Associations between an international COVID-19 job exposure matrix and SARS-CoV-2 infection among 2 million workers in Denmark

Objectives This study investigates the associations between the Danish version of a job exposure matrix for COVID-19 (COVID-19-JEM) and Danish register-based SARS-CoV-2 infection information across three waves of the pandemic. The COVID-19-JEM consists of four dimensions on transmission: two on mitigation measures, and two on precarious work characteristics. Methods The study comprised 2 021 309 persons from the Danish working population between 26 February 2020 and 15 December 2021. Logistic regression models were applied to assess the associations between the JEM dimensions and overall score and SARS-CoV-2 infection across three infection waves, with peaks in March–April 2020, December–January 2021, and February–March 2022. Sex, age, household income, country of birth, wave, residential region and during wave 3 vaccination status were accounted for. Results Higher risk scores within the transmission and mitigation dimensions and the overall JEM score resulted in higher odds ratios (OR) of a SARS-CoV-2 infection. OR attenuated across the three waves with ranges of 1.08–5.09 in wave 1, 1.06–1.60 in wave 2, and 1.05–1.45 in those not (fully) vaccinated in wave 3. In wave 3, no associations were found for those fully vaccinated. In all waves, the two precarious work dimensions showed weaker or inversed associations. Conclusions The COVID-19-JEM is a promising tool for assessing occupational exposure to SARS-CoV-2 and other airborne infectious agents that mainly spread between people who are in close contact with each other. However, its usefulness depends on applied restrictions and the vaccination status in the population of interest.


e-Appendix 1: The two stages during which the overview of reviews was conducted
We conducted the overview of reviews in two stages.Stage one was completed in 2017, when we published a report of the overview of reviews in Danish language on the home page of the Danish National Research Centre for the Working Environment (1).
For stage one, we conducted a literature search covering the time span from 2000 to 2015.We identified 72 eligible reviews and we asked experts in the fields to inform us of any potential additional eligible article that was not on the list (see e-appendix 3 below for further details on the experts).In addition, we searched manually the reference lists of all included reviews and of other key articles in the field.
We identified 47 reviews that matched our eligibility criteria at stage one.All reviews were assessed for quality and 16 reviews that were assessed to have weak quality were excluded.Thus, stage one ended with the identification of 31 reviews of moderate (21 reviews) or strong (10 reviews) quality.
At stage two, we prepared the present manuscript for the international research community with an updated literature search until November 2020.We updated our search in the electronic databases and our search in reference lists of newly identified reviews.We used exactly the same search strings as in the first round (see e-Appendix 2 below) and searched the three databases from the date we ended in stage one until November 30 th 2020.We identified several new reviews, resulting in a total of 52 eligible reviews for stage one and stage two combined, 32 of moderate and 20 of strong quality.All experts were contacted in stage one only.In addition to the journal editors and the organizations, we also contacted two international experts (Professor Michiel Kompier, Radboud University, The Netherlands and Professor Emeritus Norbert Semmer, University of Bern, Switzerland) and three experts at the National Research Centre for the Working Environment, Copenhagen, Denmark (Dr.Vilhelm Borg, Professor Anne Helene Garde, and Professor Reiner Rugulies).After we completed stage two, Professors Kompier, Semmer, Garde and Rugulies contributed to the interpretation of the results and the writing of the present paper and were included as co-authors.Place a check mark in the Yes column if the following three criteria are met:

e-Appendix 4: Health Evidence Quality Assessment Tool
1.Each primary study should be assessed for methodological quality using a standardized assessment tool/scale.2. A minimum of four out of the following seven areas should be assessed and the results described for each included primary study (the first three areas needed to be fulfilled): Data analysis • Sources of bias 3. The implication of the quality-related data on the review's findings must be addressed Q7: Are the quality of the primary studies assessed by a minimum of two authors and the method of conflict resolution described?OR: Are an inter-rater agreement Kappa score of at least 0. There is strong quality of evidence that increasing workers' influence on working time is effective for improving work-life balance.The intervention might also be effective with regard to health outcomes, but results are less consistent.

