The Incidence of Nosocomial Infection in theIntensive Care Unit, Hospital UniversitiKebangsaan Malaysia: ICU-acquiredNosocomial Infection Surveillance Program1998.. 1999

S W Rozaidi, J SUkl'O, A Dan, Department of Anaesthesiology & Intensive Care, Facultyof Medicine, Hospital Universiti Kebangsaan Malaysian, Jalan Tenteram, Bandar TunRazak, Cheras, 56000 Kuala Lumpur

Summary~

socomial infection (NI) remains one of the major causes of lCU mortality. This study'i<;!ence of lCU-acquired nosocomial infection in lCU HUKM for the years 1998 ancl 1999,

ongoing ICU·acquired nosocomial infection sUlveillance pJ"Ogram. The overall incidencewas 23%. The main types of NI was lower respiratoty tract infection 05.3%), primary bacteraemia(8.1%), ventilator associated pneumonia (5.4%), urinary tract infection (2.0%), skin infection 0.6%)central venous catheter sepsis (1.2%) and surgical skin infection (0.8%). The overall culture positive

'al infection rate was oniy 12.1%, ity from the lungs 02.6%), blood (7.3%), skin swabsd 0.6%). The main gra e organism culturecl was Acinetobacter sp. 09%) and

%) was the gra organism. The overalllCU mortality rate was 275% of whichatients who died were ttributed directly to sepsis.

leU·acquired nosocomial infection, Nosocomial infection, Intensive care unit,Incidence tion rate, Microorganism, Lower respiratory tract infection,Ventilator iated pneumonia, Primary bacteraemia, Skin infection,Epidemiolo

Introduction

One of the major causes of leu mortality andmorbidity is nosocomial infection (NI)1.2. Theseinfections not only adversely affect the lCUpatient's outcome but they also pose a financialburden to both parients ancl hospital as wellincreasing overall medical costS·\,4.

Med J Malaysia Val 56 No 2 June 200 I

There arc several ways of decreasing theincidence of NI in the ICU'. The Srudy on theEfficacy of Nosocomial Infection ControlPrograms (SENIC) had demonstrated that a well·fun infection control programmes coupled withsurveillance could decrease NT by 32%5.Furthermore, the prevention of infection in theleU requires clinicians to have knowledge of

207

ORIGINAL ARTICLE

infection rates, source and nature of infection, aswell as the anti-microbial resistance patternsfound in their own rcu.

Currently hospital-acquired infections indeveloped countries occur in 2 to 6% ofhospitalised patients6

,7, whereas in developingcountries, the rates are higher, SOllie reaching ashigh as 25%H,9, These figures depend on themethods used for detecting NT and the variablesused in determining such rates'I,

In order to be able to gauge and assess whetherpolicies and programs implemented arc able toreduce the incidence of NI, we must first have the"local" incidence or prevalence rate of lCU-NI.

The ICU HUKM is a 24-bed general ICU locatedin a 1600-bed university hospital. Both surgicaland medical cases are managed in the lCU. TheICU is run by the Department of Anaesthesiologyand Intensive Care on a 'semi-open' conceptwhere patient is co-managed by both the referringunit and lCU doctors.

The "lCU-acquired nosocomial infectionsurveillance programme" in HUKM was firststarted in JanualY 1998 with the help of theHUKM Infection Control Committee, whichcarries out its lCU rounds on a daily basis, wheremicrobiological data including sensitivity patternsare presented and discussed during the rounds, It

is currently into its third year of surveillance,

This report the first of its kind in Malaysia,describes the epidemiology of Nl in the first twoyears of implementation of the surveillanceprogramme in an adult general medical andsurgical ICU in HUKM.

Materials and Methods

Starting from the month of January 1998,information from each adult patient admitted intothe ICU was collected for the purpose of ICU­acquired nosocomial infection surveillance,

208

The information collected included patientdemographics (age, sex, race), diagnosis, reasonfor lCU admission, and their disciplines (medical,surgical), All data concerning the number andduration of devices used (central venouscatheters, pulmonary catheters, indwelling urinarycatheters, mechanical ventilation etc.) as well asnumber of device days were documented, Theuse of anti-microbial agents, status of feeding(enteral, parenteral), and agents for stress ulcerprophylaxis were included in the surveillance, Wealso recorded the development and duration ofshock (of all aetiologies) in these patients, as wellas the choice of vasopressors/inotropes used,

There are basically two forms of cultures done inour lCU; routine and specific cultures. Trachealaspirate and urine cultures are routinely done aspart of the "lCU-acquired nosocomial infectionsurveillance programme". The tracheal aspiratesare taken for all intubated patients with orwithout mechanical ventilation on the day ofaclmission, and on evelY Tuesday and Thursday,Urine cultures are taken for all patients oncontinuous bladder drainage (CBD) on day one ofCBD usage and thereof once per week, Noroutine urine cultures were done for non­catheterised patients. As for specific cultures, thelCU clinicians in charge of the patient dependingon their own clinical judgement and discretionorder such cultures.

The lCU staff nurses according to set ICUprotocol and procedures take all routine cultures.The tracheal aspirates are taken under asepticconditions with two staff nurses performing theprocedure. One staff nurse helps in securing theendotracheal or tracheostomy tube (ETT) whilethe other performs the act of tracheal aspirate byinserting a suction catheter connected to a sterilecollection container into the trachea withoutsuctioning being applied. The catheter isadvanced until resistance is felt. At this pointsuctioning is applied while the catheter is slowlywithdrawn from the airways and ETT. Once thecatheter is removed from the Err, the residual

Med J Malaysia Val 56 No 2 June 2001

secretion in the catheter tubing is furtherexpelled from the tubing by suctioning sterilewater. The specimen container is then labelledand sent for cultures.

Urine samples are taken via the specimen POltconnected to the CBD tubing. Povidine solution isapplied to the POlt, and a sterile 10ml syringewith a 23G needle connected is used to collectthe urine. The urine is then placed into a sterilecontainer and sent for cultures and sensitivity.

The lCU doctors perform blood cultures underaseptic condition. The site of venupunctui"e iscleaned with povidine and spirit and draped withsterile cloth. A sterile 10ml syringe with a 23Gneedle is used to aspirate lOml of blood. Thisblood (after changing to a new needle) is theninjected into two culture bottles (Bactec'M)aerobic and anaerobic culture bottles) and sent tothe laboratory for culture and sensitivity. If a newcentral venous line (CVC) or arterial line isinserted, and blood cultures are required, theblood specimen is then taken from these cathetersdirectly and injected into the culture bottles asthese lines are inserted under aseptic condition.

