Original Article Obesity

CLINICAL TRIALS AND INVESTIGATIONS

Bariatric Surgery Does Not Affect Food Preferences,

but Individual Changes in Food Preferences May Predict
Weight Loss
1,2, Simone Rasmussen1, Bodil Just Christensen3, Christian Ritz1, Carel W. le Roux 4,5,
Mette Søndergaard Nielsen
1 and Anders Sjödin1
Julie Berg Schmidt

Objective: Using an ad libitum buffet meal targeting direct behavior, the authors of the current study previ-
ously reported no effect of Roux-en-Y gastric bypass (RYGB) and sleeve gastrectomy (SG) surgery on food
preferences 6 months after surgery. The current study investigated changes in food preferences at 18
months after surgery and whether changes in food preferences at 6 months predicted weight loss.
Methods: Twenty food items separated into the following food categories were served at the buffet meal:
high-fat, low-fat, sweet, savory, high-fat savory, high-fat sweet, low-fat savory, and low-fat sweet. Energy
intake and intake from each of the food items were registered. Energy intake prior to the meal was
standardized.
Results: Thirty-nine subjects completed visits before surgery and 18 months following RYGB (n = 29) and
SG (n = 10) surgery. Energy intake decreased 41% (4,470 ± 209 kJ vs. 2,618 ± 209 kJ, P < 0.001), but no change
occurred in relative energy intake from any of the food categories (all P ≥ 0.23), energy density (P = 0.20), or
macronutrient intake (all P ≥ 0.28). However, changes in high-fat food intake, protein intake, energy intake,
and energy density at 6 months predicted weight loss at 18 months (P ≤ 0.02).
Conclusions: RYGB surgery and SG surgery do not affect food preferences. However, changes in food
preferences seem to be predictive of weight loss.
Obesity (2018) 0, 00-00. doi:10.1002/oby.22272

Introduction beyond an effect of dietary counseling or increased motivation for

Roux-en-Y gastric bypass (RYGB) and sleeve gastrectomy (SG) sur-
gery have been associated with changes in food preferences toward
less energy-dense foods (1-12). Studies generally report decreased
consumption of sugary and fatty foods (1-3,5,7,10-12) and increased
consumption of fruit and vegetables and protein-rich foods (3,5,7,12),
although conflicting results exists (13). Animal studies seems to
replicate results from human studies (4,14-18), suggesting that the
mechanisms driving the potential changes in food preferences are
weight loss (19).
The evidence from human studies relies on verbal reports, such as
food records, questionnaires, and interviews, which are prone to
reporting bias and underestimation of intake of unhealthy food (20).
To address this limitation, we previously targeted direct behavior
by investigating postoperative changes in food preferences using an
ad libitum buffet meal (13). Six months following RYGB and SG
surgery, total energy intake decreased 54%; however, we found no

1
Department of Nutrition, Exercise and Sports, Faculty of Science (Obesity Research),  University of Copenhagen, Copenhagen, Denmark.
Correspondence: Mette Søndergaard Nielsen (msn@nexs.ku.dk) 2 The Danish Diabetes Academy,  Odense University Hospital, Odense,
Denmark 3 Department of Food and Resource Economics, Faculty of Science, University of Copenhagen, Copenhagen, Denmark 4 Investigative
Science, Imperial College London, London, UK 5 Diabetes Complications Research Centre, Conway Institute, University College Dublin, Dublin,
Ireland.

Funding agencies: This study was carried out as part of the research program “Governing Obesity” funded by the University of Copenhagen Excellence Programme for
Interdisciplinary Research (www.go.ku.dk). Furthermore, the Danish Diabetes Academy supported by the Novo Nordisk Foundation, the Lundbeck Foundation, and the
Aase and Ejnar Danielsens Foundation funded this study.
Disclosure: CWlR was supported by grants from Science Foundation Ireland (ref 12/YI/B2480), the Health Research Board (USIRL-2016-2), and the Irish Research Council
during the conduct of the study. Furthermore, CWlR reported being on the advisory boards for Novo Nordisk and GI Dynamics, receiving a research grant from AnaBio,
and receiving honoraria for lectures and advisory work from Eli Lily, Johnson and Johnson, Sanofi Aventis, Astra Zeneca, Janssen, Bristol-Myers Squibb, and Boehringer-
Ingelheim. The other authors declare no conflict of interest.
Clinical trial registration: ClinicalTrials.gov identifier NCT02070081.
Additional Supporting Information may be found in the online version of this article.
Received: 3 May 2018; Accepted: 22 June 2018. Published online 0 Month 2018. doi:10.1002/oby.22272

