Infants in neonatal intensive care units (NICUs) are vulnerable to a variety of infections, and occupancy in the unit may correlate with risk of infection.
A retrospective cohort of infants admitted to the NICUs between 1997 and 2014. Survival analysis was used to model the relative hazard of sepsis infection in relation to two measures of occupancy: 1) the average census and 2) proportion of infants <32 weeks gestation in the unit.
There were 446 (2.3%) lab-confirmed cases of bacterial or fungal sepsis, which steadily declined over time. For each additional percentage of infants <32 weeks gestation, there was an increased hazard of 2% (hazard ratio 1.02, 95% confidence interval: 1.00, 1.03) over their NICU hospitalization. Census was not associated with risk for infection.
During times of a greater proportion of infants <32 weeks gestation in the NICU, enhanced infection-control interventions may be beneficial to further reduce the incidence of infections.
Infants in neonatal intensive care units (NICUs) are vulnerable to a variety of infections, some of which may lead to late-onset sepsis and cause severe morbidity and mortality, especially among preterm infants.1 Specific risk factors for late-onset sepsis include younger gestational age, younger maternal age and lower APGAR scores2 in addition to invasive procedures, such as the number of central lines and use of mechanical ventilation.1, 3 Risk for infection is also influenced by care practices, including hand hygiene, disinfection, and antisepsis practices, often rolled into maintenance bundles.4, 5 A breakdown in these practices may result in increased exposure to infectious pathogens from the caregivers, visitors or the environment, possibly leading to late-onset sepsis.6 Potential pathways toward a breakdown in preventable infection include a higher census in the NICU, as well as a higher proportion of premature infants, thereby putting additional demands on the staff in the unit.
A 2015 study by Julian et al.7 found the configuration of beds in the NICU (single-patient rooms versus shared open room) was not associated with the incidence of late-onset sepsis; however, a high average census upon admission was correlated with methicillin-resistant Staphylococcus aureus (MRSA) colonization in single-patient rooms. The relationship between high bed occupancy, length of stay and spread of multidrug-resistant bacteria has been observed elsewhere.8 However, data are limited for the NICU and more research has been requested.9
Complicating the study of late-onset sepsis, the epidemiology of infection may change over time. In a study of long-term trends of infection in the NICU, van den Hoogen et al.10 observed that pathogens associated with late-onset sepsis shifted over time, yet clear patterns emerged. Gram-positive bacteria were cultured in the majority of infections, including coagulase-negative staphylococci (CoNS) and S. aureus,1, 2, 10 and became a more frequent cultured agent of late-onset sepsis over the past few decades. Gram-negative bacteria (mainly Escherichia coli) steadily declined over the same time.10 Although van den Hoogen et al.10 noted fungal infections due to Candida species remained sporadic during follow-up, others have observed wide variation in the incidence of candidiasis,11 possibly related to antifungal prophylaxis. Complicating the treatment of bloodstream infections has been the rise in antimicrobial resistance, particularly with MRSA infections.1
To our knowledge, no study has examined the nuanced role of occupancy in the NICU in relationship to late-onset sepsis, thus our primary aim was to address the question, ‘Does a higher census and/or higher proportion of infants <32 weeks gestation among the departmental census increase risk for bloodstream infection leading to sepsis?’ Secondarily, we examined secular changes in cultured organisms during a 17-year period in a regional perinatal referral center.
Materials and methods
The study population was drawn from the Christiana Care Health System NICU (Newark, DE, USA), a regional perinatal referral center with 1100 admissions per year, 60 beds and the ability to flex to 70 beds. A retrospective cohort comprised all initial admissions to the NICU from 1997 through 2014, inclusive. This 17-year period corresponds to the hospital’s open-bay pod design for the NICU. Patient data were retrieved from the Neonatal Information System (Medical Data Systems, Villanova, PA, USA), a data management system capturing clinical care in the NICU, and validated by trained staff prior to finalization. The study was approved by the hospital’s Institutional Review Board.
The primary exposure was two measures of occupancy in the unit: the average census and the proportion of infants <32 weeks gestation relative to the number of occupied beds. The average census was operationalized by tallying the daily number of occupied beds in the unit and dividing by the length of stay. The proportion of infants <32 weeks gestation was operationalized by tallying the daily number of infants <32 weeks gestation in the unit and dividing by the total census. These measures were calculated for each infant's stay in the NICU until censoring or diagnosis of late-onset sepsis.
