https://orcid.org/0000-0002-3331-4853
Abstract
Outdoor workers are exposed to high levels of solar ultraviolet radiation (UVR). UVR causes skin cancer and is a risk factor for cataract and other short- and long-term health effects, but there are significant knowledge gaps regarding the exposure-response relations based on quantitative measures of UVR exposure. We developed a quantitative UVR job-exposure matrix (JEM) for the general working population of Europe.
Three experts from each of Northern, Central, and Southern Europe rated duration of outdoor work for all 372 occupations defined by the International Standard Classification of Occupations from 1988 (ISCO-88(COM)). A systematic literature search identified 12 studies providing 223 sets of summary workday UVR exposure for 49 ISCO-88(COM) occupations based on 75,711 personal workday measurements obtained from 2,645 participants and reported as arithmetic mean standard erythemal dose (SED). We combined the expert ratings with the measured occupational UVR exposure data and estimated harmonized workday UVR exposures for all 372 occupations in a linear mixed effects model.
Monotonically increasing workday UVR exposure of 0.68, 1.57, 1.80, and 2.49 SED were seen by increasing expert ratings of 0, 1 to 2, 3 to 4, and ≥5 h of daily outdoor work. The UVR exposure showed a 6-fold increase from lowest to highest exposed occupation. Farm hands, roofers, concrete placers, and other occupations within craft and related trades were among the highest exposed, while bartenders, wood-processing-plant operators, and several white-collar occupations who typically work indoor were among the lowest exposed.
This quantitative JEM for solar UVR exposure proves able to provide substantial discrimination between occupations, shows good agreement with expert assessments, and may facilitate epidemiological studies characterizing the exposure-response relation between occupational solar UVR exposure and different health effects.
Solar ultraviolet radiation (UVR) causes skin cancer and is a risk factor for cataract and other short- and long-term health effects. This study developed a job-exposure matrix (JEM) from a combination of personal solar UVR exposure measurements and expert ratings and presents quantitative UVR exposure estimates for all ISCO-88(COM) occupations. The JEM proves able to provide substantial discrimination between occupations, shows good agreement with expert assessment, and may facilitate epidemiological studies.
Introduction
Solar ultraviolet radiation (UVR) is non-ionizing electromagnetic radiation with wavelengths between 100 and 400 nanometers (nm) emitted by the sun. Approximately 95% of the UVR reaching the terrestrial surface is UVA (315 to 400 nm) and 5% UVB (280 to 315 nm). Terrestrial solar UVR exposure depends on the solar cycle, the stratospheric ozone level, cloud coverage, air pollution, altitude, surface reflection, and solar zenith angle as defined by latitude, season, and time of the day (IARC Working Group on the Evaluation of Carcinogenic Risks to Humans 2012; Vitt et al. 2020). Solar radiation with wavelengths below 290 nm is absorbed in the higher atmosphere, mainly by molecular oxygen and ozone. Most of the UVB irradiance is absorbed by ozone in the stratosphere while gases and aerosol scattering play a relatively larger role than ozone absorption for UVA. An individual’s UVR exposure furthermore depends on activity, posture, duration and degree of shade, and self-protection (clothing coverage, sunscreen, hat and sunglasses) and may not correlate well with ambient solar UVR exposure (Schmalwieser et al. 2010; Bodekær et al. 2015; Soueid et al. 2022). Personal solar UVR exposure varies across occupations, and high levels are seen in a variety of outdoor jobs (Thieden, Collins et al. 2005; Gies, Glanz et al. 2009; Schmalwieser et al. 2010, 2021; Bodekær et al. 2015; Grandahl, Eriksen et al. 2018; Peters, Pasko et al. 2019).
Personal UVR exposure is measured with dosimeters and is often reported as workday standard erythemal dose (SED). 1 SED = 100 J/m2 using the Commission Internationale de l’eclairage (CIE) action spectrum normalized to 298 nm (Diffey et al. 1997). Measured exposure depends on type of dosimeter and its location on the body (Knuschke et al. 2007; Serrano et al. 2013; Cherrie et al. 2021).
UVR exposure is the main stimulus of vitamin D synthesis in the skin, has an almost immediate effect on vitamin D levels, that are essential for a healthy musculoskeletal system, and may have a protective effect for a range of diseases (Lucas et al. 2019). The exposure-response relation between UVR exposure and vitamin D level is well characterized (Webb et al. 2021).
UVR exposure has acute adverse effects on the skin and eye and causes sunburn, photo keratitis, and immunosuppression (Harrison and Young 2002; Young 2006). Long-term or intermittent UVR exposure increases the risk of malignant melanoma, squamous cell carcinoma and basal cell carcinoma of the skin, ocular melanoma, and possibly other cancers (IARC Working Group on the Evaluation of Carcinogenic Risks to Humans 2012; Holm-Schou et al. 2020; World Health Organization 2021; Chang et al. 2022), as well as cataract and macular degeneration (Roberts 2011; Löfgren 2017). However, the exposure–response relations with quantitative measures of UVR exposure for these outcomes are not well characterized, which makes it difficult to recommend exposure levels where the accrued benefits are offset by accumulated hazards (Lucas and Ponsonby 2002; Weller 2024).
Epidemiological studies of non-acute health effects of occupational solar UVR exposure have mainly relied on surrogate exposure measures, such as occupation, work task or time worked outdoors (World Health Organization 2021), or case-by-case expert assessment (Schmitt et al. 2018). Expert assessment-based job-exposure matrices (JEM) for occupational solar UVR exposure have been developed (Kauppinen et al. 2009; Peters et al. 2012) but without fully quantitative assessment of UVR exposure, exposure–response relations cannot be characterized and quantified.
Quantitative JEMs combining job-specific individual as well as published summary exposure levels with expert ratings have proven able to characterize well-documented exposure–response relations for e.g. diesel, asbestos, silica, and endotoxin exposure in the general working population and risk of lung cancer, asbestosis, and hypersensitivity pneumonitis and is a promising way forward (Ge et al. 2020, 2020; Iversen et al. 2024, 2024).
The objective of this study is to develop a quantitative JEM for occupational solar UVR exposure in the general working population of Europe.
Material and methods
The JEM is constructed by statistical modelling of the combination of 2 datasets: (i) Ordinal expert ratings of the daily average duration of outdoor work for all occupations classified by the European version of the International Standard Classification of Occupations from 1988 (ISCO-88(COM)) (Eurostat 2023) and (ii) summary workday UVR exposure obtained with personal dosimeters published in the peer-reviewed literature.
Expert ratings
Three experts in occupational health from Norway, Sweden, and Finland (Northern Europe), 3 from Germany, United Kingdom, and Denmark (Central Europe), and 3 from Spain, Italy, and Croatia (Southern Europe) rated the average duration of outdoor work for an average worker of all 372 ISCO-88(COM) occupations as 0, 1 to 2, 3 to 4, or ≥5 h per workday. They also for each ISCO-88(COM) occupation estimated the proportion of workers that at a given point in time work outdoor (0, 1 to 24, 25 to 49, or ≥50%). This proportion was used as an estimate of the probability of outdoor work. Both rating schemes were identical to earlier rating schemes used for a quantitative JEM for daytime light exposure (Vested et al. 2019). The 3 European regions were roughly defined by a combination of latitude (35 to 45, 46 to 57, ≥58 °N) and climate (Mediterranean, oceanic, and subarctic) that may affect exposure and sun protective behavior, as well as the residence of the experts and thus knowledge about regional working conditions.
The expert ratings were conducted in 4 steps: In step 1, the 9 experts rated all occupations individually. In step 2, the 3 experts of each region met, discussed the work content of the occupations and rating discrepancies, and agreed on the ratings for each occupation within each of the respective regions. In step 3, one representative from each region met, discussed work content and rating discrepancies. In step 4, each region representative re-rated the regional ratings. Due to high inter-regional agreement (as shown later), the rating for an average worker of each occupation in Europe was obtained as the mean of the regional ratings of step 4, rounded upwards to the nearest integer.
Workday UVR exposure
Literature search
We performed a systematic search in PubMed for peer reviewed papers in English published between 1960 and May 2, 2022, to identify journal articles providing quantitative information on occupational solar UVR exposure obtained by personal sampling.
Eligibility criteria for inclusion: (i) Personal occupational solar UVR exposure measured with polysulphone film (PSF), or Gigahertz, SunSaver, or Scienterra electronic dosimeters. (ii) Solar UVR exposure reported as the erythemally effective irradiance in accordance with the sensitivity curve of the erythema action spectrum (International Commission on Illumination 2019) and expressed as standard erythema dose (SED) or erythemal weighted J/m2. (iii) Dosimeters calibrated toward a reference spectrophotometer or broadband radiometer during exposure to the sun or a well-defined UVR-source. (iv) UVR exposure expressed as the arithmetic mean SED of a workday or results provided that made it possible to compute arithmetic mean SED.
Criteria for exclusion: (i) Personal occupational solar UVR exposure measured with biological dosimeters that we could not harmonize with measurements obtained with other dosimeters (see later). (ii) Papers reporting results presented in other papers but with less measurements. (iii) Reviews. (iv) UVR exposure measurements from extreme working conditions above 3000 m, or in the Artic or the Antarctic. (v) Measurements below the European lower latitude of 34˚N (or 34 °S).
Details of the search and selection of published studies are provided in Supplementary Appendix 1.
