Evaluating the Impact of Unadjusted Confounding and Study Design on Estimated Pathogen-attributable Diarrhea...

Understanding diarrhea etiology is critical for understanding vaccine impact, but causal attribution is difficult. We evaluated the sensitivity of an existing attributable incidence estimator to unadjusted confounding and study design.

We used MAL-ED data to estimate attributable incidence by regressing diarrhea incidence on pathogen quantity and calculating attributable fractions. The partially adjusted model used mixed-effects logistic regression adjusted for co-infection, sex, test batch, and age. We evaluated the magnitude of confounding by prior immunity, antibiotic use, socioeconomic status, and breastfeeding. To understand the impact of post-diarrheal shedding, we compared estimates including and excluding stools collected ±7, 14, or 28 days from an episode. We conducted matched nested case-control and case-crossover analyses and used Poisson regression to calculate risk-based estimates of incidence.

40,406 stools samples from 1715 children were included (6625 diarrheal, 33,781 control). Antibiotic use had the greatest confounding impact, and directionality of bias was dependent on pathogen type. Shedding following diarrheal episodes did not substantially or consistently bias estimates. Nested case-control and case-crossover study designs produced similar attributable incidence estimates, and risk-based estimates were lower than odds-based estimates, ranging from 33% (Shigella) to 49% (tEPEC) episodes per 100 child-years fewer than the partially adjusted model.