OBJECTIVES: To evaluate the role of indirect measures of health needs in the use of hospital services. METHODS: An ecologic analysis was conducted in Piedmont region (4,000,000 inhabitants in North-West of Italy) using municipality aggregated data from the health information system and census population. In order to reduce heterogeneity in dimension of statistical units, the city of Turin (900,000 inhabitants) was disaggregated by neighbourhood (n=23). Linear hierarchical models were applied to 1.228 statistical units (3,528 inhabitants on average) in order to: 1) take into account the clustered nature of the data; and 2) estimate residual variability between local health units (LHU). Standardized discharge ratios were used as outcome. Direct and indirect needs measures were taken into account, specifically: standardized mortality ratios (SMR) were used as direct measure of need, while percentages of low educational level, manual work, rented dwelling, dwelling without indoor bathroom, single parent households with children, and immigrated people were used as indirect measures. All the variables were introduced in the model in a standardized form. In order to take into account the supply effect on use of health services, number of hospital beds and distance from the nearest hospital were included. Multicollinearity was evaluated using variance inflation factor estimated by standard multivariate linear regression model. Sensitivity analysis was conducted in order to evaluate the effect of ecologic bias selecting municipality on the basis of demographic dimension. RESULTS: A positive significant relationship was found for percentage of low educational level (1.38;95%CI:0.05-2.71) and percentage of rented dwellings (3.19;95%CI:1.82-4.57). Statistically significant coefficients were observed for distance (negative) and SMR (positive). Residual significant variability between LHU was observed. CONCLUSIONS: These analysis would like to contribute to a needs-based weighted population for capitation purposes in Italy. The role of geographical variability in case-mix and endogeneity in the supply should be better explored.