BJD E P I DE M I O L O G Y British Journal of Dermatology Estimating personal solar ultraviolet radiation exposure through time spent outdoors, ambient levels and modelling approaches* mez iD ,1,2,3 L. Alonso iD ,1,2,3 L. Soueid iD ,1,2,3 M. Triguero-Mas iD 4,5,6,7 A. Dalmau,1,2,3,8 J. Barrera-Go X. Basagan~a iD , 1,2,3 E. Thieden iD , H.C. Wulf, B. Diffey iD , A.R. Young iD ,11 M. Nieuwenhuijsen iD 1,2,3 and 9 9 10 1,2,3 P. Dadvand iD 1 ISGlobal, Barcelona, Spain 2 Universitat Pompeu Fabra (UPF), Barcelona, Spain 3 CIBER Epidemiologıa y Salud Publica (CIBERESP), Barcelona, Spain 4 Universitat Autonoma de Barcelona, Barcelona, Spain 5 Barcelona Lab for Urban Environmental Justice and Sustainability, Barcelona, Spain 6 Institute of Environmental Science and Technology (ICTA-UAB), Barcelona, Spain 7 Hospital del Mar Medical Research Institute (IMIM), Barcelona, Spain 8 Agencia de Qualitat i Avaluacio Sanitaries de Catalunya (AQuAS), Barcelona, Spain 9 Department of Dermatology, Bispebjerg Hospital, Copenhagen, Denmark 10 Institute of Cellular Medicine, Newcastle University, Newcastle upon Tyne, UK 11 King’s College London, St John’s Institute of Dermatology, London, UK Linked Comment: P. O’Mahoney. Br J Dermatol 2022; 186:207–208. Summary Correspondence Background Evidence on validation of surrogates applied to evaluate the personal expo- Margarita Triguero-Mas. sure levels of solar ultraviolet radiation (UVR) in epidemiological studies is scarce. Email: mtrigueromas@gmail.com Objectives To determine and compare the validity of three approaches, including (i) ambient UVR levels, (ii) time spent outdoors and (iii) a modelling approach Accepted for publication 13 August 2021 integrating the aforementioned parameters, to estimate personal UVR exposure over a period of 6 months among indoor and outdoor workers and in different Funding sources seasons (summer/winter). This work forms part of the ICEPURE project Methods This validation study was part of the European Commission-funded ICE- which was funded by the European Commission PURE project and was performed between July 2010 and January 2011 in a con- Framework Program 7 (grant no. 227020). venience sample of indoor and outdoor workers in Catalunya, Spain. We Spanish Ministry of Economy and Competitiveness: FJCI-2017-33842 (M.T.-M) and RYC-2012- developed linear regression models to quantify the variation in the objectively 10995 (P.D.). Spanish Ministry of Science and measured personal UVR exposure that could be explained, separately, by the Innovation: CEX2018-000806-S. Generalitat de ambient UVR, time spent outdoors and modelled UVR levels. Catalunya: CERCA Program. Results Our 39 participants – mostly male and with a median age of 35 years – pre- sented a median daily objectively measured UVR of 037 standard erythemal doses. Conflicts of interest The UVR dose was statistically significantly higher in summer and for outdoor None to declare. workers. The modelled personal UVR exposure and self-reported time spent out- *Plain language summary available online doors could reasonably predict the variation in the objectively measured personal UVR levels (R2 range 075–079), whereas ambient UVR was a poor predictor (R2 DOI 10.1111/bjd.20703 = 021). No notable differences were found between seasons or occupation. Conclusions Time outdoors and our modelling approach were reliable predictors and of value to be applied in epidemiological studies of the health effects of cur- rent exposure to UVR. 266 British Journal of Dermatology (2022) 186, pp266–273 © 2021 The Authors. British Journal of Dermatology published by John Wiley & Sons Ltd on behalf of British Association of Dermatologists This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. Validity of sun questionnaires and modelled exposure, L. Soueid et al. 267 What is already known about this topic? • Most studies evaluating health effects of solar ultraviolet radiation (UVR) have relied on surrogates to assess personal UVR exposure. • However, the available evidence validating these surrogates is scarce. What does this study add? • In the present study, we evaluated and compared the validity of three approaches to estimate personal UVR exposure (objectively measured using a personal dosime- ter), including (i) ambient UVR levels, (ii) time spent outdoors and (iii) a mod- elling approach, during different seasons among indoor and outdoor workers. • Modelled UVR exposure and time outdoors could predict more than three-quarters of the variation in the objectively