Bariatric surgery in patients with psychiatric comorbidity: Significant weight loss and improvement of physical quality of life

Summary Background Patients that have psychiatric comorbidity are thought to lose less weight than the general bariatric population and are therefore sometimes denied surgery. However, there is no scientific evidence for this assumption. The aim of this study is to evaluate the weight loss and health‐related quality of life (HRQoL) in patients with psychiatric disorders who undergo bariatric surgery and compare these patients with a general bariatric population. Method Patients who underwent bariatric surgery in 2015 were included. Patients who received individual counselling and had a current DSM IV axis 1 or 2 diagnosis were included in the psychiatric group (n = 163), all other patients in the generic group (n = 2362).Weight and HRQoL were assessed before and 12‐, 24‐, 36‐ and 48‐months after surgery. Data was analysed using regression analyses. Results The maximum total weight loss (TWL) was 27.4% in the psychiatric group vs 31.0% in the generic group. Difference in %TWL between the psychiatric and generic group was significant from baseline to all follow‐up moments (P < .001). Improvement of PHS was significantly higher in the generic group from baseline to 12‐month (P = .002), 24‐month (P = .0018), 36‐month (P = .025) and 48‐monthfollow‐up (P = .003). Change in mental HRQoL was only different comparing baseline to 48‐monthfollow‐up (P = .014). Conclusion Although weight loss and change in physical HRQoL was lower in patients with pre‐operative psychiatric disorders, results of this group were still excellent. Thus, patients with psychiatric diagnoses benefit greatly from bariatric surgery and these patients should not be denied weight loss surgery.


| BACKGROUND
Bariatric surgery is the most effective treatment for morbid obesity, resulting in weight loss, a reduction or resolution of comorbid conditions like type 2 diabetes, and improvement of health-related quality of life (HRQoL). [1][2][3][4][5][6] However, not all patients benefit equally from bariatric surgery. Mental health problems are thought to be indicative of a worse prognosis in terms of weight loss, since such problems could negatively influence post-operative eating behaviour and compliance. [7][8][9][10] Furthermore, there is fear of possible worsening of psychological issues after surgery. 10,11 According to the European guidelines on metabolic and bariatric surgery, psychopathology, such as severe depressions and personality disorders, is therefore considered a contra-indication for bariatric surgery. 1 Given that the prevalence of psychiatric disorders is significantly higher in patients with morbid obesity than in the general population, many patients are rejected to undergo bariatric surgery. Among patients presenting for bariatric surgery 27.3% to 41.8% had an Diagnostic and Statistical Manual of Mental Disorders (DSM) IV reported axis 1 disorder (a clinical disorder) and 22% to 24% was diagnosed with an axis 2 disorder (a personality disorder). 2,7,[12][13][14] Previous systemic reviews and a meta-analysis have attempted to demonstrate an association between psychiatric disorders and weight loss. No evidence was found that patients with pre-operative psychiatric disorders had significantly less post-operative weight loss. However, the results from the studies included in these reviews are inconsistent and sometimes show conflicting results. Furthermore, the included studies yielded evidence, which was of moderate quality, partly because the psychiatric disorders were not well-defined. 2,14,15 In addition, weight loss is not the only indicator of success after bariatric surgery: HRQoL is considered another primary outcome after bariatric surgery. 16 Although HRQoL is generally lower in patients with mental health problems, there are indications that weight loss can also reduce psychiatric symptoms in these patients and improve the overall HRQoL. 17,18 Contrary to this, it was suggested that patients with psychiatric comorbidity might have lower HRQoL after bariatric surgery. 14 Since there is a high population of patients with extreme obesity and psychiatric disorders who are at risk of developing other severe comorbidities if they remain untreated, further research is needed. In this study, we will assess a large population of patients with psychiatric disorders who have undergone bariatric surgery, focussing on both weight loss and HRQoL after surgery. The aim will be to assess 4-year weight loss and HRQoL in the total psychiatric population. Second, outcomes of four groups of psychiatric patients will be compared: patients with an axis 1 disorder, patients with an axis 2 disorder, patients with an axis 1 and 2 disorder and patients with cognitive impairment. Third, the results will be compared to the 'general' bariatric population.

| Standard treatment
All patients included in this study were being treated at the Nederlandse Obesitas Kliniek (NOK, Dutch Obesity Clinic). Treatment at the NOK involves an extensive pre-and post-operative counselling program in addition to the bariatric procedure. The counselling is conducted by an interdisciplinary team, consisting of a medical doctor, psychologist, dietician and physical therapist. 19 All patients are screened according to the international criteria for bariatric surgery, the IFSO-criteria, by the interdisciplinary team and subsequently assigned to standard (group) or individual counselling, only after being accounted mentally stable enough. 1 Most common reasons for patients to receive individual counselling are language barriers, previous bariatric surgical treatment and severe psychiatric comorbid conditions.

