UBC Faculty Research and Publications

Variations in rates of nosocomial infection among Canadian neonatal intensive care units may be practice-related Aziz, Khalid; McMillan, Douglas D; Andrews, Wayne; Pendray, Margaret; Qiu, Zhenguo; Karuri, Stella; Lee, Shoo K Jul 8, 2005

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-12887_2004_Article_75.pdf [ 549.22kB ]
JSON: 52383-1.0221599.json
JSON-LD: 52383-1.0221599-ld.json
RDF/XML (Pretty): 52383-1.0221599-rdf.xml
RDF/JSON: 52383-1.0221599-rdf.json
Turtle: 52383-1.0221599-turtle.txt
N-Triples: 52383-1.0221599-rdf-ntriples.txt
Original Record: 52383-1.0221599-source.json
Full Text

Full Text

ralssBioMed CentBMC PediatricsOpen AcceResearch articleVariations in rates of nosocomial infection among Canadian neonatal intensive care units may be practice-relatedKhalid Aziz†1, Douglas D McMillan2, Wayne Andrews6, Margaret Pendray3, Zhenguo Qiu4, Stella Karuri4, Shoo K Lee*†3,4 and The Canadian Neonatal Network5Address: 1Discipline of Pediatrics, Memorial University of Newfoundland, St. John's NF, Canada, 2Department of Pediatrics, Dalhousie University, Halifax NS, Canada, 3Department of Pediatrics, University of British Columbia, Vancouver BC, Canada, 4Centre for Healthcare Innovation and Improvement, Vancouver BC, Canada, 5Canadian Neonatal Network Coordinating Center, Vancouver BC, Canada and 6Discipline of Pediatrics, Memorial University of Newfoundland, St Johns NF, CanadaEmail: Khalid Aziz - kaziz@mun.ca; Douglas D McMillan - Douglas.McMillan@iwk.nshealth.ca; Wayne Andrews - wandrews@morgan.ucs.mun.ca; Margaret Pendray - mpendray@cw.bc.ca; Zhenguo Qiu - winston_qiu@yahoo.ca; Stella Karuri - skaruri@cw.bc.ca; Shoo K Lee* - shool@interchange.ubc.ca; The Canadian Neonatal Network - hbavinton@cw.bc.ca* Corresponding author    †Equal contributorsAbstractBackground: Nosocomial infection (NI), particularly with positive blood or cerebrospinal fluidbacterial cultures, is a major cause of morbidity in neonatal intensive care units (NICUs). Rates ofNI appear to vary substantially between NICUs. The aim of this study was to determine risk factorsfor NI, as well as the risk-adjusted variations in NI rates among Canadian NICUs.Methods: From January 1996 to October 1997, data on demographics, intervention, illnessseverity and NI rates were submitted from 17 Canadian NICUs. Infants admitted at <4 days of agewere included. NI was defined as a positive blood or cerebrospinal fluid culture after > 48 hrs inhospital.Results: 765 (23.5%) of 3253 infants <1500 g and 328 (2.5%) of 13228 infants ≥1500 g developedat least one episode of NI. Over 95% of episodes were due to nosocomial bacteremia. Majormorbidity was more common amongst those with NI versus those without. Mortality was morestrongly associated with NI versus those without for infants ≥1500 g, but not for infants <1500 g.Multiple logistic regression analysis showed that for infants <1500 g, risk factors for NI includedgestation <29 weeks, outborn status, increased acuity on day 1, mechanical ventilation andparenteral nutrition. When NICUs were compared for babies <1500 g, the odds ratios for NIranged from 0.2 (95% confidence interval [CI] 0.1 to 0.4) to 8.6 (95% CI 4.1 to 18.2) whencompared to a reference site. This trend persisted after adjustment for risk factors, and was alsofound in larger babies.Conclusion: Rates of nosocomial infection in Canadian NICUs vary considerably, even afteradjustment for known risk factors. The implication is that this variation is due to differences inclinical practices and therefore may be amenable to interventions that alter practice.Published: 08 July 2005BMC Pediatrics 2005, 5:22 doi:10.1186/1471-2431-5-22Received: 20 December 2004Accepted: 08 July 2005This article is available from: http://www.biomedcentral.com/1471-2431/5/22© 2005 Aziz et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Page 1 of 12(page number not for citation purposes)BMC Pediatrics 2005, 5:22 http://www.biomedcentral.com/1471-2431/5/22BackgroundHospital acquired (nosocomial) infection in neonatalintensive care units (NICUs) is a significant cause of mor-tality and morbidity, particularly in very low birth weight(VLBW, <1500 g) babies. VLBW infants when affected bythese infections are at increased risk for death, have alonger hospital stay, and utilize more resources than non-affected controls [1]. The 15 participating centres in theNational Institute of Child Health and Human Develop-ment Neonatal Research Network found that 11% to 32%(mean 21%) of VLBW infants have at least one episode ofnosocomial sepsis [2]. Similar variations have beendescribed in other datasets [3,4]. These studies alsoshowed that valid comparisons of rates of nosocomialsepsis between centres require adjustment for risk factors,such as patient demographics and admission illnessseverity.Risk-adjustment has also been used to look at variationsin rates of both neonatal mortality and morbidity [5]. Forexample, it has been shown that, after correcting for anumber of confounders, Canadian NICUs vary