Conclusion
Together, the three reviews in this section covered 32 studies (6 studies are included in more than one review).More than half of these studies (25) were studies with a control group, including 21 RCT studies (of which 4 are included in more than one review).
⋆⋆ Moderate quality of evidence: Consistent results from multiple reviews with a medium proportion of studies with a control group or less consistent results from multiple reviews with a high proportion of studies with a control group.Specifically: Three reviews with a high proportion of studies with a control group (Barbosa et al 2014 (29), Elliot et al 2012 (30), Spector et al 2016 ( 31)) found some positive results for knowledge, burnout, stress, or job satisfaction, however also studies with null-effects and a lack of lasting effects were found.
There is moderate quality of evidence that interventions introducing health care approach changes can lead to improvements in employees' knowledge, burnout, stress, or job satisfaction.

Conclusion
Together, the three reviews in this section covered 32 studies (0 studies are included in more than one review).More than half of these studies (18) had a control group, of which 4 studies were RCTs.
⋆⋆ Moderate quality of evidence: Consistent results from multiple reviews with a medium proportion of studies with a control group or less consistent results from multiple reviews with a high proportion of studies with a control group.Specifically: Two reviews with a high proportion of studies with a control group (Schalk et al 2010 ( 16), Daniels et al 2017 ( 17)) and one review with a medium proportion of studies with a control group (Paguio et al 2019 ( 18)) found studies with positive results but not consistently.
There is moderate quality evidence that interventions that improve various aspects of the psychosocial work environment can lead to positive effects in the psychosocial work environment or employee wellbeing.Positive outcomes were found in studies that introduced workgroup activities focusing on better communication and support and in studies using a participative approach to enhance process aspects in the work environment and the core task.

Conclusion
Together, the six reviews in this section covered 127 studies (11 studies are included in more than one review).About one third of these studies (44) had a control group, of which 5 studies were RCTs.
⋆ Low quality of evidence: Consistent results from multiple reviews with a low proportion of studies with a control group or less consistent results from multiple reviews with a medium proportion of studies with a control group.Specifically: Two reviews with a low proportion of studies with a control group (Zhang et al 2016 (9), Edwards et al 2015 (10)) and one review with a high proportion of studies with a control group (Chen et al 2014 (6)) found positive effects for competencies.
There is low quality of evidence that introducing newly educated nurses to their first job through mentoring programs or other forms of systematic and supportive introduction result in consistent results that it improves competencies.The intervention might also be effective with regard to decreasing turnover rate and improving job satisfaction, but results are less consistent.

Conclusion
Together, the six reviews in this section covered 310 studies (6 studies are included in more than one review.However, overlap could not be calculated for Avolio since they report no reference list of included studies).Almost 60% of these studies (182) were studies with a control group, of which 13 were RCTs.
Inconclusive evidence due to contradictory results: Contradictory results from multiple reviews with a medium or high proportion of studies with a control group Specifically: One review with a high proportion of studies with a control group did not find effects on psychological symptoms for employees (Gayed et al, 2018 (21)), while four reviews with a high proportion of studies with a control group found mixed results for employees' mental health (Stuber et al, 2020 (22)) or wellbeing (Kuehnl et al, 2019 (23)), or moderate effects for organizational aspects (such as employees' job satisfaction) (Collins et al, 2004 (19)) or varying effects for employees' work environment or health outcomes (Grover et al, 2016 (20)).

There is inconclusive evidence due to contradictory results for interventions of leadership trainings and development can lead to positive health and work environment effects for employees.
e-Table 5 The one employee training study that studied that found positive effects on job satisfaction, but not on stress.The study on supervisor training measured sickness absence in employees but found no effect.(Meta-analysis: No)

