key: cord-0254743-l8sozg8v
authors: Nafilyan, V.; Dolby, T.; Finning, K.; Morgan, J.; Edge, R.; Glickman, M.; Pearce, N.; Van Tongeren, M.
title: Differences in COVID-19 vaccination coverage by occupation in England: a national linked data study
date: 2021-11-11
journal: nan
DOI: 10.1101/2021.11.10.21266124
sha: 51fe17b3dd2340c4f70daee05ca5cd2d744b16a7
doc_id: 254743
cord_uid: l8sozg8v

Background: Monitoring differences in COVID-19 vaccination uptake in different groups is crucial to help inform the policy response to the pandemic. A key gap is the absence of data on uptake by occupation. Methods: Using nationwide population-level data, we calculated the proportion of people who had received two doses of a COVID-19 vaccine (assessed on 31 August 2021) by detailed occupational categories in adults aged 40-64 and estimated adjusted odds ratios to examine whether these differences were driven by occupation or other factors, such as education. We also examined whether vaccination rates differed by ability to work from home. Results: Our study population included 14,298,147 adults 40-64. Vaccination rates differed markedly by occupation, being higher in administrative and secretarial occupations (90.8%); professional occupations (90.7%); and managers, directors and senior officials (90.6%); and lowest (83.1%) in people working in elementary occupations. We found substantial differences in vaccination rates looking at finer occupational groups even after adjusting for confounding factors, such as education. Vaccination rates were higher in occupations which can be done from home and lower in those which cannot. Many occupations with low vaccination rates also involved contact with the public or with vulnerable people. Conclusions: Increasing vaccination coverage in occupations with low vaccination rates is crucial to help protecting the public and control infection, especially in occupations that cannot be done from home and involve contacts with the public. Policies such as 'work from home if you can' may only have limited future impact on hospitalisations and deaths

Whilst several studies highlight differences in vaccination coverage by ethnicity, religion, sociodemographic factors and certain underlying health conditions, there is very little evidence on how vaccination coverage varies by occupation, in the UK and elsewhere. The few study looking at occupational differences in vaccine hesitancy focus on healthcare workers or only examined broad occupational groups.

There is currently no large-scale study on occupational differences in COVID-19 vaccination coverage in the UK.

Using population-level linked data combining the 2011 Census, primary care records, mortality and vaccination data, we found that the vaccination rates of adults aged 40 to 64 years in England differed markedly by occupation. Vaccination rates were high in administrative and secretarial occupations, professional occupations and managers, directors and senior officials and low in people working in elementary occupations. Adjusting for other factors likely to be linked to occupation and vaccination, such as education, did not substantially alter the results. Vaccination rates were also associated with the ability to work from home, with the vaccination rate being higher in occupations which can be done performed from home. Policies aiming to increase vaccination rates in occupations that cannot be done from home and involve contacts with the public should be priorities

The UK began an ambitious vaccination programme to combat the COVID-19 pandemic on 8 December 2020; by 31 August 2021, 78.9% of the UK adult population had received two doses [1] .

Monitoring differences in vaccine uptake between population groups is crucial to inform the policy response. Evidence suggests that rates of COVID-19 vaccination in England differ by ethnicity, religion, socio-demographic factors and certain underlying health conditions [2, 3, 4] . However, a key evidence gap is the absence of data on uptake by occupation. Producing vaccination rates by detailed occupational categories is challenging as routinely collected data used for the analysis of vaccination uptake do not contain information on occupation. The use of surveys is limited as a large sample size is needed to precisely estimate vaccination rates in small groups.

As there is evidence that people working in occupations involving contact with patients or the public are at greater risk of COVID-19 infection and death [5, 6, 7] , it is important to ensure that vaccination coverage is high in these occupations. Measuring vaccination uptake by occupation also has implications for the management of the pandemic, especially as governments are starting to open up the economy, by ending the furlough scheme and encouraging workers to go back to the office. If vaccination rates are high in people working in occupations in which working from home is an option, then having workers back to the office would be expected to have less impact on cases and hospitalisation than if the vaccination rates are low. Similarly, imposing working from home where possible would be less effective if vaccination coverage is high in people who can work from home but low in those who cannot.

