As countries adapt to the longer-term challenges of Covid-19, occupational health professionals increasingly are being asked to advise on the fitness for work of patients who might be unusually vulnerable to the disease because of their age, ethnicity and/or comorbidities. The risk of contracting Covid-19 through work will depend on its prevalence in the local community; the extent to which the job entails either close proximity to people who could be carrying the infection or contact with material that might be contaminated by the virus; the effectiveness of any measures to reduce transmission (such as barriers or personal protective equipment); and personal immunity (e.g. as a consequence of previous infection, or perhaps in the future, immunization). More important for the individual, however, is the risk of developing serious or fatal Covid-19, which will also depend on his/her personal vulnerability should infection occur.
Early in the course of the Covid-19 epidemic, the UK government issued guidance on vulnerability from comorbidities [1,2]. More recently, others have published guidance on risk reduction for healthcare workers [3] and the broader management of return to work in the face of health risks from Covid-19 [4]. None of these documents attempts to quantify risks associated with specific comorbidities, but research is now emerging that allows more detailed and reliable assessment of vulnerability to Covid-19. This has enabled us to develop an evidence-based risk model that can be used to estimate personal vulnerability. The tool, which is intended principally to assist decisions on occupational placement of workers in the UK, was first published as a freely available online resource on 20 May 2020, and is being updated and refined periodically as relevant new data become available [5].
Our aim was to assess and compare risks of fatality in people who contract SARSCov-2 infection, according to their age, sex, ethnicity, smoking habits and various comorbidities. In preliminary searches of the published literature, no evidence could be found on risks of fatality in representative samples of people infected by the virus (including those with asymptomatic infection). However, analyses of mortality from Covid-19 in the general population could be expected to provide good proxy measures of relative risk, provided the likelihood of contracting infection did not vary importantly according to the risk factors under consideration (as might occur, for example, because of selective shielding by people with certain comorbidities). In addition, estimates of risk might be possible if data could be found on fatality rates by comorbidity in patients admitted to hospital because of Covid-19, and then combined with information about the prevalence of the same comorbidities among hospitalized Covid-19 patients as compared with the general population.
Because of the urgency to improve on earlier advice, we selectively sought papers that would provide the strongest evidence relevant to the UK and did not attempt systematically to search for, and review, all published data that might bear on the risks that we were trying to characterize. In this respect, one report initially stood out as particularly suited to our purpose. Produced by the OpenSAFELY (OS) collaborative, it presented first results from a cohort study of more than 17 million adults registered with English general practices and followed between 1 February and 25 April 2020 [6]. It used multivariate Cox regression to estimate mutually adjusted hazard ratios for death in hospital with confirmed Covid-19 in relation to risk factors determined from pseudonymized individual primary care records. As well as sex, age, ethnicity, smoking habits and multiple comorbidities, these analyses adjusted for social deprivation (which might in part indicate higher exposure through domestic crowding) and for administrative region (to allow for varying rates of infection in different parts of the country).
This study included a substantial proportion of the adult population of England, and was based on more than 5000 deaths attributed to Covid-19. Moreover, information about risk factors came from data recorded before the onset of infection, which reduced the possibility of ascertainment being biased in relation to the outcome. Nevertheless, we sought to check the plausibility of its findings, using data from four independent sources: Office for National Statistics (ONS) data on mortality from Covid-19 by sex and age in England and Wales during March 2020; ONS estimates of sex-specific odds ratios for coronavirus-related deaths by ethnic group in England and Wales; a report on outcomes, including mortality, in a cohort of 16 749 patients with Covid-19 admitted to hospitals in England, Wales and Scotland during 6 February to 18 April 2020 [7]; and data on the prevalence of comorbidities by sex and age in samples of people (intended to be nationally representative) from recent rounds of the Health Survey for England.
To produce the first iteration of our risk model, we abstracted hazard ratios for risk factors of interest from the OS report, and where possible checked their plausibility against data from the other sources mentioned above. Subsequently, new evidence became available from several further papers, including most notably a report on associations of diabetes with mortality from Covid-19 in a cohort of 61 million patients registered with English general practices [8], and another on risk factors for death related to Covid-19 in a subcohort of more than 3 million patients with diabetes [9]. Together, these allowed risk estimates for a finer classification of diabetes by type and level of control, checks on possible interactions between diabetes and other risk factors, and refined risk estimates for cardiovascular disease.
The relative risks adopted for our risk model, as at 16 July 2020, are shown in Table 1, together with qualitative assessments of the strength of evidence (‘robustness’) on which each estimate of risk is based. Smoking was not included in the model because, after allowance for other factors, it appeared not to carry any material increase in risk.
Table 1.