measured personal UVR (and are therefore of value to be applied in epidemiological studies of the health effects of UVR), whereas ambient UVR could predict only around one-fifth. • Our findings support a major role of personal behaviour in determining personal UVR exposure. Ultraviolet radiation (UVR) is a major environmental hazard estimate individual long-term UVR exposure of outdoor work- which also has health benefits.1 Sunburn, photokeratitis and ers.9,10 immunosuppression are among the short-term adverse effects To date, the available evidence evaluating the validity of the of UVR exposure while skin cancers, skin photoageing and aforementioned surrogates of UVR exposure is still limited, possibly cataracts are among adverse long-term effects.2–5 The with few studies validating UVR exposure modelling main beneficial effect of UVR is the cutaneous production of approaches.9,11–13 Moreover, studies comparing validity of dif- vitamin D which, among other possible benefits, is essential ferent surrogates of long-term UVR exposure are nonexistent. for calcium metabolism and the maintenance of the muscu- Previous studies have focused mainly on specific exposure loskeletal system.2–5 groups such as indoor workers, outdoor workers, children or The sun is the main source of UVR exposure at population adolescents for limited duration (from 1 day to a few level.3 Personal exposure to solar UVR is a function of ambi- months) without comparing the validity of surrogates in dif- ent UVR and behaviour,2,6–8 which includes time spent out- ferent subpopulations.10,14–22 Accordingly, the aim of this doors in the sun, risky behaviours (e.g. sunbathing), study was to determine and compare the validity of self- photoprotection (e.g. sunscreen use, clothing) and holiday reported time spent outdoors, ambient UVR and modelled habits.6 Epidemiological studies on the association between personal UVR levels in estimating current personal UVR expo- UVR exposure and long-term health outcomes have mainly sure among outdoor and indoor workers and during summer relied on surrogates, such as ambient UVR (or latitude as its and winter. proxy) and self-reported time outdoors, to assess personal UVR exposure. However, in a previous simulation study6 Materials and methods across six northern, central and southern European cities, we showed that each surrogate, when used alone, could only This study was conducted in the context of the ‘Impact of explain less than a quarter of the variation in personal UVR Climatic and Environmental factors on Personal Ultraviolet exposure. Our previous study also revealed that within-city Radiation Exposure and human health’ (ICEPURE) project variation (where ambient UVR is similar) was three times funded by the European Commission (FP7, grant no. greater than the variation between cities, indicating the critical 227020). role of individual behaviour in exposure to UVR. These find- ings highlighted the need to develop more comprehensive Study setting and population measures integrating both ambient UVR and individual beha- viour to assess personal exposure to UVR. In this context, Our study included 39 participants, consisting of 18 indoor modelling approaches that combine ambient UVR and individ- workers and 21 outdoor workers, living and working in Bar- ual behaviour to assess personal exposure to UVR have poten- celona province, Catalonia, Spain (2010–11). This province tial, despite having been rarely used so far in large-scale (413850° N, 21733° E) is situated in the northeast of the epidemiological studies. To the best of our knowledge, a lim- Iberian peninsula. It covers an area of approximately 7726 ited number of occupational studies have developed models to km2 including a Mediterranean coastline with a population of © 2021 The Authors. British Journal of Dermatology British Journal of Dermatology (2022) 186, pp266–273 published by John Wiley & Sons Ltd on behalf of British Association of Dermatologists 268 Validity of sun questionnaires and modelled exposure, L. Soueid et al. approximately 57 million inhabitants (2020).23 It has a Ambient ultraviolet radiation Mediterranean climate with hot and dry summers and mild winters with precipitation occurring predominantly during Ambient UVR was obtained through continuous measurement of spring and autumn months.23,24 solar UVR using modified versions of the SunSaver housed in two We recruited a convenience sample of office workers (as unshaded UVR ground stations about 50-km apart, one in Tor- indoor