Patients suspected of having CVC sepsis will havethree samples taken, one, blood from a vein awayfrom the CVC site, another blood sample directlyfrom the CVC lumen itself, and the third specimenbeing the CVC tip itself, all done under aseptictechnique. The CVC tips were processedaccording to the method by Maki et aiW

.

A sterile swab stick is used to collect specimensof suspected skin and / or wound sepsis andplaced into a sterile container.

Patients diagnosed clinically with pneumonia hada fiberoptic bronchoscopic examination andbroncho-alveolar lavage (BAL) performed onthem. The site (lung lobes or segments) forbronchoscopy is initially determined byidentifying suspected lung infected areas on chestrad{ography. Once the bronchoscopy tip is inplace, sterile water in aliquots of SOml (to a total

Med J Malaysia Val 56 No 2 June 200 I

THE INCIDENCE OF NOSOCOMIAL INFECTION

of 200mls) is injected into the side port of thebronchoscope and subsequently aspirated into asterile collecting container. A total of fourcontainers will then be sent for cultures.

Samples sent to the microbiology laboratory willundergo microscopic examination, incubation,Gram staining and subculture for organismidentification and sensitivity testing. Trachealaspirates samples would further bemicroscopically examined for pus cells under 100X magnification and graded into either scanty puscells « 4 pus cells per microscope field);moderate pus cells (5 - 10 pus cells per field) andnumerous pus cells (> 10 pus cells per field). Thedetails on how such microbiology testing iscarried out is available in most microbiologyreference books and will not be further discussedin this paper. All culture growth is recorded aslight « 15 colonies), moderate (sub confluentgrowth), and heavy (confluent growth).

Definitions

Infection, bacteraemia, sepsis, severe sepsis andseptic shock

The definitions of the above were in accordancewith the Consensus Conference of the AmericanCollege of Chest Physicians and Society of CriticalCare Medicine l1

•

Infection is defined as a microbial phen01nenoncharacterized by an inflammatory response to thepresence of microorganisms or the invasion ofnormally sterile host tissue by those organisms.Bacteraemia is defined as the presence of viablebacteria in the blood.

Sepsis is the systemic response to infection. It ismanifested by two or more of the followingconditions asa result of infection: temperature>38"C or <36"C; heart rate > 90 beats/min;respiratory rate > 20 breaths/min or PaC02 <4.3kPa «32 Torr); white blood cell count> 12 000cells/mm', or > 10% immature (band) forms.Severe sepsis is sepsis associated with organdysfunction, hypoperfusion or hypotension.

209

ORIGINAL ARTIClE

Hypoperfusion and perfusion abnormalities mayinclude but are not limited to, lactic acidosis,oliguria or an acute alteration in mental status.Septic shock is sepsis with hypotension, despiteadequate fluid resuscitation, along with thepresence of perfusion abnormalities may include,but arc not li1nited to, lactic acidosis, oliguria oran acute alteration in mental status. Patients whoare on inotropic or vasopressors agents may notbe hypoterisive at the time when perfusionabnormalities are measured.

Culture positive samples

A microbiologist classifies all samples withpositive growth of organism into 'significant','colonizer', 'normal flora' or 'contamination'.These growths are recorded as light «15 colony­forming units), moderate (sub confluent growth)or heavy (confluent growth).

All positive blood cultures are consideredclinically significant when a known pathogen iscultured coupled with clinical evidence of sepsisis present. The isolation of an organism from oneor more bottles was accepted as being clinicallysignificant for all organisms except coagulase­negative staphylococci, where isolation of two ormore bottles were required before considering itas significant given the fact of its high propensityto contaminate blood cultures.

The clinical diagnosis of infection or sepsis is leftto the lCU clinician based on his or her ownassessment and judgelnent while taking intoaccount the definitions for infection12

• Usually ahigh index of suspicion is required before anynew fever or sepsis is deelned to be due to a newinfection. For example nosocomial pneumonia issuspected when a patient previouslyaSylnptomatic develops crackles on auscultationor dullness to percussion on chest examinationcoupled with new onset of purulent sputum andchest radiographic examination showing new orprogressive infiltration, consolidation, cavitationsor pleural effusion. The diagnosis is furthersupported with positive organism cultured from

210

blood cultures; isolation of pathogen fromspecimen obtained by transtracheal aspirate,bronchial brushing or broncho-alveolar lavage.

PnelUllonia was considered ventilator associated(VAP) when its onset occurred after 48 hours ofmechanical ventilation (MV) and was judged notto have been incubated before starting MV,coupled with high clinical suspicions. Whenpossible, a fiberoptic bronchoscopicexamination and broncho-alveolar lavage (BAL)was perfonned on each of these patientssuspected to have YAP within the first 24 hoursafter the development of a new pulmonaryinfiltrate. Lower respiratory tract infections areall other forms of respiratory infections tlIat arenot due to pneumonia e.g. bronchitis,tracheobronchitis, tracheitis.

Patients are suspected to have primarybacteraemia/sepsis when they develop a newonset of fever with or without increase in totalwhite, as well as an apparent change in theiroverall condition such as hypotension,tachycardia and increasing blood lactate levelsbut with no other recognized cause (patients withcentral venous catheter infection are excludedfrom this group).

Any patient with cloudy foul smelling urine withor without suprapubic tenderness is suspected tohave a urinaly tract infection, while centralvenous line sepsis is suspected in any patientswith evidence of sepsis together with localinflammation (cellulitis) and pus production fromthe CVC insertion site, A positive culture for CVCsepsis is said to have occurred when all threecultures taken grows the same organism,

ICU-acquired nosocomial infection (NI)

The diagnosis of an Nl was done according to thestandard definitions of the Centres for DiseaseControl and Prevention (CDC)12 where there mustbe no evidence that the infection was present orincubating at the time of rcu admission. Wefurther defined an lCU-acquired infection as an

Med J Malaysia Val 56 No 2 June 2001

infection developing 24-hours after leu admissionhaving originated in the ICU, and being activelytreated. All patients diagnosed to have eithercommunity-acquired or hospital-acquired (non­leU) infections were excluded.

