Prevalence and Outcomes of Depression After Bariatric Surgery: A Systematic Review and Meta-Analysis.

Surgeons often focus on weight loss and improvement of obesity-related conditions as a primary outcome after bariatric surgery. However, the success of bariatric surgery also relies on the improvement of mental health status. Therefore, the current meta-analysis was carried out to reveal the prevalence of depressive symptoms and their subsequent impact on bariatric surgery outcomes. This study was performed following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist and the recommendation of Cochrane Collaboration. All clinical studies reporting the prevalence and/or the outcomes of depression after bariatric surgery were included in the current meta-analysis. This meta-analysis encompassed 33 articles, including a total of 101,223 patients. The prevalence of post-bariatric surgery depression was 15.3% (95% confidence intervals {CI}: 15.0-15.5%, p<0.001) among which severe, moderate, and minimal depression accounted for 1.9% (95% CI: 1.5-2.4%, p<0.001), 5.1% (95% CI: 4.4-5.8%, p<0.001), and 64.9% (95% CI: 63.3-66.5%, p<0.001), respectively. Depression is negatively correlated with weight loss (correlation -0.135; 95% CI: -0.176 to 0.093; p<0.001) and positively correlated with eating disorder (correlation 0.164; 95% CI: 0.079-0.248; p<0.001). The prevalence of post-bariatric surgery depression is relatively high reaching up to 64.9%, with almost one in five patients affected by it. Depression is associated with weight regain, eating disorders, and quality of life.

Introduction and background

Obesity is a complex health problem with a growing incidence worldwide [1]. To date, approximately 1.9 billion and 610 million adults are considered overweight and obese, respectively, representing nearly 39% of the general population [2,3]. Obesity negatively impacts all physical and mental aspects of the body. It leads to cardiovascular insufficiency, metabolic syndrome, hepatobiliary diseases, respiratory disorders, osteoarthritis, infertility, and cancer. Besides that, obesity might be associated with anxiety, low self-esteem, depression, and impaired quality of life (QoL) [4-6]. These significant consequences limit the patients’ performance, decrease their chances of getting a job due to physical appearance, increase their absenteeism frequency, and enhance isolation and addiction risks [7,8]. Obese patients are nearly 55% more vulnerable to experience depressive symptoms than the non-obese population. Furthermore, approximately 45% of bariatric surgery seekers present with depression [9,10].

A number of modalities have been proposed for treatment of obesity. Bariatric surgery is considered the safest and the most effective procedure for weight reduction, which reduces obesity-related comorbidities and improves survival [11-13]. Surgeons often focus on weight loss and improvement of obesity-related conditions as a primary outcome after bariatric surgery [14]. However, it has been widely accepted that success after bariatric surgery depends not only on weight loss but also on the improvement of mental health status [9]. While most patients show improved psychological state after bariatric surgery, a considerable proportion experience persistent psychological concerns and even worsening manifestations [15].

Patients undergoing bariatric surgery are associated with a fourfold increase in the risk of attempted suicide as compared to the general community [16,17]. Assessment of post-bariatric psychological outcomes is critical to identify morbidly obese patients who require further supportive treatment [18]. A deeper insight into the mental state of the patients undergoing bariatric surgery can contribute to a more comprehensive understanding of and identify patients at a higher risk of post-operative depression [19].

The prevalence and subsequent outcomes of depression after bariatric surgery are still unclear in the literature [20-22]. Previous investigations have focused mainly on pre-operative depression, and little is known about the impact of depression after undergoing bariatric surgery [23]. Identifying the relationship between depression and success of bariatric surgery is critical, considering that inadequate weight loss after surgery might lead to the re-emergence of obesity and its associated complications, thereby impairing the patient’s QoL [24]. Furthermore, this knowledge will help healthcare providers to identify patients at risk and employ timely and appropriate management of depression after bariatric surgery to prevent its potential long-term consequences. Therefore, the current systematic review and meta-analysis were carried out to reveal the prevalence of depressive symptoms and their subsequent effects on the short-term and long-term outcomes of bariatric surgery.