workers) and farmers (as outdoor workers) as relavit, where most outdoor workers were living (coordinates: described in Materials and Methods S1 (see Supporting Infor- 4143904, 174228), and the other in Barcelona city, where most mation). All participants gave written informed consent before indoor workers were living (coordinates: 4138522, 219421). enrolment. Ethics approval (nos 2009/3692/I and 2008/ 3017/I) was obtained from the Clinical Research Ethical Com- Questionnaire data mittee of the Parc de Salut MAR, Barcelona, Spain, to carry out this study. Data on sociodemographic characteristics and self-reported time spent outdoors were obtained through two types of questionnaires (see Materials and Methods S1 for further Data collection details). The participants answered Questionnaire 1 by tele- This study was performed over 6 months from 1 July 2010 phone interview at the end of each month during the study until 10 January 2011. period (i.e. this questionnaire was implemented six times for each participant). Through this questionnaire, we obtained data on the average time spent outdoors on working and non- Objectively measured personal ultraviolet radiation working days and holidays, as well as holiday duration and exposure location during the corresponding month. Questionnaire 2 Personal erythemally effective UVR exposure was measured was filled in and returned by the participants by post, objectively with a personal electronic dosimeter (SunSaver),25–27 1 month after the end of the study. It enquired about the which participants wore on their left wrist over any clothing, aforementioned factors included in Questionnaire 1 together every day from 7:00 till 21:00. The SunSaver was produced by with sun-protective and risky behaviours during the 6-month the Department of Dermatology, Bispebjerg Hospital, Univer- study period. sity of Copenhagen, Denmark.28 More information about the SunSaver is provided in Materials and Methods S1. The wrist Ultraviolet radiation exposure modelling was selected because this position is better tolerated during a long-term study.21,23–28 Furthermore, the UVR doses received We applied the modelling framework developed by Diffey,7 on the wrist have been shown to correlate well (50% of the which was used in our previous simulation study,6 to estimate dose) with those received on the top of the head.29 The per- the UVR exposures of participants during the course of the sonal SunSaver recorded time-stamped (every 5 min) erythe- study. This modelling framework integrates ambient UVR mally weighted UVR exposures (Figure 1) measured in levels with time spent outdoors, an exposure fraction, and standard erythemal doses (SEDs). The participants were hours of daylight to estimate the personal exposure to UVR. instructed to behave as they usually did on a regular day and to We developed two separate modelling estimates based on the remove the SunSaver when swimming or taking a shower/bath. time spent outdoors obtained from Questionnaires 1 and 2. Further details on this modelling approach are presented in Materials and Methods S1. Statistical analysis Nonparametric tests were used to generate summary statistics. Univariate regression models were developed using the log- transformed measured personal UVR dose as the outcome and each of ambient UVR levels, self-reported time spent outdoors and modelled personal UVR exposure as the predictor. We log-transformed the outcome variable (measured personal UVR exposure) in order to obtain approximate normality and linearity. The coefficient of determination (R2) was used to assess the variation in the measured personal UVR exposure explained by each surrogate. We then stratified the regression Figure 1 The personal electronic ultraviolet radiation (UVR) analyses based on occupation type (indoor vs. outdoor work- dosimeter (SunSaver) used to monitor time-stamped UVR doses. It ers) and season (winter vs. summer). The season variable was comprises a UVR sensor, a data logger and a battery mounted constructed by grouping July, August and September as the together with a digital watch in the form of a wristwatch. summer months and October, November, December and British Journal of Dermatology (2022) 186, pp266–273 © 2021 The Authors. British Journal of Dermatology published by John Wiley & Sons Ltd on behalf of British Association of Dermatologists Validity of sun questionnaires and modelled exposure, L. Soueid et al. 269 Table 1 Description of study variables, participant characteristics and exposure