Clinical Nosocomial Infection Rate (CNIR)

Defined as, the number of patients diagnosedclinically with NI per 100 patients admitted intothe ICU. In specific NI e.g. Nosocomial lowerrespiratory tract infection (NLRTI), the CNIR isdefined as, the number of patients diagnosedclinically with NLRTI per 100 patients admittedinto the ICU.

Culture Positive Nosocomial Infection Rate(CPNIR)

Defined as, the number of patients diagnosedwith Nl having positive culture samples per 100patients admitted into the ICU. In specific NI e.g.Nosocomial lower respiratOly tract infection(Nl,Rl'J), the CNIR is defined as, the number ofpatients diagnosed clinically with NLRTI per 100patients admitted into the ICU.

Device~use rateS'

Defined as, the number of device-days divided bythe number of patients-days.

Example: DU ~ Number of device-daysNumber of patient-days

DeviceMossociated infection rote6

Defined as, the number of device-associatedinfections for a specific site per 1000 device­days.

Example:VAP rate per 1000 ventilated days

= Number of ventilator-associated pneumonia X 1000Number of ventilated days

Med J Malaysia Vol 56 No 2 June 2001

THE INCIDENCE OF NOSOCOMIAL INFECTION

*For details on how to calculate the 'device-days'and 'patient-days', readers are advised to refer to theNational Nosocomial Infections Surveillance Report'.

Statistical analysis

Data are either expressed as frequency,proportion, percentage or means with standarddeviation (SD). Significance test for comparingtwo or more propOltions from independentgroups are analysed using the Chi-square test(Yates correction for continuity was used for 2 by2 tables), where else comparison of means (t-test)is used for data presented as means ± SD(MedCalc® Version 5.00.013 - Windows 95/98/NTCopyright" 1993-1999 FrankSchoojans). A 'Pvalue' of less than 0.05 (P < 0.05) was consideredstatistically significant.

Results

A total of 988 adult patients were admitted intothe general intensive care unit (JCU) during theyears 1998 and 1999, of these 23% developed NI(Table I). Majority of those admitted were fromsurgical -based disciplines (74.9%) including bothpost-operative elective and emergency surgicalcases. Nearly a third (31.6%) were admitted forpost-operative care and monitoring withoutmechanical ventilatory support. 'A significantpercentage, 11.4% were admitted with sepsis andseptic shock. Over 35% of the medical casesadmitted into ICU developed NI as compared to16% of the surgical cases. None of the Obstetric &

Gynaecology and ENT patients developed NI.

Nearly 60% of patients admitted were males, withmajority Malays (55.1%) followed by Chinese(34%) and Indians (8.5%). The sex anei· racialdistribution for patients developing NT weresimilar to the ICU demographics.

The average age of the ICU patients was 47 ± 18years old with majority in the 41 to 60 years agegroup 05.2%). The patients with NI were slightlyolder, 50 ± 16 years old but with majority in the41 - 60 years age group (40.3%).

211

ORIGINAL ARTIClE

Table IDemographics of Patients Admitted into the Intensive Care Unit

Overall Patients

n ("!o)

Patients withNosocomial Infection

n ("!o)

144 (63.2)84 (36,8)

136 (59,6)68 (29,8)20 (8,8)4 (1.8)

50 ± 1616 (7,0)28 (12,3)20 (8,8)92 (40,3)72 (31,6)

96 (42,1)48 (21.1)4 (1.7)° (0,0)

56124,6)° (0,0)24(10,51

12 (5,2)40 (17.5)24 (10,5)32 (14,0)8 (3,6)

32 (14,0)32 (14,0)8 (3,61

32 (14,0)8 (3,6)

228 (23.1)

588 159.5)400140,5)

344 (34,8)260 (26,3)64 16,5)52 (5.3)

156 (15,8144 (4,5)68 (6,8)

544 (55,11336 (34,0)

84 (8,5)24 (2,4)47 ± 18

100 (10,1)132 (13.4)136 (13,8)348 (35,2)272 127,5)

988Total number of patients

SEXMaleFemale

RACEMalayChineseIndianOthers

AGE Imean ± SD)<21 years22·30 years31 ·40 years41 ·60 years> 61 years

DISCIPLINESurgeryNeurosurgeryOrthopaedicsENTMedicalObstetric And GynaecologyOthers* *

REASON FOR ICU ADMISSIONCardio respiratory arrest 28 (2,8)Pure respiratory arrest 120 (12,1)Pneumonia 40 (4,0)Cerebral protection 168 (17)Airway protection 8 (0,8)Post operative ventilation 160 (16,2)Post operative care 312 (31.6)Sepsis 48 (4,9)Septicaemic shock 64 16.5)Close monitoring 40 14,0)The percentage is calculated per each individual group (i,e, 'overall' and 'nosocomial infection')** Others are including plastic surgery, uralogy, nephrology, neuramedical and endocrinology patients

212 Med J Malaysia Val 56 No 2 June 2001

Table IITypes of ICU-acquired Nosocomial Infection

Types on Nosocomiallnfeclion Frequency ('Yo)Overall 228 (23.1)Lower respiratory tract infection 151 115.3)Primary bacteraemia 80 (8.1)Pneumonia 53 (5.4)Urinary tract infection 20 (2.0)Skin infection 16 11.6)Central venous catheter infection 12 (1.2)Surgical skin infection 8 (0.8)* Others 12(1.21Frequency = number of palienls diagnosed with nosocomial

infectionPercentage = the number of palients diagnosed with

nosocomial infeclion per 10iailCU admission(988 patients)

* Others includes intra-abdominal and gaslro-inlestinalinfections

The main types of NI found are presented inTable II.

A total of 2252 samples from various sites weretaken from 560 patients (56.7%). Patients mayhave had several cultures done on them duringtheir admission in lCU. Seven hundred andfifty-six samples (out of the 2252 samples)were taken from 228 patients clinicallydiagnosed to have NI, of which, 572 samplesfrom 120 patients were culture positive. Thisgave a Positive Culture Nosocomial InfectionRate of 12.1% with nearly 50% of patientsclinically diagnosed N1 having negativecultures. The Gram-negative organismsconstituted 75% of nosocomial microorganismscultured. Nearly 19% were Acinetobacter :-,p.,followed by Klebsiella .\p. (17.2%) andPseudomonas sp. 04.3%). Of the Gram-positiveorganisms, MRSA (methicillin-resistantSiapbylococcus aureus) C7 .4%) remains thehighest number cultured in patients with NI.This was followed by MRSE (multi-resistant

Med J Malaysia Val 56 No 2 June 200 I

THE INCIDENCE OF NOSOCOMIAL INFECTION

Stapb. epidermidis) (3.2%). Candida sp. mainlyarising from the urine consisted of 1.6% of totalsamples cultured positive (Table III).

