Abstract
Background:
The aim of this study is to analyze medical and psychosocial variables among gender and morbidly obese (Class III) and super obese (Class IV) groups to examine their associations with anxiety and depression symptoms and to determine predictive factors for anxiety and depression symptoms.
Methods:
This cross-sectional descriptive study was conducted with a sample of 546 obese patients. Medical, sociodemographic, anxiety, depression, and eating disorder information were obtained from all participants.
Results:
The study included 358 women (65.6%) with an average age of 42.71 (Standard deviation [SD] = 10.50) years, and 188 men (34.4%) whose average age was 41.80 (SD = 9.13) years. We found that eating disorder symptoms were positive and statistically significant predictors of anxiety and depression symptoms (p < 0.001). This finding was confirmed by multiple linear regression analysis (p < 0.001).
Conclusion:
Among Class III and IV obese patients, eating disorder symptoms were clearly associated with anxiety and depression levels. Therefore, interventions directed toward these symptoms would likely be beneficial.
Introduction
O
The standard clinical guidelines for obesity surgery include a preoperative psychological evaluation (PE) as an important element for the selection of appropriate BS candidates to decrease psychosocial disturbances after surgery and to identify patients with contraindications for surgery.5–9 Current illicit drug use, active symptoms of schizophrenia, severe mental retardation, and lack of knowledge about the surgery are the most commonly cited contraindications for BS. PE is also important for detecting emotional, psychiatric, cognitive, and behavioral factors that can influence surgical success and the potential for sustained weight loss. PE not only involves identifying predictors of BS success but also includes the development of individually tailored interventions to facilitate a successful postoperative course.10–12 The medical literature suggests that psychiatric comorbidities may be more prevalent in morbidly obese candidates for BS, especially when including those with binge eating disorders, affective disorders, anxiety, depression, bulimia, tobacco addiction, and personality disorders.1,3,10,13–15 However, Kalarchian et al. 16 reported that the clinical significance of psychiatric disorders among candidates for BS is not fully understood, and Heinberg, Ashton, and Windover 17 mentioned that there is no consensus regarding what constitutes an appropriate PE before BS.
Other studies18,19 have revealed the importance of gender, supporting the idea that women score higher on measures of all of the studied psychopathological parameters, including depression, paranoid ideation, interpersonal sensitivity, social anxiety, and the number of prior mental health treatments. Furthermore, recent studies using diagnostic interviews have demonstrated that 20–60% of women interviewed have a current psychiatric disorder, such as depression or an anxiety disorder.3,14–17,20
However, despite the frequency of these assessments, no consensus exists regarding the components of an appropriate assessment or the reasons for denying surgery to a candidate. Nonetheless, there is general agreement that a PE is necessary (but not sufficient), and there is also an agreement on the factors to be assessed, such as eating behaviors, stress and coping, social support, capacity to consent, understanding of the risks and benefits of surgery, knowledge of the surgical procedure and expectations for weight loss, health outcomes, and psychosocial impact.1–4,17 However, psychological variables have not been predictive of weight loss in the majority of studies. We can find only limited evidence indicating that anxiety and depression disorders may be good predictors of weight loss.1–4,21
Therefore, it seems useful to evaluate BS candidates on the basis of comorbidities and psychological and sociodemographic variables to find possible profiles and predictors of depression and anxiety symptoms. Accordingly, the three aims of this study were as follows: (1) to analyze medical and psychosocial variables by gender among morbidly obese (Class III obesity; body mass index [BMI] >40) and super obese (Class IV obesity; BMI >50) candidates for BS, (2) to establish the relationship between these variables and anxiety and depression symptoms, and (3) to determine which medical and psychosocial variables can best predict anxiety and depression symptoms.
Materials and Methods
Study design, participants, and procedure
This cross-sectional descriptive study was conducted with a sample of obese patients treated at the Multidisciplinary Unit of Healthcare for Obese Patients in the University General Hospital of Alicante (UGHA), Spain. During the 72 months of program operation (March 2009–2015), 546 of 642 candidates complied with the entire assessment protocol, resulting in a participation level of 85.05%. Of these, 409 (74.91%) were able to start the program directly following the initial assessment, 57 (10.44%) were able to start with the recommendation that they follow a preliminary monitored diet, 63 (11.54%) were able to start with the condition that they continue mental health treatment, and 17 (3.11%) patients were refused participation, but were offered subsequent reassessment.
