Introduction

During the COVID-19 pandemic, workplaces were important contributors to the spread of severe acute respiratory syndrome corona virus 2 (SARS-CoV-2 virus) [1,2]. As with most communicable diseases, SARS-CoV-2 may constitute an occupational risk to workers [3–9]. Occupational risk of developing COVID-19 not only depends on personal risk factors and vulnerability, but likely varies substantially across occupations and environments. While some studies have explored the occupational risks of COVID-19 [6,10,11], we lack clear insight into which occupations have the highest risks of contracting COVID-19 in countries with limited legislation and no sweeping mandatory lockdowns.

There is likely considerable variation between the patterns across countries depending on the extent of lockdowns, sector specific regulations, as well as guidance and the risks in the context of less mandatory lockdowns remain to be elucidated. Sweden stands out from other nations with its comparatively limited regulations on social distancing and mask-wearing, as well as its decision to keep restaurants, primary schools, and daycare centers open [12], particularly in contrast to its Nordic neighbors, which imposed more strict lockdown measures.

Enhancing the understanding of occupational risks under fewer lockdowns and restrictions would significantly assist authorities in determining whether specific activities within certain sectors should be restricted in efforts to mitigate the transmission of SARS-CoV-2 and avoid shutting down occupations with little excess risk. Furthermore, it is essential for directing public health efforts towards curtailing future epidemics transmission of airborne viruses and enhancing safeguards for workers’ wellbeing.

To this end, we aimed to quantify the occupational risks of COVID-19-related hospital admission among workers in a Swedish population with limited legislation and lockdowns during the first two years of the pandemic.

Methods

Results

An overview of the number of employees, number of different occupations, and incidence COVID-19-related hospital admissions in the study population is presented in Table 1. Analyses were based on 607,179 individuals, 49% female, encompassing 326 different occupations. The median age was 43 years and 85% were from western Europe. Across the follow-up period, we observed 2,633 incident COVID-19 admissions, which peaked around week 52 of 2020 (Fig 1).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Overview of the study population (N = 607,179).

https://doi.org/10.1371/journal.pone.0335662.t001

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. COVID-19 related hospital admissions among employees ages 20-69 in Skåne, Sweden in 2020 and 2021.

https://doi.org/10.1371/journal.pone.0335662.g001

The distribution of sociodemographic characteristics among the reference group (JEM sum score = 0) and at-risk occupations (JEM sum score = 1–24) is presented in S2 Table. Among at-risk occupations, it was more common to be male, younger, have short or medium education, and born outside Sweden compared to the reference group. Those with missing occupational codes were more likely to be male, younger, have low education, born outside Sweden, and live alone compared to both the reference and at-risk occupations (S2 Table). Overall, individuals with missing occupational codes are typically associated with firms that have not reported to the authorities, certain types of union work, as well as project-based employment (Ludvigsson et al. 2019).

The associations between COVID-19-related admission and individual-level risk factors are presented in Table 2. As anticipated, following adjustment, risk was generally higher among males, older age groups, shorter education, foreign born, and those with higher household occupancy. Completed vaccination (minimum of two doses) against COVID-19 was associated with lower risk.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Associations between sociodemographic characteristics and COVID-19 related hospital admission 2020−21.

https://doi.org/10.1371/journal.pone.0335662.t002

Overall, healthcare occupations as a group had an elevated risk of COVID-19-related admission, with an IRR of 1.31 (95% CI: 1.13–1.51) compared to the reference group (Fig 2).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Risk of COVID-19 related hospital admission in relation to occupational group.

