Multiyear diagnostic information from prior hospitalization as a risk-adjuster for capitation payments

As part of a move toward a more market-oriented health-care system, major changes have been implemented in the Dutch social health insurance system. The competing sickness funds now receive risk-adjusted capitation payments, currently based on the age-sex distribution of the insurance portfolios. These very crude health indicators do not reflect expected costs accurately. The authors examine whether the incorporation of inpatient diagnostic information over a multiyear period can increase the accuracy of the capitation model. Using a panel data set (n approximately 50,000) comprising annual costs and diagnostic information for 5 successive years, the authors compare demographic and diagnostic models in their ability to predict future health care costs. The predictive accuracy of an age-sex-based capitation formula improves substantially when diagnostic information from an individual's prior hospitalizations is used as an additional risk-adjuster. The longer the period over which diagnostic information is available, the better is the predictive accuracy. The expected loss in 1992 for insured persons with the highest costs in 1988 decreases from 88% (demographic model) to 62% (1-year diagnostic model) and to 43% (3-year diagnostic model). The use of diagnostic information from prior hospitalizations is a promising option for improving the capitation formulae. The authors' results are relevant not only for situations where competing insurers are capitated, as in the Netherlands, but also when providers (United Kingdom) or health maintenance organizations (United States) are capitated.