Participants
Thirty-four people (2 male, 32 female) with previous obesity surgery interventions for severe obesity seeking a contouring surgery were recruited in the outpatient service of the Plastic Surgery Unit of the University of Padova. Obesity surgery patients were recruited between 2 and 10 years after their laparoscopic sleeve gastrectomy; they all had stable weights for, at least, 6 months. A group of 35 BMI-matched controls (MC) of matched age and BMI were selected from a sample of international subjects who participated in a research project about body image evaluation (3 male, 32 female), called BodyTalk project [29]. The inclusion criteria for both groups were an age between 18 and 65 years and no severe mental and medical comorbidity (e.g., no eating disorders), no neurological trauma and disorders, nor drug addictions. The exclusion criteria for the MC were that they could not have a history of obesity surgery interventions or desire to have any. Informed consent was collected from each participant. The study was approved by the local ethic committee as part of a larger study on the cognitive evaluation of obesity surgery patients, and it complies with the provisions of the Declaration of Helsinki.
Assessment Instruments
OS participants were evaluated by a trained researcher with clinical interviews, assessing weight history, behaviors and previous psychological and surgery interventions, and exclusion and inclusion criteria were applied. After providing informed consent, included participants were administered specific questionnaires about eating and weight concerns, depression, self-esteem and body evaluation, and afterwards completed three computerized tasks. The MC were tested with the same tasks using the same methodology across the different centers. The following instruments were used.
The Rosenberg self-esteem scale (RSES) [30] is a well-established self-reported 10-item test about self-esteem. For each item, there is a Likert scale response from “strongly agree” to “strongly disagree.” Results higher than 15 may indicate seriously low self-esteem.
The physical appearance comparison scale (PACS) [31] is a 5-item self-reported test used to assess the degree of physical comparison with others in various social situations. Responses are collected on a Likert scale, ranging from “never” to “always.” Higher scores indicate higher social comparison tendencies.
The body dissatisfaction subscale and the drive to thinness subscale of the eating disorder inventory (EDI) [32] are both well-established measures of specific psychopathology constructs about body shapes. Each item is on a 6-point scale, ranging from “always” to “never” and rated 0–3, and higher scores indicate higher body dissatisfaction.
The patient health questionnaire (PHQ-9) [33] is a 9-item self-reported test, and it is considered a good measure of depression. There are four answer categories from “not at all” to “almost every day”, and higher total scores indicate higher depression symptomatology.
The body image questionnaire (BIQ-20) [34] is a measure of the dynamism of one’s body (e.g., “I feel very fit”—perception of body dynamics (PBD) subscale), as well as of its rejection (e.g., “My body often annoys me”—negative evaluation of the body (NEB) subscale). There are five answer categories, ranging from “Not true” to “Absolutely true,” and higher scores indicate a more negative body image. The translation from the original Germany version to the Italian version was performed independently by an author and a professional translator, and the two versions were reviewed by two different German-speaking authors. A backward translation was performed in order to evaluate the semantic value of the questionnaire. Good reliability was found: PBD Cronbach’s α = 0.81 and NEB Cronbach’s α = 0.85.
Study Design
Semantic evaluation of body shapes and weight bias were assessed using a computerized evaluation of body shapes, as already applied to other clinical population [29]. The body shape evaluation was composed of three different tasks performed by participants and recorded using a 17″ laptop in the morning. See Fig. 1 for a visual representation.
-
1.
In the first task, the rating task, 12 body images, and 16 adjectives were presented to participants and asked to evaluate how the match was adequate. The image set was composed of a realistic human body stimulus of different weights: underweight, normal weight, overweight, and obese [35]. We selected the bodies to cover a range in BMI from 15.5 to 36.5 kg/m2. The adjective set was composed of 16 adjectives describing both physical and behavioral factors selected from literature about weight bias and already used in other studies: active, apple-shaped, attractive, clumsy, determined, feminine, heavy-set, hourglass-shaped, impulsive, insecure, lazy, open-minded, pear-shaped, smart, thin, and unfriendly [36]. The adjectives were translated into Italian by a professional translator and an author independently and then checked by another author to evaluate the differences. A third author then translated back the adjectives in order to evaluate semantic correspondence. All possible combinations were presented randomly to the participants, and the evaluation was performed with a 4-point Likert scale from “very much” to “not at all.”
-
2.
In the second task, the adjustment task, an adapted version of the body shape visualization tool, was used [36]. The participants were asked to modify a digital, realistic neutral human presented on the computer screen using eight scrollbars, each representing a principal component of the body. The goal of this task was to generate a biometrically plausible 3D body model for each specific adjective from the adjective set. The first trial was carried out without providing any adjectives, and it was used to familiarize the participants with the scrollbars. Subsequently, each adjective from the first task set was presented randomly, and participants were asked to generate a prototypic body matching the adjective. At the end of the task, participants were asked to reproduce their own body.
-
3.
In the third task, the valence assessment task, people were asked to evaluate all the presented adjective with a 5-point Likert scale from “clearly negative” to “clearly positive,” as a measurement of adjectives valence. Adjectives were presented to all groups in a randomized order.
Statistical Analysis
The analysis was conducted with IBM SPSS Statistics 25.0 (SPSS, Chicago, IL, USA). A normal distribution test was performed for all variables. The demographic variables and questionnaire scores were tested using independent t-tests. To analyze semantic evaluation of body shape, we estimated the BMI of the generated bodies and compared them between the groups.
In the adjustment task, the height, in meters, of the human model was calculated subtracting its lowest point from its highest point. The volume in cubic meters of the models was calculated as described by Zhang and Chen [37], and, dividing its volume by the average human body density (1010 kg/m3) [38] was obtained the weight of the models and was calculated the BMI.
In the valence assessment, no differences between subgroups emerged in the analysis of the assigned adjective valence (results from the valence assessment task). Adjectives were then grouped by the valences, from negative to positive according to participants’ ratings, and the average BMIs were evaluated in order to look at the mental representation of body shapes.
To evaluate weight bias, data from the rating task were aggregated with data from the valence assessment task. Thus, we obtained the valence values for body weight categories (i.e., underweight, normal weight, overweight, and obese) and we were able to split them into positive/negative adjectives. The normally distributed data was analyzed using t tests. Pearson’s correlation analysis was used to evaluate how the BMI of rated body shapes was associated with valence of the adjective. Comparisons between correlations were performed using the Fisher’s r to z transformation [39].
Body size estimation accuracy was assessed using the body perception index. It was calculated as the percentage of the ratio between the estimated BMI through the visualization tool and the actual BMI, in order to compare body perception [40]. Linear regression analysis were performed between actual BMI and own represented BMI, with the goal to evaluate if the accuracy could be explained by the participants’ BMI.
Several regression analyses were also performed between the human model’s BMI and psychological scores with stepwise approaches. The alpha was set at 0.05 for all the analyses. The effect sizes were calculated with Cohen’s delta.