Epidemiology of Shigella species and serotypes in children: a retrospective substudy of the MAL-ED observational birth cohort study

Abstract

Background: Shigellosis is a leading cause of diarrhoea in children globally. We aimed to assess the burden of specific Shigella species and Shigella flexneri serotypes, characterise their clinical syndromes and natural immunity, and evaluate their relevance as causes of diarrhoea and linear growth faltering.

Methods: In the MAL-ED birth cohort study, children younger than 17 days were enrolled from Nov 3, 2009, to Feb 29, 2012, in Bangladesh, Brazil, India, Nepal, Pakistan, Peru, Tanzania, and South Africa and were followed up for 2 years. In this retrospective substudy, we retested Shigella quantitative PCR-positive diarrhoeal and non-diarrhoeal stool samples with molecular subtyping assays. We estimated the prevalence of specific Shigella species and serotypes, estimated their associations with diarrhoea and clinical characteristics using generalised linear mixed models, estimated their associations with linear growth using linear regression, and used longitudinal infection data to estimate protection due to previous infection using the Andersen and Gill extension of the Cox model.

Findings: 1715 children (874 [51·0%] male and 841 [49·0%] female) with complete follow-up to 2 years provided 45 835 stool samples collected between age 0 and 2 years. 1294 (75·5%) children had at least one Shigella detection in a non-diarrhoeal or asymptomatic stool and 507 (29·6%) had at least one Shigella-attributed diarrhoea episode. In this substudy, we were able to type 1202 samples. The most common species and serotypes were Shigella sonnei (366 [30·4%] of 1202), S flexneri 2a (250 [20·8%]), and S flexneri 6 (296 [24·6%]). The associations of S flexneri and S sonnei detection with diarrhoea versus asymptomatic control stools were similar. Compared with diarrhoea episodes attributable to S flexneri, those attributable to S sonnei were less likely to be bloody (prevalence ratio 0·36 [95% CI 0·23 to 0·56]) or severe (prevalence ratio 0·58 [0·35 to 0·96]). The associations between asymptomatic Shigella infections and impaired linear growth at age 2 years were stronger for S flexneri than S sonnei (S flexneri length-for-age Z score difference -0·18 [95% CI -0·29 to -0·07] and S sonnei length-for-age Z score difference -0·07 [-0·21 to 0·07]). Examination of longitudinal infections showed that previous S sonnei infection was associated with a lower hazard of subsequent S sonnei diarrhoea (calibrated hazard ratio 0·41 [95% CI 0·19 to 0·90]). Otherwise, evidence for homotypic or heterotypic natural immunity was not apparent.

Interpretation: This study provides data on prevailing Shigella species and serotypes in settings with a high burden of disease, underscoring the importance of asymptomatic S flexneri infection on growth impairment and the severity of S flexneri diarrhoea. Upcoming Shigella vaccines might need to induce immune responses that improve upon those produced by natural infection.

Funding: Bill & Melinda Gates Foundation.

Figure

Association between Shigella quantity and diarrhoea by Shigella species Association is the species-specific and quantity-specific odds ratio from a generalised linear mixed-effects logistic regression model in which the outcome was diarrhoeal versus non-diarrhoeal stool, and predictors were linear and quadratic terms for pathogen quantity (using the cycle threshold value), child sex, TaqMan Array Card test batch, linear and quadratic terms for child age in 3-month intervals, a random slope for each site, and a random intercept for each individual. Shaded areas show 95% CIs.