Verbal autopsies for adult deaths: issues in their development and validation.

BACKGROUND: The verbal autopsy appears to be an attractive technique for the assessment of causes of adult death in settings where the proportion of people who die while under medical care is low. Verbal autopsies have been used extensively to assess causes of childhood deaths. The existing experience in verbal autopsies for adult deaths is limited mainly to maternal deaths. This paper aims to highlight the critical issues in verbal autopsies to assess causes of adult death which need further research.
METHODS: This paper reviews the methods of verbal autopsy used in 35 published studies and discusses issues in the development of verbal autopsies, including mortality classification, design of questionnaires, interviewers, respondents, recall periods, procedures for deriving a diagnosis and the recording of single versus multiple causes of death. It also discusses issues in the validation of verbal autopsies, including the choice of reference diagnosis and the required sample size.
RESULTS: The methodological approaches used in verbal autopsy studies have varied widely. Very few studies of the validity of verbal autopsies have been conducted. In these studies, the reported sensitivity and specificity varied widely between different causes of death and between studies.
CONCLUSIONS: The information available is inadequate to draw firm conclusions on preferred methodological approaches for verbal autopsies for adult deaths. Before these tools are used more widely for adult deaths, further research is required to compare alternative methods and to evaluate the validity of this tool in a range of settings.