The outcome of interest was culture-proven late-onset sepsis, confirmed by laboratory isolation of an infectious organism after 72-h NICU admission. We further analyzed cultured organisms to detect changes over time and grouped pathogens into one of the eight categories as follows: S. aureus, group B beta-hemolytic streptococcus, CoNS, Gram-positive other (alpha-hemolytic streptococci, group A beta-hemolytic streptococcus, Listeria monocytogenes or other Gram-positive cocci), E. coli, Gram-negative other (Acinetobacter, Enterobacter, Klebsiella, Pseudomonas, Salmonella or Serratia spp.), fungal (Candida spp.) or other (unidentified) organisms. S. aureus was further classified as methicillin sensitive or methicillin resistant (MRSA), if known.
Covariates included maternal age, race and Hispanic ethnicity; delivery method (vaginal or Cesarean section); clinical chorioamnionitis; premature rupture of membranes; gestational age at delivery (term or preterm (<37 weeks)); mechanical ventilation; and year and seasonality of admission to check for secular trends.
The effect of occupancy on the risk of late-onset sepsis in the cohort was modeled using Cox proportional hazards regression controlling for potential confounding. Cohort time was comprised of the length of stay in the NICU (until death, transfer or discharge) or time to diagnosis of sepsis. Although time to diagnosis was not of primary interest, the use of Cox regression provides an appropriate method for assessing cumulative incidence and accounts for differences in cohort time (longer lengths of stay will have a higher chance of infection).12 Covariates were included as potential confounders if they were associated with both the exposure and outcome (P<0.20) or if they changed this relationship by >10%.13 For multiple confirmed occurrences of bloodstream infections, only the first episode was modeled. Covariate violations of the Cox proportional hazards assumption were evaluated with a mix of graphical methods, global goodness of fit testing and individual tests of significance (P<0.05) for covariate interactions with cohort time in extended Cox models. The final model yielded point estimates of hazard ratios (HRs) and 95% confidence intervals (CIs) predicting the relative hazard of late-onset sepsis over the patient’s length of stay and can be interpreted as cumulative incidence relative risks.
Through a sensitivity analysis, we quantify the effects of varying the gestational age of the primary exposure from proportion of infants <32 weeks gestation to <28 weeks and <25 weeks acknowledging the increase in risk of late-onset sepsis for shorter gestations.14 All analyses were conducted using SAS 9.4 (SAS Institute, Cary, NC, USA) and R 3.2.1 (R Foundation for Statistical Computing, Vienna, Austria).
Between 1997 and 2014, there were 19 810 infants admitted to the hospital’s NICU. The mean maternal age was 28 years, 62% (n=12 274) of infants were born to non-Hispanic white mothers, 29% (n=5682) were born to non-Hispanic black mothers and 9% (n=1749) were born to Hispanic mothers. C-section deliveries accounted for 48% (n=9518) of the neonates, and the median length of stay in the NICU was 8 days (interquartile range: 4, 16). The average census during cohort follow-up was 48 occupied beds (s.d.: 7.6) and the overall mean proportion of infants <32 weeks gestation was 50% (s.d.: 8.2), indicating that during an average stay in the NICU, half of the census was comprised of infants <32 weeks gestation.
Characteristics of the cohort stratified by outcome are summarized in Table 1. There were a total of 446 (2.3% of cohort) confirmed cases of late-onset sepsis during follow-up, and the proportion of infants <32 weeks gestation was greater among cases (53%) than non-cases (50%). There was no apparent difference by the average census (48 among non-cases vs 47 among cases). The crude incidence rate of sepsis at cohort inception was 3.1 cases per 1000 NICU days, decreasing to 0.5 cases per 1000 NICU days at the conclusion of follow-up (P<0.01 for trend).
Table 2 presents the blood culture results from the 446 confirmed cases of late-onset sepsis. Gram-positive bacteria, specifically CoNS (70%) and S. aureus (11%), accounted for the majority of infections (87%). Most S. aureus infections (66%) were found to be methicillin sensitive. E. coli was cultured in 3% of qlate-onset sepsis cases (34% of total Gram-negative infections). Candida species were also found in 4% of the sepsis cases. The incidence of sepsis decreased markedly over time (Figure 1), although the relative proportions of each organism group to the total cases of sepsis per year remained similar.