signifi-cantly in their rates of death, catheter-related sepsis andintraventricular hemorrhage [6-8]. Each of these authorsconcluded that differences in outcome may be partlyattributable to practice variation. Given the complexityand intensity of care provided to VLBW infants, it is likelythat the etiology of acquired infections in NICUs is multi-factorial [9], with risk arising from the factors attributableto the host, the organism, the environment and clinicalinterventions. However, previous studies only examinedrisk factors for hospital-acquired infections in VLBWinfants. It is unclear whether risk factors are different inlarger infants with 1500 g or higher birth weights (HBW)admitted to the NICU.The aims of this study were to examine (a) risk factors forculture-proven hospital acquired infections for VLBW andHBW infants admitted to the NICU, (b) the risks associ-ated with specific clinical interventions in the NICU and(c) the risk-adjusted variations in hospital acquired infec-tion rates among Canadian NICUs.MethodsStudy populationThe study cohort included all infants admitted to 17Canadian NICUs prior to 4 days of age. Neonates withouta recorded birthweight were excluded. The study cohortwas derived from a larger study of 9506 babies admittedto 17 tertiary NICUs in the Canadian Neonatal Networkfrom January 1996 through to October 1997 [10]. Theseunits represent approximately three-quarters of all the ter-tiary NICU beds in Canada, a nation with >350 000 birthstrained abstractors with standardized definitions andoperations as part of the larger study of Canadian NICUs[10].Variable definitionsFor the purpose of this study, nosocomial infection (orNI) was defined as one or more positive single organismblood or CSF culture sampled after 48 h of admission inan infant with clinical suspicion of infection. To differen-tiate between nosocomial and primary (maternal origin)infections, the infant blood culture isolates were requiredto be different from maternal isolates or to occur at least 7days after a treated positive blood culture obtained duringthe first 48 h of life. CSF samples were obtained whenindicated according to local practices and policies. Thetime period of 2 days was chosen to exclude any neonatesborn with primary infection. Absence of NI includedthose infants who were never cultured and those with neg-ative blood and/or CSF cultures. An infection episode wasdefined as a positive culture occurring at least 7 days aftera previous treated culture or if the culture isolates were dif-ferent from the previous culture [13]. We defined nosoco-mial infections (NI) in this way because blood and CSFare normally sterile, thus a positive culture from one ofthese sites in an unwell baby is likely to be of serious clin-ical significance. Additionally, from a practical perspec-tive, these data are well defined and readily abstractedfrom patient charts.Other study definitions were in accordance with the Cana-dian Neonatal Network Data Abstractor's Manual [14]:gestational age was defined as the attending obstetrician'sbest estimate based on obstetric history and examination,and prenatal ultrasound, except where this was not avail-able or the postnatal physical assessment disagreed withobstetrical estimate by greater than 2 weeks. In that case,the pediatric estimate of gestational age was used instead.Small for gestational age meant being born below the 3rdpercentile for weight (corrected for gestation using growthcharts developed by Whitfield et al [15]). Major congeni-tal and chromosomal anomalies were specified in theAbstractor's Manual. Outborn infants were those born ata different hospital from the admitting NICU. The Scorefor Neonatal Acute Physiology, Version II (SNAP-II) wascollected on day 1 as a measure of patient acuity on theday of admission. The SNAP-II is a validated score [16]using 6 empirically weighted physiological measurementsmade during the first 12 hours after admission to theNICU. To be consistent with previous reports, SNAP-IIscore was categorized as 0 to 9, 10 to 19, 20 to 29 andgreater or equal to 30. Therapy related risk factors wereextracted from day-1 and day-3 evaluations, using a subsetof relevant variables from the Neonatal Therapeutic Inten-Page 2 of 12(page number not for citation purposes)annually and almost 30 million people [11,12]. Informa-tion about these neonates was prospectively collected bysity Scoring System (NTISS) [17]. The subset of variablesincluded the use of assisted ventilation, peripheralBMC Pediatrics 2005, 5:22 http://www.biomedcentral.com/1471-2431/5/22intravenous (IV) access, central venous access, arterial lineaccess, use of antibiotics (day 1 or 3), gavage feeding,intravenous amino acids, and intravenous fat emulsion.Data analysisUnivariate and bivariate analyses using SPSS (Chicago,Illinois) were used to describe the study population andto explore the relationships between baseline populationrisks (gestational age, birth weight, gender, Apgar score at5 minutes, small for gestational age, use of antenatal cor-ticosteroids, multiple