Weinhandl, Eric2015-11-062015-11-062015-05https://hdl.handle.net/11299/175366University of Minnesota Ph.D. dissertation. May 2015. Major: Epidemiology. Advisor: Sue Duval. 1 computer file (PDF); v, 176 pages.There are more than 450,000 patients receiving chronic dialysis in the US. Patients with end stage renal disease are statutorily eligible for Medicare, regardless of age. Due to high prevalence of poverty, many are dually enrolled in Medicare and Medicaid and are automatically enrolled in Part D (prescription drug insurance). This dissertation comprises 3 studies involving prescription drug claims in dialysis patients. In the first study, I constructed and validated risk scores based on Part D claims in incident and prevalent dialysis patients. Comorbidity indices derived from administrative data are typically based on diagnosed diseases, but the middling sensitivity of diagnosis codes limits the accuracy of these indices. In contrast, prescription drug claims are bona fide evidence of medication dispensation. I investigated aspects of newly developed scores and compared their performance in predicting risk with older scores based on diseases. In the second study, I assessed the efficacy and safety of pharmacologic inhibition of the renin-angiotensin system in dialysis patients with congestive heart failure. ACE inhibitors and ARBs are recommended for the treatment of heart failure with reduced ejection fraction. However, pivotal clinical trials all excluded patients on dialysis. I used propensity score matching to identify matched controls for treated patients that were newly dispensed an ACE inhibitor or ARB shortly after discharge from hospitalization for heart failure. I described the relative hazards of mortality and morbidity with treatment. In the third study, I assessed relative hazards of death and hospitalization associated with 4 ACE inhibitors and 2 ARBs in dialysis patients with congestive heart failure. The existence of class effects is generally presumed, but not necessarily supported by evidence regarding clinical outcomes. I showed that the most widely used agent (lisinopril) in this patient population is not necessarily associated with best outcomes.enDialysisEnd Stage Renal DiseaseHeart FailureMedicare Part DRenin-angiotensin SystemRisk ScoreThesis or Dissertation