www.obesityjournal.org  Obesity | VOLUME 00 | NUMBER 00 | MONTH 2018    1

Obesity Food Preferences Assessed by Buffet Meal Test  Søndergaard Nielsen et al.
changes in food preferences, macronutrient intake, or energy density
of the selected meal (13).
The prevailing view is that changes in food preferences following bar-
iatric surgery are due to a learned adjustment to postingestive responses
following intake of high-fat and sweet foods, commonly termed “con-
ditioned taste avoidance” (21). At 6 months, patients might still be in a
“learning period,” in which they adjust food intake to avoid postinges-
tive responses (22). Thus, we speculate whether 6 months postsurgery
was too early for behavioral changes to set in.
The aim of this study was to investigate food preferences using an
ad libitum buffet meal targeting direct behavior 18 months following
RYGB and SG surgery. We hypothesized that energy intake at 18
months after surgery was increased compared with 6 months after
surgery, thereby reaching a threshold at which enough postinges-
tive responses could be generated, resulting in a relative reduction
in intake of high-fat and sweet foods and a corresponding rela-
tive increase in intake of low-fat and savory foods. Furthermore,
because of the large variation we observed in food preferences at
6 months, we investigated whether changes in food preferences for
high-fat and sweet foods, macronutrient intake, energy density, and
total energy intake at 6 months predicted weight loss outcomes 18
months after surgery.
Methods
The study population, design, and methods have been described
previously (13). Briefly, food preferences were assessed approxi-
mately 3 months before surgery (prior to a mandatory 8% presur-
gical weight loss), as well as 6 months and 18 months following
RYGB and SG surgery. Data from the visits before surgery and 6
months after surgery have been reported previously (13). At each
visit, anthropometric data were collected in the fasted state at 9:00
am. At 3:30 pm, a picture display test was carried out, and at 4:30
pm, an ad libitum buffet meal was served. Energy intake prior to
the test meal was standardized (a 954 kJ meal and a 1,675 kJ liquid
meal were served at 10:00 am and 1:15 pm, respectively; Cambridge
Weight Plan). The liquid meal came in five different flavors: choc-
olate, chocolate mint, strawberry, wild berry, and caramel and
walnut.
The study was approved by the Scientific Ethic Committees of
the Capital Region of Denmark (Journal number H-3-2013-138)
and registered in the database www.ClinicalTrials.gov (identi-
fier NCT02070081). All study participants gave written informed
consent.
Outcomes
Buffet meal test.  The following 20 food items were served at the
buffet meal: pork rib roast, chicken fillet strips, fish cakes, nuggets,
Danish omelet, French fries, creamy potato gratin, bread (rye bread,
baguette, and crispbread), ketchup, remoulade, mayonnaise, skyr
with berries, cut raw vegetables, cut fruit, vanilla ice cream, choc-
olate sauce, cocoa meringues, biscuit cones with chocolate, sweet
licorice, and Danish pastries. The only liquid served at the meal
was water. The participants were instructed to eat according to their
2    Obesity | VOLUME 00 | NUMBER 00 | MONTH 2018
preferences and for as long as they wanted. They ate unmonitored
and unaccompanied, in order to diminish social desirability bias. The
staging and placement on the table of the 20 food items were identi-
cal at each visit.
The 20 food items varied along two dimensions associated with food
choices, fat (high or low) and taste (sweet or savory) (23), thereby
making it possible to organize the food items into separate catego-
ries (high-fat and low-fat or sweet and savory) or combined food cat-
egories (high-fat savory, low-fat savory, high-fat sweet, and low-fat
sweet) (13). Total energy intake, intake from each of the food cate-
gories, and the duration of the meal were registered. The participants
were unaware of this registration. In order to divert the attention of
the participants away from making food choices to please the investi-
gators, a blood sample was drawn after the meal, and the participants
were informed that the aim was to investigate how “eating as much as
they wished of the food they liked best” affected gut hormone levels
in the blood.
Picture display test.  Participants were presented with standardized
pictures of the following 20 food items: pork rib roast, turkey strips,
nuggets, Danish omelet, French fries, crispbread, salty crackers, skyr
with berries, carrots, cut fruit, cheese, smoked fillet (cold cuts), vanilla
ice cream, cocoa meringues, sweet licorice, Danish pastries, milk choc-
olate, pound cake, cookies, and wine gum. Pictures were displayed in
a randomized order, and the participants were instructed to choose the
food item they would most like to eat.
Previous experience of unpleasant postingestive responses. Previous
experience of unpleasant postingestive responses was assessed by a sem-
istructured qualitative interview 6 weeks, 6 months and 18 months after
surgery. Data from the interviews were quantified into a 4-point score
(0-3), with 0 indicating no experience of discomfort related to eating after
surgery, while 1, 2 and 3 indicate low, medium and high levels of discom-
fort, respectively.
Premeal hunger.  Premeal hunger was assessed by a visual analog
scale (100 mm) approximately 15 minutes prior to the buffet meal.
Anthropometric measures.  Body weight and height were measured
after an overnight fast. Percent total weight loss (%TWL) was defined
as the presurgical weight minus the weight at 18 months after surgery,
divided by the presurgical weight, and multiplied by 100.
Statistical analysis
Descriptive data summaries are presented as mean ± SD. Premeal hun-
ger, total energy intake, energy density, eating time, eating rate, and
weight changes were analyzed using linear mixed models with visit as
fixed effect. Intake from the buffet meal was analyzed using a two-stage
approach to accommodate for semicontinuous data (consisting of zero
intake and positive intake; (24)): First, a logistic mixed-effects model
was fitted to the binary data indicating if intake was 0 or not. Second, a
linear mixed model was fitted to the square root–transformed positive
intake values only. Both models included the food category-visit in-
teraction as fixed effect. Estimates from these two analyses were then
combined and back-transformed to obtain unconditional estimates of
mean intakes per category and visit and differences in mean intakes
www.obesityjournal.org
Original Article Obesity
CLINICAL TRIALS AND INVESTIGATIONS

R Foundation, Vienna, Austria) and RStudio version 1.1.383 (www.

