Weight stigma and posttraumatic stress disorder symptoms in individuals seeking bariatric surgery

Abstract

Background:

After bariatric surgery, some patients experience adverse psychiatric outcomes, including substance use, suicidality, and self-harm. These factors are commonly associated with posttraumatic stress disorder (PTSD) and related symptoms (PTSD-S) that develop following adverse childhood experiences (ACEs) and traumatic events. However, emerging evidence suggests that chronic discrimination also may contribute to PTSD-S. Weight-based discrimination is salient for people with obesity but has received little attention in relation to PTSD-S.

Objective:

Our study examined factors that may contribute to the link between experienced weight stigma (WS), which is common in individuals seeking bariatric surgery, and PTSD-S.

Setting:

Teaching hospital and surgical weight loss center in the United States.

Methods:

A total of 217 participants completed self-report surveys of experienced and internalized WS, ACEs, and PTSD-S. Demographics and trauma history were obtained from patient medical records. A stepwise multiple regression examined associations between experienced WS and internalized WS with PTSD-S, co-varying demographics, ACEs, and trauma, followed by examination of whether findings held co-varying anxiety/depressive symptoms in a participant subset (n = 189).

Results:

After accounting for covariates in step 1 and ACEs and trauma in step 2 (ΔR² = .14), experienced WS and internalized WS accounted for substantial PTSD-S variance in steps 2 and 3 (ΔR² = .12 and .13, respectively; overall model R² =.44; P < .001). Findings held after co-varying anxiety/depressive symptoms.

Conclusions:

Over and above ACEs and trauma, experienced WS and internalized WS may contribute to PTSD-S. Longitudinal research is needed to better elucidate the pathways underlying these associations.

Research shows concerning increases in postoperative psychopathology, especially substance use disorders, suicidality, and self-harm in subsets of the bariatric surgery population [1,2]. While the etiology of these conditions is complex, some studies implicate preoperative psychopathology including depression in these sequelae [1]. Yet comparatively little research has examined posttraumatic stress disorder (PTSD) or PTSD symptoms (PTSD-S) despite lifetime Diagnostic and Statistical Manual of Mental Disorders, fifth edition (DSM-5) PTSD prevalence rates as high as nearly 22% observed among individuals seeking bariatric surgery using structured diagnostic interviews [3,4]. PTSD is frequently co-morbid with depressive and substance use disorders as well as suicidality in other patient populations and is implicated as a driver of the latter psychiatric concerns [5]. Moreover, among bariatric sugery patients, early research suggests that PTSD is the most persistent postoperative psychiatric condition and that PTSD-S do not appear to remit following surgery [6–8].

Traumatic event exposure, particularly childhood maltreatment, is relatively common among patients seeking obesity surgery [9], with reported rates of physical or sexual abuse ranging from 20% to 30% [10]. A history of trauma is associated with greater preoperative psychopathology, including PTSD, depression, suicidality, and substance use disorders [11]. PTSD may be a common, if understudied, factor linking trauma exposure to other psychiatric concerns; a recent cross-sectional study observed that a lifetime PTSD diagnosis was associated with increased preoperative psychiatric impairment, including mood, eating, and substance use disorders and suicidality [12]. Yet a recent narrative review concluded that while a history of trauma appears to increase vulnerability to postoperative decrements in mental health, the current literature remains too heterogeneous to adequately characterize [10]. This heterogeneity may relate in part to the paucity of research examining PTSD-S or associated mechanisms.

One factor that may contribute to PTSD-S is experiences of weight stigma (experienced WS), referring to social devaluation of people because of their body weight [13]. Experienced WS is common among individuals seeking bariatric surgery [14] and can lead to discrimination that spans the interpersonal (e.g., inappropriate or derogatory remarks) to structural (e.g., small seats) domains [13]. Although “exposure to actual or threatened death, serious injury, or sexual violence” is required for a DSM-5 diagnosis of PTSD (p. 271, criterion A) [15], emerging research implicates PTSD-S to be associated with and/or to develop following chronic interpersonal discrimination in crosscutting domains related to race [16], sexual minority status [17], HIV status [18], and gender [19]. Experienced WS is also a chronic interpersonal stressor linked to forms of distress that are commonly co-morbid with PTSD, including depression, anxiety, and stress [13,20]. Thus, although experienced WS has not yet been examined in relation to PTSD-S, it is a key stressor that may contribute to its etiology in this population.