births, outborn status, caesareansection, maternal hypertension, congenital anomalies,admission illness severity) and NI. Chi-square test wasused to compare mortality and morbidity among infantswith and without NI, for infants with birth weight lessthan 1500 g and 1500 g or more.We used a multivariable logistic regression model todevelop a risk adjustment model for NI. The outcome var-iable was NI and independent variables were baselinepopulation risks significantly correlated with NI on bivar-iate analysis. Correlation matrix was used to examine cor-relation between variables. The full model was created byadding population risk factors, then SNAP-II score, andfinally therapy-related factors – variables were consideredto be included if their p-value was <0.10. Backward andforward selections from the full model were applied tohelp determine the parsimonious model. Finally, centreswere compared using adjusted odds ratios and 95% con-fidence intervals derived from the multivariable logisticregression analysis, using a reference centre (the hospitalwith the median incidence of nosocomial infection). Sen-sitivity analyses were conducted by excluding outlier siteswith large confidence intervals.Institutional reviewThis study was part of a larger study surveying morbiditiesand mortality in Canada, for which ethics approval foranonymous collation of data was obtained at each indi-vidual centre.ResultsPatient population and nosocomial infection episodes16538 infants in the dataset met entry criteria (age lessthan 4 days on admission). 765 (23.5%) of 3253 VLBWinfants, and 329 (2.5%) of 13244 HBW infants (41 weremissing birth weight and were excluded), developed atleast one episode of NI. Among VLBW infants who devel-oped NI, 78.7% had only 1 episode, 16.2% had 2 epi-sodes, and 5.1% had 3 or more episodes of NI, whereasamong HBW infants, 87.9% had 1 episode, 9.3% had 2episodes, and 2.8% had 3 or more episodes of NI. Themedian onset of the first infection from day of admissionOrganisms and antibioticsAmong infants with the first episode of NI, the incidenceof CSF culture positive infections was 3.7% in VLBWinfants, and 2.8% in HBW infants. The most commonorganism obtained from blood culture was coagulase neg-ative Staphylococcus (72% of isolates in VLBW, and 57%in HBW). The same was true of CSF isolates (68% and41% respectively). Gram-negative organisms were thenext most common pathogen in VLBW, whereas Grampositives, particularly group B Streptococcus, was the nextmost prevalent in HBW.Ninety percent of VLBW infants and 73.3% of HBWinfants received at least one course of antibiotics in theNICU on either day 1 and / or day 3, as recorded by NTISSInfant characteristics and therapy related factorsAmong VLBW infants, bivariate analysis showed thatinfants with NI were more likely to be outborn, deliveredby vaginal birth, and have 5 minute Apgar score <7, butless likely to be SGA (see Table 1). Infants with NI alsohad significantly lower birth weight and gestational age,and higher mean day 1 SNAP-II score, than those withoutNI. A number of therapy related risk factors were signifi-cantly associated with NI: these included use of assistedventilation, peripheral IV access, central venous access,arterial line access, use of antibiotics (on day 1 and/or day3), gavage feeding, IV amino acid, and IV fat. For HBWinfants, NI was associated with similar infant characteris-tics, but also with congenital anomalies, 5 minute Apgarscore <7, and higher SNAP-II score. Correlation betweenvariables: NI has significant correlation with all variablesidentified except for antibiotic use. However, when strati-fied by birth weight, antibiotic use has significant correla-tion with NI for HBW infants. The Spearman correlationcoefficient between CVL and parenteral nutrition is 0.26,which is significant. Furthermore, a cross-tabulation of NIby CVL controlling for IV parenteral nutrition (TPN) indi-cates significant correlation beween NI and CVL and acontingency coefficient of 0.1 for either TPN present orabsent. Therefore, there is no evidence that the presence ofTPN or CVL in the models renders the other redundant.Mortality and morbidityMortality was more strongly associated with the NI groupcompared to the group without NI for HBW infants, butnot for VLBW infants. Major morbidities (necrotizingenterocolitis, chronic lung disease at 36 weeks post-men-strual age, severe intraventricular hemorrhage/periven-tricular lesions, and severe stages (3 and 4) of retinopathyof prematurity) were more common in the NI group(Table 2).Page 3 of 12(page number not for citation purposes)was at 19 days for VLBW infants and 15 days for HBWinfants.BMC Pediatrics 2005, 5:22 http://www.biomedcentral.com/1471-2431/5/22Risk factors predictive of nosocomial infection on multivariable logistic regressionIn VLBW infants, factors associated with NI in the multi-variable logistic model were: gestational age, use ofassisted ventilation and IV fat and, and outborn status (seetable 3). Ninety-nine percent of infants receiving IV fatalso