characteristics, stratified by occupation type All Indoor workers Outdoor workers Variable N = 39 N = 18 N = 21 P-value Age, years: median (IQR) 35 (15) 29 (7) 42 (9) < 001 Sex, females, n (%) 15 (385) 13 (72) 2 (95) < 001 Education, n (%) High school 4 (10) 0 (0) 4 (19) < 001 Technical school 16 (41) 1 (55) 15 (71) University and/or higher 19 (49) 17 (94) 2 (95) Ambient UVR, SEDs: median (IQR)a 2546 (1144) 1823 (666) 2878 (566) < 001 Objectively measured personal UVR exposure, SEDs: median (IQR)a 037 (122) 008 (010) 132 (093) < 001 Summer 185 (239) 098 (082) 291 (176) < 001 Winter 051 (123) 019 (025) 130 (118) < 001 Objectively measured personal UVR exposure during 027 (030) 011 (009) 020 (030) < 001 high-risk hours, SEDs: median (IQR)a Self-reported time spent outdoors, hours: median (IQR)a Based on Questionnaire 1b 493 (460) 225 (107) 692 (212) < 001 Based on Questionnaire 2b 565 (562) 185 (107) 740 (190) < 001 Modelled personal UVR exposure, SEDs: median (IQR)a Based on Questionnaire 1b 122 (192) 071 (021) 259 (092) < 001 Based on Questionnaire 2b 130 (190) 073 (022) 261 (090) < 001 P-values correspond to a permutation test, for equal medians in the case of continuous variables, or for equal distribution in the case of cate- gorical variables. IQR, interquartile range; SEDs, standard erythemal doses; UVR, ultraviolet radiation, aDaily average measurement; bsee Mate- rials and Methods S1. January as the winter months. Stata for windows version 14 participants combined. Following stratification by season, the was used for conducting all analyses (StataCorp, College Sta- median (IQR) of daily measured personal UVR exposure was tion, TX, USA). A P-value of less than 005 was considered as 185 (239) (SEDs) in summer and 051 (123) (SEDs) in statistically significant. winter (P < 001). As expected, the median (IQR) measured personal UVR exposure in outdoor workers was significantly higher than that of indoor workers, both for the entire day Results [SEDs: 132 (093) for outdoor workers and 008 (010) for The sociodemographic characteristics of the study participants, indoor workers] and during high-risk hours (12:00–15:00) combined and separately for indoor and outdoor workers, are [SEDs: 020 (030) for outdoor workers and 011 (009) for shown in Table 1. The median age of the participants was indoor workers] (Table 1). A similar pattern was also 35 years [interquartile range (IQR) 15 years], of whom 24 observed for self-reported time outdoors and modelled per- (615%) were male. The median (IQR) daily ambient UVR sonal UVR exposure (Table 1). levels and daily measured personal UVR exposure were, The regression analyses showed that while modelled per- respectively, 2546 (1144) and 037 (122) SEDs for all sonal UVR exposure (R2 = 077) and self-reported time Table 2 Coefficients of determination (R2) for univariate regression models of the long-term exposure between modelled personal UVR exposure, self-reported time spent outdoors and ambient UVR against log-transformed objectively measured personal UVR exposure – for the overall study population, and separately for occupation type and season Overalla Indoor workersa Outdoor workersa Summera Wintera Averaged semi-annual exposure (N = 37) (N = 20) (N = 17) (N = 35) (N = 36) Questionnaire 1b Modelled personal exposure 077*** 039** 026* 065*** 058*** Self-reported time spent outdoors 075*** 015 012 046*** 043*** Ambient UVR 021** 002 00002 020** 018** Questionnaire 2b Modelled personal exposure 077*** 038** 025* 063*** 058*** Self-reported time spent outdoors 079*** 012 017 040*** 041*** Ambient UVR 021** 002 00002 020** 018* P-values: * < 005, ** < 001, *** < 0001. UVR, ultraviolet radiation, aAll values are R2; bsee Materials and Methods S1. © 2021 The Authors. British Journal of Dermatology British Journal of Dermatology (2022) 186, pp266–273 published by John Wiley & Sons Ltd on behalf of British Association of Dermatologists 270 Validity of sun questionnaires and modelled exposure, L. Soueid et al. Figure 2 Scatterplots of the daily averages of log-transformed measured personal ultraviolet radiation (UVR) exposure (standard erythemal doses, SEDs) against each of the modelled personal UVR exposures (SEDs), self-reported time spent outdoors (hours) and ambient UVR levels (SEDs) for both questionnaires (see Materials and Methods S1). outdoors (R2 = 075 and 079 for Questionnaires 1 and 2, The results of the regression analyses stratified by season respectively) could reasonably predict measured personal UVR (summer/winter) showed