Clinical Question

Do nonsteroidal anti-inflammatory drugs (NSAIDs) prevent or delay fracture healing when used for pain management?

Evidence-Based Answer

The use of NSAIDs for more than three days at higher doses during the postoperative or acute phase of fracture healing may lead to increased rates of nonunion, delayed union, and pseudarthrosis in adults. (Strength of Recommendation [SOR]: B, based on multiple systematic reviews of randomized controlled trials [RCTs], cohort studies, and case-control trials.) NSAIDs do not appear to impair fracture healing in children younger than 11 years. (SOR: B, based on multiple systematic reviews of RCTs, cohort studies, and case-control trials.)

Evidence Summary

A 2019 meta-analysis (n = 14,887) of 16 studies that included RCTs, cohort studies, and case-control trials examined the adverse effects of NSAIDs on bone healing in the setting of fracture, osteotomy, or fusion surgery.¹ The primary outcomes were nonunion, delayed union, and pseudarthrosis with a minimum follow-up of six months. The pooled analysis included 15,242 bones, of which 3,283 were exposed to NSAIDs. In 512 patients with delayed union or nonunion fractures, 226 (6.9%) had been exposed to NSAIDs and 282 (2.4%) had not, showing an increased risk with NSAID use (odds ratio [OR] = 2.1; 95% CI, 1.2 to 3.6; number needed to harm [NNH] = 23). A subgroup analysis of retrospective cohort studies in children found that exposure to NSAIDs did not result in an increased risk of delayed union or nonunion (four trials; n = 2,017 bones; 13 of 37 cases of delayed healing exposed to NSAIDs; OR = 0.6; 95% CI, 0.3 to 1.2). A subgroup analysis of low-dosage or short-duration NSAID exposure (low dosage was defined as less than 125 mg per day of diclofenac, less than 150 mg per day of indomethacin, or less than 120 mg per day of ketorolac, whereas short duration was defined as less than two weeks of treatment) did not find an increased risk of delayed union or nonunion (four trials; n = 1,109; OR = 1.7; 95% CI, 0.6 to 4.5). This meta-analysis was limited by significant heterogeneity (I² = 77.3%) and significant bimodal age distribution of the included studies, with a large gap occurring at a mean age of 18 to 35 years.