The inclusion criteria were as follows: (1) age between 18 and 60 years; (2) BMI >40 kg/m2 or BMI ≥35 kg/m2 with specific comorbidities to undergo BS; (3) evolution of morbid obesity >5 years refractory to properly supervised conservative treatments; (4) absence of endocrine pathology; (5) absence of severe or acute psychiatric disorders; and (6) ability to understand the mechanisms of weight loss through BS and the commitment to adhere to a postoperative monitoring regimen. Absolute exclusion criteria included the presence of an active eating disorder, substance abuse, unstable severe psychiatric disorders, and severe mental retardation (IQ <50). Relative exclusion criteria included moderate mood disorders, severe personality disorders, and lack of family support.
All subjects reviewed and signed a statement of informed consent detailing the purpose, procedures, and goals of the study. Any questions were addressed during this initial review session. The diagnostic interview and related testing were completed in a single day. The assessment battery was routinely administered to all patients attending the Multidisciplinary Unit of Healthcare for Obese Patients at UGHA. The psychiatrist and the psychologist in charge conducted evaluations.
This study was approved by the ethics committee of the hospital.
Measures
Background variables
Demographic information (sex, age, education, and marital status), including height and weight, medical history (diabetes mellitus, hypertension, dyslipidemia, hyperuricemia, ischemic heart disease, osteoarthritis, obstructive sleep apnea (OSA), asthma, amenorrhea/infertility, polycystic ovary syndrome, and nonalcoholic fatty liver disease), and psychiatric history (depression, anxiety, and history of physical/sexual abuse), was provided by participants using a questionnaire created for this study.
The state/trait anxiety inventory
The Spanish version 22 was used. This 40-item questionnaire measures anxiety as a state and trait. The first 20 items evaluate transient anxiety symptoms (state anxiety subscale, the state/trait anxiety inventory [STAI]-S), while the other 20 items investigate persistent personality traits related to anxiety (trait anxiety subscale, STAI-T). All responses are on a Likert scale ranging from 0 to 3. The STAI-S subscale responses ranged from “nothing” (0 points) to “a lot” (3 points). The STAI-T subscale responses ranged from “almost never” (0 points) to “nearly always” (3 points). The STAI has discriminant validity and good internal consistency. A Spanish version of the STAI was also reported to have good internal consistency (0.9–0.93 for state anxiety and 0.84–0.87 for trait anxiety).
Beck depression inventory–short version (BDI-13)
The Spanish version by Conde and Franch 23 was used. This questionnaire evaluates depression from a cognitive perspective, consisting of the 13 items most relevant to serious depression, which also most strongly correlate with the total score from the original 21-item questionnaire. Each item requires the subject to rate their feelings during the past week on a 0–3 point Likert scale. The total score ranges from 0 to 39 points. The cutoffs are as follows: 0–4 = absent or minimal depression, 5–7 = mild depression, 8–15 = moderate depression, and 16 or higher = severe depression. The correlation between the short and long form is r = 0.96.
The eating disorder inventory (EDI-2)
This is the Spanish adaptation by Garner 24 of an eating disorder symptom assessment that addresses eleven clinically relevant dimensions. The self-administered questionnaire consists of 91 items answered on a six-point Likert scale (0 = “never” to 5 = “always”). The 11 subscales of the EDI-2 are as follows: drive for thinness (DT); bulimia (B); body dissatisfaction (BD); ineffectiveness (I); perfectionism (P); interpersonal distrust (ID); interoceptive awareness (IA); fear of maturity (MF); asceticism (A); impulse regulation (IR), and social insecurity (SI). The instrument reliability for this adaptation by Garner as evaluated through Cronbach's alpha 25 ranges from 0.83 to 0.93.
