Disordered eating following bariatric surgery: a review of measurement and conceptual considerations

Abstract

Eating disorders are associated with significant medical morbidity and mortality and serious psychological impairment. Individuals seeking bariatric surgery represent a high-risk group for evidencing disordered eating and eating disorders, with some patients experiencing the persistence or onset of disordered eating postsurgery. This review synthesizes the available literature on problematic or disordered eating in the bariatric field, followed by a review of measurement and conceptual considerations related to the use of eating disorder assessment tools within the bariatric population.

Individuals seeking bariatric surgery may report engaging in certain types of disordered eating and meet eating disorder criteria. The current Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) [1] recognizes feeding disorders, including the eating disorder diagnoses of (1) anorexia nervosa (AN); (2) bulimia nervosa (BN); and (3) binge-eating disorder (BED), as well as 2 residual categories in which the full diagnostic criteria for AN, BN, or BED are not met: (4) other specified feeding or eating disorder (OSFED); and (5) unspecified feeding or eating disorders (USFED). Eating disorders are associated with significant medical morbidity and mortality [2] and serious psychological impairment [3], including a 5-fold increased risk of suicide attempts [4]. According to the Longitudinal Assessment of Bariatric Surgery–3 (LABS-3) study, 6.1% of participants seeking bariatric surgery met DSM-IV diagnostic criteria for BED, while 1.2% met BN criteria, exceeding rates reported in the general population [5,6]. Disordered eating behaviors were also quite common in the LABS-3 study: 40.4% reported loss-of-control (LOC) eating, 30.5% reported binge eating, 16.5% reported night eating, and 6.4% reported at least 1 compensatory behavior during the prior 6 months [7]. Presurgical problematic eating behaviors, however, have had little prognostic significance for postsurgical weight outcomes [7–9]. Although many of these behaviors improve after bariatric surgery, disordered eating behaviors may persist, recur, or develop de novo. In fact, studies suggest that various forms of disordered eating after bariatric surgery are associated with suboptimal weight loss trajectories and/or greater weight regain [9]. Beyond weight outcomes, changes in disordered eating may also have a cascade of effects on depression, quality of life, self-esteem, and body image [10,11].

Disordered eating behaviors (e.g., binge eating and LOC eating) and cognitions (e.g., overvaluation of weight or shape) that are clinically distressing and/or associated with impairment, but do not meet full criteria for an eating disorder, are also concerning. For instance, recent research suggests that LOC eating—that is, eating episodes occurring with a concurrent subjective sense of feeling out of control, regardless of the quantity of food consumed—is associated with psychological impairment [12,13]. Regarding disordered eating cognitions, overvaluation of weight or shape, which describes a sense of personal worth based on one’s weight or shape, is similarly associated with significant negative sequalae (e.g., greater functional impairment and depressive symptoms) [14,15].

This review synthesizes the available literature on disordered and problematic eating after bariatric surgery, followed by a review of measurement and conceptual considerations related to the use of eating disorder assessment tools within the bariatric population. Although the terms “problematic eating” and “disordered eating” are sometimes used interchangeably in the literature, problematic eating in this review refers to normative eating behaviors that are inconsistent with healthful weight loss goals (e.g., overeating), whereas disordered eating refers to behaviors associated with eating disorders (e.g., LOC eating or binge eating) or cognitions that do not meet full criteria for a standalone DSM-5 eating disorder.

Eating behavior after bariatric surgery

Normative or recommended eating after bariatric surgery

Understanding normative eating after bariatric surgery is essential to operationalize and identify postoperative problematic and disordered eating. In the early postoperative period (first 6 mo), most patients are unable to consume large amounts of food due to metabolic and anatomical changes. Patients are expected to make significant lifestyle changes as to how, what, and when they eat to tolerate food intake and have progressive and sustained weight loss. The American Society for Metabolic and Bariatric Surgery has outlined nutritional guidelines for bariatric patients [16,17]. Despite long-term weight loss success achieved by many who undergo bariatric surgery [18,19], nonadherence to recommended guidelines is common [20].

