Evidence Trace, complete trial record
In plain terms
In plain terms: Mammography screening for breast cancer has some human trial evidence supporting it, but the evidence is not yet strong enough for EBL to call it established. It is worth knowing about and worth discussing with your clinician.
Mammography screening for breast cancer
B
Supported: a good trial, backed by more
The endpoint this grade is based on: Hard clinical endpoint (mortality, major event)
The evidence as the record states it
Nelson et al. (Ann Intern Med, 2016; PMID 26756588), the review behind the US Preventive Services Task Force update: breast cancer deaths over 10 years, relative risk 0.92 (0.75 to 1.02) at ages 39 to 49 (9 trials; 3 deaths prevented per 10,000 women), 0.86 (0.68 to 0.97) at 50 to 59 (7 trials; 8 per 10,000), 0.67 (0.54 to 0.83) at 60 to 69 (5 trials; 21 per 10,000) and 0.80 (0.51 to 1.28) at 70 to 74; death from any cause not reduced; most trials used imaging and treatment now outdated. Gøtzsche and Jørgensen (Cochrane Database Syst Rev, 2013; PMID 23737396): 7 trials, about 600,000 women aged 39 to 74; the three with adequate randomization did not show a significant reduction in breast cancer deaths at 13 years (0.90, 0.79 to 1.02), the four with suboptimal randomization did (0.75, 0.67 to 0.83); no effect on death from any cause (0.99, 0.95 to 1.03); more lumpectomies and mastectomies in the screened groups (1.31, 1.22 to 1.42). Bretthauer et al. (JAMA Intern Med, 2023; PMID 37639247): no significant lifetime gain from mammography (0 days, -190 to 237).
PMID 26756588 PMID 23737396 PMID 37639247
Evidence by specific use
Breast cancer deaths, women aged 50 to 69
Grade B
Nelson et al. (Ann Intern Med, 2016; PMID 26756588): breast cancer deaths 14% lower at 50 to 59 (7 trials) and 33% lower at 60 to 69 (5 trials), 8 and 21 deaths prevented per 10,000 women over 10 years. The Cochrane review dissents: the three trials it judged adequately randomized were not significant (relative risk 0.90, 0.79 to 1.02), and only the four with suboptimal randomization were (PMID 23737396). Positive pooled trials with a dissent on trial quality: B.
PMID 26756588 PMID 23737396
Breast cancer deaths, women aged 39 to 49
Grade D
Nelson et al. (PMID 26756588) pooled 9 trials in women aged 39 to 49: breast cancer deaths relative risk 0.92 (0.75 to 1.02), about 3 deaths prevented per 10,000 women over 10 years, not statistically significant. Tested, no clear benefit: D on Route 2.
PMID 26756588
Death from any cause, women invited to mammography screening
Grade D
Death from any cause was not reduced: relative risk 0.99 (0.95 to 1.03) in the Cochrane review (PMID 23737396), not reduced in the Task Force review (PMID 26756588), and 0 days of lifetime gained in a pooled analysis (Bretthauer et al., 2023; PMID 37639247). Tested, no benefit: D on Route 2.
PMID 23737396 PMID 26756588 PMID 37639247
Why Grade B and not Grade A
Grade B reflects moderate human evidence -- typically a single strong trial, multiple smaller trials, or surrogate endpoint data without hard clinical outcome replication. To reach Grade A, Mammography screening for breast cancer would require at least one additional independent randomized trial demonstrating a hard clinical endpoint or objectively measured functional outcome in a comparable population.
What this grade covers and what it does not
Grade B on breast cancer deaths in women aged 50 to 69: pooled randomized trials find fewer deaths, 8 per 10,000 women at 50 to 59 and 21 per 10,000 at 60 to 69 over 10 years (PMID 26756588). WHY NOT A: the three trials judged adequately randomized did not show a significant reduction (PMID 23737396), the trials are decades old, and cause of death may have been misclassified in favor of screening. Grade D, tested, on breast cancer deaths at 39 to 49, where 9 pooled trials did not show a significant reduction (PMID 26756588), and on death from any cause at every age (PMIDs 23737396, 26756588, 37639247). WHAT THIS GRADE DOES NOT COVER: tomosynthesis (3D mammography), MRI or ultrasound screening, which these trials did not test; women at high inherited risk, such as BRCA carriers; women with a breast lump or other symptoms, for whom imaging is diagnosis, not screening; and women over 74.
What would move this grade
An independent randomized trial demonstrating a hard clinical endpoint -- mortality, major adverse event, or objectively confirmed functional improvement -- in an adequately powered sample, with results independently replicated, would support Grade A.
How to track whether this is working
Measured in the trials: deaths attributed to breast cancer, deaths from any cause, and numbers of breast surgeries.