Statistical analyses
Statistical analyses were conducted using IBM SPSS Statistics 19 software (SPSS, Inc., Chicago, IL). Descriptive statistics were expressed as frequencies, percentages, means, and standard deviations (SD) for variables as appropriate. Pearson's correlation was used to determine associations between age, weight, height, BMI, eating disorder symptoms, and symptoms of anxiety and depression. The χ2 and t-tests were used to determine differences in sociodemographic and anthropometric characteristics, symptoms associated with eating disorders, anxiety and depression by gender, and class of obesity. Multiple linear regression analysis was used to determine which of the symptoms associated with eating disorders contributed significantly to anxiety and depression symptoms. The level of statistical significance was set at p < 0.05.
Results
Participant characteristics
The 546 patients averaged 42.4 (SD = 10.05) years of age. There were 358 women (65.6%) with an average age of 42.71 (SD = 10.50) years, and there were 188 men (34.4%) whose average age was 41.80 (SD = 9.13) years.
The majority of the patients were married and lived with their families, had completed either primary or secondary education, were actively employed and earned average monthly incomes between 500 and 2000 Euros (Table 1).
χ2 test.
t-test.
Calculated with the total women (N = 358).
DT, drive for thinness; STAI-E, state anxiety subscale; STAI-R, trait anxiety subscale; OSA, obstructive sleep apnea; SD, standard deviation; BMI, body mass index; B, bulimia; BD, body dissatisfaction; I, ineffectiveness; P, perfectionism; ID, interpersonal distrust; IA, interoceptive awareness; MF, fear of maturity; A, asceticism; IR, impulse regulation; SI, social insecurity; BDI, depression inventory by Beck.
Bold indicates statistically significant p < 0.05.
Analysis of sociodemographic characteristics by gender revealed an association between employment situation (χ2 = 48.831; df = 5; p = < 0.001), living arrangements (χ2 = 12.065; df = 3; p = 0.007), and marital status (χ2 = 11.096; df = 3; p = 0.011), and in the morbidly obese and super obese groups, an association with monthly income (χ2 = 24.915; df = 4; p = < 0.001) and employment situation was revealed (χ2 = 16.302; df = 5; p = 0.006) (Table 1).
Patients had an average height of 165.2 (SD = 9.7) centimeters, average weight of 135.3 (SD = 25) kilograms, and average BMI of 49.52 (SD = 8.3) kg/m2.
Three hundred twenty-seven (59.89%) patients were categorized as morbidly obese (Class III; BMI greater than 40 kg/m2), and 219 (40.1%) patients were categorized as super obese (Class IV, BMI greater than 50 kg/m2). Male and female patients had statistically significant differences in average weight (t = 12.439; df = 544; p = < 0.001) and height (t = 21.270; df = 544; p = < 0.001). The morbidly obese and super obese groups also had statistically significant differences in weight (t = −17.233; df = 544; p = < 0.001) and BMI (t = −24.032; df = 544; p = < 0.001) (Table 1).
The most frequent comorbidities affecting our patients were OSA syndrome (43.6%), hypertension (38.3%), dyslipidemia (28.2%), diabetes mellitus (25.1%), osteoarthritis (24.9%), and nonalcoholic fatty liver disease (23.3%). Some women (16.6%) reported amenorrhea or infertility, and another 13.7% had polycystic ovary syndrome. Symptoms of anxiety and depression affected 26% and 20.9% of patients, respectively. There were significantly different differences in the gender scores for self-reported hyperuricemia (χ2 = 24.656; df = 1; p = < 0.001), osteoarthritis (χ2 = 8.293; df = 1; p = 0.004), OSA (χ2 = 38.254; df = 1; p = < 0.001), depression (χ2 = 11.425; df = 1; p = 0.001), anxiety (χ2 = 10.649; df = 1; p = 0.001), and physical/sexual abuse (χ2 = 7.243; df = 1; p = 0.007). There were also significant differences between the reported scores of the morbidly obese and super obese patients for dyslipidemia (χ2 = 7.927; df = 1; p = 0.005), OSA (χ2 = 4.875; df = 1; p = 0.027), depression (χ2 = 8.694; df = 1; p = 0.003), and anxiety (χ2 = 5.664; df = 1; p = 0.017) (Table 1).