Problematic and/or disordered eating after bariatric surgery

Binge-eating and LOC eating

Objective binge episodes (OBEs) are defined as a discrete period in which a person eats an unusually large quantity of food and has a sense of losing control. OBEs, required for the BED and BN diagnoses, are rarely physically possible early after bariatric surgery, but such problems can recur over time [12,21]. However, patients may still experience a sense of LOC while eating an amount of food that is not objectively large (i.e., subjective binge-eating episodes). Such problems may be reported shortly after surgery, or after an interim period [22]. Those most at risk appear to be patients who met BED criteria or engaged in OBEs prior to surgery [19]. Studies [22,23] estimate that 10%–39% of patients develop LOC eating by 2 years postsurgery, although this problem may emerge as early as 4–6 months postsurgery [13,23,24]. LOC eating is associated with other postsurgical problems [23] (e.g., vomiting), as well as an elevated risk for psychopathology (e.g., depression), less weight loss, and greater weight regain [9]. In 7-year prospective data obtained as part of the LABS-3 study using structured assessment instruments, postsurgery LOC eating was confirmed as a risk factor for less long-term weight loss and greater weight regain [8].

Grazing/picking and nibbling

Grazing or picking and nibbling (P&N) is a frequent, albeit difficult to define, problem postsurgery. Conceição and colleagues [25] surveyed 24 authors who had published research on grazing and, based on the responses, concluded that characteristics associated with the concept were highly variable among these researchers. They proposed a composite definition of P&N: an eating behavior characterized by the repetitive eating of small/modest amounts of food in an unplanned manner and/or not in response to hunger/satiety [25,26]. A difficulty with differentiating P&N from normative eating behavior is that patients are often instructed to eat multiple small meals throughout the day postsurgery. A differentiator, however, may be that P&N is often engaged in without planning. To date, the literature suggests that P&N behavior after surgery is common, affecting perhaps 18.6%–46.6% of patients [25,26], and is also linked to attenuated weight loss [27]. Some patients develop new-onset P&N postsurgery without a history of BED or OBE. Notably, P&N is widely prevalent in obesity and eating disorders in general, suggesting a considerable lack of specificity to those who have undergone bariatric surgery [28].

Eating in the absence of hunger or beyond satiety

Similarly, eating in the absence of hunger (i.e., when not hungry) or eating beyond satiety may have some negative prognostic significance for weight loss after surgery [27]. Early in the postoperative period, hunger cues diminish; thus, eating in the absence of hunger might be normative during that period, but become more problematic over time. In addition, eating when not physically hungry is a behavioral indicator of LOC eating. Yet, when this behavior develops into disordered behavior after bariatric surgery is unclear.

Night Eating Syndrome

Currently, there is limited research on Night Eating Syndrome (NES) after bariatric surgery [29]. Like P&N, the field currently lacks an agreed-upon definition of NES. Most often, NES is defined based on the timing of eating rather than the size of eating episodes. The focus on an eating pattern rather than the amount of food consumed may increase the utility within bariatric populations, as detailed above, for whom limited gastric capacity impacts the quantity of food consumed at 1 time. One study suggested that postoperative night eating symptoms improved among patients who had a preoperative depressed mood versus those without it [29], but the relationship between mood and NES is complex; other factors, including sleep, may also play a role [29,30].

Excessive dietary restriction

Very little is known regarding excessive dietary restriction following bariatric surgery. Most studies have examined dietary restraint, which examines the intention to restrict as opposed to actual restriction [8,31]. Although uncommon, a smaller percentage of individuals may develop restrictive-type eating disorders, such as AN. In some cases, patients present with a critically low body mass index (BMI) and medical co-morbidities, whereas others may not have a low BMI but nonetheless have lost excessive weight, evidence starvation, and meet criteria for AN, except for the criterion related to low weight [32,33]. Yet, there are no prevalence data available on the development of such eating disorders or other eating disorder classifications, including OSFED and USFED, after bariatric surgery.

