Nonsteroidal anti-inflammatory drugs and improved prognosis in critically Ill patients: insights from a retrospective cohort study

The clinical utility of nonsteroidal anti-inflammatory drugs (NSAIDs) in critically ill patients remains controversial due to concerns about adverse effects. This study evaluated associations between NSAID administration and mortality outcomes in intensive care unit patients. We conducted a retrospective cohort study using the MIMIC-IV database, analyzing 47,039 critically ill patients (2008–2019). Patients were categorized as no-NSAIDs users ( n = 22,139, 47.1%), aspirin-only users ( n = 21,604, 45.9%), non-aspirin NSAIDs users ( n = 1,941, 4.1%), or combination users ( n = 1,355, 2.9%). Primary endpoints were 28-day and 90-day mortality. Cox proportional hazards regression with progressive adjustment, propensity score matching, and subgroup analyses were performed. The crude 28-day mortality rates were lower across all NSAID-exposed groups compared with non-users (aspirin-only: 9.0%, non-aspirin NSAIDs: 4.8%, combination: 3.9% vs. no-NSAIDs: 13.9%, p < 0.001). In fully adjusted Cox models, all NSAID categories were independently associated with lower 28-day mortality (aspirin-only: HR = 0.59, 95% CI: 0.55–0.63; non-aspirin NSAIDs: HR = 0.63, 95% CI: 0.51–0.78; combination: HR = 0.43, 95% CI: 0.33–0.56). In subgroup analyses, stronger protective associations were observed with short-term use (≤ 3 days: HR = 0.67, 95% CI: 0.62–0.72), in patients aged ≤ 65 years (HR = 0.45, 95% CI: 0.39–0.53), those with lower comorbidity burden(Charlson index ≤ 6: HR = 0.47, 95% CI: 0.41–0.54), and mechanically ventilated patients(HR = 0.48, 95% CI: 0.44–0.53). E-value analyses indicated moderate unmeasured confounding could explain these associations. NSAID administration was associated with lower mortality in critically ill patients, with the most robust evidence for aspirin-only use. However, the effect sizes substantially exceed those of established ICU interventions, and the observed protective associations for non-aspirin NSAIDs and combination therapy did not persist in landmark or IPTW sensitivity analyses. These findings are likely inflated by contraindication-driven selection and residual confounding, and should not be interpreted as evidence supporting NSAID use for mortality reduction. Prospective randomised trials are required before any clinical recommendations can be considered.