Prevalence of rotavirus among older children and adults with diarrhea: A systematic review and meta-analysis.

BACKGROUND Older children and adults are susceptible to rotavirus, but the extent to which rotavirus affects this population is not fully understood, hindering accuracy of global rotavirus estimations. OBJECTIVE To determine what proportion of diarrhea cases are due to rotavirus among persons ≥ 5 years old and to estimate this proportion by age strata. METHODS We conducted a systematic review and meta-analysis using the PRISMA guidelines. We included studies that reported on conditional rotavirus prevalence (i.e., percent of diarrhea due to rotavirus) in persons ≥ 5 years old who were symptomatic with diarrhea/gastroenteritis and had laboratory confirmation for rotavirus infection. Studies on nosocomial infections and outbreak investigations were excluded. We collected age group-specific conditional rotavirus prevalence and other variables, such as study geography, study setting, and study type. We calculated pooled conditional rotavirus prevalence, corresponding 95% confidence intervals (95% CI), heterogeneity (I2) estimates, and prediction intervals (PI). RESULTS Sixty-six studies from 32 countries met the inclusion criteria. Conditional rotavirus prevalence ranged from 0% to 30% across the studies. The total pooled prevalence of rotavirus among persons ≥ 5 years old with diarrhea was 7.6% (95% CI: 6.2-9.2%, I2 = 99.6%, PI: 0-24%). The pooled prevalence of rotavirus among older children and adolescents was 8.7% (95% CI: 6.2-11.7%, I2 = 96%, PI:0-27%), among younger adults was 5.4% (95% CI: 1.4-11.8%, I2 = 96%, PI:0-31%), and among older adults was 4.7% (95% CI: 2.8-7.0%, I2 = 96%, PI:0-16%). Pooled conditional rotavirus prevalences did not differ by other variables. CONCLUSION In this systematic review and meta-analysis of rotavirus among persons ≥ 5 years old with diarrhea, we found relatively low pooled conditional rotavirus prevalence compared to what is typically reported for children < 5 years; however, results should be interpreted with caution as the wide prediction intervals suggest large heterogeneity.