Surgically related behaviors

Bariatric procedures may result in surgery-specific behaviors that may or may not be a function of disordered eating. To avoid over-pathologizing normative behaviors or potential consequences of bariatric surgery, assessing the cause and function of the behavior is key. For example, “dumping syndrome,” which is characterized by dizziness, diaphoresis, flushing, palpitations, and diarrhea, can occur frequently after bariatric surgery due to postsurgical behavior changes [34], and is more likely to occur in patients undergoing Roux-en-Y as compared with laparoscopic sleeve gastrectomy [35]. This is sometimes referred to as early dumping, and such symptoms can be precipitated by the ingestion of sweet, high-carbohydrate, or other foods, and generally occurs in the first 30 minutes after eating. Late dumping, probably better referred to as reactive hypoglycemia, is also characterized by these symptoms, as well as symptoms of neuroglycopenia, such as slurred speech, tremor, and shaking [35]. This generally occurs later after eating, is indicative of critically low blood glucose that affects the brain, can be quite dangerous, and requires appropriate care. Dumping, in and of itself, is not a form of disordered eating, but a byproduct of the surgery. However, if a patient decides to intentionally induce dumping to try to eliminate food or calories through diarrhea, disordered eating may be inferred. In a small case series study, 2 patients described dumping as an intentional form of purging [36], although data in this area are scarce.

“Plugging,” or dysphagia, is characterized by a sense of food getting “stuck” after eating, and may be cleared by waiting or fluid ingestion [37]. This can be problematic for patients, who may self-induce vomiting to alleviate the sensation of plugging, but it is not a sign of disordered eating per se. Notably, however, LOC eating occurring at least once weekly was associated significantly with weekly plugging in the LABS-3 consortium study [37].

Vomiting after surgery is another complex behavior with multiple potential causes and functions. Spontaneous vomiting—that is, vomiting that is not self-induced—can occur after bariatric surgery [38], but probably occurs less over time. Vomiting that appears to be spontaneously caused by bariatric surgery, however, may be due to intentional efforts to eat food(s) that cause vomiting, with the desire to influence weight or shape. Intentional self-induced vomiting may also occur in an attempt to influence weight or shape, or to alleviate physical comfort. Differentiating between various forms of vomiting is important; however, cases of self-induced vomiting for weight control have rarely been reported in the bariatric population [19].

Chewing and spitting food sometimes occurs, at times as a means to taste foods that may not be tolerated if swallowed, but also in other samples of patients with eating disorders, as well as in the general population [39]. After bariatric surgery, individuals might engage in chewing and spitting behavior to avoid gaining weight, to influence shape, to get the sensation or enjoyment from chewing, and/or to avoid physical discomfort related to the surgery. Prevalence data on this behavior after bariatric surgery are lacking.

Disordered eating assessment

Measurement

Rigorous assessment tools are needed to measure the array of aforementioned eating behaviors to facilitate accurate identification of eating disorders and disordered eating. The Eating Disorder Examination (EDE) interview [40] is the most commonly used semi-structured interview to assess disordered eating behaviors, cognitions, and general eating disorder psychopathology. The EDE was originally developed for AN and BN, but was adapted for the rigorous LABS-3 psychosocial study into what is referred to as the Bariatric Surgery Version (EDE-BSV) [19]. The EDE-BSV includes the original EDE items and assesses rich clinical data, including items which might help differentiate behaviors that are disordered from behaviors that are expected after bariatric surgery. The EDE-BSV also incorporates other self-report measures, such as the Night Eating Scale [41], into an interview format. A single study investigated the interrater reliability of the EDE-BSV, with strong support for the traditional EDE subscales and for the newly defined bariatric overeating and LOC eating episodes [42]. Agreement was less robust (κ = .60) for bariatric overeating episodes as compared to LOC eating episodes. Notably, the BSV items were not examined, and other psychometric data of the EDE-BSV are largely unavailable [42]. In addition, because the interview requires extensive training and time, many researchers and clinics elect to use self-report measures instead.

