Single-family room neonatal intensive care unit design: do patient outcomes actually change?

Elva Mankin

Medical records data were obtained from Parkland and included time periods before and after moving into a new location. NICU encounters with admission dates between January 1, 2013 and December 31, 2017 were included. The move occurred on August 20, 2015. The initial analysis included all data, but sensitivity analysis […]

Medical records data were obtained from Parkland and included time periods before and after moving into a new location. NICU encounters with admission dates between January 1, 2013 and December 31, 2017 were included. The move occurred on August 20, 2015. The initial analysis included all data, but sensitivity analysis was performed excluding admissions 90 days immediately before the move (May 22, 2015−August 19, 2015), the move date, and 90 days immediately after the move (August 21, 2015−November 18, 2015). Outcomes of interest include length of hospital stay, growth during hospital stay, time to first oral feeding, and incidence of sepsis. This study was submitted to Western Institutional Review Board and was determined to be exempt from review.

Study sites

The NICU in the old Parkland hospital consisted of multiple units, divided into pods of 4−6 beds each in an open bay configuration, totaling about 90 beds. Each unit had a centrally located multiseat caregiver workstation. In the old hospital, bed assignments were made based on acuity and patient loads, and patients were frequently moved to accommodate these needs.

In the new hospital, the NICU operates as a single unit divided into eight pods of 12 SFR each. Decentralized single-seat workstations are located between each pair of patient rooms. While there is a multiseat work area located outside each pair of pods, nurses typically stay within their assigned pod and do not use this space as a charting station. Patient room assignments are based on physician preference, with physician teams’ patients located within the same pod. Patients are not moved once assigned to a room, and nurse patient loads may be lighter than in the previous facility due to the mix of acuity within a pod. Visual examples of the NICU layouts in both the old and new hospitals have been previously published by Brittin et al. [5]

There has been a progression of baby-friendly practices over time at Parkland. While the philosophy of the ten baby-friendly standards [6] was consistent between the pre and post time periods, facility constraints such as a lack of space and privacy for breastfeeding limited adoption in the old NICU. Despite the long-term implementation of these standards, Parkland only recently received its Baby Friendly designation in July 2019 due to previously low breastfeeding rates. Otherwise, provider and interprofessional support in the NICU has not significantly changed between the study time periods. Parkland has consistently offered a wide range of services to NICU patients and their families such as lactation, social work, nutrition, psychiatric, physical medicine, child life, and on-site pharmacy.

Statistical analysis

Comparisons between the pre and post period in demographic data were assessed using univariate tests including the Wilcoxon rank-sum test for continuous variables and chi-squared analysis for categorical variables.

Length of stay (LOS)

NICU LOS was analyzed using a proportional hazards regression model, which is appropriate for skewed and non-normally distributed data. Continuous variables were assessed for functional form and if continuous form was not appropriate, the variable was categorized. If LOS was negative or equal to zero, it was counted as a data entry error and set to missing. Deaths and transfers to other hospitals were included in the model as competing risks. In addition to time period (pre, post), confounding variables included gestational age (extremely preterm (<28 wks), very preterm (28−<32 wks), moderately preterm (32−<37 wks), and term or post-term (37 wks+)), diagnosis relating to maternal substance use, FY2015 Medicare Severity Diagnosis Related Group (DRG) weights [7] (as a measure of acuity), SES index quartile (calculated using patient zip code and area-level data), race/ethnicity, and gender. Interactions with gestational age were also examined. The SES index is based on patient zip code and a combination of zip code level US Census data including percent unemployment, percent below US poverty line, median income, property values, education level, and household crowding [8]. Compared to quartiles of national block group data, 55.3% of our SES index scores were in the lowest quartile, 23.1% in the second quartile, 14.2% in the third quartile, and 7.4% in the fourth quartile. Given the relatively low SES status within our sample, we created our own quartiles based on the population of encounters at Parkland Hospital.

An interrupted time series model was explored using median LOS aggregated using 1-month periods. The initial model assumed a linear change over time and explored possible changes in slope at the time of the move. Model fit was assessed. Separate models were used for different interaction groups as deemed appropriate by the regression model. Gestational age was controlled for in the analysis. The possibility of seasonal effects was also examined using periodic adjustments.

Growth assessment

Growth during the hospital stay for NICU patients was assessed using daily weight measurements. Analysis was limited to preterm infants who were discharged home since growth patterns can be quite different for term or post-term infants in the NICU as well as for those transferred for additional care or who died. Those with two or fewer daily weight values were excluded as were children who were born elsewhere and transferred into the Parkland NICU. Analysis was conducted using cubic spline regression models with a random effect for each infant. Knot placement was based on previous work in preterm infant growth as well as visual inspection of the growth patterns and were placed at 14, 28, 42, and 70 days [9]. Gestational age was included in the model as a confounding variable.

Oral feeding

Time to first full or partial oral feeding was assessed in those in the NICU who were born preterm and discharged to home. Analysis methods (including confounders) were similar to the LOS analysis using a proportional hazards regression as well as a time series model using the median monthly time to first oral feed.


Infections in the NICU were captured using diagnoses of “Sepsis” or “Septicemia” or “Bacteremia”. Since specific time and date was not available for the diagnoses, logistic regression was used with the outcome of at least one diagnosis of infection during the NICU stay. The main comparison of interest was time period (pre, post). Potential confounding variables were gestational age categories, diagnosis relating to maternal substance use, SES index, race/ethnicity, gender, and LOS. DRG weights were not included as a confounding factor since a diagnosis of sepsis could influence the DRG weight. Interactions with time and gestational age were also examined. A time series analysis was not performed since the date of the sepsis event was not included in the data.

All regression models included sensitivity analysis to assess the impact of missing data on the model using multiple imputation through the fully conditional specification method [10]. Analysis was performed using SAS software, Version 9.4 of the SAS System for Windows (SAS Institute Inc., Cary, NC, USA) and R software, Version 3.5.1 [11]. Requests for code should be directed to the corresponding author.

Source Article

Next Post

Modern Architectural Style Considerations | Milgard Windows & Doors

Clean lines, smooth surfaces and lots of light encompass the look of a modern home. Ornamentation is kept at a minimum to focus on a large perceived space and welcome indoor-outdoor living. These one-of-a-kind homes are usually asymmetrical with flat or uniquely slanted roofs. Recommended Window and Patio Door Styles […]
Modern Architectural Style Considerations | Milgard Windows & Doors