Objectives To quantify contributions of health- and employment-related risk factors to high annual sickness absence (HASA) among a population-based cohort of older workers in England.Design Prospective cohort study.Setting 24 general practices geographically dispersed across England.Participants Men and women initially aged 50–64 years recruited as part of the Health and Employment After Fifty study.Primary and secondary outcome measures HASA, defined as a total of more than 20 days sickness absence over a 12-month interval.Results 4726 men and women were analysed, providing data on 15 333 12-month follow-up intervals. HASA was reported in 1003 (6.5%) follow-up intervals. In adjusted models, the aspects of health with the largest population-attributable fractions (PAFs) % were disabling musculoskeletal pain (25.4, 95% CI 21.2 to 29.3) and depression (11.3, 95% CI 6.1 to 16.2). Among long-term determinants of health, high body mass index had the greatest impact (PAF% 18.2, 95% CI 8.9 to 26.5). After allowance for health, high physical demands of work (PAF% 22.8) and eligibility for more generous sick pay (PAF% 25.8) made substantial contributions.Conclusions Strategies to minimise avoidable sickness absence at older ages should prioritise: reversal of recent increases in disabling mental illness; encouragement of continued activity in workers with non-specific musculoskeletal pain, supported by modification of occupational tasks if needed; timely surgery for disabling osteoarthritis; reducing obesity; and increasing opportunities for placement in work that is less demanding physically. Further research is needed to clarify the impacts of generous sick pay on patterns of sickness absence.

Objective This study aimed to quantify the risk of COVID-19-related hospital admission in spouses living with partners in at-risk occupations in Denmark during 2020-21.Methods Within a registry-based cohort of all Danish employees (N=2 451 542), we identified cohabiting couples, in which at least one member (spouse) held a job that according to a job exposure matrix entailed low risk of occupational exposure to SARS-CoV-2 (N=192 807 employees, 316 COVID-19 hospital admissions). Risk of COVID-19-related hospital admission in such spouses was assessed according to whether their partners were in jobs with low, intermediate or high risk for infection. Overall and sex-specific incidence rate ratios (IRR) of COVID-19-related hospital admission were computed by Poisson regression with adjustment for relevant covariates.Results The risk of COVID-19-related hospital admission was increased among spouses with partners in highrisk occupations [adjusted IRR (IRRadj)1.59, 95% confidence interval (CI) 1.1-2.2], but not intermediate-risk occupations (IRRadj 0.97 95% 0.8-1.3). IRR for having a partner in a high-risk job was elevated during the first three pandemic waves but not in the fourth (IRRadj 0.48 95% CI 0.2-1.5). Sex did not modify the risk of hospital admission.Conclusions SARS-CoV-2 transmission at the workplace may pose an increased risk of severe COVID-19 among spouses in low-risk jobs living with partners in high-risk jobs, which emphasizes the need for preventive measures at the workplace in future outbreaks of epidemic contagious disease. When available, effective vaccines seem essential.

Objective To map the risk of work-related SARS-CoV-2 across occupations and pandemic waves and investigate its impact on morbidity and partner-risk. Methods The cohort includes 2,4 million employees aged 20–69 with follow-up from 2020 through 2021. During this period, 261,203 employees had a positive SARS-CoV-2 test and 4416 were admitted to hospital with Covid-19 (HA). At-risk occupations defined at the 4-digit DISCO-08 level were identified using a reference population of mainly office-workers defined a priory by a job-exposure matrix (JEM). Incidence rate ratios (IRR) and effect modification by pandemic wave were computed by Poisson regression. We adjusted for demographic, social and health characteristics including household size, completed Covid-19 vaccination and occupation-specific frequency of testing. Results In addition to eight specific occupations in the healthcare sector, we found increased risk of Covid-19 related HA in bus drivers, kindergarten teachers, domestic helpers, and operators in food production (IRR from 1.5–3) and modestly increased risk of SARS-CoV-2 infection in numerous occupations outside the healthcare sector including police and security guards, supermarket attendants, receptionists, cooks, and waiters. After the first year of the pandemic, the risk fell to background levels among healthcare workers but not in other occupations. The risk of Covid-19 related HA was increased in spouses with partners in high-risk occupations (IRR 1.54, 95% CI 1.1–2.2). Employees born in low-income countries and male employees from Eastern Europe more often worked in at-risk occupations. Being foreign-born modified the risk of PCR test positivity, primarily because of higher risk among men born in Eastern Europe working in at-risk occupations (IRR 2.39, 95% CI 2.09–2.72 versus IRR 1.19 (95% CI 1.14–1.23) in native-born men). Conclusion SARS-Cov2 transmission at the workplace was common during the Covid-pandemic in spite of temporary lock-downs which emphasizes the need for improved safety measures during future epidemics.

