Five studies were included in the meta-analysis (n=unclear), including three RCTs and two matched case-control studies. Overall, the authors indicated low to moderate study quality. The RCTs reported appropriate allocation concealment and were blinded to assessment. Loss to follow-up across all included studies ranged from 0.75 to 14.1% (where reported).
No studies reported diarrhoea or rotavirus deaths. In the absence of mortality data the authors used effectiveness against serious morbidity (i.e. very severe rotavirus infection) to estimate the effect of rotavirus vaccine on diarrhoea mortality. A relative reduction of 74% (95% CI 35 to 90) was found for rotavirus vaccine (pentavalent) in children under five years old, based on one matched case-control study.
Compared with placebo, a reduction in relative risk was found in vaccinated infants for severe rotavirus infection (RRR 89.1%, 95% CI 77.9 to 94.6; three RCTs), severe gastroenteritis infection (RRR 44.2%, 95% CI 32.8 to 53.7; two RCTs), rotavirus hospitalisations (RRR 92.7%, 95% CI 77.2 to 97.6; three RCTs), gastroenteritis hospitalisations (RRR 56%, 95% CI 39 to 73; three RCTs), and any rotavirus (RRR 74.4%, 95% CI: 63.2 to 82.2; two RCTs). Evidence of significant statistical heterogeneity was found for all but severe gastroenteritis infection, where borderline heterogeneity was found and a random-effects model used.
The risk reduction of pentavalent vaccine in Latin America was 74% (95% CI:35 to 90) against very severe rotavirus, 61% (95% CI 38 to 75) against rotavirus infection, and 47% (95% CI 22 to 64) against rotavirus hospitalisation (one matched case-control study). The risk reduction of monovalent vaccine in a predominantly indigenous population in northern Australia was 57 (95% CI <0 to 83) against rotavirus hospitalisation (one matched case-control study).