Anxiety Predicts Reduced Weight Loss 30 Months after Bariatric Surgery

Abstract

Background:

Though psychosocial factors are routinely examined in pre-surgical psychological evaluations, the predictive value of some psychosocial factors on post-surgical weight loss is still relatively unknown. Additional research examining the predictive value of psychological constructs pre-operatively and long-term weight outcomes is needed to enhance the clinical utility of the pre-surgical psychological evaluations.

Objectives:

This study aimed to examine psychosocial factors as predictors of weight outcomes 30 months after bariatric surgery.

Setting:

University Hospital in the Appalachian region of United States.

Methods:

Participants included 196 adults who underwent bariatric surgery. Psychosocial data were collected as part of a routine psychological evaluation prior to bariatric surgery. Objective weight was obtained through patients’ medical record at 12, 24, and 30 months after surgery.

Results:

Linear mixed models was used to examine pre-surgical psychosocial factors predicting post-surgical weight loss (n at 12 months=153, n at 24 months= 130, n at 30 months= 92). Anxiety had a significant interaction effect with time (Est. = −0.01, p = 0.013), indicating that higher anxiety was associated with less weight loss over time. Those with severe anxiety prior to surgery lost the most weight 12 months after surgery, but also regained the most weight 30 months after surgery. Other predictors were not statistically significant.

Conclusions:

Anxiety assessed prior to surgery predicted reduced weight loss 30 months after bariatric surgery, after controlling for surgery type, baseline weight, sex, and age. Results highlight the importance of evaluation and treatment of anxiety in pre-surgical bariatric candidates.

Introduction

Psychosocial factors, including eating behavior, are routinely examined during pre-surgical psychological evaluations [1]. However, the predictive value of some various pre-surgical psychosocial factors on post-surgical weight loss is still relatively unknown or research is limited. Particularly salient variables of interest at the pre-surgical psychosocial evaluation include eating behavior as well as mood and anxiety disorders. Regarding eating behavior, specifically, binge eating behavior and binge eating disorder have been recognized as adverse factors to weight loss outcomes.

Up to 42% of patients are diagnosed with binge eating disorder prior to bariatric surgery [2]. There are mixed findings about whether pre-surgical binge eating predicts weight loss outcomes after surgery. Though numerous studies have not found a significant relationship [3], others have found that binge eating predicts reduced weight loss two years after bariatric surgery [4]. Previous research has attempted to identify potential contributors for these contradictory results. One such contributor identified includes the differing research methodology utilized for assessment of binge eating between studies [3]. Additional contributing factors for limited associations between pre-surgical psychosocial variables and post-surgical weight loss outcomes may include the focus on the presence or absence of DSM diagnosis prior to surgery versus examining transdiagnostic constructs that may elevate overall risk for suboptimal outcomes [5]. However, a majority of studies include outcomes 6 or 12 months post-surgery [3], therefore a significant gap in the literature exists for understanding these associations beyond this studied timeframe.

In addition to binge eating behaviors, additional cognitive and behavioral aspects of eating behaviors, such as restraint, disinhibition, and hunger, have also gained interest in terms of their influence on long-term weight loss outcomes [6]. Restraint is associated with lower calorie intake [7] but is not predictive of weight loss 1 year after surgery [8]. Disinhibition and hunger are associated with overeating and binge eating [9]. While hunger significantly predicted weight loss one year after bariatric surgery, no relationship was found for disinhibition [8]. Understanding the effect of eating habits on longer-term weight outcomes after bariatric surgery is needed and would be valuable for improving clinical care.

