Summary and Interpretation
We have shown that through a comprehensive multidisciplinary QI initiative sustained improvement in CLD rates can be achieved in a level 4 NICU. Our team applied core QI principles that included establishing an aim statement and key driver diagram, testing changes with iterative PDSA cycles, and measuring primary and secondary outcomes over time. This robust approach allowed us to improve CLD rates without increasing harm or other balancing measures that might increase risk for other prematurity-related morbidities (i.e., reintubation and intraventricular hemorrhage). We have successfully changed our unit culture and work is ongoing to with a revised SMART aim goal to reduce VON CLD below the first quartile for comparable NICUs (17%).
The current observed monthly BPD rates have decreased by approximately 51%, and improvements have been sustained for >18 months. Several tests of change correlated with special cause variation, the biggest of which was adoption of the “Golden Hour” protocol. Stratifying our protocols by gestational age where applicable was a key component of our findings. This process started with a detailed interrogation of our baseline data by gestational age to see where ventilator strategies clustered together. For example, we ascertained that infants ≤ 25 weeks would require prolonged ventilation and thus have focused our efforts on gentle ventilation strategies and non-invasive ventilation modes that reduce extubation failure. Conversely, infants 26–27 weeks are more likely to be successfully managed with non-invasive modes after one dose of surfactant, thus avoiding the consequences of mechanical ventilation and reducing lung injury. There were several reasons for taking a GA-based approach: (1) stratifying DR practices in this manner (instead of birth weight) allows the team to prepare in advance because birth weight is not known before delivery (2) practice changes can be targeted to specific gestational age ranges (i.e., automatic delivery room surfactant for infants ≤ 27 weeks) without unnecessary interventions or cost to other groups (3) additional resources may be needed for infants at the extremes of viability and (4) primary (CLD) and secondary (caffeine, diuretics, postnatal steroids, extubation success) outcomes can be similarly stratified to better understand barriers to change and identify additional areas for targeted PDSA cycles. Beyond GA subgroups, efforts are underway to individualize extubation success probabilities utilizing the electronic medical record, which can aid the medical team in real-time, thus reinforcing practice changes.
Sustained improvement or “holding the gains” can be challenging in complex health care systems but are made possible through several factors. First, the use of frequent audits for process measure compliance such as monitoring DR surfactant compliance allows to determine when deviations occur. Recently, one such cause was identified which prompted a chart review of the DR course of each infant, identifying contributory systems issues and offering opportunities for reeducation. There are further opportunities to use the electronic medical record to sample ventilation data with increasing frequency and develop machine learning algorithms that can project a neonate’s respiratory course. Finally, introducing other ventilation modes such as synchronized non-invasive positive ventilation was on the horizon for our unit, until the COVID-19 pandemic stretched ventilator resources and delayed its implementation.
The limitations to our initiative include that it reflects practice changes occurring in a single NICU, some of which may not be broadly generalizable to other units. Personnel availability (i.e., DR support for more involved procedures such as LISA and the NICU footprint itself (i.e., single patient rooms, proximity to delivery room) can limit broader applicability of our local practices. Additionally, a key component to any quality improvement or practice change is educating providers and staff. We are fortunate to have dedicated nurse educators to efficiently disseminate new information and data analysts to track the broad range of outcomes necessary for the success of this initiative. In addition, the population of VON CLD eligible infants skews heavily toward the older gestational ages (>28 weeks), making sustained gains difficult since CLD rates are relatively low. Finally, there were other non-respiratory management changes such as standard management of the patent ductus arteriosus and ongoing neonatal clinical trials that make it difficult to “unbundle” practice changes from other concurrent work in the NICU.
As often occurs with improvement work, we believe our learned experience from older infants can “trickle down” to lower gestational age groups as these infants are most vulnerable to lung injury. Many infants can be managed non-invasively with best practices, and those that require mechanical ventilation require careful attention to minimize ventilator-associated lung injury. To supplement our overall improvements in CLD rates, there are still ongoing efforts in our NICU targeted toward our smallest infants. A separate “Small Baby Program” quality improvement initiative is underway which will incorporate respiratory management guidelines with best practices in other disciplines (nutrition, skin care, etc.). Reducing other negative sequalae of prematurity (e.g., NEC, late onset infection, and intraventricular hemorrhage), may further potentiate our improved respiratory outcomes, thus increasing morbidity-free survival, and allowing our neonates to achieve the best possible outcome.