Fifty studies that provided 62 datasets were included in the review (n=8,504 patients). Sample size ranged from 16 to 1,111 patients. The studies included 13 RCTs (n= 1,547 patients); other studies were observational prospective or retrospective studies.
All studies: There was no significant difference between antimicrobial combination therapy and monotherapy in the risk of death or clinical failure. There was significant heterogeneity (p<0.0001; I2=45%). There was no evidence of publication bias from the funnel plot.
Datasets stratified by mortality/clinical failure rate in the monotherapy treatment group (all studies): Combination therapy was associated with a significantly lower risk of death in the subset with a death/clinical failure rate of over 25% (OR 0.54, 95% CI 0.45 to 0.66), but a significantly higher risk of death in the subset with death/clinical failure rate was below 15% (OR 1.53, 95% CI 1.16 to 2.03). There was no significant difference between treatments for the subset with intermediate risk (death/clinical failure rate 15 to 25%). No significant heterogeneity was found for any of these analyses.
Datasets stratified by presence or absence of shock/critical illness (12 studies): Combination therapy was associated with a significantly lower risk of death in the subset with shock/critical illness compared with monotherapy (OR 0.51, 95% CI 0.36 to 0.72). There was no significant difference between combination and monotherapy in subgroups without shock/critical illness. No significant heterogeneity was found for either analysis.
Meta-regression found that the only variable influencing the efficacy of combination therapy was the death rate in the monotherapy group.
Other results were also reported.