Eighteen studies were included in the review. Nine studies included 16,857 trauma patients and nine included 7,659 patients with cardiac arrest. Most of the studies were before-and-after designs or quasi-randomised trials (exact number not reported).
Sequence generation and allocation concealment of the included studies provided the largest risk of bias. Risk of bias associated with blinding of participants, personnel and outcomes was most often uncertain. Other sources of bias were considered low risk.
Trauma victims: Meta-analysis suggested that advanced life support care in trauma patients reduced the probability of survival at hospital discharge by 34% compared to basic life support care (OR 0.659, 95% CI 0.594 to 0.732; nine studies). Sensitivity analysis that excluded the largest trial (9,405 patients) produced a non-significant pooled relative effect of advanced life support (OR 0.892, 95% CI 0.775 to 1.026; eight studies).
Non trauma cardiac-arrest patients: Pooled analysis suggested that advanced life support care in non trauma cardiac arrest patients increased the odds of survival at hospital discharge by nearly 47% compared to basic life support care (OR 1.468, 95% CI 1.257 to 1.715; nine studies). Subgroup analyses suggested that advanced life support provided by physicians significantly increased the probability of survival at hospital discharge compared to basic life support (OR 2.047, 95% CI 1.593 to 2.631; six studies).
A sub-analysis of studies where survival data of patients at hospital admission were available suggested that advanced life support improved the probability of survival at hospital admission compared to basic life support (OR 1.733, 95% CI 1.537 to 1.954; four studies).
Number needed to treat calculations and Χ² and Ι² details were not reported.