Using verbal autopsy to measure causes of death: the comparative performance of existing methods
- Equal contributors
1 Institute for Health Metrics and Evaluation, University of Washington, 2301 5th Avenue Suite 600, Seattle, WA 98121, USA
2 National Institute of Public Health, Universidad 655, 62100 Cuernavaca, Morelos, Mexico
3 Johns Hopkins University, Bloomberg School of Public Health, 615 N Wolfe St #5041, Baltimore, MD 21205, USA
4 Public Health Laboratory-IdC, P.O. BOX 122 Wawi Chake Chake Pemba, Zanzibar, Tanzania
5 Public Health Foundation of India, ISID Campus, 4 Institutional Area, Vasant Kunj, New Delhi 110070, India
6 Brigham and Women's Hospital, 75 Francis St, Boston, MA 02215, USA
7 Global Development, Bill and Melinda Gates Foundation, PO Box 23350, Seattle, WA 98012, USA
8 CSM Medical University, Shah Mina Road, Chowk, Lucknow, Uttar Pradesh 226003, India
9 Dept of International Health, Johns Hopkins Bloomberg School of Public Health, E5521, 615 N. Wolfe Street, Baltimore, MD 21205, USA
10 Public Health Laboratory-Ivo de Carneri, Wawi, Chake-Chake, Pemba, Zanzibar, Tanzania
11 Johns Hopkins University, 214A Basement, Vinobapuri Lajpat Nagar-II, New Delhi 110024, India
12 Harvard School of Public Health, 677 Huntington Avenue, Boston, MA 02115-6018, USA
13 The George Institute for Global Health, The University of Sydney, 83/117 Missenden Rd, Camperdown, NSW 2050, Australia
14 Community Empowerment Lab, Shivgarh, India
15 Research Institute for Tropical Medicine, Corporate Ave, Muntinlupa City 1781, Philippines
16 Division of Nutritional Sciences, Cornell University, 314 Savage Hall, Ithaca, NY 14853, USA
17 The George Institute for Global Health, 839C, Road No. 44A, Jubilee Hills, Hyderabad 500033, India
18 Muhimbili University of Health and Allied Sciences, United Nations Rd, Dar es Salaam, Tanzania
19 School of Population Health, University of Queensland, Level 2 Public Health Building School of Population Health, Herston Road, Herston, QLD 4006, Australia
20 University of Melbourne School of Population and Global Health, Building 379, 207 Bouverie St., Parkville 3010, VIC, Australia
BMC Medicine 2014, 12:5 doi:10.1186/1741-7015-12-5Published: 9 January 2014
Monitoring progress with disease and injury reduction in many populations will require widespread use of verbal autopsy (VA). Multiple methods have been developed for assigning cause of death from a VA but their application is restricted by uncertainty about their reliability.
We investigated the validity of five automated VA methods for assigning cause of death: InterVA-4, Random Forest (RF), Simplified Symptom Pattern (SSP), Tariff method (Tariff), and King-Lu (KL), in addition to physician review of VA forms (PCVA), based on 12,535 cases from diverse populations for which the true cause of death had been reliably established. For adults, children, neonates and stillbirths, performance was assessed separately for individuals using sensitivity, specificity, Kappa, and chance-corrected concordance (CCC) and for populations using cause specific mortality fraction (CSMF) accuracy, with and without additional diagnostic information from prior contact with health services. A total of 500 train-test splits were used to ensure that results are robust to variation in the underlying cause of death distribution.
Three automated diagnostic methods, Tariff, SSP, and RF, but not InterVA-4, performed better than physician review in all age groups, study sites, and for the majority of causes of death studied. For adults, CSMF accuracy ranged from 0.764 to 0.770, compared with 0.680 for PCVA and 0.625 for InterVA; CCC varied from 49.2% to 54.1%, compared with 42.2% for PCVA, and 23.8% for InterVA. For children, CSMF accuracy was 0.783 for Tariff, 0.678 for PCVA, and 0.520 for InterVA; CCC was 52.5% for Tariff, 44.5% for PCVA, and 30.3% for InterVA. For neonates, CSMF accuracy was 0.817 for Tariff, 0.719 for PCVA, and 0.629 for InterVA; CCC varied from 47.3% to 50.3% for the three automated methods, 29.3% for PCVA, and 19.4% for InterVA. The method with the highest sensitivity for a specific cause varied by cause.
Physician review of verbal autopsy questionnaires is less accurate than automated methods in determining both individual and population causes of death. Overall, Tariff performs as well or better than other methods and should be widely applied in routine mortality surveillance systems with poor cause of death certification practices.