Objective: To develop and validate an instrument for identifying community dwelling elderly patients at increased risk for hospitalizations or emergency department (ED) encounters.
Study design: Prospective cohort study.
Patients and methods: The development cohort consisted of 411 Medicare fee-for-service patients and the validation cohort consisted of 1054 individuals enrolled in a Medicare Risk Demonstration. Baseline demographic, health status, and utilization measures were obtained from telephone interviews and mailed questionnaires. Service utilization data for the development cohort were obtained from Medicare claims files. Utilization and cost data for the validation cohort were obtained from submitted claims.
Results: Logistic regression identified 3 characteristics that were predictors of hospitalizations or ED visits during the following year in the development cohort: having 2 or more comorbidities, taking 5 or more prescription medications, and having had a hospitalization or ED encounter in the previous 12 months. A scoring system (range 0 to 9) was developed for each predictor variable and patients in the validation cohort were assigned to low (0 to 3) and high (4 to 9) risk categories. When compared with the low-risk group, the high-risk group was significantly (P < .01) more likely to be hospitalized (33% versus 14%), to have an ED visit (34% versus 15%), and to have higher per-member-per-month (PMPM) charges ($977 versus $445) during the following 12 months.
Conclusion: The Community Assessment Risk Screen (CARS) is a simple instrument that can be used to identify elderly patients who are at higher risk for health service use and increased costs.