Knowing the public health impact of occupational hazards is important for prioritization of preventive and mitigating measures and in monitoring how well they succeed. Information is needed on attributable morbidity and mortality, both globally and by national/regional jurisdiction. The best method of estimating population burdens will vary according to the nature of the hazard. One important consideration is whether health effects can be ascribed to work with confidence in the individual. Such attribution is straightforward where a disease occurs only as a consequence of occupational exposure (eg, coal workers’ pneumoconiosis, byssinosis). Alternatively, a link to occupation can sometimes be established through clinical investigation. For example, allergic contact dermatitis may confidently be attributed to work where it is associated with demonstrable sensitization to an agent encountered only in the workplace; and the role of work in an acute injury or poisoning may be clear from its circumstances and timing. Even where a disorder is not occupational in origin, it may be made worse by exposures in the workplace to an extent that can be determined in the individual case. For example, exacerbation of pre-existing asthma by occupational inhalation of irritants may be apparent from serial measurements of lung function when an employee is at, and away from, work. In such circumstances, public health burden can be estimated by aggregating data on individual cases, either across the population as a whole, or in a representative subsample. Possible sources of information include routine surveillance schemes such as the Health and Occupation Research (THOR) Network (1), data on claims for industrial injuries compensation (provided they are sufficiently accurate and complete), and ad hoc surveys in representative samples of the population. Where a disease has material fatality (eg, silicosis), counts of deaths may provide a good measure of attributable mortality. More commonly, occupational disorders are not specific to work, and there is no reliable way of determining occupational contribution in the individual case. The hazard may increase the probability and/or the average severity of a disease. For example, asbestos makes development of lung cancer more likely, while coal mine dust causes chronic obstructive pulmonary disease (COPD) through incremental loss of lung function. Either way, the need is to determine how much morbidity or mortality would be eliminated across the population, if the relevant occupational exposure were removed. To this end, epidemiological data comparing health outcomes in people according to their exposure must be combined with information on the prevalence and distribution of exposure in the population for which an estimate is sought. This is the approach underpinning the WHO/ILO analysis that is reported in Pega et al’s paper (2). Estimates of relative risk for paired combinations of occupational risk factor and disease were collated with data on the population prevalence of exposure to calculate population attributable fractions (PAF) (3), which then were multiplied by estimates of the total population impact of the disease (in terms of deaths and disability-adjusted life-years) to derive burdens attributable to occupation (2). The analysis was necessarily restricted to combinations of risk factor and disease for which there was judged to be adequate evidence, but it also has other important limitations, not all of which are acknowledged and discussed. Some of the assumed hazards are questionable. For example, occupational exposure to formaldehyde is estimated to account for some 350–400 deaths per year from leukemia. Although the International Agency for Research on Cancer has classified formaldehyde as a human carcinogen (4), that decision was controversial, and the systematic review and meta-analysis from which the relative risk was derived concluded that “on balance, these data do not provide consistent support for a relationship between formaldehyde exposure and leukemia risk” (5). Similarly, doubts have been cast on the assumed hazard of ischemic heart disease from long working hours, at least among people of higher socioeconomic status (6). A second problem lies in the ambiguous specification of some risk factors. The analysis attributes large numbers of deaths from COPD to occupational exposure to “particulate matter, gases and fumes” (2). It is unclear, however, what exactly is implied by that term. The many and varied particulates, gases and fumes that people encounter through their work differ widely in their toxicity. If the mix of such pollutants differed between the studies that were used to estimate the prevalence of exposure, and those used to estimate relative risk, then major bias is possible. Even where risk factors are specified more precisely (eg, sulfuric acid), there are challenging complexities in the characterization and classification of exposure. Impacts at an individual level, whether on risk of disease or its severity, can vary enormously