Data extraction
We extracted arithmetic mean summary workday UVR exposure (or data that made it possible to compute this) together with occupational title, measurement month, latitude, altitude, dosimeter type, dosimeter location on the body, and number of hours, workdays, and workers measured. For each published study, we generated one record per occupation, season, and latitude as the analytical unit. We defined spring as March to May, summer as June to August, autumn as September to November and winter as December to February, as we only identified measurements from the northern hemisphere. When measurements were published as several month-specific levels, we calculated the arithmetic mean season-specific exposure weighted by the number of measurement days. When published measurements were obtained across several latitudes and latitude-specific results were not presented, UVR exposure was assigned the mean latitude of the measurements.
We coded all occupational titles according to ISCO-88(COM) at the 4-digit level (Table S1). Most workers were recruited because they had occupations typically including a high proportion of outdoor work. Some studies only included measurements conducted during outdoor work, thus obtaining higher exposure than on an average workday, and we recorded this oversampling. UVR exposure reported as erythemal weighted J/m2 were converted to SED. Arithmetic mean UVR exposure was calculated if not reported and sufficient information was available in the publications (Gies et al. 2009) or was provided by the authors (Thieden et al. 2005; Grandahl et al. 2018; Wittlich 2022). From the Genesis project we had access to non-published occupational UVR exposure from several occupations all over Germany (Wittlich 2022). When latitude or altitude were not reported, we assessed this from the location where the measurements took place. There were no missing values for the other variables.
Harmonization of UVR exposure
Dosimeter type
Because measured UVR exposure may vary across dosimeter types, we harmonized UVR exposure obtained with PSF dosimeters to UVR exposure obtained with the Gigahertz X2012-10 dosimeter as previously reported by Strehl et al. (Strehl et al. 2021). In brief, 5 PSF dosimeters and 8 Gigahertz X2012 dosimeters were directly exposed to the sun when mounted on a sun tracking device that secured a constant incident angle of solar radiation of 0, 30, 110, 60, or 85° during each of 8 measurement days that typically lasted from 07:00 h to 17:00 h. A mobile weather station (MAWS201, Vaisala, Hamburg, Germany) equipped with 3 identical pyranometers were used to assess global, ambient, and reflected UVR exposure that showed good agreement with 3 static Gigahertz X20212-10 dosimeters. All measurements were conducted in May, June, and September 2020 in St. Augustin, Germany (50.8° N, 7.2° E, 65 m above sea level). According to the measurements, the PSF dosimeters overestimated exposure by a factor 1.3, and we thus used a correction factor of 0.77 (1/1.3) for all PSF measurements. This study also measured VioSpore biological dosimeters but was not able to obtain an intercalibration factor with the Gigahertz X2012 dosimeter (Strehl et al. 2021).
During 10 d in April, July, and August 2022, we followed a similar protocol comparing the Sunsaver and Scienterra dosimeters with the Gigahertz X2012-10 dosimeter during direct exposure to the sun at fixed incident angles of 0°, 30°, 60°, and 85° from 7:30 am to 3:30 pm in St. Augustin, Germany. The mean SED values were used to calculate correction factors of 1.10 and 1.0 for the Sunsaver and Scienterra dosimeters.
Dosimeter location
UVR exposure measured on different parts of the body were harmonized to upper arm exposure by multiplying by the following factors: forehead 1.16, neck 0.57, shoulder 0.59, chest 1.37, wrist 1.11, and lower back 0.81. These factors were calculated from parallel UVR exposure measurements previously obtained with PSF dosimeters on workers during outdoor sun exposure (Knuschke et al. 2007). We harmonized to the upper arm because most included measurements were conducted on the upper arm (Table 1).
Characteristics of 12 studies of personal occupational solar UVR measurements
| Reference | Latitude | Altitude (m) | Month | Dosimeter | Dosimeter location | Measurement hours | Country | Occupations, n | Partici-pants | Recordsa | Workday measurements | Oversampling of outdoor workdays |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| (Thieden, Collins et al. 2005) | 53-55° N | 20 | May-August | SunSaver | Wrist | 7 am-7 pm | Denmark, Ireland | 1 | 53 | 2 | 3388 | Yes |
| (Gies, Glanz et al. 2009) | 41-45° N | 10-330 | June-July | PSF | Wrist | 8 am-4 pm | US | 1 | 84 | 2 | 168 | Yes |
| (Schmalwieser et al. 2010) | 48° N | 200-700 | April-October | Gigahertz X2000-10 | Forehead | Workhours | Austria | 1 | 12 | 1 | 1427 | No |
| (Bodekær et al. 2015) | 41-55° N | 5-350 | May-September | SunSaver | Wrist | 7 am-7 pm | Austria, Denmark, Poland, Spain | 1 | 113 | 4 | 12976 | Yes |
| (Peters, Demers et al. 2016) | 49° N | NA | July-September | Scienterra | Forehead, chest or wrist | 8 am-5 pm | Canada | 2 | 73 | 3 | 318 | Yes |
| (Grandahl, Eriksen et al. 2018) | 55° N | 20 | April-September | Scienterra | Wrist | 7 am-7 pm | Denmark | 16 | 504 | 51 | 4641 | No |
| Modenese 2019 (Modenese et al. 2019) | 43° N | 0-10 | May | Gigahertz X2000, Gigahertz X2012 | Lower back, chest or neck | 7 am-10 am/2 pm | Italy | 1 | 7 | 4 | 18 | No |
| (Peters, Pasko et al. 2019) | 42-49° N | NA | July-September | PSF | Forehead, shoulder or wrist | Workday | Canada | 1 | 348 | 6 | 348 | Yes |
| (Moldovan et al. 2020) | 44-46° N | 90-320 | April-October | Gigahertz X2012-10 | Upper arm | 7 am- 5 pm | Romania | 1 | 27 | 6 | 724 | Yes |
| (Rydz et al. 2020) | 49-55° N | 700 (mean) | June-September | Scienterra | Forehead, chest or wrist | 7 am- 5 pm | Canada | 5 | 179 | 5 | 883 | Yes |
| (Wittlich et al. 2020) | 43-56° N | 20-340 | April-October | Gigahertz X2012-10 | Upper arm | 7 am- 5 pm | Croatia, Denmark, Italy, Romania | 1 | 104 | 14 | 1994 | Yes (Romania) |
| (Wittlich 2022) | 43-55° N | 1-999 | April-October | Gigahertz X2012-10 | Upper arm | Workday | Germany | 43 | 1141 | 125 | 48826 | No |
| Reference | Latitude | Altitude (m) | Month | Dosimeter | Dosimeter location | Measurement hours | Country | Occupations, n | Partici-pants | Recordsa | Workday measurements | Oversampling of outdoor workdays |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| (Thieden, Collins et al. 2005) | 53-55° N | 20 | May-August | SunSaver | Wrist | 7 am-7 pm | Denmark, Ireland | 1 | 53 | 2 | 3388 | Yes |
| (Gies, Glanz et al. 2009) | 41-45° N | 10-330 | June-July | PSF | Wrist | 8 am-4 pm | US | 1 | 84 | 2 | 168 | Yes |
| (Schmalwieser et al. 2010) | 48° N | 200-700 | April-October | Gigahertz X2000-10 | Forehead | Workhours | Austria | 1 | 12 | 1 | 1427 | No |
| (Bodekær et al. 2015) | 41-55° N | 5-350 | May-September | SunSaver | Wrist | 7 am-7 pm | Austria, Denmark, Poland, Spain | 1 | 113 | 4 | 12976 | Yes |
| (Peters, Demers et al. 2016) | 49° N | NA | July-September | Scienterra | Forehead, chest or wrist | 8 am-5 pm | Canada | 2 | 73 | 3 | 318 | Yes |
| (Grandahl, Eriksen et al. 2018) | 55° N | 20 | April-September | Scienterra | Wrist | 7 am-7 pm | Denmark | 16 | 504 | 51 | 4641 | No |
| Modenese 2019 (Modenese et al. 2019) | 43° N | 0-10 | May | Gigahertz X2000, Gigahertz X2012 | Lower back, chest or neck | 7 am-10 am/2 pm | Italy | 1 | 7 | 4 | 18 | No |
| (Peters, Pasko et al. 2019) | 42-49° N | NA | July-September | PSF | Forehead, shoulder or wrist | Workday | Canada | 1 | 348 | 6 | 348 | Yes |
| (Moldovan et al. 2020) | 44-46° N | 90-320 | April-October | Gigahertz X2012-10 | Upper arm | 7 am- 5 pm | Romania | 1 | 27 | 6 | 724 | Yes |
| (Rydz et al. 2020) | 49-55° N | 700 (mean) | June-September | Scienterra | Forehead, chest or wrist | 7 am- 5 pm | Canada | 5 | 179 | 5 | 883 | Yes |
| (Wittlich et al. 2020) | 43-56° N | 20-340 | April-October | Gigahertz X2012-10 | Upper arm | 7 am- 5 pm | Croatia, Denmark, Italy, Romania | 1 | 104 | 14 | 1994 | Yes (Romania) |
| (Wittlich 2022) | 43-55° N | 1-999 | April-October | Gigahertz X2012-10 | Upper arm | Workday | Germany | 43 | 1141 | 125 | 48826 | No |
aA record is defined by study, occupation, season and latitude, NA not available, PSF Polysulphone film.
