br The Prevent macrosimulation model
The Prevent macrosimulation model [8,9] was used to model projections of the number of cancer cases in the Nordic countries in the 30-year period, 2016e2045. A
more detailed description of the Prevent model can be found elsewhere [9,10], and we used a similar approach as in other Nordic studies to estimate avoidable cancers according to changes in prevalence of smoking, alcohol consumption and overweight/obesity [11e13]. We applied the Prevent model separately to each country and to the three cancer sites investigateddpostmenopausal breast (defined as breast cancer diagnosed at age 50 years and above), colon and endometrial cancerdand for three investigated hypothetical scenarios (see below). The Pre-vent model requires data on disease incidence, projected 540737-29-9 size, risk factor prevalence, relative risk (RR) estimates and changes in risk factor prevalence under hypothetical scenarios of interest.
Incidence rates, by cancer site, country, gender and age groups, were based on the incidence during the years 2009e2013 and obtained from NORDCAN [14e16]. Table 1 lists the International Classification of Diseases (ICD) codes used to define the cancer sites and the average annual number of cases in the Nordic countries. The estimated population size in the years 2016e2045, by gender and 5-year age groups, was obtained from population projections by the statistical bureaus in the respective countries [17e21].
Data from the Nordic Monitoring System on diet, physical activity and overweight in the Nordic Countries (the NORMO study), which includes self-reported sur-vey data for physical activity, were used to estimate current activity levels in the Nordic countries. The
Cancer sites, relative risk estimates and the average annual incident cases (2009e2013) in the Nordic countries.
Cancer site ICD-10 code Avg. # cases per year Relative risk
in the Nordic countries
MET-hours per week) MET-hours per week) MET-hours per week)
MET, metabolic equivalents.
information on leisure time (including transportation) physical activity was converted from hours to metabolic equivalents hours (MET-h), based on the assumption that 1 h of moderate intensity physical activity corre-sponds to 3 MET-h and that 1 h of vigorous intensity physical activity corresponds to 6 MET-h . We then categorised the MET-h per week in <3, 3 to <9, 9 to <15, 15þ and refer to these groups as high deficit, me-dium deficit and low deficit in physical activity and reference group (sufficient level of physical activity). The data used in Prevent are the proportion of individuals, by country, gender and age group, in each of the cate-gories, in 2011 and 2014. More information about the data can be found in Appendix A, together with the prevalence in each category in year 2014 by country, gender and age group.
We assumed that 15 or more MET-h per week are sufficient to avoid increased risk of cancer. This is higher than the general World Health Organisation (WHO) recommendation, but this cut-off has been used previously for estimating the population attributable fraction (PAF) of physical activity on cancer , and for cancer prevention, it is likely that the greater the amount of physical activity the greater the benefit . The RRs for low deficit (9 to <15 MET-h per week), medium deficit (3 to <9) and high deficit (<3) were estimated based on results from the World Cancer Research Fund Continuous Update Project (WCRF CUP)  and are presented in Table 1. The RRs from the WCRF CUP give the decrease in risk with increasing physical activity. We used an approach similar to Parkin  to transform the RR estimates to
RR for each of the categories of deficit in physical ac-tivity. The WCRF CUP does not present a RR per MET-h for endometrial cancer, so for endometrial cancer, we used the same RR as for breast cancer, which again is the approach used by Parkin . A more detailed description of the calculation of RR estimates is found in Appendix B. To take into account that the introduction of a change in prevalence will take some time to reach its full effect, the Prevent model includes a LAT and LAG time. During the LAT time, the risk remains unchanged, and during the LAG time, the risk among previously exposed gradually changes to reach the risk among never exposed (or unexposed). We used a LAT time of 1 year and a LAG time of 9 years, with the RR changing linearly during the LAG time.
We investigated three hypothetical scenarios A, B and C, to show the potential impact of changes in physical activity levels on the cancer burden relative to continued constant physical activity levels.
A Elimination of insufficient levels of physical activity in 2016