according to the timing, duration and intensity of exposures. An earlier report suggests that for many occupational carcinogens, exposures were classed to three levels (background/low/high) (7). However, within such broad categories, there may be substantial heterogeneity of risk. In the WHO/ILO analysis, many of the risk estimates for occupational carcinogens come from industrial cohort studies, which have tended to focus on working populations known or expected to have relatively high intensity and duration of exposure (making any risks more readily detectable). In contrast, data on the prevalence of exposure derive from studies that aimed to ascertain the full extent of exposure in the population, even if only at a modest level. Within a broad exposure category, a meta-estimate of risk from published cohort studies may not be applicable to the distribution of exposures within that category in the general population, and such incompatibility could in some cases cause population burden to be seriously overestimated. Another challenge when extrapolating risk estimates from samples to populations is the potential for effect modification. The burden of back and neck pain was assessed in relation to “occupational exposure to ergonomic factors” (2), defined as “proportion of the population who are exposed to ergonomic risk factors for low back pain at work or through their occupation” (3). However, it is unclear how well the calculation allowed for major variation between countries in the individual risk of musculoskeletal pain and disability from specified occupational activities (8, 9). One way round these problems is to estimate population burden more directly, using the same study to provide information both on the distribution of exposure in the population and the effects of that exposure. For example, a national analysis of mortality by occupation has been used to estimate excess deaths from COPD among coal miners in England and Wales (10). While such analyses have other important limitations (for example, mortality as recorded on death certificates is an imperfect marker for disease, and full account cannot be taken of changes in occupation over a lifetime), the risk estimates that they generate are an average for exposure as it occurs in the population as a whole. In this respect, provided the ascertainment of exposure is reasonably sensitive (specificity is less critical), estimates of population burden will be less prone to bias. Given the many sources of uncertainty in the WHO/ILO analysis, only some of which have been highlighted here, it is surprising that such tight 95% uncertainty ranges are reported. For example, the uncertainty in thousands of deaths globally in 2017 from occupational exposure to formaldehyde is reported as 1 to 1, and that for lung cancer from occupational exposure to diesel exhaust as 16 to 20 (3). This is a concern because findings published under the auspices of authoritative international bodies such as WHO and ILO are liable to be accepted by many without question. As occupational health researchers and practitioners, we are naturally disposed to champion the importance of health protection in the workplace, but that enthusiasm should not compromise scientific rigor. Findings that appear to support our case must be scrutinized with the same care as those that call it into question. When estimating population burdens of disease from occupational hazards, we should aim if possible to triangulate between different analytical approaches and sources of data, carefully considering and acknowledging sources of uncertainty. The potential for error will often be greatest for exposures that are relatively prevalent in the working population (eg, long working hours), for which small differences in excess relative risk can translate into substantial differences in estimated burdens of disease at population level. References 1. Carder M, Hussey L, Money A, et al. The Health and Occupation Research Network: an evolving surveillance system. Safety and Health at Work 2017;8:231-6. https://doi.org/10.1016/j.shaw.2016.12.003 2. Pega F, Hamzaoui H, Náfrádi B, Momen NC. Global, regional and national burden of disease attributable to 19 selected occupational risk factors for 183 countries, 2000-2016: A systematic analysis from the WHO/ILO Joint Estimates of the Work-related Burden of Disease and Injury. Scand J Work Environ Health. 2022;48(2):158168. https://doi.org/10.5271/sjweh.4001 3. GBD 2017 Risk Factor Collaborators. Global, regional, and national comparative risk assessment of 84 behavioural, environmental and occupational, and metabolic risks or clusters of risks for 195 countries and territories, 1990-2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet 2018;392:1923-4. https://doi.org/10.1016/S0140-6736(18)32225-6 4. International Agency for Research on Cancer. Chemical agents and related occupations. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans; Volume 100F:401-435. Lyon, France 2012. https://publications.iarc.fr/123 (accessed 30.11.21). 