This study investigates differences in vaccination rates by occupation in England, using nationwide population-level data. We calculated vaccination rates by detailed occupational categories in adults aged 40 to 64 and estimated adjusted odds ratios to examine whether these differences were driven by potential confounding factors. We also examined whether vaccination rates differed by the ability to work from home and by risk of exposure to infectious diseases.

We linked vaccination data from NHS England's National Immunisation Management System (NIMS) to the Office for National Statistics (ONS) Public Health Data Asset (PHDA) based on NHS number. The ONS 7 regression to estimate the odds of not being fully vaccinated (receiving two doses of the vaccine) by occupation, adjusting for sex, age, geographical and sociodemographic characteristics, disability status and pre-existing conditions. We estimated separate models for each occupation so that the odds ratios could be interpreted as the difference in the odds of not being vaccinated for people working in this occupation compared to all other occupations. We estimated unadjusted odds ratios, odds ratios adjusted for age and sex, and fully adjusted odds ratios.

Finally, we investigated the association between vaccination rate and the ability to work from home. We visualised the relationship using a scatter plot and estimated the strength of the association using univariate linear regression models. We also used logistic regression models to estimate the association between working from home and the odds of not being fully vaccinated, after adjusting for other characteristics. We used the same approach as for occupation and estimated standard error clustered at occupation level. Because the working from home score was standardised with a mean of 0 and a standard deviation of 1, the odds ratio for the score can interpreted as the effect of an increase by one standard deviation. Analyses were conducted using R 3.5.

Our study population included 14,298,147 adults aged 40 to 64 years who lived in England at the beginning of the pandemic and reported being employed in 2011. 88.2% of people had received two doses of a vaccine against COVID-19 by 31 August 2021; 2.7% had only received one dose, whilst 9.1%

had not received any. The average age was 52.8 years and 51.4% were female; 81.5% identified themselves as White British. 21.5% of people had at least one of the health conditions included in the QCovid risk model. Table 1 provides detailed characteristics of the sample.

[ Table 1 ]

As of 31 August 2021, vaccination rates in workers aged 40 to 64 differed by SOC Major Group.

Vaccination rates were over 90% in administrative and secretarial occupations (90.8%); professional occupations (90.7%); and managers, directors and senior officials (90.6%). Those working in elementary occupations had the lowest rate of vaccination with 83.1% of people having received two doses of the vaccine ( Table 2) .

Vaccination rates also differed markedly by occupation unit groups (Table 3 for workers. High rates of vaccination were found in education workers (particularly primary and nursery education teaching professionals, teaching assistants, senior professional of educational establishments), national and local government administrative occupation, and solicitors.

[ Table 3 ] Vaccination rates were lower than 80% in just six occupations, accounting for 259,230 (1.8%) people in our dataset. These included packers, bottlers, canners and fillers (77. 1% [76.8 -77.4 Amongst health care workers, there was some variation in vaccination rates. Most unit groups amongst health professionals had vaccination rates above average, with 91.7% of nurses and midwives being fully vaccinated and 89.4% of medical practioners. Vaccine coverage was low in health associated professionals not elsewhere classified group (e.g. acupuncturists, homeopaths and hypnotherapists) and therapy professionals not elsewhere classified (e.g. art therapists, chiropractor, Cognitive behavioural therapist and dance movement therapists). Vaccination rates were also low in care workers and home carers . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 Standardising for age or adjusting further for other factors likely to be linked to occupation and vaccination did not substantially alter the results (See Supplementary Table S3 ). Out of the 140 occupations which had an unadjusted OR for not being fully vaccinated significantly greater than one, only 7 changed to being significantly lower than one after adjusting for socio-demographic characteristics and pre-existing conditions. These included restaurant and catering managers, chefs, and taxi drivers. For another 23 occupations, the adjusted OR was not different from one. Adjusted OR for not being fully vaccinated were highest in health associated professionals not elsewhere classified ( Results were similar when looking at people who had not received any vaccine (See Supplementary   Table S4 ), suggesting that the results are not driven by people working in some occupation being more likely not have been eligible to receive their second dose.

