NHS breast screening programme: is the high incidence of interval cancers inevitable?

BMJ 1996; 313 doi: (Published 30 November 1996) Cite this as: BMJ 1996;313:1369
  1. David Asbury, clinical directora,
  2. Caroline R M Boggis, consultant radiologista,
  3. David Sheals, clinical directorb,
  4. Anthony G Threlfall, research officerc,
  5. Ciaran B J Woodman, directorc
  1. a Manchester Breast Screening Service, Withington Hospital, Manchester M20 0PT
  2. b Wigan Breast Screening Service, Royal Albert Edward Infirmary, Wigan Lane Wigan WN1 2NN
  3. c Centre for Cancer Epidemiology, Christie Hospital NHS Trust, Withington, Manchester M20 4QL
  1. Correspondence to: Professor Woodman.
  • Accepted 19 September 1996

In the NHS breast screening programme women are screened every three years; those presenting with breast cancer within three years of a negative test are considered to have an interval cancer. Unexpectedly high rates of interval cancers have been reported from the programme,1 2 but opinion is divided about what should be done.3 It is accepted that not all interval cancers could have been detected at the time of screening and that some will be true interval cancers, appearing de novo between screening rounds. We report how the occurrence of true interval cancers varies with time from screening.

Subjects, methods, and results

Interval cancers occurring before 31 March 1994 in women screened from 1 April 1988 to 31 March 1993 at the Manchester and Wigan breast screening services were identified,1 and a mammogram taken at the time of diagnosis was sought for all these cancers. Screening films were mounted on roller viewers by clerical staff; no attempt was made to replicate the screening situation, but some negative mammograms from women known not to have breast cancer were included (amounting to 10% of the total). The screening films were reviewed by three radiologists from a centre not involved in the initial assessment and consensus was reached on the presence or absence of a significant abnormality, the location of which was then checked by reference to the diagnostic films. An interval cancer was classified as a false negative when the same suspicious abnormality was present on both screening and diagnostic mammograms, as a true interval cancer when an abnormality was present only on the diagnostic mammogram, and as radiologically occult if no abnormality was present on either film. No attempt was made to classify interval cancers when a diagnostic mammogram was unavailable. Only interval cancers occurring in years for which cancer registration was complete were included in the analysis.

Two hundred and sixty interval cancers were identified; 13 were excluded as they were still awaiting radiological evaluation. Of the remaining 247 cases, 130 (53%) had a diagnostic mammogram and could be classified: 26 (39%), 51 (58%), and 53 (58%) of these presented in the first, second, and third year respectively after a negative screen. Four radiologically occult cancers have been excluded from table 1, which shows the frequency of true and false negative interval cancers with time from screening. The proportion of true interval cancers increased significantly with year from screening (2= 12.75; df=2; P<0.002).

Table 1

Frequency of true and false negative interval cancers in relation to time from screening

View this table:


Almost half of all interval cancers are diagnosed in the third year after screening.1 2 We found that the frequency of true interval cancers increases with time from screening and in the third year comprises 80% of all classifiable interval cancers. The absence of a diagnostic mammogram in many cases is an unsatisfactory but widespread finding in the NHS breast screening programme, and it is impossible to opine on the distribution of true and false negative interval cancers in these tumours. The proportion of interval cancers which can be classified has increased over time with greater clinical awareness of the importance of obtaining a diagnostic mammogram and increased provision of diagnostic mammography sets. We can, however, draw broad comparisons with other European screening programmes and trials with similar overall interval cancer rates, bearing in mind that these have a two year screening interval. The proportion of true interval cancers occurring in the first two years after screening in our series was 60%, similar to that reported in the Nijmegen programme (58%) and Stockholm trial (64%) in women aged 50–64.4 5 Recent changes designed to improve the sensitivity of the screening test are welcome, but we believe that shortening the screening interval is the only way to reduce the number of true interval cancers which will otherwise continue to present in the third year after screening.

We thank the staff of the Manchester and Wigan breast screening services for their help in the organisation of the radiological audit and the following radiologists for their participation in the audit: J Lavelle, M Wilson, A Troop, G Kelly, and P Lane.


  • Funding None.

  • Conflict of interest None.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
View Abstract