Globally, most deaths occur at home or other locations outside of health facilities, and often go unreported. This is particularly the case in countries with underdeveloped and under resourced civil registration and vital statistics (CRVS) systems. In many countries, people who die outside of health facilities are also experiencing structural inequalities and have poorer access overall to essential health services.
Estimating causes of death
Women’s deaths disproportionately go uncounted, particularly maternal deaths (deaths due to pregnancy and childbirth) and deaths resulting from gender-based violence. Counting and registering these deaths and their causes is important for measuring and addressing inequities in how people live and die. Verbal autopsy is one way to capture vital statistics data on the distribution and causes of deaths that occur outside of health facilities and can provide evidence and information needed to address inequities in the availability, accessibility, performance, and quality of essential preventive and treatment services.
Verbal autopsy is a method of capturing information about deaths that occur at home, or otherwise outside of health facilities. It is a structured interview with the caregivers of the deceased, and/or with others who are knowledgeable about the circumstances leading to the death. It can be used to determine the most likely cause of death. While a verbal autopsy does not replace a medically certified cause of death by a physician, it can help determine the likely cause of death of an individual and inform population-level estimates of the direct and contributing causes of death (as well as what Dr. Don De Savigny has termed the “causes of the causes of death”).
While verbal autopsy is an important tool for increasing equity in mortality data, its accuracy is not without gendered biases and other limitations. The accuracy of assigning cause of death may be subject to various forms of interviewer bias that disproportionately affect women and sub-populations that are marginalized or at higher risk in a given society or culture.
Minimizing bias
There are things that can be done in the verbal autopsy process to minimize bias. Here are several examples of why and how the validation and use of automated systems for analysis and interpretation of interview data obtained from the use of the 2016 and updated 2022 WHO standardized verbal autopsy instruments may help to reduce bias due to missing or misattributing symptoms and signs of the true direct or contributing causes of death:
- Unconscious bias may lead to misattributing symptoms of disease that manifest differently in men and women. Although cardiovascular disease and heart attacks are a leading cause of death among women, in a clinical setting a physician may unconsciously be biased towards diagnosing men with a heart attack while the same physician might misdiagnose similar symptoms and signs in a woman, and women’s deaths by heart attack may be often misattributed to other causes by physicians in a clinical setting.
- If a verbal autopsy is done correctly in a home or in other settings outside of health facilities, it may counteract the unconscious bias that can be present in the clinical setting, because the verbal autopsy interviewers collect in a standardized manner only the information that is reported by the respondent. They are trained to follow a standard interview with a standard set of questions. With proper training, the standardized questionnaire forces the interviewer to avoid bias in their line of questioning, which physicians may not do.
- The training that verbal autopsy interviewers receive is meant to reduce the chance of bias that might otherwise occur while asking standardized questions about sensitive issues such as gender-based violence. The use of verbal autopsy is generally known to provide more accurate information on injurious causes of deaths compared to information obtained by medically certified cause of death. This is because of the greater level of detail that is collected in the standardized verbal autopsy “injury” questions on the manner of death and circumstances of death compared to what is typically recorded in a medically certified cause of death interview form (either because the medical certification of cause of death form doesn’t collect the detail or because it is not completed properly).
- In cases where the perpetrator of a violent household death might be the deceased’s respondent during a verbal autopsy interview, or if the perpetrator is present during the interview along with another household member, verbal autopsy interviewers may be concerned for their own safety or the safety of others in the household. It may be difficult to get truthful answers from a respondent if that person is the perpetrator of violence or fears retaliatory violence from the perpetrator. Similarly, respondents may be hesitant to share information about deaths due to highly stigmatized causes such as gender-based violence, a fatal overdose due to opioid addiction, or infections with HIV or tuberculosis. The selection, training, and resourcing of verbal autopsy interviewers is important to minimize bias in their conduct of and recording of answers to the standardized WHO verbal autopsy questionnaire.
- To help avoid situations in which the verbal autopsy process could put either respondents or interviewers at risk, there is guidance in the 2022 WHO Verbal Autopsy Toolkit on developing protocols to address risky situations for verbal autopsy interviewers or cases where there is concern for others in the household.
Despite the challenges mentioned above, by ensuring that appropriate protocols and training are in place, verbal autopsies can provide critical evidence and information to measure all-cause mortality and identify demographic and other reasons for differentials in the burden of disease for countries that have under resourced and underdeveloped CRVS systems, with limited availability of medical certification of causes of death. In these countries, population-based analysis of the results of verbal autopsies can be used to improve the design and implementation of health policies and interventions that improve the lives of women, men, girls, and boys.