5. Collins JJ, Lineker GA. A review and meta-analysis of formaldehyde and leukemia. Reg Toxicol Pharmacol 2004;40:81-89. https://doi.org/10.1016/j.yrtph.2004.04.006 6. Kivimäki M, Virtana M, Nyberg ST, Batty GD. The WHO/ILO report on long working hours and ischaemic heart disease - conclusions are not supported by the evidence. Environment International 2020;144:106048. https://doi.org/10.1016/j.envint.2020.106048 7. Concha-Barrientos M, Nelson D I, Driscoll T, et al. Chapter 21. Selected occupational risk factors. In Ezzati M, Lopez AD, Rogers A, Murray CJL, editors. Comparative quantification of health risks: global and regional burden of disease attributable to selected major risk factors. Geneva: World Health Organization; 2004. 8. Coggon D, Ntani G, Palmer KT, et al. Disabling musculoskeletal pain in working populations: Is it the job, the person or the culture? Pain. 2013;154:856-63. 9. Coggon D, Ntani G, Palmer KT, et al. Drivers of international variation in prevalence of disabling low back pain: Findings from the CUPID Study. Eur J Pain. 2019;23:35-45. https://doi.org/10.1002/ejp.1255 10. Coggon D, Harris EC, Brown T, Rice S, Palmer KT. Work-related mortality in England and Wales1979-2000. Occup Environ Med. 2010;67:816-22. https://doi.org/10.1136/oem.2009.052670

Objective Mounting evidence indicates increased risk of COVID-19 among healthcare personnel, but the evidence on risks in other occupations is limited. In this study, we quantify the occupational risk of COVID-19-related hospital admission in Denmark during 2020–2021. Methods The source population included 2.4 million employees age 20–69 years. All information was retrieved from public registers. The risk of COVID-19 related hospital admission was examined in 155 occupations with at least 2000 employees (at-risk, N=1 620 231) referenced to a group of mainly office workers defined by a COVID-19 job exposure matrix (N=369 341). Incidence rate ratios (IRR) were computed by Poisson regression. Results During 186 million person-weeks of follow-up, we observed 2944 COVID-19 related hospital admissions in at-risk occupations and 559 in referents. Adjusted risk of such admission was elevated in several occupations within healthcare (including health care assistants, nurses, medical practitioners and laboratory technicians but not physiotherapists or midwives), social care (daycare assistants for children aged 4–7, and nursing aides in institutions and private homes, but not family daycare workers) and transportation (bus drivers, but not lorry drivers). Most IRR in these at-risk occupations were in the range of 1.5–3. Employees in education, retail sales and various service occupations seemed not to be at risk. Conclusion Employees in several occupations within and outside healthcare are at substantially increased risk of COVID-19. There is a need to revisit safety measures and precautions to mitigate viral transmission in the workplace during the current and forthcoming pandemics.

BACKGROUND:Multisite musculoskeletal pain is common and disabling. This study aimed to prospectively investigate the distribution of musculoskeletal pain anatomically, and explore risk factors for increases/reductions in the number of painful sites. METHODS:Using data from participants working in 45 occupational groups in 18 countries, we explored changes in reporting pain at 10 anatomical sites on two occasions 14 months apart. We used descriptive statistics to explore consistency over time in the number of painful sites, and their anatomical distribution. Baseline risk factors for increases/reductions by ≥3 painful sites were explored by random intercept logistic regression that adjusted for baseline number of painful sites. RESULTS:Among 8927 workers, only 20% reported no pain at either time point, and 16% reported ≥3 painful sites both times. After 14 months, the anatomical distribution of pain often changed but there was only an average increase of 0.17 painful sites. Some 14% workers reported a change in painful sites by ≥3. Risk factors for an increase of ≥3 painful sites included female sex, lower educational attainment, having a physically demanding job and adverse beliefs about the work-relatedness of musculoskeletal pain. Also predictives were as follows: older age, somatizing tendency and poorer mental health (each of which was also associated with lower odds of reductions of ≥3 painful sites). CONCLUSIONS:Longitudinally, the number of reported painful sites was relatively stable but the anatomical distribution varied considerably. These findings suggest an important role for central pain sensitization mechanisms, rather than localized risk factors, among working adults. SIGNIFICANCE:Our findings indicate that within individuals, the number of painful sites is fairly constant over time, but the anatomical distribution varies, supporting the theory that among people at work, musculoskeletal pain is driven more by factors that predispose to experiencing or reporting pain rather than by localized stressors specific to only one or two anatomical sites.

