Introduction

Obesity is a rising public health problem worldwide that is related to increased morbidity and mortality [1]. Bariatric surgery is one of the most effective treatment methods for significant and sustainable weight loss in severe obesity [2]. Individuals with morbid obesity also have a high rate of psychiatric disorder comorbidity in addition to their metabolic co-morbidities. Approximately 66% of bariatric surgery patients have a lifetime history of one or more Axis I psychiatric diagnoses, with 38% meeting the criteria for an Axis I diagnosis for DSM-IV at surgery [3]. One meta-analysis reported that depressive and eating disorders were the most common psychiatric disorders in bariatric surgery patients [4].

The association between obesity and psychiatric disorders is interpreted as bi-directional. A change in appetite is a symptom of depression and eating habits may deteriorate in response to negative emotional situations [5]. Mechanisms, including serotonin release dysfunction, inflammation, and hypothalamic–pituitary–adrenal (HPA) axis dysregulation, also play roles in the pathophysiology of both depressive and anxiety disorders as well as in obesity [6]. Furthermore, stigmatization due to obesity can contribute to the development of some psychiatric disorders, such as depression and anxiety disorders [7].

Psychiatric outcomes may vary depending on the type of the bariatric procedure. In general, patients undergoing different surgical techniques show improvements in their quality of life and psychopathology after the surgery [8,9,10]; however, the results are inconsistent. Some studies have also reported deteriorations in mental health in the long term, despite an initial post-surgical improvement [11,12,13]. In addition, some researchers reported that majority of the patients retained an obese view of self despite weight loss [14]. Most of the studies conducted with LSG demonstrated that depressive symptoms, self-esteem, and eating psychopathology improved after surgery especially in the short term [15,16,17,18].

An association has been noted between a patient’s preoperative psychopathology and the weight loss [19,20,21,22]; however, this association has not been identified in other studies [4, 23, 24]. Patients suffering from severe psychiatric disorders not related to obesity might experience difficulties in adapting to postoperative behavioral changes and, therefore, might lose less weight. Conversely, conditions associated with obesity, such as mild depressive symptoms or low self-esteem, may have no negative effects on weight loss [25]. On the other hand, studies investigating the association between postoperative psychopathology and weight loss have also reported inconsistent results [15, 24, 26, 27].

The previous studies that investigated associations between bariatric surgery and psychopathology mainly focused on gastric bypass and vertical banded gastroplasty (VBG). Few studies have investigated laparoscopic sleeve gastrectomy (LSG), a restrictive single-stage procedure that is relatively new in the field of bariatric surgery. LSG has been gaining popularity due to its efficacy in weight reduction and its low surgical and nutritional risks [28]. It is also an effective therapeutic option for adolescents with morbid obesity [29]. The aim of the present study was: (1) to investigate changes in depressive symptoms, self-esteem, and eating psychopathology in bariatric surgery patients prior to LSG and at the 6- and 12-month follow-ups; (2) to compare the psychopathology scores with those of healthy controls; and (3) to analyze the association between the amount of weight loss and the pre- and postoperative psychopathology scores.

Methods

Participants

This prospective study was conducted at the Surgery Department of the Selçuk University Faculty of Medicine. As a part of routine comprehensive evaluation for bariatric surgery, an endocrinologist, a surgeon, and a psychiatrist assessed the patients with obesity who sought weight loss treatment and additional consultations were made if needed. Suitability for bariatric surgery was decided by the recommendations of the current clinical guidelines (body mass index [BMI] > 40 kg/m2 or BMI > 35 kg/m2 with significant weight-related medical co-morbidities, and failure in conservative treatment) and medical professionals. In addition, the patients’ approval for the surgery was taken into consideration. Patients who were scheduled to undergo LSG from July 2015 to July 2016 were planned to be included consecutively in this study. Totally, 72 patients met the eligibility requirements for the study. Of these, 22 declined to participate in the study, leaving 50 patients in the study group. A further 50 non-obese individuals were randomly selected as the control group. All participants were between 18 and 65 years of age. Exclusion criteria for all participants included the presence of mental retardation and alcohol or substance use disorders. Two patients dropped out, leaving a final total of 48 (96%) patients who completed the 12-month follow-up.

