Brown · British journal of cancer 2018 · Epidemiological modeling and population attributable fraction analysis · n=?

The fraction of cancer attributable to modifiable risk factors in England, Wales, Scotland, Northern Ireland, and the United Kingdom in 2015.

Cited 561 times in the scientific literature.

Level 3 - non-randomized controlled study

Population attributable fraction modeling combining cohort study meta-analyses, national surveys, and cancer registry data

PubMed 29567982 · doi:10.1038/s41416-018-0029-6 · record verified 2026-08-28

What was done

The authors calculated population attributable fractions (PAFs) for combinations of modifiable risk factors and cancer types with sufficient or convincing evidence of causal association in the UK and its constituent countries (England, Wales, Scotland, and Northern Ireland) for 2015. Relative risks were derived from meta-analyses of cohort studies, risk factor exposure prevalence was extracted from nationally representative surveys, and 2015 cancer incidence was obtained from national registry data. Calculations were stratified by age, sex, exposure level, and country.

What was found

In 2015, 37.7% of all UK cancer cases were attributable to known modifiable risk factors (38.6% in males and 36.8% in females). - Across constituent nations, the attributable proportion was highest in Scotland (41.5%) and lowest in England (37.3%). - Tobacco smoking contributed the largest proportion of cases at 15.1%, followed by overweight/obesity at 6.3%. - For 10 cancer types, including lung cancer and melanoma skin cancer, over 70% of cases were attributable to known risk factors.

Why it matters

This study provides localized, country-level estimates of preventable cancer burdens across the UK, establishing that over a third of cancers remain preventable and confirming tobacco and obesity as the leading priorities for public health policy.

Limits

The findings are mathematical model estimates rather than directly observed causal outcomes. Total case and participant counts are not specified in the abstract. Calculations depend on the accuracy and availability of self-reported exposure surveys, which the authors note likely biases estimates toward underestimation. Potential multi-factor interactions and unmeasured lifestyle confounders were not detailed in the abstract.

Cited by