Health impact modelling for UVR and melanoma skin cancer

The text on this page is taken from an equivalent page of the IEHIAS-project.

As part of the EU-funded INTARESE project, which contributed to the development of this Toolbox, a case study was carried out to assess health impacts of exposures to ultra-violet radiation (UVR) under different climate change scenarios.

The assessment considered two sets of scenarios for the years 2030 and 2050: the IPCC SRES B1 scenario and SRES A2 scenario. Health outcomes considered were basal cell carcinoma (BCC), squamous cell carcinoma (SCC) and malignant melanoma (CMM). Changes in UVR exposure were estimated, taking account of changes in stratospheric ozone associated with decreasing emission of ozone depleting substances and increasing emissions of CH₄ and N₂O. Allowance was also made for future demographic changes.

Below the methods used to model non-melanoma skin cancer are decribed. Details of the working procedures are given in the attached spreadsheet.

Explanation of the method

The assessment model used Excel as a platform, in order to calculate the number of incident cases and the number of deaths at baseline. Calculations were performed for 5-year age groups (up to 85+), both for males and females, in three study areas (London, Rome and Helsinki).

In each case, future changes in ambient UVR (annual erythemal dose) were combined with exposure-response functions (ERF) for melanoma (incidence) and (future) population estimates. The ERFwas derived from Scotto and Fears’ (1987) estimation of the biological amplification factor (BAF= relative change in disease risk due to a 10% increase in UVR exposure), adjusted for age (Table 1).

Table 1. Biological amplification factor (BAF) (associated with a to a 10% increase in UVR exposure) adjusted for age (Scotto and Fears 1987)

	Face, head or neck (FHN), upper extremity (UE) *	Trunk & lower extremity (TL) **
Male	8 %	6 %
Female	10 %	5 %

* higher risk estimate; standard model setting; ** lower risk estimate

Several health impact model runs were carried out (Table 2). The baseline model and model 1 used the baseline incidence rates per age group and gender. For model 2, future incidence rate per age group and gender were calculated by applying the BAF and the future percental change in modelled ambient UVR to the baseline incidence rates. The (future) incidence rates were combined with the baseline and future population estimates in order to calculate skin cancer incidence (number of cases).

The results were then combined with the baseline and future population estimates in order to calculate skin cancer incidence (number of cases). Thus:

• Baseline model: baseline incidence rates and baseline population estimates are applied.
• Model 1: baseline incidence rates and future population estimates are applied.
• Model 2: estimated future incidence rates and future population estimates are applied.

Table 2. UVR health impact model runs for malignant melanoma

Model run	Incidence/mortality rate	Population
Baseline model (2000)	Baseline incidence/mortality rates	Baseline population size and structure
Model 1 (2030 and 2050): population change only	Baseline incidence/mortality rates remain unchanged	Future population size and structure: SRES A2 and B1
Model 2 (2030 and 2050): population change combined with changes in ambient UVR	Future incidence rate is affected by changes in ambient UVR levels: SRES A2 and B1	Future population size and structure: SRES A2 and B1

Current and future mortality (rates) per age group and gender were calculated using the age-specific incidence-mortality ratios, based on the incidence and morality rates in the EURA region as presented by Lucas et al (2006).

Attributable incidence and mortality were calculated using the upper estimate of population attributable fractions (PAF) as estimated by the WHO (Lucas et al. 2006). The upper PAF estimate for melanoma is 0.9, while the lower estimate is 0.5.

Details of the models are summarised below. (See the appended spreadsheet for a full worked example.)

Baseline model for CMM

Attributable incident cases at baseline Attributable I(baseline)x = I(baseline)x * PAF I(baseline)x = i(baseline)x * Population(baseline)x Attributable number of deaths at baseline Attributable M(baseline)x = M(baseline)x * PAF M(baseline)x= m(baseline)x * Population m(baseline)x= i(baseline)x / imr

Model 1 for CMM

The baseline incidence and mortality rates are combined with future population projections for A2 en B1 in order to explore the effects of population change only. Model 1: future attributable incident cases for selected scenarios Attributable I (scenario)x = I (scenario)x * PAF I(scenario)x= i(baseline)x * Population(scenario)x Model 1: future attributable number of deaths for selected scenarios Attributable M(scenario)x = M(scenario)x * PAF M(scenario)x = m(baseline)x * Population(scenario)x m(baseline)x = i(baseline)x / imr

Model 2 for CMM

The future incidence and mortality rates are combined with future population data. Model 2: future attributable incident cases for selected scenarios Attributable I(scenario)x = I(scenario)x * PAF I(scenario)x= i(scenario)x * Population(scenario)x i(scenario)x= i(baseline)x * [1+∆i(scenario)x] ∆i(scenario)x = 10* [1-∆E(scenario)x] * BAFx ∆E(scenario)x = [E(baseline)- E(scenario)/ E(baseline)] Model 2: future attributable number of deaths for selected scenarios Attributable M(scenario)x = M(scenario)x * PAF M(scenario)x= m(scenario)x * Population(scenario)x m(scenario)x= i(scenario)x / imr

where:

Population(baseline)_x = population size at baseline (in 100000), age/gender group x

Population (scenario)_x = population size (in 100000) in the selected scenario, age/gender group x

I(baseline)_x = number of incident cases at baseline, age/gender group x

I(scenario)_x = number of incident cases in the selected scenario, age/gender group x

i(baseline)_x = baseline incidence rate per 100000, age/gender group x

i(scenario)_x = incidence rate per 100000 in the selected scenario, age/gender group x

M(baseline)_x = number of deaths at baseline, age/gender group x

M(scenario)_x = number of deaths in the selected scenario, age/gender group x

m(baseline)_x = baseline mortality rate per 100000, age/gender group x

m(scenario)_x = mortality rate per 100000 in the selected scenario, age/gender group x

imr= incidence: mortality ratio

BAF_x = biological amplification factor, age/gender group x

PAF= population attributable fraction, UVR

E(baseline)= modelled ambient UVR at baseline

E(scenario) = modelledambient UVR in selected scenario

Selected scenarios: SRES B1 2030, SRES B1 2050, SRES A2 2030, SRES A2 2050

References

• Scotto, J. and T. Fears 1987 The association of solar ultraviolet and skin melanoma incidence among Caucasians In the United States. Cancer Investigation 5(4), 275-283.
• Lucas, R., T. McMichael, et al. 2006 Solar ultraviolet radiation: Global burden of disease from solar ultraviolet radiation. Geneva:, World Health Organization.

See also

• Changing ambient UVR and future melanoma skin cancer
• Changing ambient UVR and future non-melanoma skin cancer
• File:Melanoma health impact model Helsinki.xls