Hospital level of care and neonatal mortality in low- and high-risk deliveries: reassessing the question in Sweden by multilevel analysis.

BACKGROUND In high-risk births, the availability and concentration of neonatal resources in larger regional hospitals increases the chance of survival. The advantages of regionalization for low-risk deliveries are still unclear, but some studies have suggested that regionalization also is beneficial for low risk deliveries. The aim of the present study was to investigate both the relevance of regionalization and the concentration of neonatal resources as determinants of mortality in low- and high-risk deliveries in Sweden. METHODS Interhospital differences in 28-day neonatal mortality were analyzed distinguishing maternal and delivery factors from institutional ones. Using information from the Swedish Birth Register (1990-1995), we performed risk-stratified multilevel logistic regression analysis to study 691,742 births (first level) nested within the 66 Swedish hospitals with maternity wards (second level). RESULTS In low-risk deliveries, mortality decreased with improved access to neonatal resources. Mortality was lowest in larger regional hospitals with full access to neonatal care. This association remained unchanged after adjusting for patient mix. With regard to high-risk deliveries, mortality was higher in large county and regional hospitals than in small hospitals without access to neonatal care but, as expected, this increased risk disappeared after adjustment for patient mix. CONCLUSIONS Increased regionalization and concentration of neonatal resources for low-risk births is justified from a strictly medical point of view. From a public health perspective, closing small obstetrics units may prevent an appreciable number of deaths, but it would have only a very small impact on the risk of mortality from the individual's point of view. The cost-effectiveness of such a step remains to be analyzed from a health economics perspective.