Abstract Background This study quantifies the risk of Covid-19 among ethnic groups of healthcare staff during the first pandemic wave in England. Methods We analysed data on 959 356 employees employed by 191 National Health Service trusts during 1 January 2019 to 31 July 2020, comparing rates of Covid-19 sickness absence in different ethnic groups. Results In comparison with White ethnic groups, the risk of short-duration Covid-19 sickness absence was modestly elevated in South Asian but not Black groups. However, all Black and ethnic minority groups were at higher risk of prolonged Covid-19 sickness absence. Odds ratios (ORs) relative to White ethnicity were more than doubled in South Asian groups (Indian OR 2.49, 95% confidence interval (CI) 2.36–2.63; Pakistani OR 2.38, 2.15–2.64; Bangladeshi OR 2.38, 1.98–2.86), while that for Black African ethnicity was 1.82 (1.71–1.93). In nursing/midwifery staff, the association of ethnicity with prolonged Covid-19 sickness absence was strong; the odds of South Asian nurses/midwives having a prolonged episode of Covid-19 sickness absence were increased 3-fold (OR 3.05, 2.82–3.30). Conclusions Residual differences in risk of short term Covid-19 sickness absences among ethnic groups may reflect differences in non-occupational exposure to SARS-CoV-2. Our results indicate ethnic differences in vulnerability to Covid-19, which may be only partly explained by medical comorbidities.

Abstract Objective To explore impacts of the COVID-19 pandemic on patterns of sickness absence among staff employed by the National Health Service (NHS) in England. Methods We analysed prospectively collected, pseudonymised data on 959,356 employees who were continuously employed by NHS trusts during 1 January 2019 to 31 July 2020, comparing the frequency of new sickness absence in 2020 with that at corresponding times in 2019. Results After exclusion of episodes directly related to COVID-19, the overall incidence of sickness absence during the initial 10 weeks of the pandemic (March-May 2020) was more than 20% lower than in corresponding weeks of 2019, but trends for specific categories of illness varied. Marked increases were observed for asthma (122%), infectious diseases (283%) and mental illness (42.3%), while reductions were apparent for gastrointestinal problems (48.4%), genitourinary/gynaecological disorders (33.8%), eye problems (42.7%), injury and fracture (27.7%), back problems (19.6%), other musculoskeletal disorders (29.3%), disorders of ear, nose and throat (32.7%), cough/flu (24.5%) and cancer (24.1%). A doubling of new absences for pregnancy-related disorders during 18 May to 19 July of 2020 was limited to women with earlier COVID-19 sickness absence. Conclusions Various factors will have contributed to the large and divergent changes that were observed. The findings add to concerns regarding delays in diagnosis and treatment of cancers, and support a need to plan for a large backlog of treatment for many other diseases. Further research should explore the rise in absence for pregnancy-related disorders among women with earlier COVID-19 sickness absence.

