OpenLB
Open Library of Bioscience

BACKGROUND: The probability of repeat admissions (Pra) survey is a case-finding tool designed to identify elderly persons at risk for hospitalization. Little is known about the ability of the Pra to assess utilization patterns in a managed care setting.
OBJECTIVES: To assess the ability of the Pra survey to identify elders at risk for increased healthcare costs and to determine the utility of this tool in predicting mean per member per year (PMPY) claims among primary care sites in a managed care setting.
STUDY DESIGN: Mean PMPY paid claims, inpatient admissions, and inpatient days were compared for survey respondents with high, medium, and low Pra scores. Linear regression was used to examine the correlation (R2) between the mean Pra score for respondents and the percentage of high Pra score respondents at each primary care site in the HMO and the sites' mean PMPY paid claims.
METHODS: From a single Medicare-risk health maintenance organization (HMO) with approximately 25,000 members and 234 primary care sites in northeastern and central Pennsylvania, we contacted 24,947 enrollees about the Pra survey. A total of 17,484 (70.1%) patients in 159 of 234 (67.9%) primary care sites responded to the survey. Of these, 17,469 (99%) surveys could be scored.
RESULTS: The mean Pra score in this population was 0.27 (standard deviation = 0.108). Pra scores were stratified as follows: 4.5%, 24.1%, and 71.4% of respondents scored high (20.5), moderate (> or = 0.3 and <0.5), and low (<0.3), respectively. Patients with high Pra scores had a mean of $12,611 in PMPY claims, versus $6944 and $3038 for moderate and low scores respectively (P < .0001). The R2 between the mean Pra score for respondents at each primary care site and the sites' mean PMPY Medicare-risk claims was 0.042. If survey respondents assigned to 6 specialty care sites (hematology, oncology, rheumatology, endocrinology, nephrology, and pulmonary medicine) are included, the R2 increased to 0.176. The R2 between the percentage of high-risk individuals with a Pra score of 0.5 or higher at each primary care site and the sites' mean PMPY Medicare-risk claims was 0.0005.
CONCLUSIONS: In this Medicare-risk HMO population, the Pra survey successfully performed 2 functions: (1) the prospective identification of enrollees at risk for increased healthcare utilization and (2) identification of patient factors that accounted for as much as 17.6% of the variation in utilization between primary sites.