Using an alternative data source to examine randomization in the Canadian National Breast Screening Study

Area of Science:

• Oncology
• Public Health
• Epidemiology

Background:

• Criticisms of the Canadian National Breast Screening Study (NBSS) suggest potential bias in participant allocation.
• Concerns exist that women with prior breast disease may have been disproportionately assigned to mammography arms.

Purpose of the Study:

• To examine the randomness of participant allocation in the NBSS.
• To investigate if women with prior breast disease were non-randomly assigned to mammography groups.

Main Methods:

• Retrospective observational study using independent health insurance claims data from Manitoba, Canada.
• Comparison of self-reported breast disease history with health insurance claims prior to NBSS enrollment.
• Analysis of 9477 women with health insurance claims in the 24 months preceding NBSS entry.

Main Results:

• No significant differences in prior benign breast disease, investigation, or xeromammogram claims were found across study arms (p > 0.05).
• A small imbalance in prior breast cancer claims was noted, but subsequent claims were minimal.
• No differences were observed in the concordance between self-reported history and insurance claims across study arms.

Conclusions:

• External health insurance data provide no definitive evidence of non-random allocation of women with prior breast disease to mammography arms in the NBSS.
• Findings are specific to Manitoba participants, and generalizability to other NBSS centers requires caution.