hour.

Our study shows that of all inpatient measured capillary

glucose 5.1% is in the hypoglycaemia range (<4 mmol/l). This

is consistent with previously described rates of 3.5–5.7% for

glucose <70 mg/dl (3.9 mmol/l) in the US for non-intensive

care patients [16,17] and similar to that described in the United

Kingdom by Kerry et al. of 4.9% [18].

It has been shown previously using continuous glucose

monitoring in type 1 and 2 diabetes that glucose is more often

low in the night time than daytime (7% compared to 11% of

readings) [19] and in the Diabetes Control and Complications

Trial severe hypoglycaemia was also more common at night

[20]. Our data confirm this overnight increased risk of

hypoglycaemia in an inpatient setting and we found 61% of

all hypoglycaemia occurred between 2100 and 0900 h which

compares with the findings of Kerry et al. who found that 70%

of all hypoglycaemia occurs in this time period [18]. In 206

diabetic inpatients Bailon et al. showed that 78% more

hypoglycaemic events occurred during night shift with most

occurring between 0400 and 0500 and 0600 and 0700 h [21]. It is

possible that the increased recorded hypoglycaemia in these

studies is at least partly explicable by ascertainment bias as

large proportion of all measured glucose is done at the

beginning of early nursing shift between 0600 and 0900 h in

the morning. In our own study we were able to confirm the

high risk of overnight hypoglycaemia with the absolute

number of hypoglycaemic episodes also highest between

0600 and 0700 h. Importantly we were also able to show a

higher proportion of all measured capillary blood glucose

reading to be hypoglycaemic between 0000 and 0700 h with the

peak proportion of recorded glucose levels in the hypogly-

caemic range occurring between 0300 and 0400 h.

In an attempt to further reduce ascertainment bias we

compared the hourly peaks of monitoring (which are likely to be

predominantly driven by routine and planned monitoring

rather than by acute clinical concerns or patient symptoms)

with troughs of monitoring activity. The difference in rate of

hypoglycaemia between these peaks and troughs of measure-

ment frequency are minor suggesting that overnight increased

hypoglycaemic frequency is unlikely to be caused solely by an

increased proportion of the tests being performed because of

symptoms reported by patients. Indeed the difference in rate of

capillary blood glucose found to be in hypoglycaemic range

reduced when removing tests performed within the hour

following uncovered hypoglycaemia. Even in the severe

hypoglycaemia group (<2.5 mmol/l) who would likely be most

symptomatic there was still minimal difference between peak

and trough rates. It appears that overnight routine capillary

blood glucose testing of patients will uncover a high rate of

hypoglycaemia albeit in a group of patients who have been

deemed high risk enough to require an hourly monitoring

regime and may be on intensive intravenous or subcutaneous

insulin therapy. These data suggest that there may be a

significant burden of undiscovered hypoglycaemia in the

patients not routinely checked overnight and this should be

considered in patients at high risk of hypoglycaemia. It would

appear that the optimal time to look for hypoglycaemia

overnight would be between 0300 and 0400 h. As pointed out

by Kerry et al. this increased risk of overnight hypoglycaemia

may be worsened by early hospital meals and lack of night time

carbohydrate snack availability which should be borne in mind

when planning hospital mealtimes [18].

Our data confirm that hypoglycaemia is common in

inpatients and suggests times where monitoring might be

focused. Future research is needed to evaluate patterns of

capillary blood glucose testing to better detect and prevent

hypoglycaemia.

Conflict of interest statement

r e f e r e n c e s

[1] Sampson MJ, Crowle T, Dhatariya K, Dozio N, GreenwoodRH, Heyburn PJ, et al. Trends in bed occupancy forinpatients with diabetes before and after the introductionof a diabetes inpatient specialist nurse service. Diabet Med2006;23:1008–15.

[2] Olveira-Fuster G, Olvera-Marquez P, Carral-Sanlaureano F,Gonzalez-Romero S, Aguilar-Diosdado M, Soriguer-EscofetF. Excess hospitalizations, hospital days, and inpatientcosts among People with diabetes in Andalusia, Spain.Diabetes Care 2004;27:1904–9.

[3] American Diabetes Association. Economic costs of diabetesin the U.S. in 2007. Diabetes Care 2008;31:596–615.

[4] Umpierrez GE, Isaacs SD, Bazargan N, You X, Thaler LM,Kitabchi AE. Hyperglycaemia: an independent marker of

d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e 1 0 4 ( 2 0 1 4 ) 7 9 – 8 3 83

inpatient mortality in patients with undiagnosed diabetes. JClin Endocrinol Metab 2002;87:978–82.