Vaccination rates were associated with the ability to work from home, as assessed based on O*NET data. Figure 1 shows that vaccination rates tend to be higher in occupations in which working from home is more easily possible. The association was strong, with a R-squared of 0.208 and a F-statistics of 95.4. Results from a logistic regression model suggests that one standard deviation increase in the working from home score is associated with 0.46 [0.37 -0.58] times lower odds of not being fully vaccinated. Adjusting for geographical factors, socio-demographic characteristics and health reduced the ORs did not substantially affect the results (adjusted ORs 0.59 [0.48 -0.69]), indicating that working from home is independently associated with a reduction in the odds of not being vaccinated.

[ Figure 1 ]

Main findings of this study . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 Using whole population level linked data in England, our analysis shows that the vaccination rates of adults aged 40 to 64 years and over differed markedly by occupation. Vaccination rates were high in administrative and secretarial occupations, professional occupations and managers, directors and senior officials and low in people working in elementary occupations. Vaccination rates were also associated with the ability to work from home, with vaccine uptake being higher in occupations which can be performed from home (this relationship remained when sociodemographic characteristics were adjusted for).

Few studies have investigated how COVID-19 vaccination coverage varies by occupation. Several studies investigated vaccine hesitancy and vaccine coverage in health care workers, highlighting some differences between different job roles, with doctors having slightly higher vaccination coverage and lower hesitancy than midwives and nurses [10, 11] . We found no substantial differences in vaccination rates between nurses and medical practitioners, but found much lower vaccination rates in health associated professionals such as acupuncturists, Chiropractors and osteopaths, and in care workers and home carers. The difference between our results and those from studies focused on health care workers may stem from the age range selected (40 to 64) and recent migrants being excluded from our dataset.

Migrants make up about a fifth of the health and social care workforce [12] and people born outside of the UK tend to have lower vaccination coverage [13] .

Our findings that vaccination rates are higher among managers, directors and senior officials and in people working in professional occupations, compared with those working in elementary occupations, are in line with studies showing that vaccination rates in the UK and the US are higher in more wealthy areas and amongst people from higher socio-economic status [13, 3, 14, 4] . Our results are also consistent with a study from the US showing that vaccine hesitancy varied widely by occupation category, with hesitancy being low in people working in professional occupations, especially in the life, physical and social sciences and particularly high in people working in construction and extraction [15] .

A major strength of this study is the use of nationwide linked popula1tion-level data from clinical records and the 2011 Census. This study is the first to examine how vaccination rates vary by occupation using population-level data. Because information on occupation is not collected in electronic health records, we used data from the Census to assess people's occupation. Some surveys collect data on both . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 occupations and vaccination but face the issue that non-response is likely correlated with the propensity to be vaccinated. Having population level data based on electronic health records and the Census, which is mandatory and has a high response rate, we were able to accurately estimate vaccination rates for detailed occupational groups.

The main limitation of our study is that the information on occupation is nine years out of date. Our exposure is therefore likely to be misclassified for a proportion of people, because they have left the labour force or changed occupation since 2011, especially during the pandemic. To mitigate measurement error, we restricted our analysis to people aged 40-64 years, who have a relatively high occupational stability, as shown in an analysis of a large longitudinal household survey [7] . Exposure misclassification is nonetheless likely result in underestimating the differences in vaccination rates between occupations, especially if the exposure misclassification is random. Turnout rate may be greater in some occupations than others, especially in elementary occupations where vaccination rates are lower, which could lead to underestimate the differences in vaccination rates. However, over twothird of people aged 40 to 64 in 2019 were in the same major group as in 2011 [7] . The bias is likely to be smaller in the estimated vaccination rates by major group than by unit group.

An additional limitation is that, because the Public Health Data Asset was based on the 2011 Census, it excluded people who were living in England in 2011 but who did not take part in the Census, as well as respondents who could not be linked to the 2011-2013 NHS patient register and recent migrants. As a result, we excluded 5.4% of vaccinated people who could not be linked to the PHDA.