Review

Methods

This systematic review and meta-analysis was carried out following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [25]. An extensive systematic review of literature up to October 17, 2020, was implemented using the following databases: PubMed, Google Scholar, Web of Science (ISI), Scopus, SIGLE, Virtual Health Library (VHL), NYAM, ClinicalTrials, metaRegister of Controlled Trials (mRCT), Embase, and WHO International Clinical Trials Registry Platform. No restrictions were set in terms of patients’ age, sex, ethnicity, language, race, or place. The following keywords were used in every possible combination: “bariatric,” “sleeve,” “gastric bypass,” “gastric band,” “duodenal switch,” “depression,” and “depressive.” A further manual search was performed to comprehend all retrieved studies’ references to distinguish all additional relevant articles that were not indexed. The cross-referencing method was carried out until no other relevant article was detected.

Study selection

All clinical studies that reported the prevalence and/or the outcomes of depression after bariatric surgery were included in the current meta-analysis. This includes studies comparing the outcomes of depressed and non-depressed patients after the surgery and also single-arm studies that reported the association between depression and bariatric surgery outcomes. There were no restrictions on the patients’ age, sex, race, or place. In contrast, studies that did not report an association between depression and surgery outcomes were excluded. Furthermore, studies in which data could not be extracted, such as guidelines, review articles, animal studies, case reports, comments, letters, editorials, posters, and book chapters, were excluded without adding any restriction on langauge. The screening process of the title, abstract, and full text was performed independently to reveal potentially relevant articles that met the inclusion criteria. Discussions were carried out to resolve contradictions among reviewers.

Data extraction and quality assessment

The following data were extracted from the finalized included articles: study characteristics (the title of the included study, the second name of the first author, year of publication, study design, study period, study region, and sample size), patients’ demographic characteristics (age, sex, weight, height, body mass index {BMI}, occupation, comorbidities, family history of psychiatric illness, and pre-operative psychological status), bariatric surgery-related data (the type of surgery, initial weight loss, and intra-operative and post-operative complications), post-operative psychological status (depression screening tools, duration of the current episode, number of depressed patients, number of suicide attempts, QoL scores, and post-operative depression score), and psychological outcomes (number of depressed patients, the correlation between post-operative depression and eating disorders, weight loss, body image, regained BMI, BMI loss, and mental and physical components of QoL).

The quality of the observational studies was assessed using the National Institute of Health quality assessment tool [26]. The studies were grouped based on the quality assessment into good (quality score >65%), fair (quality score 30-65%), and bad (quality score <30%). If the parameter was controlled, the domain was considered “yes” and vice versa.

Statistical analysis

The prevalence of depression was estimated by calculating the event rate with 95% confidence intervals (CIs) for each study, followed by pooling the effect sizes of all studies to estimate the summary proportion with 95% CIs. The summary correlation and 95% CIs were computed by pooling the correlation and sample size of each relevant article. The fixed-effect model was implemented when a fixed population effect size was assumed; otherwise, the random-effects model was used. Statistical heterogeneity was appreciated using Higgins I2 statistic, at the value of >50%, and the Cochran's Q (chi-square test), at the value of p<0.10 [27]. To account for heterogeneity, the random-effects model was employed. Publication bias was assumed in the presence of an asymmetrical funnel plot and based on Egger’s regression test (p<0.10). Herein, the trim and fill method of Duval and Tweedie was used [28]. Subgroup analysis was conducted based on the severity of depressive manifestations. Data analysis was performed using the Review Manager version 5.3 (Copenhagen, Denmark: The Nordic Cochrane Centre, The Cochrane Collaboration) and Comprehensive Meta-Analysis software version 2 [29,30]. The significant difference was established at the value of p<0.05.

Results

A comprehensive systematic literature search yielded a total of 738 articles. Using EndNote X9 (London, UK: Clarivate), 347 duplicates were removed, yielding 391 articles eligible for title and abstract screening. Of these studies, 46 articles were suitable for full-text screening, and 37 articles were included for data extraction. Out of them, five studies were excluded due to overlapping data. Herein, 32 articles were included for systematic review and meta-analysis in addition to one study identified through manual search. A flow diagram illustrating the process of literature search is shown in Figure 1.

Figure 1

An illustration of the process of literature search.