The Clinicai Nosocomial LRTI Rate = 15.3% andthe Culture Positive Nosocomial LRTI Rate =

10.5%, therefore nearly 31% of the cultures donein these patients were negative. The three mainorganisms causing nosocomial LRTI wereAcinetobacter sp. (28.9%); Pseudomonas sp.(28.9%), of which 77% were Pseudomonasaeruginosa) and MRSA (9.2%). (Table III) Out ofthe 151 patients clinically diagnosed withnosocomial LRTI, only 16 patients 00.6%) werenot mechanically ventilated,

Eighty patients were diagnosed to have primarybacteraemia (Clinical Nosocomial BacteraemiaRate = 8.1%), of these the number of culturepositive blood samples was 144 in 60 patientsgiving a Positive Culture Nosocomial BacteraemiaRate of 6,1%. Majority of organisms cultured wereGram-negatives such as Klebsiella sp. (25.7%),followed by Aclnetobacter sp. (22.9%), andStrenot-rophomonas sp 01.4%). The Gram-positiveorganisms were mainly MRSE 01.4%). MRSA(5.7%) and Strep. vil1dans (5.7%) of the totalorganisms cultured in the blood.

Twelve patients were diagnosed to have centralvenous catheter sepsis (Clinical NosocomialCatheter Sepsis Rate = 1.2%). The Culture PositiveNosocomial CVC Infection Rate was 1.2%. We hadequal numbers of Gram-positive and Gram­negative organisms cultured in patients with CVCNI. These were Acinetobacter sp. (50%) and MRSE(50%). The Device-use rate for CVe's were 0.71.The mean duration of time the evc were in usefor these patients was 28.8±5.81 days ascompared to 17.6±15.71 days in patients who didnot develop CVC bacteraemia (P = 0.0139). TheCVC-associated blood stream infections (BSI) were2.5 patients per 1000 catheter days. (Table IV)

The Clinical Nosocomial UTI Rate was 2.0% and theCulture Positive Nosocomial UTI Rate was 1.6%.Klebsiella sp. (75%) was the only bacterial organism

213

ORIGINAL ARTiClE

that contributed to the infection. The remammg25% was attributed to Candida sp. All the patientswith NI had an indwelling urinaIy catheter (CBD)in place at the time of developing the NI. None ofthe patients without a catheter developed any formof UTI. The Device-use rate for CBDs was 0.79, andthe catheter-associated UTIs were 2.9 patients per1000 catheter days. (Table IV)

Most of the skin swabs (75%) taken were fromburns patients. Out of the 24 positive culturesreceived from these burns patients, 50% hadMRSA. The main Gram-negative organisms

cultured were Klehsiella sp. (16.7%), Enterohactersp. (16.7%) and Enterococcus sp. (16.7%). Therewere eight positive cultured skin swabs takenfro111 surgical skin sites in patients diagnosed withintra-abdominal sepsis. Of these, Acinetobactersp. and Klehsiella sp. (50% respectively) were thetwo main organisms cultured.

One hundred and eighty-eight patients with NIwere ventilated (26. 1% of all ventilated patients).The Device-use rate was 0.65. The averageduration of ventilation for NI patients was 13±12.5days as compared to non-NI patients, 3±2.1 days

Table IIIOrganisms Responsible for Nosocomial Infections in leu

ORGANISM 'Overall "LRTI **Pneumonia "Primary "cve Sepsis "UTI 'Skin "551 "OtherBacteraemia 'Infection

%(n=S72) %, (n=304) %(n=28) %(n=140) %(n=16) %(n=16) %(n=32) %(n=8) %(n=8)

Acinelobocler sp. 19.0 28.9 42.8 22.9 50.0 0 12.5 0 0Pseudomonas sp. 4.2 6.6 0 0 0 0 0 0 0Pseudo aeruginosa 14.7 22.3 0 2.9 0 0 0 0 20.0Klebsiella sp. 17.2 21.1 28.6 25.7 0 75.0 12.5 50.0 0Sirenofrophomonas sp 2.1 0 0 11.4 0 0 0 0 0Enlerobacler sp. 3.2 3.9 0 5.7 0 0 12.5 0 0Flavimonas sp. 0.5 0 0 2.9 0 0 0 0 0MRSA 7.4 9.2 14.3 5.7 0 0 37.5 0 20.0Sfaph. au reus 1.1 2.6 0 0 0 0 0 0 0Siaph. epidermidis 1.1 2.6 0 0 0 0 0 0 0MRSE 3.2 0 0 11.4 50.0 0 0 0 0Sirep. viridans 1.6 1.3 0 5.7 0 0 0 0 0Enferococcus sp. 1.1 0 0 2.9 0 0 12.5 0 0Claslridium deficile 1.1 0 0 0 0 0 0 0 40.0Candida sp. 1.6 1.3 14.3 0 0 25.0 0 0 0Olhers 1.1 0.2 0 2.9 0 0 0 0 0

• Results are presented as percentages:x' organism per sile X 100%Total'x' organism per site

" Percenlage:Number of organism X 100%Tolal number of organism

LRTI =Lower respiralory tract infeclion.CVC =Cenlral venous catheter.UTI =Urinary lracl infection.551 =Surgical skin infection.