Characteristics of 12 studies of personal occupational solar UVR measurements
| Reference | Latitude | Altitude (m) | Month | Dosimeter | Dosimeter location | Measurement hours | Country | Occupations, n | Partici-pants | Recordsa | Workday measurements | Oversampling of outdoor workdays |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| (Thieden, Collins et al. 2005) | 53-55° N | 20 | May-August | SunSaver | Wrist | 7 am-7 pm | Denmark, Ireland | 1 | 53 | 2 | 3388 | Yes |
| (Gies, Glanz et al. 2009) | 41-45° N | 10-330 | June-July | PSF | Wrist | 8 am-4 pm | US | 1 | 84 | 2 | 168 | Yes |
| (Schmalwieser et al. 2010) | 48° N | 200-700 | April-October | Gigahertz X2000-10 | Forehead | Workhours | Austria | 1 | 12 | 1 | 1427 | No |
| (Bodekær et al. 2015) | 41-55° N | 5-350 | May-September | SunSaver | Wrist | 7 am-7 pm | Austria, Denmark, Poland, Spain | 1 | 113 | 4 | 12976 | Yes |
| (Peters, Demers et al. 2016) | 49° N | NA | July-September | Scienterra | Forehead, chest or wrist | 8 am-5 pm | Canada | 2 | 73 | 3 | 318 | Yes |
| (Grandahl, Eriksen et al. 2018) | 55° N | 20 | April-September | Scienterra | Wrist | 7 am-7 pm | Denmark | 16 | 504 | 51 | 4641 | No |
| Modenese 2019 (Modenese et al. 2019) | 43° N | 0-10 | May | Gigahertz X2000, Gigahertz X2012 | Lower back, chest or neck | 7 am-10 am/2 pm | Italy | 1 | 7 | 4 | 18 | No |
| (Peters, Pasko et al. 2019) | 42-49° N | NA | July-September | PSF | Forehead, shoulder or wrist | Workday | Canada | 1 | 348 | 6 | 348 | Yes |
| (Moldovan et al. 2020) | 44-46° N | 90-320 | April-October | Gigahertz X2012-10 | Upper arm | 7 am- 5 pm | Romania | 1 | 27 | 6 | 724 | Yes |
| (Rydz et al. 2020) | 49-55° N | 700 (mean) | June-September | Scienterra | Forehead, chest or wrist | 7 am- 5 pm | Canada | 5 | 179 | 5 | 883 | Yes |
| (Wittlich et al. 2020) | 43-56° N | 20-340 | April-October | Gigahertz X2012-10 | Upper arm | 7 am- 5 pm | Croatia, Denmark, Italy, Romania | 1 | 104 | 14 | 1994 | Yes (Romania) |
| (Wittlich 2022) | 43-55° N | 1-999 | April-October | Gigahertz X2012-10 | Upper arm | Workday | Germany | 43 | 1141 | 125 | 48826 | No |
| Reference | Latitude | Altitude (m) | Month | Dosimeter | Dosimeter location | Measurement hours | Country | Occupations, n | Partici-pants | Recordsa | Workday measurements | Oversampling of outdoor workdays |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| (Thieden, Collins et al. 2005) | 53-55° N | 20 | May-August | SunSaver | Wrist | 7 am-7 pm | Denmark, Ireland | 1 | 53 | 2 | 3388 | Yes |
| (Gies, Glanz et al. 2009) | 41-45° N | 10-330 | June-July | PSF | Wrist | 8 am-4 pm | US | 1 | 84 | 2 | 168 | Yes |
| (Schmalwieser et al. 2010) | 48° N | 200-700 | April-October | Gigahertz X2000-10 | Forehead | Workhours | Austria | 1 | 12 | 1 | 1427 | No |
| (Bodekær et al. 2015) | 41-55° N | 5-350 | May-September | SunSaver | Wrist | 7 am-7 pm | Austria, Denmark, Poland, Spain | 1 | 113 | 4 | 12976 | Yes |
| (Peters, Demers et al. 2016) | 49° N | NA | July-September | Scienterra | Forehead, chest or wrist | 8 am-5 pm | Canada | 2 | 73 | 3 | 318 | Yes |
| (Grandahl, Eriksen et al. 2018) | 55° N | 20 | April-September | Scienterra | Wrist | 7 am-7 pm | Denmark | 16 | 504 | 51 | 4641 | No |
| Modenese 2019 (Modenese et al. 2019) | 43° N | 0-10 | May | Gigahertz X2000, Gigahertz X2012 | Lower back, chest or neck | 7 am-10 am/2 pm | Italy | 1 | 7 | 4 | 18 | No |
| (Peters, Pasko et al. 2019) | 42-49° N | NA | July-September | PSF | Forehead, shoulder or wrist | Workday | Canada | 1 | 348 | 6 | 348 | Yes |
| (Moldovan et al. 2020) | 44-46° N | 90-320 | April-October | Gigahertz X2012-10 | Upper arm | 7 am- 5 pm | Romania | 1 | 27 | 6 | 724 | Yes |
| (Rydz et al. 2020) | 49-55° N | 700 (mean) | June-September | Scienterra | Forehead, chest or wrist | 7 am- 5 pm | Canada | 5 | 179 | 5 | 883 | Yes |
| (Wittlich et al. 2020) | 43-56° N | 20-340 | April-October | Gigahertz X2012-10 | Upper arm | 7 am- 5 pm | Croatia, Denmark, Italy, Romania | 1 | 104 | 14 | 1994 | Yes (Romania) |
| (Wittlich 2022) | 43-55° N | 1-999 | April-October | Gigahertz X2012-10 | Upper arm | Workday | Germany | 43 | 1141 | 125 | 48826 | No |
aA record is defined by study, occupation, season and latitude, NA not available, PSF Polysulphone film.
Oversampling
To account for oversampling of measurements obtained only during outdoor workdays, we weighted UVR exposure by the regional expert estimates of outdoor work probability elaborated in the statistical methods section. Weighting factors of 0.13, 0.38, and 0.75 were used for expert estimates of 1 to 24%, 25 to 49%, and ≥50% probability, respectively.
Thus, all measurements were multiplied with factors for dosimeter type and dosimeter location to provide harmonized exposure as obtained with a Gigahertz X2012-10 dosimeter located at the upper arm. Measurements were corrected for oversampling of outdoor workdays as described in the Statistical methods section.
Statistical methods
We assessed intra- and inter-region agreements of the expert ratings of all 372 ISCO-88(COM) occupations by quadratic-weighted Cohen’s kappa, which takes the ordinal relationship between the rating categories into account (Sim and Wright 2005).
We fitted the harmonized arithmetic mean UVR exposure by linear mixed effect models using restricted maximum likelihood estimation (Littell 2006). The modelling was performed on the natural logarithmic scale, primarily for 2 reasons. First, this showed the best fit to the right-skewed data in terms of normality of the residuals as well as magnitude of the estimated random error as compared to the between occupations variation. Second, the implied multiplicative structure of the model seemed appropriate from an intuitive perspective (e.g. on an absolute scale, seasonal changes could be anticipated to have largest impact for persons with most outdoor work). Measurement records were weighted by the number of participants; the weights were standardized to have a mean of 1. Fixed effects were expert rating (0, 1 to 2, 3 to 4, ≥5 h), season (spring, summer, autumn), latitude (continuous), duration of daily measurements (difference between starting and ending hours, continuous). Study and ISCO-88(COM) occupation were random effects.
The random effect terms were assumed statistically independent and normally distributed with means 0 and 2 different variance components representing the between study variance and the between ISCO-88(COM) occupations variance. We obtained best linear unbiased predictions (BLUPs) of the coefficients for the ISCO-88(COM) occupations with information on UVR exposure.
To provide estimates for absolute exposure, we retransformed the model-estimated intercept on the log-scale to the original scale. We exploited that the mean of a log-normally distributed variable is the exponentiated value of the log-scale mean plus half of the log-scale variance. In this retransformation, the variance was calculated as the sum of the estimated between occupations, between studies, and residual variances. For occupations with published measurements, we provided an additional estimate obtained by adding the occupation-specific BLUPs to the relevant expert rating class means and retransforming the results from log-scale to the original scale. This retransformation used an approach similar to that described above but included only the between study and the residual variance in the variance calculation.
Data were corrected for oversampling of outdoor working days by multiplying the expert-assessed probability of outdoor work with the difference between the observed values and the estimated mean exposure for indoor work (expert rating of 0 h of outdoor work) in an iterative process.
The mixed effect model was fitted using SAS Studio 3.8 (SAS Institute Inc., Cary, NC, USA). Remaining analyses were performed using STATA 17.0 (StataCorp LP, College Station, TX, USA).
Results
The 3 experts of each of the 3 European regions showed high inter-rater agreement for duration (Cohen’s kappa 0.70 to 0.78) and probability (Cohen’s kappa 0.68 to 0.74) of outdoor work for the 372 ISCO-88(COM) occupations before they met and discussed discrepancies (step 1) (Table S2). Inter-regional agreement after one representative from each region had discussed discrepancies and re-rated the occupations (step 4) was higher; Cohen’s kappa was 0.80 for exposure duration and 0.81 for exposure probability (Table S3). The common European rating classified 164 occupations (44%) as null-exposed, 116 (31%) as exposed 1 to 2 h, 56 (15%) as exposed 3 to 4 h, and 36 (10%) as exposed ≥5 h during a workday.
A total of 12 studies published between 2005 and 2022 and conducted during spring, summer, or autumn at latitudes between 41 and 55 °N and at altitudes <1000 m above sea level fulfilled the inclusion criteria (Table 1). Most studies measured UVR exposure during a predefined period of a workday, typically 7:00 am—5:00 pm or 7:00 am—7:00 pm. Gigahertz, Scienterra, Sunsaver, and PSF dosimeters were used in 5, 3, 2, and 2 studies, respectively. The dosimeters were located on forehead, neck, shoulder, chest, upper arm, wrist, or lower back. Seven and partly one of the studies oversampled outdoor workdays.