Beyond eating habits and pathology, mood and anxiety disorders are also highly prevalent among bariatric patients and may also affect weight outcomes. The prevalence of a depressive disorder or anxiety at the time of evaluation for bariatric surgery ranges from 6.4%-31.5% whereas lifetime prevalence of an Axis I disorder is up to 72.6% [10]. Previous research has established that the presence of depression and anxiety prior to bariatric surgery is associated with less weight loss after surgery [10-12]. However, the majority of studies have focused on evaluation of examination of depression alone or depression and anxiety in combination versus anxiety only in regard to weight loss outcomes [1]; in fact, a recent systematic review that aimed to examine depression and anxiety after bariatric surgery found that of the 14 articles that met PRISMA guidelines, 6 articles assessed only depression, 8 assessed depression and anxiety, and none of them assessed anxiety only [13]. Therefore, existing research on anxiety predicting weight outcome after bariatric surgery is particularly limited.

We are aware of two studies that examined anxiety only and weight loss outcomes after surgery. These studies varied in how anxiety was assessed, when weight loss was measured after surgery, and how weight and height were measured (i.e., self-report versus objective measurements). One previous study assessed anxiety diagnoses via the Structured Clinical Interview for DSM-IV (SCID) and grouped obsessive–compulsive disorder, posttraumatic stress disorder, and generalized anxiety disorder together under the label of anxiety. Results showed that patients with a current and lifetime prevalence of anxiety lost significantly less weight at 6-12 months and 24-36 months, though weight was self-reported [12]. On the other hand, the second study assessed “frequency and severity” of anxiety on a 5-point scale and did not find a significant relationship between anxiety and weight change after surgery [14]. As such, the specific relationship between anxiety and weight loss outcomes beyond 12 months remains unclear.

Taken together, it is evident that additional research examining the predictive value of psychological constructs pre-operatively and long-term weight outcomes is needed to enhance the clinical utility of the pre-surgical psychological evaluations. Additionally, within the literature that does exist, most related studies examine weight outcomes within 6 to 12 months after surgery [3]. As such, this study aimed to examine psychosocial factors as predictors of weight outcomes at a longer interval post-operatively. As eating behavior and mental health variables represent salient possible predictors of outcomes, depression, anxiety, binge eating, and cognitive/behavioral aspects of eating (restraint, disinhibition, and hunger) prior to bariatric surgery were examined as predictors of weight outcomes 30 months after surgery.

Methods

Participants

Patients presenting for bariatric surgery at an academic medical center within the Appalachian region of the United States completed a routine pre-surgical psychological evaluation between April 2018 and November 2018 as part of the routine evaluation for surgery. Patients completed a battery of self-report measures delivered electronically via HIPAA-compliant Research Electronic Data Capture (RedCap). Patients were eligible for surgery if they were at least 18 years of age and presented with body mass index (BMI) > 40 or BMI >35 with comorbid medical conditions such as obstructive sleep apnea, type II diabetes, and/or hypertension.

Chart reviews were conducted to collect patient weight, height, and BMI, as recorded in the Electronic Medical Record. The Institutional Review Board (IRB) of the institution approved all study procedures.

Measures

Demographics.

Participants’ age, sex, race, height, weight, BMI, educational attainment, employment status, marital status, were collected in this study.

Anxiety and Depression.

The Beck Anxiety Inventory (BAI) [15] and the Beck Depression Inventory (BDI)[16] were used to assess anxiety and depression, respectively, at the initial psychological evaluation prior to bariatric surgery. The BAI is a self-report questionnaire that assesses anxiety severity over the past week and the BDI is a self-report questionnaire that assesses depression severity over the past two weeks. Both measures have Likert scales ranging from 0 (not at all) to 3 (severely), and total scores range from 0 to 63, with higher scores indicating greater symptom severity. Both measures have good psychometric properties and have been used with bariatric surgery populations [5,17,18]. The BAI was designed to target unique aspects of anxiety that are not shared with depression [19]. Of the 21 items on the BAI, 14 assess somatic symptoms (e.g., feeling hot, heart racing, feeling shaky) and 7 assess cognitions associated with anxiety and panic (e.g., fear of worst happening, fear of losing control, scared). The BDI assesses symptoms and attitudes of depression, with sample items including sadness, pessimism, loss of pleasure, concentration difficulty, and changes in appetite.

