In settings without medical certification of causes of death, verbal autopsy (VA) interviews are the most important data for cause-specific mortality analyses. In work led by a team at the University of Washington, we have developed a new statistical tool –known as InSilicoVA– for interpreting data from the standard WHO VA instrument.
Novel to InSilicoVA is that it uses a Bayesian framework to characterise uncertainty around individual causes of death as well as the distribution of causes of death at the population level. InSilicoVA has been shown to perform well compared with existing methods, including InterVA-4, and is now available in an R-package. Work is underway in the ALPHA network to quantify the validity of InSilicoVA in identifying HIV/AIDs-related deaths.