Progress and Promise: The Evolving State of Quality Improvement for Preterm Infants

This month’s issue of Pediatrics includes 2 articles detailing quality improvement (QI) efforts to improve outcomes in very preterm infants. In the article by Gentle, et al, the authors report an initiative to decrease the percentage of extremely preterm infants requiring mechanical ventilation of more than 7 days in a large level III neonatal intensive care unit (NICU).1 Using a range of interventions including consensus-based management, bedside tools to support decision-making, and empowerment of all care team members, they achieved their project aim and, notably, also observed a reduction in the rate of bronchopulmonary dysplasia (BPD) or death. In the article by Morris et al, the authors report improvement in a series of in-hospital growth parameters in very low birth weight infants (VLBW) across a group of NICUs in the California Perinatal Quality Collaborative (CPQCC).2 These authors used a collaborative model for improvement centered around broad-based guidelines for care that allowed units to adapt to their local culture. NICUs participating in the collaborative improved growth restriction at discharge and sustained that improvement for an additional 2 years, as compared with nonparticipating CPQCC NICUs that did not demonstrate improvement over the same period.The 2 reports in this issue of Pediatrics are part of a robust history of NICU-based QI that began more than 30 years ago. QI activities are now a routine part of NICU operations, and progress in the field has been documented in hundreds of reports published in the medical literature. We conducted an informal search of 4 popular journals and identified 69 NICU-based QI reports published in the past 10 years. Of these, one-third focused specifically on the very preterm population. Although most QI initiatives across the entire sample occurred in a single NICU, approximately 20% described multicenter QI efforts. The most common outcomes targeted for improvement across all reports included: antibiotic use, growth and nutrition, BPD and respiratory practices, and Golden Hour care (including hypothermia and intraventricular hemorrhage [IVH] prevention). This sampling of articles demonstrates improvements across a range of outcomes, although publication bias likely results in overreporting of successful efforts. The articles by Morris et al and Gentle et al in this issue of Pediatrics, along with this review of published QI reports, not only demonstrate the progress we have made as a field but also underscore some of the current challenges and future opportunities when applying QI methodology to improving care for very preterm infants. We offer 4 such lessons that may be learned from these reports.First, while the burden of prematurity and its associated morbidities is high, the absolute number of very preterm infants cared for in any one NICU can be small, and the frequency of associated morbidities rather low. Data from the Vermont Oxford Network (VON) Database for VLBW infants born in 2024 demonstrates that 50% of participating NICUs admit fewer than 37 VLBW infants over a year. Similarly challenging, most complications of prematurity are relatively infrequent, with the VON median hospital infection rate of 8.2%, severe IVH rate of 6%, and necrotizing enterocolitis (NEC) rate of 3.4% among VLBWs.3 To have sufficient power for learning, a single center needs to be large, as is the case for the effort reported by Gentle et al, whose NICU has ∼150 to 200 annual admissions of infants born 24 to 28 6/7 weeks, or work needs to be done as part of a network or collaborative, as reported by Morris et al. The growth of state-based perinatal quality networks, which are now present in all 50 states, along with a range of new and well-established national QI collaboratives, provides opportunities for NICUs of any size to tackle persistent quality problems for very preterm infants.4Second, there has been a longstanding tension in QI regarding the balance between standardization and personalization of care. In QI, standardization is an approach to remove unwanted variation and ensure the reliable application of evidence. However, it is imperative to remember that not all variation is harmful or undesirable. Although personalization of care contributes to significant practice variation, it is often warranted when a particular intervention should be tailored for a given patient because of his or her unique characteristics.5,6 Both of the QI reports in this issue incorporated interventions that were designed to allow for both standardization and customization of care. In the QI effort reported by Morris et al, teams used multidisciplinary nutrition rounds as a general concept to address nutrition using a standardized approach, but also to allow for pre-emptive adjustments for individual patients whose growth was faltering. The initiative described by Gentle et al used a bedside “weanogram” to standardize the clinicians’ determination of weaning eligibility that was then customized for each patient’s unique clinical situation based on blood gas and transcutaneous carbon dioxide monitoring. Given the tremendous heterogeneity in risk factors, circumstances surrounding preterm birth, developmental trajectories, and family preferences, future QI efforts for very preterm infants must strive to strike the right balance between standardization and customization. Real-time data feedback powered by machine learning (ML) and artificial intelligence (AI) has the potential to enable teams to achieve both standardization and customization. Advances in neonatal AI and ML are happening quickly, although they are not yet being incorporated into QI efforts.7Third, although tremendous progress has been made in neonatology, with increasing survival of infants at the limits of viability and reductions in complications of prematurity, we need to ask what improvements QI efforts may yield in the future. In a study from VON looking at risk-adjusted outcomes from 2005 to 2014, the authors found that by 2014, more than 75% of NICUs had learned to perform as well or better than the best 25% of NICUs performed in 2005 for most major morbidities. This impressive increase in the proportion of NICUs that learned to achieve outcomes that were achieved by only the best 10% to 25% of NICUs a decade earlier was suggested to represent the learning curve for care for VLBW infants, increased adoption of evidence in practice, and more widespread application of QI methods.8 However, will the next decade bring similar breakthrough improvements? Most of the reports of QI efforts focused on very preterm infants aim for incremental improvement by setting aims that target a 10% or 20% improvement from baseline. For example, the QI initiative reported by Morris et al aimed to reduce the percentage of VLBW infants discharged with a weight less than the 10th percentile from 42% to 33.5%, a 20% reduction. Similarly, Gentle et al aimed to reduce the proportion of infants exposed to invasive ventilation for more than 7 days from 44% to 33%, a 25% reduction. Despite the remarkable achievements in these QI initiatives, one-third of infants still experienced growth failure and prolonged exposure to mechanical ventilation. Patient-led organizations such as the NEC Society, whose aim is to build a world without NEC,9 or the large multisite collaboratives like the Solutions for Patient Safety network,10 which is working to eliminate serious harm, push us to imagine even greater possibilities for improvement. Although the modifiable risk factors and potential interventions for many morbidities of prematurity, as well as for preterm birth itself, are not fully understood, we need to continue to push ourselves as a field to set ambitious goals, rigorously apply QI methods so that we are implementing what is known at high reliability and at scale, and, at the same time, invest in research that will close the current knowledge gaps that are preventing us from achieving breakthrough improvements.Finally, the progress achieved through decades of NICU-focused QI is only worthwhile if gains are maintained. Both QI reports in this issue demonstrated moderate- to long-term sustainability. The improved performance described by Gentle et al has persisted for nearly 2 years, and Morris et al showed improved growth parameters in participating NICUs over 3.5 years. At its most fundamental level, the sustainability in these initiatives is likely due to how they engaged the people in their NICU in introducing change. The Institute for Healthcare Improvement (IHI) “Psychology of Change” white paper suggests that the main reason QI efforts are unable to sustain improvements is due to failure to engage front-line people in the project.11 By unleashing intrinsic motivation, co-designing people-driven change, co-producing in authentic relationships, distributing power, and adapting in action, project teams will be much more likely to “activate agency” in those who are impacted by change. This enables people to gain the ability to choose to act with purpose. When people are involved in the change decisions, they are invested, and they are more likely to accept and sustain change. Greater attention must be paid to the psychology of change to ensure realized gains are sustained.The author teams of the 2 QI reports in this issue of Pediatrics should be commended for their high-quality QI work to improve care and outcomes for the most vulnerable NICU patients. To continue to advance the application of QI methods to improving care for very preterm infants, as a field, we must work together in collaborative QI efforts to facilitate learning and improvement for increasingly rare events; we need to embrace new innovations in machine learning and AI to support better standardized care with personalization when needed; we need to be more aspirational in setting our improvement goals; and we must focus more on activating the agency of the entire NICU team.