Three Factor Eating Questionnaire.

The Three Factor Eating Questionnaire (TFEQ) was used to assess three dimensions of eating: a) cognitive restraint, b) disinhibition, and c) hunger [20] at the initial psychological evaluation prior to bariatric surgery. The TFEQ consists of 51 items, of which 36 are true-false items, 14 items are rated on a 4-point scale, and one question is rated on a 6-point scale ranging from 1 no restraint in eating to 6 total restraint. Scores for each dimension range from 0-21 for restraint (TFEQ-R), 0-16 for disinhibition (TFEQ-D), and 0-14 for hunger (TFEQ-H). The three factors of the Three Factor Eating Questionnaire (TFEQ) [20] have demonstrated adequate internal consistency and validity in a population with obesity [21].

Binge Eating.

Severity of binge eating symptoms was assessed using the 16-item Binge Eating Scale (BES) [22] at the initial psychological evaluation prior to bariatric surgery. This questionnaire was designed for use with adults with overweight and obesity. Participants are asked to select from one of three or four statements which most accurately describes their experience. Total scores range from 0 to 46, with higher scores indicating greater severity of binge eating. Scores are further classified into one of three categories: minimal binge eating (≤17), mild to moderate binge eating (18 – 26), and severe binge eating (≥27). The BES scale has demonstrated adequate to good internal consistency in outpatients with obesity [23] and bariatric surgery samples [24].

Weight Loss Outcomes.

Initial weight, height, and BMI were recorded from patients’ initial pre-surgical evaluation with their surgeon. Surgery type was obtained by reviewing the operative note. Attendance of 1 year follow up appointment was documented (yes or no) and weight at that appointment was recorded. If a patient did not attend this appointment, weight was obtained from another department’s visit if it was 1 year after surgery, within a one month range. Weight for 24 and 30 months was obtained from any medical visit within a one month range of the specified date. Weight and height measurements were objective measurements in clinic. Five weight loss variables were then calculated, including; (1) weight loss in kilograms (WL kg; baseline weight in kg – each time point weight in kg; (2) weight loss in BMI (BMI loss; baseline BMI – each time point BMI); (3) percent excess BMI loss (%EBMIL; (BMI loss/(baseline BMI - 25))*100);(4) percent BMI loss (%BMIL; (BMI loss / baseline BMI )*100); (5) percent excess weight loss in kg (%EWL; (WL kg / (baseline weight kg – ideal body weight in kg))*100, ideal body weight was calculated at a BMI of 25.

Results

Statistical Analysis

Statistical analyses were conducted using SAS 9.4. Descriptive statistics for categorical variables included frequencies and valid percentages and for continuous variables included means and standard deviations. The primary research question was analyzed with a variety of repeated and random effects linear mixed models (LMM) with restricted maximum likelihood (REML) estimation and Kenward-Roger degree of freedom method, where the best-fitting (i.e., smallest AIC) model included a random effect of intercept and slope of time in months, with time fit continuously, and an unstructured covariance structure, and included the time in months effect, the predictors with the predictor by time interaction terms, and covariates that include baseline weight (kg), birth sex, age (years), and surgery type. Predictors were run in separate models due to multicollinearity of several predictors (see Supplemental Tables A-D for more information). All other LMM assumptions were checked and found satisfactory. Additional models were run with different methods of calculating weight loss (including %BMIL and %WL). Models were generally consistent, so only the %BMIL models are presented here as they presented with the best model AIC.

Missing data.

