Approach

Uncertainties were explored by means of sensitivity analyses, applied to the modelled results for Helsinki, under SRES-B1 2030, model 2.  In the sensitivity analyses we assessed the effect of changing the standard model settings with regard to the following factors (see also Table 1 below):

• Changes in ambient UVR due to other factors. In the standard model setting, we assumed that important (optical) properties of the surface and the atmosphere relevant for ambient UVR (e.g. albedo, cloud cover, aerosols etc) remain unchanged. In the sensitivity analyses we explore the effect of ±10% change in modelled ambient UVR (2030).
• Cloud cover. We explored the effect of a 10% increase in cloud cover in 2030, resulting in a 3.2% decrease in modelled ambient UVR compared to the standard model setting (which assumed no change in cloud cover). Based on current annual cloud cover (International Satellite Cloud Climatology Project (ISCCP) http://home.cc.umanitoba.ca/~jander/clouds/annual.png) and data from Josefsson and Landelius (2000).
• Population attributable fraction (PAF). We used the higher PAF estimates of the WHO (Lucas et al. 2006) in our standard model setting; the lower PAF is applied in the sensitivity analyses to account for uncertainty in the PAF.
• Biological Amplification Factor (BAF; CMM only). With regard to the BAF (Scotto and Fears 1987), the FHN/UE (face, head or neck, upper extremity) result is considered as the higher risk estimate and the TL (trunk & lower extremity) result as the lower estimate. The higher risk estimate represents the standard model setting in our calculations; the lower BAF is applied in the sensitivity analyses to account for uncertainty in the BAF.
• Additive effect of increasing summer temperatures: Somestudies indicate that the future increase surface temperature could also influence the relationship between ambient UV and skin cancer, by facilitating the induction of skin cancer and/or changing exposure behaviour (Ilyas 2007; van der Leun, Piacentini et al. 2008). We will take this into account using the epidemiological results by Van der Leun et al (2008): BCC incidence rates increase with 2.9% (SE1.4) per 1°C increase in daily summer max temperature; SCC incidence rates increase with 5.5% (SE 1.6) per 1°C increase in daily summer maximum temperature. In the sensitivity analyses we explored the effect of a 1 degrees Celsius increase in summer maximum temperature between 2000 and 2030, resulting in an increase in incidence rates between  2.9% (based on additive effect for BCC) and 5.5% (based on the additive effect for SCC).
• Incidence:mortality ratios. In the standard model setting, the estimation of age- and gender-specific incidence: mortality ratios (imr) isbased on the incidence and morality rates in the EURA region as presented by Lucas et al (2006). In the sensitivity analyses, we increase these standard imr with 10% in order to account for improved treatment and increased survival changes. As a second imr alternative, we apply age-specific imr based on the incidence and morality rates in the Helsinkiarea (Helsinki and Uusimaa) from the Finish Cancer registry. The third alternative combines the previous two by increasing the Helsinki imr with 10%.
• DALY calculations. Our standard model settings are based on uniform age weighing, no discounting and no adjustments for co-morbidity in the DALY calculations. In the sensitivity we apply non-uniform age-weighing (based on WHO BoD study), a 3% discount rate (based on WHO GBD study) and adjustment of disease weights for co-morbidity at higher ages (based on malignant melanoma spreadsheets Victorian BoD study). Please note that in WP3.7-UVR, the discounting for the 2030-scenarios calculates the net present value in 2030.