On 19 October 2016 the RSS Medical and Statistics & the Law Sections held a joint afternoon of talks at The University of Manchester, about analysing data on causes of death, and issues to consider.
Michael Singleton, coroner for the Blackburn, Hyndburn & Ribble Valley region, started off the afternoon with his perspective on determining whether a death has been caused by suicide, for the purpose of registration. Michael described the Coroners & Justice Act and its influence on suicide verdicts, including ‘standard of proof’. However, he then provided several examples whereby even standard of proof is open to interpretation by the coroner at the time. For example, suicide notes are often insufficient for such a verdict. Michael finished with concluding that suicides will be under-reported from coroners’ verdicts and so analyses based on these verdicts alone will be unreliable.
Next, Dr Peter Sidebotham, reader in child health at the University of Warwick, provided evidence to a popular statistic that has been floated since the 1970’s: that one to two children per week die through causes related to child abuse or neglect. From combining Office for National Statistics (ONS) death registrations, police-recorded homicides, Serious Case Reviews (SCRs), and Child Death Overview data Peter found that in the 1970’s, the incidence of deaths through assault, unknown causes or where the intent was undetermined, was indeed around two per week among infants (children less than one year’s old), but was much lower in 1-19 year olds (and now it is closer to around one per week). These deaths amongst infants were most likely to be captured from SCRs, whereas among adolescents they were most likely to be captured from death registrations.
Tim Millar, reader from the University of Manchester Centre for Mental Health and Safety, described his work in trends of deaths due to illicit drug use and the impacts of different treatments for illicit drug users, using data from the English National Drug Monitoring Service linked to ONS death registrations. He found that although young men are more likely to die from overdose than women, this gap narrows during middle to older age. Among identified illicit drug users, risks of deaths through overdose were higher for those not enrolled on any treatment programme when compared with those on opioid therapy, and also when compared with those on psychological therapy alone.
Finally, Dr Ben Windsor-Shellard, head of mortality analysis at the ONS, gave us a whistle-stop tour of how data on causes of death are processed. For ~80% of deaths a medical doctor can certify it (and this must be done within five days of death), however for the remaining cases a coroner needs to be consulted or a post-mortem requested. Delays in registration (most likely for those due to external pregnancy-related causes) can affect analyses of mortality data. Several contributing causes can be recorded on a death certificate, but only one underlying cause may be selected, and there are strict rules for this. Unfortunately, analyses involving the underlying cause may be affected by changes over time in these rules, in coding (e.g. removal/addition of certain codes), or death certificate reform (which is likely to lead to more information being recorded). Comparability ratios are made available by ONS and the new ‘IRIS’ software attempts to establish some consistency over time.
We had a varied and clearly stimulating afternoon of talks. This was evidenced by a packed lecture theatre and some great questions from the audience. Thanks to everyone who came along!