Understanding Enteric Dysfunction as a Determinant of Post-Discharge Pediatric Outcomes in Kenya and Pakistan.

Background: This dissertation aims to establish if enteric dysfunction appears to be an important interventional target for clinical trials hoping to reduce pediatric post-discharge mortality. We also aim to generate new knowledge regarding the pathophysiology of enteric dysfunction among acutely ill children, while validating previous observations from community cohorts. Finally, we will contribute to a better understanding of the lactulose-rhamnose ratio (LRR) test’s utility and reproducibility as a clinical research tool.
Children aged 2-24 months without current diarrhea were recruited from Civil Hospital Karachi, Pakistan and Migori County Referral Hospital, Kenya. LRR tests were administered after children were clinically stable (oral feeds, not dehydrated, no oxygen needs). Similarly aged community children were pseudo-randomly selected from homes near those of the children being discharged, and were also tested using the LRR protocol. The distribution of LRR among the hospitalized and community children were compared, and potential confounders adjusted for, in linear regression of log-transformed LRR. Post-discharge growth was considered a proxy for child health and risk of mortality. Growth data were available for the hospitalized children, and associations between their LRR at discharge and subsequent changes in height-for-age (HAZ) and weight-for-age zscore (WAZ) were estimated using linear mixed effect models. Blood plasma samples from both groups were also sent for proteomic analysis. Four plasma biomarkers of systemic inflammation (CD14, CRP, TNFa, IL-6) and one of intestinal enterocyte damage (IFABP) were compared between the hospitalized and community children to understand if their association with the LRR differed across groups. Finally, extreme gradient boosted models were built to predict LRR using the full proteomic dataset, and model explanation tools were used to understand influential variables in those models.
Results: 137 hospitalized and 84 community children had LRR results available. The median age was nine months in each group. Hospitalized children had lower median MUAC (12.4 vs 13.5cm) and higher HIV infection prevalence (5% vs 1%). The mean log LRR among children being discharged was 0.45 higher (95% CI: 0.05-0.86, p=0.028) than among their community peers. However, after adjusting for weight-for-height z-score, the difference in means was reduced to 0.25 higher (95% CI: -0.20, 0.69, p=0.274). This difference was not attenuated in a sensitivity analysis including only children without HIV infection. LRR at discharge was not associated with changes in WAZ or HAZ in any post-discharge time period. There was evidence that LRR had a different relationship to CRP (p=0.044), TNFa (p=0.027), CD14 (p=0.078), and IL-6 (p=0.171) among the hospitalized as compared to the community children. Predictive models leveraging whole proteomic dataset identified a cluster of plasma proteins involved in leukocyte invasion and tissue repair pathways to be associated with increased enteric permeability. This cluster was correlated with WAZ and HAZ. The cluster was also linked to breastfeeding status and age through biomarkers of animal protein consumption and exposure to bacterial pathogens. Recent diarrhea was not correlated with any of the identified biomarkers of enteric permeability.
Conclusion: Children leaving hospital have greater enteric permeability as assessed by LRR than their community peers. However, among these hospitalized children, enteric permeability was not associated with systemic inflammation at the point of discharge and was not associated with subsequent growth. These findings cast doubt on the degree of benefit that children at hospital discharge are likely to receive from therapeutics targeting ED. Instead, these data suggest resources may be better spent on developing ED interventions for apparently healthy children in the community.