#589: Causal Inference in Nutrition Science – Daniel Ibsen, PhD

Introduction to this Episode

This episode explores how asking better questions and using stronger methods can resolve much of the confusion in nutrition science. Dr. Daniel Ibsen discusses why nutrition research often produces conflicting results and how careful methodological thinking can clarify true diet-disease relationships.

Nutrition science has unique challenges – diets are complex, people self-report their food intake imperfectly, and we canʼt easily run long-term diet experiments on people. Dr. Ibsen explains how embracing concepts like food substitution analysis, the “target trial” framework, and objective dietary assessment can strengthen evidence.

The episode centers on methodological insights that make nutrition research more reliable and actionable. Key themes include defining dietary comparisons explicitly (the “compared to what?” question), considering peopleʼs starting diets, and using causal inference techniques to design better studies.

About the Guest

Daniel B. Ibsen is an epidemiologist and nutritional scientist whose work bridges rigorous causal inference methods with real-world diet and cardiometabolic disease research. He is an Associate Professor at Aarhus University, Denmark.

Ibsen’s research spans from objective dietary biomarkers and plant-based diet patterns to how we interpret substitution models and causal claims in nutrition epidemiology.

Ibsen is working in an area of nutrition epidemiology that is undergoing methodological “rethinking”: moving from standard regression-adjustment models toward more formal causal inference frameworks (e.g., substitution models, compositional data, mediation, biomarkers). Dr. Ibsen is part of the movement that says we must be explicit about what effect weʼre estimating (e.g., what intervention or substitution) and the assumptions required for causal interpretation.

Useful Terminology for this Episode

• Diet as “compositional” – The idea that increasing one food in the diet usually means eating less of another. Nutrients and foods are consumed in proportions that must total a whole diet, so any change is relative to changes in other components.
• Isocaloric substitution – A comparison where one food is increased while an equivalent amount of calories from another food is decreased, keeping total energy intake constant.
• Causal inference – Methods and principles for determining cause-and-effect relationships from data. In nutrition, causal inference involves asking “what would happen to health if we change diet X to diet Y?” and trying to isolate that effect. It requires assumptions (like controlling confounders so that groups differ only by the diet change) and often uses frameworks that mimic a randomized trial in observational data.
• Consistency (assumption) – In causal inference, the assumption that a treatment or exposure is well-defined and uniform, so it has the same effect regardless of how or where itʼs implemented. If an exposure actually encompasses multiple different interventions (for example, very different kinds of “low-carb diets”), the consistency assumption is violated – it becomes unclear what the “true” causal effect of that broadly defined exposure is.
• Ultra-processed foods (UPFs) – Foods that are industrially formulated with many ingredients and heavy processing (NOVA Group 4).
• Food Frequency Questionnaire (FFQ) – A common dietary assessment tool in research where participants report how often they consume specific foods or beverages over a long period.
• Baseline diet – What people are eating before a change or intervention.