Over-diagnosis estimate from The Independent UK Panel on Breast Cancer Screening is based on unsuitable data

The Independent UK Panel on Breast Cancer Screening found 11% over-diagnosis in mammography screening based on data from randomized control trials (RCTs) in Malmö and Canada, claimed to be ‘RCTs without screening of [the] control group at the end of the trial’. ¹ We agree that, in order to estimate over-diagnosis, the control group should not be screened after the end of the trial. However, in order to estimate over-diagnosis, the study group should not be screened either, after the end of the trial, and the study group should be followed for a sufficiently long period after the end of screening. If this is not the case, over-diagnosis will be over-estimated, as there will be no room for the compensatory dip in breast cancer incidence expected after the end of screening. Unfortunately, both the Malmö and the Canada RCT data have problems in this respect.

The Malmö RCT, with no screening of the control group, included women born 1908–1922. ² Women born 1908–1912 were screened until the end of 1986, i.e. until they were on average 76.5 years old. With an average life-expectancy of 80.8 years, ³ this only allows for 4 years of follow-up after the end of screening. A recent study from Denmark showed that only allowing for 4 years of follow-up after the end of screening will not include the entire compensatory dip, and will thereby over-estimate over-diagnosis to be 10%. ⁴ Another problem with over-diagnosis estimates from the Malmö RCT is the fact that mammography was available outside the screening programme throughout the study. ⁵ Table 1 shows that after the end of screening for women born 1913–1922 belonging to the study group, a non-negligible proportion of the breast cancers were asymptomatic, suggesting that a non-negligible proportion of these women continued screening after the trial stopped in 1986. This would certainly result in the over-diagnosis estimate being too high, as these women cannot be followed for a sufficiently long time after the end of screening. It should also be noted that, among the women who actually stopped being screened at age 70 (i.e. women born 1918–1922), the over-diagnosis estimate was 1% (see Table 2). ⁶

Table 1

Breast cancers diagnosed 1987–1996 in invited group

Birth cohort	Clinicalcancers	Asymptomaticcancers	Cancers missinginformation
1918–1922	54	28	0
1913–1917	68	16	1
1908–1912	59	2	6

Table 2

Follow-up: incidence and relative risk for breast cancer from 1977 until end of follow-up, 31 December 2001, per age and trial group

Birth cohort	Age at randomization	Invited group			Control group			RR (95% CI) invited versus control
		Breast cancers(invasive) (N)	Person-years(py)	Incidence/1000 py*	Breast cancers(invasive) (N)	Person-years(py)	Incidence/1000 py*	All cancers	Invasive cancers
1908–1932	45–69	1320 (1200)	426,812	3.09	1205 (1116)	429,033	2.81	1.10 (1.01–1.18)	1.07 (0.99–1.17)
1908–1917	60–69	507 (469)	152,851	3.32	432 (410)	152,598	2.83	1.16 (1.02–1.32)	1.13 (0.99–1.29)
1918–1922	55–59	273 (250)	98,524	2.77	266 (252)	98,187	2.71	1.01 (0.85–1.19)	0.97 (0.82–1.16)

*All breast cancer, invasive and in situ

The two Canadian RCTs targeted women aged 40–49 ⁷ and 50–59, ⁸ respectively. However, shortly after the trials stopped, service mammography screening was implemented for women aged 50–69 in the majority of Canadian provinces from which the trial populations were recruited. ⁹ For a large proportion of the study groups, therefore, screening is unlikely to have ended at the conclusion of the trials. The Canadian study populations were followed for at least 5.5 years after the end of the trials, but given that a large number continued screening after the end of the trials, these women could not have been followed for a sufficiently long time after the end of screening.

Over-diagnosis is notoriously difficult to study, and RCT data are preferable, if available. However, none of the three RCTs where the control group was not screened at the end of the trial had the necessary follow-up period after the end of screening the study group to provide a reliable estimate of over-diagnosis.