select search filters
briefings
roundups & rapid reactions
before the headlines
Fiona fox's blog
The Office for National Statistics (ONS) has released new data and analysis looking at deaths related to COVID-19 and occupation.
Prof Sheila Bird, Formerly Programme Leader, MRC Biostatistics Unit, University of Cambridge, said:
“Yesterday’s leading Statistical Bulletin by the Office for National Statistics (ONS) concerned COVID-19 related deaths by occupation in England and Wales. The deaths analysed were registered up to and including 20 April 2020. Irrespective of cause, 82% of male deaths could be related to occupation; but only 60% of female COVID-19 deaths versus 68% for all-cause mortality.
“My focus in reading these data by gender and occupation has been first on the percentage of all-cause mortality that COVID-19 has contributed because an occupational group’s all-cause mortality-rate* reflects the intrinsic or baseline riskiness of the occupational group. Second, sadly, the lethality of COVID-19 is greater for men than for women and so a sensible plan, when looking for any additional-COVID-risk signals or COVID-prevented-signals per occupational group, is to focus first on the data for males and to seek corroboration for females who share the same occupation.
“Tables illustrates this approach for occupational groups 22 (health professionals) and 61 (caring personal service occupations) combined versus other occupations; and for the predominantly male combined occupational groups 52 (skilled metal, electrical and electronic trades) and 53 (skilled construction and building trades) versus 33 (protective service occupations) whose age-standardized COVID-19 death rate of 21 per 100,000 (95% confidence interval: 12 to 33) made the news.
“Males in occupational groups 61, 52 and 53 had similar age-standardized all-cause mortality rates of 67, 63, 67 respectively but differential COVID-19 rates.
“Age-standardized mortality rate was for male health professionals [22] was 28 versus 116 for protective service occupations [33]. Male health professionals’ COVID-19 age-standardized death-rate as 8 (95% CI: 5 to 13), substantially lower than 21 (95% CI: 12 to 33) for males in already hazardous protective service occupations. But the percentage of all-cause mortality that COVID-19 accounts for in occupational pair [22] and [61] is significantly higher than for other occupations: both for males and females.
“Finally, recall that late registration of inquest deaths in UK (Scotland excepted) means that COVID-19 deaths which have been referred for inquest in England and Wales do not yet feature in ONS’s registrations.
** for example, its age-standardized all-cause mortality rate per 100,000 population, standardized to the 2013 European Standard Population.
Prof Kevin McConway, The Open University, said:
“There’s a lot of interesting material in both of these sets of figures, but there are also some important limitations when it comes to making sense of the findings. I’ll describe just two of these.
“First, on the analysis of COVID-19 deaths by occupation, it’s important to note that these results really can’t tell us much about the actual causes of differences in death rates between different occupations. The ONS researchers rightly made adjustments for the age patterns of different occupations, particularly in comparing death rates for occupational groups with the rates in the general population. That makes sense, given that many previous investigations have shown that death rates from COVID-19 rise quite rapidly with age. The ONS report also presents its finding separately for males and females, which again makes sense because previous work has found that death rates for men are generally quite a bit higher than those for women. However, ONS did not make similar statistical adjustments for other factors that have also been found, in other studies (including by ONS), to be correlated with death rates. ONS make this explicit in their report, where the say that “In the analysis we adjusted for age, but not for other factors such as ethnic group, place of residence or deprivation.” That means that we can’t use these results to say whether the high death rates, for, say, male taxi and bus drivers arise because of some special features of their work environment, or because different proportions of them come from ethnic minorities than is the case in the general population, or because they are more likely to live in areas where infection and death rates have been high, or because they live in areas where people tend to be poor or deprived. One can speculate or state hypotheses about the balance between these different potential causes, but this data set and analysis won’t reliably help with that process.
“What’s perhaps less obvious is that even a large amount of statistical adjustment wouldn’t necessarily help as much as we’d like with actually defining what causes what here. The problem is that this is observational data – we can’t assign people at random to work as bus drivers or as something else. In analysing causes and effects in observational studies, adjusting for other variables has a role, but exactly what should be adjusted for depends on knowledge about what has the capability of affecting what. Interpreting the results from adjusted analyses as if they tell you precisely what is causing what is very often a mistake. In the case of this ONS study, more adjustments might have thrown some light on causation, but they might not have – and in any case, because more adjustments were not made, the data don’t help all that much in disentangling different causes.
“One might hope that something could perhaps be deduced about cause and effect by making use of the findings of the other ONS study, about potential exposure of different occupations to risk from the new coronavirus. However, that brings me to my other point – I don’t think it could help much. That’s because the data behind these assessments of potential risk seem to me to be rather problematic in relation to the current situation. They are based on (US) data for different occupations from before the current situation arose. Two variables are involved – for each occupation, the extent to which people in a given occupation have to be in close physical proximity to other people to do their work, and the frequency with which people in the occupation have to be exposed to “disease or infections”. Ben Humberstone of ONS points out in his blog that these quantities were measured before the current pandemic, and that working practices may have changed. Currently, the group with the highest level of potential exposure on both of these variables, is dental nurses – unsurprisingly because you can’t do dental procedures without close proximity to someone’s mouth, and because many of the people they work with will have infections, of the gums for instance, so they would typically be in contact with infections daily. But currently many dental clinics and practices are closed, so that the occupational risk to dental nurses who normally work there may be zero. In other occupational groups, measures of social distancing may well have changed the proximity measure. In addition to this, it’s not clear that such a general measure of frequency of exposure to disease or infections, recorded before the pandemic, would provide the right information to assess potential exposure to transmission of the virus, given its pattern of prevalence in the community, and given that it is not associated with all other diseases or infections in a consistent way. So I’m not sure that this ONS analysis tells us very directly who is most at risk in current conditions. (That said, I must admit to being relieved that my occupational group, “Actuaries, economists and statisticians”, comes very close to the bottom on proximity to others, and pretty low on exposure to disease and infection as well.)”