Measures and procedure

Considering the objectives of the study, a sociodemographic data form was created by the researchers to collect sociodemographic variables. The Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, 4th Edition Axis I Disorders (SCID-I) was used to assess the psychiatric co-morbidities and to determine the competence for surgery. Each participant was assessed for depressive symptom level with the Hamilton Depression Rating Scale (HDRS, 17-item version), for eating disorder psychopathology with the Eating Disorder Examination Questionnaire (EDE-Q, 33-item version) and for self-esteem with the Rosenberg Self-esteem Scale (RSES- first 10-item). All scales were in the Turkish forms, and the validity and reliability of all scales have been confirmed [30,31,32].

The EDE-Q, which assesses eating disorder psychopathology, comprises four subscales (dietary restraint, eating concerns, weight concerns, and shape concerns) and yields a global score. Items are rated on seven point scales (0–6), with higher scores reflecting greater severity or frequency. Most items are specific to current symptomatology, encompassing the previous 28-day period.

Patients were evaluated at least 2 weeks prior to surgery (t0) and at 6 months (t1) and 12 months (t2) after LSG. Healthy controls were evaluated only once.

The Sociodemographic Data Form, SCID-I, HDRS, EDE-Q, and RSES were applied to patients at t0. The t1 and t2 assessments of patients consisted of all of these scales except for SCID-I. All assessments were conducted face to face by a psychiatrist.

All participants were also measured for their weight (kg) and BMI (kg/m2) by medical professionals. The percent excess weight loss (%EWL; [weight loss between T0 and follow-up/excess weight at T0]*100) of the patients were also calculated during the follow-up visits.

Statistical analysis

Statistical analyses were conducted using the Statistical Package for the Social Sciences (SPSS) v20.0. Descriptive statistics included the mean and standard deviation (SD) was used for continuous data, whereas frequency and percentages were used for discrete data. The Kolmogorov–Smirnov/Shapiro–Wilk test was conducted to evaluate the compliance of the data with a normal distribution. The two independent groups were compared using the Chi-square/Fisher’s exact test for discrete data, the Student t test for continuous data with a normal distribution, and the Mann–Whitney U test for continuous data without a normal distribution.

In the comparison of dependent variables of patients, paired-samples t test (for 2 measurements)/repeated measures ANOVA (for more than 2 measurements) was used for continuous data with a normal distribution, while Wilcoxon (2 measurements)/Friedman test (> 2 measurements) was used for continuous data without a normal distribution. The Bonferroni correction was applied to the post-hoc tests.

A series of multiple linear regression analyses was performed to predict BMI at t1/t2 as a dependent variable using the baseline scores of each scale separately after controlling for baseline BMI, gender, and age. Regression analyses were also repeated separately for the percent change in scores of each scale. Assumptions for regression analyses were checked for autocorrelation of residuals using the Durbin–Watson test and for multicollinearity by assessing the tolerance and the variance inflation factor; these assumptions were met. A p value < 0.05 (two-tailed) was considered statistically significant.

Results

The study sample consisted of 48 patients and 50 controls. No statistically significant difference was identified between patients and control group in terms of age, gender, education, and employment. At t0, 31% of the patients (n = 15) had psychiatric disorders, including six patients with eating disorders (5 with binge eating disorder, 1 with bulimia nervosa), five with depressive disorders, and four with anxiety disorders. In the control group, two participants were diagnosed with depressive disorders and one with anxiety disorders. No patients or controls were on psychotropic treatment. Details are given in Table 1.

Table 1 Sociodemographic variables of patients and controls
Full size table

A statistically significant decrease was identified in the patient BMI levels (t0 > t1 > t2). The BMI levels were statistically significantly higher for all patient measurements when compared to the control group. The %EWL was 50.93 ± 9.87 at the t1 and 69.70 ± 10.44 at the t2 follow-up visits. A significant difference was identified between the t1 and t2% EWL levels (p < 0.001).

A statistically significant and progressive decrease was identified in the HDRS and RSES scores of patients when compared to their previous measurements. The RSES levels were higher for all patient measurements when compared to the controls. The HDRS scores were higher at t0 and t1 for the patients than for the controls, while no difference was identified at t2 (t0, p < 0.001; t1, p < 0.001; t2, p = 0.068). These results are summarized in Table 2.

