Is scaling-up community maternal verbal autopsies a viable approach to inform action to reduce maternal mortality?

Key messages

  • Community verbal autopsies are recommended in maternal death surveillance and response systems. Potentially they can inform action to prevent further deaths by improving our understanding of: The circumstances surrounding deaths occurring in the community; individual and community-level factors contributing to delays for those deaths occurring at facilities; and family perspectives of treatment received at facilities. However, undertaking verbal autopsies at scale may not always be feasible.
  • Maternal deaths are rare events and, particularly in settings that are experiencing rapid increases in institutional delivery rates, are increasingly occurring in facilities.
  • Community verbal autopsies entail a high training cost, with large numbers of community health workers each covering relatively small geographical areas. Their low skill, high turnover and infrequency of conducting verbal autopsies generate poor quality information and limited new insights. The collection of data with limited use at a sensitive time also raises ethical concerns.
  • In low resource settings, other strategies could be considered to achieve the primary purposes of maternal death surveillance and response, for example strengthening community based vital registration systems for better notification of deaths; strengthening facility-based maternal death reviews with a focus on the community delays contributing to facility-based deaths; and intermittent qualitative research by skilled researchers.


The primary goal of maternal death surveillance and response (MDSR) is to eliminate preventable maternal mortality by obtaining and strategically using information to guide public health actions and monitoring their impact. The overall objectives of MDSR are to provide information that effectively guides immediate as well as longer term actions to reduce maternal mortality; and to count every maternal death, permitting an assessment of the true magnitude of maternal mortality and the impact of actions to reduce it[1]. The MDSR process typically recommends community verbal autopsies, as well as facility based maternal death reviews. However, my experience of working with facility based maternal death reviews and community verbal autopsies in Kenya, Nepal, Bangladesh and India, both within standalone research projects and the government health system, has led me to wonder: Are community verbal autopsies at scale an appropriate use of funds in a low resource setting to achieve the MDSR goal and objectives?

Factors to consider

VA expert opinion_MDSR Sept 2017 newsletter

Before implementing verbal autopsies at scale, the following factors should be considered:

  • Proportion of maternal deaths occurring in the community: Many contexts are witnessing an increase in institutional deliveries, particularly where maternity user fees have been removed and/or incentive payments have been provided to clients accessing services. As a result, a lower percentage of maternal deaths occur in the community.
  • Cost: Typically, community health workers/volunteers are trained to undertake community verbal autopsies at scale. Given that they each cover relatively small geographical areas, a large number of community health workers require training, resulting in high training costs. Furthermore, there is frequently high turnover within this cadre, resulting in areas either not having someone trained to conduct verbal autopsies, or the additional cost of training new recruits.
  • Rare events: Maternal deaths are a rare event, especially in a small geographical area. This means that many of those trained will never actually conduct a verbal autopsy, and for those who do the time lag since training may mean that much of what was learnt in training has been forgotten.
  • Technical skills: Community health workers assigned to the task of conducting verbal autopsies at scale, are often low skilled. As a result, the information collected through verbal autopsies is often too weak to enable a cause of death to be assigned, or to provide a clear understanding of the events leading up to the death and contributing factors, thus failing in its purpose.
  • Sensitive time: Verbal autopsies should ideally be conducted as soon as possible after the event to obtain as much accurate information as possible. However, a maternal death is a sensitive time for relatives, and when relying on a large pool of community workers to conduct the interviews, often after a long time lag since training, it is hard to ensure these are undertaken in a sensitive manner.
  • Ethics: The sensitive time, and the fact that the information may be too weak to use for the means it was designed, raises questions as to whether it is ethical to collect verbal autopsy data at scale.
  • Meeting MDSR goal and objectives: Given the above factors it should be considered whether verbal autopsies are the most effective means for meeting the desired goal, namely to prevent maternal deaths. There may already be a good understanding of the barriers preventing access to health services, or it may be feasible to collect this information through lower resource and quicker qualitative data collection, and instead channel resources assigned to verbal autopsies to helping communities overcome these barriers.
  • Monitoring: One of the objectives of MDSR is to count every maternal death, including those that occur in the community. This can still be done through a strong notification system that does not need to rely on verbal autopsies.

Suggested way forward

In a context with limited resources, the following approaches could be considered to obtain community based information to guide public health actions to reduce maternal mortality and monitor their impact:

  • Count every maternal death: Strengthen the identification and notification of maternal deaths in the community to enable the true magnitude of maternal deaths to be measured. Where feasible this should be done alongside the strengthening of the vital registration system, rather than as a parallel duplicate data collection system.
  • Strengthen facility-based maternal death reviews: With a higher percentage of deaths occurring in facilities it is important to ensure all facility deaths are reviewed within the recommended timeframe. Committees should meet regularly to identify relevant action points and to ensure that these action points are followed up.
  • Strengthen documentation of community delays contributing to facility-based deaths: Facility based maternal death reviews should document as much detail as possible regarding any problems occurring at the facility level, however, they are also an opportunity to document community-based delays, or inappropriate treatment, which may have contributed to the deaths. Admittedly this is only relevant to deaths that occur at facilities, and those conducting the reviews will be limited in regard to the amount and accuracy, of information they have access to regarding what happened prior to arrival. However, they can still provide a valuable insight into delays in the decision to seek formal healthcare and in accessing formal care. These can feed into a separate community-related action plan and be fed back to community workers to take action. It is a feasible way of focusing on the community, without the need for additional resources.
  • Intermittent in-depth qualitative research, including verbal autopsies by trained researchers: The challenges faced by the community could be captured through other, cheaper, means, for example focus group discussions on the barriers preventing utilisation of health services. Community verbal autopsies could be conducted intermittently by trained researchers covering larger geographically areas, ideally in combination with a strong community notification system. This was done successfully within the Nepal Maternal Mortality and Morbidity Study 2008/09, using a prospective surveillance system for one year and covering a population of 3.2 million.

This expert opinion piece was written by DrĀ Sarah Barnett, Technical Specialist at Options.


[1] World Health Organization. (2013). Maternal death surveillance and response: Technical guidance. Information for action to prevent maternal death. Geneva: WHO.