received IV amino acid, so these interventions werecombined. SNAP-II scores were combined into ≥10 and0–9 (the reference) groups because the parameters for thehigher score groups [10–19, 20–29, > = 30) appeared tobe similar. Compared to infants with a gestational age ofgreater than 28 weeks, lower gestation was increasinglymore likely to be associated with NI. SNAP-II score ≥10was of borderline significance.In HBW infants, factors associated (p < 0.05) with NI inthe multivariable logistic model were: gestational age, useoutborn status, use of antibiotics and presence of congen-ital anomalies. Again, IV fat and IV amino acid could beused interchangeably in the model. Infants who receivedparenteral nutrition were more likely to have NI. Centralvenous access was associated with higher risk for NI, aswas use of peripheral IV access.Variations in nosocomial infection rates among NICUs: site comparisonsFor VLBW infants, cross-tabulation of site and NI showedsignificant variation between sites (chi-square = 248.7, df= 16, P < 0.001). Crude rates of NI ranged from 6.7% to74.5% of infants having at least one episode of NI. Thecrude incidences of NI for 8 hospitals were significantly (p< 0.05) different from the reference hospital (site O) (Fig-ure 1). Site variations persisted after adjustment for base-line patient risk factors (gestational age, admission dayTable 1: Characteristics of infants with and without nosocomial infections (NI), for birth weights less than 1500 g and 1500 g or more (asterisks indicate significant difference at p < 0.05 level between infants with and without NI in each birthweight group)Birth weight Less than 1500 g 1500 g or moreCharacterististics With NI (n = 765) Without NI (n = 2488) With NI (n = 329) Without NI (n = 13244)Gestation (mean weeks ± sd) 27.2 +/- 2.3 28.8 +/- 2.8* 35.8 +/- 3.6 36.3 +/- 3.2*Birth weight (mean gm ± sd) 957 +/- 259 1010 +/- 267* 2611 +/- 806 2758 +/- 796*Day 1 SNAP-II (mean ± sd) 18.2 +/- 12.5 14.1 +/- 13.0* 11.9 +/- 12.5 4.9 +/- 8.5*Small for gestational age (%) 9.7 12.6* 2.8 2.7Antenatal steroids (%) 71.4 68.2 21.1 20.4Multiple birth (%) 26.7 28.5 7.9 12.5*Outborn (%) 24.6 12.7* 59.9 23.1*Male (%) 55.3 51.8 59.9 58.1Cesarean delivery (%) 47.5 55.2* 38.9 36.3Maternal hypertension (%) 18.1 19.4 10.4 11.6Apgar @ 5 mins < 7 (%) 26.9 20.0* 21.4 11.9*Congenital Anomalies (%) 10.6 7.9* 48.6 10.9*Table 2: Mortality and morbidity among infants with and without nosocomial infections, for infants with birth weight less than 1500 g and 1500 g or more (asterisks indicate significant difference at p < 0.05 level between infants with and without nosocomial infections within each birth weight group).Birthweight Less than 1500 g 1500 g or moreOutcome (%) With NI (n = 765) Without NI (n = 2488) With NI (n = 329) Without NI (n = 12915)Mortality 8.7 8.6 8.5 1.3*Necrotizing enterocolitis 13.4 4.4* 8.8 0.4*Chronic lung disease 38.7 18.5* 25.7 4.4*Severe intraventricular hemorrhage 10.5 8.2* 5.3 2.8*Severe retinopathy of prematurity 14.6 8.1* 0.0 0.0Survival without major morbidity 46.3 69.5* 74.8 96.0*Page 4 of 12(page number not for citation purposes)of central venous access, peripheral intravenous access,and use of IV fat or IV amino acid, assisted ventilation,SNAP-II score, outborn status) using multivariate logisticregression analysis (Figure 1), with the highest odds ratioBMC Pediatrics 2005, 5:22 http://www.biomedcentral.com/1471-2431/5/22being 8.6 (95% confidence interval (CI) 4.1 to 18.2), andthe lowest 0.2 (95% CI 0.1 to 0.4) when compared to thereference site (Figure 1).For HBW infants, significant variation was also presentbetween sites (chi-square = 360.0, df = 16, P < 0.001).Crude rates of NI ranged from 0.1% to 17.0 %. When onlysite variables were in the regression model, 6 hospitalshad significantly different (p < 0.05) NI rates from the ref-erence hospital (F). Significant site differences largely cor-rected after adjustment for baseline population risks(gestational age, admission day SNAP-II score, outbornstatus) (Figure 2).Comparison of Figures 1 and 2 shows that hospitals withlow and high incidences of NI among VLBW infants alsotended to have lower and higher incidences of NI amonglarger infants. When sites P and N were excluded from theanalysis, site A was also found to have significantly higherincidence of NI than the median hospital for both VLBWand HBW infants. (Figures 3 and 4)DiscussionThe above results represent data from 17 tertiary NICUs,making up approximately three-quarters of level III neo-natal cots in Canada. Canada has publicly-funded healthcare with a highly regionalized neonatal-perinatal caresystem in which each tertiary NICU is the referral centrefor its geographic region. Consequently, our resultsapproximate a population-based study for three quartersof the Canadian population.The bivariate analysis suggests that infants who experienceat least one episode of nosocomial bacteremia or nosoco-mial bacterial meningitis are smaller, more preterm, andtransported from another centre. VLBW infants with NIwere more