Anxiety and depression
Women scored slightly higher in reported state and trait anxiety. The gender difference was not statistically significant for state anxiety, but was statistically significant for trait anxiety (t = −2.985; df = 539; p = 0.003) (Table 1).
The average BDI score for all patients was 7.49 (SD = 5.3), between the scores for mild depression (5–7) and moderate depression (8–15). Overall, 33.7% of patients had no or minimal depression (BDI score 0–4), 24.5% reported mild depression (5–7), 33.9% had moderate depression (8–15), and 7.9% had severe depression. We found no significant differences between men and women or between the morbidly obese and super obese groups (Table 1).
Symptoms related to eating disorders
An overview of EDI-2 scores reported by gender, morbidly obese and super obese subgroups, and overall group are shown in Table 1.
Women scored significantly higher on the following EDI subscales: DT (t = −3.094; df = 543; p = 0.002), BD (t = −4.156; df = 544; p = < 0.001), and I (t = −2.717; df = 543; p = 0.003). Men scored higher score in the P subcategory (t = 2.962; df = 539; p = 0.003).
Relationship between anxiety and depression with symptoms related to eating disorders
The correlations between the dimensions of the EDI-2 and the BDI and STAI-S/STAI-T were positive and statistically significant (p < 0.01). However, a correlation between age, weight, height, and BMI and anxiety and depression disorders or symptoms associated with the eating disorder was not observed (Table 2).
p < 0.01.
p < 0.05.
W, weight; H, height; A–E, state anxiety subscale; A–R, trait anxiety subscale.
Gray shading represents a relationship between the symptoms of anxiety and depression and symptoms associated with the eating disorder.
For multiple linear regression analysis, dependent variables included the BDI, STAI-S, and STAI-T, and the EDI-2 dimensions were possible predictor variables. The EDI-2 variables ineffectiveness (I) (β = 0.479), IR (β = 0.279), DT (β = 0.089), SI (β = 0.164), and bulimia (B) (β = 0.132) were all significantly associated with symptoms of depression (R2 = 0.513, F = 112.342, p < 0.001).
The variables ineffectiveness (I) (β = 0.874), IA (β = 0.232), SI (β = 0.378), and bulimia (B) (β = 0.319) were significantly associated with State Anxiety (R2 = 0.344, F = 68.946, p = < 0.001).
In addition, the variables ineffectiveness (I) (β = 0.960), IA (β = 0.385), SI (β = 0.484), DT (β = 0.217), and IR (β = 0.335) were significantly associated with Trait Anxiety (R2 = 0.558, F = 134.098, p = < 0.001) (Table 3).
Discussion
This study analyzed medical and psychosocial variables in obese patient candidates for BS and aimed to establish a relationship between these variables and symptoms of anxiety and depression. The results suggest that the sociodemographic profile of the respondents is consistent with that reported in the literature,1–4,25–27 although there was a greater percentage of women (65.6%) and morbidly obese patients (59.9%) in this sample. We found a relationship between gender and employment situation, living arrangements and marital status, as well as between morbidly obese and super obese groups with monthly income and employment situation. A higher percentage of women worked at home, earned between 500 and 1000 euros, and were widowed. We also found that a higher percentage of morbidly obese patients earned >2000 euros and that a higher percentage of super obese patients worked at home and were unemployed. Although few studies have analyzed the differences in sociodemographic characteristics among gender and morbidly obese and super obese groups, similar results have been obtained in other studies4,25 (Table 1). Morbid obesity leads to social and occupational isolation, as well as the development of severe comorbidity. 27 It therefore follows that obese individuals feel rejected by society and may develop serious emotional and psychological problems. BS usually leads to a significant improvement of psychopathological parameters as patients lose the social stigma associated with weight loss and enjoy improved social recognition. This leads to less anxiety and fear of social interactions, as well as the emergence of new positive plans for the future.11,28
In this study, we found differences in weight and height according to gender, consistent with results of systematic reviews previously conducted in patients with morbid obesity.3,4
The results of the STAI indicated statistically significant gender differences in trait anxiety, but not in Trait Anxiety. These results agree with the assumption that men tend to deny Trait Anxiety, which naturally results in higher scores for women than for men. 22 BDI scores using the cutoffs established by Conde and Franch 23 revealed an average score of 7.49 (SD = 5.3), just between the scores for mild and moderate depression, with no significant differences between genders and morbidly obese and super obese groups.