TABLE 1 Characteristics of the subjects (n = 39)
Male/female 6/33
Age (y) 40.3 ± 9.2
Body weight (kg) 130.5 ± 22.0
BMI (kg/m2) 45.0 ± 6.8
RYGB/SG surgery 29/10
Patients with type 2 diabetes 10
Time from baseline to surgery (d) 92 ± 37
Data are shown as mean ± SD or frequency.
SG, sleeve gastrectomy; RYGB: Roux-en-Y gastric bypass.
between combinations of category and visit. Differences between type
of surgery and between the groups with highest and lowest energy
intake were analyzed using linear mixed models with a visit-group
interaction or a food category-visit-group interaction (intake from the
buffet meal) as fixed effect. Binary outcomes from the picture display
test were analyzed using logistic mixed-effects models (one model per
category) with visit as fixed effect. All models were adjusted for age
and gender and included subject-specific random effects. Associations
between changes in relative intake at the buffet meal at 18 months and
previous experience of postingestive responses were evaluated using
simple linear regression adjusted for age and gender. The same model
was used to evaluate associations between changes in relative intake
at the buffet meal at 6 months and 18 months and weight loss out-
come at 18 months after surgery. Results are reported as mean ± SE
(intake as mean percentages). P values < 0.05 were considered signif-
icant. Statistical analyses were carried out using R (version 3.3.2; The
rstudio.com).
Results
Participants
Sixty-one subjects were originally recruited and 41 completed the visit
at 6 months post surgery. Results were reported previously (13). Prior
to the 18 months follow-up visit at, two subjects were excluded due to
pregnancy, leaving a total sample of 39 subjects with complete data
from all three visits. Twenty-nine subjects underwent RYGB surgery,
and ten subjects underwent SG surgery. Baseline characteristics are
summarized in Table 1.
Weight loss
Eighteen months after surgery, %TWL was 31% (40.9 ± 2.2 kg,
P < 0.001) (Supporting Information Figure S1). Weight loss was larger
in RYGB compared with SG patients (44.8 ± 3.4 kg vs. 30.0 ± 4.4 kg,
P = 0.01) and in subjects without diabetes compared with subjects with
diabetes (45.7 ± 3.1 kg vs. 26.4 ± 5.8 kg, P = 0.01).
Buffet meal test and premeal hunger
From before surgery to 18 months after surgery, total energy intake
at the buffet meal decreased 41% (4,470 ± 209 kJ vs. 2,618 ± 209 kJ,
P < 0.001) and eating rate decreased 18% (242 ± 13 kJ/min vs. 198 ± 13
kJ/min, P = 0.01). No changes in hunger or energy density of the meal
were observed (Table 2).

TABLE 2 Intake, eating time, and eating rate at buffet meal (n = 39)

Postsurgery, Postsurgery, P value, P value,

Presurgery 6 mo 18 mo 6 mo 18 mo
Premeal hunger (cm) 5.7 ± 0.5 5.0 ± 0.5 5.5 ± 0.5 0.20 0.68
Total energy intake (kJ) 4,470 ± 209 2,096 ± 209 2,618 ± 209 < 0.001 < 0.001
Energy density (kJ/g) 7.2 ± 0.3 7.3 ± 0.3 7.5 ± 0.3 0.72 0.20
Eating time (min) 19.5 ± 1.1 13.5 ± 1.1 14.9 ± 1.1 < 0.001 < 0.001
Eating rate (kJ/min) 279 ± 16 204 ± 16 236 ± 16 < 0.001 0.01
Intake from combined categories
High-fat savory (%) 48 ± 2 48 ± 4 51 ± 3 0.95 0.53
High-fat sweet (%) 7 ± 1 7 ± 1 7 ± 1 0.91 0.73
Low-fat savory (%) 40 ± 2 37 ± 3 36 ± 3 0.50 0.31
Low-fat sweet (%) 5 ± 1 8 ± 2 6 ± 1 0.16 0.62
Intake from separate categories
High-fat (%) 55 ± 3 56 ± 4 61 ± 4 0.88 0.23
Low-fat (%) 45 ± 3 44 ± 4 39 ± 4
Sweet (%) 20 ± 2 21 ± 3 19 ± 3 0.71 0.89
Savory (%) 80 ± 2 79 ± 3 81 ± 3
Intake macronutrients
E% carbohydrate 35 ± 1 34 ± 2 36 ± 2 0.59 0.75
E% fat 44 ± 2 45 ± 2 46 ± 2 0.70 0.59
E% protein 20 ± 1 21 ± 2 18 ± 2 0.88 0.28
Data are shown as mean ± SE. Changes compared with presurgery were found using linear mixed models adjusted for age and gender and with subject as random effect.
E%, energy percentage.

www.obesityjournal.org  Obesity | VOLUME 00 | NUMBER 00 | MONTH 2018    3

 Obesity

TABLE 3 Intake at buffet meal stratified by type of surgery

RYGB patients (n = 29) SG patients (n = 10) Overall
P value,
Postsurgery, Postsurgery, P value, P value, Postsurgery, Postsurgery, P value, P value, between
Presurgery 6 mo 18 mo 6 mo 18 mo Presurgery 6 mo 18 mo 6 moa 18 moa groups
Premeal hunger 5.6 ± 0.4 5.5 ± 0.4 5.8 ± 0.4 0.98 0.60 6.5 ± 1.6 4.0 ± 1.6 5.0 ± 1.6 < 0.001 < 0.01 0.12
(cm)

