Performance of four computer-coded verbal autopsy methods for cause of death assignment compared with physician coding on 24,000 deaths in low- and middle-income countries

Table 4 Partial chance-corrected concordance of computer-coded verbal autopsy methods versus physician-coded verbal autopsy reference standards

Dataset	Test cases	Open-source random forest		Open-source tariff method		InterVA-4		Average for top cause (%)
		Top (%)	Top 3 (%)	Top (%)	Top 3 (%)	Top (%)	Top 3 (%)
China	400	33	55	32	64	N/A	N/A	33
Institute for Health Metrics and Evaluation	400	31	54	32	48	N/A	N/A	32
Million Death Study	6100	55	81	48	70	38^a	60^a	47
Agincourt	2900	42	75	38	62	39	56	40
Matlab	1600	45	72	48	68	45	59	46
Average		41	67	40	62	41	58

Top cause represents accuracy of the CCVA method’s most probable cause matching the cause assigned by PCVA; Top 3 represents whether CCVA’s three most probable causes contain the cause assigned by PCVA. Averages calculated across CCVA methods only use results for the top cause. ^aThe Million Death Study dataset used for InterVA-4 contained a sample of 552 cases, in which we attempted to extract additional InterVA-4 indicators from the narratives.

ISSN: 1741-7015