Energy Density of Unappetizing Foods Predicts Weight Gain
1. Increasing energy density of
unappetizing foods is
associated with lower weight
gain in adolescents
Jennifer R. Sadler
Department of Nutrition
University of North Carolina, Chapel Hill
2. How do food preferences
relate to weight change?
7. Does the relationship between energy
density and preference predict weight
change over time in adolescents?
Food
Preference
Weight
Change
8. Does the relationship between energy
density and preference predict weight
change over time in adolescents?
Food
Preference
Weight
Change
Hypothesis
High ED foods
Weight Gain
Low ED foods
14. IMAGE RATE
• 103 food images
• Varied energy density
• Representative food groups/types
• Mix of natural & empty
backgrounds
15. Sample
• 133 adolescents recruited in Oregon
• Baseline (age 15):
– BMI
– Image Rate Paradigm
• 3 years of follow-up anthropometrics
– 117 completed 1 or more years of follow-up
16. Analysis
• Per participant, selected food image
groups
– Appetizing: 32 highest rated images
– Unappetizing: 32 lowest rated images
• For each image group, calculated
average palatability rating and average
energy density
17. Analysis
• Calculated BMI change as ∆BMI/years
• Used linear regression to test if BMI
change (slope) was predicted by image
rate results
18. Analysis
• Calculated BMI change as ∆BMI/years at
year 4
• Used linear regression to test if BMI
change (slope) was predicted by image
rate results
Model 1
∆BMI = avg. rating of appetizing foods +
avg. rating of unappetizing foods
19. Analysis
• Calculated BMI change as ∆BMI/years at
year 4
• Used linear regression to test if BMI
change (slope) was predicted by image
rate results
Model 2
∆BMI = avg. ED of appetizing foods + avg.
ED of unappetizing foods
20. Analysis
• Controlled for self-reported hunger (pre
Image Rate paradigm), dietary restraint,
sex, and baseline BMI
21. Analysis
• Controlled for self-reported hunger (pre
Image Rate paradigm), dietary restraint,
sex, and baseline BMI
• Analysis completed in RStudio (v. 1.1.453).
22. Analysis
• Controlled for self-reported hunger (pre
Image Rate paradigm), dietary restraint,
sex, and baseline BMI
• Analysis completed in RStudio (v. 1.1.453).
• Model significance thresholded at
Bonferroni-corrected p < 0.025. Variable
significance thresholded at Bonferroni-
corrected p < 0.007.
23. Sample Characteristics (n=117)
Sex 55% Female
Baseline Age 15.0 ± 0.9 years
Baseline BMI 21.2 ± 2.3 kg/m2
Race/Ethnicity 76% Non-Hispanic White
Self-Reported Hunger (0-
100)
11.0 ± 4.0
Dietary Restraint 16.4 ± 6.3
Pubertal Development
Stage (0-4)
2.6 ± 0.4
∆BMI (per year) 0.5 ± 0.7
25. Significant difference in ratingsDensity
0
0.005
0.01
0.015
Average Rating
-100 -50 0 50 100
Appetizing
Unappetizing
26. No significant difference in energy
density
Appetizing
Unappetizing
1.0
0.75
0.5
0.25
0
1.0 1.5 2.0 2.5 3.0
Density
Average Energy Density (kcal/g)
31. Model 1: Ratings of appetizing and unappetizing
foods does not predict change in BMI
4
2
0
-2
∆BMI
-100 1000
average rating of foods
Appetizing
Unappetizing
32. Model 1: Ratings of appetizing and unappetizing
foods does not predict change in BMI
4
2
0
-2
∆BMI
-100 1000
average rating of foods
Ratings of food
images does not
significantly
predict BMI
change
(Model: p = 0.15)
Appetizing
Unappetizing
33. Model 2: Energy density of unappetizing foods
significantly predicts change in BMI
4
2
0
-2
∆BMI
1.5 2.0 2.5
average energy density of foods
3.01.0
Appetizing
Unappetizing
34. Model 2: Energy density of unappetizing foods
significantly predicts change in BMI
4
2
0
-2
∆BMI
1.5 2.0 2.5
average energy density of foods
3.01.0
Appetizing
Unappetizing