Crude model adjusted for sex and age (10-year groups). Fully adjusted model adjusted for sex, age (10-year groups), education (3 groups), country of origin (4 categories), number of household members (0, 1, 2, 3, 4+), and COVID-19 vaccination (from date of second vaccination until end of follow-up).

https://doi.org/10.1371/journal.pone.0335662.g002

Within healthcare, three out of 13 occupations with >500 employees had an elevated risk of COVID-19-related admission; nursing professionals (IRR 1.54; 95% CI: 1.16–2.03), healthcare assistants (IRR 1.28; 95% CI: 1.08–1.52), and nursing aides in private homes (IRR 1.42; 95% CI: 1.07–1.88) (Table 3). Some indication of excess risk was also noted for physiotherapists (IRR 1.67; 95% CI: 0.83–3.37), recreational therapists (IRR 1.47; 95% CI: 0.65–3.30), dental assistants and therapists (IRR 1.45 (95% CI: 0.77–2.73), and x-ray technicians (IRR 1.79; 95% CI: 0.74–4.34). Psychological therapists, dentists, and midwifery professionals did not appear to have an elevated risk of COVID-19-related admission.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Risk of COVID-19 related hospital admission in relation to occupation. Incidence rate ratios (IRR) with 95% confidence limits (CI) relative to employees in all occupations with unlikely occupational exposure to SARS-CoV-2^a.

https://doi.org/10.1371/journal.pone.0335662.t003

In the educational sector, no elevated risk was observed overall (IRR 1.03; 95% CI: 0.86–1.23). However, among secondary education teachers some indication of elevated risk was noted IRR 1.31; 95% CI: 0.85–2.04).

For the transportation sector, an overall excess risk was observed (IRR 1.34; 95% CI: 1.10–1.65), with bus and tram drivers having the highest risk, IRR 1.68 (95% CI: 1.27–2.23). Car, taxi, and van drivers also appeared to have higher risks. Although the point estimate was elevated, the wide confidence intervals reflect limited statistical power due to a low number of hospital admissions.

No elevated risk was observed in the retail sales sector overall or in specific occupations. Similarly, for accommodation, food, building, personal and protective services, and recreational activity sectors, no elevated risks were noted overall or across specific occupations. However, there was a higher risk of COVID-19 admission among fast food preparers with an IRR of 1.76 (95% CI: 0.90–3.42) (Table 2).

Across an additional 62 professions with >500 employees, few occupations were associated with risk of COVID-19-related admission (S3 Table). We did observe excess risk among human resource managers (IRR 1.82; 95% CI: 0.97–3.42), while lower risk was noted among building electricians (IRR 0.52; 95% CI: 0.27–1.00), building frame and related trades workers (e.g., scaffolders, assemblers of prefabricated buildings/houses) (IRR 0.50; 95% CI: 0.25–1.00), as well as software developers (IRR 0.60; 95% CI: 0.39–0.93) (S3 Table).

Overall, when restricting analyses to participants with a diagnosis of COVID-19 in diagnostic position one (n = 2,124), IRRs were not markedly different (S4 Table), with the exception of primary school teachers (IRR 1.27; 95% CI: 0.77–2.11 versus IRR 1.03; 95% CI: 0.78–1.37), waiters and bartenders (IRR 1.42; 95% CI: 0.73–2.76 versus IRR 1.14; 95% CI: 0.61–2.14), and kitchen helpers (IRR 1.21; 95% CI: 0.83–1.75 versus IRR 1.07; 95% CI 0.76–1.50) where IRRs were somewhat higher. Conversely, occupations within the transportation sector were generally lower. In additional sensitivity analyses, only small changes were observed when using a case definition with higher specificity for COVID-19 (e.g., only using the ICD10 code U07.1, n = 2,443) (S5 Table).

Discussion

In this Swedish register-based study we observed increased risks of COVID-19-related admission in many patient-facing occupations across the healthcare sector as well as in multiple occupations within the transportation sector, specifically bus and taxi drivers. However, no apparent excess risks were observed in the educational sector or retail sales sectors and a decreased risk was observed among electricians, some construction workers, and software developers. Despite the absence of widespread lockdowns in Sweden, the magnitude of occupational risks for childcare, education, retail, and service-oriented professions appeared broadly similar to those reported in neighboring countries with extensive lockdowns. However, such comparisons should be interpreted cautiously given differences in study design, background incidence, and other contextual factors [3,8,17].