Background Clustering of observations is a common phenomenon in epidemiological and clinical research. Previous studies have highlighted the importance of using multilevel analysis to account for such clustering, but in practice, methods ignoring clustering are often employed. We used simulated data to explore the circumstances in which failure to account for clustering in linear regression could lead to importantly erroneous conclusions. Methods We simulated data following the random-intercept model specification under different scenarios of clustering of a continuous outcome and a single continuous or binary explanatory variable. We fitted random-intercept (RI) and ordinary least squares (OLS) models and compared effect estimates with the “true” value that had been used in simulation. We also assessed the relative precision of effect estimates, and explored the extent to which coverage by 95% confidence intervals and Type I error rates were appropriate. Results We found that effect estimates from both types of regression model were on average unbiased. However, deviations from the “true” value were greater when the outcome variable was more clustered. For a continuous explanatory variable, they tended also to be greater for the OLS than the RI model, and when the explanatory variable was less clustered. The precision of effect estimates from the OLS model was overestimated when the explanatory variable varied more between than within clusters, and was somewhat underestimated when the explanatory variable was less clustered. The cluster-unadjusted model gave poor coverage rates by 95% confidence intervals and high Type I error rates when the explanatory variable was continuous. With a binary explanatory variable, coverage rates by 95% confidence intervals and Type I error rates deviated from nominal values when the outcome variable was more clustered, but the direction of the deviation varied according to the overall prevalence of the explanatory variable, and the extent to which it was clustered. Conclusions In this study we identified circumstances in which application of an OLS regression model to clustered data is more likely to mislead statistical inference. The potential for error is greatest when the explanatory variable is continuous, and the outcome variable more clustered (intraclass correlation coefficient is ≥ 0.01).

BACKGROUND:Chronic pain is a common cause of health-related incapacity for work among people in the UK. Individualised placement and support is a systematic approach to rehabilitation, with emphasis on early supported work placement. It is effective in helping people with severe mental illness to gain employment, but has not been tested for chronic pain. OBJECTIVE:To inform the design of a definitive randomised controlled trial to assess the clinical effectiveness of individualised placement and support for people unemployed because of chronic pain. METHODS:A mixed-methods feasibility study comprising qualitative interviews and focus groups with key stakeholders, alongside a pilot trial. STUDY PARTICIPANTS:Primary care-based health-care professionals, employment support workers, employers, clients who participated in an individualised placement and support programme, and individuals aged 18-64 years with chronic pain who were unemployed for at least 3 months. INTERVENTION:An individualised placement and support programme integrated with a personalised, responsive pain management plan, backed up by communication with a general practitioner and rapid access to community-based pain services. OUTCOMES:Outcomes included stakeholder views about a trial and methods of recruitment; the feasibility and acceptability of the individualised placement and support intervention; study processes (including methods to recruit participants from primary care, training and support needs of the employment support workers to integrate with pain services, acceptability of randomisation and the treatment-as-usual comparator); and scoping of outcome measures for a definitive trial. RESULTS:All stakeholders viewed a trial as feasible and important, and saw the relevance of employment interventions in this group. Using all suggested methods, recruitment was feasible through primary care, but it was slow and resource intensive. Recruitment through pain services was more efficient. Fifty people with chronic pain were recruited (37 from primary care and 13 from pain services). Randomisation was acceptable, and 22 participants were allocated to individualised placement and support, and 28 participants were allocated to treatment as usual. Treatment as usual was found acceptable. Retention of treatment-as-usual participants was acceptable throughout the 12 months. However, follow-up of individualised placement and support recipients using postal questionnaires proved challenging, especially when the participant started paid work, and new approaches would be needed for a trial. Clients, employment support workers, primary care-based health-care professionals and employers contributed to manualisation of the intervention. No adverse events were reported. CONCLUSION:Unless accurate and up-to-date employment status information can be collected in primary care health records, or linkage can be established with employment records, research such as this relating to employment will be impracticable in primary care. The trial may be possible through pain services; however, clients may differ. Retention of participants proved challenging and methods for achieving this would need to be developed. The intervention has been manualised. TRIAL REGISTRATION:Current Controlled Trials ISRCTN30094062. FUNDING:This project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 25, No. 5. See the NIHR Journals Library website for further project information.

