Evaluating 2016 WHO Verbal Autopsy data

If you are doing Verbal Autopsy surveys using the 2016 WHO Verbal Autopsy instrument, this may be of interest to you. (The Survey Solutions team has an implementation of this questionnaire. I also have a version which handles partial birth and death dates and a French translation).

For a long time InterVA has been kind of the standard algorithm to determine the cause of death from these VA data. This seems to be no longer so. The Institute for Health Metrics and Evaluation (IHME) at the University of Washington has developed SmartVA as an alternative because of serious deficiencies of InterVA.

SmartVA is based on the PHMRC Shortened Questionnaire, a shortened version of the 2016 WHO VA questionnaire. While many questions in the Shortened Questionnaire have an exact equivalent in the 2016 WHO VA questionnaire, they use different variable names and some variables have to be derived from multiple variables in the original questionnaire.

There are also some tricks to improve SmartVA performance significantly (e.g. by including keywords from the Open Narrative section).

In short:

If you have VA data from the 2016 WHO Verbal Autopsy questionnaire and want to use SmartVA for cause of death determination, you’ll face a complex task of preparing the correct inputs. I have gone through this and can offer you assistance if needed.

We are working on a paper at the HIGH Institute of Global Health at the University of Heidelberg, comparing InterVA and SmartVA and describing the challenges of implementing SmartVA.

This paper is also going to show the use of the Verbal Autopsy results as part of the calculation of Disability-Adjusted Life Years (DALY).