{
  "abstract": "Objective To provide researchers with clear guidance on reporting key information to support bias assessment.Methods Researchers frequently work within the constraints of existing data and fixed study designs, whether analyzing completed studies, exploring new hypotheses with established datasets, or combining data from multiple studies of varying quality. In such contexts, it is crucial to assess potential confounding, information bias (e.g., measurement error or misclassification), and selection bias—either by the study team or by systematic reviewers and hazard assessors.Results Directed acyclic graphs can help identify potential biases, but appropriate reporting is essential for meaningful assessment. For confounding, important data include the identification of negative control outcomes and exposures that are related to the confounder, exposure period, confounder prevalence across groups, and its association with the exposure. For information bias, researchers should report the sensitivity and specificity of exposure measurements, validity of exposure and outcome measurement, along with relevant references, interview quality by case or control status and both unadjusted and adjusted risk estimates. For selection bias, useful information includes definitions and distributions of participants and non-participants (e.g., cases and controls), loss to follow-up rates in key subgroups, baseline exposure status, and proportions of participants with prevalent exposures at baseline or time zero. When such data are available in publications, quantitative bias analyses become feasible and substantially easier.Conclusion Researchers with access to individual-level data are encouraged to report the necessary information to enable robust bias assessment. This will support quantitative bias analysis and facilitate the inclusion of study findings in systematic reviews and hazard identification efforts.Acknowledgements This work was produced by the IARC Study Reporting Considerations Group: Lin Fritschi, Terry Boyle, Brigid M. Lynch, Scott Weichenthal, and Irina Guseva Canu.",
  "authors": [
    {
      "affiliations": [
        "Unisanté, University of Lausanne, Switzerland"
      ],
      "name": "Irina Guseva Canu"
    }
  ],
  "title": "8000014 Appraising bias with access to original data",
  "uid": "0c670fcd-0707-5172-860e-eb77eea28f3a"
}