Comparative Efficacy of Liberal vs. Conservative Oxygenation Strategies in Adult ICU Trial Populations: A Random-Effects Meta-Analysis

Objective The use of arterial oxygen targets in mechanically ventilated adults remains controversial. Higher oxygenation and FiO₂ may speed liberation but risk hyperoxemia-related complications; lower targets may increase hypoxemia risk and end-organ damage. This meta-analysis evaluates whether liberal versus conservative oxygen targets impact 28-day ventilator-free days (VFD-28), mortality, or ICU length of stay (LOS). Methods MEDLINE, Embase, and trial registries were searched for randomized trials comparing liberal (≥96% SpO₂ or >100 mmHg PaO₂) or usual oxygenation targets vs. conservative (88-96% SpO₂ or PaO₂ 55-80 mmHg) oxygen targets in mechanically ventilated adults. The primary outcome was VFD-28, reported as mean difference (MD) defined as Conservative minus Liberal with negative values favoring the Liberal group. The results were pooled using DerSimonian–Laird random-effects modeling. When only medians and IQRs were reported, means and SDs were estimated using the Wan method (n ≥ 25); range-only/small-n cases used Hozo. Secondary outcomes included all-cause mortality (risk ratio [RR]) and ICU LOS (MD). Heterogeneity was assessed using τ² and I²; sensitivity analyses included Hartung-Knapp and leave-one-out tests. Trials were aligned on the prespecified 90-day mortality endpoint Results Five trials (n = 18,905) were included. VFD-28 favored liberal targets: pooled MD −0.58 days (95% CI −1.06 to −0.10; negative favors liberal), with low heterogeneity. For 90-day mortality, there was no statistically significant difference between the two groups: pooled RR 1.02 (95% CI 0.98 to 1.06; I² = 0%). ICU LOS also showed no clear difference: pooled MD 0.50 days (95% CI −0.48 to 1.47) with substantial heterogeneity. Findings were robust to cluster design-effect adjustments and sensitivity analyses. Given the few studies in this analysis, the power to detect small-study effects was limited. Discussion Liberal oxygen targets confer a modest VFD benefit without significant differences in ICU LOS or mortality. Further research is warranted to assess potential differences in adverse outcomes such as ventilator-associated lung injury, pneumonia, or ICU-acquired weakness.