[Skip to Content]
[Skip to Content Landing]
February 19, 1997

Risk-Adjustment Methods Based on Health Status and Functional Status-Reply

Author Affiliations

HealthSystem Minnesota Minneapolis

JAMA. 1997;277(7):531. doi:10.1001/jama.1997.03540310029022

In Reply.  —Dr Allman suggests that self-reported survey information should be used for risk adjustment for persons older than 65 years. He suggests that self-reported functional status measures perform just as well, if not better, than claims-based systems.Although the SF-36 model did predict health care expenditures within 5% of actual expenses for a higher proportion of "high-risk" groups than did adjustment using ambulatory care groups, we found that performance of risk-adjustment systems varied depending on the measure of predictive accuracy chosen. Taking this factor into account and considering issues of administrative feasibility, we concluded that a claims-based or encounter-based system is preferable to a survey-based method.Allman correctly suggests that some issues of administrative feasibility, such as surveying illiterate and non-English-speaking subjects, can be overcome. Even so, significant feasibility problems remain for survey-based risk adjusters. For example, the suggestion that SF-36 data be collected every 5 years does not correspond to