Statistical analysis

A detailed statistical analysis plan is provided separately (Annex 6). The principal features of the main analyses are as follows.

Analysis as two separate trials

The findings will be monitored, analysed and reported as two separate trials. One is a trial among younger women (randomly allocated at age 47-49 to one additional screening invitation or control). The other is a trial among older women (randomly allocated at age 71-73 to one additional screening invitation or control).

Restriction to 2 million younger women and 1 million older women

Of the 4.5 million women randomised, only 3.0 million are included in the main analyses population. Women were included only if, when randomised, they were the right age, alive, linkable to routinely collected NHS England electronic records, without a history of cancer or other breast disease, and likely to accept if invited (based on their previous screening history). All these criteria were based unbiasedly on information recorded before randomisation and were chosen blind to analyses of mortality differences between invitees and controls. None of these criteria excluded significantly different proportions of invitees and controls.

Effects of actually having one additional screening VISIT

The primary and main subsidiary analyses will be of the effects of actually having one additional screening visit (from adherence-corrected analyses: see next section).

Parallel analyses will also be provided of the effects of being randomly allocated to receive one additional screening invitation (from intent-to-treat analyses). These under-estimate, perhaps by about one-quarter, the effects of actually having an additional screening. For, the difference between invitees and non-invitees in the proportions actually screened was about three-quarters, both in younger women and in older women.

Primary and main subsidiary endpoints

The primary endpoint is breast cancer mortality (based only on the certified underlying cause of death). The main subsidiary endpoints are invasive breast cancer (overall, and subdivided by tumour characteristics such as diameter, grade, ER status and stage), in situ breast cancer, use of mastectomy, use of systemic breast cancer therapy (particularly chemotherapy), and use of radiotherapy for breast disease.

Mortality from a breast cancer diagnosed < 4 years after randomisation

A few years after just one single additional screening invitation, the annual incidence rates of newly diagnosed breast cancer may well become similar in invitees and in controls. No material effect of the random allocation should be expected on mortality from breast cancers that are diagnosed after this convergence of incidence rates, so the mortality analyses focus on deaths from breast cancers that had been diagnosed only a few years after randomisation.

Mortality after younger women reach age 55 and older women reach age 75

No material effect of the random allocation should be expected on breast cancer mortality during the first few years after randomisation. Hence, the primary analyses are restricted to the later breast cancer deaths.

Primary analyses

The primary analyses will be of the effects of one additional screening visit on numbers of deaths from a breast cancer diagnosed < 4 years after randomisation that occurred after reaching age 55 (for younger women) or after reaching age 75 (for older women).

Although the primary analyses will disregard oestrogen receptor (ER) status, corresponding analyses of mortality from ER+, ER– or ER unknown breast cancer will also be given.

Main subsidiary analyses

The main subsidiary analyses will be of breast cancer incidence and treatment. Additional subsidiary analyses will investigate any effects of screening on other outcomes that are available from linkage with routinely collected NHS records. The final primary and subsidiary analyses will be accompanied by appropriate health economic analyses. Analyses of other mortality and all-cause mortality will be given, but uninformative.

Allowance for cluster randomisation

Although AgeX is cluster-randomised the primary analyses of breast cancer mortality will average only about 0.1 breast cancer deaths per cluster, so they will have virtually the same statistical power as individually randomised comparisons of similar size would have had.