While checking for potential confounding and assessing models for significant time-varying exposure effects, multiple methods showed that there were no significant violations to the proportional hazards assumptions. All time interactions in the extended models were non-significant and were supported by visual inspection of the hazard curves. Thus the final model included all main effects and no interactions with cohort time or other covariates. Maternal race/ethnicity, gestational age, mechanical ventilation, seasonality and year of admission were found to be potential confounders and were also included in the final model with the exposures of average census and proportion of infants <32 weeks gestation.
For each additional percentage of infants <32 weeks gestation in the unit, neonates had an increased late-onset sepsis hazard of 2% (HR 1.02, 95% CI: 1.00, 1.03) over their NICU hospitalization, controlling for potential confounding (Table 3). Varying the gestation to <28 weeks and <25 weeks, the risk remained: HR 1.03 (95% CI: 1.01, 1.04) and HR 1.02 (95% CI: 0.99, 1.06), respectively (data not shown). The average census was not an independent predictor of infection. Neonates with black or Hispanic mothers had higher late-onset sepsis hazards of 51% (HR 1.51, 95% CI: 1.24, 1.85) and 55% (HR 1.55, 95% CI: 1.11, 2.17), respectively, than the non-black and non-Hispanic reference group. Compared with term delivery, preterm infants had in excess of a twofold average increase in the association with late-onset sepsis (HR 2.35, 95% CI: 1.39, 3.96). Being on mechanical ventilation was strongly associated with increased risk (HR 9.34, 95% CI: 6.55, 13.32), compared with no ventilation.
We detected secular changes in the cohort. For each later year in the date of admission, there was a corresponding 8% decrease in risk for sepsis in the cohort (HR 0.92, 95% CI: 0.89, 0.94). Infants admitted during the spring had an increased risk for sepsis compared with infants admitted during the winter (HR 1.33, 95% CI: 1.02, 1.74), controlling for potential confounding including admission year. Fall and summer admissions were not apparently different from winter.
Given that very low birth weight neonates (<1500 g) are particularly vulnerable to infections in the NICU,13 we calculated stratum-specific estimates for the proportion of infants <32 weeks gestation exposure but did not detect any effect modification (P=0.43, data not shown). We further hypothesized a priori that this exposure would be modified by the census but also did not detect any effect modification (P=0.58, data not shown).
In this retrospective cohort analysis, we assessed the changing incidence of late-onset sepsis during a 17-year period in the hospital’s NICU and tested two measures of occupancy with the hypothesis that an increase in exposure would correlate to an increase in infection. We found that the overall incidence of bloodstream infections leading to sepsis declined over time; however, types of cultured pathogens did not markedly differ during the study. Further, we noted that infants present in the NICU at any time when there are proportionally more infants of shorter gestations had an increased risk for late-onset sepsis.
The decrease in temporal incidence of sepsis may be partially attributed to specific hospital-wide quality improvements, including central line insertion and maintenance bundles, antifungal prophylaxis and a focus on hand hygiene and gloving. Invasive procedures, such as central line insertion, are commonplace in the NICU and present an increase in the risk for infection.15 In a study of 18 NICUs in New York state, Schulman et al.16 examined preuse and postuse of central line maintenance checklists and noted a 67% decrease in central-line-associated bloodstream infections. In a 2015 Cochrane review, the use of antifungal agents for prophylaxis of invasive fungal infections was associated with an 80% decrease in invasive infections.17 Finally, proper hand hygiene is a well-known correlate of protection against infection in the NICU and has been found to reduce the incidence of infections in repeated studies.18, 19, 20 Others have noted that hand hygiene prior to nonsterile gloving may further reduce bloodstream infections in the NICU21 and from 2001 to 2010 was universally practiced in our NICU.
Consistent with other studies of bacterial and fungal infections leading to late-onset sepsis in the NICU, the majority of cultured organisms were CoNS, S. aureus, E. coli and Candida species.1, 10, 22 Most S. aureus infections were found to be methicillin sensitive, with MRSA infections peaking in the early 2000s (data not shown). The decline in MRSA may be associated with a specific NICU quality improvement program and processes focusing on screening and treating for MRSA colonization proactively in our hospital. However, some MRSA infections may remain endemic, even with strict adherence to hygiene practices or other prevention measures.23
There are limited studies on the relationship of census to bloodstream infections in the NICU. Most recently, Julian et al.7 noted an average 31% increase in the association of census with MRSA colonization in single-patient rooms; this effect was not observed in open-unit rooms. In an outbreak of Salmonella in a NICU in Tennessee, the horizontal transmission throughout the unit was believed to be influenced by a high census.24 To our knowledge, this is the first study to investigate the effect of census and the proportion of infants based on gestational age as occupancy risks for sepsis over time.