Conclusion
Together, the six reviews in this section covered 83 studies (9 studies are included in more than one review).Each of these 9 studies were used in two out of the six reviews, none of them in more than two reviews.The most overlap of studies was between the reviews from Montano et al 2014 and Corbiere et al 2009 (6 studies were used in both of these reviews).The majority of the 83 studies (66) were studies with a control group, of which 26 were RCTs.
⋆⋆ Moderate quality of evidence: consistent results from multiple reviews with a medium proportion of studies with a control group or less consistent results from multiple reviews with a high proportion of studies with a control group.Specifically: Six reviews with a high proportion of studies with a control group (i.e.all of the reviews in this group) found that at least half of the intervention studies led to some positive effects.In a few cases, the positive effects were only found for those that were at high risk (Lee et al 2014 (44)) or those that received a high dose of the intervention (Romppanen et al 2016 (43)).Two of the six reviews found a tendency for better effects from combined individual and organizational-level interventions (Corbiere et al 2009 (41)) or from more comprehensive interventions tackling many organizational aspects at once (Montano et al 2014(40)).The outcome variables covered five categories: quality of work (e.g.job demands, job control, social support), psychological resources and responses (e.g.coping skills), physiology (e.g.cholesterol), complaints (e.g.stress, burnout), and absenteeism.For the metaanalysis an overall effect based on these outcomes was calculated The meta-analysis showed no overall effect for the five organizational-level interventions.

There is moderate quality of evidence that organizational-level interventions that aim to improve employees' various health and wellbeing outcomes can
(Meta-analysis: Yes)

Stress indicators and burnout
The three studies all showed some tendencies or potentials for positive effects, but due to too few studies it was not possible to draw a firm conclusion.

Conclusion
Together the 6 reviews in this section covered 47 studies (6 studies are included in more than one review).Each of these 6 studies were used in two out of the six reviews, none of them in more than two reviews.The most overlap of studies was between the reviews from Mimura et al 2002 and Ruotsalainen et al 2015 (2 studies were used in both of these reviews).The majority of the studies (43 of 47) were studies with a control group, of which at least 28 were RCT studies (one review (van der Klink) did not provide detailed information about this aspect).
Inconclusive evidence due to contradictory results: Contradictory results from multiple reviews with a medium or high proportion of studies with a control group.Specifically: Two (Richardson et al, 2008 (48)

Conclusion
Together these two reviews identified 6 studies (0 studies are included in more than one review).
Inconclusive evidence due to lack of studies: High or moderate quality systematic reviews/meta-analyses that cannot draw conclusions due to limited number of studies.Specifically: Although all six identified studies show some positive effects on retention, both reviews do not reach a conclusion due to too few studies.Both reviews pointed out that interventions that initiate multiple strategies at the same time may be better suited to sustaining older employees.
There is inconclusive evidence due to lack of studies about the effect of organizational-level interventions on employee retention.e-Appendix 6: Overlap between primary studies in the three review groups about burnout, various health and wellbeing outcomes and stress.

Reviews that focus on burnout
The eight reviews that focused on the outcome burnout included a total of 125 primary studies.Of these, 19 studies (15.2%) were included in more than one review.Five of the 19 studies were used in three reviews, while 14 studies were used in two reviews.The most overlap between two reviews (9 studies) were between the review by DeChant et al 2019 (36)

Reviews with that focus on various health and wellbeing outcomes
The six reviews that focused on various health and wellbeing outcomes included a total of 83 primary studies.Of these, nine studies (10.8%) were included in more than one review.Each of these nine studies were used in two of the six reviews, none of them in more than two reviews.The most overlap of studies was between the reviews from

Reviews that focus on stress
The six reviews that focused on the outcome stress included a total of 47 studies.Of these, six studies (12.8%) were included in more than one review.Each of these six studies were used in two out of the six reviews, none of them in more than two reviews.The most overlap of studies was between the reviews from

⋆⋆ Moderate quality of evidence
High or moderate quality systematic reviews/meta-analyses demonstrating consistent results from multiple reviews with a medium proportion of studies (25% to 50%) with a control group or less consistent results from multiple reviews with a high proportion of studies (>50%) with a control group.

⋆ Low quality of evidence
High or moderate quality systematic reviews/meta-analyses demonstrating consistent results from multiple reviews with a low proportion of studies (<25%) with a control group or less consistent results from multiple reviews with a medium proportion of studies (25% to 50%) with a control group.

