Cancer incidence inconsistency between UK Biobank participants and the population: a prospective cohort study

Abstract

Background

While the UK Biobank has been widely used for cancer research, its representativeness of the population in terms of cancer incidence has not been thoroughly investigated.

Methods

We conducted a prospective cohort study of 466,163 UK Biobank participants who were cancer-free at recruitment. Standardised incidence ratios (SIRs) were calculated for all cancers combined and for 25 cancers, by comparing incidences for the participants with the UK national incidences. Variations in SIR by age, sex and deprivation measures were investigated.

Results

Over a median follow-up period of 12 years, 47,535 participants had a cancer diagnosis. The SIR for all cancers combined was 0.90 (95% CI: 0.89, 0.91). The SIR increased with age and deprivation (P = 10⁻⁹). The SIRs of 17 cancers differed from 1 (Bonferroni-adjusted P < 0.05): for prostate cancer and melanoma the SIRs were 1.2 and for the other 15 cancers the SIRs ranged from 0.43 to 0.93. The SIRs of 13 cancers differed by deprivation: the greater the deprivation, the lower the SIRs for prostate cancer and melanoma, and the higher the SIRs for the other 11 cancers.

Conclusions

The overall cancer incidence was 10% lower for the UK Biobank participants compared with the population, with most cancers having a lower incidence that increased with deprivation. Irrespective of their causes, the inconsistencies could bias UK Biobank research results related to absolute cancer risks, such as the development and/or validation of cancer risk models and penetrance estimates for cancer susceptibility genes.