Table 1 summarizes the most commonly used eating disorder self-report measures in the bariatric field, with detailed descriptions (e.g., psychometric properties, administration time, and costs) of each measure (see Parker et al. [43] for a systematic review of these measures at the pre-bariatric phase). Importantly, very few studies have evaluated the psychometric properties of these measures at the pre- or postoperative bariatric stage. At the preoperative stage, comparisons between the EDE interview and the self-report questionnaire version (EDE-Q) [44] suggest lower levels of agreement between the 2 assessment tools than those previously observed in other clinical populations [45]. Furthermore, the limited literature does not support the use of the original factor structures of many measures at the pre- or postsurgical phase [43]. Without a valid factor structure, components like subscale scores are not meaningful. Failure to replicate factor structures suggests that the bariatric surgery population may have unique characteristics that are distinct from those of traditional eating disorder samples. Indeed, in the first set of studies examining the factor structure of the EDE-Q among individuals seeking bariatric surgery, the original factor structures were not supported; rather, alternative factor structures were found [46]. Parker and colleagues [47] replicated and extended these findings at the pre- and postsurgical stages; analyses did not support the original factor structures of widely used eating disorder measures, including the EDE, EDE-Q, and Three-Factor Eating Questionnaire (TFEQ) [48]. Importantly, however, these findings were based on individuals who had undergone the laparoscopic adjustable gastric banding procedure [47]. Only 1 study has examined the factor structure of an eating disorder measure in the postoperative phase following sleeve gastrectomy. Carr and colleagues [49] found that among individuals with LOC eating following laparoscopic sleeve gastrectomy, the Eating Loss of Control Scale (ELOCS) [50] generated 2 factors, instead of the traditional 1 factor supported in nonbariatric populations [50]. To our knowledge, no study has examined the factor structures of eating disorder measures among individuals who underwent Roux-en-Y gastric bypass.

Table 1.

Available psychometric properties and administration information for commonly used eating disorder measures in the bariatric field

Scale	Items	Description	Available psychometric properties				Administration information
			Internal consistency	Test–retest	Factor structure	Means/SD, cut points	Time to complete, min	Cost to use	Requires permission to use
Binge Eating Scale (BES) [82]	16	Assesses binge-eating behaviors and binge-related feelings and cognitions Each item consists of 4 multiple-choice statements reflecting a range of severity (score 0–3) Clinical cutoff scores are provided	.83–.90 [68–73]	Not reported	Not tested	Means/SD (Pre/Post) Validated cut scores (Pre) [71]	5	No	No
Dutch Eating Behavior Questionnaire (DEBQ) [89]	33	Assesses aspects of eating behavior Items are assessed on a 5-point Likert scale Includes 2 factors: (1) eating in response to specific emotions and (2) eating in response to diffuse emotional states Includes 3 subscales: (1) Restrained Eating, (2) Emotional Eating, and (3) External Eating	Not reported	Not reported	Not tested	Means/SD (Pre/Post)	10	No	No
Eating Disorder Diagnostic Scale (EDDS) [90]	22	Assesses DSM-4-TR criteria for AN, BN and BED Consists of a combination of Likert scores, dichotomous scores, frequency scores, and open-endedquestions (e.g., height and weight)	.86 [74]	Not reported	Not tested	None	5–10	No	No
Eating Disorder Examination Questionnaire (EDE-Q) [44]	28–41 or 7*	Adapted from the Eating Disorder Examination (EDE) semi-structured interview Assesses frequency of each listed behavior over the last 28 days. Includes 4 subscales: Restraint, Weight Concern, Shape Concern, and Eating Concern	Original .60–.89 [46,56,75–77] Revised/Alternative .69–.98 [46,47,56,78]	Not reported	Original not supported [46,47,56,78]	Means/SD (Pre/Post)	5–15	No	No
Eating Disorder Inventory (EDI) [91]	64	Assesses psychological and behavioral traits common in AN and BN Items are assessed on a 6-point Likert scale Includes 8 subscales: Drive for Thinness, Bulimia, Body Dissatisfaction, Ineffectiveness, Perfectionism, Interpersonal Distrust, Interoceptive Awareness, and Maturity Fears	.38–.88 [53,79]	Not reported	Not tested	Means/SD (Pre/Post) Inferential comparisons Bariatric (Post) to general and ED sample [97]	20	Yes	No
Eating Loss of Control Scale (ELOCS) [50]	18	Assesses frequency and severity of loss-of-control eating Items include frequency counts and other questions, rated on an 11-point Likert scale	.88–.92 [49]	Not reported	Original not supported [49]	Means/SD (Post)	10	No	No
Emotional Eating Scale (EES) [92]	25	Assesses desire to eat in response to 25 negative affective states Items are measured on an 5-point Likert scale Includes 3 subscales: Anger/Frustration, Anxiety, and Depression	.72–.92 [77,80,81]	Not reported	Not tested	Mean/SD (Pre/Post) Means for upper/lower quartile (Pre) [80]	5	No	No
Night Eating Questionnaire (NEQ) [41]	14	Assesses the behavioral and psychological symptoms of night eating syndrome Items are assessed on a 15-pointLikert scale Provides cutoff scores for probable night-eating syndrome	.80 [82]	Not reported	Not tested	Mean/SD (Pre/Post)	5–10	No	No
Questionnaire on Eating and Weight Patterns (QEWP) [54]	28	Assesses the presence or absence of binge-eating episodes, the frequency of such episodes, and additional required features for the diagnosis of DSM-4 BED criteria Items include both dichotomous queries and frequency assessments	Not reported	Not reported	Not tested	Proportion BED/LOC (Pre/Post) Agreement (Kappa) with other Dx measures (Pre) [58,93,94]	5	No	No
3-Factor Eating Questionnaire† (TFEQ) [48]	51	Assesses cognitive restraint, disinhibition, and hunger responsiveness Items are presented in both True-False and 5-point Likert response formats	.73–.87 [47,78,83]	Not reported	Original not supported [47,78]	Mean/SD (Pre/Post)	20	No	No
Yale Food Addiction Scale (YFAS) [95]	25	25-item instrument assessing eating behavior in relation to highly palatable foods (i.e., high-sugar and/or high-fat foods) Majority of items provide a Likert response format of increasing frequency Can be scored in 2 ways: (1) A symptom count version indicating the number of dependence symptoms experienced in the past 12 mo; and (2) a diagnostic threshold version determining whether 3 or more symptoms are present and whether there is clinically significant impairment or distress	.82–.94 [55,81,84–89]	Not reported	Original factor structure replicated [55,90]	Mean/SD (Pre/Post)	10	No	No