[5] Krinsley JS. Association between hyperglycaemia andincreased hospital mortality in a heterogeneouspopulation of critically ill patients. Mayo Clin Proc2003;78:1471–8.

[6] Frisch A, Chandra P, Smiley D, Peng L, Rizzo M, Gatcliffe C,et al. Prevalence and clinical outcome of hyperglycaemia inthe perioperative period in noncardiac surgery. DiabetesCare 2010;33:1783–8.

[7] Evans NR, Dhatariya KK. Assessing the relationshipbetween admission glucose levels, subsequent length ofhospital stay, readmission and mortality. Clin Med2012;12:137–9.

[8] Turchin A, Matheny ME, Shubina M, Scanlon J, GreenwoodB, Pedergrass M. Hypoglycemia and clinical outcomes inpatients with diabetes hospitalised in the general ward.Diabetes Care 2009;32:1153–7.

[9] Nirantharakumar K, Marshall T, Kennedy A, Narendran P,Hemming K, Coleman JJ. Hypoglycaemia is associated withincreased length of stay and mortality in people withdiabetes who are hospitalised. Diabet Med 2012;29:445–8.

[10] Tan HK, Flannagan D. The impact of hypoglycaemia onpatients admitted to hospital with medical emergencies.Diabet Med 2013;30:574–80.

[11] American College of Endocrinology and American DiabetesAssociation. Consensus statement on inpatient diabetesand glycemic control: a call to action. Diabetes Care2006;29:1955–62.

[12] American Diabetes Association. Standards of medical carein diabetes-2013. Diabetes Care 2013;36(Suppl. 1):S11–66.

[13] National Diabetes Inpatient Audit 2012. Published March2013; 2013, Available at http://www.hscic.gov.uk/ [lastaccessed July 2013].

[14] R Core Team. R: a language and environment for statisticalcomputing. Vienna, Austria: R Foundation for StatisticalComputing; 2012, ISBN: 3-900051-07-0, Available at http://www.R-project.org/ [last accessed July 2013].

[15] Joint British Diabetes Society. The hospital management ofhypoglycaemia in adults with diabetes mellitus. NewcastleUpon Tyne: Joint British Diabetes Society; 2010.

[16] Swanson CM, Potter DJ, Kongable GL, Cook CB. Update oninpatient glycaemic control in hospital in the United States.Endo Pract 2011;17:853–61.

[17] Cook CB, Konglable GL, Potter DJ, Abad VJ, Leija DE,Anderson M. Inpatient glycaemic control: a survey of 126US hospitals. J Hosp Med 2009;9:E7–14.

[18] Kerry C, Mitchell S, Sharma S, Scott A, Rayman G. Dailytemporal patterns of hypoglycaemia in hospitalised peoplemay reveal potentially correctable factors. Diabet Med 2014.http://dx.doi.org/10.1111/dme.12256.

[19] Bode BW, Schwartz S, Stubbs HA, Block JE. Glycemiccharacteristics in continuously monitored patients withtype 1 and type 2 diabetes: normative values. Diabetes Care2005;28:2361–6.

[20] The DCCT research Group. Epidemiology of severehypoglycaemia in the diabetes control and complicationstrial. Am J Med 1991;90:450–9.

[21] Bailon RM, Cook CB, Hovan MJ, Hull BP, Seifert KM, Miller-Cage V. Temporal and geographical patterns ofhypoglycaemia among hospitalised patients with diabetesmellitus. J Diab Sci Technol 2009;3:261–8.

trends in recorded capillary blood glucose and hypoglycaemia in hospitalised patients with diabetes

Documents

hypoglycaemia wilms' tumour* -...

microbiological testing of adults hospitalised with ... ·...

avoiding hypoglycaemia toolkit...y&h ahsn – avoiding...

89017 capillary kh lim brochure - dou yee...

hypoglycaemia final

legionnaires’ disease among patients hospitalised with...

hypoglycaemia in clinical_diabetes

impaired counterregulatory responses to hypoglycaemia...

neonatal hypoglycaemia overview & sugar babie's trial

management of diabetic emergencies:...

the hospital management of hypoglycaemia in adults · pdf...

hypoglycaemia diabetes outreach (june 2011). 2 hypoglycaemia...

multicentre study of hospitalised patients with sports

malnutrition in hospitalised patients and clinical ... ·...

column for capillary electrophoresis capillary

suicide and hospitalised self-harm in australia

hospital management of hypoglycaemia in adults with diabetes

hypoglycaemia in type 1 diabetes: technological treatments...

capillary thermostatting in capillary electrophoresis ·...

hospitalised child