Evaluating InterVA for Cause of Death Attribution
Author Information
Author(s): Rafael Lozano, Michael K. Freeman, Spencer L. James, Benjamin Campbell, Alan D. Lopez, Abraham D. Flaxman, Christopher J.L. Murray
Primary Institution: Population Health Metrics Research Consortium (PHMRC)
Hypothesis
How accurately does InterVA assign causes of death compared to physician-certified verbal autopsies?
Conclusion
InterVA performs worse than physician-certified verbal autopsies in assigning causes of death.
Supporting Evidence
- InterVA achieved a chance-corrected concordance of 24.2% for adults.
- InterVA's cause-specific mortality fraction accuracy was 0.546 for adults.
- InterVA performed worse for neonates, with a chance-corrected concordance of 6.3%.
- InterVA's performance was compared to physician-certified verbal autopsies (PCVA) across multiple datasets.
Takeaway
InterVA is a tool that helps figure out why people died, but it doesn't do as good a job as doctors looking at the same information.
Methodology
The study used clinical diagnostic gold standards to assess the performance of InterVA on 12,542 verbal autopsy cases across different age groups.
Potential Biases
InterVA may overpredict or underpredict certain causes based on the input data and its design limitations.
Limitations
InterVA was challenged to predict causes it was not built to identify, and some items were not mapped to the PHMRC survey.
Participant Demographics
The study included adult, child, and neonatal deaths from multiple countries.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website