A comparison of two methods for identifying frail medicare-aged persons

This article compares the efficacy of two screening methods to identify frail Medicare-aged persons using self-report questionnaire data: a clinical judgment method developed by nurse and social worker professionals in a community-based long-term care department and an empirical research method previously developed by the Center for Health Research using computer formulas and stepwise logistic regression coefficients. A sub-aim was to see whether the empirical method proved robust over time by measuring aggregate utilization and mortality in frail and nonfrail cohorts, which would increase the interest of physicians, managed care organizations, and other agencies providing service to Medicare beneficiaries. The clinical judgment method classified 88% of the population correctly. The empirically derived method surpassed the clinical judgment method by correctly classifying an additional 4% of the population (92%), primarily by reducing the false-positive rate by 24%. Persons identified empirically as frail in 1990 had a mortality rate over a decade of 82% compared with 34% of the nonfrail. The inpatient use rate was 150%) greater in the frail cohort, and the total days spent in any type of care setting during the stud), period (1990-1998) were 388 per frail person compared with 84 per nonfrail person. The clinical staff adopted the empirical method, which employs centralized survey services and customized software and has been successfully disseminated to managed care organizations in 14 states at a cost of $7 per survey participant. This research-based screening process is now used prospectively to increase staff efficiency, improve resource targeting, an support aging in place.