Impact of Shigella infections and inflammation early in life on child growth and school-aged cognitive outcomes: Findings from three birth cohorts over eight years

Abstract

Background: Shigella infections cause inflammation, which has been hypothesized to mediate the associations between Shigella and child development outcomes among children in low-resource settings. We aimed to assess whether early life inflammation and Shigella infections affect school-aged growth and cognitive outcomes from 6-8 years of age.

Methodology/principal findings: We conducted follow-up assessments of anthropometry, reasoning skills, and verbal fluency in 451 children at 6-8 years of age in the Brazil, Tanzania, and South Africa sites of MAL-ED, a longitudinal birth cohort study. We estimated the associations between Shigella burden and inflammation with linear growth at 2, 5, and 6-8 years of age, and with the cognitive test scores using linear regression and adjusting for potential confounding variables. We also assessed whether inflammation mediated the associations between Shigella and school-aged outcomes using a regression-based approach to mediation analysis. A high prevalence of Shigella was associated with a 0.32 (95% CI: 0.08, 0.56) z-score lower height-for-age z-score (HAZ) at 6-8 years compared to a low prevalence of Shigella. Intestinal inflammation had a smaller association with HAZ at 6-8 years. Shigella burden had small and consistently negative associations with cognitive outcomes in Brazil and Tanzania, but not South Africa, and the estimates were not statistically significant. Systemic inflammation was strongly associated with lower verbal fluency scores in Brazil (semantic fluency z-score difference: -0.57, 95% CI: -1.05, -0.10; phonemic fluency z-score difference: -0.48, 95% CI: -0.93, -0.03). There was no evidence that intestinal inflammation mediated the association between Shigella and HAZ or cognitive outcomes.

Conclusions/significance: While Shigella infections were consistently associated with long-term deficits in linear growth, the estimates of the negative associations between Shigella and cognitive outcomes were imprecise and only observed in the Brazil and Tanzania sites. Systemic inflammation was strongly associated with lower semantic and phonemic fluency scores in Brazil only, highlighting the site-specificity of effects.

Fig 1

Associations between Shigella prevalence (A) and quantity (B) in the first 2 years of life with linear growth and cognitive outcomes at 6–8 years of age among 451 children in the Brazil, South Africa, and Tanzania MAL-ED cohorts. Mean Z-score difference estimates and 95% confidence intervals were derived from linear regression models adjusting for site, age at the 6–8 year assessment, enrollment weight-for-age z-score (or enrollment length-for-age z-score for height outcomes), sex, socioeconomic status, exclusive breastfeeding in the first 6 months, maternal height, and the burden of each of the 12 most prevalent pathogens identified in the first 2 years of life (excluding Shigella).

Fig 2

Associations between intestinal (A) and systemic (B) inflammation in the first 2 years of life with linear growth and cognitive outcomes at 6–8 years of age among 451 children in the Brazil, South Africa, and Tanzania MAL-ED cohorts. Mean Z-score difference estimates and 95% confidence intervals were derived from linear regression models adjusting for site, age at the 6–8 year assessment, enrollment weight-for-age z-score (or enrollment length-for-age z-score for height outcomes), sex, socioeconomic status, exclusive breastfeeding in the first 6 months, maternal height, and the burden of each of the 12 most prevalent pathogens identified in the first 2 years of life (excluding Shigella).

Fig 3

Associations between Shigella and inflammation in the first 2 years of life with linear growth at 2, 5, and 6–8 years of age among 451 children in the Brazil, South Africa, and Tanzania MAL-ED cohorts. Mean Z-score difference estimates and 95% confidence intervals were derived from linear regression models adjusting for site, enrollment length-for-age z-score, sex, socioeconomic status, exclusive breastfeeding in the first 6 months, maternal height, and the burden of each of the 12 most prevalent pathogens identified in the first 2 years of life (excluding Shigella).