2. Introduction

2.1 Background, and rationale for long-term follow-up

In the UK, women of age 50-70 are invited about every 3 years for two-view digital mammographic screening, with the first invitation generally at ages 50-52 and the 7th and last at ages 68-70. The Independent UK panel on breast cancer screening concluded in 2012 that 7 such invitations at ages 50-70 would reduce breast cancer mortality by about 43 per 10,000 (an average of 6 per 10,000 per invitation), but that the effects of additional screening outside the range 50-70 remained uncertain.¹

AgeX seeks to address this uncertainty by providing large-scale randomised evidence on the effects of one additional screening at ages 47-49 and, separately, the effects of one additional screening at 71-73.  It randomised 4.5 million women (2.8 million younger and 1.7 million older), half allocated to be sent one additional screening invitation and half not.

Of them, 3 million (2 million younger and 1 million older) are included in the main analyses population (Section 5). Randomisation excludes all possibility of bias between invitees and controls, and appropriate analyses (Section 6) can yield unbiased estimates of the effects of one additional screening visit, and of one additional screening invitation.

Randomisation lasted from mid-2009 (when the pilot phase began) to March 2020, when COVID temporarily disrupted the national breast screening program. Screening the last few invitees was delayed by this disruption for some months but was eventually completed. By 2024 the entire AgeX database had been consolidated and linked for long-term (purely electronic) follow-up to all relevant national datasets, and pre-randomisation characteristics had been defined from sources that cannot have been affected by the random allocation.

To assess the long-term (for a decade or two after randomisation) direct and indirect effects of one additional screening, AgeX is continuing low-cost electronic follow-up (involving no contact with participants). This involves linkage to routinely collected NHS England records of breast cancer incidence and treatment, of hospital episode statistics, and of cause-specific mortality (as coded by the Office for National Statistics), but no other contact with the NHS or the screening program. 

The AgeX data monitoring and ethics committee (DMEC) will continue during long-term electronic follow-up to review annually (as it has been doing throughout the entire study) confidential interim analyses, particularly of mortality. The first two published reports on mortality are scheduled to be in 2027-28 (the interim analysis, of mortality to 12.2026) and 2032-33 (the final analysis, of mortality to 12.2031) unless the DMEC recommends release of earlier mortality results.

Because recruitment was completed some years before scientifically informative evidence on mortality could reasonably have been expected, in reports of the scheduled interim and final mortality analyses no statistical allowance will be made for the DMEC having periodically examined the data. For, follow-up will continue anyway, so the interim and final analyses cannot be affected by any DMEC reports. This 2024 statistical analysis plan (SAP) is for the 2025-33 follow-up phase of AgeX. It describes the pre-specified primary and main subsidiary analyses, based on version 7.0 of the AgeX study protocol (earlier versions specified similar analyses), and adheres to guidelines for SAP content.²

2.2 Primary objective

The primary objective is to assess long-term effects (for a decade or two, depending on each woman’s date of randomisation) of one additional screening on breast cancer mortality. The primary analyses will assess the effects of actually having had one additional screening visit, and parallel analyses will assess the effects of having being randomly allocated to receive one additional screening invitation.

2.3 Main subsidiary objectives

The main subsidiary objectives are to assess long-term effects of one additional screening visit on breast cancer incidence (partly to help assess the extent of over-diagnosis) and on the use of mastectomy, of breast-related radiotherapy, and of systemic breast cancer therapy, particularly chemotherapy (to assess the net effects of screening on the eventual extent to which these treatments are given).

Again, analyses parallel to these main subsidiary analyses will assess the effect of one additional screening invitation. Health economic analyses will accompany the final analyses of breast cancer incidence, treatment and mortality.