Predictors of health resource use by disabled older female Medicare beneficiaries living in the community.

Objectives: To identify specific clinical factors that could best predict resource use by disabled older women.
Design: Cross-sectional.
Setting: Urban community in Baltimore, Maryland.
Participants: One thousand two community-dwelling, moderately to severely disabled, female Medicare beneficiaries aged 65 and older, from the Women's Health and Aging Study I (WHAS).
Measurements: WHAS data were merged with participants' 1992-1994 Medicare claims data for the year after baseline evaluation, reflecting inpatient, outpatient, home-based, and skilled-nursing services. The independent contributions of factors hypothesized to predict health expenditures were assessed, using chi-square and regression analyses, with the logarithm of Medicare expenditures as the primary outcome.
Results: Demographic factors were not associated with Medicare expenditures. Factors associated with expenditures in bivariate analyses included heart disease (1.4x), chronic obstructive pulmonary disease (1.3x), diabetes mellitus (1.1x), smoking, comorbidity, and severity of disability, as well as low creatinine clearance, serum albumin, caloric expenditure, or skinfold thickness. Heart disease, diabetes mellitus, and low skinfold thickness remained significant after adjustment for other factors.
Conclusion: Heart disease, diabetes mellitus, and low skinfold thickness are important independent predictors of 1-year Medicare expenditures by disabled older women. Many other variables that reflect disease, disability, nutrition, or personal habits have less predictive ability. Most demographic factors are not predictors of expenditures in this population. Focusing on the best predictors may facilitate more-effective risk adjustment and creation of related health policies.