Five RCTs were included in the review (n=795 initially enrolled, n=689 analysed). Sample sizes ranged from 50 to 300. Two RCTs scored 5 on the Jadad scale (high quality). One study scored 4. Two RCTs scored 2.
Incidence of ventilation-associated pneumonia (five RCTs): Patients who underwent mechanical ventilation and received probiotics reported statistically significantly fewer incidences of ventilation-associated pneumonia compared with controls (OR 0.55, 95% CI 0.31 to 0.98, random-effects model). There was no evidence of statistical heterogeneity according to the fixed-effect model.
Subgroup analyses that included only RCTs with similar probiotic regimens and similar administration techniques did not significantly alter the results. However, after removal of the RCT with high incidence of ventilation-associated pneumonia, the results were no longer statistically significant. Subgroup analyses by underlying microorganism showed no significant differences in incidence of ventilation-associated pneumonia between patients who received probiotics and control.
Length of ICU stay (three RCTs): Patients who received probiotics had shorter ICU stays compared with controls when a fixed-effect model was used (WMD -0.99 days, 95% CI -1.37 to -0.61) but not when a random-effects model was used. Eliminating the RCT that reported the longest ICU stay did not significantly alter the results.
There were no statistically significant differences between probiotic and control groups in all-cause mortality during ICU stay (four RCTs), in-hospital stay (two RCTs), duration of mechanical ventilation (three RCTs) or diarrhoea (two RCTs).
Results for other secondary outcomes were reported in the review.