Table 2 Clinical variables of controls and patients
Full size table

Comparison of female and male patients in terms of HDRS and RSES scores at t0, t1, and t2 follow-ups revealed no statistically significant difference (p > 0.05) (Table 2). Moreover, the percent change from baseline scores also showed no statistically significant difference between male and female patients (HDRS, change between t0–t1 p = 0.117, t0–t2 p = 0.946; RSES t0–t1 p = 0.544, t0–t2 p = 0.232).

No statistically significant correlation was detected between age and the percent change from baseline in HDRS and RSES scores of patients (p > 0.05).

A statistically significant and progressive decrease was identified in the EDE-Q weight concern and shape concern subscale scores of patients when compared to their previous measurements. The EDE-Q eating concern subscale score showed statistically significant decrease from t0 to t1, however, showed statistically significant increase from t1 to t2. A statistically significant decrease was detected in EDE-Q dietary restraint subscale score from t0 to t1, but no difference was identified between t1 and t2. A statistically significant decrease in EDE-Q total score was noted at t1 compared to t0, but no difference was identified between t1 and t2.

The EDE-Q total and all subscale scores were higher for all patient measurements when compared to the controls. These results are summarized in Table 3.

Table 3 EDE-Q variables of controls and patients
Full size table

Comparison of female and male patients in terms of EDE-Q total and subscale scores at t0, t1, and t2 follow-ups revealed no statistically significant difference (p > 0.05) (Table 3). In addition, the percent change from baseline scores also showed no statistically significant difference between male and female patients (EDE-Q total, change between t0–t1 p = 0.307, t0–t2 p = 0.666; dietary restraint t0–t1 p = 0.413, t0–t2 p = 0.223; eating concerns t0–t1 p = 0.768, t0–t2 p = 0.364; weight concerns t0–t1 p = 0.055, t0–t2 p = 0.117; shape concerns t0–t1 p = 0.768, t0–t2 p = 0.838).

No statistically significant correlation was detected between age and the percent change from baseline in EDE-Q total and subscale scores of patients (p > 0.05).

A series of multiple linear regression analyses were performed to investigate the association between the baseline scores of each scale (EDE-Q total, HDRS and RSES) separately and the BMI at t1/t2, after controlling for the baseline BMI, gender, and age (Table 4). Analyses were also performed for the percent change in scores from baseline for each scale separately (Table 5). Only the percent change of EDE-Q total was associated with BMI at t1. No association was found between BMI and other baseline values or percent changes in scores for any scale (p > 0.05).

Table 4 Regression analyses of baseline variables to predict BMI at t1/t2
Full size table
Table 5 Regression analyses of percent change of scale scores to predict BMI at t1/t2
Full size table

In addition, the above-mentioned regression analyses were repeated for the EDE-Q subscales. A statistically significant association was only identified for the percent change of EDE-Q dietary restraint subscale and BMI at t1 (B = − 0.021, p = 0.036). No association was found between BMI and other baseline values or percent changes in scores for any EDE-Q subscales (p > 0.05).

Discussion

Self-esteem and depression have a known relationship [33], and patients with obesity are reported to have lower self-esteem and higher depressive symptoms when compared to non-clinical populations [34]. Our results presented here are, therefore, in line with the existing literature. A significant progressive improvement was identified in depressive symptoms and self-esteem in our patients at the 6- and 12-month follow-ups after LSG. Similar results were also reported in short-term studies conducted with LSG [15,16,17, 27] and other surgical techniques [4, 8, 19, 24, 35].

The depressive symptoms of our patients were similar to those in the control group at the 12-month follow-up visit. Ostherhues et al. also reported no difference between patients and controls in terms of depression after sleeve gastrectomy (median = 19 months) [36]. Likewise, Aasprang et al. reported that a reduction in depressive symptoms in patients observed during the first year after biliopancreatic diversion with duodenal switch was followed by a slightly increase from the 2–5-year follow-up; however, the symptoms at the 5-year follow-up were similar to those of the controls [9]. On the other hand, in our patients, the self-esteem levels were still lower than in the control group at the 12-month follow-up. Taken together, these data indicate that depressive symptoms and self-esteem tend to improve, particularly after the first year (often called the “honeymoon phase”) of bariatric surgery.