SD = standard deviation; DSM-4 = Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition; DSM-4-TR = Diagnostic and Statistical Manual of Mental Disorders, Forth Edition Text Revision; AN = anorexia nervosa; BN = bulimia nervosa; BED = binge-eating disorder; ED = eating disorder; LOC = loss-of-control; Dx = diagnosis.

^*The 7-item version of this scale has different subscales than those included in the description.

^†Measure was later revised and re-named to “Eating Inventory” [96].

The scarcity of psychometric data negatively impacts both the clinical utility of measures and the generalizability of findings. Two general concepts, reliability and validity, help determine the quality of the measures used and, ultimately, the research findings. Reliability refers to the consistency of a measure [51], and it includes internal consistency (degree to which all items measure the same construct) and test–retest (consistency across time or raters) [52]. Within the bariatric field, internal consistency is inconsistently reported. Estimates range from weak (.38) [53] to excellent (.98) consistency [47]. Caution is warranted when using scales that demonstrate weak internal consistency, as the single scale (or subscale) is likely not measuring a single unitary construct. In some cases, revised scales developed using bariatric samples have shown improvements in reliability, including revised versions of the EDE-Q [47]. To our knowledge, the test–retest reliability of eating disorder measures in the bariatric literature has been completely untested. This suggests that we do not know whether the eating disorder measures would produce similar findings if a given measure was repeated [52]. The evidence suggests that testing the reliability of most measures used within the bariatric field is warranted.

Validity refers to the degree to which the underlying latent construct is the cause of the item scores, and it conceptually reflects whether or not the measure is assessing what it purports to measure [52]. The most common forms of validity include content validity, criterion validity, and construct validity. Content validity refers to the degree to which a scale adequately samples from the universe of possible items. This is conceptually difficult to establish, but review by experts can increase confidence in content validity [52]. Within the bariatric field, the majority of measures were developed by eating disorder experts broadly. Although limited, measures such as the EDE-BSV and the Repetitive Eating Questionnaire (RepEAT-Q) were developed by bariatric experts, which increases confidence in the content validity. Criterion-related validity describes an empirical association between the measure of interest and related and well-validated measures [52]. For measures intended to diagnose types of eating disorders, establishing the level of agreement is a form of criterion validity. Based on the studies reviewed, only the Questionnaire on Eating and Weight Patterns [54] has available estimates of criterionvalidity. Construct validity refers to the degree to which the underlying latent construct is the cause of item scores, and it can include a range of psychometric properties, such as a replicated factor structure. Like reliability, untested or poor replication of a factor structure can seriously undermine confidence in findings.

Finally, in addition to the measurement issues specific to bariatric surgery populations, measurement within the field suffers from some of the main limitations observed in self-report measurement broadly (e.g., evidence to support use of the measures in heterogenous samples). Whether constructs such as body image concerns or other cognitive symptoms of eating disorder psychopathology differ across age, gender, and racial and ethnic groups in those undergoing bariatric surgery remains unknown. In order to better characterize either similarities or dissimilarities across groups, more information is needed related to the validity of self-report measurements within diverse bariatric samples.