very similar coefficients of determi- exposure, ambient UVR was a relatively poor predictor (R2 = nation (R2) for each surrogate for winter and summer, as well 021) (Table 2). Figure 2 shows scatterplots of these associa- as between questionnaires (Table 2). We observed that mod- tions in which we mostly noted two different clusters of elled personal UVR exposure was the best predictor of mea- observations, which could indicate differences between the sured personal UVR exposure in summer (R2 = 065 or 063 indoor and outdoor workers or between the summer and for Questionnaires 1 and 2, respectively) and winter (R2 = winter seasons. Consequently, we performed the regressions 058 for both questionnaires). Variation of measured personal stratified by two groups of occupation and two seasons. UVR exposure explained by self-reported time outdoors was The regression analyses stratified by occupation showed that slightly lower, with R2 = 046 or 040 for summer and 043 modelled personal UVR exposure was the best predictor of or 041 for winter, based on Questionnaires 1 and 2, respec- measured personal UVR exposure among indoor workers (R2 tively. On the other hand, the variation of measured personal = 039 and 038 for Questionnaires 1 and 2, respectively), UVR exposure explained by the ambient UVR was the lowest whereas modelled personal UVR exposure explained less varia- among the evaluated surrogates (modelled personal UVR tion in measured personal UVR exposure for outdoor workers exposure, self-reported time outdoors and ambient UVR) (R2 but was still the best predictor (R2 = 026 and R2 = 025 for = 020 for summer and R2 = 018 for winter, based on Ques- Questionnaires 1 and 2, respectively) (Table 2). Furthermore, tionnaires 1 and 2, respectively). the regression analyses stratified by occupation showed lower R2 than when occupations were combined. This was expected Discussion as, within groups, the differences in time spent outdoors and consequently in UVR levels between workers were lower and To the best of our knowledge, this study is one of the first to more difficult to predict. (i) validate a modelling approach to assess personal UVR British Journal of Dermatology (2022) 186, pp266–273 © 2021 The Authors. British Journal of Dermatology published by John Wiley & Sons Ltd on behalf of British Association of Dermatologists Validity of sun questionnaires and modelled exposure, L. Soueid et al. 271 exposure; (ii) compare the validity of modelled personal UVR implications in reducing the burden of data collection in exposure, self-reported time spent outdoors and ambient UVR future epidemiological studies of the health effects of UVR to assess personal UVR exposure; and (iii) compare the valid- exposure. ity of this combination of surrogates of UVR exposure among In our study, outdoor workers received approximately 16 indoor and outdoor workers as well as across different seasons times more erythemal UVR than indoor workers, as recorded (summer/winter). Moreover, the 6-month duration of the by the SunSavers. Such higher exposure is expected, especially study is, to our knowledge, one of the longest continuous when indoor workers probably spent time outdoors only later periods among the studies assessing the validity of these surro- during the day, when ambient UVR was low. Furthermore, as gates, enabling us to address seasonal variations in personal expected, our measured personal UVR exposure levels in sum- behaviour. We found that individually, modelled personal mer were higher than those in winter. Moreover, our UVR exposure and self-reported time spent outdoors could observed predictive capability of different surrogates across predict more than three-quarters of the variation in the mea- strata of occupation (indoor vs. outdoor) and season (winter sured personal UVR levels, whereas ambient UVR levels could vs. summer) were comparable. However, our stratified analy- predict only around one-fifth of this variation. When we strat- ses resulted in smaller coefficients of determination (R2), ified our analyses by season and occupation, we observed sim- maybe because the variation due to important factors deter- ilar predictive capability of each surrogate between winter and mining personal UVR, namely occupation and season, were summer as well as between indoor and outdoor workers. effectively removed from these analyses. We found that modelled personal UVR exposure (based on One strength of this study was the high compliance rate, a combination of self-reported time