It has been showed that comorbidity of depression and obesity is common and a reciprocal association was found between them. Mechanisms such as inflammation and HPA-axis dysregulation are found in both obesity and depression. Obesity is also frequently accompanied by several psychopathologies, such as stigmatization, body dissatisfaction, maladaptive eating patterns, and low self-esteem, which can also contribute to the development of depression [37]. Improvements in these factors might be pathways explaining the significant improvement in depression after bariatric surgery. Nevertheless, a number of studies have highlighted that the initial improvement in weight loss, depression, and self-esteem occurring in the short term after surgery might decrease in the long term [11,12,13, 38].

Our results revealed no association between preoperative depressive symptoms and weight loss. A previous meta-analysis revealed that preoperative depression was not consistently associated with differences in weight outcomes [4]. Similar results were also reported in studies conducted with LSG [15, 39]. For example, Semanscin-Doerr et al. reported no difference in the %EWL at the 12-month follow-up after LSG in patients with and without preoperative current depressive disorder [40].

Studies investigating the relationship between postoperative depressive symptoms and the weight loss following different surgical methods have reported varying results [26]. In our study, no association was found between the percent change in postoperative depressive symptoms and weight loss. Similarly, Rieber et al. reported no correlation between the level of weight loss and either the depressive scores at the 1-year follow-up point or the changes in depressive scores from the pre-LSG [15]. Similar to the depressive symptoms, the pre- and postoperative self-esteem levels of our patients showed no significant association with weight loss. However, the number of studies evaluating self-esteem in obesity is quite limited and the results are inconsistent [41,42,43,44].

Anatomical restriction, side effects, and neurobiological changes due to bariatric surgery might override the association between weight loss and depression and self-esteem, especially in the early postoperative period [20]. It has been reported that the relationship between these factors and weight loss has increased in the long term when the influence of the surgical procedures has waned [19, 20]. On the other hand, in the early postoperative period, the depressive symptoms and self-esteem may improve even without substantial weight loss. Factors such as patient expectations and patients taking an active role in changing their lives might affect this improvement [45]. In addition, the severity of the preoperative psychopathology and its relationship with obesity may have different effects on weight loss. A systematic review reported that mild psychopathologies such as low self-esteem and depression resulting from obesity may predict a positive effect on weight loss, whereas serious psychopathologies not associated with obesity may have negative effects [25]. Therefore, it is important for clinicians to consider psychiatric and psychosocial assessments in this process to increase the success of surgery especially in the long term. Clinical practice guidelines suggest both preoperative and postoperative assessments by mental health professionals to identify patients who are likely to have difficulties adjusting to changes after bariatric surgery [46].

Patients with obesity seeking bariatric surgery frequently have eating psychopathologies [4, 47] and at rates higher than in the normal population [34, 48, 49]. Our results also revealed a higher EDE-Q total and subscale scores in patients at the preoperative stage than in the control group. The EDE-Q total score of the patients improved at the 6-month follow-up when compared to the preoperative period. However, this initial improvement stabilized after 6 months and showed no further improvement. The data in the literature also indicate that eating psychopathology generally decreased in the short term after surgery [15, 18, 47]. However, in our study, EDE-Q total and all subscale scores at the 12-month follow-up were still significantly higher than in the controls. Similarly, Figura et al. reported that some eating scores improved, but they remained higher than in the normal population at 19 months after LSG [48].

Assessment of EDE-Q subscale scores revealed that shape and weight concern scores improved progressively during the follow-up. This improvement may be associated with progressive weight loss as well as improvements in self-esteem and depressive symptoms seen in the postoperative first year. A study reported an association between overvaluation of weight/shape and depressive symptom severity [50]. It was also reported that body dissatisfaction scores of patients with obesity who undergo LSG improved in the postoperative period [48]. On the other hand, in our study, dietary restraint subscale score improved in the first 6 months after LSG, but then stayed stabile. Eating concern subscale score improved in the first 6 months after LSG, but worsened after 6 months.