Study characteristics

This meta-analysis encompassed 33 articles, including a total of 101,223 patients. There were 76.33% (27,674/36,282) females. At the baseline, the mean age of the included patients ranged from 32.2 to 47.61 years. The mean BMI ranged from 42.02 to 51.8 kg/m2, and the average pre-operative depression score ranged from 7.7 to 20.2. The mean follow-up period ranged from six to 45.6 months. Of the included studies, three studies showed fair quality, and the remaining articles were of good quality [31-33]. The Funnel plot was found to be symmetrical, which indicates no publication bias (Table 1, Figure 2).

Table 1

Demographic characteristics of the included studies.

*Range

**Median and range

***Mean and range

BDE: Beck Depression Inventory; EDE-Q: Eating Disorder Examination-Questionnaire; BSQ: Body Shape Questionnaire, HADS: Hospital Anxiety and Depression Scale; HAM-D Scale: Hamilton Depression Scale; SF-36: Medical Outcomes Study Short Form-36 Health Survey; IPAQ: International Physical Activity Questionnaire; PHQ-9: Patient Health Questionnaire-9; RSES: Rosenberg Self-Esteem Scale; CESD: Center for Epidemiologic Studies Short Depression Scale; NR: non-reported

S. no.

Study ID

Study region

Study design

Study period

Sample size (number)

Gender (female) (number)

Age (mean± SD)

Type of Procedure

BMI (Mean± SD)

Psychological Assessment

Follow-up Period

Quality Assessment

Gastric banding (number)

Roux-en-Y gastric bypass (number)

Gastric Sleeve (Number)

Duodenal Switch (Number)

%

Decision

1.

Brandão et al., 2016 [34]

Portugal

Retrospective observational and cross-sectional study

January 2009 and June 2013

75

64

(63-64)*

47

19

9

0

44.75 (34.53-59.82)**

BDI, EDE-Q, BSQ

(18-46) month*

78%

Good

2.

Sousa et al., 2014 [35]

Portugal

Retrospective

NR

52

43

44.04 (10.87)

38

5

9

0

NR

BDI

(22-132) month*

75%

Good

3.

Andersen et al., 2010 [36]

Norway

Prospective cohort study

NR

50

28

37.9±7.9

0

0

0

50

NR

HADS>8

2 years

80%

Good

4.

de Zwaan et al., 2011 [37]

Germany

Prospective cohort study

NR

107

75

37.5±9.7

76

31

0

0

49.4±7.4

DSM-IV

(24-36) month*

80%

Good

5.

Freire et al., 2020 [38]

Brazil

Retrospective

1999 and 2004

96

75

40.2± 10.1

0

96

0

0

50±8.2

BDI

2 years

78%

Good

6.

Pinto et al., 2017 [39]

Brazil

Prospective cohort study

NR

60

51

34.7±9.2

NR

NR

NR

NR

46.04±7.52

BDI-SF>4

NR

67%

Good

7.

Nijamkin et al., 2013 [40]

USA

Prospective cohort study

NR

144

120

44.5±13.5

0

144

0

0

35.95±5.9

BDI-II

18 months

76%

Good

8.

Mitchell et al., 2014 [41]

USA

Randomized clinical trial

February 2006 and February 2009

2,146

1,685

46 (37.54)

539

1,507

NR

NR

45.9 (41.8-51.4)**

BDI

2 years

87%

Good

9.

Jans et al., 2018 [42]

Flemish

Randomized clinical trial

December 2012 until March 2016

54

NR

29.4±4.3

2

45

2

0

28.1±5.1

Dutch pregnancy-validated Edinburgh Depression Scale

45.6±29.9

85%

Good

10

Alabi et al., 2018 [43]

Mexico

Retrospective

January 2015 and January 2016

73

56

38.1±9.1

NR

NR

NR

NR

38.8±3.8

BDI-II

12 months

83%

Good

11

Nicolau et al., 2017 [44]

Spain

Retrospective

NR

60

47

45.5±9.4

NR

NR

NR

NR

48.4±7.6

BDI-II, SF-36 Health Survey Spanish version

46.48±18.1

75%

Good

12

Bressan et al., 2019 [33]

Brazil

Cross-sectional study

2015 and 2016

71

54

39.8±10.3

NR

NR

NR

NR

NR

BDI-II, Rosenberg Self-Esteem Scale

NR

55%

Fair

13.