A total of 223 study-occupation-season-latitude records provided UVR exposure estimates for 49 occupations based on 75,711 personal workday measurements obtained from 2,645 participants (Table S4). The 49 occupations represented 1 occupation classified as null-exposed by the experts, 8 (16%) as exposed 1 to 2 h, 21 (43%) as exposed 3 to 4 h and 19 (39%) as exposed ≥5 h. Overall, the harmonized and probability weighted UVR exposure levels were lower than the original exposure levels, especially for those obtained with PSF dosimeters, dosimeters located at neck and shoulder, and for measurements oversampled on outdoor workdays.
Table 2 presents the mixed-effects model estimates (β) and the corresponding arithmetic mean ratios with 95% confidence intervals for the fixed effects along with the arithmetic means for a reference workday in the highest exposed expert rating class as well as the 3 remaining rating classes during summer at latitude 50 and based on 8-h measurements. UVR exposure increased monotonically with higher expert rating and was more than 3-fold higher for the highest (≥5 h) compared with the lowest rating (0 h). Autumn and spring UVR exposure were 46% and 92% of the summer exposure. UVR exposure declined with increasing latitude (3% per degree). No significant trend was seen by number of daily hours measured.
Fixed effects model parameters from a linear mixed effects model for personal occupational solar UVR exposure
| Model parameters | Number of records | β-estimate (95% CI) | Arithmetic mean ratio (95% CI) |
|---|---|---|---|
| Expert rating, duration of outdoor work, hours | |||
| 0 | 3 | −1.30 (−2.21; −-0.39) | 0.27 (0.11; 0.68) |
| 1–2 | 21 | −0.46 (−0.88; −0.04) | 0.63 (0.41; 0.96) |
| 3–4 | 78 | −0.32 (−0.63; −0.02) | 0.72 (0.53; 0.98) |
| ≥5 | 121 | Ref. a | Ref. a |
| Season | |||
| Spring | 67 | −0.08 (−0.19; 0.02) | 0.92 (0.82; 1.03) |
| Summer | 89 | Ref. | Ref. |
| Autumn | 67 | −0.77 (−0.89; −0.66) | 0.46 (0.41; 0.52) |
| Latitude (°), reference 50° N | 223 | −0.03 (−0.06; −0.00) | 0.97 (0.94; 1.00) |
| Duration of measurement (hours), reference 8 hours | 223 | 0.03 (−0.12; 0.17) | 1.03 (0.89; 1.19) |
| Model parameters | Number of records | β-estimate (95% CI) | Arithmetic mean ratio (95% CI) |
|---|---|---|---|
| Expert rating, duration of outdoor work, hours | |||
| 0 | 3 | −1.30 (−2.21; −-0.39) | 0.27 (0.11; 0.68) |
| 1–2 | 21 | −0.46 (−0.88; −0.04) | 0.63 (0.41; 0.96) |
| 3–4 | 78 | −0.32 (−0.63; −0.02) | 0.72 (0.53; 0.98) |
| ≥5 | 121 | Ref. a | Ref. a |
| Season | |||
| Spring | 67 | −0.08 (−0.19; 0.02) | 0.92 (0.82; 1.03) |
| Summer | 89 | Ref. | Ref. |
| Autumn | 67 | −0.77 (−0.89; −0.66) | 0.46 (0.41; 0.52) |
| Latitude (°), reference 50° N | 223 | −0.03 (−0.06; −0.00) | 0.97 (0.94; 1.00) |
| Duration of measurement (hours), reference 8 hours | 223 | 0.03 (−0.12; 0.17) | 1.03 (0.89; 1.19) |
aThe estimated intercept on the logarithmic scale was 0.71 (95% CI: 0.31; 1.11) corresponding to an estimated arithmetic mean of 2.49 (95% CI: 1.66; 3.72) in the highest exposed expert rating class during summer at latitude 50 and based on 8-hour measurements. The estimated exposures for the lower 3 expert rating classes were 0.68 (95% CI: 0.26; 1.75) for 0 h, 1.57 (95% CI: 0.95; 2.59) for 1 to 2 h, and 1.80 (95% CI: 1.20; 2.70) for 3 to 4 h.
Fixed effects model parameters from a linear mixed effects model for personal occupational solar UVR exposure
| Model parameters | Number of records | β-estimate (95% CI) | Arithmetic mean ratio (95% CI) |
|---|---|---|---|
| Expert rating, duration of outdoor work, hours | |||
| 0 | 3 | −1.30 (−2.21; −-0.39) | 0.27 (0.11; 0.68) |
| 1–2 | 21 | −0.46 (−0.88; −0.04) | 0.63 (0.41; 0.96) |
| 3–4 | 78 | −0.32 (−0.63; −0.02) | 0.72 (0.53; 0.98) |
| ≥5 | 121 | Ref. a | Ref. a |
| Season | |||
| Spring | 67 | −0.08 (−0.19; 0.02) | 0.92 (0.82; 1.03) |
| Summer | 89 | Ref. | Ref. |
| Autumn | 67 | −0.77 (−0.89; −0.66) | 0.46 (0.41; 0.52) |
| Latitude (°), reference 50° N | 223 | −0.03 (−0.06; −0.00) | 0.97 (0.94; 1.00) |
| Duration of measurement (hours), reference 8 hours | 223 | 0.03 (−0.12; 0.17) | 1.03 (0.89; 1.19) |
| Model parameters | Number of records | β-estimate (95% CI) | Arithmetic mean ratio (95% CI) |
|---|---|---|---|
| Expert rating, duration of outdoor work, hours | |||
| 0 | 3 | −1.30 (−2.21; −-0.39) | 0.27 (0.11; 0.68) |
| 1–2 | 21 | −0.46 (−0.88; −0.04) | 0.63 (0.41; 0.96) |
| 3–4 | 78 | −0.32 (−0.63; −0.02) | 0.72 (0.53; 0.98) |
| ≥5 | 121 | Ref. a | Ref. a |
| Season | |||
| Spring | 67 | −0.08 (−0.19; 0.02) | 0.92 (0.82; 1.03) |
| Summer | 89 | Ref. | Ref. |
| Autumn | 67 | −0.77 (−0.89; −0.66) | 0.46 (0.41; 0.52) |
| Latitude (°), reference 50° N | 223 | −0.03 (−0.06; −0.00) | 0.97 (0.94; 1.00) |
| Duration of measurement (hours), reference 8 hours | 223 | 0.03 (−0.12; 0.17) | 1.03 (0.89; 1.19) |
aThe estimated intercept on the logarithmic scale was 0.71 (95% CI: 0.31; 1.11) corresponding to an estimated arithmetic mean of 2.49 (95% CI: 1.66; 3.72) in the highest exposed expert rating class during summer at latitude 50 and based on 8-hour measurements. The estimated exposures for the lower 3 expert rating classes were 0.68 (95% CI: 0.26; 1.75) for 0 h, 1.57 (95% CI: 0.95; 2.59) for 1 to 2 h, and 1.80 (95% CI: 1.20; 2.70) for 3 to 4 h.
In Table S5, we present the estimated variance components of the statistical model. In the basic model, that only included study and ISCO-88(COM) occupation as random effects; the variance between studies and occupations, and the residual variance constituted 10, 42, and 48% of the total variance. When we in the intermediate model added season, latitude, and duration of the measurements as fixed effects, the total variance as well as the variance between studies and between ISCO-88(COM) occupations increased while the residual variance decreased and the model fit improved. When we in the full model furthermore added expert ratings, the variance between occupations decreased by 24%, while minor changes were seen for between studies and residual variances.
Table 3 presents model-based UVR exposure for the expert ratings, and full-model UVR exposure for the 49 occupations with UVR exposure measurements. For most occupations (65%), UVR exposure estimates were based on one study. Only one occupation, secretaries, was classified with an expert rating of 0 h and showed a model-based exposure of 0.62 SED.