Sensitivity analysis was conducted examining model differences between those with complete weight data across all time points (n=76), and the results when all participants (n=196) were included using the REML estimation method in the LMM. Differences were found between models; notably, model effects were generally strengthened for the complete case models, suggesting bias in the estimates due to MNAR or MAR attrition. Further assessment was done for all measures at each time point with a missing data indicator; at baseline only race was significantly different, with all 8 participants of non-White race having missing data at some point (Fisher’s exact p=0.024). No other differences were noted at baseline for all demographic and weight measurements. At 12-months and 24-months, the weight change variables did not differ by the missing data indicator. However, at 30-months, the change in excess weight variables approached but did not reach statistical significance (p=0.06), suggesting those with missing data at earlier time points that came back for the final assessment may have been marginally less successful in their weight loss efforts (e.g., %EWL in kg, all data n = 76, M=50.8 v. missing data n = 10, M = 40.1). Although none of the variables presented with statistically significant differences indicating MNAR data, we suspect some missing data may be at least MAR, which all case inclusion in a LMM with REML estimation can estimate with less bias and more appropriate “intent to treat” than including only those with complete cases.

We ran another sensitivity analysis to check whether the 30 month time point was consistent with the 12 and 24 time points by running identical models but excluding the final time point; model results demonstrated consistent estimates but attenuated p-values, demonstrating consistency and increased power in including the final time point. Thus, we presented the model with all time points.

Results

Descriptive statistics can be found in Table 1. A total of 196 participants had weight data at baseline and underwent surgery for weight loss. Most of the participants were White (n=188, 96%). Table 2 includes weight loss for all outcomes variables. On average, participants lost the most weight at 12-months (M=37.9 kg), and then regained some weight over time, at 24- (M=34.9 kg) and 30- (M=32.7 kg) months.

Table 1.

Demographics and Descriptives (N=196)

	n (%)
Birth Sex
Female	162 (82.65%)
Male	34 (17.35%)
Race
Caucasian	188 (95.92%)
Non-Caucasian	8 (4.08%)
Marital Status
Single/never married	36 (18.37%)
Married	117 (59.69%)
Divorced	24 (12.24%)
Separated	7 (3.57%)
Widowed	10 (5.10%)
Remarried	2 (1.02%)
Education
High school not completed	3 (1.53%)
High school or GED	50 (25.51%)
Some college	58 (29.59%)
Associate’s degree	32 (16.33%)
Bachelor’s degree	35 (17.86%)
Master’s degree or higher	18 (9.18%)
Employment
Full-time	131 (66.84%)
Part-time	8 (4.08%)
Disabled	13 (6.63%)
Retired	8 (4.08%)
Unemployed	20 (10.20%)
Homemaker	13 (6.63%)
Student	6 (1.53%)
Surgery Type
Gastric sleeve	180 (91.84%)
Gastric bypass	16 (8.16%)
Anxiety (categories)
Minimal	138 (70.41%)
Mild	33 (16.84%)
Moderate	21 (10.71%)
Severe	4 (2.04%)
Depression (categories)
Minimal	143 (72.96%)
Mild	22 (11.22%)
Moderate	29 (14.80%)
Severe	2 (1.02%)
	M (SD)
Age	43.27 (11.18)
TFEQ Cognitive Restraint	13.60 (3.81)
TFEQ Disinhibition	6.06 (3.01)
TFEQ Hunger	3.72 (2.90)
Binge Eating Scale	10.80 (6.18)
Beck Depression Inventory	7.84 (7.51)
Beck Anxiety Inventory	7.57 (8.07)

Note. TFEQ = Three Factor Eating Questionnaire. BES = Binge Eating Scale. BDI = Beck Depressive Inventory. BAI = Beck Anxiety Inventory.

Table 2.