The nature of an occupancy effect is dependent on several factors, including its measure and definition. We considered two measures of occupancy: the census and the proportion of infants based on gestational age. There is no standard method of modeling the NICU census. We operationalized census based on follow-up time in the cohort (analogous to the area under the curve), corresponding to an aggregate exposure altering a patient’s risk of infection, rather than a one-time exposure. Census can be potentially measured in other ways. For example, an average number occupied beds upon admission has been used.7 Although this may place a patient at an altered risk of infection proximal to the admission process, we believe checking the daily census during the entire stay (or until the diagnosis of sepsis) more accurately captures the census exposure effect. Further, the proportion of infants <32 weeks gestation may be a more meaningful measure of occupancy, as a higher census with relative few premature infants places fewer demands on the unit as compared with a lower census with relatively high number of premature infants. Although the proportion of infants <32 weeks gestation effect reported in this study may appear modest (2% average increase in incidence for each percent change), if two otherwise similar infants are present in the NICU during markedly different periods, the effect could be quite dramatic. For example, a neonate present when the proportion of infants <32 weeks gestation is 60% may have an average 40% risk increase for late-onset sepsis compared with a neonate present when the proportion is 40%.
In searching for the causal pathway relating the very preterm exposure to an increased risk for late-onset sepsis, we view this measure of occupancy as a proxy for the true risk factor(s). For example, it may reflect acuity of the unit, increased visitation, change in clinical care based on time constraints or an alteration of simple prevention strategies, including hand hygiene and gloving. Any one of these factors could be a contributor towards increased pathogens present in the unit and/or a breakdown in hand hygiene practices, central line care and so on. Although we did not have specific markers for factors such as staffing ratio or proper use of alcohol-based hand sanitizers (or hand washing), these factors have been associated with a risk for infection in other NICUs.17, 18, 25, 26 An acuity-adjusted measure of occupancy, such as a census weighted for nurse-to-patient ratio, may be a more informative measure.27 Nevertheless, the finding of an increased risk of infection suggests a need for additional awareness in the NICU of infection-control procedures when there are a greater proportion of infants of shorter gestations.
In addition to our principal findings, we noted several other interesting associations. Infants admitted to the NICU of non-Hispanic black and Hispanic mothers were at an increased risk of late-onset sepsis, relative to non-Hispanic white mothers. This finding is consistent with other literature on the relationship between race, ethnicity and neonatal morbidity, and while the exact role that race and ethnicity play has been well studied, its effects remains unclear.28 The finding of a seasonal effect of late-onset sepsis indicated that the warmer months experienced a higher incidence of bloodstream infections. Given the diversity of pathogens cultured, the effect is an average of many organisms, rather than attributed to a specific organism. Seasonality of specific bacterial organisms leading to infection has previously been described.29 For example, E. coli infections have been found to peak during the warmest months in Minnesota.30 Nationally, Acinetobacter species have been found highest during July through October compared with November through June,31 and worldwide Klebsiella pneumoniae has been found to peak during the four warmest months of the year.32 Consistent throughout these studies are the increase in bacterial infections during the warmer months, which may prove useful for infection-control awareness during summer.
There are limitations to this work. In addition to the aforementioned exposure ‘proxy’ effect, other hospital NICUs may have differing infection-control procedures. Thus generalizability should be evaluated if a high proportion of critically ill infants triggers certain hospital infection-control measures that are different from the current institution and may modify the exposure-to-outcome relationship. Study strengths include a large population with adequate statistical power and use of a database with nurse-reviewed clinical data.
In this study, we demonstrated that a proportionally greater number of infants <32 weeks gestation may increase the likelihood of a bloodstream infection leading to late-onset sepsis. Other institutions can readily adapt this occupancy measure, and other markers of acuity can be substituted (for example, proportion of very low birth weight infants). Our data suggest that late-onset sepsis is decreasing in incidence and infection-prevention strategies should specifically include measures of occupancy in the NICU.
