Participants
The data are retrieved from the “Oslo Bariatric Surgery Study” (OBSS), which is a prospective study of two cohorts of patients that will be followed over a 10-year period. The OBSS focuses on identification of psychosocial predictors of behavioral change and weight loss maintenance. All participants are assessed with self-report questionnaires at five time points; pretreatment (baseline), 1, 3, 5, and 10-years post-treatment.
Two groups of patients were recruited: patients scheduled for bariatric surgery (surgical group) and patients starting a conservative weight reduction treatment (non-surgical group). The inclusion criteria for both groups were BMI ≥40 or BMI ≥35 kg/m2, with obesity related comorbidity, age ≥18 years, and the ability to understand and comply with the study procedures. The patients in the surgery group had previous failed attempts of sustained weight loss using conservative measures. Only patients with current, previous established or suspected psychiatric disorder were evaluated by a psychologist or a psychiatrist. There was no routine preoperative psychological screening. The patients were recruited from the Center for Morbid Obesity and Bariatric Surgery at Oslo University Hospital, between February 2011 and September 2013, after they had participated in a preoperative mandatory course. The course consisted of 36 h (10 meetings), which included topics like treatment options, diet, physical activity, emotions, motivation, and behavior change. Of the total number of patients awaiting bariatric surgery during the study period (N = 728, 79.4 % women, mean age = 46.4, SD = 9.6, and mean BMI = 45.8, SD = 6.4), 222 were excluded due to their inclusion in other ongoing studies (see Fig. 1). Of the remaining 506 patients, 318 consented to participate, of whom 301 (49.5 %) responded to the pre-surgery questionnaires. Of the intention to treat sample used in this study, 12 respondents (4 %) did not proceed with surgery but were not excluded as they did not differ on any of the study variables.
Patients in the non-surgical group were recruited between February 2011 and June 2014 from two rehabilitation centers: the Tonsåsen Rehabilitation Center and the Haugland Rehabilitation Center. At both centers, the treatment programs combine diet, physical activity, and counseling, during three 2-week residential stays over an 8-month period. Of the 420 patients at the Tonsåsen Rehabilitation Center, and the 130 patients at the Haugland Rehabilitation Center, 207 (48.6 %) and 57 (43.8 %) respectively, agreed to participate in the present study and completed the questionnaires prior to the start of treatment. Data from both samples were collapsed as the patients from both rehabilitation centers did not differ according to gender, age, education, or BMI (p > 0.05). The total response rate was 47.4 % (N = 261).
The Regional Ethics Committee for Medical Research (2009/1248a) approved the study protocol and all participants gave informed written consent before enrollment.
Assessments
Socio-demographic variables consisted of gender, age, educational background, and partner status. Weight was measured using a platform scale (Seca 635, III; 0–300 kg), with patients wearing light clothing and no shoes, on the day of approval for bariatric surgery and the first day of conservative treatment. Height was measured using a wall-mounted adjustable instrument.
Two family obesity indices consisted of two questions about participants’ childhood and family members (parents/siblings) with obesity, and two questions about their partners and offspring with obesity, with the response options of (0) no and (1) yes. The sum of the response of the two questions in each index ranged from 0 (none with obesity), to 2 (both with obesity).
Current behavior and dieting history
Current unhealthy eating habits were assessed using four items specifically constructed for this study (frequency of snacking between meals, snacking on sweets between meals, drinking soda between meals, and night eating) with response options ranging from (1) never to (5) always. Physical activity during the previous week was assessed using the International Physical Activity Questionnaire (IPAQ)-Short Form [26]. The mean scores were calculated by weighting the type of activity by energy requirements and reported as metabolic equivalent values (MET). Alcohol consumption during the past 12 months was assessed by measuring the frequency of consuming one or more units of alcohol during the past year, with response options ranging from (1) never consumed alcohol to (9) daily or almost daily consumption. Intoxication was evaluated using response options ranging from (0) no alcohol use to (5) visibly intoxicated 10 times or more. Self-monitoring of weight was measured by one item with response options ranging from (1) almost never to (7) more than once a day. Dieting and weight loss history were measured by single questions from the Survey for Eating Disorders (SED) [27] and Weight and Lifestyle Inventory (WALI) [28]. The questions addressed the number of past successful weight loss attempts >10 kg, the total number of times patients had participated in organized weight loss programs, and the number of times during the past year patients had been on a self-initiated diet lasting for 3 days or more. We also calculated the percentage of years of dieting relative to age. Weight loss methods were assessed using 11 items with the response options of (0) no and (1) yes [27]. Based on exploratory factor analyses, three types of methods were identified i.e., restricting the amount of food, restricting the type of food (e.g., fat or sugar), and unhealthy strategies (e.g., using laxatives, drugs, or vomiting).
