Breast cancer screening in an urban, Swedish population Aspects of non-attendance, interval cancers and over-diagnosis

Abstract: Service screening with mammography was implemented in Sweden in the late 80's, following the results from trials in Sweden and abroad. A high rate of attendance, a high diagnostic accuracy and treatment in accordance with established guidelines are key circumstances for an effective screening programme. These factors have to be continuously monitored. One trial was carried out 1976-86 in the city of Malmö, Sweden: the Malmö Mammographic Screening Trial (MMST). 42 000 women aged 45-69 were randomly allocated either to be invited to screening or to constitute a control group which was not screened. After the termination of the trial the Malmö Mammographic Service Screening Programme (MMSSP) was implemented in 1990. All women aged 50-69 are regularly invited to screening. The transition from trial to service screening programme provides a natural, experimental setting for epidemiological studies of factors of significance for the effectiveness of screening in different time periods and under different screening premises.

The aim of the present thesis was to focus on three issues in an urban, Swedish population of relevance for the effectiveness of mammographic screening: non-attendance, interval cancers and over-diagnosis.

In study I rates of non-attendance ranged from 23-43% between 17 residential areas. There was a high correlation between rate of non-attendance and a socio-economic score in the areas, r=-0.78. Women living in less affluent areas participated to a lower extent. In study II it was demonstrated that attendance was lower in the MMSSP than in the MMST, 65% versus 74%. Stage distribution and survival among non-attenders seemed to have improved in the MMSSP compared to in the MMST. Several individual socio-economic factors, such as marital status, employment status, foreign background and income were identified as predictors of non-attendance in the MMSSP. The interval cancer rate was 1.5/1000 women screened in the MMSSP, study III. The prognosis for women with interval cancer was significantly better in the MMSSP compared to the MMST. Interval cancer cases in the MMSSP did not have any worse prognosis compared to cases in an unscreened group of women. In study IV it was shown that 15 years after the end of the MMST, the rate of over-diagnosis was 10% in the invited group compared to the control group in the ages 55-69 years at randomisation.

In conclusion, the results in study I and II may be used to improve the attendance rates in screening by allocating resources to appropriate groups and areas. Women should be informed about the risk for interval cancers and over-diagnosis, studies III and IV. However, interval cancers seem to be less of a problem in the current screening situation. The rate of over-diagnosis must be evaluated together with the benefits of screening in terms earlier diagnosis and treatment and reduced breast cancer mortality.