Socioeconomic Deprivation and Hospital Length of Stay

Abstract

Socioeconomic deprivation is not easily measurable in hospital information systems. However, its identification is essential, as it is associated with morbidity and hospital length of stay (LOS). We aimed at studying the feasibility of using routinely recorded individual and area-based socioeconomic indicators, and assessing their relation with LOS. In a cross-sectional study we collected area-based socioeconomic deprivation indicators from French census databases and individual ones from the 2009 medical and administrative databases of a French referral maternity hospital. The principal outcome was the LOS for delivery. Individual level socioeconomic deprivation indicators included preferential insurance scheme (health insurance allocated to poor persons). Nine area-based socioeconomic deprivation indicators were aggregated at the census tract and commune levels. The relation between socioeconomic deprivation and LOS was studied using multilevel models. The well-documented relation between socioeconomic deprivation and preterm delivery was firstly studied in these models as a validation step. The linkage between aggregated and individual data was possible for the 3471 women included. The median LOS was 5 days. In multivariable analysis adjusted for age (P=0.02), twinning (P=0.0001), delivery mode (P<0.0001), drug addiction (P<0.0001), diagnosis-related group severity level (P<0.0001), and unemployment rate (P=0.002) were associated with an increased LOS. Identifying deprived patients in hospital databases using routinely collected area-based indicators is feasible. The relation of these latter with LOS is consistent with previous studies. Further multicenter investigations are needed to confirm the interest of using such indicators for cost and morbidity predictions.