Effectiveness of Alberta Family Integrated Care on infant length of stay in level II neonatal intensive care units: a cluster randomized controlled trial

Globally, the estimated rate of preterm birth is 10.6% (regional range 8.7–13.4%) [1]. The preterm birth rate in Alberta, 9.2% in 2018–2019, is the highest among Canadian provinces [2]. Approximately 15% of preterm infants are born at < 32 weeks gestational age (GA) [1] and require care in a level III neonatal intensive care unit (NICU) [3],while 85% are born between 32 and 36 weeks GA [1] and may require care in a level II NICU [3]. The unexpectedness of preterm birth leaves parents feeling overwhelmed, anxious, depressed, isolated, and unprepared to interact with, and care for, their infant [4‐6]. Preterm birth and experiences in the NICU disrupt the early parent-infant relationship, which is critical for brain and biological development [7‐9]. In pursuit of optimal infant care, parents are often unintentionally marginalized [10‐12]. Models that integrate families into the care team show promise, but lack well-defined components, theoretically driven underpinnings, and contextualization for levels of NICU care.

Building on a foundational clinical program in Estonia [13, 14], Family Integrated Care (FICare) is a model of care developed to involve and support families as part of their infants’ care team in level III NICUs [15, 16]. In a pilot study, compared to a retrospectively matched control group (n = 62), at 21 days post-enrolment infants receiving FICare (n = 31) showed an increased rate of weight gain and breastfeeding, while mothers reported less stress [15]. In an international cluster randomized controlled trial (cRCT) of FICare in level III NICUs, compared to the standard care group (n = 891), at 21 days post-enrolment infants in the FICare group (n = 895) showed significantly increased weight gain trajectory and high frequency breastmilk feeding (> 6 times per day), and mothers reported less stress and anxiety [16].

With these positive findings, there was momentum to implement level III FICare in level II NICUs in Alberta. To address concerns about goodness of fit from provincial stakeholders, including nurses, allied health professionals, physicians, hospital administrators, parents, and researchers, we adapted the level III model to the level II NICU setting in Alberta and conducted a cRCT. The aim of the trial was to evaluate the clinical effectiveness and costs of Alberta FICare™ [17]. In this paper, we report the primary outcome, infant hospital length of stay (LOS), and the secondary outcomes of hospital readmissions and emergency department visits to 2 months corrected age (CA), feeding type at discharge, and maternal psychosocial distress and parenting self-efficacy at discharge. We hypothesized that compared to standard care, Alberta FICare™ would (1) decrease infant hospital LOS by 10%, or 1.6 days, without a concomitant increase in hospital readmissions and emergency department visits, and (2) increase rates of breastmilk feeding, reduce maternal psychosocial distress, and increase parenting self-efficacy at discharge.

Between December 15, 2015 and July 26, 2018, we enrolled 765 infants and 654 mothers; 375 infants/325 mothers at intervention sites and 390 infants/329 mothers at standard care sites (Fig. 1). An equal proportion of infants (6%) in both groups were excluded, most commonly due to becoming ineligible (e.g., transfer to a level III NICU) or their mothers withdrawing from the study. The final sample consisted of 353 infants/308 mothers in the intervention group and 365 infants/306 mothers in the standard care group. See Table 2 for infant and maternal characteristics. Compared to the standard care group, the intervention group had a significantly higher proportion of mothers who were Caucasian and born in Canada (χ² P = .01 for both variables). However, in univariate regression analyses, neither of these maternal variables were associated with infant LOS (born in Canada vs not: R² = .000, P = .74; Caucasian vs non-Caucasian: R² = .002, P = .24).

The reduction in hospital LOS with Alberta FICare™ contrasts with a meta-analysis of 19 RCTs of FCC interventions for preterm infants and parents that found no difference in LOS [29]. There may be several reasons for these differences, including considerable variability in components of the FCC interventions (e.g., educational, psychosocial, NICU environment), context of NICU care (e.g., Asia, Europe, Australia/New Zealand, and North America), and GA of infants (23–42 weeks). It may be that multicomponent interventions, such as FICare, may be required to demonstrate differences in LOS. However, the FICare level III NICU cRCT did not demonstrate a group difference in LOS [16].

Differences in results regarding LOS between our study and the FICare level III NICU cRCT may be attributed to differences in the FICare models. In contrast to the FICare level III NICU model where staff education and support were considered key components of the model, in the Alberta FICare™ trial these were considered part of the implementation process. In our study, we monitored intervention fidelity and provided booster doses of training, which is essential to ensuring interventions are delivered as intended [30]. In contrast to the FICare level III NICU model [15, 16] where parents typically received a printed orientation package and attended facilitated group education sessions, in Alberta FICare™ Parent Education used a multimodal approach including an evidence-informed Parent Education Pathway that HCP signed off as parents acquired knowledge and skills. Parent education was supplemented by the free Life’s Little Love app that parents could download on their personal devices. The FICare level III NICU model [15, 16] emphasized creating a physical environment where parents could stay at the bedside, while in Alberta FICare™ greater emphasis was placed on creating a unit culture where parents felt welcomed and included in the care team. Unlike the FICare level III cRCT [16], we screened for postnatal depression and referred mothers in both groups to existing social work services. To signal the importance of parental presence in the NICU as a precursor to parental participation in care, we provided parking passes to families in the intervention group for the duration of hospitalization. Many mothers and some fathers recorded their thoughts and feelings in the investigator-designed parent journals, which may have been therapeutic [31].