Patients in our study presented an EDI-2 profile very similar to that found in the literature. 26 Accordingly, we found statistically significant differences between men and women in the EDI-2 subscales DT, BD, I, and P, similar to the results obtained in the Spanish questionnaire adaptation. 24 The authors found that men and women reflected different personal, physical, and social attitudes toward eating behaviors. In addition, women had significantly greater DT, BD, I, and IA scores, while males had only slightly higher scores in the P subcategory. Statistically significant differences were found between morbidly obese and super obese groups only in the BD subscale.
In our sample, we found very high DT and BD scores compared to the general Spanish population. The subscale scores for B, I, ID, IA, MF, A, and SI were very similar, and the P and IR subscale scores were lowest compared to the general Spanish population. 24
Similarly, Garner, Olmsted, and Polivy 29 applied the EDI to a group of obese women and another group of formerly obese women and found that the currently obese scored higher than the original nonclinical group in the DT, BD, and B scales.
Excess weight is a major cause of comorbidities that can lead to further morbidity and mortality. In our study, male patients experienced more hyperuricemia and OSA, while female patients experienced more osteoarthritis, anxiety, depression, and physical/sexual abuse. These results are similar to those of another study 30 that showed a positive relationship between overweight and obesity states and glucose intolerance, dyslipidemia, type 2 diabetes, hypertension, osteoarthritis, kidney failure, cardiovascular diseases (atrial fibrillation and heart failure), respiratory problems (asthma), and severe mental disorders (bipolar disorder and schizophrenia). In addition, we found that higher BMI was associated with depression. 30 It is therefore important for health professionals to help control and manage obesity to better control and prevent the numerous comorbidities that may affect mortality and quality of life.
Our patient rejection rate was at the lower end of the limits described by Sarwer et al., 11 who cited a rejection rate due to psychiatric reasons of between 3% and 20%. This is likely because most of our patients already underwent a prior mental health assessment in their local health zone. Nonetheless, all patients were reassessed at the Multidisciplinary Unit of Healthcare for Obese Patients with a specific protocol to harmonize the evaluation of patients from all health zones. The low percentage of excluded patients may also be explained by the low number of patients with psychoactive substance abuse screened for our study. Substance abuse is one of the primary reasons for exclusion from studies of this nature.9–11 In addition, prior stabilization of psychiatric disorders by mental health services helped to avoid contraindications to BS and exclusion from our study.
Numerous studies have been conducted to predict weight loss in heterogeneous groups of severely obese patients.3,26,28 However, very few of these studies have linked the symptoms associated with eating disorders to anxiety and depression. In this study, we found a positive and statistically significant relationship between these variables. Moreover, it was demonstrated through multiple linear regression analysis that the factors that best predicted depression were ineffectiveness, IR, DT, SI, and bulimia. The best predictors of State Anxiety were ineffectiveness, IA, SI, and bulimia. Finally, the best predictors of Trait Anxiety were ineffectiveness, IA, SI, DT, and IR.
The findings of the present study indicate that morbidly obese patients with symptoms related to eating disorders are likely to have high anxiety and symptoms of depression. Therefore, interventions directed toward symptoms of eating disorders will likely benefit these patients. Improved knowledge of these interventions could contribute to improved treatment and prevention of anxiety and depression, which, as many studies have shown,3,26,28,30 can contribute to enhanced weight loss after BS.
The use of an incidental sample, although representative, may be considered a limitation of this study. Furthermore, we propose a future study involving multivariate analysis and structural equation modeling with the variables studied to better understand the factors that affect weight loss after BS.
Future research should take pre- and postoperative weight loss into consideration when investigating the effects of intervention programs on symptoms associated with eating disorders to improve symptoms of anxiety and depression.
Footnotes
Author Disclosure Statement
No competing financial interests exist.