Measuring AIDS Mortality with Verbal Autopsy
Author Information
Author(s): Lopman Ben A, Barnabas Ruanne V, Boerma J. Ties, Chawira Godwin, Gaitskell Kezia, Harrop Tara, Mason Peter, Donnelly Christl A, Garnett Geoff P, Nyamukapa Constance, Gregson Simon
Primary Institution: Imperial College London
Hypothesis
Can verbal autopsy effectively classify deaths as AIDS-related in a rural Zimbabwean population?
Conclusion
The study found that 69% of deaths were correctly classified as AIDS/non-AIDS using verbal autopsy data.
Supporting Evidence
- The algorithm achieved a sensitivity of 66% and specificity of 76% in the test dataset.
- Eight predictors were identified as significant for classifying AIDS deaths.
- The study used a gold standard of HIV status to validate the verbal autopsy method.
Takeaway
Researchers asked family members about symptoms of people who died to figure out if their death was due to AIDS, and they got it right 69% of the time.
Methodology
A verbal autopsy questionnaire was administered to caregivers of adults who died, and an algorithm was created to classify deaths based on reported symptoms.
Potential Biases
Potential bias due to reliance on caregiver recall and the social stigma surrounding AIDS.
Limitations
The study may not generalize to populations with different symptom profiles or healthcare access.
Participant Demographics
Participants were adults from rural Zimbabwe, with known HIV status.
Statistical Information
P-Value
0.05
Confidence Interval
95% CI, 65%–77%
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website