Integrating families into the care of their infant in NICU requires a shift in culture where parents are educated and supported by HCP to provide care for their infant from the time of admission. As parents become more confident and assume greater responsibility for the care of their infant(s) in the NICU, it is critical for safe delivery of care to have clearly defined roles for parents and HCP. Thus, the Relational Communication component of Alberta FICare™ may be an important component to prevent assumptions about the family’s situation and their readiness and willingness to provide care, and understand patterns of interactions that enable integration of families into the care team [19].

Rates of any breastmilk feeding at discharge were high in both groups, with 95 and 96% of infants in the intervention and standard care group, respectively, receiving breastmilk only or combination of breastmilk and formula. These rates were higher than those reported in the FICare level III NICU cRCT [16], where the rate of any breastmilk feeding at discharge was 75% in the FICare group and 81% in the standard care group (P = .004). Without data specific to infants in NICU, we speculate that failure to find group differences in our study may be attributed to high rates of breastfeeding initiation in Alberta [32].

Although effects favored Alberta FICare™, we found no significant group differences in maternal anxiety, depressive symptoms, and stress, which contrasts with the results of the FICare level III NICU cRCT [16] and a meta-review of diverse interventions for parents of preterm infants [33]. The meta-review did not include FICare interventions, and most effective interventions included a home visitation component [33]. In the FICare level III NICU cRCT, investigators measured differences in maternal outcomes within a 21-day window [16]. In the Alberta FICare™ cRCT, we measured maternal outcomes between admission and discharge. If parental ability to provide care at home is a criterion for discharge, then we speculate our lack of group difference may be related to timing of measurement rather than the effect of Alberta FICare™. Our results are similar to Ingram and colleagues who did not find any change in parenting self-efficacy in their before and after study of family-centred neonatal discharge planning with infants born at 27–33 weeks GA [34].

As hypothesized, we did not find group differences in proportions of infants requiring hospital readmissions and emergency department visits. Our result related to readmission is contrary to the results of a meta-analysis of 4 RCTs (3 from China, 1 from Iran) of FCC interventions in NICU that reported significantly lower readmission rates in the FCC group versus the standard care group [29]. The studies included in the meta-analysis reported small sample sizes (range 46–130), < 10 readmissions in both groups across studies, and wide range of infant GAs [29]. We speculate that differences in results between our cRCT and the meta-analysis [29] may be attributed to small samples and different GAs. To our knowledge, ours is the first cRCT to report emergency department visits as an outcome related to FICare interventions.

Our study has several strengths. First, the theoretical underpinnings of Alberta FICare™ will enable a better understanding of the potential mechanisms underlying its effect on outcomes. Future research is needed to identify components that make the greatest contribution to effectiveness and should be implemented in resource constrained settings. Second, our cRCT was designed and implemented with input from key stakeholders; parents were included in design and training for HCP at intervention sites. Co-design strengthens the relevance and sustainability of intervention [35]. Third, all available level II NICUs in Alberta participated and completed the study. Finally, quarterly site visits to monitor intervention fidelity increased confidence that Alberta FICare™ was implemented as designed, and the cRCT design minimized contamination across sites.

This study has several limitations. First, components of the Alberta FICare™ intervention (required time spent in NICU and parking passes) were intentionally designed to facilitate parental presence. Although parental presence may have increased parental confidence and shortened LOS, these intervention components were designed as part of the Alberta FICare™ bundle; effects of individual components cannot be disentangled from the whole. Health system policy changes that provide parking passes to all parents of infants in NICU and repeated evaluation of Alberta FICare™ may identify the most effective components of the intervention. Although there was a statistically significant group difference in parental time spent in NICU (9.00 vs. 7.79 h for Alberta FICare™ and standard care groups, respectively), a 1.21-h difference is unlikely to be clinically relevant. Second, the rate of missing maternal data differed between groups; 87% of mothers in the intervention group completed both surveys versus 71% of mothers in the standard care group. Since data on type of feeding was reported by mothers on the discharge survey, this resulted in more missing infant feeding data for infants in the standard care group compared to infants in the Alberta FICare™ group. Third, Alberta has a single health services delivery system with many standardized structures and processes. Thus, results may not be generalizable to other systems of care. Fourth, there were insufficient sites in the province for a cRCT and this study was underpowered for the planned analysis.

Alberta FICare™ in level II NICUs reduced hospital LOS in preterm infants born between 32^0/7 and 34^6/7 weeks GA, without concomitant increases in readmissions and emergency department visits. A small number of sites in a single jurisdiction and inclusion of a select group of mothers and infants limit generalizability of findings. Country-wide replication of the trial in all level II NICUs would increase generalizability.