Much effort is put into determining why people die - which usually means determining the "medical cause of death" - or, to put it simply, what was the disease process or event leading to the death. In some parts of the world, this has been done routinely for hundreds of years, but in other places deaths are still not routinely followed up.
But maybe the medical cause is not the only item of interest after someone dies? Irrespective of what the disease process might be, it's also interesting to understand which deaths occur because health systems don't work as well as they should do, or after emergencies that happen in circumstances where no help is available.
I've previously written about tools from the World Health Organization for undertaking verbal autopsies - interview procedures for following up deaths. The material from these interviews is often processed using computer models like InterVA-5 (downloadable from www.interva.net), with the main objective of assigning medical causes of death to each case.
But InterVA-5 also includes a new feature to derive circumstances of mortality categories (COMCATs) for each death, without any additional input or cost. These categories are:
Traditional practices or beliefs influenced health seeking behaviour and the pathway to death
Sudden, urgent or unexpected conditions leading to death, which probably precluded life-saving actions
Lack of recognition or awareness of serious disease (e.g. symptoms or severity) negatively influenced health seeking behaviour
Inability to mobilise and use resources (e.g. material, transport, financial) hindered access to health care
Problems in getting health care despite accessing health facilities (e.g. related to admissions, treatments and medications)
Death occurred in circumstances that could not reasonably have been averted (e.g. very elderly or recognised terminal conditions)
A combination of the above categories affected the pathway to death; no single factor predominated
We've tried this out on a dataset from the Agincourt project in rural South Africa. All the details of the COMCATs from Agincourt are described here.
This chart shows the basic ideas - on the left hand side, the basic groups of medical causes of death are shown for each age group from 2012 to 2016. On the right hand side, for the same deaths, the COMCATs are shown. For example, for children who died, emergency circumstances often applied. In middle age, lack of resources or lack of recognition of the problem often led to death. And for the elderly, an appreciable proportion of deaths were inevitable - we all have to die some time! But if you were responsible for health services in the Agincourt area, there are also positive messages. For the under-5s, in addition to mortality rates falling, deaths following a failure to recognise the seriousness of a child's condition also fell over time.