Risk factor | Relative risk | Equivalent added years of agea | Robustness of risk estimate |
---|---|---|---|
Female sex | 0.6 | -5 | Moderately robust |
Ethnicity | |||
Asian or Asian British | 1.5 | 4 | Moderately robust |
Black | 1.7 | 5 | Moderately robust |
Mixed | 1.4 | 3 | Provisional |
Other non-white | 1.3 | 3 | Provisional |
Body mass index (kg/m2) | |||
30–34.9 | 1.3 | 3 | Provisional |
35–39.9 | 1.6 | 5 | Provisional |
≥40 | 2.4 | 9 | Provisional |
Hypertension (according to actual age, years) | |||
20–26 | 3.3–3.6 | 12 | Provisional |
27–33 | 3.0–3.3 | 11 | Provisional |
34–39 | 2.7–2.9 | 10 | Provisional |
40–44 | 2.4–2.6 | 9 | Provisional |
45–49 | 2.2–2.4 | 8 | Provisional |
50–54 | 2.0–2.1 | 7 | Provisional |
55–57 | 1.8–1.9 | 6 | Provisional |
58–61 | 1.6–1.8 | 5 | Provisional |
62–64 | 1.5–1.6 | 4 | Provisional |
65–67 | 1.3–1.4 | 3 | Provisional |
68–70 | 1.2–1.3 | 2 | Provisional |
71–72 | 1.1 | 1 | Provisional |
≥73 | 1 | 0 | Provisional |
Heart failure | 2.2 | 8 | Provisional |
Other chronic heart disease | 1.3 | 3 | Provisional |
Cerebrovascular disease | 2.2 | 8 | Provisional |
Asthma | |||
Mild (no requirement for oral corticosteroids in past year) | 1.1 | 1 | Moderately robust |
Severe (requiring oral corticosteroids in past year) | 1.4 | 3 | Moderately robust |
Chronic respiratory disease (excluding asthma) | 1.9 | 6 | Moderately robust |
Diabetes | |||
Type 1 | |||
HbA1 ≤58 mmol/ mol in past year | 2.0 | 7 | Moderately robust |
HbA1 >58 mmol/ mol in past year | 2.7 | 10 | Moderately robust |
HbA1c unknown | 3.3 | 12 | Moderately robust |
Type 2 and other | |||
HbA1 ≤58 mmol/ mol in past year | 1.5 | 4 | Moderately robust |
HbA1 >58 mmol/ mol in past year | 2.0 | 7 | Moderately robust |
HbA1c unknown | 2.3 | 8 | Moderately robust |
Chronic kidney disease | |||
Estimated GFR 30–60 ml/min | 1.5 | 4 | Moderately robust |
Estimated GFR < 30 ml/min | 3.0 | 11 | Moderately robust |
History of dialysis or end-stage renal failure | 3.7 | 13 | Moderately robust |
Non-haematological cancer | |||
Diagnosed <1 year ago | 1.7 | 5 | Provisional |
Diagnosed 1–4.9 years ago | 1.2 | 2 | Provisional |
Diagnosed ≥5 years ago | 1 | 0 | Provisional |
Haematological malignancy | |||
Diagnosed <1 year ago | 2.8 | 10 | Provisional |
Diagnosed 1–4.9 years ago | 2.5 | 9 | Provisional |
Diagnosed ≥5 years ago | 1.6 | 5 | Provisional |
Liver disease | 1.8 | 6 | Provisional |
Chronic neurological disease other than stroke or dementiab | 2.6 | 9 | Provisional |
Organ transplant | 3.6 | 12 | Provisional |
Spleen diseasesc | 1.4 | 3 | Provisional |
Rheumatoid/lupus/ psoriasis | 1.2 | 2 | Provisional |
Other immunosuppressive conditiond | 1.8 | 6 | Provisional |
aAdded years for hypertension are calculated from relative risks before rounding.
bMotor neurone disease, myasthenia gravis, multiple sclerosis, Parkinson’s disease, cerebral palsy, quadriplegia, hemiplegia and progressive cerebellar disease.
cIncludes splenectomy, or spleen dysfunction (e.g. from sickle cell disease).
dIncludes HIV, conditions inducing permanent immunodeficiency (ever diagnosed), aplastic anaemia, and temporary immunodeficiency recorded within the past year.
A notable feature of Covid-19 is that mortality rates increase exponentially with age. In these circumstances, vulnerability from other risk factors can conveniently be expressed in terms of the added years of age that would give an equivalent increase in risk [10]. Table 1 quantifies the vulnerabilities associated with demographic variables and comorbidities as age equivalents as well as relative risks. If it is assumed that when risk factors are present in combination, their relative risks multiply (the normal default assumption in regression analyses such as in the OS report), then combined effects can be estimated by summing the age equivalent for each. Moreover, by adding the summed age equivalents to the person’s true age, it is possible to generate a summary measure of personal vulnerability, which we have termed ‘Covid-age’. It represents the age of a healthy white male with equivalent vulnerability (white males being the largest demographic group in the UK workforce).
Covid-ages can be translated into estimated case–fatality rates. This is complicated by uncertainties about the prevalence of asymptomatic infection, but from the limited data that are available, it is estimated that a Covid-age of 47 years might correspond approximately to a case fatality rate of 2 per 1000 [5]. For each additional 10 years of age, case fatality increases by a factor of 2.8.