likely to have been delivered vaginally. Noso-comial infection rates will vary from institution to institu-tion if these demographic factors vary from region toregion. Adjusted data, to some extent, correct for thesegeographic variations.Our data show that significant variations in nosocomialinfection do occur between units, with crude rates ranging6.7% to 74.5% among VLBW infants, and 0.1% to 17.0%among HBW infants. These variations persist after adjust-ment for gestation, outborn status and admission illnessseverity. These findings are not unique to hospitalacquired infections: using data from the same populationof infants, other authors have identified between-site var-iations in the crude and adjusted rates of both mortalityand a number of major neonatal morbidities, includingchronic lung disease, retinopathy of prematurity, catheter-related sepsis and intraventricular lesions [6-8,10,18].There seems little doubt that these differences in outcomeexist, the important question remains as to the underlyingreasons for the differences.Multivariate analysis suggests that, in addition to gesta-tion and acuity, a number of therapeutic interventions areassociated with nosocomial infection. These includemechanical ventilation, vascular access (central andperipheral), and parenteral nutrition. These factors arepredictive in cohorts of both VLBW and HBW babies –supporting their place as clinically significant risks factorsfor NI. The preponderance of NI among HBW babies withcongenital anomalies may reflect surgical interventions,or the possibility of immunocompromise. We recognizethat there are likely to be other, unrecorded demographicfactors or interventions, that may account for risk, such asTable 3: Risk factors associated with nosocomial infection on multivariate analysis, for infants with birth weight less than 1500 g and 1500 g or moreBirth weight Less than 1500 g Odds Ratio (95% confidence intervals)1500 g or more Odds Ratios (95% confidence intervals)Gestation ≤24 weeks 2.7 (1.4 – 5.3) 4.0 (0.5 – 32.7)Gestation 25–28 weeks 2.2 (1.2 – 4.1) 6.1 (1.6 – 23.1)Gestation 29–32 weeks 1.1 (0.6 – 2.0) 1.9 (1.3 – 2.6)SNAP II ≥10 1.2 (1.0 – 1.6) 1.0 (0.8 – 1.3)Peripheral intravenous line 0.9 (0.7 – 1.1) 1.9 (1.1 – 3.2)Central venous Line 0.9 (0.8 – 1.2) 1.6 (1.2 -2.1)Assisted ventilation 1.5 (1.1 – 2.0) 2.9 (1.9 – 4.6)Parenteral nutrition 3.9 (3.0 – 5.2) 5.1 (3.8 – 6.9)Outborn status 2.0 (1.6 – 2.5) 1.9 (1.4 – 2.6)Use of antibiotics 0.9 (0.4 – 2.3) 2.9 (1.6 – 5.0)Congenital anomalies 1.2 (0.9 – 1.6) 2.9 (2.2 – 3.8)Page 5 of 12(page number not for citation purposes)had higher illness severity on admission. They have lower5-minute Apgar scores and are more likely to have beenfeeding practices or ethnicity – these remain unaccountedfor in our study.BMC Pediatrics 2005, 5:22 http://www.biomedcentral.com/1471-2431/5/22Crude and risk adjusted odds ratios of nosocomial infection rates by hospital for infants <1500 g birth weight (point estimates and 95% confidence interval limi s h wn)Figure 1Crude and risk adjusted odds ratios of nosocomial infection rates by hospital for infants <1500 g birth weight (point estimates and 95% confidence interval limits shown). Adjustment is for gestational age, admission day SNAP-II score, and outborn status. NICUOdds-Ratios051015G L C K D J A M O P E B H Q F I NPanel A: Crude VariationNICUOdds-Ratios051015G C D A L K J O B H E M Q F I P NPanel B: Adjusted VariationPage 6 of 12(page number not for citation purposes)Site O, the site closest to the median risk, was used as the reference site.BMC Pediatrics 2005, 5:22 http://www.biomedcentral.com/1471-2431/5/22Crude and risk adjusted odds ratios of nosocomial infection rates by hospital for infants ≥1500 g birth weight (point estimates and 95% confidence interval limi s h wn)Figure 2Crude and risk adjusted odds ratios of nosocomial infection rates by hospital for infants ≥1500 g birth weight (point estimates and 95% confidence interval limits shown). Adjustment is for gestational age, admission day SNAP-II score, and outborn status. NICUOdds-Ratios02468101214G J C D M B L P F K O E Q H A I NPanel A: Crude VariationNICUOdds-Ratios02468101214G J P O D C B E K L A F H N M Q IPanel B: Adjusted VariationPage 7 of 12(page number not for citation purposes)Site O, the site closest to the median risk, was used as the reference site.BMC Pediatrics 2005, 5:22 http://www.biomedcentral.com/1471-2431/5/22Crude and risk adjusted odds ratios of nosocomial infection rates by hospital (with sites P and N excluded) for infants <1500 g bi th weight (point estimates and 95% c nfidence interval limits hown)Figure 3Crude and risk adjusted odds ratios of nosocomial infection rates by hospital (with sites P and N excluded) for infants <1500 g birth weight (point estimates and 95% confidence interval limits shown). Adjustment is for gestational age, admission day SNAP-II score, and outborn status.024681012G L C K D A J M O E B H Q F IPanel A: Crude VariationNICUOdds-Ratio024681012G J D L C B K F M O E H Q A IPanel B: Adjusted