Overall model
significantly
predicts change
in BMI
p = 0.0036
35. Model 2: Energy density of unappetizing foods
significantly predicts change in BMI
4
2
0
-2
∆BMI
1.5 2.0 2.5
average energy density of foods
3.01.0
Appetizing
Unappetizing
Overall model
significantly
predicts change
in BMI
p = 0.0036
ED of appetizing
food images does
not significantly
predict BMI
change
(p = 0.07)
36. Model 2: Energy density of unappetizing foods
significantly predicts change in BMI
4
2
0
-2
∆BMI
1.5 2.0 2.5
average energy density of foods
3.01.0
Appetizing
Unappetizing
Overall model
significantly
predicts change
in BMI
p = 0.0036
ED of appetizing
food images does
not significantly
predict BMI
change
(p = 0.07)
ED of
unappetizing
food images
significantly
predicts BMI
change
(p = 0.0005)
37. Model 2: Energy density of unappetizing foods
significantly predicts change in BMI
4
2
0
-2
∆BMI
1.5 2.0 2.5
average energy density of unappetizing images
Slope = -0.8
Controlled for
hunger, dietary
restraint, sex
and baseline
BMI
38. Conclusions
• Energy density of food images rated as
unappetizing predicts three year weight
change in adolescents
39. Conclusions
• Energy density of food images rated as
unappetizing predicts three year weight
change in adolescents
• Effect is controlled for…
– Baseline BMI
– Hunger
– Dietary Restraint
– Sex
40. Conclusions
• Energy density of food images rated as
unappetizing predicts three year weight
change in adolescents
• Effect is controlled for…
– Baseline BMI
– Hunger
– Dietary Restraint
– Sex
• Ratings of ED foods did not predict change in
BMI
42. Conclusions
• Dislike of high ED foods protects against
weight gain
– Preference Food Choice
• Dislike as a target for intervention
43. Conclusions
• Dislike of high ED foods protects against
weight gain
– Preference Food Choice
• Dislike as a target for intervention1
1. (Legget et al, 2013)
44. Conclusions
• Dislike of high ED foods protects against
weight gain
– Preference Food Choice
• Dislike as a target for intervention
• Limitation: Preferences change over time
45. Conclusions
• Dislike of high ED foods protects against
weight gain
– Preference Food Choice
• Dislike as a target for intervention
• Limitation: Preferences change over time
• Valence of preference matters
47. References
• Drewnowski, A. (1998). Energy density, palatability, and satiety: implications for weight
control. Nutrition reviews, 56(12), 347-353.
• Potter, C., Griggs, R. L., Ferriday, D., Rogers, P. J., & Brunstrom, J. M. (2017). Individual variability
in preference for energy-dense foods fails to predict child BMI percentile. Physiology &
behavior, 176, 3-8.
• Nederkoorn, C., Houben, K., Hofmann, W., Roefs, A., & Jansen, A. (2010). Control yourself or just
eat what you like? Weight gain over a year is predicted by an interactive effect of response
inhibition and implicit preference for snack foods. Health Psychology, 29(4), 389.
• Dietz, W. H. (1994). Critical periods in childhood for the development of obesity. The American
journal of clinical nutrition, 59(5), 955-959.
• Van Strien, T., Frijters, J. E., Bergers, G. P., & Defares, P. B. (1986). The Dutch Eating Behavior
Questionnaire (DEBQ) for assessment of restrained, emotional, and external eating
behavior. International journal of eating disorders, 5(2), 295-315.
• Drewnowski, A. (1997). Taste preferences and food intake. Annual review of nutrition, 17(1), 237-
253.
• Legget, K. T., Cornier, M. A., Rojas, D. C., Lawful, B., & Tregellas, J. R. (2015). Harnessing the
power of disgust: a randomized trial to reduce high-calorie food appeal through implicit priming. The
American journal of clinical nutrition, 102(2), 249-255.
