Incomplete reporting of clinically significant acute rejection episodes in the national kidney transplant registry.

Administrative claims data could provide a unique opportunity to identify acute rejection (AR) events using specific antirejection medications and to validate rejected data reported to the Organ Procurement and Transplantation Network. This retrospective cohort study examined differences in registry-reported events and those identified using claims data among adult kidney transplant recipients from 2012 to 2017 using Standard Analysis Files from the US Renal Data System. Rejection rates, survival estimates, and center-level differences were assessed using each approach. Among 45 880 first-time kidney transplant recipients, we identified 3841 AR events within 12 months of transplant reported by centers in the registry; claims data yielded 2945 events. Of all events occurring within 12 months of transplant, 48.5% were reported using registry only, 32.9% were identified using claims only, and 18.6% were identified using both approaches. A 3-year death-censored graft survival probability was 90.0%, 88.4%, and 81.2% (P < .001) for ARs identified using registry only, claims data only, and both approaches, respectively. The large discordance between registry-reported and claims-based events suggests incomplete and potentially inaccurate reporting of events in the Organ Procurement Transplant Network registry. These findings have important implications for analyses that use AR data and underscore the need for improved capture of clinically meaningful events.