Variation in the use of alternative levels of hospital care for newborns in a managed care organization.
OBJECTIVE(S): To assess the extent to which variation in the use of neonatal intensive care resources in a managed care organization is a consequence of variation in neonatal health risks and/or variation in the organization and delivery of medical care to newborns. STUDY DESIGN: Data were collected on a cohort of all births from four sites in Kaiser Permanente by retrospective medical chart abstraction of the birth admission. Likelihood of admission into a neonatal intensive care unit (NICU) is estimated by logistic regression. Durations of NICU stays and of hospital stay following birth are estimated by Cox proportional hazards regression. RESULTS: The likelihood of admission into NICU and the duration of both NICU care and hospital stay are proportional to the degree of illness and complexity of diagnosis. Adjusting for variation in health risks across sites, however, does not fully account for observed variation in NICU admission rates or for length of hospital stay. One site has a distinct pattern of high rates of NICU admissions; another site has a distinct pattern of low rates of NICU admission but long durations of hospital stay for full-term newborns following NICU admission as well as for all newborns managed in normal care nurseries. CONCLUSIONS: Substantial variations exist among sites in the risk-adjusted likelihood of NICU admission and in durations of NICU stay and hospital stay. Hospital and NICU affiliation (Kaiser Permanente versus contract) or affiliation of the neonatologists (Kaiser Permanente versus contract) could not explain the variation in use of alternative levels of hospital care. The best explanation for these variations in neonatal resource use appears to be the extent to which neonatology and pediatric practices differ in their policies with respect to the management of newborns of minimal to moderate illness.