Abstract 11631: Is It Possible to Predict Which Patients With Atherosclerotic Cardiovascular Disease Will Cost the Most During Long-Term Follow-Up? Results From the IMPRES Registry.

Background: Atherosclerotic disease (ASCVD) is the leading cause of mortality and healthcare expenditures. Annual indirect and direct costs of ASCVD are estimated to be $330 billion. Although many patients are diagnosed with ASCVD, it is usually only a small percentage of them that are responsible for most of the costs. If it were possible to predict which patients are expected to cost the most, targeted approaches to best manage them could be developed. Methods: A total 62,070 Intermountain Healthcare patients ≥18 years of age with a first encounter for ASCVD (CAD, CVD, and PAD) between January 1, 1999 and December 31, 2013 who survived the index ASCVD event were studied. Patients were stratified into high-cost (>$4,500) and lower-cost (≤$4,500) categories by their average annual costs (calculated by summing the average outpatient, emergency, and inpatient costs per month for each year divided by the number of years contributed). Multivariable analysis was used to determined baseline independent predictors of highest cost. Results: Overall, 2,805 (4.5%) patients fell into the high-cost group (mean = $14,541 [median = $8,018] vs. mean = $526 [median = $223] in the lower-cost group). Total cost over long-term follow-up was $150,932 vs. $34,788. The strongest multivariable predictors of high-cost included age, a PAD diagnosis, diabetes, renal failure, heart failure, COPD, and malignancy (Table). A high-cost risk model was developed and shown to be highly predictive of a high-cost patient (AUC=0.738 [0.729, 0.748], p<0.0001). Conclusion: A small (<5%) population of patients diagnosed with ASCVD have more than 27 times as much average annual per-month cost, and more than 4 times as much over long-term follow-up, than the remaining 95%. Many of the expected high-risk baseline clinical characteristics account for this large disparity in cost. However, by quantitatively integrating these into a multivariable risk model, high-cost patients can best be pre-identified, so they may be selected to receive special attention. [ABSTRACT FROM AUTHOR]