*Geek Box: Comparison Levels in Nutrition Research
One issue for nutrition interventions that differ from drug interventions, is that nutrient intake tends to have a narrow range of variability. Think about it; it would be uncommon to go from eating 5% carbohydrate one day to 55% the next. If we look at dietary fat, for example, individuals tend to have relatively consistent levels of intake that may fluctuate over time only by a few percentage points. In contrast, in a drug trial, an intervention is able to compare the effects of an exposure [the drug] to zero exposure, i.e., a true placebo; when you compare something to nothing, there will generally always be some degree of effect size.
However, in nutrition, even the ‘control’ group will have at least adequate levels of a nutrient, because it is considered unethical to expose people in a trial to a deficiency. This often means that there isn’t that much a difference in levels of intake between two comparison groups, and it can result in ’null’ findings, or ‘weak’ effect sizes. This can be complicated if, for example, the trial will examine a nutrient or food as the exposure, but participants habitually consume that food or nutrient before the trial. One way to try and address this is to get participants to exclude certain foods or macronutrients [as opposed to micronutrients, which may result in nutrient deficiencies], in the run-in to the trial. The second, and more important, way to address this is to create a sufficient contrast in the exposure. For example, in this study, the MUFA-diet had 22% from MUFA, while the control had 8%. This may then create a big enough difference to detect an effect of the intervention, compared to the control.