Experienced WS can increase awareness of one’s stigmatized identity, resulting in the self-application of harsh weight-related stereotypes (internalized WS) [21,22]. In other diverse populations, internalized stigma related to HIV status [23], sexual minority status [24], and mental illness [25] has been cross-sectionally and/or prospectively linked to PTSD-S. Internalized WS is commonly reported among individuals seeking bariatric surgery [26] and is implicated as an important mechanism of the link between experienced WS and distress [27]. Yet no research has yet examined the link of internalized WS to PTSD-S in any population, including in relation to experienced WS. In aggregate, these data suggest that experienced WS and internalized WS may be understudied factors that increase vulnerability to developing PTSD-S.

This cross-sectional study assessed 2 hypotheses: (1) experienced WS will be associated with PTSD-S after accounting for adverse childhood experiences (ACEs) and criterion A events, and (2) internalized WS will account for an incremental variance in PTSD-S after accounting for covariates, ACEs, criterion A events, and experienced WS.

Methods

Participants and procedure

Participants were enrolled through a parent prospective trial from June 2015 to 2019 from a Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program–accredited center in the Northeast in the United States. Individuals seeking bariatric surgery (i.e., preoperative), irrespective of anticipated surgical procedure, were eligible to participate. Exclusion criteria included (1) those under18 years of age, (2) those seeking revisional operations on previous weight loss procedures, and (3) those unable to read or speak English. The study was advertised as assessing “adverse interpersonal experiences and health in bariatric surgery patients.” Recruitment occurred via mailings with study advertisements, patients’ surgical weight loss providers, and bariatric surgery support group meetings. Data collection was monitored by research staff completely independent of clinical care. Patients were informed that no provided information would be shared with their surgical weight loss team and that their participation would not impact their eligibility for bariatric surgery. The institutional review board–approved study personnel administered written informed consent with eligible participants. Pertinent deidentified data from the medical record were merged with a self-report survey administered via Qualtrics (Qualtrics International, Inc, Provo, Utah) or a paper-based survey per patient preference. Amazon gift cards ($10) were provided for study participation. Institutional review board approval was secured by the University of Connecticut and Hartford Hospital.

Of patients who completed study surveys (n = 229; 75% of those enrolled [n = 306] and 33% of those potentially eligible [n = 704]), 222 had data available for all constructs of interest. Of these, 5 were excluded because of class I obesity, resulting in a final sample of 217 patients. Participants in this analysis did not differ from those enrolled in the study who did not complete surveys (n = 77) on ethnicity, insurance status, age, or body mass index (BMI) in χ² analyses and independent t tests (P ˃ .05). However, survey completers were less likely to be female (χ²(1, n = 294) = 83.58; P < .001) or to self-report Black or Other relative to White race (χ²(2, n = 294) = 6.55; P = .036). Of note, our prior work in an overlapping sample assessed different research questions [28].

Measures

Data extracted from the patient medical records included demographics (age, sex, and Medicaid as a proxy for socioeconomic status), objectively measured BMI (at the date closest to survey completion), and reported history of DSM-5 criterion A traumatic event (yes/no for presence or absence, coded by a trained research assistant from the semistructured psychological assessment for surgery). Remaining constructs were assessed via the study survey.

Adverse childhood experiences were assessed with the ACE Checklist [29], which measures 10 categories of childhood maltreatment (yes/no; e.g., childhood physical or sexual abuse). An ACE score ˃4 represents the threshold for accelerated probability of adverse health effects, including depression, alcoholism, and suicidality [30]. The ACE Checklist has been used in bariatric surgery samples [31] and has shown good test–retest reliability (˃.65) [32].

