Abstract
OBJECTIVES: After completion of the prevalent screening round of the breast cancer screening programme in Limburg, The Netherlands, incidences started rising once again. This increase was contrary to expectations, which had predicted a slightly higher, but stable, incidence after the first screening round. The trends in incidence were studied to find explanations for the observed rise in incidence. SETTING: Breast cancer screening programme in mid-Limburg and southern Limburg, the Netherlands. METHODS: The data files of the breast cancer screening programme and the Maastricht cancer registry were linked to evaluate the effect of breast cancer screening. Only the first primary breast tumour was included in the evaluation. RESULTS: The second peak of incidence after the prevalent screening round was 45% higher than the incidences before the start of the screening. Also, the decrease in incidence of large and node positive tumours was interrupted. Compared with national detection rates, the number of screen detected cancers was lower before 1995 and higher after that year. After 1997, incidence decreased again of all breast cancers, but also of large and node positive tumours. The incidence of node positive tumours showed large fluctuations, probably due to the introduction of the sentinel node procedure and immunohistochemistry. In 1999, incidence of large tumours and node positive tumours was 18% and 28%, respectively, lower than before the start of the screening. CONCLUSIONS: An increase in the background incidences and improved detection in the screening programme most likely explain this trend. The improved detection after 1995, and the lower than desirable decrease in large tumours, indicate that the screening performance was not optimal before 1996. The incidence of node positive tumours cannot be used any more as an indicator of the success of the screening programme because of detection bias.