Research blog: Systematic biases in death certification - a job for the Medical Examiner?

Katherine Sleeman is an NIHR Clinician Scientist and Honorary Consultant in Palliative Medicine at the Cicely Saunders Institute, King’s College London. In this blog she discusses her recent paper on death certification in dementia, published in the journal Age and Ageing, which was co-authored by NIHR Maudsley BRC researchers Dr Gayan Perera and Professor Robert Stewart.  

It is frequently said that there are just two universal certainties: death and taxes. While HMRC is responsible for ensuring that taxes are paid, information about who dies, where, and how, is gathered through death certification.

Dementia is a public health priority of increasing importance. In 2014, it was reported that dementia had overtaken cancer and cardiovascular disease as the most common cause of death for women in England. We have previously shown that the proportion of death certificates in England where dementia was mentioned as a cause of death doubled between 2001 and 2010.

But what is unclear is why dementia deaths appear to be increasing. Is it due to an increasing prevalence of dementia in our ageing society? Due to increased detection of dementia, perhaps? Or does this increase simply represent an improvement in death certification practices over time?

With this question in mind, we set out to investigate death certification in dementia.

We used the NIHR Maudsley Biomedical Research Centre (BRC) Clinical Record Interactive Search (CRIS) to identify people with a clinical diagnosis of dementia. Individual linkage of these clinical records to ONS mortality data determined which patients had died, and defined our cohort. Our primary outcome was whether or not dementia was given as a cause of death on the death certificate.

We included information on 7,115 people who had a clinical diagnosis of dementia and who died over an 8 year period (2006-2013). We found dementia was given as a cause of death in just 53.6% of cases. As expected, dementia was more likely to be written on the death certificate if patients were older, and if they had more severe cognitive impairment. Perhaps less expected, we found that dementia was more likely to be written on the death certificate for people with Alzheimer’s disease compared to other subtypes of dementia.

Death certification of dementia became more likely with each year over the study period, suggesting that at least part of the highly publicized increase in prevalence of dementia determined using mortality data is a result of improved certification practices. Last, we found that the place in which the person died was strongly associated with whether or not dementia was mentioned: people who died in care homes had almost twice the odds of being certified with dementia compared to those who died at home, even when other factors such as age and the degree of cognitive impairment were controlled for.

Reliable information on deaths and their causes is essential to monitor disease burden and trends, to assess public health programmes, to guide policy, and to decide priorities for research. Our study provides evidence that changes in certification practices are likely to have contributed to the increase in prevalence of dementia identified using mortality data. Even so, the burden of dementia measured using mortality data remains a considerable underestimate. We found important biases affecting whether or not dementia is recorded, which may indicate a persisting lack of awareness of dementia as a terminal disease.

Concerns over variable and poor quality death certification have led to the proposed introduction of Medical Examiners who will independently ascertain information on causes of death for every person who dies in England and Wales. It is likely that systematic biases in death certification occur not just in dementia, but in other pathologies too. Identifying and correcting these will be an essential role of the Medical Examiner.

You can respond to the consultation on death certification reforms until 15th June 2016.

This blog was commissioned and originally published by the British Geriatrics Society, on the British Geriatrics Society blog.


Tags: CRIS - Informatics - Clinical and population informatics - Publications -

By NIHR Maudsley BRC at 12 May 2016, 11:43 AM


Back to Blog List