A pre-post quasi-experimental study was conducted. Following research ethics approval by the Research Ethics Office of the Jewish General Hospital (Federal Assurance Number 0796), data were collected from mothers who provided written informed consent in the OW unit from February to December 2014. The pod/SFR was scheduled to open mid-2015, however this was delayed until January 2016. To allow time for the transition and adjustment of the staff to the new unit, post-occupancy data collection started one year after the transition beginning May 2017 and were collected to May 2018.
Participants and setting
Mothers were included if their infant was hospitalized in the NICU for at least 2 weeks and was considered stable by the attending neonatologist at the time of recruitment; they were able to read English or French and provide informed consent; and they were living less than one hour from hospital during the hospitalization. They were not included if: they would not be caring for the infant after discharge (e.g., foster placement); the infant had a major congenital anomaly, or sensory handicap; or they had given birth to multiples.
The primary outcome in this study was maternal NICU-stress measured with the PSS: NICU. Based on a previous study (15), the standard deviation for the PSS: NICU score was expected to be around 1.2 and a difference of 0.75 is considered clinically significant. A sample size of 56 mothers in the OW and 56 mothers in the pod/SFR unit would provide 80% power to detect such a difference. To allow for attrition, 70 mothers were enrolled in the OW and 80 in the pod/SFR unit.
The study was conducted at a Canadian level 3 NICU that transitioned from a 34-bed OW design to a 40-bed combination design, consisting of three pods of six beds for level 3 intensive care (1:2 nurse-patient ratio), two pods similar in size of six beds for level 2 care (1:3 ratio), and ten SFRs for level 1 care (1:4 ratio) (Figure 1: Pod/SFR unit design). In this new unit, all infants are admitted to a pod and, once their condition permits, moved to a SFR when they require step-down care. Therefore, in contrast to most previous NICU design studies our pod/SFR participants experienced two types of design: their infant was admitted to a 6-bed pod and moved to a SFR for step-down care prior to discharge.
The former OW was a 34-bed level-3 unit of 400-m2 in one space with around 550 admissions per year, including about 115 infants born very low birthweight (less than 1500 grams). Six neonatologists and 94 nurses were on staff. There was one room designated for mothers to express breast milk, and a parent room with one sofa bed located down a hallway outside the unit for over-night stays. Parents could place a chair at their infant’s bedside. The OW had florescent lighting and windows on only one of the four walls.
The current 1125-m2 pod/SFR unit is situated in a newly constructed wing, and has new ventilators, monitors, and robotic arms. There are 7 neonatologists and 91 nurses on staff who work in both the pods and SFR areas. There are two rooms dedicated for parents to sleep overnight on a full-size bed (these must be reserved), and each SFR includes a parent lounger chair that does not lay fully flat but can also be used for overnight stays. In the new unit there is one shared parent shower room, and a dining area with kitchen facilities just outside the entrance is designated for their use. In the new unit, the lighting is indirect florescent; and in all pods, there are large windows with blinds for light control.
Both pre- and post-move parents had access to the unit 24-hours per day and were invited to attend weekly parent meetings on topics of interest. Although several staff nurses were trained lactation consultants, there was no dedicated position for lactation support in the former OW nor the current pod/SFR.
All the major study dependent variables were assessed with reliable and valid measures, most previously used in studies of mothers of NICU infants. The revised Parental Stress Scale: Neonatal Intensive Care Unit (PSS: NICU) (22, 23) assessed the primary dependent variable: maternal stress related to the experience of their infants’ hospitalization. This self-report measure is the gold standard for measuring parent stress in this context and taps stress related to NICU sights and sounds, feeling restricted in their role as caregiver, and the infant’s behavior and appearance. Higher scores reflect greater stress. Internal consistency is excellent, and there is evidence of construct validity (22, 24).
The Edinburgh Postnatal Depression Scale (EPDS) (25) is widely used to assess perinatal depressive symptomatology with demonstrated high internal consistency and validity for detecting major depression in the perinatal period. We used a cut-off score of 12 or more to indicate possible clinical depression as this score has optimal sensitivity for a diagnosis of major depression (26). Mothers' perceptions of staffs' ability to provide the infant and family with family-centered care during hospitalization was measured with the Family Centered Care Questionnaire - Parents (FCCQ) (27). The FCCQ examines respect, collaboration and support; and both validity and reliability are established.
The Nurse Parent Support Tool (NPST) is designed to measure parents’ perceptions of nursing support during their child’s hospitalization (28), and higher scores indicate higher support. Internal consistency is high and the NPST discriminates among parents with different levels of support (28). The General Sleep Disturbance Scale (GSDS) (29) assesses subjective sleep disturbance. The GSDS has well-established validity and reliability in parents of newborns, and a score above 42 indicates sleep disturbances.
The Breastfeeding Self-Efficacy Scale - Short Form (BSES-SF) measures mothers’ breastfeeding confidence (30). Higher scores indicate greater self-efficacy. Reliability estimates indicate excellent internal consistency, and it has been validated with NICU mothers (31). The Readiness for Hospital Discharge Scale (RHDS) assessed mothers’ perception of their preparedness for leaving the NICU. Items measure the parents’ preparedness (parent subscale), as well as their perception of how ready their infant is to go home (child subscale) (32). Predictive validity has been established (32).
At enrolment, data on maternal age, education, marital and employment status, parity, country of birth and other demographic characteristics were also collected with a background questionnaire. At enrolment and again in the week prior to their infant’s discharge, participants reported the number of hours and timing of their presence in the NICU each day for the previous 7-day period.
Data collection procedures
Following research ethics approval, research staff approached mothers who met inclusion criteria and agreed to learn about the study. Those who subsequently provided written informed consent were issued an on-line secure website address at enrolment to access and complete questionnaires about stress, depressive symptoms, support, family-centered care, and sleep disturbances. They also responded to the socio-demographic questionnaire and described their presence. At enrolment, mothers took twenty to thirty minutes to complete the questionnaires. Paper copies of all questionnaires were available for those who preferred. At discharge, data about their infant’s medical condition and birth were extracted from the hospital medical record by research staff.
In the week prior to their infant’s anticipated discharge from hospital, mothers completed the breastfeeding self-efficacy and readiness for discharge questionnaires and again responded to the question about their presence in the NICU. For both groups retention was very good: 80% of mothers completed data collection at discharge in the OW group and 77% in the pod/SFR group.
Range and logic checks were conducted for all study variables. To describe the sample demographic and clinical characteristics, percentages were computed for categorical variables and means and standard deviations for continuous variables. To compare group characteristics, t-test and Wilcoxon rank sum test were used for continuous variables. Chi square and Fisher’s exact test were used for categorical variables.
T-test and Wilcoxon rank sum test were conducted to compare the groups’ unadjusted means for the maternal dependent variables, and analysis of covariance (ANCOVA) was used to adjust for covariates. All tests performed were two-sided and a p value of less than or equal to 0.05 was considered significant.