Modeling risk-adjusted capitation rates for Umbria, Italy

We studied the potential effect of refining per capita financing in Italy by risk adjustment using severity of illness as well as age and gender. Data were drawn from hospital, pharmaceutical, and demographic files for the entire population of the Umbrian region of Italy in 1997 and 1998. Hospitalization data from 1997 were used to classify patients into severity of illness categories which were hypothesized to be at risk for higher health services costs in 1998. Data on costs in 1998 were developed from hospital and pharmaceutical administrative data. Coefficients from 1997 models were used to develop predicted 1998 costs. Predicted costs in 1998 were compared to observed costs. Disease Staging models identified 155 unique clinical risk adjustment categories. These categories included 5.3% of the Umbrian population in 1997, who accounted for 21.6% of costs in the next year. In prediction models of future year costs using Umbrian data, R(2) values for Disease Staging models were 0.16, compared to values of 0.07 for a risk adjustment model used by Medicare. By identifying groups within the overall population who were more severely ill and who used more resources, these models can be used to assist health care planners estimate health care resources such as facilities, manpower, and programs.