This synthesis brings together findings on cause-specific mortality documented by means of verbal autopsies applied to over 110,000 deaths across Africa and Asia, within INDEPTH Network sites.

Methods: Developments in computerised methods to assign causes of death on the basis of data from verbal autopsy (VA) interviews have made possible these standardised analyses of over 110,000 deaths from 22 African and Asian Health and Demographic Surveillance System sites in the INDEPTH Network. In addition to previous validations of the InterVA-4 probabilistic model, these wide-ranging analyses provide further evidence of the applicability of this approach to assigning the cause of death. Plausible comparisons with existing knowledge of disease patterns, as well as substantial correlations with out-of-model parameters such as time period, country, and other independent data sources were observed.

Findings: Substantial variations in mortality between sites, and in some cases within countries, were observed. A number of the mortality burdens revealed clearly constitute grounds for public health actions. At an overall level, these included high maternal and neonatal mortality rates. More specific examples were childhood drowning in Bangladesh and homicide among adult males in eastern and southern Africa. Mortality from non-communicable diseases, particularly in younger adulthood, is an emerging cause for concern. INDEPTH's approach of documenting all deaths in particular populations, and successfully assigning causes to the majority, is important for formulating health policies.

Future directions: The pooled dataset underlying these analyses is available at the INDEPTH Data Repository for further analysis. INDEPTH will continue to fill cause-specific mortality knowledge gaps across Africa and Asia, which will also serve as a baseline for post-2015 development goals. The more widespread use of similar VA methods within routine civil registration systems is likely to become an important medium-term strategy in many countries.