Experienced WS was assessed with the Stigmatizing Situations Inventory-Brief (SSI-B) [33], which measures varied forms of experienced WS (e.g., physical barriers and comments about one’s weight by doctors and children). The first 173 participants were administered the original 10-point SSI-B ranging from 0 (never) to 9 (several times per week). The remaining participants (n = 56) were administered an 8-item version ranging from 0 (never) to 7 (several times per week), an approach used in prior research with bariatric surgery patients [34] to minimize the relatively low mean values and standard deviations previously observed in the initial validation study of the full SSI in this population [14]. A single scale was created for this analysis by converting the anchors for each subset of participants using percentage frequencies that were then combined [35]. The SSI-B has good reliability and validity [33]. For the converted scale in this sample, Cronbach’s α = .91.

Internalized WS was assessed with the Weight Stigma Internalization Scale-Modified (WBIS-M) [36,37], which assesses the extent to which prejudicial beliefs about people with obesity are applied to oneself (e.g., “I hate myself for my weight”). Items were rated from 1 (strongly disagree) to 7 (strongly agree). The WBIS-M has shown good reliability and validity [26] and been used with bariatric surgery patients (α = .86) [26].

PTSD-S were assessed with the Posttraumatic Stress Checklist, Civilian version, (PCL-C) [38]. Symptoms assessed correspond with DSM-4 (1994) diagnostic criteria to generate a symptom severity score. Past-month symptoms are assessed in relation to lifetime “stressful experiences” rather than a criterion A traumatic event. Items are ranked on a 5-point Likert scale ranging from 1 (not at all) to 5 (extremely). A provisional PTSD diagnosis was characterized using the symptom cluster method [39]; PTSD criteria were met if symptomatic responses were reported for ˃1 cluster B item, ˃3 cluster C items, and ˃2 cluster D items and total symptom severity score ≥35. Data analyses used the symptom severity score. The PCL-C has shown good reliability and validity [38] and has been employed as a measure of PTSD-S in prior studies without specific reference to a criterion A event [23,40] as well as in studies with people with overweight/obesity (α = .94) [41].

Anxiety and depressive symptoms were assessed in a subset of participants with available data from the patient psychological chart (n = 189 had overlapping availability of both measures) using 2 widely used screeners: the Generalized-Anxiety Scale-7 (GAD-7) [42] and the Beck Depression Inventory II (BDI-II) [43]. Both scales are broadly used in clinical and research settings, have shown good internal consistency, and have been used in research with individuals seeking bariatric surgery [44,45]. Cronbach’s α is unavailable in this sample because the data-pulling procedure recorded total scores only.

Analyses

Data were examined for missing values and outliers. All available cases were analyzed (complete case analysis; n = 217). Skewness and kurtosis were within recommended parameters for regression analysis (i.e., skewness <2.1; kurtosis <7.1) [46]. All analyses used SPSS version 27.0 (IBM, Armonk, NY, USA) and controlled for age, BMI, sex, ethnicity, race, and insurance status as a proxy for socioeconomic status given their associations with WS and/or PTSD [21,47–49]. To test our first and second hypotheses, a hierarchical multiple regression model was used to examine the proportion of variance in PTSD-S accounted for by experienced WS (step 3) after accounting for demographic covariates in step 1 (hypothesis 1), followed by ACEs and history of trauma in step 2, and to examine the subsequent incremental variance accounted for in PTSD-S by internalized WS in step 4 (hypothesis 2). Last, we explored whether findings held after accounting for anxiety and depressive symptoms in the subset of participants with available data (n = 189) by repeating the hierarchical multiple regression analysis and changing only the addition of anxiety and depressive symptoms in step 1 to the model.

Results

Participants were predominantly female (n = 177; 81.6%), non-Hispanic (n = 178; 82%), and self-identified as primarily White (n = 143; 65.9%), Black (n = 40; 18.4%), Multiracial (n = 14; 6.5%), or Other (n = 16; 7.4%), with a mean age of 42.09 ± 11.99 years. Levels of obesity included class II (BMI 35 or < 40 kg/m²; n = 29; 13.1%) and class III (BMI ≥ 40 kg.m²; n = 188; 86.6%), with a mean BMI of 48.01 ± 8.53 kg/m². Most participants used private insurance (n = 137; 63.1%) or Medicaid (n = 68; 31.3%), and nearly all reported experiencing WS at least once in their lives (SSI-B > 0; n = 210; 96.8%).