48. 1.5
0
-1
∆BMI
1.5 2.0 2.5
average energy density of unappetizing images
Male
Female
Significant interaction of sex on relationship between energy
density of unappetizing food images and BMI change
Editor's Notes
Thank you!
People inherent prefer highly energy dense foods, which theoretically contributes to repeated overconsumption and weight gain.
(Drewnowski, A. (1998). Energy density, palatability, and satiety: implications for weight control. Nutrition reviews, 56(12), 347-353.)
However, tests of this relationship have produced mixed results.
Potter et al found no relation between preference for high ED foods and child BMI-percentile in children 3-14 years old.
While Nederkoorm found that preference for snack foods interacted with inhibition to predict prospective weight gain in adults.
However, when teasing apart this relationship, its important to consider that preference has valence.
Typically, studies focus on foods that individuals like. But what about foods that they dislike?
This question is of upmost importance during adolescents, since adolescents gain autonomy in their food choices, allowing their preferences to more strongly impact their weight.
We hypothesized that preference for high
Our data came from participants in the Chocolate study, which was a study lead by Eric Stice at the Oregon Research Institute.
Our data came from participants in the Chocolate study, which was a study lead by Eric Stice at the Oregon Research Institute.
Participants were 15 years old at baseline. They completed a baseline assessment
Our data came from participants in the Chocolate study, which was a study lead by Eric Stice at the Oregon Research Institute.
BMI change was calculated by regressing BMI at each year of follow-up against time to calculate a BMI slope.
We used two linear regression models to test if Body mass variability is represented by distinct functional connectivity patterns
BMI change was calculated by regressing BMI at each year of follow-up against time to calculate a BMI slope.
We used two linear regression models to test if Body mass variability is represented by distinct functional connectivity patterns
Ran two models and each model had seven variables with controls
Ran two models and each model had seven variables with controls
Ran two models and each model had seven variables with controls
A little over half of the sample were women.
The sample was predominantly 15 years old and healthy weight at baseline. Most of sample did not report high dietary restraint, and most of the sample was undergoing puberty.
Over the three years of follow-up, on average participants gained half a BMI unit per year.
Star indicates significant difference between female and male participants
If we look at the images that most commonly were rated as “appetizing” some clear highly ED foods are found, like chocolates, cookies, and french fries.
BUT there also is a consistent preference for low ED fruits.
If we look at the images that most commonly were rated as “appetizing” some clear highly ED foods are found, like chocolates, cookies, and french fries.
BUT there also is a consistent preference for low ED fruits.
A similar trend is found in the images most commonly rated as unappetizing – there are high ED food items, like trail mix and chicken wings, and there are low ED items like grilled vegetable skewers and wild rice.
A similar trend is found in the images most commonly rated as unappetizing – there are high ED food items, like trail mix and chicken wings, and there are low ED items like grilled vegetable skewers and wild rice.
Test interactions
Test interactions
Test interactions
Looking at our two variables of interest, we found that the ED of appetizing food images did not significantly predict change in BMI.
However, the ED of unappetizing food images DID predict change in BMI. This relationship was highly significant.
Here I’ve plotted the relationship between energy density of unappetizing food images and change in BMI alone.
The slope of this line is negative .8, so with a 1 kcal per gram increase in the average energy density of unappetizing food images, we would expect a 0.8 decrease in BMI change per year. Over our three years of follow-up, this would protect against a 2.4 unit increase in BMI.
So in a sample of adolescents, followed for three years, we found that the energy density of food images rated as unappetizing predicted BMI, and the relationship was negative such that increasing the energy density of unappetizing food images was associated with lower weight gain over three years.
This effect is seen controlling for baseline BMI, self-reported hunger before rating the images, dietary restraint, and biological sex.
We found that ratings of these images did not predict change in BMI either.
This may be acting through food choice, but the study did not directly assess food choice, so we can’t be sure.