Four RCTs were included (n=1,271 patients, range 40 to 791). Three were high quality trials (scoring 3 out of 5 points) and one low quality trial (scoring 2 points).
There was no statistically significant difference between the prone and supine position groups in overall intensive care unit (ICU) mortality (four RCTs). However among the most severely-ill patients, the prone position was associated with significantly lower ICU mortality (OR 0.34, 95% CI 0.18 to 0.66; two RCTs; n=195 patients).
There was no statistically significant difference between the prone and supine position groups in ventilator-associated pneumonia (three RCTs), ICU length of stay (one RCT), duration of mechanical ventilation (two RCTs) or pneumothorax (two RCTs).
In the prone positioning group, there was a significantly higher rate of new or worsening pressure sores (OR 1.49, 95% CI 1.17 to 1.89; three RCTs; n=1,135 patients).
There was no significant statistical heterogeneity for any analysis.
All results presented above were by intention-to-treat.