Inconclusive evidence due to lack of studies
High or moderate quality systematic reviews/meta-analyses that cannot draw conclusions due to limited amount of studies.

Inconclusive evidence due to contradictory results
High or moderate quality systematic reviews/meta-analyses demonstrating contradictory results from multiple reviews with a medium (25% to 50%) or high proportion (>50%) of studies with a control group.
Consistent results: almost all studies that measure that effect show an effect in the same direction (or show consistently the absence of an effect) Less consistent results: only some of the studies that measure that effect show an effect in the same direction. e-

# 1 ("
workplace marker) employee*[Text Word]) OR workplace[MeSH Terms] OR job site*[Text Word]) OR (organization and administration[MeSH Terms]) OR organization*[Text Word] OR organisation*[Text Word] OR organizations[MeSH Terms] OR team[Text Word] OR teams[Text Word] OR "work location"[Text Word] OR "workplace"[Text Word] OR "work place"[Text Word] OR "work site"[Text Word] OR worker*[Text Word] OR occupational[Text Word] OR work[Text Word]) OR workplace[MeSH Terms] OR workplace* OR work place* OR worksite* OR work site* Filters: Journal Article; English AND #2 (psychosocial marker) autonomy[Title/Abstract]) OR communication[Text Word]) OR control[Title/Abstract]) OR demands[Text Word] OR demand[Text Word) OR empowerment[Text Word]) OR feedback[MeSH Terms]) OR feedback[Title/Abstract]) OR involve*[Title/Abstract]) OR "job stress"[Text Word]) OR "occupational stress"[Text Word]) OR participation[Text Word]) OR psychological[Title/Abstract]) OR psychosocial[All Fields] OR "role clarification"[Text Word]) OR "role conflict"[Text Word] OR "role conflicts"[Text Word] OR "role imbalance"[Text Word] OR "job security"[Title/Abstract]) OR "shift work"[Text Word] OR "shift worker"[Text Word] OR "shift workers"[Text Word]) OR social capital[MeSH Terms] OR "social capital"[Text Word] OR "social networks"[Text Word] OR "social network"[Text Word)) OR social support[MeSH Terms] OR "social relationships"[Text Word] "social relationship"[Text Word OR stress, psychological[MeSH Terms]) OR (stress[Title/Abstract] OR stressor*[Title/Abstract] OR stress, physiological[MeSH Terms] OR support[Title/Abstract] OR "task characteristics"[Text Word] OR "task structure"[Text Word] OR decision making[MeSH Terms] OR internal external control[MeSH Terms]) OR "team work"[Text Word] OR trust[MeSH Terms] OR trust[Title/Abstract] OR "working time"[Text Word] OR participative[Text Word] OR "flexible working conditions"[Text Word] OR collaborat*[Title/Abstract] OR "occupational health psychology"[Text Word] OR leader*[Text Word]) OR manager*[Text Word]) OR management[Title/Abstract]) OR "job strain"[Text Word]) OR ("effort reward imbalance" OR "effort-reward imbalance") OR bullying[MeSH Terms] OR bullying[Text Word]) OR ("organizational justice" OR "organisational justice") OR "job insecurity"[Text Word]) OR ("working hour"[Text Word] OR "working hours"[Text Word]) OR motivation[MeSH Terms] OR motivation[Text Word]) OR engagement[Title/Abstract]) OR job satisfaction[MeSH Terms] OR "job satisfaction"[Text Word]) OR discriminat*[Title/Abstract] OR "workplace democracy"[Text Word]) OR influenc*[Title/Abstract] OR ("organizational citizenship behavior" OR "organisational citizenship behaviour") OR ("organization citizenship behavior" OR "organisation citizenship behaviour") OR "team training"[Text Word] OR "team building"[Text Word] OR "job design"[Text Word]) OR "self management"[Title/Abstract] OR self manag*[Title/Abstract] OR conflict[MeSH Terms] OR conflict[Title/Abstract] OR well-being[Text Word] Filters: Journal Article; intervention studies[MeSH Terms] OR intervention*[Text Word]) OR "intervention study"[Text Word] OR "intervention studies"[Text Word] Filters: Journal Article; English OR trial[Title/Abstract]) OR transformation*[Text Word] cochrane database syst rev"[Journal]) OR search*[Title/Abstract]) OR meta-analysis [Publication Type]) OR MEDLINE[Title/Abstract]) OR (systematic[Title/Abstract]) AND review[Title/Abstract]) Filters: Journal Article; English #1 AND #2 AND #3 AND #4 Are the population, intervention, and outcomes clearly described in the research question or inclusion criteria?Q2: Were appropriate inclusion criteria used to select primary studies?Q3: Did the authors describe a search strategy that was comprehensive? (at least two strategies from each column) Circle all strategies used • Health databases • Handsearching • Psychological databases • Key informants • Social science databases • Reference lists • Educational databases • Unpublished • Other Q4: Did the search strategy cover an adequate number of years?Q5: Did the authors describe the level of evidence in the primary studies included in the review?Circle the appropriate level of evidence • Level I  RCTs only • Level II  Non-randomized, cohort, case-control • Level III  Uncontrolled studies Q6: Did the review assess the methodological quality of the primary studies?
lead to positive effects.The interventions covered in the six reviews focusing on various health and wellbeing outcomes used a variety of approaches.The effective interventions included working time related changes, improvement in communication, increasing social support, organizational changes, clinical supervision, social support and employee involvementinterventions with the aim to prevent stress reactions (e.g.participation in decision making, Social support groups) (5 out of 36 studies): 5 studies with random assignment to treatment and control conditions Stress indicators, including psychological (most common), physiological, and organizational outcomesThe meta-analysis showed no overall effect for the five organizationalthe aim to reduce occupational stress (e.g.participation in decision making, employee committees, increase individual and group psychosocial coping resources) (5 out of 48 studies): 5 high quality experimental or quasiexperimental studies with control groups

