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

Disclosures

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

In This Article

Results

A total of 4,341 studies were initially screened. Twenty-six studies were considered for inclusion, of which seven[21–27] were excluded for various reasons (see Table 1, Supplemental Digital Content 2, http://links.lww.com/JAAOS/A151). Two studies[28,29] were included only after additional information was obtained from the original authors. Ultimately, 19 studies were considered for final analysis[5,6,28–44] (see Table 2, Supplemental Digital Content 2, http://links.lww.com/JAAOS/A151), although three of these studies were excluded from analysis because of zero events in both the exposed and nonexposed groups. Data from the included 16 studies were extracted for analysis (see Table 3, Supplemental Digital Content 2, http://links.lww.com/JAAOS/A151). The pooled analysis included a total of 15,242 bones, of which 3,283 were exposed to NSAIDs and 11,959 were not. A total of 512 cases of delayed union or nonunion were reported, of which 226 had been exposed to NSAIDs and 286 had not, resulting in an OR of 2.07 with a 95% CI of 1.19 to 3.61, indicating an increased risk of delayed union or nonunion with NSAID exposure (Figure 2).

Figure 2.

Forest plot showing all included studies. Three studies with zero events in both the exposed and nonexposed arms were excluded from the final analysis. Values are presented on a log scale to normalize their graphical distribution.

A subgroup analysis of studies only involving pediatric patients was performed (n = 4 studies; Figure 3). The subgroup analysis included a total of 2,017 bones, of which 1,181 were exposed to NSAIDs and 836 were not. A total of 37 cases of delayed union or nonunion were reported, of which 13 had been exposed to NSAIDs and 24 had not. Despite the overall effect, this subgroup analysis suggested that exposure to NSAIDs in this population did not result in an increased risk of delayed union or nonunion (OR, 0.58; CI, 0.27 to 1.21). A subgroup analysis of adult studies (n = 12 studies; Figure 4), however, continued to demonstrate an increased risk of delayed union or nonunion (OR, 2.93; CI, 1.61 to 5.33). The adult studies included a total of 13,225 bones, with 2,102 of which were exposed to NSAIDs and 11,123 that were not. This included 475 cases of delayed union or nonunion, of which 213 had been exposed to NSAIDs and 262 that had not. A meta-regression analysis that directly tested the effect of age confirmed a significant difference between studies involving pediatric patients and those of adult patients (P = 0.007).

Figure 3.

Forest plot showing studies with pediatric patients only. One study with zero events in both the exposed and nonexposed arms was excluded. Values are presented on a log scale.

Figure 4.

Forest plot showing studies with adult patients only. Two studies with zero events in both the exposed and nonexposed arms were excluded. Values are presented on a log scale.

Further analysis on the adult-only subgroup examined the type of bone studied in each article. For this analysis, we removed one study,[42] which examined the pelvis. There were six studies of long bones consisting of a total of 12,030 bones, with 1,349 that had been exposed to NSAIDs and 10,681 that had not. A total of 328 cases of delayed union or nonunion were reported, of which 89 had been exposed to NSAIDs and 239 had not. This analysis revealed a markedly increased risk of delayed union or nonunion in studies with long bones (n = 6 studies; OR, 2.34; CI, 1.12 to 4.90). An additional five studies involving the spine included a total of 1,127 patients, with 706 who were exposed to NSAIDs and 421 who were not. A total of 125 cases of delayed union, nonunion, or pseudarthrosis were reported, of which 106 had been exposed to NSAIDs and 19 had not. Subgroup analysis of this cohort similarly demonstrated a markedly increased risk of bone healing complication (n = 5 studies; OR, 4.90; CI, 1.45 to 16.58). However, meta-regression between these study cohorts did not suggest a significant difference (P = 0.33).

A total of four studies of adult patients with a low dose or short duration of treatment were reported (Figure 5). A total of 1,109 bones were included in this analysis, with 510 that had been exposed to NSAIDs and 599 that had not. A total of 98 cases of delayed union or nonunion were reported, of which 52 had been exposed to NSAIDs and 46 had not. This cohort did not demonstrate a markedly increased risk of delayed union or nonunion as a result of exposure to NSAIDs (OR, 1.68; CI, 0.63 to 4.46).

Figure 5.

Forest plot showing only studies with a low dose or short duration of NSAID exposure. One study with zero events in both the exposed and nonexposed arms was excluded. Values are presented on a log scale.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....