spent outdoors and ambi- especially considering our relatively long study period. A total ent UVR levels) and self-reported time outdoors were equally of 949% of questionnaires were returned along with data good and better predictors of measured personal UVR expo- from 949% of the SunSavers. Low compliance has been a sure than ambient UVR. Previous studies9,11–13 exploring UVR problem in previous studies of the validity of the surrogates exposure modelling approaches include an algorithm of exposure to UVR.15,17,19 Furthermore, our study included (Genesis-UV) to retrospectively estimate lifetime occupational both rural (outdoor workers) and urban (indoor workers) exposure, and a 3D numeric model (SimUVex/SimUVex v2) populations. Also, the continuous study period of 6 months that estimates site-specific erythemal doses received at an indi- was sufficient to capture most seasonal behavioural variability. vidual level; therefore, it is not possible to compare their find- Our study also faced some limitations. Our nonrandom ings with ours. However, our observation that self-reported convenience sampling as well as differences in cultures, sun time outdoors and a modelling approach could better predict habits and personal behaviours could limit the generalizability measured personal UVR exposure compared with the ambient of our findings to other populations and settings. Moreover, UVR is in line with our previous simulation study,6 which considering that our study area was relatively small, we did showed that ambient UVR alone could explain only 16% of not have a large variation in ambient UVR levels, which could the variability in the annual measured personal UVR exposure have resulted in our underestimation of the proportion of the whereas self-reported time spent outdoors along with ambient variation in the personal UVR exposure that could be UVR increased the explained variation to 40%. Furthermore, explained by the variation in ambient UVR levels, especially in previous studies6,30 have shown a larger within-city (where indoor workers who spent much of their time indoors. Fur- ambient UVR is similar) variation in personal UVR exposure thermore, our modest sample size could have limited our sta- compared with between-city variation, supporting a more tistical power to detect patterns, especially for our subanalysis important role of personal behaviour than ambient UVR in of indoor/outdoor workers. The model we used7 to assess determining personal exposure to UVR. exposure to UVR assumes that outdoor exposure is symmetri- Moreover, previous studies9,10,14,16,20,22,31 exploring ambi- cal around the solar noon, but this assumption is less likely to ent UVR as a surrogate of measured personal UVR exposure be true for indoor workers, whose exposure more likely over a period of 7 days to 7 months have shown a predictive occurs early and late in the day. Moreover, the absence of dif- capability (R2) ranging from 002 to 074, which is in line ferences between the questionnaires could have been biased with our findings for this surrogate (R2 = 021). Studies by Questionnaire 2 being tied to having undertaken Question- exploring the ability of the self-reported time spent outdoors naire 1. Additionally, the SunSavers were not completely to predict measured personal UVR exposure were performed waterproof, leading to some loss of data. Finally, we did not over a period of 4–26 days and reported a predictive capabil- ask about high-risk hours outdoors in our questionnaire and, ity ranging between 003 and 074,15–17,19–21 while for this we measured it only objectively. This aspect could be added surrogate our observed R2 ranged from 075 to 079. to questionnaires in future epidemiological studies. We did not observe any notable difference between the two In conclusion, in our sample of adult indoor and outdoor questionnaires that we used to obtain information on time workers, modelled personal UVR exposure and self-reported spent outdoors. This observation could indicate that the same time outdoors could equally predict over three-quarters of the data validity can be obtained by collecting data monthly (as in variation in our objectively measured personal UVR exposure Questionnaire 1) or 1 month after finishing the 6-month-long over a period of 6 months. In contrast, ambient UVR was a study (as in Questionnaire 2). This finding can have important relatively poor predictor of the variation in the measured © 2021 The Authors. British Journal of Dermatology British Journal of