Recorded presentation

Ciaran S. Phibbs, PhD¹, Laurence C Baker, PhD², Susan K Schmitt, MS¹, Beate Danielsen, PhD³, Roderic H. Phibbs, MD⁴, and Aaron B. Caughey, MD⁵. (1) Health Economics Resource Center, VA Medical Center, 795 Willow Road, Menlo Park, CA 94025, 650-493-5000-1-22813, cphibbs@stanford.edu, (2) Department of Health Research & Policy, Stanford University School of Medicine, Box 5092, Stanford, CA 94305-5092, (3) Health Information Solutions, 2545 Clubhouse Drive West, Rocklin, CA 95765, (4) Department of Pediatrics, University of California, San Francisco, 3rd and Parnassus Avenues, San Francisco, CA 94143, (5) Department of Obstetrics and Gynecology, University of California, San Francisco, 3rd and Parnassus Avenues, San Francisco, CA 94143

Previous work by ourselves and others have consistently found that mortality was lower when births occurred in hospitals with higher levels of neonatal intensive care units (NICUs) and higher NICU patient volumes. All of these studies used cross sectional analysis from one or two years of data. We examined these relationships over a 10 year period. The longer time period yields more precise estimates of the level of care and patient volume effects due to greatly increased sample sizes and the ability to observe mortality as levels of care and patient volumes change over time. We used the 1991-2000 California birth/infant death cohort files, linked to the hospital discharge abstracts for mothers and infants. Logistic regression was used to control for the effects of patient volume and level of neonatal intensive care unit (NICU) care at the hospital of birth, demographics, type of insurance coverage, birth weight, and maternal and neonatal diagnoses on neonatal mortality. Regionalization of the care for low weight infants in California is limited. In 2000 only 1/3 of the very low birth weight (<1500g, VLBW) infants delivered in a hospital with a high-volume tertiary NICU. Almost all of the smaller and lower level NICUs were located in major urban areas. Over the 1990s the was a shift in the delivery of VLBW infants from hospitals with large tertiary NICU to hospitals with smaller and lower level NICUs. While there was a significant decline in neonatal mortality over this time period, the mortality differences across levels of care at the hospital of birth grew. NICU patient volume and level of NICU at the hospital of birth were important predictors of mortality. Mortality was lowest in those hospitals with high-volume, tertiary NICUs. For each 1 patient increase in average NICU census, the odds of mortality dropped by 1% (P<0.001). The volume effect was found for both overall NICU census and for the number of VLBW infants treated. This includes some high-volume NICUs that don’t provide neonatal surgery, but have all of the other characteristics of regional NICUs. Further, over the course of the 1990s, the difference in relative mortality between births in hospitals with large tertiary NICUs and those in hospitals with low-volume, lower level NICUs increased markedly. It should be a policy priority to refer as many premature deliveries as possible to hospitals with high-volume NICUs.

Learning Objectives: At the end of the session participants will understand

• The extent that care for high-risk infants has shifted from tertiary to community hospitals.
• The the difference in mortality between tertiary and community hospitals has grown over time.
• That hospital of birth is a more important determinate of mortality than highest level of NICU care provided.

Keywords: Perinatal Outcomes, High Risk Infants

Presenting author's disclosure statement:
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.

Recorded presentation