| Patient and data selection
Patients were selected from a prospective database, provided that they underwent bariatric surgery in 2015. Data was collected up to April 2020. The reason for individual counselling was assessed through screening of the electronic patient record. Only patients who received individual counselling and had a current psychiatric diagnosis or reported a cognitive impairment were included for analysis in the 'psychiatric group'. All patients who underwent bariatric surgery in the same year who received standard (group-) counselling were included in the 'generic group'.

| Psychiatric diagnosis
To define the psychiatric diagnosis the following variables were selected from the electronic patient record: reason for individual counselling, past and current psychiatric diagnosis and/or cognitive impairment and the use of psychotropic drugs.
The diagnoses were than grouped into DSM IV axis 1 and axis 2 disorders: clinical disorders vs personality disorders respectively. All patients with current psychiatric disorders on axis 1 and/or axis 2 were assigned to respectively axis 1 group, axis 2 group and axis 1 + 2 group. Although cognitive disorders are strictly taken a subgroup of the DSM IV axis 2, previous investigation has demonstrated that cognitively impaired patients show slightly deviant behaviour. [20][21][22] Therefore, patients who had no known psychiatric disorder but reported to be cognitively impaired (IQ was not formally assessed in the NOK), were assigned to a fourth group: the cognitive impairment (CI) group.
Patients who were prescribed a psychotropic drug other than minor tranquillizers such as benzodiazepines, but had no known psychiatric diagnosis were also considered to have a current diagnosis on axis 1. Mental health problems requiring psychiatric care were also included. Moreover, patients with personality traits were also included in the axis 2 group.
At the NOK, eating disorders are considered a contra-indication for bariatric surgery and therefore not included in this analysis.

| Body weight and other parameters
Body weight was assessed at pre-operative screening and 12-, 24-, 36-and 48-months after surgery. Height was assessed during preoperative screening. Body mass index (BMI in kg/m 2 ) and percentage of total weight loss (%TWL) were calculated for baseline (only BMI) and follow-up. The following parameters were also registered: age, gender, type of surgery and whether this was primary or secondary procedure.

| Health-related quality of life
The perceived HRQoL was evaluated at pre-operative screening and 12-, 24-, 36-and 48-months after surgery using the RAND-36 questionnaire. The RAND-36 is one of the most used questionnaires to assess HRQoL in bariatric patients. 16 It consists of 36 questions and 9 scales and can be used to calculate two subscales: physical health summary (PHS) and mental health summary (MHS). Scores range from 0 to 100, where a higher score represents a higher HRQoL.

| Statistical analysis
Continuous variables were visually inspected and tested for normality by the Shapiro-Wilk test. Descriptive statistics summarized baseline patients' characteristics. Differences in baseline characteristics between the psychiatric and generic group were assessed with t-test and chi-square test. Subsequently patients were divided in the four psychiatric groups: axis 1 group, axis 2 group, axis 1 + 2 group and CI group. Differences between the groups regarding baseline characteristics were analysed with one-way ANOVA. These analyses were performed using SPSS (version 24) statistical software.
Then a linear mixed model was conducted to assess how %TWL changed from baseline to 12, 24, 36 and 48 months in the psychiatric group. First, the change of %TWL over all follow-up moments was assessed with random intercept, thereby the model considers different intercepts for each patient. Then type of surgery, gender, baseline BMI and age were added to the model as fixed effects. In the last part of the model, the differences within the psychiatric groups were assessed (effect modification of the four groups). In a second mixed model, the differences between the psychiatric and generic group were studied. The same model was performed for the changes in the two RAND subtotal scores: the PHS and MHS.
All assumptions for regression analysis were met. These analyses were performed using STATA, version 13 (StataCorp. 2013. Stata 13 Base Reference Manual. College Station, Texas: Stata Press). Findings were considered statistically significant if the P-value was <.05.

| Study population
A total of 163 patients were included in the psychiatric group and 2362 in the generic group. In the psychiatric group, there were significantly less females (66.9% vs 79.5%, P < .001). There were also less patients who underwent a primary RYGB (61.3% vs 76.0%, P < .001).
Mean age and baseline BMI were not significantly different between these groups.

| Distribution of psychiatric disorders
In the psychiatric group 58.6% had a psychiatric disorder on axis 1; 6.8% had a disorder on axis 2; 16.7% had diagnoses on both axis 1 and 2; 17.9% presented with a cognitive impairment. The most common clinical disorder was a mood disorder (54.9%), with a prevalence of 63.2% in the axis 1 group and a prevalence of 48.1% in the axis 1 + 2 group (Table 1). Most common personality disorder was a cluster B disorder (9.0%), with a prevalence of 44.4% in the axis 1 + 2 group; whereas the axis 2 group had no prevalence of cluster B disorders.

| Weight loss in psychiatric group
In the psychiatric group, mean BMI was 31.9 kg/m 2 (±6.

