Predictors of mortality in outpatient Geriatric Evaluation and Management clinic patients

Background. This study describes characteristics and predictors of survival in an outpatient Geriatric Evaluation and Management (GEM) population. Methods. Prospective evaluation and longitudinal follow-up of consecutive patients (N = 636) seen in a GEM Clinic between January 1986 and September 1991. Results. The typical patient was 78.4 years of age, White, female (73%), unmarried (66%), and living with a spouse or relative (47%). Although two-thirds were demented, most were independent in Activities of Daily Living (ADL) and partially dependent in Instrumental Activities of Daily Living (IADL). Subjects were followed for an average of 25 months. In bivariate analysis, IADL was the strongest predictor of survival (O.R. = 4.4). Higher ADL, better cognitive status, lower comorbid illness, and lack of recent hospitalization were also predictive of survival. In stepwise logistic regression, only IADL (O.R. = 4.2) and comorbid illness (O.R. = 1.5) predicted survival. In Kaplan-Meier Lifetable Analysis, survival at two years was 91% in the least dependent IADL group while survival was 75% in the most dependent group. Comorbid illness was the only factor that improved prediction of survival above that seen with IADL alone. When subjects are stratified by both function and illness, mortality was 36% in the ill and disabled group and 8% in those of high function and limited illness. Conclusions. IADL and comorbid illness scores offer a means of stratifying subjects for risk of death and may be useful in evaluating and comparing mortality experience in outpatient GEM and control populations. Stratification may increase the likelihood that studies aimed at improving survival will detect a difference resulting from the intervention.