Shane AL, Stoll BJ . Recent developments and current issues in the epidemiology, diagnosis, and management of bacterial and fungal neonatal sepsis. Am J Perinatol 2013; 30 (2): 131–141.
Cohen-Wolkowiez M, Moran C, Benjamin DK, Cotten CM, Clark RH, Benjamin DK Jr et al. Early and late onset sepsis in late preterm infants. Pediatr Infect Dis J 2009; 28 (12): 1052–1056.
Hilde Verstraete E, De Coen K, Vogelaers D, Blot S . Risk factors for health care-associated sepsis in critically ill neonates stratified by birth weight. Pediatr Infect Dis J 2015; 34 (11): 1180–1186.
Helder OK, Brug J, van Goudoever JB, Looman CW, Reiss IK, Kornelisse RF . Sequential hand hygiene promotion contributes to a reduced nosocomial bloodstream infection rate among very low-birth weight infants: an interrupted time series over a 10-year period. Am J Infect Control 2014; 42 (7): 718–722.
Shepherd EG, Kelly TJ, Vinsel JA, Cunningham DJ, Keels E, Beauseau W et al. Significant reduction of central-line associated bloodstream infections in a network of diverse neonatal nurseries. J Pediatr 2015; 167 (1): 41–6.e1-3.
Bentlin MR, Rugolo LM, Ferrari LS, Brazilian Neonatal Research Network (Rede Brasileira de Pesquisas Neonatais). Practices related to late-onset sepsis in very low-birth weight preterm infants. J Pediatr (Rio J) 2015; 91 (2): 168–174.
Julian S, Burnham CA, Sellenriek P, Shannon WD, Hamvas A, Tarr PI et al. Impact of neonatal intensive care bed configuration on rates of late-onset bacterial sepsis and methicillin-resistant Staphylococcus aureus colonization. Infect Control Hosp Epidemiol 2015; 25: 1–10.
Kaier K, Meyer E, Dettenkofer M, Frank U . Epidemiology meets econometrics: using time-series analysis to observe the impact of bed occupancy rates on the spread of multidrug-resistant bacteria. J Hosp Infect 2010; 76 (2): 108–113.
Sherenian M, Profit J, Schmidt B, Suh S, Xiao R, Zupancic JA et al. Nurse-to-patient ratios and neonatal outcomes: a brief systematic review. Neonatology 2013; 104 (3): 179–183.
van den Hoogen A, Gerards LJ, Verboon-Maciolek MA, Fleer A, Krediet TG . Long-term trends in the epidemiology of neonatal sepsis and antibiotic susceptibility of causative agents. Neonatology 2010; 97 (1): 22–28.
Benjamin DK Jr, Stoll BJ, Gantz MG, Walsh MC, Sánchez PJ, Das A et al. Neonatal candidiasis: epidemiology, risk factors, and clinical judgment. Pediatrics 2010; 126 (4): e865–e873.
Kleinbaum D, Klein M . Survival Analysis. Springer Science+Business Media: New York, NY, USA, 2012.
Rothman KJ, Greenland S, Lash TL (eds). Modern Epidemiology. Lippincott Williams & Wilkins: Philadelphia, PA, USA, 2008.
Hornik CP, Fort P, Clark RH, Watt K, Benjamin DK Jr, Smith PB et al. Early and late onset sepsis in very-low-birth-weight infants from a large group of neonatal intensive care units. Early Hum Dev 2012; 88 (Suppl 2): S69–S74.
Mahieu LM, De Muynck AO, Ieven MM, De Dooy JJ, Goossens HJ, Van Reempts PJ . Risk factors for central vascular catheter-associated bloodstream infections among patients in a neonatal intensive care unit. J Hosp Infect 2001; 48 (2): 108–116.
Schulman J, Stricof R, Stevens TP, Horgan M, Gase K, Holzman IR et al. Statewide NICU central-line-associated bloodstream infection rates decline after bundles and checklists. Pediatrics 2011; 127 (3): 436–444.
Austin N, Cleminson J, Darlow BA, McGuire W . Prophylactic oral/topical non-absorbed antifungal agents to prevent invasive fungal infection in very low birth weight infants. Cochrane Database Syst Rev 2015; 10: CD003478.
