Annex 1

Effects of changing the details of the definition of the age range 

The seemingly trivial distinction between birth-year age (current year minus birth year) and birthday age caused problems with breast screening that were not clearly recognised until 2018. If this year’s birthday has already happened then these two ages are the same, otherwise the birthday age is 1 less than the birth-year age; on average, therefore, it is 6 months less.

When 3-yearly breast screening began in 1988 England was divided into about 80 areas, each with its own screening unit. Over the subsequent decades, few areas changed. The large majority of units adopted a 3-year cycle, reviewing successive GP lists and inviting the women on each list according to birth-year age (50-70, since the mid-2000s). So, birth-year age could equally well have been 50, 51 or 52 at the first invitation, and 68, 69 or 70 at the last.

Hence, 1/3 of all women in these areas got their first invitation at birth-year age 50, at a time when half (ie, 1/6 of all women) had not yet reached their 50th birthday. Likewise, 1/3 got their last invitation at birth-year age 68, at a time when half (1/6 of all women) had not yet reached their 68th birthday. 3 years later these women, being of birth-year age 71, would not be invited again, although still of birthday age 70. This affected about 1/6 of all women reaching age 70, and was later considered inappropriate.

Although breast screening invitations traditionally depended on birth-year age, and continued to do so, new wording introduced by Public Health England (PHE) in 2013-14 inadvertently specified the upper age limit in terms of birthday age (by saying that 3-yearly screening would continue “until the 71st birthday”). It was not realised that changes in invitation procedures were implied, but from then on a discrepancy existed, affecting substantial numbers of women per year, between what the invitation procedures were doing and what they were said to be doing.

During 2018 there was much concern about use of birth-year age 50-70 rather than birthday age 50-70 to decide who to invite, but it should be noted that it made no difference to the frequency or total number of invitations. On average it made invitations start, and so end, younger by 6 months of age, but there is no good evidence as to whether this was better or worse for women.

In 2009, AgeX began randomising whether or not to continue inviting women who would otherwise have been just too old to be invited, which at that time meant those of birth-year age 71-73. (It also began randomising whether or not to offer an extra invitation at younger ages.)

In July 2016 new PHE software was introduced that, in addition to inviting birth-year age 50-70, provided the option (not always used) of also inviting women of birth-year age 71 who were still of birthday age 70. This software should have ensured that use of this option would exclude such women from AgeX (as half would get no invitation), but it did not. This was the only relevant IT error; there was no error in 2009, when the trial began.

From July 2016 to January 2018 some women of birthday age 70 entered the trial. During 2018 those allocated no further screening were informed by PHE that they could still request screening, and most were also offered a specific 2018 screening appointment. As most women of birthday age 70 who entered the trial after mid-2016 would get an invitation regardless of their allocation, all are excluded from the primary analyses (but will be reported on separately).

Early 2018 concerns about women of age 70 not being offered an invitation centred at first on the trial, even though it had never had any involvement with, or access to, the patient invitation software. Enquiries later in 2018 clarified the fundamental role of the inadvertent discrepancy (which was unrelated to AgeX) between what the patient invitation software had been doing ever since the UK screening program began and the newly prepared 2013-14 specification.

This discrepancy has since been resolved; since September 2018, invitations are based on birthday age. This means the first invitation should arrive at birthday age 50, 51 or 52 (on average 6 months later than it used to) and the last should arrive at birthday age 68, 69 or 70 (again, on average 6 months later than it used to). It also means that women entering AgeX since September 2018 are of birthday age 47-49 or of birthday age 71-73. (NB To allow for delays before screening, randomisation selects ages 46 years 8 months to 49 years 10 months.)