Frequency and proximity of contact with infected persons remain key contributing factors to being infected with SARS-CoV-2 at the workplace. Our findings are in general agreement with earlier studies which demonstrated elevated risk of COVID-19 across multiple occupations in the healthcare and transportation sectors [3,5,6,8–10,18–23]. For example, a UK study in 900,000 National Health Service workers also reported elevated risks among nurses, care assistants, and health professionals, but also in dentists and midwifery professionals, a finding not observed in the present study [6]. Similarly, a Norwegian case-control study of around 3.5 million persons reported higher risks among nurses, physicians, dentists and physiotherapists [8]. A recent Danish register-based study of 2.4 million Danish employees, conducted by the present research group using comparable methodology, also found elevated risks for several occupations within healthcare, but also medical laboratory technicians, hospital porters, and psychologists [3]. In the present study, point estimates were elevated in nine of the 13 healthcare occupations, although, within some occupations, analyses were hampered by the low number of cases. Studies have shown that over the pandemic period the occupational risks in healthcare workers declined, which is likely explained by increased access and adequate use of personal protective equipment (PPE), testing frequency, as well as early access to vaccination [4,17,24]. Although risk estimates were generally elevated across healthcare professionals, our results do not suggest that Sweden’s limited lockdown strategy substantially altered the occupational risk of COVID-19-related hospital admission in this group. Given that healthcare professionals were largely exempt from restrictions in many countries, such differences in occupational risk between Sweden and countries with stricter lockdowns may not be expected [3].

The transportation sector also appeared to be important to virus acquisition and this is also in line with prior studies [3,7,8,10]. We observed the highest risk in bus and tram drivers, which highlights that these professions would have likely benefited from increased training in infection control guidelines as well as PPE.

We found no apparent elevated risk in the education sector overall, which agrees with some, but not all studies [4,7,8]. A study from Norway reported higher risks in the education sector, but only for the second wave of the pandemic (i.e., 18 July to 18 December 2020) [8]. Furthermore, a Swedish case control study with recruitment from October 2020 to December 2021, reported higher risk of both severe COVID-19 (based on hospital admissions) and SARS-CoV-2 infection (based on positive polymerase chain reaction test) among primary school teachers and early childhood educators compared to occupations with low risk of exposure [7]. Our findings of no overall elevated risk are in agreement with one Danish study, which also found no increased risk of hospital admissions for COVID-19 in the educational, retail sales sectors, and some service occupations [3]. However, the Danish study did observe elevated risks among childcare workers providing daytime care for 4–7-year-old children, whereas we found no elevated risk. Although secondary schools, universities, and some businesses implemented teleworking, Sweden generally remained open, which contrasts with other countries where many childcare centers as well as primary schools were closed or where teaching took place digitally from home. Our findings support the argument that perhaps primary school closures should be carefully considered and possibly avoided in certain contexts, particularly, due to the negative impacts on children’s mental health due to closures [25]. Nonetheless, evidence from other countries is mixed, and our study does not address infection risk among children. Decisions regarding primary school closures should be based on multiple considerations, including both potential impacts on transmission and the well-being of children.

Some occupations were associated with lower risk of COVID-19-related hospital admission compared to those with the lowest JEM sum score, namely electricians, some types of construction workers, and software developers. The lower risk among certain construction workers (e.g., building frame and trade workers) is in line with previous research and is likely related to outdoor working environments [7,26]. However, not all studies have shown this, for example a British and a US study both found that outdoor work was associated with elevated risk for SARS-CoV-2 infection [17,27]. For software developers, the option of working from home was encouraged by both employers and Swedish healthcare authorities, which likely resulted in fewer contacts with infected persons and consequently reduced the risk of infection. Lower risk of SARS-CoV-2 infection in many highly skilled occupations has also been reported by others, which likely had to do with early implementation of teleworking [3]. For example, in Skåne from quarter four 2020 to quarter two 2021 37–41% of those ages 17–74 years worked from home at least one day a week [28].

