Self-Selection Bias in Randomized and Observational Studies on Screening Mammography: A Quantitative Assessment

Area of Science:

• Oncology
• Biostatistics
• Public Health

Background:

• Observational studies on screening mammography effectiveness are susceptible to self-selection bias.
• Women attending screening differ in characteristics from those who do not.
• The Dr quantity method aims to correct for bias by comparing breast cancer death risk.

Purpose of the Study:

• To quantify self-selection bias in screening mammography studies.
• To estimate the Dr quantities necessary for bias correction.

Main Methods:

• Self-selection and Dr quantities were quantified using Swedish randomized trials.
• Self-selection in cohort studies was estimated using all-cause mortality relative risks.
• Dr quantities for observational studies were derived from cohort study self-selection estimates.

Main Results:

• Self-selection bias was 2.10 in Swedish trials and 1.78 in cohort studies.
• The required Dr quantity for observational studies was estimated at 1.53.
• The median Dr quantity used in 19 studies was 1.16, indicating undercorrection.

Conclusions:

• Women not attending screening have approximately double the risk of breast cancer death.
• This elevated risk is independent of screening mammography's efficacy.
• Most observational studies likely overestimate screening mammography effectiveness due to insufficient bias correction.