Attributable risk - methodology

The supplement estimates the numbers and percentages of cancer cases in the UK in 2010 that were linked to exposure to a range of lifestyle and environmental factors earlier in life.1 

The proportion (or percentage) of such attributable cancers is the population attributable fraction (PAF), which provides an estimate of the total effect of a risk factor.

For most of the exposures, the PAF was calculated using how common the exposure was for each sex and age group in the population around the year 2000; the difference between the actual exposure and a theoretical optimum (by age group and sex); and the relative risk per unit difference.1 For tobacco smoking, the method developed by Peto et al (1992) was used; this assumes that smoking is the cause of almost all lung cancers, and uses the difference between lung cancer rates in non-smokers and the observed rates to estimate population exposure to smoking. A single study was used as the source of the expected number of lung cancers in the absence of smoking, and the relative risks for the other smoking-related cancers.2,3

Parkin et al (2011c)4 provides a summary of the PAFs for each exposure studied, by cancer type. The PAFs reflect the effect of one cancer risk factor, independently of other causes. One type of cancer can be attributed to multiple causes, and there is an overlap between having exposure to different risk factors. This means that summing the PAFs for each risk factor would overestimate the burden of that cancer resulting from a combination of them. Instead, the individual PAFs were combined by applying the percentage of cases attributable to one risk factor to the cases remaining after those linked to other risk factors have already been deducted (and this can be done in any order). This is based on the assumption that the risk factors are independent of each other, and that their relative risks can be multiplied when two or more are present. 

Throughout the supplement non-melanoma skin cancers were excluded.

section updated 07/12/11

• The risks of exposure were taken from published systematic reviews and meta-analyses of published studies.
• Risk factor exposure distributions were obtained from surveys usually at Great Britain level, mainly from around the year 2000. The predicted number of cancer cases in 2010 (by cancer type, sex, and five year age group) was obtained by using cancer incidence rates from Great Britain for 1975 to 2007, applied to UK population projections, to get the 2010 data.5 Prostate and breast cancer projections take into account the effect of screening.

section updated 07/12/11

For most risk factors, an interval of 10 years was assumed to be present between exposure and cancer incidence. This was based on the mean interval between measurement of exposure and the cancer outcome in the studies used as the source of data on the relative risks. The method used for tobacco smoking does not require estimation on the basis of past exposure.

The prevalence of exposures  was only available in the national surveys for those aged 15 and over. This means the PAFs could only be calculated for adults aged 25 and over, after allowing for the 10-year latency period. As only around 1% of cancers occur before the age of 25, the impact on the estimates will be minimal.

The strength of the evidence for each exposure is discussed in the individual sections.1  For example, the risk from smoking tobacco is certain and well-quantified, for dietary factors there is more controversy.

Quantifying the likely impact of preventive interventions requires complex scenario modelling. So for example, while 50% of colorectal cancer cases are attributable to lifestyle (diet, alcohol, physical inactivity and overweight), it has been estimated that only about half of this number are preventable over a 20-year period.6

PAFs should not be used to indicate the percentage of cancers that can currently be prevented by practical means without reference to the supplement, where some of the uncertainties and assumptions are discussed in detail.

section updated 07/12/11