Measurement errors in the assessment of exposure to solar ultraviolet radiation and its impact on risk estimates in epidemiological studies.

To date, many studies addressing long-term effects of ultraviolet radiation (UVR) exposure on human health have relied on a range of surrogates such as the latitude of the city of residence, ambient UVR levels, or time spent outdoors to estimate personal UVR exposure. This study aimed to differentiate the contributions of personal behaviour and ambient UVR levels on facial UVR exposure and to evaluate the impact of using UVR exposure surrogates on detecting exposure-outcome associations. Data on time-activity, holiday behaviour, and ambient UVR levels were obtained for adult (aged 25-55 years old) indoor workers in six European cities: Athens (37°N), Grenoble (45°N), Milan (45°N), Prague (50°N), Oxford (52°N), and Helsinki (60°N). Annual UVR facial exposure levels were simulated for 10,000 subjects for each city, using a behavioural UVR exposure model. Within-city variations of facial UVR exposure were three times larger than the variation between cities, mainly because of time-activity patterns. In univariate models, ambient UVR levels, latitude and time spent outdoors, each accounted for less than one fourth of the variation in facial exposure levels. Use of these surrogates to assess long-term exposure to UVR resulted in requiring more than four times more participants to achieve similar statistical power to the study that applied simulated facial exposure. Our results emphasise the importance of integrating both personal behaviour and ambient UVR levels/latitude in exposure assessment methodologies.