to (peer reviewed journal and English language #1 AND #2 AND #3 AND #4 e-Appendix 3: List of experts
The European Public Health Association (EUPHA), Section on Social Security, Work and Health • European Academy of Occupational Health Psychology (EAOHP) • Scientific Committee on Epidemiology in Occupational Health (EPICOH) • Conference on the Prevention of Work-Related Musculoskeletal Disorders (PREMUS) • International Journal of Occupational and Environmental Health • International Archives of Occupational and Environmental Health • American Journal of Industrial Medicine • Annals of Occupational Hygiene • Occupational Medicine • Journal of Occupational Health • Stress Organisations: • International Commission on Occupational Health.Scientific Committee on Work Organisation and Psychosocial Factors (ICOH-WOPS) •

Appendix 5: Detailed description of the included reviews and synthesis e-Table 5.1: Synthesis of the 30 reviews that examined specific organizational-level interventions (group 1) 1. Changes in working time arrangements (4 reviews/76 studies)
Q10: Do the data support the author's interpretation?TOTAL SCORE Quality Assessment Rating: Strong: 8-10 Moderate: 5-7 Weak: 4 or less e-

⋆⋆⋆ Strong quality of evidence: Consistent
results from multiple reviews with high or moderate quality and with a high proportion of studies with a control group.Specifically: Three reviews (Joyce et al 2010 (2), Nijp et al 2012 (3), Bambra et al 2008 (4)) with a high proportion of studies with a control group found positive results for work-life balance.

found that increased control can lead to positive health effects. There is moderate quality of evidence that interventions that increase employee control can lead to positive health effects for employees. However, not all studies found positive results, which partly might be due to incomplete implementation. Interventions that were conducted for economic reasons seem to have a tendency for negative health effects. 3. Health care approach changes (3 reviews/32 studies))) Reference (Job groups) Quality of review Type of Intervention (number of studies relevant for this overview) and study design of relevant studies Outcomes Main results (Meta-analysis: Yes/No) Proportion of studies with a control group
Staff training in dementia care(19 out of 19 studies): 11 RCT studies, 3 nonrandomized quasi-experimental controlled studies, 5 quasi-experimental one-group time series studies.Knowledge, burnout, job satisfaction, sense of competence, and self-efficacy 16 studies found improvement in at least one staff outcome.Knowledge increased most frequently (in six of seven interventions).(Meta-analysis: No)

