• Jeremy R Beitler, Shahzad Shaefi, Sydney B Montesi, Amy Devlin, Stephen H Loring, Daniel Talmor, and Atul Malhotra.
• Division of Pulmonary and Critical Care Medicine, Brigham and Women's Hospital, 221 Longwood Avenue, BLI 036, Boston, MA, 02115, USA, jbeitler@partners.org.
• Intensive Care Med. 2014 Mar 1; 40 (3): 332-41.

PurposeProne positioning for ARDS has been performed for decades without definitive evidence of clinical benefit. A recent multicenter trial demonstrated for the first time significantly reduced mortality with prone positioning. This meta-analysis was performed to integrate these findings with existing literature and test whether differences in tidal volume explain conflicting results among randomized trials.MethodsStudies were identified using MEDLINE, EMBASE, Cochrane Register of Controlled Trials, LILACS, and citation review. Included were randomized trials evaluating the effect on mortality of prone versus supine positioning during conventional ventilation for ARDS. The primary outcome was risk ratio of death at 60 days meta-analyzed using random effects models. Analysis stratified by high (>8 ml/kg predicted body weight) or low (≤ 8 ml/kg PBW) mean baseline tidal volume was planned a priori.ResultsSeven trials were identified including 2,119 patients, of whom 1,088 received prone positioning. Overall, prone positioning was not significantly associated with the risk ratio of death (RR 0.83; 95% CI 0.68-1.02; p = 0.073; I (2) = 64%). When stratified by high or low tidal volume, prone positioning was associated with a significant decrease in RR of death only among studies with low baseline tidal volume (RR 0.66; 95% CI 0.50-0.86; p = 0.002; I (2) = 25%). Stratification by tidal volume explained over half the between-study heterogeneity observed in the unstratified analysis.ConclusionsProne positioning is associated with significantly reduced mortality from ARDS in the low tidal volume era. Substantial heterogeneity across studies can be explained by differences in tidal volume.

14 keywords...

hide…