Motivation, goal attainment, and expectations
Motivation to lose weight was measured with the response to a single question, ranging from (1) not to (10) extremely motivated; patients were also requested to rate the degree of social influence on their decisions to seek treatment, on a scale from (1) no influence to (5) strong influence. Readiness to restrict food intake and readiness to increase physical activity were measured on a scale from (1) not ready to (10) trying to change, extrapolated from the Readiness and Motivation Interview [29] and the trans-theoretical model of change [30]. Weight loss goal was measured using one question from the Goals and Relative Weight Questionnaire [31], from which the relative difference (%) between participants’ actual weight before treatment and their goal weight was calculated. A higher percentage indicates a higher expectation of weight loss. Outcome expectations were operationalized by asking the respondents to “indicate how likely you believe it is that you will feel this way three years after the operation/treatment” on a set of 9 items with scores ranging from (1) no to (10) high expectations. The scale was developed for this study. The factor analysis of the responses yielded two factors: well-being (e.g., satisfied with the amount of weight lost, general appearance, self-esteem, and feeling good about oneself), and social competence (e.g., improved sex life, being outgoing, personal success, and fewer concerns). General perceived self-efficacy, i.e., a strong belief in one’s ability to master new behaviors or situations, was assessed using the 10-item General Perceived Self-efficacy Scale [32].
Self-evaluation, emotional distress, protective factors, and social environment
Body image was evaluated using two subscales of the Multidimensional Body-Self Relations Questionnaire: the Body Areas Satisfaction Scale (BASS; 9 items) and Appearance Orientation (14 items) [33]. A high mean score on the BASS indicates satisfaction with body and weight, while a high mean score on the subscale Appearance Orientation signifies preoccupation with one’s appearance and body. Self-esteem was assessed using a 4-item short version of Rosenberg’s Self-Esteem Scale [34]. The Emotion Regulation Questionnaire [35] was used to measure the ability to regulate negative emotions by redefining the situation (6 items) or by merely suppressing them (4 items). Anxiety and depression symptoms were measured using the Hospital Anxiety and Depression Scale (HADS) [36]. Higher sum score (0–21) reflects higher level of anxiety and depression symptoms. A cut-off score >10 indicates a probable anxiety or mood disorder. The frequency of binge eating episodes in the past 3 months was assessed using one item from the Survey for Eating Disorders [37], scored as (1) never, (2) previously, or (3) now, in conformance with the DSM-IV (Diagnostic and Statistics Manual for mental disorders) definition of binge eating symptoms. Stressful life events, impact, and coping were assessed with a scale developed at the University of Tromsø by measuring the frequency of having experienced 17 negative events (e.g., death or illness in the family, job loss, divorce, violence, and sexual abuse). The impact of each event was rated by participants from (1) very low to (4) very high, and coping with the event was rated from (1) badly to (3) well. Relationship satisfaction was measured using the 5-item Relationship Assessment Scale [38] with scores ranging from (1) little to (4) much satisfaction. The Resilience Scale for Adults [39] was used to assess factors that may protect against maladjustment. It consists of the five subscales: family cohesion (6 items), social competence (7 items), social resources (7 items), personal strength (10 items), and structured style (4 items). All items had a semantic differential scale format with scores ranging from 1 to 7.
Statistical analyses
Analysis of covariance (ANCOVA) was used to compare the differences between the two study groups on all of the variables. As there were statistically differences in gender, age, and BMI between the two study groups, the group means were adjusted for possible confounders by entering gender, age, and BMI as covariates. The effect size, partial eta squared (ηp2) was reported, indicating the proportion of variance explained by a variable that is not explained by the covariates. The magnitude of the effect size for the partial eta squared is 0.01 (small), 0.06 (medium), and 0.14 (large), according to Cohen’s guidelines [40]. Multiple logistic regression analyses were conducted to identify the variables that uniquely predicted group membership. In these analyses, all study variables showing significant mean differences between the two treatment groups (p < 0.05) in the ANCOVA were included simultaneously, while controlling for BMI, age, and gender.