Beyond crude mortality rates: what can small-scale surveys tell us about age-specific mortality in humanitarian emergencies?

Heudtlass, Peter Louis [UCL] Masquelier, Bruno [UCL] Guha-Sapir, Debarati [UCL]

Mortality data is a key indicator to assess severity and to identify vulnerable populations in humanitarian emergencies. Due to time and resource constraints, data is often only disaggregated by age to identify children under five years of age. The motivation for using this cut-off is based on the widespread view that children under five are particularly vulnerable in emergencies. Puzzlingly, the few studies analyzing vulnerability of the under-five age group do not support this view. Therefore, this paper aims to provide new empirical evidence on emergency-related mortality risks by age groups using a large dataset from South Sudan. We extracted mortality and demographic data from 221 small-scale mortality surveys conducted between 2000 and 2012 in current day South Sudan. Most reviewed surveys only disaggregate mortality data by under-five and over-five years of age. We used generalized linear regression models to estimate mortality trends and interaction terms to test for differences in emergency-related excess mortality between age groups. We observed a decline in mortality rates after the 2005 peace agreement. The decline in under-five mortality is not significantly different from the rest of the population and in both age groups mainly due to a reduction in diarrhea- and malnutrition-related mortality. Children under five and the population five and over show a similar risk profile. Higher under-five mortality during emergencies reflects higher mortality in stable situations. Thus, also in South Sudan, the empirical evidence does not support the under-five cut-off in emergencies. The lack of disaggregation of mortality and demographic data in the available surveys prevented us from testing for other possible high risk groups (older children, elderly, women). Age- and sex-specific disaggregation of mortality data in emergencies is needed – but the choice of age categories is crucial.