Confidence intervals and interval cancers ... needles and haystacks?

The measurement of interval cancer rates will be critical to the successful evaluation of the breast cancer screening programme. The number of interval cancers expected in a district will depend on the population size, the test sensitivity and the background incidence of breast cancer. The present study was undertaken to measure the incidence of breast cancer in Northern Ireland, and to assess the practical problems that might be encountered in ascertaining interval cancers. Given a local yearly incidence of 182/100,000, only 14 interval cancers per year might be expected in Northern Ireland (population 1.5 million). The completeness of the local cancer register would not ensure reliable detection of this relatively small number of tumours. To give larger samples sizes and thus narrow the confidence intervals of measured rates, the results from several smaller regions could be aggregated. Different regions may have different mechanisms of ascertainment and this may make the interpretation of these rates more difficult.