VariationNICUOdds-RatioPage 8 of 12(page number not for citation purposes)BMC Pediatrics 2005, 5:22 http://www.biomedcentral.com/1471-2431/5/22Crude and risk adjusted odds ratios of nosocomial infection rates by hospital (with sites P and N excluded) for infants ≥1500 g bi th weight (point estimates and 95% c nfidence interval limits hown)Figure 4Crude and risk adjusted odds ratios of nosocomial infection rates by hospital (with sites P and N excluded) for infants ≥1500 g birth weight (point estimates and 95% confidence interval limits shown). Adjustment is for gestational age, admission day SNAP-II score, and outborn status.024681012G J C D M B F L K E O Q H A IPanel A: Crude VariationNICUOdds-Ratio024681012G J D M C B K F L O E H Q A IPanel B: Adjusted VariationNICUOdds-RatioPage 9 of 12(page number not for citation purposes)BMC Pediatrics 2005, 5:22 http://www.biomedcentral.com/1471-2431/5/22There is little doubt from this study that nosocomial infec-tion is associated with significant in-hospital morbidity,but the results on mortality are more subtle. Unadjustedmortality in VLBW infants was not increased by an epi-sode of nosocomial sepsis. A possible explanation is thatthe effect size is small because most deaths among VLBWinfants occur in the first few days of life from other causes,for example cardiorespiratory failure, whereas NI occursmore commonly at a later age. In addition, coagulase neg-ative staphylococcal infections (the commonest isolate inNICU) are reported to have a relatively low mortality inVLBW infants (2% or less in three large cohort studies [19-21]), making a small contribution to overall mortalitythat may not achieve statistical significance.Between-site analysis confirmed that inter-NICU variationin NI rates persisted after risk adjustment for birth weight,gestation and illness severity in VLBW infants. The 3 siteswith lower than median incidence of NI for VLBW infants,also had a lower incidence of NI for HBW infants. Threeout of 4 sites with higher than median incidence of NI forHBW infants also had higher incidence of NI for VLBWinfants. The impression that overall site performance isconsistent for infants at different birth weight groups sup-ports the contention that factors other that those recordedand adjusted for in this study, probably related to inter-ventions or the NICU environment, contribute to the riskof nosocomial infection – especially in VLBW infants,with whom most of the burden of NI lies.We have seen that nosocomial infection rates vary acrossCanada and are significantly affected by clinical interven-tions. However, a number of important factors have notbeen adjusted for, particularly those related to organismvirulence or infectivity, and population susceptibility.One could hypothesize that organism virulence or infec-tivity varies across the country, as does the susceptibilityof population groups. There is some evidence to supportvariation in genetic susceptibility: Ahrens et al [22]reported that VLBW neonates with specific CD14 genemutations had a greater rate of neonatal sepsis than thosewithout. A more detailed study of population demo-graphics and organism subtypes may reveal whether thesefactors impact on nosocomial infection rates. Similarly,environmental factors such as unit layout and clinicalpractices may affect nosocomial infection rates – theextensive list of potential confounders may require analternative research methodology to discover associationand/or causation. Bloom et al [23] reported decreasedrates of nosocomial infections by sharing informationabout differences between NICUs with high and lownosocomial infection rates, and addressing these varia-tions in clinical practice. The Canadian Neonatal Networknosocomial infection rates, which will hopefully answersome of these questions.A limitation of our study is that it is possible that somepositive blood or CSF cultures are in fact false positives.Struthers et al [24] estimated that 5% of positive bloodcultures were false, suggesting that the vast majority ofconfirmed bacteremia is true. However, it might be arguedthat even false positive results represent a significant bur-den of illness, and including these episodes may fairly rep-resent this burden. Another limitation of this type ofobservational study is the difficulty in differentiatingcause and effect. The observational data in this study werepresented using univariate, unadjusted and risk-adjustedmultivariate models. Unfortunately, these analysescannot account for the temporal relationships betweenevents, nor do they adjust for risk factors that are notincluded in data collection. However, the strengths of thelogistic regression model are its robustness and extensiveuse in similar studies. Future studies should take temporalrelationships into account, permitting time-sensitive anal-ysis, better inference of cause and effect, and assessment ofattributable risk.ConclusionRates of nosocomial infection in Canadian NICUs varyconsiderably, even after correction for known risk factors.The data suggest that this