Our analysis highlights substantial differences in vaccination coverage between occupations, with vaccination rates being higher in professional occupations and managers, directors and senior officials and lower in people working in elementary occupations. These differences are only partially explained by other factors, including education, suggesting that some occupations are independently associated with vaccination status. People working in elementary occupations may have little job control and therefore may find it more difficult to attend vaccination or may be put off by the potential side effects, which could prevent them from working. Existing evidence suggests that occupational characteristics, such as job strain, can affect health behaviours, such as compliance with treatment [16] . Vaccine hesitancy may also be driving differences in uptake. The drivers of vaccine hesitancy are complex, and occupation may play a part. Indeed, we find that vaccine coverage is low in health associated professionals such as acupuncturists and osteopaths, and therapy professionals such as art therapist, chiropractor, cognitive behavioural therapists and dance movement therapists. They may be more likely to reject mainstream medicine than other groups. The rates were also low in fitness instructors, who may be more likely to believe they are very healthy and therefore may not need the vaccine.

In our study population of older worker adults in England, 88.2% had received two doses of a vaccine against COVID-19 by 31 August 2021 -a public health triumph. However, many occupations with lower vaccination rates involved contact with the public (taxi and cab drivers, sales and retail assistant, waiters and waitresses) or with vulnerable people (care workers and home carers). Therefore, increasing vaccination coverage in these occupations would be crucial to help protect the public and control infection. Our results also show that vaccination rates are associated with the ability to work from home, with vaccination rates typically being higher in occupations which can be done from home.

Therefore, policies such as 'work from home if you can' may only have limited impact on hospitalisations and deaths, as the vaccination rates are higher in the occupations that can be done from home and lower in those which cannot. Efforts should be made to increase vaccination rates in occupations that cannot be done from home. This includes many elementary occupations where it may be difficult to get time off work to be vaccinated, or where workers may not have the spare time or resources to access vaccination or are worried about missing work because of the side effects of the vaccines. As these are some of the more deprived groups in society, increasing vaccination rates in these groups would help reduce inequalities in vaccination coverage. Our results could also help inform future vaccination strategy and future models of COVID-19 transmission. " BMJ Open, vol. 11, no. 7, p. e053402, 2021. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101/2021.11.10.21266124 doi: medRxiv preprint Williamson and et al., "Trends and clinical characteristics of COVID-19 vaccine recipients: a federated analysis of 57.9 million patients' primary care records in situ using OpenSAFELY," medRxiv, 2021.

[ 4 ] ONS, "Coronavirus and vaccination rates in people aged 50 years and over by socio-demographic characteristic, England: 8 in COVID-19 vaccination coverage in England: a national linked data study," medRxiv, 2021.

[ 9 ]

ONS, "Which jobs can be done from home?," 2020.

[Online]. Available:

https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/employmentandemployeetypes/articles/whichjobsc anbedonefromhome/2020-07-21. 

Funding . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 This work was funded by ONS and supported by funding through the National Core Study "PROTECT" programme, managed by the Health and Safety Executive on behalf of HM Government

Ethical approval was obtained from the National Statistician's Data Ethics Advisory Committee (NSDEC(20)12). full access to all data in the study and takes responsibility of the integrity of the data and the accuracy of the data analysis. VN affirms that the manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned have been explained

Tables and Figures   Table 1 Characteristics of the study population . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ;

1 

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101/2021.11.10.21266124 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101/2021.11.10.21266124 doi: medRxiv preprint

At least one condition identified as a risk factor for COVID-19 in the QCovid risk model

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101/2021.11.10.21266124 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 Social services managers and directors 25,084 6.5 (6.2 -6.9) 1. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint Sales and retail assistants 514,651 10.0 (9.9 -10.1) 3.6 (3.5 -3.6) 86.4 (86.3 -86.5)

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 Train and tram drivers 13,616 7.5 (7.0 -7.9) 1. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 

Vehicle body builders and repairers 13

Boat and ship builders and repairers 5

Electricians and electrical fitters 89