Prof Paul Hunter, Professor in Medicine, UEA, said:
“The latest ONS analysis on reported deaths from COVID 19 amongst people of working age paints a very stark impact of the social class divide in the UK. From these basic analyses, it is not absolutely clear why these differences exist, though for some it is only too obvious.
“One of the key findings has been the high death rates in care workers and social care workers (this does not include health care workers). Male social care workers have more than twice the risk of death than do male health care workers. We have heard repeatedly over recent weeks of the failure to deliver sufficient PPE to care homes and care workers and the high death rate in this occupational group is a probable consequence of that failure.”
Prof Neil Pearce, Professor of Epidemiology and Biostatistics, London School of Hygiene & Tropical Medicine, said:
“This important report confirms that in the working age population Covid-19 is largely an occupational disease. This is not just for health care and social care workers, but for many other occupations that involve contact with people. The highest Covid-19 death rates are for security guards, with high rates also for taxi drivers and chauffeurs, bus and coach drivers, chefs, sales and retail assistants, construction workers, and service occupations (including hospital porters, kitchen and catering assistants, and waiters).
“The authors note that the findings are adjusted for age, but not for other factors such as ethnic group and place of residence; pre-existing conditions such as diabetes and obesity may also vary by occupational group.
“Nevertheless, the findings are striking, and emphasise that we need to look beyond health and social care, and that there is a broad range of occupations which may be at risk from Covid-19. These are many of the same occupations that are now being urged to return to work, in some instances without proper safety measures and PPE being in place.
“The observations are almost certainly due to exposure in these jobs – exposure to people. It is also well-known that working class men (and women) have poorer health than more wealthy people. But here we see the excess specifically in the working class jobs that involve contact with the public. This is not at all surprising – it’s not rocket science to say that if you are a bus driver and you have constant contact with the public (perhaps without proper PPE) then you are more likely to catch Coronavirus.
“For example, taxi and bus drivers (who have contact with the public) have very high Covid-19 death rates, whereas heavy truck drivers (who mostly don’t have public contact) don’t have high rates.”
Prof Keith Neal, Emeritus Professor in the Epidemiology of Infectious Diseases, University of Nottingham, said:
“This is another important analysis of deaths by the Office for National Statistics.
“There are a number of standout features of the report:
“1. The rate amongst health care workers is similar to comparable people. This suggests that PPE is working, or that by the time patients are ill enough to be admitted to hospital, often after over a week, they may be less infectious, or a combination of both factors.
“2. A high rate amongst care workers. This is consistent with what we know about outbreaks in care homes. This may reflect poorer provision and effective use of PPE, or that they are dealing with people early in their COVID-19 infections when they may be asymptomatic and or more infectious, or a combination of both.
“3. A high rate amongst men in transport occupations in which they are exposed to many people. The higher rate in the confined spaces of taxis and similar vehicles is consistent with what we know about indoor transmission.
“4. The higher number of occupations where men have higher death rates may be contributing to the higher overall mortality in men.
“5. Higher mortality in chefs and other occupations will reflect significantly higher transmission risks prior to lockdown restrictions which has important policy implications for restarting work.”
Dr Michael Head, Senior Research Fellow in Global Health, University of Southampton, said:
“The ONS are producing some excellent and important data to support our understanding of the burden of COVID-19 and non-COVID disease. On Sunday evening, the Prime Minister announced many people will be able to go back to work, but had not provided guidance to employers or employees – this is promised later on Monday. In between these announcements, the ONS provide data that shows significantly higher death rates in several occupations where social distancing is difficult. This can have huge implications for transmission within the construction, manufacturing and indeed education sectors (for the proposed phased reopening of primary schools). There is much still to disentangle around COVID-19 risk factors and how best to use that information to manage aspects such as population return to work. However, right now, the forthcoming guidance for safe working simply has to provide extensive detail on how each sector is expected to manage their staff and working environments. Employers and employees need that reassurance.
“We already know that risk factors for dying from COVID-19 include for example (a) being male, (b) having underlying conditions, and (c) ethnicity. The ONS data suggests certain occupations may be more at risk than others. So, taking the points above, a ‘male diabetic Asian bus driver’ is at higher risk than a ‘female healthy Caucasian company director’ of similar age. But, we don’t yet fully understand how these risk factors all mix together and the relative ‘weight’ that should be applied to each risk factor individually and in combination with each other.”
‘Coronavirus (COVID-19) related deaths by occupation, England and Wales: deaths registered up to and including 20 April 2020’: https://www.ons.gov.uk/releases/covid19relateddeathsbyoccupationenglandandwalesdeathsregistereduptoandincluding20thapril2020
‘Which occupations have the highest potential exposure to the coronavirus (COVID-19)?’: https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/employmentandemployeetypes/articles/whichoccupationshavethehighestpotentialexposuretothecoronaviruscovid19/2020-05-11
ONS blog, ‘Understanding the impact of coronavirus on the workforce’: https://blog.ons.gov.uk/2020/05/11/understanding-the-impact-of-coronavirus-on-the-workforce/
All our previous output on this subject can be seen at this weblink:
www.sciencemediacentre.org/tag/covid-19
Declared interests
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.”
None others received.