The Community Assessment Risk Screen (CARS): Identifying elderly persons at risk for hospitalization or emergency department visit

Objective: To develop and validate an instrument for identifying community dwelling elderly patients at increased risk for hospitalizations or emergen cy 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 individual s enrolled in a Medicare Risk Demonstration. Baseline demographic, health s tatus, 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 th e validation cohort were obtained from submitted claims. Results: Logistic regression identified 3 characteristics that were predict ors of hospitalizations or ED visits during the following year in the devel opment cohort: having 2 or move comorbidities, taking 5 or more prescriptio n medications, and having had a hospitalization or ED encounter in the prev ious 12 months. A scoring system (range 0 to 9) was developed for each pred ictor 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 ho spitalized (33% versus 14%), to have an ED visit (34% versus 15%), and to h ave higher per-member-per-month (PMPM) charges ($977 versus $445) during th e following 12 months. Conclusion: The Community Assessment Risk Screen (CARS) is a simple instrum ent that can be used to identify elderly patients who are at higher risk fo r health service use and increased costs.