Incidence of adverse events with topical and oral nonsteroidal anti-inflammatory drugs for chronic knee pain: a systematic review with meta-analysis and meta-regression

OBJECTIVE: To quantify the incidence of adverse events (AEs) with topical and oral NSAIDs for chronic knee pain and evaluate factors associated with AE incidence. METHODS: Randomized controlled trials (RCTs) of topical or oral non-steroidal anti-inflammatory drugs (NSAIDs) reporting AEs in people with chronic knee pain were included. Searches were conducted in Medline, Embase, CINAHL, CENTRAL, and Web of Science from inception to January 2026. Two reviewers independently extracted AE data, assessed risk of bias, and rated certainty of evidence. RESULTS: Among 137 RCTs (13,634 participants), oral NSAIDs were associated with an incidence of all reported AEs of 25.1% (95% confidence interval (CI) 20.8-30.4%), with 19.3% (95% CI = 14.6-22.4%) of participants experiencing one or more AE. For topical NSAIDs, incidences were 7.8% (95% CI = 4.5-13.5%), and 2.4% (95% CI = 0.6-5%). Gastrointestinal disorders accounted for 64.5% of categorized AEs with oral NSAIDs, whereas skin disorders accounted for 56% with topical NSAIDs. COX-2 selective NSAIDs had an incidence of all reported AEs of 19.4% (95% CI = 13.3-28.3%), and non-selective NSAIDs of 34.5% (95% CI = 27.2-43.6%). The incidence of serious AEs could not be reliably estimated because of their rarity. Age, sex, and BMI did not explain detectable differences in AE incidence. All analyses were supported by very-low- or low-certainty evidence. CONCLUSION: Overall, the study-level incidence of AEs associated with topical and oral NSAIDs is low, and serious AEs are rare. Our findings can inform clinical decision-making alongside established evidence on treatment efficacy and individual patient factors. PROTOCOL REGISTRATION: www.crd.york.ac.uk/prospero identifier is CRD420261301656.