| ISCO-88(COM) | ISCO-88(COM) text | Studies | Recordsc | Participants | Workdays | Expert ratingd | Expert estimatese | Full- model estimatesf |
|---|---|---|---|---|---|---|---|---|
| 9211 | Farm-hands and laborers | 1 | 3 | 12 | 455 | ≥ 5 | 2.49 | 3.38 |
| 7131 | Roofers | 2 | 8 | 113 | 3650 | ≥ 5 | 2.49 | 3.28 |
| 7123 | Concrete placers, concrete finishers, and related workers | 2 | 9 | 78 | 2296 | ≥ 5 | 2.49 | 3.22 |
| 7111 | Miners and quarry workers | 1 | 3 | 15 | 711 | 3-4 | 1.80 | 2.91 |
| 7122 | Bricklayers and stonemasons | 3 | 23 | 145 | 2857 | ≥ 5 | 2.49 | 2.70 |
| 7129 | Building frame and related trades workers not elsewhere classified | 3 | 7 | 134 | 2069 | ≥ 5 | 2.49 | 2.68 |
| 9312 | Construction and maintenance laborers: roads, dams and similar constructions | 5 | 15 | 581 | 5171 | ≥ 5 | 2.49 | 2.68 |
| 8113 | Well drillers and borers and related workers | 1 | 3 | 3 | 75 | ≥ 5 | 2.49 | 2.66 |
| 7136 | Plumbers and pipe fitters | 1 | 3 | 8 | 76 | 3–4 | 1.80 | 2.50 |
| 7112 | Shotfirers and blasters | 1 | 3 | 3 | 55 | ≥ 5 | 2.49 | 2.47 |
| 7214 | Structural-metal preparers and erectors | 1 | 3 | 15 | 620 | 3–4 | 1.80 | 2.43 |
| 7245 | Electrical line installers, repairers, and cable jointers | 1 | 3 | 68 | 1834 | 3–4 | 1.80 | 2.38 |
| 2211 | Biologists, botanists, zoologists, and related professionals | 1 | 2 | 3 | 56 | 3–4 | 1.80 | 2.36 |
| 6112 | Gardeners, horticultural and nursery growers | 4 | 9 | 249 | 8864 | ≥ 5 | 2.49 | 2.26 |
| 3212 | Agronomy and forestry technicians | 1 | 3 | 14 | 571 | ≥ 5 | 2.49 | 2.19 |
| 8332 | Earth-moving and related plant operators | 2 | 6 | 64 | 2552 | 3−4 | 1.80 | 2.19 |
| 9330 | Transport laborers and freight handlers | 1 | 3 | 62 | 2830 | 3−4 | 1.80 | 2.19 |
| 3213 | Farming and forestry advisers | 1 | 3 | 8 | 259 | 3−4 | 1.80 | 2.15 |
| 7423 | Woodworking machine setters and setter-operators | 1 | 3 | 9 | 362 | 1−2 | 1.57 | 2.14 |
| 6111 | Field crop and vegetable growers | 2 | 7 | 158 | 13424 | ≥ 5 | 2.49 | 2.10 |
| 4142 | Mail carriers and sorting clerks | 2 | 6 | 83 | 2659 | 3−4 | 1.80 | 2.08 |
| 9162 | Sweepers and related laborers | 2 | 6 | 74 | 1766 | ≥ 5 | 2.49 | 2.07 |
| 7124 | Carpenters and joiners | 2 | 6 | 74 | 1932 | 3−4 | 1.80 | 2.06 |
| 9313 | Building construction laborers | 1 | 3 | 3 | 123 | ≥ 5 | 2.49 | 2.03 |
| 7233 | Agricultural- or industrial-machinery mechanics and fitters | 1 | 3 | 60 | 2912 | 1−2 | 1.57 | 1.92 |
| 9161 | Garbage collectors | 2 | 5 | 66 | 2812 | ≥ 5 | 2.49 | 1.87 |
| 8312 | Railway brakers, signalers and shunters | 1 | 3 | 7 | 208 | ≥ 5 | 2.49 | 1.77 |
| 2148 | Cartographers and surveyors | 2 | 5 | 30 | 1224 | 3−4 | 1.80 | 1.74 |
| 6130 | Crop and animal producers | 2 | 4 | 69 | 3807 | ≥ 5 | 2.49 | 1.71 |
| 3474 | Clowns, magicians, acrobats and related associate professionals | 1 | 2 | 5 | 123 | 1−2 | 1.57 | 1.69 |
| 6152 | Inland and coastal waters fishery workers | 2 | 7 | 15 | 320 | ≥ 5 | 2.49 | 1.69 |
| 3475 | Athletes, sports persons and related associate professionals | 2 | 4 | 40 | 532 | 3−4 | 1.80 | 1.56 |
| 5169 | Protective services workers not elsewhere classified | 3 | 6 | 122 | 1244 | 3−4 | 1.80 | 1.55 |
| 7133 | Plasterers | 1 | 3 | 8 | 306 | 3−4 | 1.80 | 1.53 |
| 7213 | Sheet-metal workers | 1 | 3 | 3 | 71 | 3−4 | 1.80 | 1.53 |
| 8163 | Incinerator, water-treatment and related plant operators | 1 | 3 | 5 | 185 | 1−2 | 1.57 | 1.52 |
| 7141 | Painters and related workers | 1 | 1 | 1 | 26 | 3−4 | 1.80 | 1.51 |
| 8324 | Heavy truck and lorry drivers | 1 | 3 | 26 | 1230 | 1−2 | 1.57 | 1.51 |
| 5162 | Police officers | 1 | 3 | 3 | 75 | 3−4 | 1.80 | 1.50 |
| 8334 | Lifting-truck operators | 1 | 3 | 38 | 327 | 3−4 | 1.80 | 1.50 |
| 6141 | Forestry workers and loggers | 1 | 3 | 18 | 813 | ≥ 5 | 2.49 | 1.47 |
| 2213 | Agronomists and related professionals | 1 | 3 | 3 | 95 | 3−4 | 1.80 | 1.39 |
| 2142 | Civil engineers | 1 | 1 | 16 | 49 | 1−2 | 1.57 | 1.13 |
| 3145 | Air traffic safety technicians | 1 | 3 | 14 | 241 | 1−2 | 1.57 | 1.07 |
| 3143 | Aircraft pilots and related associate professionals | 1 | 3 | 4 | 166 | 1−2 | 1.57 | 0.96 |
| 8141 | Wood-processing-plant operators | 1 | 3 | 6 | 210 | 3−4 | 1.80 | 0.90 |
| 3320 | Pre-primary education teaching associate professionals | 1 | 3 | 62 | 3005 | 3−4 | 1.80 | 0.63 |
| 4115 | Secretaries | 1 | 3 | 40 | 403 | 0 | 0.68 | 0.62 |
| 5123 | Waiters, waitresses and bartenders | 1 | 3 | 8 | 60 | 3−4 | 1.80 | 0.53 |
| ISCO-88(COM) | ISCO-88(COM) text | Studies | Recordsc | Participants | Workdays | Expert ratingd | Expert estimatese | Full- model estimatesf |
|---|---|---|---|---|---|---|---|---|
| 9211 | Farm-hands and laborers | 1 | 3 | 12 | 455 | ≥ 5 | 2.49 | 3.38 |
| 7131 | Roofers | 2 | 8 | 113 | 3650 | ≥ 5 | 2.49 | 3.28 |
| 7123 | Concrete placers, concrete finishers, and related workers | 2 | 9 | 78 | 2296 | ≥ 5 | 2.49 | 3.22 |
| 7111 | Miners and quarry workers | 1 | 3 | 15 | 711 | 3-4 | 1.80 | 2.91 |
| 7122 | Bricklayers and stonemasons | 3 | 23 | 145 | 2857 | ≥ 5 | 2.49 | 2.70 |
| 7129 | Building frame and related trades workers not elsewhere classified | 3 | 7 | 134 | 2069 | ≥ 5 | 2.49 | 2.68 |
| 9312 | Construction and maintenance laborers: roads, dams and similar constructions | 5 | 15 | 581 | 5171 | ≥ 5 | 2.49 | 2.68 |
| 8113 | Well drillers and borers and related workers | 1 | 3 | 3 | 75 | ≥ 5 | 2.49 | 2.66 |
| 7136 | Plumbers and pipe fitters | 1 | 3 | 8 | 76 | 3–4 | 1.80 | 2.50 |
| 7112 | Shotfirers and blasters | 1 | 3 | 3 | 55 | ≥ 5 | 2.49 | 2.47 |
| 7214 | Structural-metal preparers and erectors | 1 | 3 | 15 | 620 | 3–4 | 1.80 | 2.43 |
| 7245 | Electrical line installers, repairers, and cable jointers | 1 | 3 | 68 | 1834 | 3–4 | 1.80 | 2.38 |
| 2211 | Biologists, botanists, zoologists, and related professionals | 1 | 2 | 3 | 56 | 3–4 | 1.80 | 2.36 |
| 6112 | Gardeners, horticultural and nursery growers | 4 | 9 | 249 | 8864 | ≥ 5 | 2.49 | 2.26 |
| 3212 | Agronomy and forestry technicians | 1 | 3 | 14 | 571 | ≥ 5 | 2.49 | 2.19 |
| 8332 | Earth-moving and related plant operators | 2 | 6 | 64 | 2552 | 3−4 | 1.80 | 2.19 |
| 9330 | Transport laborers and freight handlers | 1 | 3 | 62 | 2830 | 3−4 | 1.80 | 2.19 |
| 3213 | Farming and forestry advisers | 1 | 3 | 8 | 259 | 3−4 | 1.80 | 2.15 |
| 7423 | Woodworking machine setters and setter-operators | 1 | 3 | 9 | 362 | 1−2 | 1.57 | 2.14 |
| 6111 | Field crop and vegetable growers | 2 | 7 | 158 | 13424 | ≥ 5 | 2.49 | 2.10 |
| 4142 | Mail carriers and sorting clerks | 2 | 6 | 83 | 2659 | 3−4 | 1.80 | 2.08 |
| 9162 | Sweepers and related laborers | 2 | 6 | 74 | 1766 | ≥ 5 | 2.49 | 2.07 |
| 7124 | Carpenters and joiners | 2 | 6 | 74 | 1932 | 3−4 | 1.80 | 2.06 |
| 9313 | Building construction laborers | 1 | 3 | 3 | 123 | ≥ 5 | 2.49 | 2.03 |
| 7233 | Agricultural- or industrial-machinery mechanics and fitters | 1 | 3 | 60 | 2912 | 1−2 | 1.57 | 1.92 |
| 9161 | Garbage collectors | 2 | 5 | 66 | 2812 | ≥ 5 | 2.49 | 1.87 |
| 8312 | Railway brakers, signalers and shunters | 1 | 3 | 7 | 208 | ≥ 5 | 2.49 | 1.77 |
| 2148 | Cartographers and surveyors | 2 | 5 | 30 | 1224 | 3−4 | 1.80 | 1.74 |
| 6130 | Crop and animal producers | 2 | 4 | 69 | 3807 | ≥ 5 | 2.49 | 1.71 |
| 3474 | Clowns, magicians, acrobats and related associate professionals | 1 | 2 | 5 | 123 | 1−2 | 1.57 | 1.69 |
| 6152 | Inland and coastal waters fishery workers | 2 | 7 | 15 | 320 | ≥ 5 | 2.49 | 1.69 |
| 3475 | Athletes, sports persons and related associate professionals | 2 | 4 | 40 | 532 | 3−4 | 1.80 | 1.56 |
| 5169 | Protective services workers not elsewhere classified | 3 | 6 | 122 | 1244 | 3−4 | 1.80 | 1.55 |
| 7133 | Plasterers | 1 | 3 | 8 | 306 | 3−4 | 1.80 | 1.53 |
| 7213 | Sheet-metal workers | 1 | 3 | 3 | 71 | 3−4 | 1.80 | 1.53 |
| 8163 | Incinerator, water-treatment and related plant operators | 1 | 3 | 5 | 185 | 1−2 | 1.57 | 1.52 |
| 7141 | Painters and related workers | 1 | 1 | 1 | 26 | 3−4 | 1.80 | 1.51 |
| 8324 | Heavy truck and lorry drivers | 1 | 3 | 26 | 1230 | 1−2 | 1.57 | 1.51 |
| 5162 | Police officers | 1 | 3 | 3 | 75 | 3−4 | 1.80 | 1.50 |
| 8334 | Lifting-truck operators | 1 | 3 | 38 | 327 | 3−4 | 1.80 | 1.50 |
| 6141 | Forestry workers and loggers | 1 | 3 | 18 | 813 | ≥ 5 | 2.49 | 1.47 |
| 2213 | Agronomists and related professionals | 1 | 3 | 3 | 95 | 3−4 | 1.80 | 1.39 |
| 2142 | Civil engineers | 1 | 1 | 16 | 49 | 1−2 | 1.57 | 1.13 |
| 3145 | Air traffic safety technicians | 1 | 3 | 14 | 241 | 1−2 | 1.57 | 1.07 |
| 3143 | Aircraft pilots and related associate professionals | 1 | 3 | 4 | 166 | 1−2 | 1.57 | 0.96 |
| 8141 | Wood-processing-plant operators | 1 | 3 | 6 | 210 | 3−4 | 1.80 | 0.90 |
| 3320 | Pre-primary education teaching associate professionals | 1 | 3 | 62 | 3005 | 3−4 | 1.80 | 0.63 |
| 4115 | Secretaries | 1 | 3 | 40 | 403 | 0 | 0.68 | 0.62 |
| 5123 | Waiters, waitresses and bartenders | 1 | 3 | 8 | 60 | 3−4 | 1.80 | 0.53 |
aUVR exposure harmonized to a Gigahertz X2012-10 dosimeter, dosimeter location at the upper arm and, for some studies, weighted by the exposure probability of the occupation. Estimates are presented for the summer season, a latitude of 50 N and a measurement duration of 8 h.
