Interval cancers as an indicator of performance in breast screening

The literature shows that breast screening performance varies between centers. Those who monitor the performance of screening programs must use appropriate methods to demonstrate quality. Analysis of interval cancer rates provides one such measure. This study analyses interval cancers over a ten-year period in a single center of the UK National Breast Screening Program, offering screening to a population of 45,000 women.

108,948 screening examinations were undertaken between 1987 and 1996. Sequential interval cancer rates were studied and compared with the results of the Swedish Two-County trial. More detailed analysis of the first prevalence round was undertaken.

Improvement in rates occurred after three years. Two-year time bands show some fluctuation in rates, indicating variation in performance. For one time period, rates equivalent to the Two County Trial were achieved, but this was not consistently maintained. Overall rates equate to performance reported elsewhere in the UK (25% in the first year after screening, 48% in the second year after screening and 48% in the third year after screening, equivalent Two County rates are 17%, 32% and 57%). The prevalence round analysis shows that undiagnosed cancers arose after the film reading stage in 88% of interval cases. The team was able to distinguish minor signs from significant lesions effectively. Assessment procedures were effective once the patient had been recalled.

Interval cancer rates fell initially but reached a plateau. Recall procedures were effective in making the diagnosis, and cases were missed at the film reading stage. Consistency of performance may be important in achieving low rates.