Twenty-four percent of participants (n = 51) endorsed experiencing 4 or more ACEs. Twenty-five percent of the sample (n = 54) met the criteria for provisional DSM-4 PTSD diagnosis per the symptom cluster scoring method (see Measures). Per the psychology report, 21.7% of participants (n = 47) reported experiencing a criterion A traumatic stressor. Only 8.8% (n = 19) of the overall sample reported both criterion A and met criteria for a provisional PTSD diagnosis.

Means and intercorrelations for main study constructs are presented in Table 1; all were associated in hypothesized directions (P < .05). Some data on key constructs for this analysis are reported elsewhere with an overlapping sample [28]. These include demographic comparisons between survey completers and those not enrolled in the study and comparisons by covariates on study constructs. Here we present comparisons between covariates and study constructs elsewhere unreported: trauma history and PTSD-S. BMI was associated with PTSD-S (r = .182; P = .007), criterion A trauma (r = .141; P =.038), and age inversely with experienced WS (r = −.145; P =.033). No other findings were significant (P ˃ .05).

Table 1.

Descriptive statistics and intercorrelations for main study variables

Measure	EWS	IWS	PTSD-S	ACE	Trauma
EWS	__
IWS	.382**
PTSD-S	.477**	.559**	__
ACE	.284**	.247**	.366**	__
Trauma	.201**	.144*	.232**	.403**	__
M	.22†	3.85	34.26	2.30	__
SD	.17	1.39	14.90	2.11	__
n	217	217	217	217	217

EWS = experienced weight stigma; IWS = internalized weight stigma; PTSD-S = posttraumatic stress disorder symptoms; ACE = adverse childhood experiences; M = mean; SD = standard deviation.

Trauma (psychology chart): n = 49 (22.2%) reported a history of trauma.

^*P < .05.

^**P < .01.

^†Mean percentage frequency on 0–100 scale ranging from never to weekly.

Assessing proportion of variance in PTSD-S accounted for by experienced WS and internalized WS

As shown in Table 2, the overall model accounted for 43.7% of variance in PTSD-S (F(10,206) = 16.02; P < .001). After accounting for covariates in step 1 (ΔR² = .04; P = .138) and ACEs and history of trauma in step 2 (ΔR² = .144; P < .001), experienced WS accounted for a significant 12.1% in step 3 (P < .001; hypothesis 1). Modeled following the antecedent steps, internalized WS accounted for 12.8% in step 4 (P < .001; hypothesis 2). Findings held when the analysis was repeated co-varying for anxiety and depressive symptoms in the available participant subset (F(12,176) = 16.28; P < .001).

Table 2.

Regression models reporting unstandardized (B) and standardized betas (β) and standard errors (SE) for predictors of posttraumatic stress disorder symptoms

Variable	Model 1			Model 2			Model 3			Model 4
	B	SE	β	B	SE	β	B	SE	β	B	SE	β
Age	−.05	.09	−.04	−.03	.09	−.03	−.004	.08	−.003	.05	.07	.04
BMI	.30	.12	.17*	.24	.12	.14*	−.14	.13	−.08	−.06	.11	−.04
Ethnicity	2.43	3.02	.06	2.72	2.79	.07	1.07	2.60	.03	−1.90	2.39	−.05
Race	.78	1.95	.03	−.86	1.83	−.03	−2.00	1.70	−.08	−1.83	1.54	−.07
SES	−1.27	1.81	−.05	−2.24	1.69	−.09	−1.38	1.57	−.06	−.06	1.43	−.003
Sex	−2.66	2.68	−.07	−3.85	2.50	−.10	−3.94	2.31	−.10†	−2.70	2.10	−.07
ACE				2.40	.49	.34***	1.67	.47	.24***	1.21	.43	.17**
Trauma				3.56	2.50	.10	2.99	2.32	.08	2.23	2.10	.06
EWS							37.80	6.28	.43***	25.00	5.98	.29***
IWS										4.38	.64	.41***
R 2			.04			.19***			.31***			.44***

BMI = body mass index; SES = socioeconomic status; ACE = adverse childhood experiences; EWS = experienced weight stigma; IWS = internalized weight stigma.

