Response to Drs. Igor Burstyn and George Luta’s letter: Advice on better utilization of validation data to adjust odds ratios for differential exposure misclassification (recall bias)
Bibliographic record
Abstract
We want to thank Drs. Burstyn & Luta (1) for their recognition of our recent study (2) suggesting that estimates of breast cancer risk following retrospective self-reported night shift work are inflated by recall bias. The main strength of the study was a gold standard based on individual, prospective, objective and detailed information on night shift work that allowed validation of self-reported night shift work obtained after breast cancer was diagnosed – the usual situation for case–control studies (3). The study confirmed what textbooks have long taught but rarely documented empirically (4–6). We also want to thank Drs. Burstyn & Luta for their advice on how we could have utilized this precious dataset not only for simple but also probabilistic and Bayesian quantitative bias analyses. Even if highly instructive, this may still require strong statistical involvement. Using data provided in our paper, Drs. Burstyn & Luta’s bias-corrected odds ratio (OR) estimate of breast cancer following night shift work was centered around 1.0 (95% credible interval 0.3–1.7) and suggests that recall bias could completely, and not only partly as in our analysis (OR 1.05; correctly computed 95% confidence interval 0.88–1.27), explain the observed associations between night shift work and breast cancer found in case–control studies with retrospective self-reported exposure information. This finding strengthens our concern that breast cancer studies based on retrospective self-reports of night shifts may not provide convincing evidence. The gold standard of this validation study was based on a cohort of healthcare workers with day-by-day night shift information from a pay roll register and has earlier been used for breast cancer risk assessment showing no increased risk (7). A recent Swedish study using comparable data neither showed an overall increased risk (8). However, these studies included only information on recent night shift work. The next step should be a follow up of the cohorts when information on more distant night shift work becomes available with an emphasis on analyses that explore the timing of shift work on breast cancer risk. References 1. Burstyn I, Luta G. Advice on better utilization of validation data to adjust odds ratios for differential exposure misclassification (recall bias). Scand J Work Environ Health – online first. https://doi.org/10.5271/sjweh.4226 2. Vestergaard JM, Haug JN, Dalbøge A, Bonde JP, Garde AH, Hansen J et al. Validity of self-reported night shift work among women with and without breast cancer. Scand J Work Environ Health. 2024;50(3):152–7. https://doi.org/10.5271/sjweh.4142. 3. Cordina-Duverger E, Menegaux F, Popa A, Rabstein S, Harth V, Pesch B et al. Night shift work and breast cancer: a pooled analysis of population-based case-control studies with complete work history. Eur J Epidemiol. 2018; 33(4):369–79. https://doi.org/10.1007/s10654-018-0368-x. 4. Berrington de González A, Richardson DB, Schubauer-Berigan MK. Statistical methods in cancer research, Volume V. Bias assessment in case–control and cohort studies for hazard identification. Lyon, France: International Agency for Research on Cancer; 2024. 5. Checkoway H, Pearce N, Kriebel D. Research Methods in Occupational Epidemiology. New York: Oxford University Press 2004. p372. 6. Rothman KJ. Modern Epidemiology. Boston/Toronto: Little, Brown and Company; 1986. p358. 7. Vistisen HT, Garde AH, Frydenberg M, Christiansen P, Hansen AM, Hansen J et al. Short-term effects of night shift work on breast cancer risk: a cohort study of payroll data. Scand J Work Environ Health. 2017;43(1):59–67. https://doi.org/10.5271/sjweh.3603. 8. Gustavsson P, Bigert C, Andersson T, Kader M, Härmä M, Selander J et al. Night work and breast cancer risk in a cohort of female healthcare employees in Stockholm, Sweden. Occup Environ Med. 2023;80(7):372–6. https://doi.org/10.1136/oemed-2022-108673. Henrik Albert Kolstad, MD,1, 2 Jesper Medom Vestergaard, MIT,1 Jens Peter Bonde, MD,3 Sadie Costello, PhD,4 Annett Dalbøge, PhD,1 Åse Marie Hansen, PhD,5, 6 Ann Dyreborg Larsen, PhD,6 Anne Helene Garde, PhD 5, 6 1 Occupational and Environmental Medicine, Danish Ramazzini Centre, Aarhus University Hospital, Aarhus, Denmark. 2 Department of Clinical Medicine, Aarhus University, Denmark. 3 Department of Occupational and Environmental Medicine, Bispebjerg and Frederiksberg Hospital, Denmark. 4 Environmental Health Science, School of Public Health, University of California, Berkeley, USA. 5 The National Research Centre for the Working Environment, Denmark. 6 Department of Public Health, University of Copenhagen, Denmark. Correspondence to: Henrik Kolstad, Occupational and Environmental Medicine, Danish Ramazzini Centre, Aarhus University Hospital, Aarhus, Denmark. [E-mail: kolstad@clin.au.dk]
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.010 | 0.002 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.002 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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.
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 teacher head, 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".