4    Obesity | VOLUME 00 | NUMBER 00 | MONTH 2018

Total energy intake 4,718 ± 243 2,252 ± 243 2,782 ± 244 < 0.001 < 0.001 3,754 ± 400 1,643 ± 399 2,140 ± 401 < 0.001 < 0.001 0.16
(kJ)
Energy density 7.6 ± 0.4 7.6 ± 0.4 8.1 ± 0.4 1.00 0.06 6.4 ± 1.3 6.8 ± 1.3 6.5 ± 1.2 0.56 0.97 0.52
(kJ/g)
Eating time (min) 18.9 ± 1.4 14.3 ± 1.4 16.2 ± 1.4 < 0.001 < 0.001 19.2 ± 2.7 9.4 ± 2.7 9.2 ± 2.6 < 0.001 < 0.001 < 0.01
Eating rate (kJ/min) 299 ± 16 203 ± 16 238 ± 16 < 0.001 < 0.001 212 ± 34 193 ± 33 224 ± 32 0.18 0.52 0.048
Intake combined categories
High-fat savory (%) 50 ± 3 51 ± 4 52 ± 4 0.79 0.71 43 ± 5 37 ± 7 47 ± 6 0.45 0.59 0.44
High-fat sweet (%) 6 ± 1 6 ± 2 5 ± 1 0.91 0.64 8 ± 2 10 ± 4 14 ± 4 0.78 0.20
Low-fat savory (%) 40 ± 3 34 ± 4 36 ± 4 0.25 0.40 42 ± 5 49 ± 7 36 ± 6 0.41 0.46
Low-fat sweet (%) 5 ± 1 9 ± 2 7 ± 2 0.83 0.20 6 ± 2 4 ± 3 2 ± 2 0.55 0.09
Intake separate categories
High-fat (%) 56 ± 3 59 ± 5 63 ± 4 0.66 0.19 52 ± 6 52 ± 9 64 ± 7 0.99 0.21 0.54
Low-fat (%) 44 ± 3 41 ± 5 37 ± 4 48 ± 6 48 ± 9 36 ± 7
Sweet (%) 19 ± 3 21 ± 4 21 ± 3 0.57 0.63 22 ± 4 19 ± 6 16 ± 4 0.69 0.26 0.86
Savory (%) 81 ± 3 79 ± 4 79 ± 3 78 ± 4 81 ± 6 84 ± 4
Intake macronutrients
E% carbohydrate 34 ± 2 34 ± 2 36 ± 2 0.95 0.36 40 ± 3 34 ± 5 35 ± 4 0.26 0.37 0.20
E% fat 46 ± 2 45 ± 3 45 ± 2 0.90 0.87 40 ± 3 46 ± 5 47 ± 5 0.35 0.20
E% protein 20 ± 1 21 ± 2 18 ± 2 0.94 0.36 20 ± 3 21 ± 4 17 ± 4 0.85 0.59
Data are shown as mean ± SE.
a
Changes compared with presurgery were found using linear mixed models adjusted for age and gender and with subject as random effect.
E%, energy percentage; RYGB, Roux-en-Y gastric bypass; SG, sleeve gastrectomy.

www.obesityjournal.org
Food Preferences Assessed by Buffet Meal Test  Søndergaard Nielsen et al.
 TABLE 4 Body weight and intake, eating time, and eating rate at buffet meal in subjects with high energy intake and low energy intake at buffet meal 18 mo after
surgery
Low energy intake group (n = 13) High energy intake group (n = 13)

Postsurgery, Postsurgery, Overall P value,

Presurgery 18 mo P value Presurgery 18 mo P value between groupsa
Original Article

Female/male 12/1 - - 9/4 - - -

www.obesityjournal.org 
RYGB/SG surgery 7/6 - - 11/2 - - -
Body weight (kg) 124.0 ± 4.1 83.8 ± 4.1 < 0.001 135.9 ± 5.6 96.1 ± 5.6 < 0.001  > 0.99
Premeal hunger (cm) 5.5 ± 0.8 4.6 ± 0.8 0.43 5.6 ± 0.8 6.0 ± 0.8 0.62 0.32
Total energy intake (kJ) 3,463 ± 292 1,295 ± 292 < 0.001 5,513 ± 407 3,931 ± 407 < 0.01 0.37
CLINICAL TRIALS AND INVESTIGATIONS

Energy intake (kJ/kg body 31.5 ± 4.1 19.2 ± 4.0 < 0.001 42.1 ± 4.0 43.6 ± 4.0 0.79 0.02
weight)
Energy density (kJ/g) 6.6 ± 0.4 6.0 ± 0.4 0.16 7.6 ± 0.5 9.0 ± 0.5 < 0.01 < 0.01
Eating time (min) 19.9 ± 1.8 9.7 ± 1.8 < 0.001 19.7 ± 2.0 19.2 ± 2.0 0.79 < 0.01
Eating rate (kJ/min) 186 ± 17 150 ± 17 0.06 299 ± 33 238 ± 33 0.18 0.61
Intake from combined categories
High-fat savory (%) 41 ± 4 27 ± 6 0.048 48 ± 4 57 ± 5 0.20 < 0.01
High-fat sweet (%) 4 ± 1 2 ± 1 0.20 9 ± 2 16 ± 3 0.05
Low-fat savory (%) 48 ± 4 65 ± 6 0.02 40 ± 4 22 ± 4 < 0.001
Low-fat sweet (%) 7 ± 2 6 ± 3 0.75 3 ± 1 6 ± 2 0.21
Intake from separate categories
High-fat (%) 46 ± 5 33 ± 8 0.16 59 ± 5 73 ± 5 0.048 0.03
Low-fat (%) 54 ± 5 67 ± 8 41 ± 5 27 ± 5
Sweet (%) 17 ± 3 13 ± 4 0.40 15 ± 3 24 ± 4 0.10 0.39
Savory (%) 83 ± 3 87 ± 4 85 ± 3 76 ± 4
Intake macronutrients
E% carbohydrate 37 ± 3 42 ± 5 0.39 35 ± 3 36 ± 3 0.81 < 0.001
E% fat 42 ± 3 38 ± 5 0.52 45 ± 3 48 ± 3 0.47
E% protein 22 ± 3 20 ± 4 0.78 20 ± 2 16 ± 2 0.21
Data are shown as mean ± SE.
aChanges compared with presurgery were compared using linear mixed models adjusted for age and gender and with subject as random effect.

