Selection of sources of evidence

The initial search yielded 610 studies for title and abstract review, of which four duplicates were removed. Forty-seven studies were selected for full-text review, of which 37 were excluded at this stage, leaving 10 studies included in this scoping review (Figure 1).

Figure 1.

PRISMA flow diagram.

Figure 1. Long description

The flowchart begins with the identification phase, where 610 studies are sourced from databases/ registers, specifically MEDLINE. No references are obtained from other sources such as citation searching or grey literature. Four references are removed due to duplicates identified manually and by Covidence. In the screening phase, 606 studies are screened, with 559 studies excluded. Forty-seven studies are sought for retrieval, and all are successfully retrieved. These 47 studies are assessed for eligibility, with 37 studies excluded for various reasons including non-English language, review paper status, wrong setting, wrong outcomes, wrong comparator, wrong intervention, wrong study design, wrong patient population, and wrong format. Finally, 10 studies are included in the review.

Characteristics of included studies

The 10 included studies were from two countries (the United States and the Netherlands), and consisted of six retrospective cohort studies, two prospective cohort studies, one matched case-control study, and one study with unclear methodology (Table 1). Of these, eight studies compared outcomes before and after moving to a new ward, whereas two studies compared neonatal rooms with differing designs within the same hospital.

Table 1.

Characteristics and results of 10 studies evaluating neonatal room design and Healthcare-Associated Infection (HAI) outcomes

Table 1. Long description

A table with 10 rows and 7 columns comparing studies on neonatal room design and healthcare-associated infection outcomes. Columns include Study, Setting, Design, N, Comparison, HAI outcomes assessed, Additional non-HAI outcomes assessed, and Results: Effect of SFR on HAI outcomes. Each row provides details for a specific study, including the study name, location, study design, number of participants, comparison groups, healthcare-associated infection outcomes assessed, additional non-HAI outcomes assessed, and the results of the study on HAI outcomes. The table captures various studies conducted in the United States and the Netherlands, with different study designs and comparisons.

SFR, Single Family Room; NICU, Neonatal Intensive Care Unit; CLABSI, Central Line-Associated Bloodstream Infection; MDRO, Multidrug-Resistant Organism; MRSA, Methicillin-Resistant Staphylococcus aureus; MSSA, Methicillin Sensitive Staphylococcus aureus.

*A significant decrease is indicated by ↓; a significant increase is indicated by ↑. The = symbol indicates no significant change in that outcome.

Four studies assessed sepsis incidence. CLABSIs, MDROs, and MRSA were each examined by two studies. Additional infection outcomes assessed by individual studies included Klebsiella pneumoniae, Escherichia coli, Enterobacter cloacae, MSSA, third-generation cephalosporin-resistant bacteria (3G-CRB), and nonspecific HAIs.

Sepsis

An important infectious outcome and high-risk factor for neonatal morbidity is sepsis. Yet, studies reported conflicting findings regarding the association between NICU design and sepsis incidence. One prospective single center study estimated that unadjusted neonatal sepsis incidence was reduced from 11% (14 cases) in the open-bay unit (OBU) to 6% (6 cases) in the SFR unit. ^{Reference Domanico, Davis, Coleman and Davis12} Similarly, transitioning from OBU to SFR was associated with decreased odds of late-onset sepsis (OR 0.486, 95% CI 0.293–0.807), and decreased length-of-stay in a Dutch level II NICU. ^{Reference Van Veenendaal, Van Der Schoor and Heideman13} Notably, this study identified decreased parenteral nutrition following implementation of family-integrated care in SFRs as a primary mediator of this association. ^{Reference Van Veenendaal, Van Der Schoor and Heideman13} Conversely, a 5-year study of nearly 10,000 NICU encounters in a public safety-net hospital found no significant change in sepsis incidence before and after a switch from OBU to SFR. ^{Reference Puumala, Rich and Roy7} Furthermore, in the same study, although a decline in sepsis incidence was observed among extremely preterm infants following the transition to a SFR design, the change was associated with an increase in sepsis incidence among term and postterm infants. ^{Reference Puumala, Rich and Roy7} In sum, two studies found SFRs decrease sepsis incidence, whereas one study found no significant change, and another study found increased sepsis among term and postterm infants in SFRs. These contrasting findings may be due to changes in staffing patterns, other parallel quality improvement efforts targeting prematurity, feeding regimens, or data availability for sepsis diagnoses. Nevertheless, these findings suggest that architectural factors alone may not broadly improve patient outcomes, but there may be subsets of patients for whom NICU design elements may offer an advantage.

