claim expense in a given year (Rosenbloom and Gertman, 1984; Alexandre, 1988; Berk et al., 1988). For example, a report on one large University of California indemnity health plan showed that just 227 individuals accounted for 42 percent of all reimbursable hospital charges for the 1982/83 contract year (Prudential Insurance Company, 1984 cited in Luft et al., 1985). In another university health plan, 0.4 percent of the Blue Cross enrollees filing claims accounted for 21 percent of reimbursements (Luft et al., 1985). How these individuals choose among health plans can have a marked effect on a plan's costs.5 Such selection is not easy to predict using the usual demographic measures of risk.
Sixth, if health status comparisons focus only on averages and not on what happens to the most or least healthy, they may be insensitive to some forms of risk selection. For example, one recent study of Medicare beneficiary enrollment in 23 HMOs (Lichtenstein et al., 1991) found that 9 HMOs showed favorable selection and 14 showed neutral selection when the comparisons were based on mean health status for each plan. In contrast, when the extremes were compared (i.e., proportions of the most disabled and the least disabled enrolled in each plan), 10 additional HMOs showed favorable selection. None of the 23 HMOs experienced unfavorable selection on either measure, and 4 showed neutral selection on both.
Seventh, even if the extent of risk selection is considered modest as reflected in the current data, health plans' actions may still be influenced by a strong fear of adverse selection or a strong conviction that the benefits of favorable selection are significant. The consequence may be the same protective strategies that would result from documented evidence of serious selection effects.
Risk selection and risk segmentation raise both philosophical and practical questions for employers, health plans, and public policymakers. These questions involve the fundamental, interrelated issues of equity, access, cost, and quality. This section considers how risk selection may affect each issue and concludes by discussing the implications of the Americans with Disabilities Act of 1990.
Americans are still debating a basic ethical question in health care, that is, whether all people should be guaranteed some appropriate level of health