E%, energy percentage; RYGB, Roux-en-Y gastric bypass; SG, sleeve gastrectomy.

Obesity | VOLUME 00 | NUMBER 00 | MONTH 2018    5

Obesity
Obesity Food Preferences Assessed by Buffet Meal Test  Søndergaard Nielsen et al.

TABLE 5 Mean frequency of food choices at picture display test from each combined food category

Postsurgery, Postsurgery, P value, P value,

Presurgery 6 mo 18 mo 6 mo 18 mo
High-fat savory (%) 31 ± 10 21 ± 9 31 ± 10 0.39 1.00
High-fat sweet (%) 7 ± 4 0 ± 0 0 ± 0 1.00 1.00
Low-fat savory (%) 36 ± 8 63 ± 9 51 ± 9 0.04 0.23
Low-fat sweet (%) 12 ± 7 6 ± 5 8 ± 5 0.30 0.50
Data are shown as mean ± SE. Changes compared with presurgery were found using linear mixed models adjusted for age and gender and with subject as random effect.

Figure 1 Associations between percent total weight loss (%TWL) at 18 mo after surgery and changes in (A) energy intake from before surgery to 6 mo after surgery, (B)
energy density from before surgery to 6 mo after surgery, (C) relative intake of protein from before surgery to 6 mo after surgery, (D) relative intake of high-fat food from
before surgery to 6 mo after surgery, and (E) relative intake of high-fat food from before surgery to 18 mo after surgery. Associations were evaluated using simple linear
regressions adjusted for age and gender.

