Evidence Supporting the Value of Surgical Procedures: Can We Do Better?

There is an acknowledged need for higher-quality evidence to quantify the benefit of surgical procedures, yet not enough has been done to improve the evidence base. This lack of evidence can prevent fully informed decision-making, lead to unnecessary or even harmful treatment, and contribute to wasteful expenditures of scare health care resources. Barriers to evidence generation include not only the long-recognized technical difficulties and ethical challenges of conducting randomized surgical trials, but also legal challenges that limit incentives to conduct surgical research as well as market-based challenges that make it difficult for those funding surgical research to recoup investment costs. These legal and market dynamics differ substantially from those surrounding new drug or device development. Nevertheless, obstacles could be overcome and overall expenditures could be reduced if a share of federal health care agency budgets were reallocated to generating randomized trial data, standardizing outcome measures, and conducting observational studies analogous to those that have been facilitated for drugs via the Food and Drug Administration’s Sentinel Initiative. Until better quality evidence is available, ethical principles require adequate disclosure of the limited evidence base supporting current surgical procedures.

FDA-Catalyst—Using FDA’s Sentinel Initiative for large-scale pragmatic randomized trials: Approach and lessons learned during the planning phase of the first trial

Background: The US Food and Drug Administration’s Sentinel Initiative is well positioned to support pragmatic clinical trials. FDA-Catalyst combines direct contact with health plan members and/or providers with data in the Sentinel infrastructure. Here, we describe the rationale, feasibility analyses, and lessons learned from the planning phase of the first large pragmatic trial conducted using the Sentinel Initiative’s delivery system capabilities—IMplementation of a randomized controlled trial to imProve treatment with oral AntiCoagulanTs in patients with Atrial Fibrillation (the IMPACT-AFib trial). Methods: During the planning phase, we convened representatives from five commercial health plans, FDA, study coordinating centers, and a patient representative for protocol development, institutional review board preparation, and other activities. Administrative claims data from the plans were included in a retrospective cohort analysis to assess sample size for the trial. Members ≥30 years old with ≥365 days of medical/pharmacy coverage, ≥2 diagnosis codes for atrial fibrillation, a guideline-based indication for oral anticoagulant use for stroke prevention, and no evidence of oral anticoagulant use in the 365 days prior to the index atrial fibrillation diagnosis in 2013 were included. Exclusions for the analysis included other conditions requiring anticoagulation, history of intracranial hemorrhage, and gastrointestinal bleed. We calculated rates of oral anticoagulant use, transient ischemic attack or stroke, and bleeding in the 365 days following the index atrial fibrillation diagnosis. Results: A total of 44,786 members with atrial fibrillation with no evidence of recent oral anticoagulant use were identified. In total, 87% (n = 38,759) were classified as having a guideline-based indication for oral anticoagulants. Of those, 33% (n = 12,867) had a new oral anticoagulant dispensed during the following year, 15% (n = 5917) were hospitalized for stroke or transient ischemic attack, and 9% (n = 3469) for bleeding events. This information was used to develop the trial protocol including sample size, power calculations, and level of randomization. Conclusion: Sentinel infrastructure generated preliminary data that supported planning and implementation of a large pragmatic trial embedded in health plans. This planning identified unanticipated challenges that must be addressed in similar trials.