What is it about?

Confounding is a central problem in epidemiology. It influences measures of association; that is, measures that contrast occurrence of the outcome (most often disease) between groups of people or populations with different conditions, for example, being male or female, smokers or nonsmokers, or children born by different modes of delivery. This chapter gives an introduction into the problem of confounding, into methods to identify it, and into methods to control for it. Confounding is a problem of all fields within epidemiology, but there are specific examples in pediatric epidemiology that are discussed in this chapter.

Featured Image

Why is it important?

Confounding is a threat to valid inference from observational studies and thus has always to be considered and potentially to be accounted for. Confounding in pediatric epidemiology is not different from confounding in other fields of epidemiology. However, there are ample examples related to central hypotheses in pediatric epidemiology - like Barker’s hypothesis of fetal origins of disease - and to often encountered traits or perinatal indicators like birth weight, gestational age, or reproductive factors like spontaneous abortion.


This chapter is part of an interesting book on Pediatric Epidemiology edited by Wieland Kiess, Carl-Gustav Bornehag, and Chris Gennings (ISBN: 978-3-318-06122-2; DOI:10.1159/isbn.978-3-318-06123-9). It was fun to contribute to this compilation of important aspects. Have a look at the other chapters too, they may impact on how you conduct your next study and/or on how you interpret a study of interest.

PD Dr. med. Jon Genuneit
Universitat Leipzig

Read the Original

This page is a summary of: How to Deal with Confounding, November 2017, Karger Publishers, DOI: 10.1159/000481329.
You can read the full text:




The following have contributed to this page