Pairing Observational and Patient-level Clinical Trial Data to Assess Cardiovascular Risk Reduction with Catheter Ablation for Atrial Fibrillation

Project: Research project

Project Details


1  PROJECT SUMMARY 2  Atrial fibrillation (AF) poses a considerable disease burden but has few therapies proven to improve survival. 3  Catheter ablation, currently indicated only for symptom relief, holds great promise for reducing morbidity and 4  mortality. The ongoing CABANA trial will soon provide an initial answer regarding whether or not to expand the 5  indication for ablation from symptom control, currently used in only 5% of AF patients, to cardiovascular risk 6  reduction in broader AF populations. There will be an urgent need to assess the generalizability of the trial, 7  because inappropriately expanding the use of ablation may expose a large population to unnecessary harm, 8  whereas forgoing a potential beneficial treatment may miss the opportunity to help many patients. 9  Observational datasets with high external validity can be used to assess the generalizability of randomized 10  controlled trials (RCTs). Conversely, rigorous RCTs with high internal validity can be used to assess the 11  internal validity of observational data. The overall objective of this project is to demonstrate a novel framework 12  that pairs an RCT with observational data to inform clinical best practice. Taking advantage of the 13  complementary strengths of RCTs and observational data, this proposed project will describe the 14  representativeness of CABANA patients to real-world patients using a large observational database, 15  OptumLabs Data Warehouse (OLDW), while simultaneously assessing the degree of confounding inherent to 16  observational studies using OLDW to compare AF treatments. Aim1 will (1a) estimate the prevalence and 17  describe the characteristics of real-world patients who would be excluded from CABANA; and (1b) identify 18  patient populations under-represented in CABANA. Aim2 will (2a) assess the agreement of treatment effects 19  between CABANA and trial-like OLDW patients; and (2b) investigate if obtaining additional clinical data from 20  electronic health records (EHR) to supplement claims data would reduce residual confounding. The OLDW 21  contains rich data on a large and heterogeneous population, e.g., insurance claims for over 130 million 22  enrollees and EHR for over 40 million patients, and thus is optimally suited for this project. The proposed work 23  is innovative, because it seeks to shift the evaluation of treatment towards integrating RCTs and observational 24  studies in real time, rather than waiting years after the RCT publication to provide observational evidence. 25  Furthermore, this study will take a major step towards the understanding of the extent of confounding in 26  observational data due to the unique access to patient-level clinical trial data and the ability to link claims and 27  EHR to assess for residual confounding. The findings will inform practice by identifying patients to whom 28  extrapolation from CABANA should be cautioned, and will motivate future research by determining which 29  populations require further investigation. The results will also guide future observational research by providing 30  evidence on the internal validity of observational studies and how to best minimize confounding.
StatusNot started