Effect of NSAIDs on Bone Healing Rates: A Meta-analysis

Benjamin M. Wheatley, MD; Kyle E. Nappo, MD; Daniel L. Christensen, MD; Ann M. Holman, MLS; Daniel I. Brooks, PhD; Benjamin K. Potter, MD


J Am Acad Orthop Surg. 2019;27(7):e330-e336. 

In This Article


Our review of the orthopaedic literature suggests that the effect of NSAIDs on delayed union or nonunion seems to be strongly related to patient age. Not only was there no pooled effect in the pediatric subgroup but none of the studies inuded in that subgroup demonstrated an increased risk of delayed union or nonunion with exposure to NSAIDs. It is important to note the bimodal age distribution of the included studies with a large gap occurring at a mean age of 18 to 35 years. This gap made it difficult to analyze age as a continuous variable. A lack of studies in early adulthood is important because although NSAID exposure has been cited as a potential risk factor for developing a delayed union or nonunion, so has increasing age.[3] Most adult studies included in our analysis contained an older mean age and, as such, could have contributed to confounding the two effects. Better understanding of how NSAIDs affect healing across the human life span will require additional studies including a more representative distribution of ages and/or dedicated studies of younger adults.

Previous studies have suggested that NSAIDs may have different effects in long bones compared with the spine. Although there was a numerically higher effect of delayed union or nonunion in studies that involved the spine, a meta-regression between these two groups of studies did not find a significant difference between them.

The results of the subgroup analysis with a low dose or short duration of treatment suggest that NSAIDs administered in this fashion do not markedly increase the risk of a bone healing complication. The biologic basis for this finding may be that relatively less potent or brief prostaglandin suppression allows for fracture healing to continue at a slower rate (ie, low-dose exposure) or resume after NSAID withdrawal (ie, short-duration exposure) in patients thus treated.[37,45] However, care should be taken in interpreting this finding. This subgroup only included 4 of the 16 evaluable studies, and they were not consistent in their findings: although 2 found no effect, the other two found an increased risk. Therefore, because of the low number of included studies and patients in this cohort, a definitive answer cannot be given with regard to dosing or duration of treatment. However, our findings do suggest that short-duration or low-dose NSAID exposure represents a decreased risk of delayed union or nonunion versus prolonged or high-dose exposure.

Higher quality studies demonstrated a decreasing amount of variability compared with the studies with a lower NOS grade. The study with the lowest NOS grade included in our analysis also demonstrated the highest OR and the largest CI; however, this study was included in the analysis because it met the a priori criteria by having an NOS grade of five or greater. Of note, if this study were to have been considered an outlier or otherwise unacceptable and removed, the pooled effect would have been reduced but still significant (OR, 1.82; CI, 1.09 to 3.02).

Limitations of the present meta-analysis include that the available studies were conspicuously lacking in both randomized controlled trials and studies that consisted of young adult patients. The latter issue may affect the generalizability of the available data. Heterogeneity among the included studies makes them difficult to compare. The degree of heterogeneity may have contributed to the lack of statistically significant findings between some groups. We attempted to address this heterogeneity through the use of a random-effects model, which should have better generalizability across these varied designs and patient populations.

Based on the available literature, NSAID exposure seems to increase the risk of a delayed union or nonunion in a healing bone. No significant effect exists in the pediatric population. In contrary to previous reports, we did not find a differential effect based on the type of bone. In our analysis, a low dose or short duration of NSAID treatment did not show an increased risk of bone healing complication, but this analysis was limited by a low number of included studies. Despite an increasing body of literature regarding the effect of NSAIDs in healing bones, a definitive answer remains elusive. Many confounding factors may play a role in the safety of NSAIDs in the setting of a healing bone. Additional high-quality and randomized studies on this topic are still wanting, which may provide a more definitive answer for specific patient, surgery and injury populations, and alternate NSAID dosing regimens. Although NSAIDs provide an appealing potential adjunct or alternative to opioid medications, based on our findings, they should be avoided in patients at an increased risk of nonunion or delayed union, and our understanding of NSAID effects on bone healing remains incomplete.