Breast cancer screening involves taking mammograms (x-rays) of women's breasts to search for signs of cancer. This study investigates the impact of test threshold, screening interval (frequency) and age of eligibility on intermediate outcomes, and health outcomes such as mortality and morbidity. This observational study links breast cancer screening, cancer registry and mortality registration data to answer these questions.

Breast cancer screening involves taking mammograms (x-rays) of women's breasts to search for signs of cancer. Different countries give different versions of breast screening, because there is no clear evidence which is best. This study investigates the effect of three variations to breast screening on outcomes for women screened.

Firstly, the test threshold. When radiologists examine the women's mammograms it is often not clear whether cancer is present. The test threshold means the tendency of radiologists to recall more or fewer women for further tests. In England overall 4% of women are recalled for further tests because their mammograms show suspicious signs, but different radiologists vary between 2% and 10%. This study investigates how recalling different proportions of women affects their long term outcomes.

Secondly the interval between screening invitations. In the UK breast screening is every 3 years, which is the longest time between screens in the world. In the US it is every year or 2 years and in Europe every 2 years. This study uses the natural variation in English screening interval to investigate the impact on women's outcomes.

Thirdly the age women are invited for breast screening. The NHS Cancer Plan in England resulted in extension of the upper age limit of eligibility for breast cancer screening from 64 to 70. This study investigates the change on women's outcomes.

The primary outcomes are health outcomes or close approximations, such as mortality, overdiagnosis of cancer which would never have caused harm, stage shift in cancer diagnosis, treatment (to reflect treatment associated morbidity), and false positive recall to assessment (which is known to induce anxiety). Secondary outcomes are intermediate outcomes with known but more proximal links to health outcomes: number of cancers detected at screening and their characteristics, and number of interval cancers detected between screening rounds.

This is an observational study linking women's screening records to cancer registry and mortality records from 1988 to 2018.

This observational study began on 1st January 2021, building on the POSTBOx study (NCT04365114). POSTBOx evaluates the impact of one or two readers on women's outcomes, POSTBOx primary outcomes 4 and 5 and secondary outcome 1 were dependent on obtaining follow up funding, which was achieved in this ATHENA-M project. The ATHENA-M funding also adds two additional exposures (screening interval and age of eligibility), and expands analysis of the test threshold exposure. Both projects were significantly delayed in data linkage and transfer, the investigators expect complete data transfer to the analysis team at Warwick between February and April 2022.

Inclusion Criteria:

• Women invited for breast cancer screening by the English NHS Breast Screening Programme

Exclusion Criteria:

• Missing or corrupted data for NHS number, so linkage between databases is not possible
• For analysis of test threshold and screening interval: women who did not attend routine English NHS Breast screening service within the specified date and age range, even if they attended symptomatic breast cancer services, high risk (family history) breast screening services, or if they were referred for mammograms by their general practitioner

Women invited for a Breast Cancer Screening appointment in England between 1st January 1988 and 31st March 2018. For test threshold and screening interval data this is further limited to women attending at least one breast screening appointment meeting the age and date eligibility requirements. The population will be further limited to women invited for breast cancer screening after 1st January 1998, if accurate data linkage cannot be achieved before this date in the absence of a unique identifier (before the widespread implementation of NHS number)

The investigators do not have Office for Data Release permissions to share these data more widely, but can assist other researchers in applying for access.