214 Med J Malaysia Val 56 No 2 June 2001

THE INCIDENCE OF NOSOCOMIAL INFECTION

Table IVDevice-use Ratio (DU) and Device-associated Nosocomial Infections Rates for Mechanical

Ventilation, Urinary Catheters, and Central Venous Catheters in the ICU, HUKM Compared withICU, Jordan University Hospital (JUH) and National Nosocomial Infections Surveillance (NNIS)

Parameter HUKM Rate JUH Rate NNIS Rates Percentile(mean/median) Compared with NNIS

Mechanical ventilationDU 0.65 0.46 0.38/0.37 > 90YAP 11.9 19.1 11.3/10.1 < 75

Urinary catheter0.75/0.76DU 0.79 0.75 < 75

UTI 2.9 15.6 5.2/5.1 < 25Central venous catheter

DU 0.71 0.47 0.47/0.47 > 90CYC 2.5 3.0 4.5/4.6 dO

.," = ventilator associated pneumonia rate= urinary catheter associated-inFection rate= central venous catheter·associated inFection rate

(Table IV). Ventilated NI patients hada higher mortality rate when compared toventilated non-NT patients (46% versus 34.6%respectively; p=0.0066). A total of 53 patientswere diagnosed clinicaIly to have ventilator­associated pneumonia (VAP) but only 28 patientswere confirmed YAP based on the BAL results.The other 25 patients diagnosed YAP could not

have their diagnosis supported by a BAL becauseof the unavailability of the bronchoscope duringtheir admission. The incidence rate for VAP inpatients who had BAL confirmation was 6.3patients per 1000 ventilated days, while theoverall incidence rate of VAP taking account thosewho did and did not have a BAL was 11.9 patientsper 1000 ventilated days. (Table IV) (Table V)

Table VDuration of Mechanical Ventilation and the Mortality Rate

Mechanical Ventilated Patients Mechanical Ventilated Patientswith Nosocomial Infection with No Nosocomial Infection

(n = 200) (n = 520)Total days 2528 1416Mean (daysl ± SD 13 ± 12.5 3 ± 2.1Mortality 1%1* 92 (46.0) 180 (34.6)'Percentage: Number of martalily Fram each group X 100%

Frequency of each group

p value is p < 0.0001 comparing mean days of ventilation For both groups of patients.p value is p = 0.0066 comparing mortality For both groups of patients.

Med J Malaysia Vol 56 No 2 June 2001 215

ORIGINAL ARTICLE

The main Gram-negative organisms causingVAP were Acinetobacter sp. (42.8%), andKlebsiella sp. (28.6%). MRSA (14.3%) were theonly Gram-positive organism causing YAP.Candida sp. made up of the remaining 14.3%organism detected.

A total of 284 patieots (28.5%) had shock in1CU, of these, 112 patients 09.4%) werepatients diagnosed to have N1 (Group A).(Table VI) The average duration of shock inpatients with NI was 7±7.1 days, as comparedto non-NI patients (Group A,), 3±1.7 days(PA<O.OOOl). The mortality rate in these patientswith shock however reflected differentlybetween the two groups. Patients with no NTtended to have a higher mortality rate (81.4%versus 64.3%; P" ~ 0.0027).

Eighty-four (75%) NI patients had septic shock(Group B). These patients had an average ICUstay of 8±7.5 days as compared to Non-NI patientswith septic shock (Group B,) having 3±1.9 days

(P"<O.OOOl). The mortality rate for the Group Bpatients also differed, 61.9% for NI patients inseptic shock versus 92% in non-NT patients inseptic shock (p"<0.0005).

The average duration of non-septic shock patientswith NI (Group C) was 4±4.9, as compared tonon-NI patients (Group c,), 2±1.3. This washighly significant (PC~0.00l8). However there wasno statistical differences seen when the mortalityrate was compared between the Group C patients(71.4% versus 66.7%; P"=0.8330).

The length of ICU stay differed between patientswith NI and those without (13±12.4 days versus3±2.8 days; P<O.OOOl). It was noted that patientswith NI also had a higher mortality rate C.4r ":._versus 23.7%; P<0.0005} These patients tli4 (\1

had longer ICU stay (21.06±11.24 days \ \J~12.27±11.34 days; P < 0.000l) when compar .C\those without NI (Table VII} The mortality r3\NT patients when compared to the genlpopulation was 9.30/0 (the overall ICU mortal

(Group A)

Table VIDuration of Shock and the Mortality Rate in ICU Patients with

Nosocomial Infection (NI) and No-nosocomial InfectionPatients with Patients with Patients with Patients with Patients with

NI in Shock NI in Septic NI in Non- No NI in No NI inShock Septic Shock Shock Septic Shock

(Group B) (Group C) (Group A,) (Group B,)

Patients withNo NI in Non­

Septic Shock(Group C,)

72176

21.3

48 (66.7)92 (92.0)140 (81.4)

84 28 172 100672 112 488 312

8 4 3 3

7.5 4.9 1.7 1.9pB < 0.0001 pC = 0.0018

52 (61.9) 20 (71.4)p value" pA2 < 0.0027 pB' < 0.0005 pC' =0.8330

112784

7

7.1p value* rr' < 0.0001Mortality (%1 72 (64.3)

FrequencyTotal daysMean (days)SD

Percentage: Number of patients dead from each group X 100%Frequency of each group

, p values comparing patients with nosocomial infection and patients with no·nosocomial infection Groups A, Band C's durotionof shock.

" p values comparing patients with nosocomial infection and patients with no-nosocomial Groups A, Band C's mortality.

216 Med J Malaysia Vol 56 No 2 June 2001

THE INCIDENCE OF NOSOCOMIAL INFECTION

Table VIIThe Duration of leu Admission and Mortality Rate

Total daysMean (daysl ± SDMortality 1%1*Total mortality daysMean (mortality daysl ± SD

Patients withNosocomial Infections

(n =228)3044

13 ± 12.492(40.41

648821.06 ± 11.24

Patients withNo-nosocomial Infections

(n =760)2548

3 ± 2.8180 (23.7)

834012.27 ± 11.34

* Percentage: Mortalily from each group X 100%Frequency of each group

p value: p < 0.0001 comparing the means of ICU stay for both groups of patients.p value: p < 0.0005 comparing mortalily for both groups of patients.p value: p < 0.0001 comparing the means of ICU stay for patients who died.

rate is 27.5%), Mortality directly attributed tosepsis was present in 56 patients with Nl (60.9%).It is regretted that we were unable to determinethe post-lCU hospital mortality.

There was no particular pattern noted when wecompared the number of infection sites tomortality, ventilation days, shock days and LOS.The patients with three sites of infection tcnded tohave a higher mortality rate (66.7%), those withtwo sites having longer ventilated days (mean 26days) and those with four sites having longerduration of shock (mean 11 days), whilst the LOSwas longest with patients having two infectionsites (mean 24 days). Surprisingly there were nodeaths in patients with five sites of infection.

