ABSTRACT
Objective: To develop predictive models of high-cost acute coronary syndrome (ACS) patients using demographic, disease, and treatment characteristics.
Study design: This was a retrospective, administrative claims analysis utilizing pharmacy, medical, and eligibility data from a large US managed care organization.
Methods: ACS was defined by ICD-9 codes for unstable angina (UA) or acute myocardial infarction (AMI). New onset patients (without ACS claims) in the prior six months were identified for the time period 07/01/99–06/30/01, and followed up to 12 months, health plan disenrollment, or death. Cost was measured as that incurred during the initial episode plus subsequent follow-up or during the subsequent follow-up only. Patients were dichotomized as high-cost (top 20%) or low-cost (bottom 80%), based on total costs. Logistic regression was used to examine the association for being classified as high-cost.
Results: A total of 13 731 patients were included: 51.7% with UA, 39.6% with AMI and 8.7% with both UA and AMI. The mean age was 54.2 years and 68.2% were male. A number of co-morbidities (hypertension, diabetes, heart failure, etc.) predicted high-cost patients. Among medications, prior ACE inhibitor use predicted high-cost patients. While revascularization procedures, in general, were strong predictors of high-cost, revascularization during the index ACS episode (opposed to revascularization during the follow-up) decreased the odds of being high-cost (odds ratio [95% CI] 0.615 [0.506–0.748]).
Conclusion: High-cost patients with new onset ACS can be predicted by some characteristics, but many of these characteristics are non-modifiable co-morbidities. Payers and providers may find opportunities for clinical and cost-saving interventions for these patients.
Notes
* Presented as an abstract at the 17th Annual Academy of Managed Care Pharmacy meeting, April 22, 2005