Comparability of the quality of verbal autopsy data from telephonic interviews versus face-to-face interviews in Agincourt, rural northeast South Africa

Abstract

Background The Medical Research Council (MRC) and University of the Witwatersrand (WITS) Rural Public Health and Health Transitions Research Unit have relied on face-to-face interviews to collect verbal autopsy data since 1992. In March 2020, all in-person interviews conducted by the Wits Agincourt Research Unit were suspended and replaced with telephonic interviews due to the COVID-19 pandemic. Mortality data has a significant role in measuring the health status of a population as well as informing health policy, programme implementation and evaluation. It is important that quality data is collected when monitoring mortality indicators to ensure accurate and complete information is available for health analysts. Literature argues that the mode of data collection contributes to the quality of data and some studies have shown that data collected using telephonic interviews are inferior to those collected using face-to-face interviews. The effect on the quality of data induced by the change in method in 2020 to collect verbal autopsy data by the MRC/Wits Agincourt Research Unit has not been assessed. Aim The purpose of this study was to assess if the quality of verbal autopsy data collected by the MRC/Wits Agincourt Research Unit through telephonic interviews during the COVID-19 pandemic was comparable to the verbal autopsy data collected through face-to-face interviews prior to the pandemic and after. The study also aimed to assess the reliability of verbal autopsy data that was collected telephonically. Methods This was a descriptive research study that used data profiling techniques to assess the accuracy, completeness and consistency of verbal autopsy data gathered from the Agincourt Health and Socio-demographic Surveillance System population during the periods 1 April 2020-31 March 2021 using telephonic interviews and 1 April 2019-31 March 2020 and 1 April 2021-31 March 2022 through face-to-face interviews. The Kruskal-Wallis H Test was used to measure if the data quality results from the three datasets were statistically comparable. Results The verbal autopsy data that was collected via telephonic interviews during the pandemic and the verbal autopsy data that was collected after the pandemic via face-to-face interviews had vi similar overall data quality score of 93.8%. The data that was collected before the pandemic had an overall data quality score of 91.6%. At 95% confidence interval the Kruskal Wallis test result (p-value = 0. 89) showed that there was no difference in the mean rank quality assessment results between data collected face-to-face before and after the COVID-19 pandemic and that collected telephonically during the pandemic. Conclusion Our results show that the verbal autopsy data was comparable and there was not much variation in the quality of data collected via telephonic interviews versus that collected via face-to-face interviews in all three data quality dimensions that were measured. Where the results varied slightly, the difference was not statistically significant. Verbal autopsy data collected telephonically is reliable and can be used as a method of data collection by researchers.