Discussion

Nosocomial infection can involve any organ orsystem, particularly in those where instnlmentationand device use is the highest e.g. urinary catheters,intubations and mechanical ventilation, centralvenous catheterisation etc. The relative incidenceshave generally relnained constant over the yearsapalt from differences in microorganism and anti­microbial sensitivity pattern13•

Med J Malaysia Val 56 No 2 June 200 I

In this paper, we did not specificaliy look atpredisposing and risk factors leading to NI,neither did we attempt to discuss the antibiogramand organism sensitivity as this was thought to bebeyond the scope of this paper. Despite databeing collected on device usage, directcorrelation testing was not performed as themajor reason for this paper was looking at theincidence of NI in the lCU.

Currently there are two main significant andimportant studies on the epidemiology of ICU­acquired nosocomial infections available forcomparison. These are the National NosocomialInfections Surveillance System (NNIS)'·" and theEuropean Prevalence of Infection in IntensiveCare (EPIC) Study".

The NNIS is a database established in 1970involving lnedical institutions in the United Statesof America conducted by the Hospital InfectionsProgram to collect high quality nosocomialinfection surveillance data that can be aggregatedinto a national database. The data available ispublished regularly allowing comparisonsamongst institutions following NNIS methodologyto be lnade.

217

ORIGINAL ARTICLE

The EPIC on the other hand is a I-day prevalencestudy encompassing 1417 ICUs, providing a totalof 10,038 completed case reports forms. It is thelargest study of its kind in Europe.

The study conducted by the Jordan UniversityHospital QUH)" was looking at the incidence rateof NI in their hospital comparing it with the NNISgroup. This study allowed comparison to bemade with a developing countIy, and as suchwould provide important results for comparisonto be made against our study.

The overall NT rate in our institution was 23%,which was comparable to other similar ICUs QUHgroup=16.20/0, and 25% in a medical leu in SaudiArabia)". In the EPIC study, the overallprevalence rate was 20.6% (ranging 9.7 to 31.6%depending on the ICU studied). Such differenceseen in the prevalence rate in the EPIC studystems out from how each leu views and conductsits infection policy and protocols. Even though itwas not specificallY looked at in the EPIC studybut the differences between the prevalence ratesnoted amongst the various countries werethought to be due to differences in reu practice,

Despite the high numbers of culture samplestaken in our leU patients, nearly 50% of thecultures from patients with NI were negativecompared to only 15% in the EPIC study. Our lowrates could be attributed to patients being onSOllie form of anti-microbial prior to developingNl. The high positive culture rates in the EPICstudy was thought to have reflected possiblecontamination of the sample or the process ofcolonization, which is a universal phenomenon incritically ill patients l5

•

There were also differences in organisms culturedbetween the various studies. In the EPle study themost frequently reported isolates wereEnterobacteriaceae (34.4%) (Predominantly Escb.coli, Klebsiella sp and Enterobacter sp.) which wassimilar to other quoted studies18

. In our study,Acinetobaete,- sp. 09%) was the predominantorganism cultured. It was reported less than 6% in

218

the NNIS, whilst for the Gram-positive group,Stapb. aureus 8.5% (versus 30.1% in EPIC versus10.9% in NNIS) remains the most frequentorganislll cultllred,

The three main sites of NT in our patients werefrom lower respiratory tract 05.3%), primalybacteraemia (8.1%) and YAP (5.4%). Lowerrespiratory tract infections alone accounted for66% of Nl. The EPIC study noted pneumonia(47%), lower respiratOly tract infections 08%) andUTI (18%) as their three main causes of NI in theICU. The diagnosis used In this study as well asthe JUH was based on the NNIS definitions, whichitself used the eDe definitions l2

•

Pneumonia and lower respiratory tract infectionsare the most common cause of NI in the criticallyill patients but third commonest hospital NI afterUTI and surgical wound infections lH

• It is difficultto determine precisely the incidence ofnosocOlliial pneUlllonia in the leU, as the clinicaldiagnostic criteria used have low specificity. Theincidence ranges from 10% to 65%19.

The most common organisms causing nosocomialpneumonia in the NNIS study are Staph. aureus(20%) and Pseudomonas aeruginosa (21%). Inour own leU this was not so. Acinetobacter spwere the predominant organism cultured (28.7%).This was velY lliuch higher when compared tothe NNIS series (4.0%). At the time of writing thispaper, preliminalY results for the year 2000 stillshowed Acinetobacter sp as. the major organismcausing NI pneumonia. We were currently in theprocess of identifying why the incidence of thisorganism was high in our leU (culturing ofcircuits, humidifier and dryers), Further more,such high incidences seen in our leU may be theresult of over diagnosing. The true frequency ofNI cause by Acinetobacter sp. is ditllcult to assessbecause isolation of this organism in clinicalspecimens may reflect colonization rather thaninfection20

• The distinction between the two in thecritically ill patient may at times be diffitult whichmay be the case for OUf high percentage as wellas aUf unavailability to do EAL in all patients

Med J Molaysia Val 56 No 2 June 200]

diagnosed with pneumonia. What is known is thatit is strongly associated with mechanicallyventilated patients occurring in 4.0% to 26% ofventilated patients due to the moist environmentof the ventilator circuits and humidifier<o. As forother organisms, a recent prospective cohortstudy identified Staph. aureus (27%) as the mostfrequent organism causing nosocomialpneumonia21 with 20% of patients dying within aweek of the first positive culture. It was noted thatthose suffering from Pseudomonas sp pneumoniahad a higher mortality rate (45% versus ]4% fromother organismP. In our series 61.1% of patientswith Pseudomonas sp pneumonia died comparedto 32.6% dying from other organisms.

The incidence of clinically diagnosed VAP in ourstudy was 11.9 patients per 1000 ventilator days(versus J1.JH ~]9.1). When compared to NNIS, ourrates were less then the 751h percentile, thusconsidered a 'High' outlier where 75% of thehospitals in the NNIS database had lower rates(ratios) and 25% had higher rates (ratios) of YAP.This finding may point to a defect or problem inour care and technique of managing patients onmechanical ventilators. It could also reflect thehigh utilization of mechanical ventilation in ourleU (ventilator-device-use ratio of 0.65 - > 90th

percentile of the NNlS database).

