1.4.SciPub_171.jpg

Statistical Methods in Cancer Research Volume V: Bias Assessment in Case–Control and Cohort Studies for Hazard Identification

IARC Scientific Publication No. 171

Edited by Berrington de González A, Richardson DB, Schubauer-Berigan MK

2024

ISBN-13

978-92-832-2228-6

Other languages

No other languages


Observational epidemiology is used to identify the causes of cancer and other chronic diseases, to determine the effectiveness of interventions, and to understand reasons for differences in disease rates over time or across locations. Reviewers of observational epidemiology studies must judge whether a causal interpretation is supported, including whether chance, bias, and confounding can be reasonably ruled out.

This Scientific Publication was supported by a scientific workshop on epidemiological bias assessment in hazard identification, convened by IARC and funded in part by the United States National Cancer Institute. It summarizes the wide range of practical methods that can be used by a reader or reviewer of a publication to assess the potential impact of confounding, information bias (including differential and non-differential exposure and outcome misclassification), or selection bias on the results of an epidemiological study. The methods presented include indirect approaches, for example negative control outcomes or exposures and proxies, and other approaches, such as sensitivity analyses.

The methods are presented in a way that is accessible to epidemiologists and other research workers who do not have extensive statistical training, as well as to statisticians who do not have extensive epidemiological training. The concepts are illustrated with numerous practical examples, taken from cancer epidemiology, that recur throughout the chapters, as well as links to online code or spreadsheets.

The book also outlines the process for integrating these bias assessments into the evidence synthesis. The process that is outlined uses the wide array of methods described in the book to consider all informative studies.

Finally, it is hoped that this book will encourage authors to apply a wider range of direct and indirect bias assessments in their primary research publications. Broader adoption of these analyses will enhance the quality of the original papers and further improve the interpretation of the evidence in subsequent reviews.

Worked spreadsheets for Examples (as listed in Annex 2)
Chapter 7: Example 7.14 and Example 7.16
 
R code
Chapter 4: Example 4.21 and Example 4.25
Chapter 6 and Annex 3: Example 6.9