In the last decade, many studies have examined associations between poor psychosocial work environment and depression. We aimed to assess the evidence for a causal association between psychosocial factors at work and depressive disorders. We conducted a systematic literature search from 1980 to March 2019. For all exposures other than night and shift work and long working hours, we limited our selection of studies to those with a longitudinal design. We extracted available risk estimates for each of 19 psychosocial exposures, from which we calculated summary risk estimates with 95% confidence intervals (PROSPERO, identifier CRD42019130266). 54 studies were included, addressing 19 exposures and 11 different measures of depression. Only data on depressive episodes were sufficient for evaluation. Heterogeneity of exposure definitions and ascertainment, outcome measures, risk parameterization and effect contrasts limited the validity of meta-analyses. Summary risk estimates were above unity for all but one exposure, and below 1.60 for all but another. Outcome measures were liable to high rates of false positives, control of relevant confounding was mostly inadequate, and common method bias was likely in a large proportion of studies. The combination of resulting biases is likely to have inflated observed effect estimates. When statistical uncertainties and the potential for bias and confounding are taken into account, it is not possible to conclude with confidence that any of the psychosocial exposures at work included in this review is either likely or unlikely to cause depressive episodes or recurrent depressive disorders.

Background: The NHS is the biggest employer in the UK. Depression and anxiety are common reasons for sickness absence among staff. Evidence suggests that an intervention based on a case management model using a biopsychosocial approach could be cost-effective and lead to earlier return to work for staff with common mental health disorders. Objective: The objective was to assess the feasibility and acceptability of conducting a trial of the clinical effectiveness and cost-effectiveness of an early occupational health referral and case management intervention to facilitate the return to work of NHS staff on sick leave with any common mental health disorder (e.g. depression or anxiety). Design: A multicentre mixed-methods feasibility study with embedded process evaluation and economic analyses. The study comprised an updated systematic review, survey of care as usual, and development of an intervention in consultation with key stakeholders. Although this was not a randomised controlled trial, the study design comprised two arms where participants received either the intervention or care as usual. Participants: Participants were NHS staff on sick leave for 7 or more consecutive days but less than 90 consecutive days, with a common mental health disorder. Intervention: The intervention involved early referral to occupational health combined with standardised work-focused case management. Control/comparator: Participants in the control arm received care as usual. Primary outcome: The primary outcome was the feasibility and acceptability of the intervention, study processes (including methods of recruiting participants) and data collection tools to measure return to work, episodes of sickness absence, workability (a worker’s functional ability to perform their job), occupational functioning, symptomatology and cost-effectiveness proposed for use in a main trial. Results: Forty articles and two guidelines were included in an updated systematic review. A total of 49 of the 126 (39%) occupational health providers who were approached participated in a national survey of care as usual. Selected multidisciplinary stakeholders contributed to the development of the work-focused case management intervention (including a training workshop). Six NHS trusts (occupational health departments) agreed to take part in the study, although one trust withdrew prior to participant recruitment, citing staff shortages. At mixed intervention sites, participants were sequentially allocated to each arm, where possible. Approximately 1938 (3.9%) NHS staff from the participating sites were on sick leave with a common mental health disorder during the study period. Forty-two sick-listed NHS staff were screened for eligibility on receipt of occupational health management referrals. Twenty-four (57%) participants were consented: 11 (46%) received the case management intervention and 13 (54%) received care as usual. Follow-up data were collected from 11 out of 24 (46%) participants at 3 months and 10 out of 24 (42%) participants at 6 months. The case management intervention and case manager training were found to be acceptable and inexpensive to deliver. Possible contamination issues are likely in a future trial if participants are individually randomised at mixed intervention sites. Harms: No adverse events were reported. Limitations: The method of identification and recruitment of eligible sick-listed staff was ineffective in practice because uptake of referral to occupational health was low, but a new targeted method has been devised. Conclusion: All study questions were addressed. Difficulties raising organisational awareness of the study coupled with a lack of change in occupational health referral practices by line managers affected the identification and recruitment of participants. Strategies to overcome these barriers in a main trial were identified. The case management intervention was fit for purpose and acceptable to deliver in the NHS. Trial registration: Current Controlled Trials ISRCTN14621901. Funding: This project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 25, No. 12. See the NIHR Journals Library website for further project information.

