Objective: To assess the contribution of maternal and newborn characteristics to variation in neonatal intensive care use across regions and hospitals.
Study design: This was a retrospective population-based live birth cohort of newborn infants insured by Texas Medicaid in 2010-2014 with 2 subcohorts: very low birth weight (VLBW) singletons and late preterm singletons. Crude and risk-adjusted neonatal intensive care unit (NICU) admission rates, intensive and intermediate special care days, and imaging procedures were calculated across Neonatal Intensive Care Regions (n = 21) and hospitals (n = 100). Total Medicaid payments were calculated.
Results: Overall, 11.5% of live born, 91.7% of VLBW, and 37.6% of infants born late preterm were admitted to a NICU, receiving an average of 2 days, 58 days, and 5 days of special care with payments per newborn inpatient episode of $5231, $128 075, and $10 837, respectively. There was little variation across regions and hospitals in VLBW NICU admissions but marked variation for NICU admissions in late preterm newborn infants and for special care days and imaging rates in all cohorts. The variation decreased slightly after health risk adjustment. There was moderate substitution of intermediate for intensive care days across hospitals (Pearson r VLBW -0.63 P < .001; late preterm newborn -0.53 P < .001).
Conclusions: Across all risk groups, the variation in NICU use was poorly explained by differences in newborn illness levels and is likely to indicate varying practice styles. Although the "right" rates are uncertain, it is unlikely that all of these use patterns represent effective and efficient care.
Keywords: geographic variation; healthcare costs; perinatal care; physician practice patterns; premature infant.
Copyright © 2019 Elsevier Inc. All rights reserved.