Article Text
Statistics from Altmetric.com
To the Editor We appreciate the editorial by Burden and Timpson1 highlighting collider bias as a potential explanation for our findings of a differential association between diabetes and atrial fibrillation (AF) by ethnicity as well as by heart failure (HF) type.2 Collider bias applies in aetiologic research where causal associations between two independent variables are spuriously created due to stratification by a causally related health outcome. First we point out that ours was not aetiologic research, and we specifically acknowledged that our observational study design does not allow ascertainment of causality. We nonetheless agree that the diabetes–AF paradox was one of our most intriguing observations, raising questions that, while not answered by …
Footnotes
Contributors All authors have contributed to the planning and writing of the response letter.
Competing interests None declared.
Provenance and peer review Commissioned; internally peer reviewed.
Patient consent for publication Not required.