Weight loss outcomes and anxiety across 30 months

	Baseline (N=196)		12 months (n=53)		24 months (n=130)		30 months (n=92)
	M	SD	M	SD	M	SD	M	SD
Kg	135.55	24.42	97.41	19.89	102.17	21.03	106.43	23.14
Lb.	298.86	53.86	214.75	43.85	225.24	46.37	234.64	51.00
BMI	48.71	7.35	35.11	6.48	36.49	6.68	37.81	7.25
WL (kg)	--	--	37.94	14.94	34.88	17.53	32.69	15.20
BMIL	--	--	13.62	5.13	12.42	6.13	11.62	5.25
%EBMIL	--	--	59.22	20.31	52.78	21.94	49.07	20.61
%BMIL	--	--	27.73	9.10	25.00	10.62	23.39	9.74
%EWL (kg)	--	--	59.08	20.26	52.64	21.87	48.95	20.47
Anxiety Category (BAI)	Baseline %BMIL		12 months %BMIL		24 months %BMIL		30 months %BMIL
Minimal	--	--	12.19	8.98	25.44	10.58	24.43	9.76
Mild	--	--	29.62	6.95	28.07	7.96	24.59	7.87
Moderate	--	--	25.67	11.25	17.03	11.93	13.34	9.14
Severe	--	--	37.94	12.40	26.15	9.53	21.56	3.48

Note. WL = weight loss in kg. BMIL = weight loss in BMI. %EBMIL = percent excess BMI loss. %BMIL = percent BMI loss. %EWL = percent excess weight loss in kg. The following are the clinical cut offs for the Beck Anxiety Inventory (BAI): minimal anxiety is 0-9, mild anxiety is 10-16, moderate anxiety is 17-29, and severe anxiety is 30-63.

Across all models, weight loss generally decreased over time, from 12 to 30 months loss (Est. range: −0.23, −0.12, p-value range: 0.005, 0.13), higher baseline weight was associated with more weight loss (Est. range: 0.07, 0.08, p-value range: 0.011, 0.017), older age was associated with less weight loss (Est. range: −0.15,−0.14, p-value range: 0.005, 0.13), and those undergoing laparoscopic sleeve gastrectomy (SG) had less weight loss than those undergoing laparoscopic Roux-en-Y gastric bypass (LRYGB) loss (Est. range: −9.33, −8.98, p-value range: 0.0003, 0.0005). After 30 months, those who underwent LRYGB lost on average 12.24 kg more weight than those who had SG (SG, M=31.76, SD=14.78; LRYGB, M=44.00, SD=16.81). Anxiety had a significant interaction effect with time (Est. = −0.01, p = 0.013), indicating that higher anxiety was associated with less weight loss over time. Other predictors were not statistically significant. See Table 3 for model values for all outcome variables.

Table 3.

Linear mixed model fixed effect results for %BMIL from time 12 months to 30 months (N=196)