Yuan et al., 2019 [45]

USA

Retrospective claims data from Aetna

2008 and 2016

64,090

NR

46.19±13.59

NR

NR

NR

NR

NR

BDI-II

748 days

88%

Good

14.

Osterhues et al., 2017 [10]

Germany

Randomized clinical trial

September 2015 and March 2016

103

80

43.30±11.69

NR

NR

NR

NR

NR

HADS ≥8

NR

68%

Good

15.

Booth et al., 2015 [46]

UK

A controlled interrupted time-series

January 1, 2000, to April 30, 2012

3,045

2,406

45.9±10.2

NR

NR

NR

NR

44±8.3

NR

(2-3)* years

78%

Good

16.

Elwan et al., 2014 [47]

Egypt

Prospective cohort study

January 2012 and June 2014

30

22

33.80±9.61

0

0

15

0

46.0±1.55

HAM-D Scale

19.56±6.92 month

81%

Good

17.

Lu et al., 2018 [48]

Taiwan

Retrospective from National Health Insurance Research Database of Taiwan

2001 to 2009

2,102

1,425

32.2±9.8

NR

NR

NR

NR

NR

NR

NR

68%

Good

18.

Timofte et al., 2018 [49]

Romania

Prospective cohort study

NR

7

3

NR

0

0

7

0

NR

Montgomery-Asberg Depression Rating Scale

NR

71%

Good

19.

Susmallian et al., 2019 [32]

Israel

Prospective, midterm follow-up study

January 2013 to December 2014

253

NR

41.65±11.05

0

0

253

0

42.02±5.03

NR

NR

59%

Fair

20.

Sivas et al., 2020 [50]

Turkey

Prospective cohort study

January 2016 and May 2017

27

23

37.1±10.4

NR

NR

NR

NR

46.2±5.2

BDI-II, IPAQ

NR

69%

Good

21.

Sait et al., 2019 [51]

Saudi Arabia

Cross-sectional study

July 2013 and July 2017

214

184

NR

0

32

177

0

NR

PHQ-9

NR

72%

Good

22.

Porcu et al., 2011 [52]

Brazil

Prospective cohort study

NR

50

NR

NR

NR

NR

NR

NR

NR

BDI, the Hospital Scale of Anxiety and Depression (I-TAD)

NR

45%

Fair

23.

White et al., 2015 [16]

USA

Prospective cohort study

NR

357

NR

43.7±10

NR

NR

NR

NR

51.2±8.3

BDI-II, EDEQ, Short Form-36 Health Survey

24 months

73%

Good

24.

Martens et al., 2020 [53]

USA

Prospective cohort study

2015-2017

1,991

1,573

47.61±11.63

0

324

1,667

0

47.42±8.04

Patient Health Questionnaire

NR

69%

Good

25.

Lu et al., 2019 [54]

USA

Prospective cohort study

NR

103

103

44.1±11.7

NR

NR

NR

NR

45.3±6.2

CESD short scale

NR

71%

Good

26.

Barzin et al., 2020 [55]

Iran

Prospective cohort study

March 2014 to March 2016

685

581

38.7±10.9

0

242

443

0

45.1±6.0

BDI-II

NR

75%

Good

27.

Lagerros et al., 2017 [56]

Sweden

Retrospective from National Health Insurance Research Database of Taiwan

2008 and 2012

22,539

16,961

41.3

0

22,539

0

0

NR

ICD-diagnoses F32-F33 forms

546 (2-730) days***

88%

Good

28.

Méa et al., 2017 [57]

Brazil

Cross-sectional observational study

NR

20

11

NR

NR

NR

NR

NR

NR

BDI-II

NR

71%

Good

29.

Matini et al., 2014 [58]

Iran

Prospective observational study

May 2012 to May 2013

67

55

36.8±8.5

NR

NR

NR

NR

48.8±4.7

HDRS

6 months

67%

Good

30.