CLABSIs

Findings on the association between NICU design and CLABSIs were similarly mixed. One study reported a CLABSI decrease from 10.1 per 1,000 device days in the OBU to 3 per 1,000 device days in SFRs. ^{Reference Walsh, Mccullough and White5} However, this study was primarily focused on nurses’ perceptions of the ward design and lacked methodological information regarding the neonatal population used to calculate incidence. In contrast, a rigorous Dutch study that performed univariate and multivariate analyses of more than 700 infants born at <32 weeks gestational age found no statistically significant difference in HAI incidence, including CLABSI, upper respiratory tract infections, urinary tract infections, or other outcomes such as necrotizing enterocolitis and antibiotic use between the OBU and SFR groups. ^{Reference Jansen, Lopriore and Berkhout14}

MRSA and MSSA colonization

To broaden our review, we included studies that evaluated colonization in the NICU. Surprisingly, we found that SFR design may increase colonization rates in neonates. Although one single-center retrospective study in a hospital with SFRs and OBUs found no significant difference in MRSA colonization rates, late-onset sepsis, or mortality between infants cared for in the two layouts, they noted that higher average daily census was associated with increased MRSA colonization only in the SFR (hazard ratio 1.31, 95% CI 1.02–1.68). ^{Reference Julian, Burnham and Sellenriek15} Across both unit configurations, the only factor that consistently reduced MRSA colonization was hand hygiene compliance upon entry and exit. ^{Reference Julian, Burnham and Sellenriek15} This finding is consistent with a 2023 quality improvement study, identified while reviewing supporting literature, that found an increase in MRSA infection and colonization after transitioning to a SFR design, which was hypothesized to be associated with reduced hand hygiene in the SFR NICU. ^{Reference Barrett, Fleiss and Hansen16} Additionally, one of the included single-center studies that performed metagenomic sequencing on stool and groin swabs from infants before and after transitioning from an OBU to SFR NICU found that infants in the SFR harbored a higher abundance of antibiotic resistance and virulence genes despite lower overall antibiotic usage. ^{Reference Hourigan, Subramanian and Hasan17} Although the individual bacterial, viral, and fungal taxa were similar between cohorts, skin microbial profiles differed significantly between groups (P = .0001). ^{Reference Hourigan, Subramanian and Hasan17}

MDROs

In another challenge to the assumption that SFRs may be protective for infants, a single-center study that collected weekly throat and rectal surveillance cultures from 1,293 infants during the two years before and after transitioning from an OBU to SFR design found that the transition to SFRs did not significantly reduce rates of colonization with MDROs or 3G-CRB. ^{Reference Van Der Hoeven, Bekker and Jansen18} Moreover, in a 2-year period, another single-center study in the Netherlands identified one MDRO cluster among four patients in an OBU and four MDRO clusters among nine patients following the change to a SFR design. ^{Reference Van Der Hoeven, Jansen and Kraakman19} Together, these findings challenge the assumption that SFRs may be protective for infants and highlight potential multifactorial reasons for these patterns. Reasons include staffing model changes related to SFRs, Hawthorne effect in open bay settings which may improve hand hygiene compliance, limited direct exposure to the environment in open bay settings due to space restrictions, and population-level changes to MDRO prevalence.

Summary of findings

Ultimately, in this scoping review of 10 observational studies, we found mixed and conflicting evidence regarding the association between NICU design and HAIs. Despite limited comparative analyses in the included studies, there were general reports of a decline in incidence of infectious outcomes when NICUs were converted from open bay to SFR, however the data interpretation relied heavily on the context and HAI outcome of interest. These contrasting findings highlight the need for rigorous research to test assumptions about the inherent infection prevention benefits of SFRs and underscore the importance of rigorous hand hygiene practices, staffing, and environmental cleaning, irrespective of unit design. What these studies clearly establish is that structural design alone does not guarantee better HAI outcomes. ^{Reference Julian, Burnham and Sellenriek15,Reference Van Der Hoeven, Bekker and Jansen18}

Limitations

Our study has several limitations. First, we excluded non-English publications which may have limited our search to healthcare settings with more resources and fewer OBUs. However, the paucity of data on this topic and our detailed search strategy suggests that we may have captured most of the available literature focused on NICUs. Secondly, our review incorporated multiple infectious outcomes including sepsis, colonization, and HAIs typically included in infection prevention surveillance. Limiting outcomes to only HAIs or CLABSIs or device-associated infections may have shown more consistent associations with NICU design. Additionally, all included studies were observational, making it difficult to account for confounders that mitigate the relationship between unit design and HAIs. Many of these studies were also before and after comparisons, in which practice or workflow changes during the unit transition may have played a role in HAI incidence. For example, physical expansions in NICU space may be associated with a corresponding increase in patient volume; therefore, outcomes attributed to the design change may reflect staffing adjustments or differences in patient population driven by increased capacity. Additionally, as the threshold for active resuscitation has shifted toward lower gestational age, it is possible that the “after” cohorts may have been systematically different and of higher acuity than the “before” cohorts, due to the evolution of neonatology practice, rather than physical changes in unit structure.

To ameliorate these limitations, future studies should consider controlling for gestational age distribution and illness severity across study periods, utilizing risk-adjusted outcome metrics, tracking unit census data as a covariate, and using concurrent controls when possible.