variation is due, in part, to dif-ferences in clinical practices, and therefore may be amena-ble to interventions that alter practice. More study, andalternative study designs, may be required to evaluatecontributing factors and effect practice change.Canadian Neonatal NetworkShoo K. Lee (Coordinator, Canadian Neonatal Network;Centre for Healthcare Innovation and Improvement, Van-couver, BC); Wayne Andrews (Charles A. Janeway ChildHealth Centre, St John's, NF); Ranjit Baboolal (NorthYork Hospital, N. York, ON); Jill Boulton (St Joseph'sHealth Centre, London, ON; previously Mt Sinai Hospi-tal, Toronto, ON); David Brabyn (Royal Columbian Hos-pital, New Westminster, BC); David S.C. Lee (St Joseph'sHealth Centre; London, ON); Derek Matthew (VictoriaGeneral Hospital (Victoria, BC); Douglas D. McMillan(IWK-Grace Health Centre for Women, Children andFamilies, Halifax, NS; formerly Foothills Hospital,Calgary, AB); Christine Newman (Hospital for Sick Chil-dren; Toronto, ON); Arne Ohlsson (Mt Sinai Hospital,Toronto, ON; formerly Women's College Hospital,Toronto, ON); Abraham Peliowski (Royal Alexandra Hos-pital, Edmonton, AB); Margaret Pendray (Children's &Women's Health Centre of British Columbia (Vancouver,BC); Koravangattu Sankaran (Royal University Hospital,Page 10 of 12(page number not for citation purposes)is embarking on a multicentre study to examine the effectof multiple evidence-based practice interventions onSaskatoon, SK); Barbara Schmidt (Hamilton Health Sci-ences Corporation, Hamilton, ON); Mary Seshia (HealthBMC Pediatrics 2005, 5:22 http://www.biomedcentral.com/1471-2431/5/22Sciences Centre, Winnipeg, MB); Anne Synnes (Children'sand Women's Health Centre of British Columbia, Van-couver, BC; formerly Montreal Children's Hospital, Mon-treal, PQ); Paul Thiessen (Children's & Women's HealthCentre of British Columbia, Vancouver, BC); RobinWalker (Children's Hospital of Eastern Ontario andOttawa General Hospital, Ottawa, ON); Robin Whyte(IWK-Grace Health Centre for Women, Children andFamilies, Halifax, NS).Coordinating centre staffHolly Bavinton; Stella Karuri; Zhenguo Qiu; Sarka Lisonk-ova, Shawn Stewart.AbbreviationsNosocomial infection (NI)Neonatal Intensive Care Unit (NICU)Score for Neonatal Acute Physiology, Version II (SNAP-II)Neonatal Therapeutic Intensity Scoring System (NTISS)Very low birth weight (VLBW)Higher birth weight (HBW)Cerebrospinal fluid (CSF)Standard deviation (SD)Confidence interval (CI)Competing interestsThe author(s) declare that they have no competinginterests.Authors' contributionsKA interpreted data and drafted the manuscript. DDM,WA and MP were site investigators. ZQ performed statisti-cal analysis and interpretation. SKL was the principalinvestigator and drafted the manuscript. All these individ-uals read and approved the final manuscript. The CNNrepresents all site investigators, and was responsible fororganization and administration of the SNAP study, andsubsequent data flow.AcknowledgementsRuth Little for editorial assistance and Li-Yin Chien for assistance with data analysis.This study was supported by Grant 40503 and Grant 00152 from the Med-ical Research Council of Canada. Additional funding was provided by the B.C.'s Children's Hospital Foundation; Calgary Regional Health Authority; gram, Health Care Corporation of St John's; The Neonatology Program, Hospital for Sick Children; Lawson Research Institute; Midland Walwyn Capital Inc; Division of Neonatology, Hamilton Health Sciences Corpora-tion; Mount Sinai Hospital; North York General Hospital Foundation; Saint Joseph's Health Centre; University of Saskatchewan Neonatal Research Fund; University of Western Ontario; Women's College Hospital.References1. Stoll BJ, Gordon T, Korones SB, Shankaran S, Tyson JE, Bauer CR,Fanaroff AA, Lemons JA, Donovan EF, Oh W, Stevenson DK, Ehrenk-ranz RA, Papile LA, Verter J, Wright LL: Late-onset sepsis in verylow birth weight neonates: a report from the National Insti-tute of Child Health and Human Development NeonatalResearch Network.  J Pediatr 1996, 129(1):63-71.2. Stoll BJ, Hansen N, Fanaroff AA, Wright L, Carlo WA, EhrenkranzRA, Lemons JA, Donovan EF, Stark AR, Tyson JE, Oh W, Bauer CR,Korones SB, Shankaran S, Laptook AR, Stevenson DK, Papile LA,Poole WK: Late-onset sepsis in very low birth weightneonates: the experience of the NICHD Neonatal ResearchNetwork.  Pediatrics 2002, 110(2 Pt 1):285-91.3. Stevenson DK, Wright LL, Lemons JA, Oh W, Korones SB, Papile LA,Bauer CR, Stoll BJ, Tyson JE, Shankaran S, Fanaroff AA, Donovan EF,Ehrenkranz RA, Verter J: Very low birth weight outcomes of theNational Institute of Child Health and Human DevelopmentNeonatal Research Network, January 1993 through Decem-ber 1994.  Am J Obstet Gynecol 1998, 179:1632-1639.4. Brodie SB, Sands KE, Gray JE, Parker RA, Goldmann DA, Davis R,Richardson DK: Occurrence of nosocomial bloodstream infec-tions in six neonatal intensive care units.  Pediatr Infect Dis J2000, 19(1):56-65.5. Horbar JD, Badger J, Lewit EM, Rogowski J, Shiono PH: Hospital andpatient characteristics associated with variation in 28-daymortality rates for very low birth weight infants.  