50: Hospital-level variation in term nicu admission rates

Abstract

Neonatal intensive care unit (NICU) admissions are known to vary by hospital, and NICU admission rates among term, non-anomalous infants have been proposed as a metric of obstetric quality. The objective of this study was to determine what percentage of the variation was attributed to systemic differences between hospitals and their practices after accounting for a hospital’s patient case mix. This is a retrospective cohort study using data collected as part of the Consortium for Safe Labor study. Women who delivered term (≥37 weeks), vertex-presenting, singleton, non-anomalous, liveborn infants in hospitals with a NICU were included. The primary outcome was NICU admission. Multilevel mixed effect models were used and included a hospital-specific random effect. The total variance and the proportion of the variance attributed to the hospital was calculated before and after adjustment for maternal demographics (age, race, insurance status) and pregnancy characteristics (parity, gestational age, diabetes, hypertension, tobacco use, drug use, HIV, abruption, and birth weight). Adjusted odds ratios were obtained, controlling for demographics, pregnancy characteristics, and the random-effect of the hospital. 146,714 infants from 9 hospitals were included. Figure 1 shows the variation in NICU admission rates (range 3.0 – 9.8%). Prior to adjustment for patient case mix, 11.4% of the variation was attributed to the hospital. With the addition of maternal demographics, the hospital-level variation was slightly decreased at 10.7%. However, the majority of the variation was explained by the pregnancy characteristics of patients at each hospital. When these characteristics were added to the model, the residual hospital variation was only 2.0%. The adjusted odds ratios for the pregnancy characteristics associated with NICU admission are shown in Table 1. The term, non-anomalous NICU admission rate varied three-fold across the nine, geographically diverse hospitals that participated in the Consortium for Safe Labor. However, the residual variation attributed to hospital-specific practices was minimal (2.0%) after accounting for each hospital’s case mix. The lack of variation attributed to hospital-specific practices argues against this rate being used as a quality metric in obstetrics. Further research is needed to study NICU admission variation in more diverse settings to assess the generalizability of these findings.View Large Image Figure ViewerDownload Hi-res image Download (PPT)