OBJECTIVE: We designed a case-referent study to investigate the effect of mammographic screening at the individual level, looking at the association of breast cancer death with screening history. METHODS: The study population included all women aged 50-75 in the province of Limburg, the Netherlands who had been invited to the screening program from 1989 to 2006. From this population, 118 cases originated who died of breast cancer in 2004 or 2005. The screening history of these cases was collected and compared with a sample of the invited population. The breast cancer death rate in the screened relative to the unscreened women was estimated as the odds ratio (OR). This OR was adjusted for self-selection bias, the difference in baseline risk for breast cancer death between screened and unscreened women. RESULTS: Analysis of the data showed a breast cancer mortality reduction of 70% in the screened versus the unscreened women (OR = 0.30, 95% CI 0.14-0.63). The magnitude of self-selection was estimated specifically for Limburg. After correction for self-selection bias, the effect of screening increased to 76% (OR = 0.24, 95% CI 0.10-0.58). CONCLUSION: Screening resulted in a remarkable reduction in breast cancer mortality. Contrary to findings in other countries, adjustment for self-selection in Limburg had no influence on the impact of screening. Thanks to a well-organized centralized screening program, similar results are expected in other regions of the Netherlands.
OBJECTIVE: We designed a case-referent study to investigate the effect of mammographic screening at the individual level, looking at the association of breast cancer death with screening history. METHODS: The study population included all women aged 50-75 in the province of Limburg, the Netherlands who had been invited to the screening program from 1989 to 2006. From this population, 118 cases originated who died of breast cancer in 2004 or 2005. The screening history of these cases was collected and compared with a sample of the invited population. The breast cancer death rate in the screened relative to the unscreened women was estimated as the odds ratio (OR). This OR was adjusted for self-selection bias, the difference in baseline risk for breast cancer death between screened and unscreened women. RESULTS: Analysis of the data showed a breast cancer mortality reduction of 70% in the screened versus the unscreened women (OR = 0.30, 95% CI 0.14-0.63). The magnitude of self-selection was estimated specifically for Limburg. After correction for self-selection bias, the effect of screening increased to 76% (OR = 0.24, 95% CI 0.10-0.58). CONCLUSION: Screening resulted in a remarkable reduction in breast cancer mortality. Contrary to findings in other countries, adjustment for self-selection in Limburg had no influence on the impact of screening. Thanks to a well-organized centralized screening program, similar results are expected in other regions of the Netherlands.