Linking Centers for Medicare & Medicaid Services data with prospective DCOR trial data: methods and data comparison results

Abstract

The Dialysis Clinical Outcomes Revisited (DCOR) trial was a large randomized, multicenter 3-year trial comparing the effects of sevelamer with calcium-based binders on mortality, hospitalization, morbidity, and medical costs in hemodialysis subjects. Dialysis Clinical Outcomes Revisited was prospectively designed to link subjects to the Centers for Medicare & Medicaid Services End-Stage Renal Disease (CMS ESRD) database to collect additional baseline characteristic data and to enhance outcome evaluation. Subjects were linked to the CMS ESRD database by means of an algorithm using several patient identifiers. Some baseline characteristic data were collected exclusively from the CMS ESRD database. Mortality and hospitalization end points were obtained from the CMS ESRD database and compared with similar data collected prospectively into a case-report form (CRF) database. Of the 2103 patients who participated in the DCOR study, 2101 were successfully linked to the CMS ESRD database. Patient baseline data showed that treatment groups were well-balanced, except that a higher proportion of subjects in the calcium-based binder group had atherosclerotic heart disease. Calculated mortality rates were similar between databases, but more deaths were identified in the CMS than in the CRF database. These additional deaths were verified through several sources. More hospitalizations were also detected in the CMS than in the CRF database. The CMS database was a good source of death end points and hospitalization occurrence. Linking patients to the data-rich CMS ESRD database allowed assessment of additional important secondary end points at a relatively low cost compared with prospective data collection.