Rationale: It is commonly stated that mortality from acute respiratory distress syndrome (ARDS) and acute lung injury (ALI) is decreasing.

Objectives: To systematically review the literature assessing ARDS mortality over time and to determine patient- and study-level factors independently associated with mortality.

Methods: We searched multiple databases (MEDLINE, EMBASE, CINAHL, Cochrane CENTRAL) for prospective observational studies or randomized controlled trials (RCTs) published during the period 1984 to 2006 that enrolled 50 or more patients with ALI/ARDS and reported mortality. We pooled mortality estimates using random-effects meta-analysis and examined mortality trends before and after 1994 (when a consensus definition of ALI/ARDS was published) and factors associated with mortality using meta-regression models.

Measurements and main results: Of 4,966 studies, 89 met inclusion criteria (53 observational, 36 RCTs). There was a total of 18,900 patients (mean age 51.6 years; 39% female). Overall pooled weighted mortality was 44.3% (95% confidence interval [CI], 41.8-46.9). Mortality decreased with time in observational studies conducted before 1994; no temporal associations with mortality were demonstrated in RCTs (any time) or observational studies (after 1994). Pooled mortality from 1994 to 2006 was 44.0% (95% CI, 40.1-47.5) for observational studies, and 36.2% (95% CI, 32.1-40.5) for RCTs. Meta-regression identified study type (observational versus RCT, odds ratio, 1.36; 95% CI, 1.08-1.73) and patient age (odds ratio per additional 10 yr, 1.27; 95% CI, 1.07-1.50) as the only factors associated with mortality.

Conclusions: A decrease in ARDS mortality was only seen in observational studies from 1984 to 1993. Mortality did not decrease between 1994 (when a consensus definition was published) and 2006, and is lower in RCTs than observational studies.