Our culture positive primary bacteraemia rate of8.1% was very much lower then the NNIS rates of]7.0% but higher then the St. Thomas' Hospitalgroup of 3.7%". The differences could probablybe under diagnosing of bacteraemia in our study.Patients may have been misdiagnosed to havesystemic inflammatOly response syndrome (SIRS)due to other causes rather then sepsis. Thedifferentiation between SIRS and sepsis may bedifficult in the critically ill patient with culturenegative results. Further more, differences seenbetween various Ieus could be related todifferences in laboratory culture techniques. Wetake more than one blood culture set fromdifferent sites to help exclude contamination butsuch techniques are not standard practice in most

Med j Malaysia Vol 56 Na 2 June 200 I

THE INCIDENCE OF NOSOCOMIAL INFECTION

European hospitals". In the EPIC study,laboratory culture techniques were not described.The pathogenesis of nosocomial blood-borneinfection is commonly associated with prolongeduse of intravascular devices such as central linesand·. dialysis catheters. The study by MaId et al"found that 21% of pulmonaty artery cathetersbecome colonized and that 1.1% of catheterinsertions arc associated with blood streaminfection. Other studies have quoted colonizationrates up to 60%". The EPIC study further supportsthe high associated risk with intravasculardevices. In our study such correlation was notsought but the cve usage ratio in our ICUpatients was 0.71, which was way above the 90th

percentile of the NNIS. The incidence ofcolonization was also not available, as we do notsend our CVC tips for routine cultures. Such highusage of CVC should logically attrihute to higherinfection rates, as most studies wouldindicate3,!3,2V l but this was not so in our series. TheevC-associated infection was 2.5 patients per1000 catheter days «50th percentile of the NNIS;J1.JH ~ 3.0 patients per 1000 catheter days). Againthe low incidence seen in our series was probablydue to under diagnosing rather than 'true'favourable results.

Organisms commonly associated with catheterrelated sepsis are the gram-positive organisms i.eStaph aUl'eus and Stapb. epidennidio"'. In theNNIS study, Coagulase-negative staphylococci07%) were more con1l11only reported in centralline associated sepsis in medical ICU. In ourseries, the two major gram-negative organismscultured were Klebsiella sp (25.9%) andAcinetobacter sp. (22.9%) where as the NNISgroup cultured only Acinetobacter sp. (2.0%),and Klebsiella sp (4.0%). Edgeworth JD et al."had quoted the incidence of nosocomialbacteraemia in their series as Acinetobacter sp(2.0%) and Klebsiella sp (9.0%) withPseudomonas sp (19%) still the highest gram­negative organism cultured, where elsecoagulase-negative staphylococcus (12%) thehighest gram-positive organism cultured.

219

ORIGINAL ARTICLE

Surprisingly nosocomial UTIs did not occurmuch in our ICU despite of the high usage ofurinary catheters (94.3% of ICU patientscatheterised versus 75.2% and 95% in otherstudies)16,21, The CBD-device-use ratio was 0.79making it below the 75'" percentile of the NNIS(jUE ~ 0.75). The CBD-associated UTI rate was2.9 patients per 1000 catheter days «25'"percentile of NNIS; JUE~15.57). Our low UTIrates were probably reflecting the strict aseptictechnique of CBD insertion and close infectionmonitoring carried out in our lCU. Then againsuch differences could be due to underdiagnosing UTI in these ICU patients. Thediagnosis of UTI may be overlookcd especiallywhen common symptoms and signs of UTI suchas dysuria, frequency, and suprapubic pain maybe missed in the unconscious patient. We tendedto have a higher suspicion of UTI when 'dirty'urine is seen in oLlr patients.

As expected there were differences seen inmortality rates amongst the various studies. Such

.,;. differences are also associated with the type ofICU as well as type of infection the patient issuffering from. The differences may also reflectdifferences in intensive care practice and patientselection rather than any real differences inabsolute standards of carel~.

The EPIC study clearly showed significantcorrelation between the prevalence rate of rcu­acquired infections and the mortality rate (16.8%;ranging 8.4% to 28.5%, R'~0.68). The risk ofdying from a NI was highest when patientsdeveloped nosocomial pneumonia (13% to55%)19, laboratory-proven bloodstreambacteraemia and clinical sepsis (31 % to 35%)25. Inour seties, the tisk of dying from primalybacteraemia was higher then from nosocomialpneumonia (47.8% versus 41.2%). These ratesquoted may be misleading, as patients may havemore than one infection occurring at a time. Asin our own study, the more sites of infection apatient had did not translate into a highermortality rate; neither did it show any particular

220

pattern in length of mechanical ventilation,duration of shock or ICU length of stay. It wouldhave been better if in our study we were able toadjust these patients base on the degree of organdysfunction or failure they had and see how thisinfluenced the mortality rate, duration ofventilation, shock and ICU stay. It wasunfortunate that the JUH and NNIS study did nothave any figures on their leU mortality rate.

The shOltcoming of this study is the referencesused for comparison of data. The EPIC study is aI-day point prevalence study, and as such onlyprovides a snapshot in time, and in comparisonwith incidence studies, may actually overestimatea problem. The authors of the EPIC study furthernoted the possibility of selection bias regardingthe identification and voluntary participation ofthe rcus surveyed. However such bias may havebeen reduced by the fact that data acquisitionwas multidisciplinary.

In the NNIS study, the percentage of organismcultured was mainly hospital-wide rather thanICU specifically. As such the NNIS rates may belower than those expected for an ICU. Acomparison of organisms cultured from variousinstitutions is important but mainly of academicrather than clinical interest. It is more importantand clinically relevant to know your ownhospitals or ICU organism profile. Furtherlimitations in using the NNlS are the infrequencyand unavailability of culturing in some of theparticipating hospitals.

It is further noted that information with regardsto patients underlying and/or concomitantdiseases, as well as their clinical status and leuseverity scores were not included in this study.Even though it is important for such informationto be included especially when one is lookingfor a 'cause-effect' of nosocomial infection inICU, the authors felt this was not in keepingwith the main objective of this study looking atincidence of NI and as such was not included inthe study.

