Effect of Screening on Breast Cancer Mortality
The accessibility of the data for international audit and verification also varies, with formal audit having been undertaken only in the Canadian trials. In fact, the author of one trial (Kopparberg) refused to respond to queries about methodology or to submit raw data for independent review.
All of these studies are designed to study breast cancer mortality rather than all-cause mortality, because of the infrequency of breast cancer deaths relative to the total number of deaths. When all-cause mortality in these trials was examined retrospectively, only the Edinburgh trial showed a significant difference, which could be attributed to socioeconomic differences. The meta-analysis (follow-up methods) of the four Swedish trials also showed a small but significant improvement of all-cause mortality.
The trials are listed chronologically.
Health Insurance Plan, United States 1963 [4,5]
|Age at entry: 40 to 64 years.|
|Randomization: Individual, but with significant imbalances in the distribution of women between assigned arms, as evidenced by menopausal status (P < .0001) and education (P = .05). |
|Sample size: 30,000 to 31,092 in study group and 30,565 to 30,765 in control group. |
|Consistency of reports: Variation in sample size reports. |
|Intervention: Annual two-view mammography (MMG) and CBE for 3 years. |
|Control: Usual care. |
|Compliance: Nonattenders to first screening (35% of the screened population) were not reinvited.|
|Contamination: Screening MMG was not available outside the trial, but frequency of CBE performance among control women is unknown.|
|Cause of death attribution: Women who died of breast cancer that had been diagnosed before entry into the study were excluded from the comparison between the screening and control groups. However, these exclusions were determined differently within the two groups. Women in the screening group were excluded based on determinations made during the study period at their initial screening visits. These women were dropped from all further consideration in the study. By design, controls did not have regular clinic visits, so the prestudy cancer status of control patients was not determined. When a control patient died and her cause of death was determined to be breast cancer, a retrospective examination was made to determine the date of diagnosis of her disease. If this was prior to the study period then she was excluded from the analysis. This difference in methodology has the potential for a substantial bias in comparing breast cancer mortality between the two groups, and this bias is likely to favor screening.|
|External audit: No. |
|Follow-up duration: 18 years. |
|Relative risk of breast cancer death, screening versus control (95% confidence interval [CI]): 0.71 (0.55-0.93) at 10 years and 0.77 (0.61-0.97) at 15 years. |
|Comments: The MMGs were of poor quality compared with those of later trials, because of outdated equipment and techniques. One should remember that the intervention consisted of both MMG and CBE. Major concerns about trial performance are the validity of the initial randomization and the differential exclusion of women with a prior history of breast cancer. |