Our study has multiple strengths. Firstly, it constitutes a major extension of prior publications, as Sweden, being one of few countries with no sweeping mandatory lockdowns (particularly for childcare and primary school education), serves as an important benchmark for comparison. Secondly, we had access to regional and national registries with high accuracy, coverage, and validity pertaining to occupation as well as COVID-19 diagnoses. Thirdly, we were able to control for vaccination status.

Some weaknesses should be acknowledged. We likely missed some occupations that could be at risk due to the limited number of people (e.g., occupations with <500 employees). In addition, since we only accounted for occupation at a single time point (November 2019) one could expect some degree of misclassification since some employees would have started, stopped, or changed jobs during follow-up. Although we used four-digit level occupational codes, these potentially comprise multiple occupations leading to some truncation of occupations. Residual confounding from factors such as pre-existing chronic diseases, social contacts outside the household and workplace, or use of public transportation are additional sources of potential bias. Moreover, outside of healthcare professionals, it is unclear which occupations and individuals had access to and utilized PPE and could be a source of interaction. Since we use hospital-based admissions, we had no data on rapid antigen tests taken at home or other testing centers, and cases thus likely only constitutes more severe COVID-19. Moreover, these two outcomes are not interchangeable and equating them would be misleading. Hospital admissions reflect a more severe spectrum of COVID-19 and are influenced by factors beyond infection risk itself, including age and comorbidities. Consequently, our findings should not be interpreted as direct estimates of occupational risk of contracting COVID-19, but rather as estimates of occupational differences in the risk of progressing to a hospital admission with COVID-19. The intended causal pathway is that occupation influences infection risk, which in turn influences hospitalization; however, alternative pathways are also possible, such as occupation influencing healthcare-seeking behavior, which could independently affect hospitalization risk. Clarifying this distinction is important to avoid misinterpreting hospitalization risk as a direct surrogate for infection risk. We also had no way of disentangling infections acquired at work from those outside work, and exposure outside work or comorbidities are relevant for admission [23,29,30]. However, we adjusted for family size which likely limited this source of potential confounding. Also, multiple comparisons could result in spurious findings and some results should be interpreted with caution, particularly those with a limited number of cases (e.g., psychological therapists, dentists, and midwifery professionals). While we did not stratify by pandemic wave, variant, or vaccination period, our focus was on overall occupational differences across the full two-year period; future research could explore how temporal changes may have influenced these patterns. Lastly, the COVID-19-JEM utilized in the present study has not been validated, and was constructed using subjective expert ratings, and could result in exposure misclassification. Moreover, the JEM was developed shortly after the pandemic began, when knowledge about SARS-CoV-2 transmission and variants was uncomprehensive [18,31]. Although it effectively assesses occupational risk, it should be updated with current knowledge. For example, ventilation is now recognized as a crucial factor due to the major role of airborne transmission [32,33]. While considering work location (indoors vs. outdoors) is valuable, incorporating ventilation aspects would also provide additional insights. Nevertheless, limited and highly variable data on ventilation rates in different workplaces make it difficult to provide reliable job-level assessments.

In general, understanding the risk of occupationally acquired COVID-19 is complex and multifaceted as risk factors often present simultaneously in vulnerable populations. The risks can span workplace characteristics (e.g., ventilation), preventive actions (e.g., COVID-19 testing, ability to maintain distance, or availability of PPE), as well as the social environment (e.g., mandated lockdowns). Moreover, an employee’s sociodemographic characteristics may also influence mode of commuting as well as domestic environment, and these factors often change over time. Overall, the progression of the pandemic and the related occupational risks must be interpreted within the specific context of each country, particularly in the present study, as Sweden had no mandatory lockdowns.

In this Swedish register-based study, our findings support and extend previous work demonstrating elevated risk of COVID-19-related hospital admission among patient-facing healthcare professionals. Our findings also suggest that occupations outside the healthcare sector, namely public transportation workers, may also be at increased risk of COVID-19-related hospital admission. Overall, Sweden’s minimal lockdown measures did not appear to result in higher occupational risks for workers in childcare, education, retail, and service sectors.

Supporting information

Acknowledgments

None.