Our analysis has important limitations. Currently, the risk assessments are derived largely from a single study, albeit with checks on plausibility using other sources. Data on some risk factors were incomplete and, although the study was large, its findings, in particular for rarer comorbidities, are liable to statistical uncertainties. Risks associated with some comorbidities may have been inappropriately attenuated by adjustment for deprivation. In the absence of evidence to the contrary, we have assumed as a first approximation that relative risks from different factors multiply, but that may not always be true. For example, evidence has now emerged that although hypertension is not associated with higher mortality when results are averaged across all ages, it carries an increased risk in younger adults [11]. Also, some risk estimates may have been liable to residual bias because of differences in exposure to infection.
A further limitation is the heterogeneity of some categories of comorbidity. For example, chronic pulmonary disease aggregates various disorders, each with a range of severity. In the future, data should emerge that allow evidence-based risk assessments for more specific sub-categories of disease. Meanwhile, clinical judgement should be applied when considering how risks might vary within a broader, aggregated category of comorbidity, taking as a starting point the estimated risk for the category as a whole. Similarly, clinical judgement is required for comorbidities on which epidemiological evidence is not yet available.
We expect that some of these limitations can be addressed as further evidence becomes available. Meanwhile, we believe that our assessment of vulnerability offers an improvement on what has been available previously. We caution against simplistic rules for decisions based only on the risks that it estimates. It does not remove the need for clinical judgement, and there are other important considerations when managing occupational risks from Covid-19—for example, the practicability of different possible control measures, the expected prevalence of infection in the local population, the personal value judgements of the individual worker and prevailing advice from government (which may be driven by a need to control demands on healthcare services as well as individual risk). With these caveats, we hope that it will prove a useful contribution to decisions about fitness for work during the Covid-19 pandemic in the UK adult population.
References
- 1. https://www.gov.uk/government/publications/guidance-on-shielding-and-protecting-extremely-vulnerable-persons-from-covid-19/guidance-on-shielding-and-protecting-extremely-vulnerable-persons-from-covid-19. (20 May 20, date last accessed).
- 2. https://www.gov.uk/government/publications/staying-alert-and-safe-social-distancing/staying-alert-and-safe-social-distancing. (20 May 20, date last accessed).
- 3. Khunti K, de Bono A, Browne I, et al. . Risk Reduction Framework for NHS Staff at risk of COVID-19 Infection. https://www.fom.ac.uk/wp-content/uploads/Risk-Reduction-Framework-for-NHS-staff-at-risk-of-COVID-19-infection-12-05-20.pdf. (19 May 2020, date last accessed).
- 4. Lalloo D, Roy M, Macdonald E. COVID-19 Return to Work Guide for Health Professionals Advising Patients and Employers.https://www.som.org.uk/SOM_RTW_guide_health_professionals_COVID-19_FINAL.pdf. (19 May 2020, date last accessed).
- 5. Coggon D, Croft P, Cullinan P, Williams A, on behalf of the Joint Occupational Health COVID-19 Group. Covid-19 Medical Risk Assessment.Association of Local Authority Medical Advisers.2020. https://alama.org.uk/covid-19-medical-risk-assessment/. (19 May 2020, date last accessed). [Google Scholar]
- 6. The OpenSAFELY Collaborative, Williamson E, Walker AJ, et al. . OpenSAFELY: Factors Associated with COVID-19-Related Hospital Death in the Linked Electronic Records of 17 Million Adult NHS Patients. 10.1101/2020.05.06.20092999 (13 May 2020, date last accessed). [DOI]
- 7. Docherty AB, Harrison EM, Green CA et al. . Features of 16,749 Hospitalised Patients With COVID-19 Using the ISARIC WHO Clinical Characterisation Protocol. 10.1101/2020.04.23.20076042 (13 May 2020, date last accessed). [DOI] [PMC free article] [PubMed]
- 8. Barron E, Bakhai C, Kar P, et al. . Type 1 and Type 2 Diabetes and COVID-19 Related Mortality in England: A Whole Population Study. https://www.england.nhs.uk/wp-content/uploads/2020/05/valabhji-COVID-19-and-Diabetes-Paper-1.pdf. (19 June 2020, date last accessed). [DOI] [PMC free article] [PubMed]
- 9. Holman N, Knighton P, Kar P, et al. . Type 1 and Type 2 Diabetes and COVID-19 Related Mortality in England: A Cohort Study in People With Diabetes. https://www.england.nhs.uk/wp-content/uploads/2020/05/Valabhji-COVID-19-and-Diabetes-Paper-2-Full-Manuscript.pdf (19 June 2020, date last accessed).
- 10. Spiegelhalter D. How old are you really? Communicating chronic risk through ‘effective age’ of your body and organs. BMC Med Inform. Decis. Making 2016;16:104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Williamson EJ, Walker AJ, Bhaskaran K, et al. . OpenSAFELY: factors associated with COVID-19 death in 17 million patients. Nature 2020. https://www.nature.com/articles/s41586-020-2521-4(4 August 2020, date last accessed). [DOI] [PMC free article] [PubMed] [Google Scholar]