Identifying effect modifiers of corticosteroids in acute respiratory distress syndrome: a meta-regression analysis of randomized controlled trials

Systemic corticosteroids are administered in acute respiratory distress syndrome (ARDS) for their anti-inflammatory effects. However, the factors that influence their therapeutic efficacy remain poorly understood. We conducted a systematic review and meta-analysis of randomized controlled trials (RCTs) evaluating corticosteroid therapy in adults with non-COVID ARDS. Comprehensive searches of PubMed, CENTRAL, and Web of Science were performed through October 2024. A random-effects model was used to pool the risk ratios (RRs) for short-term all-cause mortality. Inverse-variance-weighted univariable meta-regression analyses were conducted to assess associations between study-level characteristics and treatment effects. Subgroup analysis was performed based on the era of lung-protective ventilation (LPV) adoption using a random-effects model. Adverse events including intensive care unit (ICU)-acquired weakness, superinfection, and barotrauma were also evaluated. The credibility of subgroup effects was evaluated using Instrument for assess the Credibility of Effect Modification Analyses (ICEMAN), and certainty of evidence was assessed using Grading of Recommendations Assessment, Development and Evaluation (GRADE) framework. Eleven RCTs encompassing 1,441 patients met inclusion criteria. Meta-regression identified year of publication as the only potential of effect modifier (coefficient: -0.014; 95% confidence interval [CI]: -0.031 to 0.002; p = 0.082). Subgroup analysis of short-term mortality showed a lower RR in trials conducted after LPV implementation (RR 0.65; 95% CI: 0.50 to 0.86) than in earlier trials (RR 0.86; 95% CI: 0.73 to 1.02), although a between-subgroup difference was not apparent (p = 0.095). Both ICEMAN credibility of subgroup effect and GRADE certainty of evidence were moderate. A sensitivity analysis using the longest available mortality time point showed a consistent directional pattern (RR 0.71; 95% CI 0.54 to 0.94 after LPV, compared with RR 0.86; 95% CI 0.73 to 1.01 before LPV; p = 0.253). Analyses of adverse events suggested a potential increase in ICU-acquired weakness, no clear increase in superinfection, and a possible reduction in barotrauma. Benefit appears larger in more recent trials conducted in the LPV era; this may reflect improvements in ventilatory care, optimization of steroid regimens, and other secular changes in non-COVID ARDS, highlighting the need for further stratified research. PROSPERO CRD42024597366.