Dermatology (2022) 186, pp266–273 published by John Wiley & Sons Ltd on behalf of British Association of Dermatologists 272 Validity of sun questionnaires and modelled exposure, L. Soueid et al. personal UVR exposure. We did not observe major differences 6 Dadvand P, Basaga~ na X, Barrera-G omez J et al. Measurement errors in the relative validity of each surrogate to predict measured in the assessment of exposure to solar ultraviolet radiation and its personal UVR exposure between indoor and outdoor workers impact on risk estimates in epidemiological studies. Photochem Photo- biol Sci 2011; 10:1161–8. or between summer and winter. Moreover, we did not find 7 Diffey B. A behavioral model for estimating population exposure any notable benefit from obtaining monthly data on self- to solar ultraviolet radiation. Photochem Photobiol 2008; 84:371–5. reported time outdoors compared with a single acquisition 8 Schmalwieser AW, Siani AM. Review on nonoccupational personal 1 month after the study. These findings could enable future solar UV exposure measurements. Photochem Photobiol 2018; 94:900– epidemiological studies of the health effects of long-term UVR 15. exposure to optimize their assessment of this exposure. At the 9 Wittlich M, Westerhausen S, Kleinespel P et al. An approximation same time, our findings highlight, again, the critical role of of occupational lifetime UVR exposure: algorithm for retrospective assessment and current measurements. J Eur Acad Dermatology Venereol individual behaviour in determining personal exposure to 2016; 30:27–33. UVR. Further multicentric studies are required to replicate our 10 Moldovan HR, Wittlich M, John SM et al. Exposure to solar UV findings in other settings with different cultures and climates, radiation in outdoor construction workers using personal dosime- relying on a larger sample size as well as different study areas try. Environ Res 2020; 181:108967. to maximize the variation in personal behaviour as well as 11 Vernez D, Milon A, Vuilleumier L et al. A general model to predict ambient UVR level. individual exposure to solar UV by using ambient irradiance data. J Expo Sci Environ Epidemiol 2015; 25:113–18. 12 Cheng W, Brown R, Vernez D, Goldberg D. Estimation of individ- Acknowledgments ual exposure to erythemal integral calculation. Sensors (Basel) 2020; 20:4068. This work forms part of the ICEPURE project which was 13 Religi A, Moccozet L, Farahmand M et al. SimUVEx v2: a numeric funded by the European Commission Framework Program 7 model to predict anatomical solar ultraviolet exposure. Proceedings (grant no. 227020). We are thankful to Mr Juan R. Moreta of the 2016 SAI Computing Conference (13–15 July 2016, Lon- from the State Meteorological Agency (AEMET) for supplying don, UK). SAI 2016; 2016:1344–8. us with the ambient UVR data. M.T.-M. is funded by a Juan 14 Cahoon EK, Wheeler DC, Kimlin MG et al. Individual, environmen- tal, and meteorological predictors of daily personal ultraviolet de la Cierva fellowship (FJCI-2017-33842) and P.D. is funded radiation exposure measurements in a United States cohort study. by a Ram on y Cajal fellowship (RYC-2012-10995); both are PLoS One 2013; 8:e54983. awarded by the Spanish Ministry of Economy and Competi- 15 Cargill J, Lucas RM, Gies P et al. Validation of brief questionnaire tiveness. We acknowledge support from the Spanish Ministry measures of sun exposure and skin pigmentation against detailed of Science and Innovation through the ‘Centro de Excelencia and objective measures including vitamin D status. Photochem Photo- Severo Ochoa 2019–2023’ Program (CEX2018-000806-S), biol 2013; 89:219–26. and support from the Generalitat de Catalunya through the 16 Chodick G, Kleinerman RA, Linet MS et al. Agreement between diary records of time spent outdoors and personal ultraviolet radi- CERCA Program. The sponsor or funding organizations had no ation dose measurements. Photochem Photobiol 2008; 84:713–18. role in the design or conduct of this research, and all the 17 Dwyer T, Blizzard L, Gies PH et al. Assessment of habitual sun authors declare that they have no conflict of interests. exposure in adolescents via questionnaire – a comparison with objective measurement using polysulphone badges. Melanoma Res 1996; 6:231–9. References 18 Gies P, Roy C, Toomey S et al. Solar UVR exposures of primary school children at three locations in Queensland. Photochem Photobiol 1 Alfredsson L, Armstrong BK, Allan Butterfield D et al. Insufficient 1998; 68:78–83. sun exposure has become a