Predictor	Variable	Estimate	SE	Df	t	p
Anxiety (BAI)	Intercept	31.87	6.03	164	5.29	<.0001
	Baseline weight	0.07	0.03	160	2.46	0.015
	Time	−0.12	0.05	123	−2.35	0.020
	Sex	1.99	1.99	157	1.00	0.318
	Age	−0.15	0.06	162	−2.47	0.015
	Surgery type	−8.99	2.49	155	−3.61	0.0004
	Predictor	0.21	0.11	139	1.91	0.058
	Time*Predictor	−0.01	0.00	121	−2.52	0.013
Depression (BDI)	Intercept	32.00	6.06	165	5.28	<.0001
	Baseline weight	0.08	0.03	159	2.58	0.011
	Time	−0.16	0.06	126	−2.82	0.006
	Sex	2.24	2.00	157	1.12	0.266
	Age	−0.14	0.06	162	−2.38	0.019
	Surgery type	−9.17	2.49	155	−3.69	0.0003
	Predictor	0.09	0.12	148	0.69	0.490
	Time*Predictor	−0.01	0.01	136	−1.31	0.192
Cognitive Restraint (TFEQ)	Intercept	32.18	6.71	186	4.80	<.0001
	Baseline weight	0.07	0.03	159	2.55	0.012
	Time	−0.22	0.14	130	−1.51	0.134
	Sex	1.95	2.03	157	0.96	0.337
	Age	−0.14	0.06	161	−2.40	0.018
	Surgery type	−9.23	2.50	155	−3.70	0.0003
	Predictor	0.07	0.24	145	0.30	0.768
	Time*Predictor	0.00	0.01	130	0.04	0.967
Disinhibition (TFEQ)	Intercept	33.18	6.15	172	5.40	<.0001
	Baseline weight	0.07	0.03	159	2.41	0.017
	Time	−0.23	0.09	130	−2.52	0.013
	Sex	1.89	2.04	157	0.93	0.355
	Age	−0.15	0.06	162	−2.45	0.015
	Surgery type	−8.98	2.52	155	−3.56	0.0005
	Predictor	0.08	0.30	152	0.27	0.786
	Time*Predictor	0.00	0.01	131	0.19	0.849
Hunger (TFEQ)	Intercept	32.29	6.12	167	5.28	<.0001
	Baseline weight	0.07	0.03	160	2.56	0.012
	Time	−0.18	0.06	128	−2.79	0.006
	Sex	2.11	1.99	157	1.06	0.291
	Age	−0.15	0.06	162	−2.41	0.017
	Surgery type	−9.15	2.50	155	−3.66	0.0003
	Predictor	0.20	0.32	145	0.62	0.539
	Time*Predictor	−0.01	0.01	130	−0.63	0.528
Binge Eating (BES)	Intercept	33.26	6.12	169	5.44	<.0001
	Baseline weight	0.08	0.03	159	2.56	0.011
	Time	−0.22	0.08	126	−2.85	0.005
	Sex	2.41	2.06	158	1.17	0.242
	Age	−0.14	0.06	161	−2.31	0.022
	Surgery type	−9.33	2.51	155	−3.71	0.0003
	Predictor	−0.07	0.15	148	−0.45	0.656
	Time*Predictor	0.00	0.01	127	0.16	0.872

Note. Baseline weight in kg. Time is in months. The reference group for sex is female. The reference group for surgery type is gastric bypass. BAI = Beck Anxiety Inventory. BDI = Beck Depressive Inventory. TFEQ = Three Factor Eating Questionnaire. BES = Binge Eating Scale

Post-hoc Analyses.

A t-test revealed no significant difference in anxiety by surgery type (SG or LRYGB), t(194)= −0.13, p = 0.900. Average anxiety score for those who had SG were 7.54 (SD=8.10) and average anxiety scores for those who had LRYGB were 7.81 (SD=7.99)

Though anxiety was a continuous variable for the main analyses, it was categorized into minimal (0-9 on BAI), mild (10-16 on BAI), moderate (17-29 on BAI), and severe (30-63 on BAI), in accordance with the measure’s cut off points, to further clinical interpretation. Table 2 shows that those with minimal anxiety initially lost the least amount of weight at 12 months (M=12.19, SD=8.98) but continued to lose weight at the 24- (M=25.44, SD=10.58) and 30- (M=24.43, SD=9.76) month marks. In contrast, all other anxiety categories lost more weight but did not maintain loss. When included in a linear mixed model, those with severe anxiety lost the most weight (Est. = 16.45, p = 0.003) and also regained the most weight, in reference to those with minimal anxiety (Est. = −0.61, p = 0.014) and after controlling for covariates.

Anxiety diagnoses, according to ICD-10 codes (i.e., anxiety, generalized anxiety disorder, social anxiety disorder, and panic), were extracted from patient medical records. 37.24% (n = 73) of participants had an anxiety diagnosis. The presence of an anxiety diagnosis, versus no anxiety diagnosis, predicted weight loss (Est. = 3.94, p = 0.033). Of those who had severe anxiety on the BAI, 100% (n = 4) had an anxiety diagnosis. Of those who had moderate anxiety on the BAI, 86% (n = 18) had an anxiety diagnosis. For participants who had mild anxiety on the BAI, 70% (n = 23) had an anxiety diagnosis. Finally, for participants who had minimal anxiety on the BAI, 20% (n = 28) had an anxiety diagnosis.