6    Obesity | VOLUME 00 | NUMBER 00 | MONTH 2018 www.obesityjournal.org

Original Article
CLINICAL TRIALS AND INVESTIGATIONS
No postoperative changes were observed in relative energy intake
from the combined food categories (high-fat savory, high-fat sweet,
low-fat savory, and low-fat sweet), the separate food categories (high-
fat versus low-fat and sweet versus savory), the different macronutri-
ents (Table 2), or from any of the 20 food items served at the buffet
(Supporting Information Table S1).
Experiences of unpleasant postingestive responses assessed at 6
weeks, 6 months, and 18 months after surgery were not associated
with changes in relative intake of high-fat or sweet foods at the buffet
meal (all P ≥ 0.07). All of the 39 subjects (100%) reported having one
or more experiences with unpleasant physiological responses to food
after surgery.
Similar decreases in energy intake were seen in RYGB and SG patients
(41% vs. 43%, respectively; RYGB: 4,718 ± 243 kJ vs. 2,782 ± 244
kJ, P < 0.001 and SG: 3,754 ± 400 kJ vs. 2,140 ± 401 kJ, P < 0.001,
P = 0.16). Furthermore, the stratified analysis showed no changes in
relative energy intake from the combined and separate food categories
and macronutrients in RYGB (all P ≥ 0.19) or SG patients (all P ≥ 0.09),
and there was no differences between surgical groups (all P ≥ 0.20)
(Table 3). However, we found a trend for a reduced relative intake
of low-fat sweet foods in SG patients (6 ± 2% vs. 2 ± 2%, P = 0.09).
Furthermore, relative energy intake from the combined and separate
food categories did not differ between patients with or without diabetes
(all P ≥ 0.44).
In a post hoc analysis separating data into the highest (n = 13) and
lowest (n = 13) tertile of total energy consumed at the buffet meal
18 months after surgery, we found a difference in relative intake
from the combined categories (P < 0.01), the separate categories
(P = 0.03), and macronutrients (P < 0.001) across tertiles. In the sub-
group of patients with the lowest energy intake at 18 months, the
relative intake from the high-fat savory category decreased, whereas
relative intake from the low-fat savory category increased (P = 0.048
and P = 0.02, respectively). In contrast, in the subgroup with the
highest energy intake at 18 months, the relative intake from the low-
fat savory category decreased after surgery (P < 0.001). Furthermore,
pre- to postoperative changes in eating time, energy density of the
meal, and total energy consumed per kilogram body weight dif-
fered between groups (P < 0.01, P < 0.01, and P = 0.02, respectively)
(Table 4).
Picture display test
The subjects’ choice of picture from the combined groups, high-
fat savory, high-fat sweet, low-fat savory and low-fat sweet, did not
change from before surgery to 18 months after surgery (all P ≥ 0.23)
(Table 5).
Prediction of weight loss outcomes
Changes between presurgery and 6 months postsurgery for total
energy intake (P = 0.02, R 2  = 0.24), energy density (P < 0.001,
R 2 = 0.39), protein intake (P = 0.01, R 2 = 0.26), and intake of high-fat
foods (P = 0.01, R 2 = 0.26) predicted %TWL at 18 months (Figure 1).
Furthermore, changes between presurgery and 18 months postsur-
gery for intake of high-fat foods (P = 0.01, R 2 = 0.26) were associated
with %TWL at 18 months (Figure 1). A tendency was also found
www.obesityjournal.org 
Obesity
for changes between presurgery and 18 months postsurgery for
total energy intake (P = 0.09, R 2 = 0.18) and energy density (P = 0.07.
R 2 = 0.20). No associations were found between intake of sweet foods
and %TWL at 18 months after surgery (P ≥ 0.35) or between intake
from the remaining macronutrients and %TWL at 18 months after
surgery (all P ≥ 0.19).
Discussion
We confirmed our previous findings, that RYGB surgery and SG sur-
gery do not affect food preferences when assessed by an ad libitum
buffet meal targeting direct behavior. Total energy intake at the meal
decreased after surgery, but there were no changes in food preferences.
Thus, the reduction in calories was simply caused by eating smaller
portions of the same food items. However, changes in energy intake
and food preferences seem to be predictive of the postoperative weight
loss at 18 months.
This is the first study to investigate changes in food preferences fol-
lowing bariatric surgery using an ad libitum buffet meal test. Our
results contradict those of previous studies using verbal report mea-
sures, emphasizing the need for further investigation of food pref-
erences targeting direct behavior. Mechanistic studies investigating
the effect of surgery on taste sensitivity report decreased (25-28) or
unchanged (29,30) sucrose detection following surgery. Furthermore,
appetitive behavior for sweet and fatty food also seems to be reduced
following surgery (31), and the mechanisms have further been elu-
cidated by functional neuroimaging studies showing a reduced
reward-related response to pictures of palatable foods (32-34). These
changes are all consistent with studies reporting changes in food
preferences following bariatric surgery. However, it is uncertain if
changes in taste sensitivity and food reward translate to a modifica-
tion of food selection and preferences; thus, implications for actual
food consumption behavior are not possible to infer from these stud-
ies. Animal models, however, allow for assessment of more direct
behavior, using the two-bottle preference test and a cafeteria diet
paradigm. Using animal models, it is possible to examine changes in
food preferences in strictly controlled experiments, while excluding
the potential impact of social stigma associated with unhealthy eat-
ing behavior seen in humans. Interestingly, animal studies replicate
previous findings in humans, reporting a decrease in preferences for
high-fat and sweet foods following surgery (4,14-18). These results,
however, are not in line with the findings in present study target-
ing direct behaviors of food intake in humans. The limited exposure
of rats to different diets compared with the large range of different
foods humans are exposed to throughout life could be a potential
explanation for these discrepancies. Furthermore, rodent models
have much less biological variations than human models, increasing
the likelihood of detecting differences in food preferences on the
group level.
We hypothesized that 18 months after surgery, energy intake at
the buffet meal would increase compared with 6 months after sur-
gery, thereby potentially reaching a threshold at which postinges-
tive responses would be sufficient to lead to a substantial shift in
food preferences for most patients. This hypothesis was based on
the view that the effect of RYGB and SG surgery on food prefer-
ences is caused by a learned adjustment to postprandial visceral sig-
nals rather than by a change in experienced palatability (35). This
Obesity | VOLUME 00 | NUMBER 00 | MONTH 2018    7