The observed reduction in eating psychopathology, especially in the short term after surgery, can be affected by factors such as anatomical restriction due to the bariatric surgery itself and side effects caused by too rapid or too abundant food consumption. The physical limitation may also have an indirect effect by reducing the sense of “loss of control over eating”, which accompanies many maladaptive eating behaviors. However, the dynamics responsible for an eating disorder could survive after surgery [45, 51]. For example, the risk of eating psychopathology after surgery is high in patients with a preoperative eating psychopathology [24, 52, 53]. In addition, preoperative binge eating might convert to postoperative grazing [51, 54] which is one of the most frequent problematic eating behavior after surgery [49]. Our results also revealed that preoccupying with eating and dietary restraint worsened despite an initial improvement. This result supports that cognitive features of eating psychopathology may persist especially after the initial effects of surgery have waned. Therefore, patients should be referred to mental health professionals, when needed, to deal with maladaptive eating behaviors before surgery and during follow-up. Moreover, given obesity’s multifactorial origin and its deleterious outcomes, a comprehensive management is recommended which can be accomplished by a multidisiplinary team to deal with different aspects of obesity and its related disorders as well as to provide different treatment options. An obesity team should be composed of several health professionals such as an endocrinologist, a bariatric surgeon, a psychiatrist, a clinical nutritionist, and a physiatrist together with other health professionals (i.e., dietitian, psychologist, physiotherapist, and nurse) [55].

According to our results, weight loss had no association with preoperative eating psychopathology. Similar results have been reported in most studies conducted with LSG [15, 18, 39]. Further evidence exists for a relationship between weight loss and postoperative eating psychopathology [18, 27, 48]. Contrary to expectations, in our study, the postoperative EDE-Q total score was associated with weight loss at the 6-month follow-up, whereas no association was found at the 12-month follow-up. The same result was also detected for dietary restraint subscale. It has been reported that dietary restraint is associated with binge eating episodes [56]. However, the relationship between eating psychopathology and weight loss is expected to increase over time, as the physical restriction and side effects due to surgery affect short-term eating behaviors (binge eating, grazing, etc.). The EDE-Q total and subscale scores used in this study particularly assesses the cognitive features of eating psychopathology (restraint, eating concerns, shape concerns, and weight concerns), but not the behavioral eating symptoms (frequency of binge eating, etc.), so this might have affected our results.

Our study had several limitations, including the short follow-up period, the limited number of patients, and the lack of use of obesity-specific scales. Although EDE-Q provides a general view of eating psychopathology, lack of assessment of the eating behaviors with more specific scales is another limitation. In addition, the analyses were made on the scale scores, and no diagnostic-based analysis could be performed due to the small sample size. Although patient and control groups were homogeneous in terms of sociodemographic variables, medical comorbidity frequency was higher in the study group. Besides, no inference can be made regarding the causality of the postoperative variables and the weight loss. Sample variability might also be reduced in the present study, since the patients who were approved for surgery and could adapt to the preoperative program were selected only from those undergoing LSG. These factors would restrict the generalizability of our results. Besides, differences in the factors, such as the study design, sample size, assessment instruments, and follow-up period in studies, complicate interpretation and comparison of results.

Conclusion

The present study is an attempt to provide an overall view of the important factors involved in the association between bariatric surgery and psychiatry. Our results revealed a progressive improvement in depressive symptoms and self-esteem in LSG patients for the 1-year follow-up after surgery. Eating psychopathology also improved, but eating concern worsened after 6 months. The depressive symptoms of patients reached a similar level at 12 months to those in the control group; however, self-esteem was still lower and eating psychopathology was still higher in the patients than in the control group at the 12-month follow-up. No association was identified between pre- and postoperative psychiatric factors and weight loss at the 12-month follow-up. In conclusion, apart from weight loss, clinicians should closely monitor patients during the course of follow-up in terms of psychiatric conditions and should adopt a multidisciplinary approach, when needed, for an improved psychiatric outcome. Further longitudinal studies assessing more variables in a larger sample could provide more comprehensive information about the association with psychiatric conditions and bariatric surgery.