Stage distribution of breast cancer: a basis for the evaluation of breast screening programmes.

Stage at presentation can be predictive of future mortality rates. A change in stage distribution will be indicative of the effect of a programme. All registrations for 1985 aged over 50 years were staged from case notes and, to assess whether a stage drift is already occurring, cases that presented in 1975 were also staged. The accuracy and completeness of the Wessex Cancer Register for this particular cancer was assessed and further sources of registration identified. There was a significant difference between 1975 and 1985 stage distributions. If the Southampton and Salisbury service achieves the same success as the Gavleborg County trials, 4.61 deaths will be prevented at 5 years for each annual cohort of cases. For individual breast screening programmes, effectiveness is better monitored by analysis of stage distribution than of mortality rates.