Obesity Food Preferences Assessed by Buffet Meal Test  Søndergaard Nielsen et al.
learned adjustment was hypothesized to be caused by conditioned
taste avoidance in which the triggering foods can still be consumed,
as long as the patients learn to reduce consumption to avoid certain
postingestive signals or responses (36). Thus, one possibility was
that energy intake at 6 months was too low to trigger these postpran-
dial visceral signals and, consequently, a shift in food chosen at the
buffet meal or that learned adjustment to conditioned taste avoidance
progresses over time so that 6 months postsurgery was too early for
behavioral changes to set in. However, we were not able to confirm
our hypothesis, even though energy intake increased 25% between 6
and 18 months after surgery.
To fully explore our hypothesis, we divided the subjects into tertiles
according to total energy consumed after surgery. Food preferences
changed in both the low and high energy intake group in opposite direc-
tions. However, these changes were contrary to our hypotheses. In the
high energy intake group, patients consumed more food from the high-
fat category; hence, this group may either have had a higher threshold
for triggering postprandial visceral signals or a less profound biological
response to these visceral signals compared with the group with lower
energy intake.
Thus, we can preclude that our previous findings 6 months after sur-
gery were caused by a low energy intake or an early assessment of
food preferences occurring before adaptation to conditioned taste
avoidance.
We found no associations between postoperative experiences of nega-
tive responses related to eating and food preferences at the buffet meal.
This result is not caused by a low incidence of subjects reporting post-
prandial discomfort since all patients reported one or more experiences
with unpleasant physiological responses to food after surgery. As early
as 6 weeks after surgery, 95% of the subjects reported postprandial
discomfort, indicating that patients, despite receiving nutritional coun-
seling as part of the bariatric surgery program, encountered food that
caused visceral malaise.
Even though we found no consistent changes in food preferences
for the entire cohort of patients, changes in high fat intake, protein
intake, total energy intake, and energy density seem to be predic-
tive of the postoperative weight loss. The decrease in energy intake,
energy density, and intake of high-fat foods and the increase in pro-
tein intake from before surgery to 6 months after surgery predicted
the extent of %TWL at 18 months. The association between changes
in high-fat food and %TWL at 18 months after surgery was still sig-
nificant when analyzing changes in intake from before surgery to
18 months after surgery. These results suggest that changes in food
preferences, especially in the early postoperative period, could be
important markers for weight loss success following surgery because
they explained up to 39% of the variability in later weight loss. With
the potential of identifying patients in need of additional support to
improve their weight loss success, this association should be further
investigated.
In spite of a larger weight loss in RYGB compared with SG patients,
we found no changes in food preferences when stratifying subjects
according to surgery or differences between surgical groups. Energy
consumed at the meal was higher in RYGB patients; however, the
8    Obesity | VOLUME 00 | NUMBER 00 | MONTH 2018
decrease in intake from before surgery to after surgery was compa-
rable between RYGB and SG patients (41% and 43% respectively).
Studies comparing the effect of RYGB and SG surgery on food
preferences are scarce. In a pilot study, El Labban et al. did not
detect any major differences in food intake and preferences between
RYGB and SG operated patients (37). Nance et al. compared pre-
to postoperative changes in eating behavior and detected similar
changes after RYGB and SG surgery (30). Thus, in spite of different
anatomic changes following RYGB and SG surgery, these proce-
dures seem to have similar outcomes on dietary intake and food
preferences.
When measuring food preferences by means of verbal report, we
found no change in subjects’ choice of picture from the different
food categories between presurgery and 18 months postsurgery.
However, at the assessment 6 months after surgery, patients reported
an increased preference for low-fat savory foods compared with
before surgery, which is in line with the dietary advice received as
part of the bariatric surgical program. These results could indicate
that patients are more aware of the nutritional recommendations in
the early postoperative period, whereas this awareness seems to fade
with time after surgery.
Assessing food preferences by targeting direct behavior is a major
strength of this study. However, the experimental setting of the meal
has some limitations, which have previously been described (13).
Briefly, we cannot preclude that the social desirability of healthy
eating might influence the choice of food at the buffet meal, and
whether the assessed behavior at the buffet meal test reflects actual
behavior outside the experimental setting can be questioned. The
standardized liquid meals consumed prior to the buffet meal were
predominantly “sweet meals” (e.g., chocolate and chocolate mint
flavor). This could have reduced intake of sweet foods at the buf-
fet meal because of reduced liking for sweetness caused by sen-
sory-specific satiety (38). Such an effect would be present at all
visits, and also before surgery, potentially limiting the ability to
detect a decrease in preference for sweet foods following surgery.
Additionally, the low number of SG participants (n = 10) limited the
ability to make specific conclusions regarding the effect of SG sur-
gery on food preferences.
Conclusion
Data from this 18-month follow-up study confirmed our previous
findings that RYGB surgery and SG surgery do not consistently af-
fect food preferences in humans. Energy intake was still consider-
ably decreased; however, we found no overall changes in relative
intake of high-fat, low-fat, sweet, and savory foods; macronutrient
intake; or energy density of the meal. These results conflict with
previous results based on verbal report measures, emphasizing the
need for further studies investigating changes in food preferences by
targeting direct behavior. This is especially important if changes in
energy intake and food preferences can predict postoperative weight
loss, as our data indicate, because this will strengthen identification
of subjects in need of additional early support to optimize weight
loss.O
© 2018 The Obesity Society
www.obesityjournal.org
Original Article
CLINICAL TRIALS AND INVESTIGATIONS
References
1. Miller GD, Norris A, Fernandez A. Changes in nutrients and food groups in-
