An examination of the methods for assessing preventability of cancer

Edward Giovannucci at the Harvard TH Chan School of Public Health is researching the different ways in which the proportion of potentially preventable cancers is estimated

  • Topic: Combination of cancers
  • Institution: Harvard School of Public Health
  • Country: United States
  • Status: Ongoing
Researcher: Edward Giovannucci

Estimating the percentage of cancers that could be prevented in a population if certain lifestyle choices are changed is important. This would have implications for public health recommendations, and inform the distribution of medical and research resources.

The preventability of cancer can be worked out as the number of cancer cases in a population that could be prevented if lifestyles were modified to minimise risk. The proportion of potentially preventable cancers is called the Population Attributable Risk (PAR).

Estimates for the preventability of cancer have varied widely, in part due to different data sources and differences in assessing the relevant risk factors. Moreover, there are two methods in general use to estimate the PAR, which may produce differing results.

The approach most commonly used determines the relative risks (size of the effect) for each risk factor (exposure), estimates the prevalence of the risk factor in a population and uses standard formulae to calculate PAR.

The second approach identifies a low risk group and compares this with actual rates in the population (for example, if the low-risk group had half the age-standardised rate as the US population, presumably half of the cancers are potentially preventable if the population achieved the low-risk profile). This new project will compare the results of these two methods and, if they differ, determine the factors accounting for this.

This project also aims to assess the influence of various factors on the PAR for diet, physical activity and body weight on cancer. Five factors have been identified as influencing the PAR:

  • The time lag between an exposure and the risk for cancer. For example, an exposure from childhood and its effect on cancer risk in adulthood.
  • Dose-response relation with cancer; to what extent does the volume and intensity of a risk factor, such as alcohol consumption, affect cancer risk.
  • The joint contributions of multiple risk factors; how following the whole package of World Cancer Research Fund’s Cancer Prevention Recommendations is more likely to reduce cancer risk than changing just one factor or behaviour.
  • Studies that determine relative risks (which lead to working out the PAR) often rely on single measurements of diet which do not reflect long-term dietary patterns.
  • Smoking, as an important risk factor, complicates PAR calculations.

Reviewing the effect of the different approaches as well as understanding and assessing how these diet, physical activity and body weight factors influence PAR estimates is the broad objective of this project.

Grant publication

> Population attributable risk for colorectal and breast cancer in England, Wales, Scotland, Northern Ireland, and the United Kingdom [version 2; peer review: 2 approved] (AMRC Open Research, March 2021; revised March 2022)