Childhood mortality during and after acute illness in Africa and south Asia: a prospective cohort study

Abstract

Background: Mortality among children with acute illness in low-income and middle-income settings remains unacceptably high and the importance of post-discharge mortality is increasingly recognised. We aimed to explore the epidemiology of deaths among young children with acute illness across sub-Saharan Africa and south Asia to inform the development of interventions and improved guidelines.

Methods: In this prospective cohort study, we enrolled children aged 2-23 months with acute illness, stratified by nutritional status defined by anthropometry (ie, no wasting, moderate wasting, or severe wasting or kwashiorkor), who were admitted to one of nine hospitals in six countries across sub-Saharan Africa and south Asia between Nov 20, 2016, and Jan 31, 2019. We assisted sites to comply with national guidelines. Co-primary outcomes were mortality within 30 days of hospital admission and post-discharge mortality within 180 days of hospital discharge. A priori exposure domains, including demographic, clinical, and anthropometric characteristics at hospital admission and discharge, as well as child, caregiver, and household-level characteristics, were examined in regression and survival structural equation models.

Findings: Of 3101 children (median age 11 months [IQR 7-16]), 1120 (36·1%) had no wasting, 763 (24·6%) had moderate wasting, and 1218 (39·3%) had severe wasting or kwashiorkor. Of 350 (11·3%) deaths overall, 234 (66·9%) occurred within 30 days of hospital admission and 168 (48·0%) within 180 days of hospital discharge. 90 (53·6%) post-discharge deaths occurred at home. The proportion of children who died following discharge was relatively preserved across nutritional strata. Numerically large high-risk and low-risk groups could be disaggregated for early mortality and post-discharge mortality. Structural equation models identified direct pathways to mortality and multiple socioeconomic, clinical, and nutritional domains acting indirectly through anthropometric status.

Interpretation: Among diverse sites in Africa and south Asia, almost half of mortality occurs following hospital discharge. Despite being highly predictable, these deaths are not addressed in current guidelines. A fundamental shift to a child-centred, risk-based approach to inpatient and post-discharge management is needed to further reduce childhood mortality, and clinical trials of these approaches with outcomes of mortality, readmission, and cost are warranted.

Funding: The Bill & Melinda Gates Foundation.

Figure 1

Cumulative hazard curves, stratified by nutritional status at hospital admission Data provided are adjusted for age, sex, and site. (A) Mortality in the first 30 days following hospital admission (2935 child-months). (B) Post-discharge mortality (15 999 child-months). (C) Monthly mortality rates following discharge from hospital.

Figure 2

Structural equation models of relationships between domains of characteristics and 30-day mortality and post-discharge mortality (A) Relationships between domains of hospital admission and pre-existing characteristics and 30-day mortality. (B) Relationships between domains of hospital discharge, hospital admission, and pre-existing characteristics and post-discharge mortality. Significant associations (p<0·05) are indicated by solid arrows and non-significant associations by dashed arrows. Significant negative associations are indicated by the – symbol. For significant associations, the arrow thickness corresponds to the effect size. Each outcome was adjusted for age and sex and was modelled with a random intercept at the site level. The full estimation results are provided in the appendix (pp 44–45). Model results were weighted using sampling and lost to follow-up weights.