The Limits of the Verbal Autopsy for Cause of Death Determination

Abstract

Background: Accurate information on causes of death (CoD) is essential to reduce mortality. In low-income settings, data on CoD is obtained using imprecise methods and knowledge of what people die of is based on assumptions, preventing to achieve full effectiveness of health interventions, monitor their impact, and thus targeting effective policies to reduce mortality. Methods: We have validated a computer-coded verbal autopsy method -the InterVA-4- using individual and population metrics to determine CoD against the gold standard, the complete diagnostic autopsy (CDA), in 316 deceased patients of different age groups who died in a tertiary-level hospital in Maputo, Mozambique between 2013 and 2015. Findings: We found a low agreement of the InterVA-4 model across all age groups at the individual (Kappa statistics ranging from -0·030 to 0·232, lowest in stillbirths and highest in adults) and population levels (chance-corrected cause-specific mortality fraction accuracy ranging from -1·00 to 0·62, lowest in stillbirths, highest in children). The sensitivity of the InterVA-4 model in identifying infectious diseases was low for all infectious diseases (0% for tuberculosis, diarrhea, and disseminated infections, 32% for HIV-related infections, 33% for malaria and 36% for pneumonia). Twenty-six maternal deaths were assigned to eclampsia but only four patients were actually eclampsia. Interpretation: The agreement of the InterVA-4 model against the CDA was low at the individual and at the population level. These findings are key to revise current tools used to establish CoD in high-burden mortality areas, and will help in improving VA techniques and developing more precise methods. Funding Statement: This work has been funded through grants from the Bill and Melinda Gates Foundation and the Spanish Ministry of Health. Declaration of Interests: The authors do not have any conflict of interest. Ethics Approval Statement: This study received approval by the Clinical Research Ethics Committee of the Hospital Clinic of Barcelona, Spain (File 2013/8677) and the National Bioethics Committee of Mozambique (Ref. 342/CNBS/13).