The Office for National Statistics (ONS) have published an update to their data set looking COVID-19 related deaths in the UK by occupation.
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“This analysis by ONS is an updated version of a previous release from six weeks ago (11 May) that covered much of the same ground. In general, the main patterns in this new release match those in the previous report; differences between men and women in the ways that occupation is related to death rates, quite scarily high death rates for some occupations such as male taxi drivers, chauffeurs and chefs, higher death rates (for both men and women) for social care workers than health care workers. Also, the main limitations of the data are unchanged. ONS adjusted their analysis to allow for differences in age between different groups, but they did not adjust in their main analyses for other factors such as ethnic group, place of residence, or deprivation, that are known from other studies to be associated with death rates from COVID-19. And in any case, these are observational data. There are bound to be many other differences between occupational groups, apart from just the jobs that people do, and these other differences may be the actual causes of differences in death rates, not just the work that people do.
“However, there are several further analyses in this report that were not included in the previous one. ONS have done more work, and they now have more data than they had six weeks ago. They have examined some other factors that might relate to the ways that COVID-19 deaths are associated with occupation. In looking at all of these, we need to remember the overall point that the study is observational and can’t tell us for sure what causes what – but these new analyses give food for thought anyway. Some of them split up people according to the Index of Multiple Deprivation (IMD) of the neighbourhood where they lived. The IMD is an overall measure of deprivation that takes into account ‘factors such as income, employment, health, education, crime, the living environment, and access to housing within an area’. It doesn’t tell you everything, it averages over all these different kinds of deprivation, and it describes the neighbourhood where people live, not the people individually. But it does tell you something. For men, the major group of occupations that had the highest overall death rate are what are known as ‘elementary occupations’. This is a broad term, and the grouping includes a diverse set of occupations – but, roughly speaking, they would generally involve fairly simple and often routine tasks, that might require hand-held tools and often some physical effort. Within this group, the death rates (adjusted for age) for men living in the most deprived neighbourhoods, were much higher than – over double – the rates in the least deprived neighbourhoods. The same is generally true for women working in the group of occupations that had the highest death rates for women, namely caring, leisure and other service occupations. But it is also true for men working in occupations that generally have higher social status and, often, higher incomes – the group known as ‘managers, directors, and other senior officials’. Men in that group have considerably lower death rates from COVID-19 than men in elementary occupations, whatever the IMD measure for the neighbourhood where they live. But it remains the case that, for men in what might be thought of as this group of more privileged and better-off people (on average), the death rates for those living in the most deprived neighbourhoods are about three times as high as the rates in the least deprived neighbourhoods.
“Other analyses of a similar kind looked at the region where people live, and in particular compared London with everywhere else in England and Wales. For men, the death rates in some occupations that have particularly high rates, such as taxi drivers and security guards, have been far higher in London than elsewhere. And analyses that look at ethnicity also show differences. In men, the ONS analysts picked out 17 occupations that had higher death rates than average. These included taxi drivers, security guards and chefs, but also many others that were not so extreme overall such as nurses, nursing assistants, sales assistants, and care workers. Of the 17 occupations, 11 had proportions of people in Black and Asian ethnic groups that are significantly higher than the proportions in the general population. Something similar was observed for women, though the ONS statisticians only had sufficient data to look at a smaller number of occupations – just four. But of these four, two (nurses, and care workers) again had significantly higher proportions of women from Black and Asian ethnic groups than in the overall population. But the data cannot tell us the extent to which these differences between ethnic groups are related to where people live (the region, or how deprived the neighbourhood is), to education, to health, to poverty, or to a whole range of other factors that might well come into the picture.”
Prof Sheila Bird, Formerly Programme Leader, MRC Biostatistics Unit, University of Cambridge, said:
“The Office for National Statistics (ONS) has updated its analysis of COVID-19 deaths by occupation to encompass death-registrations in England and Wales between 9 March and 25 May 2020 for persons aged 20-64 years.
“The analysis published today included 268 COVID-mention deaths among social workers (97 males and 171 females; gender-specific rates of 50 and 19 per 100,000) and 272 COVID-mention deaths among healthcare workers (130 males and 142 females; gender-specific rates of 30 and 11 per 100,000).
“The analysis published today has a major limitation (see page 22): the analysis does not include COVID-mention deaths of social care workers or healthcare workers which have been referred to the coroner for investigation. On 12 May 2020, Scotland’s Lord Advocate advised that all such deaths in Scotland should be referred to procurators fiscal (the Scottish equivalent of England’s coronial system); and that his advice applied retrospectively.
“ONS offers the unhelpful information that the median delay between date of death and the date of death-registration was 4 days for the included COVID-mention deaths. Median tells us nothing about the 95th percentile of the delay distribution for included COVID-mention deaths; and nothing at all about the number of excluded suspect COVID-deaths of health and social care workers which have not yet been registered because they have been referred to the coroner and the coroner’s investigation is not yet complete. Bizarrely, in England and Wales (unlike Scotland), not even the fact-of-those-deaths has to be notified to ONS until the coronial process has been completed.
“Based on Scottish data, about one in six COVID-mention deaths for health and social care workers may have been referred to coroners and so we could be missing around 100 such COVID-mention deaths if coronial investigations are still underway. The 95th percentile for the delay distribution for social care and healthcare workers’ COVID-mention deaths could be insightful about intermediate coronial completion-times.
“I await information, requested in early May 2020, from the Chief Coroner for England about the number of suspect COVID-mention deaths of healthcare or social care workers that were referred to coroners.”
Prof Neil Pearce, Professor of Epidemiology and Biostatistics, London School of Hygiene & Tropical Medicine (LSHTM), said:
“This updated report from ONS strengthens the evidence that, for working age people, COVID-19 is largely an occupational disease. The occupations which are at increased risk are those which involve regular contact with patients or the public. These include security guards, taxi drivers, bus drivers, chefs, and sales and retail workers. All of these occupations should be supplied with appropriate PPE and required to use it – this is a workplace health and safety issue. It is not just health and social care workers who need PPE – it is anyone working with the public.
“As you would expect, there are also increased risks for social care workers. However, the rates are not so high for health care workers. This probably reflects a major problem with the data that ONS has to work with. For this group, many probable Covid-19 deaths are being referred to Coroners. In England and Wales, these deaths are not ‘counted’ until the Coroner has completed their investigation – which can take months or even years. In the meantime, these deaths don’t appear in the official statistics. This problem doesn’t happen in Scotland because act-of-death must by law be registered within 8 days of death having been ascertained. An emergency system should be in place in England and Wales so that the Office for National Statistics is informed immediately about any death referred to coroners so that fact-of-death is registered.”
“Coronavirus (COVID-19) related deaths by occupation, England and Wales: deaths registered up to and including 25 May 2020” was publsihed by the Office for National Statistics on Friday 26 June
All our previous output on this subject can be seen at this weblink:
Prof Kevin McConway: “Prof McConway is a member of the SMC Advisory Committee, but his quote above is in his capacity as a professional statistician.”
Prof Sheila Bird: SMB leads for the Royal Statistical Society on the need for legislation in England, Wales and Northern Ireland to end the late registration of coroner-referred deaths.
None others received.