Grilo et al., 2007 [59]

USA

Prospective cohort study

NR

137

NR

42.3±10.2

0

137

0

0

51.8±7.9

BDI-II, Short Form-36 Health Survey

12 months

75%

Good

31.

Smith et al., 2020 [60]

USA

Prospective cohort study

2006-2009

2,308

1,816

45.5±11.4

0

2,308

0

0

NR

BDI

3 years

67%

Good

33.

Ivezaj et al., 2014 [61]

USA

Prospective cohort study

NR

107

94

42.7±10.5

0

107

0

0

51.7±7.8

BDI, EDE-Q, SF-36, and RSES

12 months

75%

Good

Figure 2

Funnel plot to assess publication bias across prevalence studies.

A total of 27 articles, including 98,757 patients, reported the prevalence of post-bariatric depression. Pooling the data revealed a prevalence rate of 15.3% (95% CI: 15.0-15.5%, p<0.001) (Figure 3) [10,16,32,33,36-40,42-49,51-58,60,61]. Subgroup analysis among patients with depression revealed that prevalence of severe depression was 1.9% (95% CI: 1.5-2.4%, p<0.001). The prevalence of moderate depression was 5.1% (95% CI: 4.4-5.8%, p<0.001), whereas the prevalence of mild and minimal depression was 12.7% (95% CI: 11.8-13.7%, p<0.001), and 64.9% (95% CI: 63.3-66.5%, p<0.001) (Figure 4, panels A-D) [33,41,43,47,55,57,60].

Figure 3

Pooling of the prevalence of post-bariatric depression with subgroup analysis.

Pooling the data revealed a prevalence rate of 15.3% (95% CI: 15-15.5%, p<0.001).

Figure 4

Prevalence of post-bariatric surgery depression.

The image shows the prevalence rate of (A) minimal depression: 64.9% (95% CI: 63.3-66.5%, p<0.001); (B) mild depression: 12.7% (95% CI: 11.8-13.7%, p<0.001); (C) moderate depression: 5.1% (95% CI: 4.4-5.8%, p<0.001); (D) severe depression: 1.9% (95% CI, 1.5-2.4%, p<0.001).

Impact of depression on bariatric surgery outcomes

Weight Loss and BMI

The association between post-bariatric depression and weight loss was reported in three articles, including 2,173 patients. In the random-effects model (p=0.048, I2=67%), there was a statistically significant negative association between post-operative depression and weight loss (correlation -0.135; 95% CI: -0.176 to -0.093; p<0.001). Conversely, there was no statistically significant association between post-bariatric surgery depression and BMI (correlation 0.011; 95% CI: -0.093 to -0.115; p=0.836) (Figure 5, panels A and B) [16,34,35,37,53,61].

Figure 5

Correlation between depression and surgery outcomes.

(A) Weight loss, correlation: -0.135, 95% CI: -0.176 to -0.093, p<0.001; (B) BMI loss, correlation: 0.011, 95% CI: -0.093 to -0.115, p=0.836; (C) eating disorder, correlation: 0.164, 95% CI: 0.079-0.248, p<0.001; (D) quality of life (physical component), correlation: -0.010, 95% CI: -0.083 to 0.063, p=0.785; and (E) quality of life (mental component), correlation: 0.217, 95% CI: 0.145-0.286, p<0.001.

Eating Disorder

Three studies, including 516 cases, evaluated the correlation between post-bariatric depressive manifestations and eating disorders. There was a statistically significant positive association (correlation 0.164; 95% CI: 0.079-0.248; p<0.001) between post-operative depression and eating disorders in the random-effects model (p=0.109, I2=54%) (Figure 5, panel C) [16,35,61].

Quality of Life

The impact of post-bariatric surgery depression on the mental component of the QoL was assessed among 704 patients from four studies. In the random-effects model (p<0.001, I2=98%), pooling the effect sizes revealed a statistically significant association between post-bariatric depressive manifestations and mental component of QoL (correlation 0.217; 95% CI: 0.145-0.286; p<0.001). However, there was no statistically significant association between post-operative depression and the physical component of QoL (correlation -0.010; 95% CI: -0.083 to 0.063; p=0.785) (Figure 5, panels D and E) [10,16,50,59,61].