Discussion

This study aimed to examine psychosocial factors as predictors of weight outcomes 30 months after bariatric surgery. Bariatric surgery emerged as an effective intervention within the sample, with patients achieving weight loss over time. Higher baseline weight was associated with greater weight loss. Surgery type also significantly affected weight outcomes, such that patients with SG had less weight loss compared to patient with LRYGB.

Of the predictors considered in the main analyses, only anxiety emerged as a significant predictor of weight outcomes 30 months after bariatric surgery. Anxiety, at the time of evaluation prior to bariatric surgery, predicted reduced weight loss 30 months after surgery. Those with severe anxiety prior to surgery lost the most weight 12 months after surgery, but also regained the most weight 30 months after surgery. The presence of an anxiety diagnosis also significantly predicted weight outcomes after 30 months.

These results are consistent with some previous research suggesting a potential link between anxiety and weight loss outcomes, though prior research on anxiety predicting weight outcome after bariatric surgery is limited. The present results are consistent with findings from de Zwann and colleagues (2011) which showed that an anxiety predicted poorer weight loss outcomes up to 36 months after bariatric surgery [12]. Other studies that have examined comorbid anxiety and depression prior to surgery are associated with reduced weight loss outcomes. For example, a study that examined 153 bariatric surgery patients found that anxiety, according to DSM-IV criteria, predicted decreased weight loss after 4 years [11]; however, given that “anxiety and/or depression” was examined in this study, the results for anxiety alone are unclear. Several explanations for anxiety’s impact on weight outcomes may be possible. Social anxiety disorder is common among obese and bariatric samples and has a lifetime prevalence rate of 7.5% [25]. Evidence suggests that social anxiety impacts overeating and emotional eating behavior among bariatric patients, which leads to decreased weight loss [26]. Another study found that attachment anxiety (i.e., fear of social rejection and abandonment) predicted decreased weight loss after bariatric surgery and this effect was medicated by dietary adherence [27]. Those with anxiety may be more sensitive to changes in social situations involving eating and the way others perceive them after bariatric surgery, especially given the numerous eating habit modifications that are recommended after bariatric surgery.

Another consideration is the physiological aspect of anxiety and the subsequent effect on eating habits. The experiential avoidance model of anxiety suggests that worry prevents people with clinical anxiety from remaining “in contact” with their internal experience [28]. Such disconnection or “mindlessness” may interfere with interpretation of hunger signaling and cessation cues. As such, patients may have a deficit in perceiving physiological hunger and may eat beyond times of physical hunger.

Finally, the psychosomatic theory of emotional eating states that individuals with obesity have low interoceptive awareness and have difficulty distinguishing between hunger and other feelings of discomfort, and therefore end up eating in response both [29]. Thus, responding to numerous different cues by eating would likely lead to consuming greater overall calorie intake over time, which may cause sub-optimal weight loss or weight regain in bariatric patients. Indeed, prevalence rates of emotional eating in bariatric surgery candidates are high [30]. This rationale is especially salient given that in the present study, anxiety was assessed via the Beck Anxiety Inventory, which has a disproportionate emphasis one assessing physiological aspects of anxiety [15]. Of the 21 items of the BAI, over half of the items (n=13) items describe physical sensations. As such, the BAI may identify individuals who have physical sensations of discomfort based on anxiety, which has the potential to contribute to eating behavior in response to these cue instead of hunger signaling. These results highlight the importance of evaluation and treatment of anxiety in pre-surgical bariatric candidates.

