Causal Inference
Long-term cardiovascular outcomes following bariatric surgery: Reconciling seemingly conflicting evidence. Sebastien Haneuse* Sebastien Haneuse Luke Benz Valerie A. Smith Matthew L. Macejewski David Arterburn
To-date there have been no randomized trials of the impact of bariatric surgery on incident cardiovascular disease (CVD). Nevertheless, a large body of observational evidence exists in support of bariatric surgery being associated with reduction in risk of CVD, as well as a wide range of other adverse outcomes. Two recent studies, however, with one based on claims data from Optum and another based on electronic health record (EHR) data from the Veterans Administration (VA), found no evidence of benefit regarding CVD, arguing that much of that prior work suffers from various sources bias, including confounding, information bias and surveillance bias, and from design decisions that compromise whether one can conceive of a corresponding target trial. In turn, renewed calls for randomized trials have been put forward. In this work, we first present new results based on a target trial emulation that mimics the emulation employed in the VA study but with EHR data from Kaiser Permanente. The results provide evidence of a protective effect of bariatric surgery in relation to risk of CVD and, therefore, are consistent with the majority of the literature. We then examine possible mechanisms by which the discrepant results can be reconciled, including issues of statistical validity that arise from small samples and whether recent work on transportability indicates that we should not expect results to always be concordant. We conclude with a discussion of the role of conservatism associated with “clinical trial thinking”, what standards we should be using as we consider the work of others in the literature and the role that evidence triangulation may play in the future.