Taken together, relatively few studies have examined psychometric properties of eating disorder measures in the bariatric field, particularly after bariatric surgery, when eating behaviors are nuanced and potentially more complex compared to other populations that do not have an altered gastrointestinal anatomy. There is limited psychometric testing of disordered eating measurement tools and limited support of original factor structures in the bariatric field. Indeed, of the 12 commonly used self-report measures listed in Table 1, factor analyses were conducted for only 4 measures (EDE-Q, ELOCS, TFEQ, and Yale Food Addiction Scale [YFAS]); only 1 study with 1 of these measures—the YFAS, which is a controversial construct not clearly defined as disordered eating—replicated the original factor structure [55]. Future studies with large and diverse sample sizes are needed to appropriately and accurately examine factor structures, and greater rigor in reporting psychometric properties (i.e., various forms of reliability and validity, as well as means and standard deviations for the overall participant group and subgroups as appropriate) is warranted. Based on the limited psychometric literature to date, the use of the EDE-Q [47,56] and TFEQ with alternative factor structures is recommended to examine eating disorder psychopathology prior to bariatric surgery, while the ELOCS with 2 factors is recommended to examine LOC eating after bariatric surgery [49]. Importantly, however, the authors note the need for replication of the structure within independent bariatric samples.

Conceptual considerations

In addition to psychometric limitations of eating disorder measures in the bariatric field, a discussion of conceptual measurement issues is warranted for this population. The majority of disordered eating examined among bariatric populations has included eating behaviors thought to promote suboptimal weight outcomes, such as LOC eating. Little is known about the entire spectrum of disordered eating, including more restrictive disordered eating behaviors and cognitions (e.g., preoccupation with weight/shape), which can be impairing. This is likely because the original eating disorder diagnostic criteria were not developed with bariatric surgery in mind and are not well suited to characterize a full range of eating pathology in postsurgery patients.

Indeed, many of the well-known eating disorder measures may not conceptually apply to postoperative bariatric patients. For example, the EDE interview and self-report versions include items designed to capture disordered dietary restraint. The primary dietary restraint items ask: “have you been consciously trying to restrict the amount of food you eat, whether or not you succeeded?”; “have you been consciously trying to avoid certain foods, whether or not you succeeded?”; and “have you been trying to follow certain definite rules regarding your eating?” Higher scores are indicative of greater eating-disorder psychopathology. However, one would expect individuals to endorse such items after bariatric surgery. In fact, scoring higher might be suggestive of greater adherence to postoperative dietary guidelines and may be adaptive in some instances, at least conceptually. Similarly, a higher score on weight dissatisfaction is considered more severe or problematic, given the original intention of this measure with more restrictive eating disorders. For individuals seeking bariatric surgery, however, higher scores of weight dissatisfaction might not be problematic or might not capture the same construct as originally intended among individuals with extremely low weights.

The field also lacks consensus on how to define an OBE in a postbariatric population. According to the DSM-5, the amount of food eaten is objectively larger than most others in a similar circumstance would eat. In this circumstance or context (having undergone bariatric surgery), defining “an unusually or objectively large” amount of food is complex and is a matter of debate [57]. “Unusually or objectively large” might be influenced greatly by the duration since having the operation. This has important implications, because rates of OBEs postsurgery could be higher if “unusually large” is defined based on what constitutes a portion size for someone who has undergone bariatric surgery as opposed to someone without a bariatric surgery history.

Cognitive features of eating disorders are often over-looked in the bariatric literature. For example, measurements of eating disorder psychopathology among bariatric surgery patients would be incomplete without at least a brief assessment of overvaluation of weight and shape. Several existing measures commonly used among bariatric surgery patients (e.g., EDE and EDE-Q) include 2 applicable Likert-scale questions (i.e., “has your weight influenced how you think about [judge] yourself as a person?” and “has your shape influenced how you think about [judge] yourself as a person?”), and questions similar in content and brevity would likely benefit future eating disorder measurement development efforts among this population. Similarly, other cognitive constructs, such as preoccupation with food/eating or weight/shape, might shed light on important risk factors or maintenance mechanisms of disordered eating after bariatric surgery.