6. Prevention of workplace violence (4 reviews/68 studies) Reference (Job groups) Quality of review Type of Intervention (number of studies relevant for this overview) and study design of relevant studies Outcomes Main results (Meta-analysis: Yes/No) Proportion of studies with a control group
four reviews in this section covered 68 studies (5 studies are included in more than one review).Less than half of these studies(27)were studies with a control group, including 4 RCT studies.Three reviews with a medium proportion of studies with a control group found positive effects of workplace violence prevention interventions on employees' confidence, knowledge, and competencies.
With regard to effects for violence ⋆ Low quality of evidence: Consistent results from multiple reviews with a low proportion of studies with a control group or less consistent results from multiple reviews with a medium proportion of studies with a control group Specifically: Three reviews (Kynoch et al, 2011 (28); Price et al, 2015 (25); Tölli et al, 2017 (27)) found that staff training can improve confidence, while Kynoch et al (2011) (28) and Price et al (2015)(25)found that it can also increase knowledge.However, with regard to decreasing violence we found low quality evidence.Two reviews with a medium proportion of studies with a control group found less consistent positive results: one review found less clear results for violence reduction compared to other outcomes(Price et  al, 2015 (25)) while another review found more clear results for violence reduction compared to other outcomes(Tölli et al, 2017 (27)).

.2: Synthesis of the 22 reviews that examined interventions with focus on employees' health, wellbeing, or labor market retention (8 reviews/125 studies) 1. Burnout (8 reviews/125 studies) Reference (Job groups) Quality of review Type of Intervention (number of studies relevant for this overview) and design of relevant studies
(33) of the 19 studies were used in 3 reviews, while 14 studies were used in 2 reviews.The most overlap between two reviews (9 studies) were between the review by DeChant et al 2019(36)andWest et al 2016 (38)which both are reviews about burnout interventions among physicians.About a third of the 125 studies(38)were studies with a control group, of which 35 were RCTs.Consistent results from multiple reviews with high or moderate quality and with a high proportion of studies with a control group.Specifically: Two reviews(Panagioti et al 2016 (32), Awa et al 2010(33)) with a high proportion of studies with a control group found that interventions that either are exclusively organizational or that use a combination of individual and organizational components can reduce burnout.This finding is supported by several other reviews with a medium or low proportion of studies with a control group.

is contradictory evidence about the ability of organizational-level interventions to reduce stress. The six reviews that focused on interventions to reduce stress covered a variety of interventions including participation in decision making, increased influence at work, social support groups, employee committees, improving support or mentoring, improving communication skills, organizational changes in distribution of work tasks, reduce stressors, changing content of care, improving work schedules, combination of organizational with individual changes. 4. Retention (2 reviews/6 studies) Reference (Job groups) Quality of review Type of Intervention (number of studies relevant for this overview)
(49)n der Klink et al, 2001(49)) out of five reviews with a high proportion of studies with a control group found that organizationallevel interventions to reduce stress had no effects.One review (Ruotsalainen et al 2015 (51)) found effects for only one specific type of intervention (scheduling), one review found that positive effects from organizational interventions lasted longer than from individual interventions(Giga et al, 2003 (50)), and one review found weak evidence for combined interventions(Naghieh et al 2015 (52)).

Table 6 .1: Overlap of primary studies in reviews on burnout
and West et al 2016 (38)which both were reviews about burnout interventions among physicians.

Table 6 .2: Overlap of primary studies in reviews on various health and wellbeing outcomes
Montano et al 2014 (40)andCorbiere et al 2009 (41)(six studies were used in both of these reviews).

Table 6 .3: Overlap of primary studies in reviews on stress Reviews that focus on stress
(51)ra et al 2002 (53)and Ruotsalainen et al 2015(51)(two studies were used in both of these reviews). e-