When considering the health impact of foods, it is important to consider "compared to what?". Increasing the amount of a certain food or nutrient in the diet, typically implies a displacement of another.

While comparisons are more obvious in trials, in epidemiology food substitution models can be useful to help us determine the health effects of increasing/decreasing intake of a food, food group or nutrient.

However, these models are often misinterpreted and miscommunicated as if they are a game of "rock, paper, scissors", where one food beats another, and the losing food must be removed from the diet or considered harmful to health.

In this episode we discuss the problem of treating substitution analyses as food-ranking contests, rather than context-dependent comparisons shaped by the comparator, the unit of substitution, the baseline diet, and the outcome being studied.

• [01:30] Misuse of "compared to what?"
• [06:39] What substitution models do
• [10:43] Specified vs unspecified substitution
• [16:57] Why the units used matter
• [26:45] Example: organic vs conventional produce
• [31:22] When substitutions are useful
• [34:35] If legumes beat fish, does that mean fish intake should be zero?
• [44:31] Naive vs bias-adjusted: artificial sweeteners case study
• [49:14] Checklist: how to interpret food substitution analyses

Links:

• Go to episode page (all study references linked)
• Join the Sigma newsletter for free
• Subscribe to Sigma Nutrition Premium
• Subscribe to Alinea Nutrition Education Hub
• Enroll in the next cohort of our Applied Nutrition Literacy course
• Episode #472: Compared To What?
• Episode #589: Causal Inference in Nutrition Science – Daniel Ibsen, PhD