Med J Malaysia Val 56 No 2 June 200 I

Conclusion

leU-acquired nosocomial infections are and still amajor problem in reus. It results in increase reustay as well as extra costs attributable toinfection'l,<i,24, This study has clearly documentedthe high incidence rate of lCU-acquired infection,and the importance of lower respiratory tractinfection, pneumonia and bacteraemia inincreasing ICU mortality. The need for vigiiantsurveillance as well as good infection controlpolicy and management have shown in otherstudies to help reduce the incidence, cost andmortality of these groups of ICU patients'. In thisstudy such conclusions could not be drawn, asdata prior to the surveillance programme was notavailable for comparison. Nevertheless we did

THE INCIDENCE OF NOSOCOMIAL INFECTION

reduce the cost. of unnecessary IDutine urinecatheter tip and sample cultures (estimated savedcost of RM16, 000 per year), as these cultures arenot done anymore in our ICU as part of thesurveillance programme.

Acknowledgements

The authors wouid like to thank the Departmentof Microbiology, Faculty of Medicine, UniversitiKebangsaan Malaysia and the Infection ControlCommittee, HUKM particularly AssociateProfessor Dr Nordiah Awang and The PharmacyDepartment, BUKM for their kind assistance inproviding data and heip in the ICU-acquiredNosocolnial Infection Surveillance Program.

1. Wakefield D. Understanding the costs ofnosocomial infections. In: Wenzel RP, editor.Prevention and control of nosocomial infections.2nd ed. Baltimore: William & Wilkins; 1993: 21-41.

2. Weinstein RA. Nosocomial infection update. EmergInfect Dis 1998; 4, 416-20.

3. Weinstein RA: Epidemiology and control ofnosocomial infections in adult intensive care units.Am J Med 1991; 91(Suppi 3B), 1795-1845.

4. Haley RW. Measuring the costs of nosocomialinfections: methods for estimating economicburden on the hospitaL Am J Med 1991; 91 (Suppl3B), 532-538.

5. Haley RW, Culver DE, White JW, Morgan WM,Emori TG, Munn VP et al. The efficacy of infectionsUlveillance and control programs in preventingnosocomial infection in u.s. hospitals. Am JEpidemioi 1985; 121, 182-205.

6. Naticmal Nosocomial Infections Surveillance (NNIS)Report, Data SummalY from October 1986 - April1996, Issued May 1996: A report from the NationalNosocomial Infections Surveillance eNNIS) System.Am J Infect Controi 1996; 24(5} 380-88.

Med J Malaysia Val 56 No 2 June 200 I

7. Haley RW, Culver DH, White. JW, Morgan WM,Emori TG. The nationwide nosocomial infectionrates: a new need for vital statistics. Am JEpidcmioi 1985; 12L 159-67.

8. Nettleman MD. Global impact of infection control.In: Wenzel RP, editor. Prevention and control ofnosocomial infections. 2nd ed. Baltimore: William& Wiikins; 1993; 13-20.

9. Western KA, 5t John R, Shearer LA. Hospitalinfection control-an international perspective.Infect Control 1982; 3: 453-5.

10. MaId DG, \{Teise CE, Safarin HW. A

semiquantitative method for identifyingintravenous-catheter-related infection. N Engi JMed 1977; 296, 1305-309.

11. American Coiiege of Chest Physicians/Society ofCritical Care Medical consensus conference:Definitions for sepsis and organ failure andguidelines for the use of innovative therapies insepsis. Crit Care Med. 1992; 20: 864-74.

12. GarnerJS,Jarvis WR, Emori TG, Horan TC, HughesJM. CDC definitions for nosocomial infections, AmJ Infect ControL 1988; 16, 128-40.

221

ORIGINAL ARTIClE

]3. Edgeworth ]D, Treacher DF, Eykyn S]. A 25-yearstudy of nosocomial bacteremia in an adult intensivecare unit. Crit Care Med. ]999; 27(8} ]42]-428.

14. Robert AW. Nosocomial infection update. EmergingInfectious Diseases. 1998; 4(3). Located at:http://www.cdc.gov/ncidodleicllvo14no3/weinstein.htm.

15. VincentJL, Bihari DJ, Suter PM, Bruining HA, WhiteJ, Nicolas-Chanoin MH, et at. The prevalence ofnosocomial infection in intensive care units inEurope: Results of the European Prevalence ofInfection in Intensive Care (EPIC) Study. ]AMA.]995; 274(8); 639-44.

]6. Khuri-Bulos NA, Shennak M, Agabi S, Saleh S, AlRawashdeh 5, Al Ghanem 5, et at. Nosocomialinfections in the intensive care units at a universityhospital in a developing country: Comparison withNational Nosocomial Infections SurveillanceIntensive care units rates. Am J of Infect Control.]999; 27(6): 547-52.

17. Jarvis WR, White JW, Munn VP et al. Nosocomialinfection swveillance. MMWR CDC Surveill Summ.]983; 33, 9SS-21SS.

18. Dal Nogare AR. Nosocomial pneumonia in themedical and surgical patients. Risk factors andprimary management. Med Clin North Am. 1994;78, ]08]-1090.

222

19. Grap MJ, Munro CL. Ventilator-associatedpneumonia: Clinical significance and implicationsfor nursing. Heart Lung. ]997; 26(6): 4]9-29.

20. Forster DH, Daschner PD. Acinetobacter Species asNosocomial Pathogens. Eur J Clin Microbiol InfectDis ]998; 17,73-77-

21. Taylor GD, Buchanan-Chell -M, Kirkland T,McKenzie M, Wiens R. Bacteremic nosocomialpneumonia. A 7-year experience in one institution.Chest ]995; 108, 786-88.

22. MaId DG, Stolz 55, Wheeler 5, Mermel LA. Aprospective, randomised trial of gauze and twopolyurethane dressings for site care of pulmonmyartety catheters: implications of cathetermanagement. Crit Care Med 1994; 22: 1729-737.

23. Richards MJ, Edwards JR, Culver DH, Gaynes RP.Nosocomial infections in medical intensive careunits in the United States. Crit Care Med. 1999;27(5} 887-92.

24. Pry DE,Fly RV, Borzotta AP. Nosocomial blood­borne infection secondary to intravascular devices.Am] Surg. ]994; ]67, 268-72.

25. Pittet D, Tarara D, Wenzel RP. Nosocomial blood­stream infection in critically ill patients. Excesslength of stay, extra costs and attributablemortality. ]AMA ]994; 27], ]598-601.

Med J Malaysia Vol 56 No 2 June 2001