Discussion

Bariatric surgery procedures are associated with clinically significant weight reduction, improvements in overweight-related comorbidities, and prolonged life expectancy [62]. It positively affects the patients’ physical and mental aspects of life, including daily activities, social relationships, body image, eating behavior, sexual life, and mental health. There is relative individual variation in the weight reduction after surgery, and some patients might experience worsening of their psychological health status [63,64]. Despite the growing body of evidence related to bariatric surgery outcomes, there is limited literature regarding the impact of the surgery on psychological outcomes [15,65]. Therefore, this meta-analysis was performed to assess the prevalence of post-bariatric depressive manifestations and evaluate how these manifestations affect surgery outcomes.

Our systematic review and meta-analysis revealed that approximately one in every five patients who underwent bariatric surgery would experience depression at any interval within three years after surgery. The proportion of patients at risk to develop minimal depression after bariatric surgery was considerably high (more than 50% of bariatric surgery seekers). These findings are comparable with Courcoulas et al. who reported a decline of mild depression manifestations from 28% to 9.8% six months after surgery, followed by new rise to 12.2% and 15.6% in the second and third years after surgery, respectively [66]. In the short-term period, post-bariatric depressive manifestations might not have a significant impact on weight regain. Instead, initial weight reduction is related mainly to the bariatric surgery-induced metabolic changes rather than behavioral or psychological factors.

Most of the weight reduction occurs during the first year after bariatric surgery. This period of rapid weight loss is rewarding for patients to lose more weight. However, after this period, the weight loss plateaus, requiring patients to adopt overly restrictive and long-term nutritional and behavioral modifications to lose any additional weight [67]. The resultant loose skin and plateauing of body weight after rapid weight loss are associated with a high risk of body dissatisfaction [68]. These situations are accompanied by unrealistic expectations regarding rapid weight loss and body contouring, which puts the patients under more stress [69]. Patients at a higher risk of post-bariatric depression should be subjected to close monitoring. This includes exhaustive pre-operative assessment of depression and psychological disorders, along with employing timely and effective anti-depressive interventions [70]. This could enhance the effectiveness of the surgery, amplify weight reduction after surgery, and improve the long-term QoL. However, further studies with an adequate long-term period are required for comprehensively understanding the trajectory of depressive manifestations and weight regain after bariatric surgeries.

Identifying factors associated with long-term suboptimal weight loss in patients seeking bariatric surgery is of great importance to minimize the risk of revision surgery, psychological illness, and costs associated with suboptimal weight reduction [71,72]. In this meta-analysis, post-bariatric depression was associated with weight regain, eating disorders, and poor QoL. These results reinforce the close association between obesity and depressive manifestations, wherein both conditions could be dependent on each other [73,74]. In this regard, Geerts et al. reported that suboptimal weight loss after bariatric surgery was associated with impulsive eating, eating disorders, and depression [75]. Switzer et al. reported a strong association between rebound weight gain and depressive manifestations after bariatric surgery [76]. In a systematic review, Hindle et al. reported a significant association between early post-operative weight loss, eating adaptation, and later long-term weight loss. However, the evidence regarding the association between early post-operative psychological disturbance and later weight gain was not sufficient and inconstant to reach a definitive conclusion [23].

To the best of our knowledge, this is the first systematic review that gathered the rapidly emerging controversial evidence regarding the prevalence of post-bariatric surgery depression and its subsequent impact on the surgery outcomes. However, some limitations should be acknowledged. The majority of the included articles were of observational design, revealing a potential risk of selection bias. There was significant heterogeneity between the included studies. This heterogeneity might stem from different demographic characteristics, assessment methods, and surgical techniques. Due to the short follow-up periods, the long-term prevalence of depression and its impact on bariatric surgery could not be assessed.

Conclusions

The prevalence of post-bariatric surgery depression is high. Depression is associated with weight regain, eating disorders, and impaired QoL. The integration of these findings in healthcare protocols can help healthcare providers identify patients at a higher risk of depression and enhance bariatric surgery outcomes by stratifying the patients to the most appropriate and effective treatment in a timely fashion. However, further studies need to be conducted to tackle the limitations of the current meta-analysis.