Abstract
BACKGROUND: Vital data on births, deaths, and causes of death are primarily captured by civil registration and vital statistics systems, which are vastly lacking or inadequately operating in resource-constrained settings. Out-of-health facility deaths remain prevalent and often pass without adequate medical certification, leading to gaps in understanding the medical, social, logistical, and health system circumstances contributing to these deaths. Verbal autopsy (VA), as a standardized and scalable method, is well designed to fill gaps by eliciting causes of death information at the population level. Circumstances Of Mortality CATegories (COMCAT) is a newly integrated concept within VA to identify and quantify likely circumstantial factors contributing to deaths, mainly from social and health system aspects. COMCAT, previously tested in South Africa and Saudi Arabia, show potential to systematically categorize circumstances of death at a population scale. This study intends to verify the process of COMCAT by assessing its plausibility and relevance in generating sensible applications in Indonesia. METHODS: InterVA-5, a machine learning-based VA tool, was used for processing and interpreting medical and non-medical (COMCAT) causes of death for data collected between 2017 and 2021 in the Sleman Health and Demographic Surveillance System. Cause-specific mortality fractions and the corresponding COMCAT categories were derived for each cause of death. RESULTS: Lack of recognition of the illness severity among families attributed mostly to deaths in the Sleman region. The proportions and ranks of each COMCAT were consistent with the known health information of the population in the setting, which speaks to the plausibility of these findings. The novel geo-mapping component of this tool application adds value to monitoring "hot spots" and their social and health system determinants. CONCLUSION: Geo-temporal COMCAT information shows sensible applications of the utility of the VA by producing plausible circumstantial information on population mortality in time and space.