real public health problem. Int J Environ 19 Glanz K, Gies P, O’Riordan DL et al. Validity of self-reported solar Res Public Health 2020; 17:1–15. UVR exposure compared with objectively measured UVR expo- 2 Lucas RM, Yazar S, Young AR et al. Human health in relation to sure. Cancer Epidemiol Biomarkers Prev 2010; 19:3005–12. exposure to solar ultraviolet radiation under changing stratospheric 20 Køster B, Søndergaard J, Nielsen JB et al. The validated sun expo- ozone and climate. Photochem Photobiol Sci 2019; 18:641–80. sure questionnaire: association of objective and subjective mea- 3 Public Health England. Report RCE-30. Ultraviolet radiation, vita- sures of sun exposure in a Danish population-based sample. Br J min D and health: report of the independent Advisory Group on Dermatol 2017; 176:446–56. Non-ionising Radiation. March 2017; 10–14. Available at: https:// 21 Thieden E,  Agren MS, Wulf HC. Solar UVR exposures of indoor www.gov.uk/government/publications/ultraviolet-radiation-and- workers in a working and a holiday period assessed by personal vitamin-d-the-effects-on-health (last accessed 12 September 2021). dosimeters and sun exposure diaries. Photodermatol Photoimmunol Pho- 4 van der Rhee H, de Vries E, Coomans C et al. Sunlight: for better tomed 2001; 17:249–55. or for worse? A review of positive and negative effects of sun 22 Kovacic J, Wittlich M, Malte S, Macan J. Personal ultraviolet radia- exposure. Cancer Res Front 2016; 2:156–83. tion dosimetry and its relationship with environmental data: a lon- 5 Vecchia P, Hietanen M, Stuck BE et al. Protecting workers from gitudinal pilot study in Croatian construction workers. J Photochem ultraviolet radiation. International Commission on Non-Ionizing Photobiol B Biol 2020; 207:111866. Radiation Protection, in collaboration with: International Labour 23 Statistical Institute of Catalonia. Population on 1 January. Pro- Organization and World Health Organization, 2007; 12–29. Avail- vinces. 2020. Available at: https://www.idescat.cat/pub/?id= able at: http://www.who.int/uv/publications/Protecting_Workers_UV_ aec&n=245&lang=es (last accessed 28 September 2021). pub.pdf (last accessed 12 September 2021). British Journal of Dermatology (2022) 186, pp266–273 © 2021 The Authors. British Journal of Dermatology published by John Wiley & Sons Ltd on behalf of British Association of Dermatologists Validity of sun questionnaires and modelled exposure, L. Soueid et al. 273 24 Council of Europe. Barcelona, Spain – Intercultural City. 2018. 30 Lucas R, McMichael T, Smith W, Armstrong B. Solar ultravio- Available from: https://www.coe.int/en/web/interculturalcities/ let radiation: global burden of disease from solar ultraviolet barcelona (last accessed 27 September 2021). radiation. Environmental Burden of Disease Series, no. 13. Geneva: 25 Bodekær M, Harrison GI, Philipsen P et al. Personal UVR exposure World Health Organization, Public Health and the Environ- of farming families in four European countries. J Photochem Photobiol ment, 2006. B Biol 2015; 153:267–75. 31 Sun J, Lucas RM, Harrison S et al. The relationship between ambi- 26 Heydenreich J, Wulf HC. Miniature personal electronic UVR ent ultraviolet radiation (UVR) and objectively measured personal dosimeter with erythema response and time-stamped readings in a UVR exposure dose is modified by season and latitude. Photochem wristwatch. Photochem Photobiol 2005; 81:1138–44. Photobiol Sci 2014; 13:1711–18. 27 Heydenreich J, Wulf HC. Personal electronic UVR dosimeter mea- surements: specific and general uncertainties. Photochem Photobiol Sci Supporting Information 2019; 18:1461–70. 28 Thieden E, Philipsen PA, Heydenreich J, Wulf HC. UV radiation Additional Supporting Information may be found in the online exposure related to age, sex, occupation, and sun behavior version of this article at the publisher’s website: based on time-stamped personal dosimeter readings. Arch Dermatol Materials and Methods S1 Estimating personal solar ultravio- 2004; 140:197–203. let radiation exposure through time spent outdoors, ambient 29 Thieden E, Agren MS, Wulf HC. The wrist is a reliable body site for personal dosimetry of ultraviolet radiation. Photodermatol Photoim- levels and modelling approaches: Study setting and popula- munol Photomed 2000; 16:57–61. tion; Data collection; Questionnaires. © 2021 The Authors. British Journal of Dermatology British Journal of Dermatology (2022) 186, pp266–273 published by John Wiley & Sons Ltd on behalf of British Association of Dermatologists