Overall compilation of adverse effects of non-steroidal anti-inflammatory drugs: a hypothesis-free systematic investigation using a nationwide cohort study

Background Non-steroidal anti-inflammatory drugs (NSAIDs) are widely used for osteoarthritis (OA), despite various adverse effects (AEs). Previous studies were often limited by small sample sizes, a focus on only predefined outcomes, and an imbalanced research coverage across NSAID subtypes. These factors can cause confirmation or heterogeneity bias, and in clinical practice, focusing on only well-known AEs may lead to the overlooking of other potential AEs. To address this, we conducted a hypothesis-free screening of AEs within a large, single cohort. Methods Using a nationwide South Korean cohort, we selected 888,909 newly diagnosed OA patients with health screening data between 2010 and 2014. The first three characters of ICD codes were considered as potential AEs and their effects were evaluated. To reduce reverse-causation bias, we first used chi-square and Poisson tests to identify significant indications, and excluded the corresponding ICD codes. Time-dependent survival analysis was conducted, defining NSAID users as patients with any annual medication possession ratio (MPR) ≥ 0.1. Additionally, a self-controlled case series analysis was conducted, defining the risk period as up to 6 months after NSAID intake. Further, we assessed the association between five NSAID subtypes (aceclofenac, meloxicam, loxoprofen, celecoxib, and naproxen) and AEs, and compared their adjusted hazard ratios (aHRs) with each other. Results We confirmed previously reported AEs (e.g., anemia, cerebrovascular and cardiorenal diseases). The risk of nephrotoxicity varied significantly by NSAID type, with loxoprofen (aHR = 3.95 [95% CI, 1.56–10.00]), celecoxib (aHR = 2.44 [95% CI, 1.68–3.53]), and naproxen (aHR = 4.7 [95% CI, 2.16–10.24]) showing statistically comparable risks, all of which were significantly higher than that of meloxicam (aHR = 1.22 [95% CI, 0.68–2.19]). Conclusion Our findings enhance the understanding of NSAID safety profiles by identifying dose–response and duration–time AEs. They also contribute to better diagnosis and management of AEs while providing valuable guidelines for both patients and clinicians.