take following laparoscopic Roux-en-Y gastric bypass (RYGB). Obes Surg
2014;24:1926-1932.
2. Laurenius A, Larsson I, Melanson KJ, et al. Decreased energy density and
changes in food selection following Roux-en-Y gastric bypass. Eur J Clin Nutr
2013;67:168-173.
3. Olbers T, Björkman S, Lindroos A, et al. Body composition, dietary intake, and en-
ergy expenditure after laparoscopic Roux-en-Y gastric bypass and laparoscopic ver-
tical banded gastroplasty: a randomized clinical trial. Ann Surg 2006;244:715-722.
4. le Roux CW, Bueter M, Theis N, et al. Gastric bypass reduces fat intake and prefer-
ence. Am J Physiol Integr Comp Physiol 2011;301:R1057-R1066.
5. Ullrich J, Ernst B, Wilms B, Thurnheer M, Schultes B. Roux-en Y gastric bypass
surgery reduces hedonic hunger and improves dietary habits in severely obese sub-
jects. Obes Surg 2013;23:50-55.
6. Ammon BS, Bellanger DE, Geiselman PJ, Primeaux SD, Yu Y, Greenway FL.
Short-term pilot study of the effect of sleeve gastrectomy on food preference. Obes
Surg 2015;25:1094-1097.
7. Ernst B, Thurnheer M, Wilms B, Schultes B. Differential changes in dietary habits
after gastric bypass versus gastric banding operations. Obes Surg 2009;19:274-280.
8. Coluzzi I, Raparelli L, Guarnacci L, et al. Food intake and changes in eating behav-
ior after laparoscopic sleeve gastrectomy. Obes Surg 2016;26:2059-2067.
9. Molin Netto BD, Earthman CP, Farias G, et al. Eating patterns and food choice
as determinant of weight loss and improvement of metabolic profile after RYGB.
Nutrition 2017;33:125-131.
10. Halmi KA, Mason E, Falk JR, Stunkard A. Appetitive behavior after gastric bypass
for obesity. Int J Obes 1981;5:457-464.
11. Coughlin K, Bell RM, Bivins BA, Wrobel S, Griffen WO. Preoperative and postop-
erative assessment of nutrient intakes in patients who have undergone gastric bypass
surgery. Arch Surg 1983;118:813-816.
12. Kenler HA, Brolin RE, Cody RP. Changes in eating behavior after horizontal gas-
troplasty and Roux-en-Y gastric bypass. Am J Clin Nutr 1990;52:87-92.
13. Nielsen MS, Christensen BJ, Ritz C, et al. Roux-En-Y gastric bypass and sleeve
gastrectomy does not affect food preferences when assessed by an ad libitum buffet
meal. Obes Surg 2017;27:2599-2605.
14. Shin AC, Zheng H, Pistell PJ, Berthoud HR. Roux-en-Y gastric bypass surgery
changes food reward in rats. Int J Obes (Lond) 2011;35:642-651.
15. Mathes CM, Letourneau C, Blonde GD, le Roux CW, Spector AC. Roux-
en-Y gastric bypass in rats progressively decreases the proportion of fat calo-
ries selected from a palatable cafeteria diet. Am J Physiol Integr Comp Physiol
2016;310:R952-R959.
16. Mathes CM, Bohnenkamp RA, Blonde GD, et al. Gastric bypass in rats does not de-
crease appetitive behavior towards sweet or fatty fluids despite blunting preferential
intake of sugar and fat. Physiol Behav 2015;142:179-188.
17. Chambers AP, Wilson-Perez HE, McGrath S, et al. Effect of vertical sleeve
gastrectomy on food selection and satiation in rats. Am J Physiol Metab
2012;303:E1076-E1084.
18. Wilson-Perez HE, Chambers AP, Sandoval DA, et al. The effect of vertical sleeve
gastrectomy on food choice in rats. Int J Obes (Lond) 2013;37:288-295.
19. Manning S, Pucci A, Batterham RL. Roux-en-Y gastric bypass: effects on feeding
behavior and underlying mechanisms. J Clin Invest 2015;125:939-948.
www.obesityjournal.org 
Obesity
20. Poppitt SD, Swann D, Black AE, Prentice AM. Assessment of selective under-re-
porting of food intake by both obese and non-obese women in a metabolic facility.
Int J Obes Relat Metab Disord 1998;22:303-311.
21. Kapoor N, Al-Najim W, le Roux CW, Docherty NG. Shifts in food preferences after
bariatric surgery: observational reports and proposed mechanisms. Curr Obes Rep
2017;6:246-252.
22. Mathes CM, Spector AC. Food selection and taste changes in humans after
Roux-en-Y gastric bypass surgery: a direct-measures approach. Physiol Behav
2012;107:476-483.
23. Finlayson G, King N, Blundell JE. Is it possible to dissociate “liking” and “want-
ing” for foods in humans? A novel experimental procedure. Physiol Behav
2007;90:36-42.
24. Smith VA, Preisser JS, Neelon B, Maciejewski ML. A marginalized two-part model
for semicontinuous data. Stat Med 2014;33:4891-4903.
25. Bueter M, Miras AD, Chichger H, et al. Alterations of sucrose preference after
Roux-en-Y gastric bypass. Physiol Behav 2011;104:709-721.
26. Altun H, Hanci D, Altun H, et al. Improved gustatory sensitivity in morbidly
obese patients after laparoscopic sleeve gastrectomy. Ann Otol Rhinol Laryngol
2016;125:536-540.
27. Burge JC, Schaumburg JZ, Choban PS, DiSilvestro RA, Flancbaum L. Changes in
patients’ taste acuity after Roux-en-Y gastric bypass for clinically severe obesity. J
Am Diet Assoc 1995;95:666-670.
28. Scruggs DM, Buffington C, Cowan GSM. Taste acuity of the morbidly obese before
and after gastric bypass surgery. Obes Surg 1994;4:24-28.
29. Pepino MY, Bradley D, Eagon JC, Sullivan S, Abumrad NA, Klein S. Changes in
taste perception and eating behavior after bariatric surgery-induced weight loss in
women. Obesity (Silver Spring) 2014;22:E13-E20.
30. Nance K, Eagon J, Klein S, Pepino M Effects of sleeve gastrectomy vs. Roux-en-Y
gastric bypass on eating behavior and sweet taste perception in subjects with obe-
sity. Nutrients 2017;10. https://doi.org/10.3390/nu10010018
31. Miras AD, Jackson RN, Jackson SN, et al. Gastric bypass surgery for obesity de-
creases the reward value of a sweet-fat stimulus as assessed in a progressive ratio
task. Am J Clin Nutr 2012;96:467-473.
32. Scholtz S, Miras AD, Chhina N, et al. Obese patients after gastric bypass sur-
gery have lower brain-hedonic responses to food than after gastric banding. Gut
2014;63:891-902.
33. Ochner CN, Stice E, Hutchins E, et al. Relation between changes in neural respon-
sivity and reductions in desire to eat high-calorie foods following gastric bypass
surgery. Neuroscience 2012;209:128-135.
34. Ochner CN, Kwok Y, Conceição E, et al. Selective reduction in neural responses to
high calorie foods following gastric bypass surgery. Ann Surg 2011;253:502-507.
35. Gero D, Steinert RE, le Roux CW, Bueter M. Do food preferences change after
bariatric surgery? Curr Atheroscler Rep 2017;19:38.
36. Abdeen G, le Roux CW. Mechanism underlying the weight loss and complications
of Roux-en-Y gastric bypass. Review. Obes Surg 2016;26:410-421.
37. El Labban S, Safadi B, Olabi A. The effect of Roux-en-Y gastric bypass and sleeve
gastrectomy surgery on dietary intake, food preferences, and gastrointestinal symp-
toms in postsurgical morbidly obese Lebanese subjects: a cross-sectional pilot
study. Obes Surg 2015;25:2393-2399.
38. He W, Boesveldt S, Delplanque S, de Graaf C, de Wijk RA. Sensory-specific sati-
ety: Added insights from autonomic nervous system responses and facial expres-
sions. Physiol Behav 2017;170:12-18.
Obesity | VOLUME 00 | NUMBER 00 | MONTH 2018    9