Mukerji A, Narciso J, Moore C, McGeer A, Kelly E, Shah V . An observational study of the hand hygiene initiative: a comparison of preintervention and postintervention outcomes. BMJ Open 2013; 3 (5): 1–7.
Pessoa-Silva CL, Hugonnet S, Pfister R, Touveneau S, Dharan S, Posfay-Barbe K et al. Reduction of health care associated infection risk in neonates by successful hand hygiene promotion. Pediatrics 2007; 120 (2): e382–e390.
Won SP, Chou HC, Hsieh WS, Chen CY, Huang SM, Tsou KI et al. Handwashing program for the prevention of nosocomial infections in a neonatal intensive care unit. Infect Control Hosp Epidemiol 2004; 25 (9): 742–746.
Kaufman DA, Blackman A, Conaway MR, Sinkin RA . Nonsterile glove use in addition to hand hygiene to prevent late-onset infection in preterm infants: randomized clinical trial. JAMA Pediatr 2014; 168 (10): 909–916.
Testoni D, Hayashi M, Cohen-Wolkowiez M, Benjamin DK Jr, Lopes RD, Clark RH et al. Late-onset bloodstream infections in hospitalized term infants. Pediatr Infect Dis J 2014; 33 (9): 920–923.
Popoola VO, Budd A, Wittig SM, Ross T, Aucott SW, Perl TM et al. Methicillin-resistant Staphylococcus aureus transmission and infections in a neonatal intensive care unit despite active surveillance cultures and decolonization: challenges for infection prevention. Infect Control Hosp Epidemiol 2014; 35 (4): 412–418.
Boehmer TK, Bamberg WM, Ghosh TS, Cronquist A, Fornof ME, Cichon MK et al. Health care-associated outbreak of Salmonella Tennessee in a neonatal intensive care unit. Am J Infect Control 2009; 37 (1): 49–55.
Cimiotti JP, Haas J, Saiman L, Larson EL . Impact of staffing on bloodstream infections in the neonatal intensive care unit. Arch Pediatr Adolesc Med 2006; 160 (8): 832–836.
Parry GJ, Tucker JS, Tarnow-Mordi WO, UK Neonatal Staffing Study Group. Relationship between probable nosocomial bacteraemia and organisational and structural factors in UK neonatal intensive care units. Qual Saf Health Care 2005; 14 (4): 264–269.
Gagliardi L, Corchia C, Bellù R, Coscia A, Zangrandi A, Zanini R . What we talk about when we talk about NICUs: infants' acuity and nurse staffing. J Matern Fetal Neonatal Med 2015; 19: 1–20.
Petrova A, Mehta R, Anwar M, Hiatt M, Hegyi T . Impact of race and ethnicity on the outcome of preterm infants below 32 weeks gestation. J Perinatol 2003; 23 (5): 404–408.
Perencevich EN, McGregor JC, Shardell M, Furuno JP, Harris AD, Morris JG Jr et al. Summer peaks in the incidences of Gram-negative bacterial infection among hospitalized patients. Infect Control Hosp Epidemiol 2008; 29 (12): 1124–1131.
Al-Hasan MN, Lahr BD, Eckel-Passow JE, Baddour LM . Seasonal variation in Escherichia coli bloodstream infection: a population-based study. Clin Microbiol Infect 2009; 15 (10): 947–950.
McDonald LC, Banerjee SN, Jarvis WR . Seasonal variation of Acinetobacter infections: 1987-1996. Nosocomial Infections Surveillance System. Clin Infect Dis 1999; 29 (5): 1133–1137.
Anderson DJ, Richet H, Chen LF, Spelman DW, Hung YJ, Huang AT et al. Seasonal variation in Klebsiella pneumoniae bloodstream infection on 4 continents. J Infect Dis 2008; 197 (5): 752–756.
We thank Gina Moore, Christiana Care Health System, for her assistance with data abstraction and interpretation and Josh Kessler for his assistance in data collection. This work was supported by INBRE grant NIH-NIGMS (p20 BM103446).
The authors declare no conflict of interest.
About this article
Cite this article
Goldstein, N., Eppes, S., Ingraham, B. et al. Characteristics of late-onset sepsis in the NICU: does occupancy impact risk of infection?. J Perinatol 36, 753–757 (2016). https://doi.org/10.1038/jp.2016.71
Advances in Neonatal Care (2021)