Bibliographic record
Abstract
The Canadian Breast Screening Studies 1 and 2 (CNBSS) were set up to estimate the effect of breast cancer screening on mortality from the disease (1,2). Since the first publications of mortality results, they have been the subject of considerable debate, mainly focusing on the possibility of nonrandom allocation to screening or control, the appropriateness or otherwise of design aspects of the studies, and the quality of mammography (1–4). The year 2021 saw the online publication of an article by Yaffe et al documenting eyewitness statement that random allocation to the screening or control arms was a clear protocol violation in at least one center, and that symptomatic patients were routinely recruited to the trial in another (5). This bombshell is augmented by two articles in the current issue of the Journal of Breast Imaging (6,7). The first article reports on a survey of professional staff involved with the CNBSS, including radiologists, both within the trial and independent reviewers, radiographic technologists, study coordinators, and others (6). The survey explores issues of clarity of design, training of staff, inclusion/exclusion, integrity of random allocation, and quality of mammography. While the survey is pragmatic rather than systematic, it is clear from the responses that the quality of mammography was often inadequate, and that symptomatic women were not excluded from the CNBSS. For staff at several centers, training was inadequate, and there was little transparency of trial processes. This last is particularly important. As noted in the second article in this issue, the chief methodological concern raised by commentators on CNBSS was the possibility of the randomization being a clear protocol violation to preferentially include women with a suspicion of breast cancer in the screening arm, evidenced by an excess of advanced tumors at initiation in this arm (4). In the past, systematic reviewers have dismissed this possibility, on the basis of similar distributions of breast cancer risk factors in the total populations in the screening and control arms (8). First, this ignores the fact that the shift of a small but crucial number of women with pre-existing symptomatic breast cancer would have no discernible effect on the risk factor distributions in the entire randomized population (5). Second, and more importantly, now that there is eyewitness statement that this differential allocation did indeed occur in at least one center (5,7), there can no longer be any justification for assuming that it did not happen or did not bias the results of CNBSS. The second article also documents other major concerns with CNBSS, notably the inclusion/exclusion policy, which allowed recruitment of symptomatic women and the poor quality of mammography (7). It also points out a lack of transparency of study policy, which allowed considerable variation among centers in terms of interpretation of recruitment and allocation procedures. Both papers conclude that CNBSS can no longer be considered safe to include as evidence to inform screening policy. Sadly, this reader can only come to the same conclusion. One might ask how much it matters in 2022. Unfortunately, CNBSS is still cited prominently as evidence against breast cancer screening, particularly in the age subgroup 40–49 years (9), and the above indicates that such citation is inappropriate. What should be done to remedy the situation? In the first instance, one should congratulate the witnesses who have spoken out about shortcomings of CNBSS, notably those who have reported serious issues with allocation (5,7). This must have taken some courage. Second, systematic reviews and meta-analyses can no longer justify the inclusion of CNBSS as providing evidence of adequate quality on the efficacy of breast cancer screening. There is an urgent need to re-review the evidence excluding CNBSS, particularly in terms of target age groups for screening. None declared.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.328 | 0.556 |
| Meta-epidemiology (narrow) | 0.002 | 0.003 |
| Meta-epidemiology (broad) | 0.004 | 0.003 |
| Bibliometrics | 0.021 | 0.041 |
| Science and technology studies | 0.015 | 0.010 |
| Scholarly communication | 0.012 | 0.004 |
| Open science | 0.015 | 0.007 |
| Research integrity | 0.004 | 0.006 |
| Insufficient payload (model declined to judge) | 0.010 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".