bFor 323 occupations no publications reporting UVR exposures were identified.
cStudy–occupation–season–latitude records.
dHours of outdoor work.
eModel-based arithmetic mean SED levels for the expert rating classes.
fModel-based arithmetic mean SED levels including fixed effect of expert rating class and BLUPs for occupations with published measurements.
| ISCO-88(COM) | ISCO-88(COM) text | Studies | Recordsc | Participants | Workdays | Expert ratingd | Expert estimatese | Full- model estimatesf |
|---|---|---|---|---|---|---|---|---|
| 9211 | Farm-hands and laborers | 1 | 3 | 12 | 455 | ≥ 5 | 2.49 | 3.38 |
| 7131 | Roofers | 2 | 8 | 113 | 3650 | ≥ 5 | 2.49 | 3.28 |
| 7123 | Concrete placers, concrete finishers, and related workers | 2 | 9 | 78 | 2296 | ≥ 5 | 2.49 | 3.22 |
| 7111 | Miners and quarry workers | 1 | 3 | 15 | 711 | 3-4 | 1.80 | 2.91 |
| 7122 | Bricklayers and stonemasons | 3 | 23 | 145 | 2857 | ≥ 5 | 2.49 | 2.70 |
| 7129 | Building frame and related trades workers not elsewhere classified | 3 | 7 | 134 | 2069 | ≥ 5 | 2.49 | 2.68 |
| 9312 | Construction and maintenance laborers: roads, dams and similar constructions | 5 | 15 | 581 | 5171 | ≥ 5 | 2.49 | 2.68 |
| 8113 | Well drillers and borers and related workers | 1 | 3 | 3 | 75 | ≥ 5 | 2.49 | 2.66 |
| 7136 | Plumbers and pipe fitters | 1 | 3 | 8 | 76 | 3–4 | 1.80 | 2.50 |
| 7112 | Shotfirers and blasters | 1 | 3 | 3 | 55 | ≥ 5 | 2.49 | 2.47 |
| 7214 | Structural-metal preparers and erectors | 1 | 3 | 15 | 620 | 3–4 | 1.80 | 2.43 |
| 7245 | Electrical line installers, repairers, and cable jointers | 1 | 3 | 68 | 1834 | 3–4 | 1.80 | 2.38 |
| 2211 | Biologists, botanists, zoologists, and related professionals | 1 | 2 | 3 | 56 | 3–4 | 1.80 | 2.36 |
| 6112 | Gardeners, horticultural and nursery growers | 4 | 9 | 249 | 8864 | ≥ 5 | 2.49 | 2.26 |
| 3212 | Agronomy and forestry technicians | 1 | 3 | 14 | 571 | ≥ 5 | 2.49 | 2.19 |
| 8332 | Earth-moving and related plant operators | 2 | 6 | 64 | 2552 | 3−4 | 1.80 | 2.19 |
| 9330 | Transport laborers and freight handlers | 1 | 3 | 62 | 2830 | 3−4 | 1.80 | 2.19 |
| 3213 | Farming and forestry advisers | 1 | 3 | 8 | 259 | 3−4 | 1.80 | 2.15 |
| 7423 | Woodworking machine setters and setter-operators | 1 | 3 | 9 | 362 | 1−2 | 1.57 | 2.14 |
| 6111 | Field crop and vegetable growers | 2 | 7 | 158 | 13424 | ≥ 5 | 2.49 | 2.10 |
| 4142 | Mail carriers and sorting clerks | 2 | 6 | 83 | 2659 | 3−4 | 1.80 | 2.08 |
| 9162 | Sweepers and related laborers | 2 | 6 | 74 | 1766 | ≥ 5 | 2.49 | 2.07 |
| 7124 | Carpenters and joiners | 2 | 6 | 74 | 1932 | 3−4 | 1.80 | 2.06 |
| 9313 | Building construction laborers | 1 | 3 | 3 | 123 | ≥ 5 | 2.49 | 2.03 |
| 7233 | Agricultural- or industrial-machinery mechanics and fitters | 1 | 3 | 60 | 2912 | 1−2 | 1.57 | 1.92 |
| 9161 | Garbage collectors | 2 | 5 | 66 | 2812 | ≥ 5 | 2.49 | 1.87 |
| 8312 | Railway brakers, signalers and shunters | 1 | 3 | 7 | 208 | ≥ 5 | 2.49 | 1.77 |
| 2148 | Cartographers and surveyors | 2 | 5 | 30 | 1224 | 3−4 | 1.80 | 1.74 |
| 6130 | Crop and animal producers | 2 | 4 | 69 | 3807 | ≥ 5 | 2.49 | 1.71 |
| 3474 | Clowns, magicians, acrobats and related associate professionals | 1 | 2 | 5 | 123 | 1−2 | 1.57 | 1.69 |
| 6152 | Inland and coastal waters fishery workers | 2 | 7 | 15 | 320 | ≥ 5 | 2.49 | 1.69 |
| 3475 | Athletes, sports persons and related associate professionals | 2 | 4 | 40 | 532 | 3−4 | 1.80 | 1.56 |
| 5169 | Protective services workers not elsewhere classified | 3 | 6 | 122 | 1244 | 3−4 | 1.80 | 1.55 |
| 7133 | Plasterers | 1 | 3 | 8 | 306 | 3−4 | 1.80 | 1.53 |
| 7213 | Sheet-metal workers | 1 | 3 | 3 | 71 | 3−4 | 1.80 | 1.53 |
| 8163 | Incinerator, water-treatment and related plant operators | 1 | 3 | 5 | 185 | 1−2 | 1.57 | 1.52 |
| 7141 | Painters and related workers | 1 | 1 | 1 | 26 | 3−4 | 1.80 | 1.51 |
| 8324 | Heavy truck and lorry drivers | 1 | 3 | 26 | 1230 | 1−2 | 1.57 | 1.51 |
| 5162 | Police officers | 1 | 3 | 3 | 75 | 3−4 | 1.80 | 1.50 |
| 8334 | Lifting-truck operators | 1 | 3 | 38 | 327 | 3−4 | 1.80 | 1.50 |
| 6141 | Forestry workers and loggers | 1 | 3 | 18 | 813 | ≥ 5 | 2.49 | 1.47 |
| 2213 | Agronomists and related professionals | 1 | 3 | 3 | 95 | 3−4 | 1.80 | 1.39 |
| 2142 | Civil engineers | 1 | 1 | 16 | 49 | 1−2 | 1.57 | 1.13 |
| 3145 | Air traffic safety technicians | 1 | 3 | 14 | 241 | 1−2 | 1.57 | 1.07 |
| 3143 | Aircraft pilots and related associate professionals | 1 | 3 | 4 | 166 | 1−2 | 1.57 | 0.96 |
| 8141 | Wood-processing-plant operators | 1 | 3 | 6 | 210 | 3−4 | 1.80 | 0.90 |
| 3320 | Pre-primary education teaching associate professionals | 1 | 3 | 62 | 3005 | 3−4 | 1.80 | 0.63 |
| 4115 | Secretaries | 1 | 3 | 40 | 403 | 0 | 0.68 | 0.62 |
| 5123 | Waiters, waitresses and bartenders | 1 | 3 | 8 | 60 | 3−4 | 1.80 | 0.53 |
| ISCO-88(COM) | ISCO-88(COM) text | Studies | Recordsc | Participants | Workdays | Expert ratingd | Expert estimatese | Full- model estimatesf |
|---|---|---|---|---|---|---|---|---|
| 9211 | Farm-hands and laborers | 1 | 3 | 12 | 455 | ≥ 5 | 2.49 | 3.38 |
| 7131 | Roofers | 2 | 8 | 113 | 3650 | ≥ 5 | 2.49 | 3.28 |
| 7123 | Concrete placers, concrete finishers, and related workers | 2 | 9 | 78 | 2296 | ≥ 5 | 2.49 | 3.22 |
| 7111 | Miners and quarry workers | 1 | 3 | 15 | 711 | 3-4 | 1.80 | 2.91 |
| 7122 | Bricklayers and stonemasons | 3 | 23 | 145 | 2857 | ≥ 5 | 2.49 | 2.70 |
| 7129 | Building frame and related trades workers not elsewhere classified | 3 | 7 | 134 | 2069 | ≥ 5 | 2.49 | 2.68 |
| 9312 | Construction and maintenance laborers: roads, dams and similar constructions | 5 | 15 | 581 | 5171 | ≥ 5 | 2.49 | 2.68 |
| 8113 | Well drillers and borers and related workers | 1 | 3 | 3 | 75 | ≥ 5 | 2.49 | 2.66 |
| 7136 | Plumbers and pipe fitters | 1 | 3 | 8 | 76 | 3–4 | 1.80 | 2.50 |
| 7112 | Shotfirers and blasters | 1 | 3 | 3 | 55 | ≥ 5 | 2.49 | 2.47 |
| 7214 | Structural-metal preparers and erectors | 1 | 3 | 15 | 620 | 3–4 | 1.80 | 2.43 |
