Thirteen trials (n=1,559) were included in the review: 11RCTs and 2 quasi-randomised trials.
The quality of the included studies was generally high, Nine studies had adequate allocation concealment, nine standardised or described cointerventions, and all used an intention-to-treat analysis.
Mortality (10 trials): no overall statistically significant difference was found between prone and supine positioning (RR 0.96, 95% CI: 0.46, 1.28, p=0.52). The subgroup analysis revealed no difference in either trials using short-term prone positioning or in those using prolonged prone positioning; there was also no difference between patients with acute lung injury or respiratory failure and others.
Oxygenation (9 trials): prone ventilation resulted in an increased ratio of partial pressure of oxygen to inspired fraction of oxygen of between 23 and 34% on post-randomisation days 1 to 3, measured at the end of the prone manoeuvre. This ratio remained elevated by between 6 and 9% after return to a supine position.
Ventilator-associated pneumonia (6 trials): there was a significantly lower rate of pneumonia in patients in the prone position groups compared with the supine groups (RR 0.81, 95% CI: 0.66, 0.99, p=0.04).
Duration of ventilation (6 trials): there was a trend towards shorter duration of ventilation for patients in the prone position groups (WMD -0.9 days, 95% CI: -1.9, 0.1, p=0.06).
Number of ventilator-free days (4 trials): there was no significant difference between the groups in the number of ventilator-free days.
Adverse events (8 trials): there was a higher incidence of pressure ulcers in patients in the prone position groups (RR 1.36, 95% CI: 1.07, 1.71, p=0.01; 6 trials). There were no significant differences in any of the following adverse events: endotracheal tube obstruction, accidental extubation, loss of central venous or arterial access, thorocostomy tube dislodgement, pneumothorax, cardiac arrest. However, an alternative analysis using data from an additional trial reporting adverse event data in a different format did produce a statistically significantly higher risk of endotracheal tube obstruction in the prone position groups (RR 2.46, 95% CI: 1.33, 4.55, p=0.004).
There was no evidence of statistical heterogeneity in any of the analyses.
There was some evidence of publication bias from the funnel plot analysis, which was not confirmed by statistical tests.
Further results were reported on the CMAJ website. See Web address at end of abstract.