This study assesses the efficacy of a NICU redesign from OPBY to SFRs within a flagship teaching hospital and referral center that is part of a larger health system in the Southeastern region of the United States. The NICU is a 45-bed department that provides care to premature and sick neonates 24 hours a day, seven days a week. The NICU in the study provides Level II and Level III care, including neonates 23 weeks and weighing less than one pound. Level II involves advanced care for neonates who are stable without complications but require special care and frequent feedings. Level III involves specialized care for neonates with high acuity conditions. The NICU has a high volume of patients with prematurity, respiratory distress, use of high frequency jet ventilation, hypoglycemia, and suspected sepsis.

The NICU initially had an OPBY design with multiple neonates admitted to multiple large rooms, with staff assigned based on patient acuity within the room. The OPBY NICU had an average staff of 65 full-time equivalent registered nurses (including on weekends) with 10.19 as the average number of registered nurse-hours per inpatient day. In 2020, a new Women’s and Children’s wing was constructed with a SFR design. In the first year, the SFR unit had an average staff of 71 full-time equivalent registered nurses (including on weekends) with 10.16 as the average number of registered nurse-hours per inpatient day. The first year also coincided with the COVID-19 pandemic, thus presenting a unique opportunity to examine the impact of the pandemic on unit design transitions. The results describe policy and procedure changes enacted in response to COVID-19.

Survey responses from parents/guardians of neonates discharged from the NICU in the OPBY versus SFR settings as well as interviews with NICU nurses were collected and analyzed for common themes.

The hospital contracted with a third-party vendor to administer satisfaction surveys to parents/guardians of neonates who were discharged from the NICU during a two-year period, including all data one year when the NICU had an OPBY design, and one year when the NICU had a SFR design. The data use agreement with the health system currently only allows the years of data presented in the manuscript. As noted elsewhere, in future works, plans are in place to obtain additional years of data.

The survey asked a series of questions about parent perspectives on care and clinician-parent communication using five-point Likert scales with the options (1) very poor; (2) poor; (3) fair; (4) good; and (5) very good. The survey evaluated parent perspectives including an overall assessment and feedback on nursing care, physician care, inclusiveness and responsiveness, and wait time. The survey also asked about parent feedback on communication including nurse-parent communication, physician-parent communication, and communication about tests and treatment.

Survey data were imported into Stata Version 15.0 for analysis. Two-sided Wilcoxon-Mann-Whitney tests were used to compare ordinal level dependent variables from parents whose neonate received care in the OPBY design to those who received care in the SFR. The data meet the assumptions of the Wilcoxon-Mann-Whitney non-parametric test. Dependent variables are at the ordinal level, and observations in each group are independent of each other. Levene’s tests assessed homogeneity of variance and identified equal variances across groups for all outcomes. Both the medians and interquartile ranges are presented for each group for all outcomes. The code used to generate these results is available upon request.

We report statistically significant results as p < 0.05 and marginally significant results as 0.05 < p < 0.10. Using an a priori power analysis for a Wilcoxon-Mann-Whitney test conducted in G*Power, it was determined that a total sample size of 86 across two years, or 43 parents from each NICU design, is needed to achieve power for the analyses assuming alpha level of 0.05, power of 0.80, an effect size of 0.50, allocation ratio of 1.0, and two-tailed distribution. The parental surveys were an unmatched convenience sample with a total sample size of 88, including 55 in the OPBY design and 33 in the SFR design.

Qualitative semi-structured interviews were facilitated to determine perspectives of NICU nurses. A semi-structured interview guide was developed by: (1) drafting questions; (2) receiving input from academic team members and clinical partners; (3) revising the guide to address feedback; (4) pilot testing the interview guide; and (5) refining the guide based on lessons learned from the pilot testing. The interviews presented in the manuscript were conducted among nurses in the SFR setting, and almost all also had provided care in the OPBY setting. Interview questions asked nurses to reflect on their experience providing care in the OPBY setting, providing care in the SFR setting, and then comparing and contrasting the two settings. The interview protocol is provided in Supplemental File 1.

NICU nurses were recruited by: (1) emailing NICU staff about the interviews and linking to a screener to complete if interested in participating; (2) posting flyers with tear-off tabs containing a QR code to the screener; and (3) having NICU staff champions spread the word about the interviews. The screener was an online form that asked demographic questions. The NICU nurses who completed the screener were then emailed to schedule interviews. In an interview reminder email, respondents were asked to review a digital version of the informed consent. This research was approved by the university’s and healthcare system’s institutional review boards and the healthcare system’s nursing research council. All methods were performed in accordance with the relevant guidelines and regulations.

A two-person team facilitated the interviews, including an experienced moderator and a notetaker. At the beginning of the interview, informed consent was obtained from all interviewees. Interviews followed a semi-structured interview protocol and lasted about 60 to 90 minutes. Respondents received $50 Visa gift cards for participating in the interviews. The interview sample included 11 percent of nurses who provided care on the NICU. All nurses were White, non-Hispanic women. Nurses varied in their experience in the NICU-care setting, ranging from one to more than 15 years of experience. A plurality of nurses had their bachelor’s degrees and worked between 30 to 39 hours per week. Nurses varied in the days and shifts worked in the NICU including those working weekdays, weekends, day shifts, and night shifts.

Interviews were recorded with respondent permission and auto-transcribed via Zoom. Auto-transcriptions were manually cleaned to create verbatim transcripts, which were imported into NVivo for coding. Using coding best practices, a team of coders open coded the data to identify topics and used the results of the open coding to develop a codebook with code names, code definitions, examples of quotations, and inclusion/exclusion criteria. Interviews were conducted until code and meaning saturation were reached. Focused coding, interrater reliability assessment, consensus building, and codebook revision were performed in multiple rounds. In each round, two team members engaged in focused coding to code a subset of the transcripts using the codebook.

Interrater reliability was assessed between the two coders with the goal of achieving Cohen kappa coefficients of greater than 0.80 for each code, which signals high consistency of coding. In instances in which the coders achieved coding reliability of equal to or lower than 0.80 for a particular code, coders met to review the segments with differences in coding, discuss the rationale for their coding decisions, and reach consensus on coding decisions. This sometimes resulted in subsequent revisions to both the coding and codebook. After completing this process, coding reports were run on the coded data. Content analysis was conducted of coding reports to systematically analyze and identify patterns within and across codes and sub-populations.