| 7245 | Electrical line installers, repairers, and cable jointers | 1 | 3 | 68 | 1834 | 3–4 | 1.80 | 2.38 |
| 2211 | Biologists, botanists, zoologists, and related professionals | 1 | 2 | 3 | 56 | 3–4 | 1.80 | 2.36 |
| 6112 | Gardeners, horticultural and nursery growers | 4 | 9 | 249 | 8864 | ≥ 5 | 2.49 | 2.26 |
| 3212 | Agronomy and forestry technicians | 1 | 3 | 14 | 571 | ≥ 5 | 2.49 | 2.19 |
| 8332 | Earth-moving and related plant operators | 2 | 6 | 64 | 2552 | 3−4 | 1.80 | 2.19 |
| 9330 | Transport laborers and freight handlers | 1 | 3 | 62 | 2830 | 3−4 | 1.80 | 2.19 |
| 3213 | Farming and forestry advisers | 1 | 3 | 8 | 259 | 3−4 | 1.80 | 2.15 |
| 7423 | Woodworking machine setters and setter-operators | 1 | 3 | 9 | 362 | 1−2 | 1.57 | 2.14 |
| 6111 | Field crop and vegetable growers | 2 | 7 | 158 | 13424 | ≥ 5 | 2.49 | 2.10 |
| 4142 | Mail carriers and sorting clerks | 2 | 6 | 83 | 2659 | 3−4 | 1.80 | 2.08 |
| 9162 | Sweepers and related laborers | 2 | 6 | 74 | 1766 | ≥ 5 | 2.49 | 2.07 |
| 7124 | Carpenters and joiners | 2 | 6 | 74 | 1932 | 3−4 | 1.80 | 2.06 |
| 9313 | Building construction laborers | 1 | 3 | 3 | 123 | ≥ 5 | 2.49 | 2.03 |
| 7233 | Agricultural- or industrial-machinery mechanics and fitters | 1 | 3 | 60 | 2912 | 1−2 | 1.57 | 1.92 |
| 9161 | Garbage collectors | 2 | 5 | 66 | 2812 | ≥ 5 | 2.49 | 1.87 |
| 8312 | Railway brakers, signalers and shunters | 1 | 3 | 7 | 208 | ≥ 5 | 2.49 | 1.77 |
| 2148 | Cartographers and surveyors | 2 | 5 | 30 | 1224 | 3−4 | 1.80 | 1.74 |
| 6130 | Crop and animal producers | 2 | 4 | 69 | 3807 | ≥ 5 | 2.49 | 1.71 |
| 3474 | Clowns, magicians, acrobats and related associate professionals | 1 | 2 | 5 | 123 | 1−2 | 1.57 | 1.69 |
| 6152 | Inland and coastal waters fishery workers | 2 | 7 | 15 | 320 | ≥ 5 | 2.49 | 1.69 |
| 3475 | Athletes, sports persons and related associate professionals | 2 | 4 | 40 | 532 | 3−4 | 1.80 | 1.56 |
| 5169 | Protective services workers not elsewhere classified | 3 | 6 | 122 | 1244 | 3−4 | 1.80 | 1.55 |
| 7133 | Plasterers | 1 | 3 | 8 | 306 | 3−4 | 1.80 | 1.53 |
| 7213 | Sheet-metal workers | 1 | 3 | 3 | 71 | 3−4 | 1.80 | 1.53 |
| 8163 | Incinerator, water-treatment and related plant operators | 1 | 3 | 5 | 185 | 1−2 | 1.57 | 1.52 |
| 7141 | Painters and related workers | 1 | 1 | 1 | 26 | 3−4 | 1.80 | 1.51 |
| 8324 | Heavy truck and lorry drivers | 1 | 3 | 26 | 1230 | 1−2 | 1.57 | 1.51 |
| 5162 | Police officers | 1 | 3 | 3 | 75 | 3−4 | 1.80 | 1.50 |
| 8334 | Lifting-truck operators | 1 | 3 | 38 | 327 | 3−4 | 1.80 | 1.50 |
| 6141 | Forestry workers and loggers | 1 | 3 | 18 | 813 | ≥ 5 | 2.49 | 1.47 |
| 2213 | Agronomists and related professionals | 1 | 3 | 3 | 95 | 3−4 | 1.80 | 1.39 |
| 2142 | Civil engineers | 1 | 1 | 16 | 49 | 1−2 | 1.57 | 1.13 |
| 3145 | Air traffic safety technicians | 1 | 3 | 14 | 241 | 1−2 | 1.57 | 1.07 |
| 3143 | Aircraft pilots and related associate professionals | 1 | 3 | 4 | 166 | 1−2 | 1.57 | 0.96 |
| 8141 | Wood-processing-plant operators | 1 | 3 | 6 | 210 | 3−4 | 1.80 | 0.90 |
| 3320 | Pre-primary education teaching associate professionals | 1 | 3 | 62 | 3005 | 3−4 | 1.80 | 0.63 |
| 4115 | Secretaries | 1 | 3 | 40 | 403 | 0 | 0.68 | 0.62 |
| 5123 | Waiters, waitresses and bartenders | 1 | 3 | 8 | 60 | 3−4 | 1.80 | 0.53 |
aUVR exposure harmonized to a Gigahertz X2012-10 dosimeter, dosimeter location at the upper arm and, for some studies, weighted by the exposure probability of the occupation. Estimates are presented for the summer season, a latitude of 50 N and a measurement duration of 8 h.
bFor 323 occupations no publications reporting UVR exposures were identified.
cStudy–occupation–season–latitude records.
dHours of outdoor work.
eModel-based arithmetic mean SED levels for the expert rating classes.
fModel-based arithmetic mean SED levels including fixed effect of expert rating class and BLUPs for occupations with published measurements.
Increasing UVR exposure of 0.68, 1.57, 1.80, and 2.49 SED were seen for expert ratings of 0, 1 to 2, 3 to 4 ≥5 h of outdoor work. The highest exposure was observed in farm hands, roofers, and concrete placers and other occupations within craft and related trades (main ISCO-88(COM) group 7). The lowest exposure was observed in waiters and bartenders, wood-processing-plant operators, and several white-collar occupations. The UVR exposure of the highest exposed occupation was 6-fold that of the lowest exposed. Figure 1 depicts how the occupation specific UVR exposure estimates vary around the corresponding expert ratings.
Model-based standardized erythema dose (SED) (arithmetic means) in measured occupations compared to expert rating of duration of outdoor work. Circles represent the expert ratings and the 4-digit numbers the 49 ISCO-88(COM) codes of the individual occupations.
Discussion
We constructed a quantitative JEM for 372 occupations with estimates of personal solar UVR exposure expressed as workday SED. The JEM was based on a combination of arithmetic mean UVR exposure measured across Europe, Canada, and the United States reported in the literature and a priori expert ratings. This approach allowed us to assign an exposure estimate to occupations with no available exposure measurements.
The experts rated the daily average duration of outdoor work for each occupation and not the workday solar UVR exposure as such. Still, this proxy measure showed monotonically increasing UVR exposure by increasing rating and a more than 3-fold higher exposure for the highest compared with the lowest rating. This indicates that we can be quite confident with the provided UVR exposure estimates for occupations with no measurements relying solely on the expert ratings. Full-model UVR exposure estimates for occupations with measurements further differentiated exposure estimates beyond that captured by the expert ratings and showed a 6-fold increase from the lowest to the highest exposed occupation.