Post-hoc analyses revealed an interesting trajectory of how severity of anxiety symptoms impacts weight loss over time after bariatric surgery. Those who had minimal anxiety prior to surgery lost the least weight after 12 months, compared to those with greater anxiety; however, they continued to lose weight by 24 months and essentially maintained weight by 30 months after surgery. Those with mild, moderate, and severe anxiety lost the most weight initially (at 12 months) but then regained weight, ultimately having a poorer net weight loss compared to those with minimal anxiety by 30 months after surgery. Further, those with severe anxiety regained the most weight by 30 months after surgery. It is possible that those with greater anxiety prior to surgery initially perform best within the first 12 months after surgery as anxiety may contribute to increased adherence with post-surgical recommendations. However, as the “honeymoon” phase of bariatric surgery fades, those with greater anxiety may struggle to continue that high level of adherence in the maintenance phase of bariatric surgery, ultimately leading to poorer outcomes than those with minimal or no anxiety. The mechanisms by which severe anxiety contributes to this trajectory following surgery is an area for future research.

Binge eating and depression were not significant predictors of weight outcomes in this study. Non-significant results for binge eating and weight outcomes are consistent with some previous research, including a recent systematic review and meta-analysis [3]. A recent study found that depression did not impact weight outcomes, but atypical depressive symptoms were associated with BED and binge eating severity [31]. A recent study from the Longitudinal Assessment of Bariatric Surgery-2 (LABS-2) examined changes in depressive symptoms over time and the effect on weight outcomes up to 7 years after surgery [32]. Though a majority of patients experienced a decrease in severity of depressive symptoms over time after surgery, various subgroups emerged dependent on the trajectory of depressive symptoms after surgery. Classification into suboptimal subgroups impacted weight outcomes, such that those who experienced the greatest decrease in depressive symptoms soon after surgery had the best weight loss outcomes [32]. It is likely that the change in depression symptoms and its’ impact on weight outcomes is more nuanced than the present examination allowed.

Strengths of the study include a large sample of bariatric patients with weight outcomes at a longer post-operative interval than many other studies currently available in the literature. Given that retention is a challenge with longitudinal studies, the fact that numerous patients receive additional medical services beyond just bariatric services at the same hospital is a unique strength of the study sample; weight from any medical appointment within the appropriate time frame was used, which allowed for more complete data collection at follow up time points. Further, all weight variables and surgery type were extracted from medical and operative notes, thereby increasing the accuracy as opposing to self-report data collection. There was minimal missing data for the predictor variables and the statistical analysis utilized were powerful and able to account for missing data.

Study limitations include that testing was not confidential given that it was a part of the clinical pre-surgical psychological evaluation. Social desirability may influence results or bias the pre-surgical psychosocial variables towards underreporting [33]. The study’s sample is majority Caucasian (96%) and findings may not generalize to other clinics with a more racially and ethnically diverse bariatric population. Also, binge eating with loss of control was not specifically assessed and distinguished in this study; this is a limitation because studies that identify “loss of control” eating in assessment of binge eating consistently demonstrate a significant relationship with weight outcomes, whereas those that do not may fail to demonstrate this association [34]. Finally, it is possible that the missing data was not missing not random; therefore, attrition may bias estimates.

Conclusions

This study found that anxiety assessed prior to surgery predicted reduced weight loss 30 months after bariatric surgery, after accounting for surgery type, baseline weight, sex, and age as covariates. Those with severe anxiety prior to surgery lost the most weight 12 months after surgery, but also regained the most weight 30 months after surgery. Results highlight the importance of evaluation and treatment of anxiety in pre-surgical bariatric candidates. Future research should examine specific types of anxiety (e.g., generalized, social, panic) to further understanding of how pre-operative anxiety impacts long-term weight outcomes after bariatric surgery.

Highlights.

Anxiety assessed prior to surgery predicted reduced weight loss after bariatric surgery

Those with severe anxiety initially lost the most weight, compared to those with minimal anxiety

Those with severe anxiety also regained the most weight after 30 months

Evaluation and treatment of anxiety in bariatric candidates is important for outcomes