Efficacy and safety of anti-inflammatory drug-assisted treatment of symptoms in patients with schizophrenia: a meta-analysis

Although anti-inflammatory agents have been explored as adjunctive treatments for schizophrenia, findings remain inconsistent. While some meta-analyses suggest benefit, concerns persist regarding methodological heterogeneity, language bias, and the influence of publication bias on reported effect sizes (ES). This updated meta-analysis aims to re-evaluate the efficacy and safety of anti-inflammatory drugs in schizophrenia, with a critical focus on robustness after accounting for these biases. We conducted a systematic review and meta-analysis of randomized controlled trials (RCTs) evaluating anti-inflammatory drugs as adjunctive therapy in schizophrenia. Databases (PubMed, Embase, Scopus, Web of Science) were searched from inception to December 31, 2024, with restriction to English-language publications. Standardized mean differences (Hedges’s g) were calculated using a random-effects model. Heterogeneity was assessed using I², and publication bias was evaluated via Egger’s test and trim-and-fill analysis. Sensitivity analyses were performed to assess robustness. Seventy-four RCTs involving 5,484 participants were included. Initial analyses suggested significant benefits for aspirin (ES = 0.64), celecoxib (ES = 0.49), estrogens (ES = 0.59), fatty acids (ES = 0.25), minocycline (ES = 0.38), N-acetylcysteine (NAC; ES = 0.60), monoclonal antibodies (mAbs; ES = 0.49), and pregnenolone (ES = 0.20). However, substantial heterogeneity (I² > 50%) and significant publication bias were detected for several agents, including estrogens (p < 0.001) and minocycline (p = 0.01). Trim-and-fill analysis indicated that the observed effects for estrogens and minocycline were likely inflated, with adjusted ES falling below conventional thresholds for statistical significance in some cases. Sensitivity analyses excluding high-risk-of-bias studies further weakened the evidence for several compounds. No significant benefit was found for statins or varenicline. While some anti-inflammatory agents initially appear beneficial as adjunctive treatments in schizophrenia, these findings are tempered by high heterogeneity, potential language bias (due to exclusive inclusion of English-language studies), and significant publication bias. After correction for these biases, the clinical significance of many reported effects diminishes. Current evidence does not robustly support routine clinical use of these agents outside of research settings. Future large-scale, long-term, and globally representative RCTs with rigorous reporting standards are needed to determine whether any anti-inflammatory strategy offers reproducible and clinically meaningful benefits.