Pediatrics1997, 99:149-156.6. Sankaran K, Chien LY, Walker R, Seshia M, Ohlsson A, Lee SK, TheCanadian Neonatal Network: Variations in mortality ratesamong Canadian neonatal intensive care units.  CMAJ 2002,166:173-178.7. Chien LY, McNab Y, Aziz K, Andrews W, McMillan DD, Lee SK, TheCanadian Neonatal Network: Variations in central venous cath-eter related infection risks among Canadian neonatal inten-sive care units.  Pediatr Infect Dis J 2002, 21:505-511.8. Synnes A, Chien LY, Peliowski A, Baboolal R, Lee SK, The CanadianNeonatal Network: Variations in Intraventricular HemorrhageIncidence Rates among Canadian Neonatal Intensive CareUnits.  J Pediatr 2001, 138:525-531.9. Edwards WH: Preventing nosocomial bloodstream infectionin very low birth weight infants.  Semin Neonatol 2002,7(4):325-33.10. Lee SK, McMillan D, Ohlsson A, Pendray M, Synnes A, Whyte R,Chien LY, Sale J, the Canadian Neonatal Network: Variations inpractice and outcomes of the Canadian NICU Network :1996–7.  Pediatrics 2000, 106:1070-9.11. Statistics Canada. Births and birth rate   [http://www.statcan.ca/english/Pgdb/People/Population/demo04a.htm]. Accessed February23, 200112. Statistics Canada. Population   [http://www.statcan.ca/english/Pgdb/People/Population/demo02.htm]. Accessed February 23, 200113. Chien LY, MacNab Y, Aziz K, Andrews W, McMillan DD, Lee SK, theCanadian Neonatal Network: Variations in central venous cath-eter-related infection risks among Canadian neonatal inten-sive care units.  Pediatr Infect Dis J 2002, 21:505.14. The Canadian Neonatal Network™: The Canadian NeonatalNetwork™ Data Abstractor's Manual.  Vancouver, B.C; 1996. 15. Whitfield M: British Columbia Provincial Growth Chart.  Van-couver, Canada: British Columbia Children's Hospital; 1996. 16. Richardson DK, Corcoran JD, Escobar GJ, Lee SK: SNAP-II andSNAPPE-II : Simplified newborn illness severity and mortal-ity risk scores.  J Pediatr 2001, 138(1):92-100.17. Gray JE, Richardson DK, McCormick MC, Workman-Daniels K,Goldmann DA: Neonatal therapeutic intervention scoring sys-tem: a therapy-based severity-of-illness index.  Pediatrics 1992,Page 11 of 12(page number not for citation purposes)Dalhousie University Neonatal-Perinatal Research Fund; Division of Neonatology, Children's Hospital of Eastern Ontario; Child Health Pro-90(4):561-7.Publish with BioMed Central   and  every scientist can read your work free of charge"BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime."Sir Paul Nurse, Cancer Research UKYour research papers will be:available free of charge to the entire biomedical communitypeer reviewed and published immediately upon acceptancecited in PubMed and archived on PubMed Central BMC Pediatrics 2005, 5:22 http://www.biomedcentral.com/1471-2431/5/2218. Ohlsson A, Synnes AR, Seshia MMK, Newman CJ, Schmidt B, ChienLY, Lee SK: Variations in the Site Specific Rates of Bronchop-ulmonary Dysplasia (BPD) among Infants Admitted to 17Neonatal Intensive Care Units (NICUs) in Canada – Implica-tions for Bench Marking.  Pediatr Res 2000, 47(4):371A.19. Isaacs D: A ten year, multicentre study of coagulase negativestaphylococcal infections in Australasian neonatal units.  ArchDis Child Fetal Neonatal Ed 2003, 88:F89-F93.20. Karlowicz MG, Buescher ES, Surka AE: Fulminant late-onset sep-sis in a neonatal intensive care unit, 1988-and the impact ofavoiding empiric vancomycin therapy.  Pediatrics 2000,106(6):1387-90.21. Stoll BJ, Hansen N, Fanaroff AA, Wright LL, Carlo WA, EhrenkranzRA, Lemons JA, Donovan EF, Stark AR, Tyson JE, Oh W, Bauer CR,Korones SB, Shankaran S, Laptook AR, Stevenson DK, Papile LA,Poole WK: Late-onset sepsis in very low birth weightneonates: the experience of the NICHD Neonatal ResearchNetwork.  Pediatrics 2002, 110(2 Pt 1):285-91.22. Ahrens P, Kattner E, Köhler B, Härtel C, Seidenberg J, Segerer H,Moller J, Gopel W: Genetic Factors in Neonatology StudyGroup. Mutations of genes involved in the innate immunesystem as predictors of sepsis in very low birth weightinfants.  Pediatr Res 2004, 55:652-656.23. Bloom BT, Craddock A, Delmore PM, Kurlinski JP, Voeloker M, Land-fish N, Rodriguez-Pierce M, Swanton D, Rossi J, Ehlen J, Harmon C,Deterding J, Houser F: Reducing acquired infections in theNICU: Observing and implementing meaningful differencesin process between high and low acquired infection ratecentres.  J Perinatol 2003, 23:489-492.24. Struthers S, Underhill H, Albersheim S, Greenberg D, Dobson S: Acomparison of two versus one blood culture in the diagnosisand treatment of coagulase-negative staphylococcus in theneonatal intensive care unit.  J Perinatol 2002, 22:547-549.Pre-publication historyThe pre-publication history for this paper can be accessedhere:http://www.biomedcentral.com/1471-2431/5/22/prepubyours — you keep the copyrightSubmit your manuscript here:http://www.biomedcentral.com/info/publishing_adv.aspBioMedcentralPage 12 of 12(page number not for citation purposes)


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items