^*P <.05.

^**P <.01.

^***P <.001.

^†P ˃.10.

Discussion

Among individuals seeking bariatric surgery in our sample, experienced WS and internalized WS together accounted for a quarter of variance in PTSD-S. We also found experienced WS and its correlates remained associated with PTSD-S after accounting for demographics, ACEs, and criterion A events as well as symptoms of anxiety and depression in a participant subset, implicating WS as an understudied mechanism of PTSD-S in this population.

Our findings offer preliminary support for the hypothesis that experienced WS and its correlates may contribute to PTSD-S in this population and implicate its internalization as a potential vulnerability factor. Our findings align with an increasing body of research that suggests that chronic discrimination may contribute to PTSD-S in diverse populations [16]. Some such studies have been critiqued for not accounting for DSM-relevant traumatic stressors [16], although our findings suggest that the link of WS and PTSD-S may persist even after accounting for criterion A stressors. It remains possible, however, that the persistent experienced WS and PTSD-S link in our sample is better attributed to other trauma-related variance unqueried or unreported at the psychology evaluation.

Consistent with prior research, adverse interpersonal experiences were common within our sample of individuals seeking bariatric surgery. Nearly all participants reported experiencing WS at least once in their lives, with about one-tenth reporting 4 or more ACEs. During the psychology assessment, approximately one-fifth reported a criterion A traumatic experience. Although prior research suggests that unstructured interviews underestimate PTSD prevalence rates [50], our findings generally mirror prior research with people seeking bariatric surgery that shows rates of physical or sexual abuse ranging from 20% to 30% [10]. Our study complements these findings alongside those of Salwen et al. [51], who showed even greater elevations in trauma (31%–61%) alongside experienced WS in this population.

Twenty-five percent of our sample met criteria for a current provisional PTSD diagnosis using the PCL-C, greater than those reporting a criterion A traumatic stressor at the surgical psychological evaluation (22%) and higher than any study to date has reported for people seeking bariatric surgery. This may relate in part to our use of the PCL-C and its anchor to “stressful experiences”; other studies have reported PTSD/PTSD-S using either semistructured or structured psychiatric diagnostic assessments. While this may be viewed as a limitation, our study has the advantage of administering the PCL-C outside the context of the psychological assessment for surgery, feasibly leading to less underreporting of symptoms than has been documented in this context [50]. Notably, when considering those who endorsed criterion A and met PCL-C criteria for provisional PTSD using the symptom-cluster method, the PTSD rate dropped to 8.6%, similar to other current PTSD prevalence rates in this population assessed using structured diagnostic assessments [3,12].

To inform treatment development, an important next step to build on these findings will be assessing the longitudinal and/or momentary contributions of WS and its internalization to PTSD-S, as well as objectively assessed PTSD. Such work also will benefit from examining whether antecedent trauma increases subsequent vulnerability to adverse WS-related sequelae, as inferred by a recent study that found ACEs associated with internalized WS among women with overweight/obesity [52]. It is possible that earlier trauma may lead to weight gain and obesity (e.g., via binge eating) [53], which, in turn, increases exposure to experienced WS and internalized WS, compounding PTSD-S. While beyond the parameters of this work, these findings suggest that future research may benefit from further examining how specific ACEs, criterion A traumas, and experienced and internalized WS interact over time to shape the development of PTSD-S as well as varied postoperative trajectories. Further, assessing the intersectionality of WS with other traumatic and/or minority stressors in the prediction of concerning postoperative sequelae, including depression, substance use, and suicidality/self-harm, is also a critical area for future research.

