Background: Food and Drug Administration-mandated postmarket studies for transcatheter aortic valve replacement in low-risk populations plan to use passively collected registry data linked to claims for long-term follow-up out to 10 years. Therefore, it is critically important to understand the validity of these claims-based end points. We sought to evaluate the ability of administrative claims with International Classification of Diseases-Tenth Revision (ICD-10) codes to identify trial-adjudicated end points and reproduce treatment comparisons of aortic valve replacement in the Evolut Low Risk Trial.
Methods: We linked Evolut Low Risk trial patients to the Medicare Provider Analysis and Review database. We calculated sensitivity, specificity, positive predictive value, negative predictive value, and κ agreement statistic of claims to detect clinical end points through 2 years in trial patients. We additionally compared end points across treatment arms using trial-adjudicated outcomes versus claims-based outcomes.
Results: Trial-adjudicated deaths were perfectly identified by claims. Claims had good performance in identifying trial-adjudicated disabling stroke (sensitivity 68.8%, specificity 99.0%, positive predictive value 64.7%, negative predictive value 99.1%, κ=0.66) and pacemaker placement (sensitivity 85.2%, specificity 98.4%, positive predictive value 90.4%, negative predictive value 97.5%, κ=0.86), but more modest performance in identifying trial-adjudicated myocardial infarction (κ=0.46) and vascular complications (κ=0.45). There was no difference between treatment arms for the primary end point of death or disabling stroke using trial data (hazard ratio, 0.83 [95% CI, 0.41-1.68]) or claims data (hazard ratio, 0.89 [95% CI, 0.43-1.81]; interaction P=0.71).
Conclusions: Claims-based end points performed well in ascertaining death, disabling stroke, and pacemaker placement and were able to reproduce principal trial findings. These results support the selective use of claims-based end points for transcatheter aortic valve replacement postmarketing surveillance.
Registration: URL: https://www.clinicaltrials.gov; Unique identifier: NCT02701283.
Keywords: aortic valve; clinical trial; stroke; transcatheter aortic valve replacement.