P P Glasziou1, A J Woodward, C M Mahon. 1. Department of Social and Preventive Medicine, University of Queensland Medical School, Brisbane.
Abstract
OBJECTIVE: To carry out a systematic quality review and meta-analysis of all randomised trials of mammographic screening that included women aged under 50 years. DATA SOURCES: Reports of randomised trials of mammographic screening were identified via MEDLINE and checks of the bibliographies of retrieved articles and reviews. DATA SYNTHESIS: Identified trials were assessed for: (i) method of randomisation; (ii) documented comparability of baseline data; (iii) standardised criteria for breast cancer death; (iv) blinded review of cause of death; (v) completeness of follow-up; and (vi) use of an "intention-to-treat analysis". Seven randomised trials including almost 160,000 women aged under 50 were studied. The combined estimate of relative risk was 0.95 (95% confidence interval, 0.77-1.18), a statistically non-significant reduction of 5%. Adjustment for the cluster randomisation of two trials, and for degree of compliance, did not substantially change this result. CONCLUSIONS: These analyses suggest little, if any, benefit for women under 50 years of age. The results are not explained by the quality of the trials or the radiology. We recommend that women in this age group intending to be screened should be fully informed of these results.
OBJECTIVE: To carry out a systematic quality review and meta-analysis of all randomised trials of mammographic screening that included women aged under 50 years. DATA SOURCES: Reports of randomised trials of mammographic screening were identified via MEDLINE and checks of the bibliographies of retrieved articles and reviews. DATA SYNTHESIS: Identified trials were assessed for: (i) method of randomisation; (ii) documented comparability of baseline data; (iii) standardised criteria for breast cancer death; (iv) blinded review of cause of death; (v) completeness of follow-up; and (vi) use of an "intention-to-treat analysis". Seven randomised trials including almost 160,000 women aged under 50 were studied. The combined estimate of relative risk was 0.95 (95% confidence interval, 0.77-1.18), a statistically non-significant reduction of 5%. Adjustment for the cluster randomisation of two trials, and for degree of compliance, did not substantially change this result. CONCLUSIONS: These analyses suggest little, if any, benefit for women under 50 years of age. The results are not explained by the quality of the trials or the radiology. We recommend that women in this age group intending to be screened should be fully informed of these results.