Pending the outcomes of such research, and alongside broader multilevel efforts to shift systemic WS, it is possible that individual-level approaches that address healthy coping with experienced WS while buffering the impact of internalized WS may prove helpful in reducing WS-related PTSD-S. To date, an array of approaches has shown early promise in reducing internalized WS, including targeted cognitive behavioral therapy with standard behavioral weight loss [54] and approaches aligned with a weight-inclusive approach, including acceptance and commitment therapy, compassion-focused therapy [55], and mindful yoga [56].

Strengths and limitations

This study has several notable strengths. Very few, if any, studies to our knowledge have examined correlates of PTSD-S in any bariatric surgery patient population, including in relation to WS. Our control of ACEs and patient reports of trauma (i.e., criterion A events) as well as depressive/anxiety symptoms in a participant subset, increases our confidence that the observed association between experienced WS and PTSD-S is not merely due to antecedent traumatic stressors uncaptured by our measurements. Moreover, our assessment of most study variables, including ACEs, through a survey administered independent of the psychological evaluation for surgery suggests that corresponding measures may be less amenable to underreporting biases previously observed in this population [50]. Last, more than one third of our participants self-identified as non-White, providing increased sample demographic diversity relative to prior work, if also pointing to the need for analyses that address intersectionality of identities in relation to health and offering room for improvement. WS may interact with other minority stressors in contributing to behavioral health outcomes, and traumatic stressors may further impact these associations, an important focus of future research.

Several limitations also warrant mention. Use of a cross-sectional design limits our ability to ascertain directionality or causality of the assessed constructs. Relatedly, sample generalizability is limited given our adaptation of the experienced WS scale (SSI-B) and study advertisements for individuals with “adverse interpersonal experiences.” There are also limitations related to our assessment of provisional PTSD and PTSD-S. The PCL-C is a dated measure and has been replaced by the PCL-5, which corresponds to DSM-5 PTSD criteria [15]. Additionally, the PCL-C is not anchored in relation to a specific trauma, although we endeavored to address this limitation by using (1) the symptom cluster scoring method and (2) a conservative PTSD-S severity score of ≥35 and then assessing the proportion thereof reporting criterion A at the psychology assessment for surgery [57]. Yet trauma history was indirectly queried at the psychological assessment, potentially compounding underreporting and contributing to the observed experienced WS and PTSD-S association. It is recommended that future research use the gold standard assessment for PTSD (Clinician-Administered PTSD Scale [CAPS]) or structured interviews to provide a more comprehensive assessment [57]. Future research would benefit from anchoring assessment of PTSD-S specifically in relation to a qualifying criterion A event (e.g., using the widely used clinical screening tool PCL-5 with the criterion A component or with the Life Events Checklist and Extended Criterion A) [58], weigh-trelated bias, and more severe forms of weight-related stigma (e.g., weight-related verbal or physical abuse).

It is also important to note that only a third of potentially eligible participants completed study surveys, pointing to self-selection bias and limiting the generalizability of findings. Indeed, our advertisement of a study on “adverse interpersonal experiences” may have yielded a sample self-selecting with relatively greater levels of trauma and stigma than the overall bariatric surgery patient population or those more comfortable openly reporting such experiences. An important question for future research is exploring how the observed findings may differ or replicate in the overall population of bariatric surgery patients, attending to a lens that accounts for intersectionality of diverse identities. Last, the role of depression should be considered in these sequelae because it may predict increased vulnerability to the development of WS-related PTSD-S symptoms and/or emerge a downstream outcome.

Conclusion

The etiology of postoperative elevations in depression, substance use disorders, and suicidality and self-harm among subsets of bariatric surgery patients remains poorly understood. PTSD and PTSD-S represent an understudied risk factor that is associated with each of these outcomes. Better understanding of contributing mechanisms will aid theory-informed research and development of screening, prevention, and treatment targets. Our study observed WS-linked factors to account for greater variance in PTSD-S when compared with adverse childhood experiences and history of criterion A traumatic stressors in individuals seeking bariatric surgery. Alongside ongoing critical efforts to reduce WS at structural and interpersonal levels, future research is needed to determine longitudinal associations between these factors and temporality and clinical implications.