Control of Confounding and Reporting of Results in Causal Inference Studies. Guidance for Authors from Editors of Respiratory, Sleep, and Critical Care Journals

• Published in: Annals of the American Thoracic Society Vol. 16; no. 1; pp. 22 - 28
• Main Authors: Lederer, David J., Bell, Scott C., Branson, Richard D., Chalmers, James D., Marshall, Rachel, Maslove, David M., Ost, David E., Punjabi, Naresh M., Schatz, Michael, Smyth, Alan R., Stewart, Paul W., Suissa, Samy, Adjei, Alex A., Akdis, Cezmi A., Azoulay, Élie, Bakker, Jan, Ballas, Zuhair K., Bardin, Philip G., Barreiro, Esther, Bellomo, Rinaldo, Bernstein, Jonathan A., Brusasco, Vito, Buchman, Timothy G., Chokroverty, Sudhansu, Collop, Nancy A., Crapo, James D., Fitzgerald, Dominic A., Hale, Lauren, Hart, Nicholas, Herth, Felix J., Iwashyna, Theodore J., Jenkins, Gisli, Kolb, Martin, Marks, Guy B., Mazzone, Peter, Moorman, J. Randall, Murphy, Thomas M., Noah, Terry L., Reynolds, Paul, Riemann, Dieter, Russell, Richard E., Sheikh, Aziz, Sotgiu, Giovanni, Swenson, Erik R., Szczesniak, Rhonda, Szymusiak, Ronald, Teboul, Jean-Louis, Vincent, Jean-Louis
• Format: Journal Article
• Language: English
• Published: United States American Thoracic Society 01.01.2019
• Subjects: Algorithms; Bias; Causality; Confounding Factors, Epidemiologic; Critical Care; Design; Epidemiology; Guidelines as Topic; Humans; Hypotheses; Hypothesis testing; Life Sciences; Lung cancer; Models, Statistical; Periodicals as Topic; Pulmonary Medicine; Research Design - standards; Research methodology; Sleep; Sleep Medicine Specialty; Variables; New York; Greenland; confounding factors; epidemiology; research design; causality
• ISSN: 2329-6933; 2325-6621; 2325-6621
• DOI: 10.1513/AnnalsATS.201808-564PS

[...]we urge authors to consider using causal models when testing causal associations. [...]all other methods of controlling for confounding involve implicit assumptions about causal effects, which are not transparent to the reader. The observed associations between these covariates and the outcome have not been subject to the same approach to control of confounding as the exposure. [...]residual confounding and other biases often heavily influence these associations. [...]we recommend that P values only rarely be presented in isolation (exceptions may include "omics" studies and tests for interaction).