ObjectiveTo quantify occupational risks of COVID-19 among healthcare staff during the first wave (9 March 2020–31 July 2020) of the pandemic in England.MethodsWe used pseudonymised data on 902 813 individuals employed by 191 National Health Service trusts to explore demographic and occupational risk factors for sickness absence ascribed to COVID-19 (n=92 880). We estimated ORs by multivariable logistic regression.ResultsWith adjustment for employing trust, demographic characteristics and previous frequency of sickness absence, risk relative to administrative/clerical occupations was highest in ‘additional clinical services’ (care assistants and other occupations directly supporting those in clinical roles) (OR 2.31 (2.25 to 2.37)), registered nursing and midwifery professionals (OR 2.28 (2.23 to 2.34)) and allied health professionals (OR 1.94 (1.88 to 2.01)) and intermediate in doctors and dentists (OR 1.55 (1.50 to 1.61)). Differences in risk were higher after the employing trust had started to care for documented patients with COVID-19, and were reduced, but not eliminated, following additional adjustment for exposure to infected patients or materials, assessed by a job-exposure matrix. For prolonged COVID-19 sickness absence (episodes lasting >14 days), the variation in risk by staff group was somewhat greater.ConclusionsAfter allowance for possible bias and confounding by non-occupational exposures, we estimated that relative risks for COVID-19 among most patient-facing occupations were between 1.5 and 2.5. The highest risks were in those working in additional clinical services, nursing and midwifery and in allied health professions. Better protective measures for these staff groups should be a priority. COVID-19 may meet criteria for compensation as an occupational disease in some healthcare occupations.Trial registration numberISRCTN36352994.

Objective To explore the patterns of sickness absence in National Health Service (NHS) staff attributable to mental ill health during the first wave of the COVID-19 epidemic in March–July 2020. Design Case-referent analysis of a secondary dataset. Setting NHS Trusts in England. Participants Pseudonymised data on 959 356 employees who were continuously employed by NHS trusts during 1 January 2019 to 31 July 2020. Main outcome measures Trends in the burden of sickness absence due to mental ill health from 2019 to 2020 according to demographic, regional and occupational characteristics. Results Over the study period, 164 202 new sickness absence episodes for mental ill health were recorded in 12.5% (119 525) of the study sample. There was a spike of sickness absence for mental ill health in March–April 2020 (899 730 days lost) compared with 519 807 days in March–April 2019; the surge was driven by an increase in new episodes of long-term absence and had diminished by May/June 2020. The increase was greatest in those aged >60 years (227%) and among employees of Asian and Black ethnic origin (109%–136%). Among doctors and dentists, the number of days absent declined by 12.7%. The biggest increase was in London (122%) and the smallest in the East Midlands (43.7%); the variation between regions reflected the rates of COVID-19 sickness absence during the same period. Conclusion Although the COVID-19 epidemic led to an increase in sickness absence attributed to mental ill health in NHS staff, this had substantially declined by May/June 2020, corresponding with the decrease in pressures at work as the first wave of the epidemic subsided.

COVID-19 is an occupational hazard in healthcare staff. Using data on 959,356 individuals continuously employed by National Health Service (NHS) Trusts in England from 01-01-2019 to 31-07-2020, we applied logistic regression to explore demographic and occupational risk factors for COVID-19 sickness absence during 09-03-2020 to 31-07-2020. The overall prevalence of COVID-19 sickness absence was 9.7% and that for prolonged absence (lasting >14 days) was 2.2%. In a fully adjusted model with any COVID-19 absence as an outcome, there was no association with sex, but risk was lower above age 55 years, and higher in non-white (especially Asian) ethnic groups and people with more frequent sickness absence for any reason during 2019 (Figure 1). Risk was also higher among people whose work was classed (using a job-exposure matrix) as involving direct care of patients expected to have higher prevalence of COVID-19 than the general population (exposure code <3), and in several staff groups, including additional clinical services, registered nursing and midwifery professionals, and allied health professionals. Patterns of risk for prolonged sickness absence were broadly similar, except that ORs increased progressively with age rather than falling and were even higher for non-white vs. white ethnicity. These findings point to priorities for improved measures to protect against risk of SARS-CoV-2 infection causing COVID-19 absence in healthcare staff.