Statistical Challenges and Solutions for Real-World Data and Comparative Effectiveness Research

• Activity Number: 218
• Type: Topic Contributed
• Date/Time: Monday, August 4, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Social Statistics Section
• Author(s): Wei Shen*+ and Douglas Faries
• Companies: Eli Lilly and Company and Eli Lilly and Company
• Keywords: Comparative Effectiveness ; Real-world Evidence ; Network Meta-analysis ; Unmeasured Confounding ; Bayesian Methods ; Observational Research

Abstract

Timely and credible comparative effectiveness research is critical for guiding the appropriate use of medical resources including assignment of the best intervention for each patient. However, information on competing treatments from head to head randomized controlled trials is often not available. In this work, we present case studies of the use of network meta-analyses (NMA) and sensitivity analysis for unmeasured confounding to address some of the statistical challenges and provide comparative effectiveness information to healthcare payers, providers, and patients. We first present a case study of the use of Bayesian NMA to provide comparative effectiveness information between multiple interventions for osteoarthritis. The flexibility of the Bayesian NMA framework is demonstrated, including adjustment for study level variations and utilizing the posterior probability distribution to provide ranking of competing interventions. A second case study describes the use of recently proposed approaches for assessing the impact of unmeasured confounding in healthcare claims database research comparing the cost of interventions for Type II diabetes.