Ten-year trends in clinical characteristics and outcome of children hospitalized with severe wasting or nutritional edema in Malawi (2011-2021): Declining admissions but worsened clinical profiles

Abstract

Background: Severe acute malnutrition (SAM) constitutes a substantial burden in African hospitals. Despite adhering to international guidelines, high inpatient mortality rates persist and the underlying contributing factors remain poorly understood.

Objective: We evaluated the 10-year trend (2011-2021) in clinical factors and outcomes among children with severe wasting and/or nutritional edema at Malawi's largest nutritional rehabilitation unit (NRU).

Methods: This retrospective study analyzed trends in presentation and outcomes using generalized additive models. The association between clinical characteristics and mortality or readmission was examined and key features were also related to time to either mortality or discharge.

Results: 1497 children (53%, females) were included. Median age at admission (23 months, IQR 14, 34) or anthropometry (i.e., weight-for-age, height-for-age and weight-for-height) did not change over the 10-years. But the prevalence of edema decreased by 40% whereas dehydration, difficulty breathing, and pallor became more common. Yearly HIV testing increased but positive-detection remained around 11%. Reporting of complete vaccination dropped by 49%, and no reduction in 'watch' antibiotic usage was detected. Overall admissions declined but mortality remained around 23% [95%CI; 21, 25], and deaths occurred earlier (5.6 days [95%CI; 4.6, 6.9] in 2011 vs. 3.5 days [95%CI; 2.5, 4.7] in 2021; p<0.001). Duration of hospitalization was shortened and readmissions surged from 4.9% [95%CI; 3.3, 7.4] in 2011 to 25% [95%CI; 18, 33] in 2021 (p<0.001). Age, wasting, having both dehydration and diarrhea, or having vomiting, cough, or difficulty breathing were associated with mortality but these associations did not show any interaction over time.

Conclusion: Over 10 years, mortality risk remained high among hospitalized children with SAM and coincided with worsened clinical presentation at admission and increased readmission. Longitudinal data from major NRUs can identify shifts in clinical profiles or outcomes, and this information can be leveraged to promote earlier care-seeking, improved risk stratification, and implementation of more patient-centered treatments.

Fig 1. Trends over 10 years (2011–2021) in clinical characteristics of children with severe acute malnutrition admitted at Moyo NRU.

Frequency per year of (A) Age groups as indicated by legend; (B) Reported breastfeeding in children less than 12 months of age; (C) Nutritional edema; and selected (D) Integrated Management of Childhood Illness (IMCI) danger signs. Linear and non-linear trends were tested with generalized additive models, full results are presented in S1, S3 Tables. Grey dashed line shows linear fit, and significance indicated at right: n.s., non-significant, *p<0.05, **p<0.01, ***p<0.001.

Fig 2. Trends over 10 years in children with severe acute malnutrition admitted at Moyo NRU.

(A) HIV status as per legend; (B) Vaccination status (black line) and unreported vaccination (red); (C) Antibiotic usage as per WHO-Aware classification (Access, black line; and Watch, red); (D) Mortality; (E) Days to discharge (black line) or to death (red); and (F) Readmission. Linear and non-linear trends tested with general additive models, full results presented in S5–S8 Tables. Grey dashed line shows linear fit, and significance indicated at right: n.s., non-significant, *p<0.05, **p<0.01, ***p<0.001.

Fig 3. Differences in the daily probability of mortality or discharge associated with selected clinical features at admission in children admitted to the Moyo NRU over the 10-year period.

Estimated cumulative incidence curves represent the probability of death (in black) or discharge (in grey) at any given day of hospitalization depending on having (A) either lower WHZ (i.e., WHZ < -3.5 z-score) or lower age (i.e., < 2 years) or both (indicated in red); (B) only diarrhea or dehydration, or both (indicated in red); (C) vomiting; (D) edema; (E) difficulty breathing; or (F) cough, as indicated by legends. Group differences were tested using competitive risk analysis which showed that both the rate of mortality and of discharge differed depending on the presence of these clinical features (Results detailed in S10, S11 Tables). WHZ, weight-for-height/length z-score.