Have you ever considered the accuracy and reliability of hospital death certification? Prior to my recent research project, I certainly hadn’t. Accurate certification is not only important for legalities sake, or for families in understanding the cause of death of their relative, but it is pivotal in determining public health initiatives and health service provision allocation. Inaccuracies in cause of death documentation could therefore potentially result in over or under-provision of resources in certain areas. There are publications reporting error rates of up to 40% in cause of death documentation, as well as audits concluding that simple educational measures improve accurate certification by at least 13%.

As part of my research, I was tasked to independently validate the cause of death of a number of deceased patients with a fellow doctor by reviewing patient records, and formulate our own individual opinions on death causation. The outcome of this task was interesting for two reasons. Firstly, it highlighted gaps in my knowledge regarding accurate completion of death certification, and secondly, it unearthed repeated errors in recorded certificates, when reviewed for comparison.

The main point I learnt was the importance in getting to grips with the underlying cause of death: the initiator in the chain of events that led directly to death. With that in mind, the death certificate should tell a story. The patient’s initiating event may have been a cerebrovascular accident (CVA), but as a result of the ensuing immobility, they may have died of pneumonia. Frequently, when reviewing recorded death certificates, cases like this would have had “1a” documented as “pneumonia” with no underlying cause. When considering coding and resource allocation, it is vital that the underlying cause (the CVA) be mentioned on the lowest line of part 1.

Another learning point was to avoid describing terminal events or modes of dying. Many official death certificates used phrases like “cardiorespiratory arrest” as the final cause of death (1a). Everybody ultimately dies of cardiorespiratory arrest so in describing the cause of death, that phrase becomes redundant. Rather, the underlying cause of the arrest should be detailed.

Lastly, organ failure alone was frequently documented as a cause of death. As organ failure can be due to an unnatural cause, such as poisoning, it is imperative that the underlying cause for the organ failure be documented. For example, the final cause of death may be cardiac failure, but if the failure is underpinned by ischaemic cardiomyopathy, which in turn was caused by myocardial infarction, these must be documented as 1b and 1c respectively.

Aside from these errors highlighted, it is important to consider why they were made in the first place. Junior doctors are often responsible for completing hospital death certificates and, having finished the foundation programme last year, I was alarmed that some of this information was new to me. It is part of the undergraduate syllabus to teach medical students how to fill out death certificates but, in my experience, no further teaching was offered as a foundation doctor. As publications have shown a reduction in certification error after educational intervention, perhaps completion of death certification training should be additionally offered to junior doctors. The ONS have published clear and thorough guidance on completing medical certificates; I recommend doctors refresh their memories on death certification by reading these.

I declare that that I have read and understood the BMJ Group policy on declaration of interests and I have no relevant interests to declare.

Rebecca Mills is about to commence core medical training in London and has been undertaking research in cardiology for the past year.