The category rated by the experts as 0 h of outdoor work during an average workday (i.e. indoor workers) showed UVR exposure above null. This is most likely because the UVR exposure reported in the included papers was typically defined for a workday rather than for work hours, and may include outdoor hours before, during, or after indoor work hours.
We observed lower UVR exposure during spring and autumn compared with during summer and decreasing UVR exposure with increasing latitude, as expected (IARC Working Group on the Evaluation of Carcinogenic Risks to Humans 2012). We observed a negligible trend in UVR exposure with increasing duration of the daily measurements, which could reflect that the longer measurements included more hours during the morning and the evening with low ambient UVR exposure of little impact on all workday UVR exposure.
We observed that the total variance of the linear mixed effects model increased when we included season, latitude, and duration of measurements as fixed effects in addition to study and occupation as random effects (intermediate model). This was slightly counterintuitive, although not completely unexpected given the non-nested and highly unbalanced structure of the data. However, a substantial reduction of the residual variance and a material increase in the between-occupations variance, which is a measure of the discriminatory power of the model, confirms the relevance of including these effects. Occupations were completely nested within the expert rating classes, and the reduced variance between occupations we observed when we included the expert rating (full model), was as expected and furthermore confirmed the ratings’ relevance.
Strengths and limitations
We only included UVR exposure measurements obtained with personal dosimeters, which is crucial when assessing individual UVR exposure because personal exposure correlates poorly with ambient exposure (Schmalwieser et al. 2010; Bodekær et al. 2015; Soueid et al. 2022).
Occupations characterized by much outdoor work were heavily overrepresented among the measured occupations when compared with the expert ratings of duration of outdoor work. On one hand, this is a strength because the variation is expected to increase with higher exposure, on the other hand, the uncertainty due to the very few measurements of low exposed occupations is a limitation.
Several of the 12 studies did not provide UVR exposure for randomly selected workdays for a given occupation but specifically for days with outdoor work. The reported UVR exposure is thus expected to be higher than that of an average worker because most outdoor jobs include some extent of indoor work. Therefore, we weighted the exposure from these studies by the expert estimates of exposure probability. Selection of outdoor workers may have affected all studies to some extent, especially for occupations with little outdoor work, but perhaps not for studies that measured UVR exposure for several weeks or months (Grandahl et al. 2018; Wittlich 2022). This may imply that low exposed occupations are assigned a too high exposure, which will deflate true exposure-response relations in future epidemiological studies applying the JEM.
We harmonized UVR exposure for dosimeter locations other than the upper arm by published correction factors obtained with PSF dosimeters (Knuschke et al. 2007). We assume that these factors are also applicable for the electronic dosimeters, however, this may to some extent not be the case, e.g. due to differences in spectral and angular responsivity and sensitivity.
All studies used calibrated dosimeters and the electronic dosimeters provided comparable SED levels as expected. However, the PSF dosimeter provided considerably higher exposure levels than the electronic dosimeters. We accounted for this by a correction factor obtained from published parallel measurements with PSF dosimeters and Gigahertz X2012-10 electronic dosimeters (Strehl et al. 2021).
Comparable variances in UVR exposure were seen between occupations and between studies. This indicates that we were not able to control for variation across seasons, latitudes, and duration of daily measurements as well as other factors, such as clouds, reflection, ozone level, and air pollution, that may vary day-by-day because we relied only on aggregated level information. Future access to individual level hour-by-hour and day-by-day information, that is provided by most electronic dosimeters combined with information on the daily ozone level, is expected to improve the occupational UVR exposure estimates significantly.
Most occupations (86%) were represented by no measurements; and of the 14% of occupations with measurement, most were represented by only one study, which is a major limitation. The participants in the measured occupations may not be representative for an average worker with regards to extent of outdoor work. This may, for instance, be the case for waiters, waitresses, and bartenders, showing the overall lowest UVR exposure. At the one extreme, they may work in night clubs, never exposed to solar UVR. At the other extreme, they may be full-shift UVR exposed if working at an outdoor café. This limitation can only be alleviated by including more studies from a variety of worker populations within the same occupation.
We identified no studies reporting UVR exposure during winter. This may affect the application of the JEM on populations living at southern, but not populations living at northern European latitudes (Vitt et al. 2020).
We only searched PubMed for peer reviewed articles in English reporting personal occupational UVR exposure, and this is a limitation. Searching for existing measurement results indexed in other databases and published in the gray literature and reported in other languages could have increased the number of relevant articles from other populations and occupations throughout the year. The optimal approach would, however, be to conduct a measurement campaign of personal UVR exposure. Participants should be invited to represent occupations with different extent of outdoor work, e.g. as proposed by the expert raters of the current study.
This JEM does not account for UVR exposure from artificial industrial sources or protective factors, such as clothing coverage, sunscreen, hat, sunglasses, or other self-protection that may vary across workers, occupations, and countries. This information is needed for the populations on which this JEM will be applied to obtain the relevant occupational exposure at the skin and eye. Information on UVR exposure during time off work is necessary to have the complete UVR exposure profile, which is the ultimate measure for assessing UVR exposure related health effects. However, only small differences in solar UVR exposure during leisure time across occupations and time spent outdoor during days off have been observed for in- and outdoor workers (Grandahl et al. 2018; Daugaard et al. 2019). Still, UVR exposure from sunbeds should also be considered (IARC Working Group on the Evaluation of Carcinogenic Risks to Humans 2012).
The performance of the JEM should be examined by studying its ability to confirm known short and long-term relationships between UVR exposure and for instance vitamin D and skin cancer (IARC Working Group on the Evaluation of Carcinogenic Risks to Humans 2012; Webb et al. 2021).
Comparison with other studies
We included all studies fulfilling the inclusion criteria, consequently no directly comparable studies are available to check for analogous findings. We therefore compared with studies reporting UVR exposure by other statistics and obtained with other dosimeters. An earlier study of solar UVR exposure of strawberry pickers in Italy (43 °N) during spring and summer showed workday geometric mean SED of 0.48 to 1.24, based on Gigahertz X2000-5 dosimeters located at the forearm (Nardini et al. 2014). Another study of vineyard workers in Italy (43 °N) during July, showed a workday median SED of 5.9 based on PSF dosimeters located at the arm (Siani et al. 2011). These estimates are lower and higher, respectively, than the 1.60 SED we estimated for gardeners, horticultural and nursery growers during summer at 50 °N. A study of construction workers working entirely outdoor during July in Spain (39 °N), who were measured with a biological dosimeter (VioSpore) located at the chest or shoulder, showed a median workday SED of 6.1 (Serrano et al. 2013), which is much higher than our estimate of 1.29 SED for building construction laborers at 50 °N.
Even if none of these results are directly comparable, they indicate a high variability in UVR exposure within occupations as also shown in our data. Therefore, when applied for epidemiological research, the occupation and latitude-specific UVR estimates may be supplemented with sensitivity analyses using the model-based expert rating UVR exposure.
Application
The JEM can be used in epidemiological studies to estimate personal occupational UVR exposure based on the participants’ work histories and the latitude of the worksites as spring, summer, or autumn workday exposure by applying this formula: Occupational UVR exposure during summer (Table 3) × arithmetic mean ratio for spring, summer, or autumn (Table 2) × arithmetic mean ratio for latitudelatitude worksite—50 (Table 2). Spring UVR exposure for a roofer in Madrid (40°N) will then be 3.28 SED × 0.92 × 0.9740-50 = 4.09 SED. The JEM only provides estimates for spring, summer, and autumn and not for winter and thus not for a whole year nor an annual average. At northern latitudes, the average of spring, summer, and autumn exposure will tend to overestimate the annual average relatively more than at southern latitudes. When applied on populations from northern Europe, (SEDspring + SEDsummer + SEDautumn + 0winter)/4 should, however, correspond well with the annual average, because UVR exposure is close to null during winter. These considerations also apply to long term cumulative exposure expressed as SED-years that ideally should be based on the annual average SED (in line with annual packs in pack-years for smoking).
Conclusion
This is the first quantitative JEM providing a biologically relevant metric, the SED, for occupational solar UVR exposure based on the combination of aggregated personal measurements and expert consensus ratings. The JEM proves able to provide substantial discrimination between occupations, shows a monotonic increase with increasing expert rating and a 6-fold ratio between the highest and lowest exposed occupations. Solar UVR exposure is ubiquitous and not restricted to specific industries or occupations, and the JEM is designed for epidemiological studies of the general working population and may, keeping the mentioned limitations in mind, be applicable for epidemiological studies of exposure–response relations between quantitative measures of solar UVR exposure and different health effects in Europe.
Supplementary material
Supplementary material is available at Annals of Work Exposures and Health online.
Author contributions
ETW, KP, and HAK designed the study. ETW and KP identified studies and extracted data. ETW and MFG analyzed data. IAB, MPCC, AD-H, KG, JM, AM, HN, SS, KS, and MW rated the extent of outdoor work. SC, TH, PAP, and SW estimated calibration factors for dosimeter types. HAK wrote the first draft of the manuscript and is the guarantor of the manuscript. All the authors interpreted the data and assisted in the drafting of the manuscript. HAK and KP have full access to the data, and HAK had the final responsibility to submit the manuscript. All authors approved the final version of the manuscript.
Funding
This work was part of the Exposome Project for Health and Occupational Research (EPHOR) that is funded by the European Union’s Horizon 2020 research and innovation programme under grant agreement number 874703.
Conflict of interest
The authors declare no conflict of interest relating to the material presented in this Article. Its contents, including any opinions and/or conclusions expressed, are solely those of the authors.
Data availability
The data and the JEM is available on request.
