MétaCan
Menu
← Back to cohort

Metformin to treat breast cancer: an application of the prevalent new-user design for observational studies with no active comparator

2025· article· en· W4411635310 on OpenAlexafffund
Charles Khouri, Sophie Dell’Aniello, Hui Yin, Laurent Azoulay, Samy Suissa

Bibliographic record

VenueJournal of Clinical Epidemiology · 2025
Typearticle
Languageen
FieldBiochemistry, Genetics and Molecular Biology
TopicMetabolism, Diabetes, and Cancer
Canadian institutionsMcGill University Health CentreJewish General Hospital
FundersCanadian Institutes of Health ResearchCanada Foundation for Innovation
KeywordsObservational studyMetforminMedicineOncologyBreast cancerInternal medicineComparatorCancerInsulinEngineering

Abstract

fetched live from OpenAlex

OBJECTIVES: Numerous observational studies have reported significant reductions in cancer outcomes, including breast cancer in women, with metformin use. However, most studies were affected by immortal time bias. We assessed whether metformin use in women diagnosed with breast cancer is associated with lower breast cancer-related and all-cause mortality and illustrate the impact of immortal time bias on the results. STUDY DESIGN AND SETTING: The Clinical Practice Research Datalink was used to identify a base cohort of all women with a new diagnosis of breast cancer and with type 2 diabetes, at least 18 years of age, between 1998 and 2020. We employed the prevalent new-user design to match metformin initiators 1:1 with nonusers on a prior diabetes diagnosis and time-conditional propensity scores. We also used the naïve approach that introduces immortal time when classifying metformin users. Hazard ratios (HRs) and 95% CIs of all-cause and breast cancer-related death were estimated. RESULTS: The base cohort included 13,314 women newly diagnosed with breast cancer and with type 2 diabetes, before (n = 4761) and after (n = 8553) their breast cancer diagnosis, of which 5047 initiated metformin during follow-up. The prevalent new-user design included 4923 metformin initiators and 4923 matched nonusers. The HRs of breast cancer-related and all-cause mortality were 1.12 (95% CI: 0.98-1.28) and 0.96 (95% CI: 0.89-1.05), respectively. The naïve approach, among women with diabetes at cohort entry, which included 1354 metformin users and 3407 metformin nonusers, resulted in adjusted HRs of 0.45 (95% CI: 0.40-0.50) and 0.58 (95% CI: 0.54-0.62) for breast cancer and all-cause mortality. CONCLUSION: In this study, the use of metformin was not associated with a reduced risk of breast cancer-related and all-cause mortality. Using the flawed approach not accounting for immortal time bias, we confirmed the implausible beneficial effects of metformin on breast cancer and all-cause mortality reported in previous studies. PLAIN LANGUAGE SUMMARY: Observational studies have reported that the antidiabetic drug metformin can increase the survival of women with breast cancer. However, these studies were shown to have a flaw in their analysis, called "immoral time bias", known to exaggerate the benefit of a drug. We used a cohort of over 13,000 women with breast cancer to investigate the effectiveness of metformin on reducing mortality in women with breast cancer, using both the flawed and a correct time-matched approach. Using the flawed approach, we confirmed the implausible beneficial effects of metformin on breast cancer-related and on all-cause mortality reported in previous studies. Using the correct time-matched approach, we found that the use of metformin was not associated with these beneficial effects, confirming the impact of the flaw.

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 imitation

Not 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.

metaresearch head score (Codex)0.170
metaresearch head score (Gemma)0.163
Version: metacan-v3-hybrid-931329e0061cValidation status: machine_predicted_unvalidated
Candidate categoriesMetaresearch
Consensus categoriesnone
DomainCandidate signal: Methods · Consensus signal: none
Study designCandidate signal: Observational · Consensus signal: none
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.830
Threshold uncertainty score0.900

Distilled classifier scores by category (both heads)

CategoryCodexGemma
Metaresearch0.1700.163
Meta-epidemiology (narrow)0.0030.002
Meta-epidemiology (broad)0.0070.012
Bibliometrics0.0030.002
Science and technology studies0.0030.003
Scholarly communication0.0030.003
Open science0.0050.004
Research integrity0.0060.004
Insufficient payload (model declined to judge)0.0060.000

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.

Opus teacher head0.297
GPT teacher head0.520
Teacher spread0.222 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.

Study designObservational
DomainMethods
GenreEmpirical

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".

Quick stats

Citations1
Published2025
Admission routes2
Has abstractno

Explore more

Same venueJournal of Clinical Epidemiology→Same topicMetabolism, Diabetes, and Cancer→French-language works237,207→