| Psychiatric disorders and HRQoL
There were no differences between the psychiatric groups in the change of PHS at any of the follow-up moments (Table 3). Change in MHS scores at 12 months differed significantly between the axis 1 and axis 2 group (P = .018) and between axis 1 + 2 and axis 2 group at 12 months (P = .006). Three large meta-analysis concluded that there was no difference in weight when comparing patients with and with psychiatric comorbidity. 2,14,15 This study contradicts these results, since a significant difference was found between the psychiatric group and the 'general'

| HRQoL: psychiatric vs generic group
bariatric population. However, the 5% lower weight loss in our study should not be the reason to deny patients with psychiatric comorbidity bariatric surgery. Especially since previous studies have shown that T A B L E 3 RAND-36 scores in the psychiatric group, for each of the separate psychiatric disorders, presented as mean ± SD Abbreviations: BL, baseline; %TWL, % total weight loss; 12 M, 12 months follow-up; 24 M, 24 months follow-up; 36 M, 36 months follow-up; 48 M, 48 months follow-up. α = significant difference compared to axis 2 group, P < .05; β = significant difference compared to axis 1 + 2 group, P < .05. even a 5% reduction of body weight in patients with obesity can already lead to great health improvements, like lower cholesterol levels, decreased blood pressure and improved beta cell function and insulin sensitivity. 23,24 In addition, a TWL above 20% has previously been described a successful result. The average TWL of 20.9% after 4 years in the psychiatric group is therefore considered sufficient. 23 In our study, post-operative weight loss seems to be different between the psychiatric subgroups. Previous studies could not relate axis 1 disturbances, like depression and anxiety, to lower weight loss. 2,14,15,25 In our study however, patients with an axis 1 diagnosis, which were mostly anxiety and depressive disorders, had the lowest weight loss (16.8% after 4 years) of all four groups. It could be that in our group the axis 1 disorders were more serious, since patients with mild axis 1 disorders are mostly treated in group counselling.
Surprisingly, weight loss was very high in patients with an axis 2 disorder, TWL was 30.0%, while other studies showed a negative effect of personality disorders. 14,15 This might be explained by the fact that only specific personality disorders seem to interfere with weight loss. 26 Moreover, patients in this study had a strict multidisciplinary program before and after surgery, which can be a reason for improved weight loss results. However, the current results might also be explained by the relatively small number of patients and low follow-up rate in the axis 2 group, with only three patients left after 4 years of follow-up. As a result, it remains difficult to draw conclusions for this subgroup.
Another interesting finding was that patients with a cognitive impairment demonstrated relatively high weight loss, which was significantly better than the axis 1 group and almost the same as the generic group. Possibly psychotropic medication, which is only prescribed to patients with psychiatric disorders and not to patients who are solely cognitively impaired, might limit weight loss. In addition, the NOK provides structured care and demands that patients have a wellfunctioning support system involved in the care program. This might also explain that the supposed deficits in executive function and memory in these patients do not limit their weight loss as much as one might expect.
Physical HRQoL improved in all psychiatric patients, with no differences between the groups. Even though actual scores were higher in the generic group, there were no differences between the psychiatric and generic patients when comparing changes between follow-up moments. Therefore, the changes in physical HRQoL after surgery were comparable between the psychiatric group and the generic group.
In the psychiatric group, mental HRQoL only improved up to 12 months and then decreased. At 4-yearfollow-up, the mean score was 3.5 point below the MHS score before surgery. It seems that the improvement in mental HRQoL in the psychiatric group is only temporarily. In the generic group mental HRQoL also declined after Our study has some limitations. First, this is a retrospective study and therefore the main limitation is the assessment of psychiatric diagnosis using the electronic patient record. Second, only patients receiving individual counselling were assessed for inclusion in the psychiatric group. As a consequence, it is possible that patients assigned to the generic group could have also suffered from a (less severe) psychiatric disorder. Third, there was a risk of selection bias as patients who suffered from severe mental health problems, and therefore considered not psychiatrically stable, were not operated and therefore not included. Finally, 48-months compliance was 48.5%, this is a known problem in bariatric patients. This is in concordance with previous research, but by conducting a mixed model analysis, we attempted to limit study confounding. 27 However, since this low compliance especially resulted in low numbers of patients in the different groups, we cannot draw conclusions when comparing the groups.

| CONCLUSION
The current European guideline on bariatric surgery considers non-stabilized psychotic disorders, severe depression and personality disorders to be a contra-indication, unless specifically advised by a psychiatrist experienced in obesity. 1 Based on our findings, it might be too early to update the guideline, since we do not have knowledge about the risk of worsening of psychiatric symptoms after the surgery.
However, our findings show that patients with psychiatric disorders should not be excluded from surgery as they show significant weight loss and improvement of physical HRQoL. Given that the health risks related to morbid obesity are severe, physicians should carefully asses a patient before deciding to reject a candidate on mental health grounds. A structured, multidisciplinary perioperative counselling program for this group of patients seems to improve the success of the intervention.

ETHICS STATEMENT
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
For this type of study, formal consent is not required.

DATA AVAILABILITY STATEMENT
The data are not publicly available due to privacy restrictions, but are available from the author (V. M. M.) on reasonable request.