Many animal studies have investigated the impact of nonselective NSAIDs and selective COX-2 inhibitors on fracture healing. Nearly all those experiments focused on chronic drug administration following simulated long-bone fractures. One concern regarding the clinical relevance of those animal studies is that the “fractures” are often created by open means, which results in cortical devascularization and which may not accurately simulate the most common long-bone fracture pathophysiology in humans. Nevertheless, many orthopaedic surgeons have used the results of those animal studies to limit—or even stop—their use of NSAIDs to treat postfracture pain.
In the July 15, 2020 issue of The Journal, George et al. use a large private-insurance database to investigate the association between postfracture prescriptions filled for NSAIDS (both selective COX-2 inhibitors and nonselective types) and the subsequent diagnosis of a nonunion at 1 year postinjury. Administrative database research is more useful for generating hypotheses than for proving or disproving them, and these authors (along with Commentary writer Willem-Jan Metsemakers, MD, PhD) rightly point out that adequately powered randomized trials are needed to more fully address this issue.
Still, I was a bit surprised by the finding that nonselective NSAIDs were not associated with the diagnosis of nonunion while selective COX-2 inhibitors were. It seems to me that, given the sparse and conflicting clinical evidence today, a brief course of NSAIDs for fracture-related pain management should be included for patients while we await answers from studies with more robust research designs.
Marc Swiontkowski, MD