Although the existing adolescent research suggests parallels to the adult literature—namely, general decreases in eating disorder pathology following surgery—there are still outstanding questions. For example, it is not clear whether the self-report measures that exist for younger populations are developmentally appropriate [11]. Additional research on a wider breadth of behaviors with larger samples of adolescents would be important to help answer outstanding questions about the presentation of youth receiving bariatric surgery.

Gaps and future research implications

Taken together, while reviewing current and future research in this area, it is important to keep in mind the following limitations. First, the majority of disordered eating examined among bariatric populations has included eating behaviors thought to promote suboptimal weight outcomes, such as LOC eating. Little is known about the entire spectrum of disordered eating, including more restrictive disordered eating behaviors, compensatory behaviors, and cognitive features of disordered eating. Second, the original eating disorder diagnostic criteria were not developed with bariatric surgery in mind and are not well suited to characterize eating pathology in postsurgery patients. Third, current eating disorder measures used in bariatric surgery research were developed with nonbariatric patients, many of which included predominantly White women presenting with restrictive eating. As such, the nuances of disordered eating behavior after bariatric surgery may not be captured fully by the field’s current assessment instruments.

With these limitations in mind, we offer the following recommendations. First, efforts should be made to study the psychometric properties of existing measures at multiple time points (before surgery, acute postsurgery, and long-term follow-up). For example, psychometric testing of the EDE and EDE-BSV is needed to determine the utility of both instruments before and after bariatric surgery. Second, efforts should be made to develop new reliable and valid measures specifically for bariatric surgery. There are a few noteworthy examples of this approach in assessing health-related quality of life among bariatric patients, including the Bariatric and Obesity Specific Surgery [59] and the Quality of Life for Obesity Surgery [60] instruments. Though not widely used, both measures include eating behavior subscales and demonstrate psychometric strengths, including estimates of internal consistency [59,60], test–retest reliability [59], improved content validity [59,60], and a supported factor structure among bariatric patients [59,60]. In addition to considering health-related quality of life, new measures are needed to understand eating-disordered cognitions and behaviors and the bio-psychosocial factors that motivate patients to initiate and maintain these behaviors before and after bariatric surgery. More research is also needed to characterize normative eating after surgery, which will inform assessments of problematic or disordered eating after surgery.

In addition to investigating existing self-report measures and developing new ones, attention should be paid to methods that might reduce the limitations associated with traditional self-report measures, including recall bias and social desirability/approval bias [61]. Methods such as daily diaries can reduce recall biases and may have improved ecological validity, though some data suggest the impact of recall bias is less for people with eating disorders as compared with other clinical groups [62,63]. Photographic food records have shown some promise [64,65]. Furthermore, laboratory-based feeding paradigms would be useful to help gain a better understanding of normative and disordered eating before and after bariatric surgery. Emerging technology, such as sensors and other objective measures, have the potential to assess patterns of eating behavior [66]. Early evidence, however, suggests problems with adherence that would need to be addressed before wide-spread adoption could be achieved [67,68].

Conclusions

To date, disordered eating is common among individuals seeking bariatric surgery and tends to decrease significantly following surgery. Existing literature suggests that postsurgical disordered eating is associated with attenuated weight loss or weight regain following bariatric surgery. Relatively few data, however, are available on the full spectrum of disordered eating, including more restrictive eating and diagnoses. Further, our current eating-disorder classification system and assessment tools are limited for pre- and postoperative bariatric surgery populations. The current tools, which were originally designed for individuals with more restrictive, low-weight eating disorders, might not accurately or reliably capture complex and nuanced eating after bariatric surgery. Thus, despite the proliferation of research in this area over the past 2 decades, results should be interpreted with caution due to the lack of psychometric data of many of these measures among bariatric samples. Future research should examine comprehensive psychometric properties (i.e., various reliability and validity indices) of commonly used eating-disorder measures at preoperative and postoperative phases to develop greater confidence in the extant data and provide future measurement directions. Without appropriate psychometric testing, it is unclear whether new measures are needed for this population. Importantly, however, even with psychometric testing of current measures, it will be important to consider whether current measures miss specific disordered-eating characteristics after bariatric surgery. Future research should consider rigorous scale-development techniques for disordered eating that are unique to bariatric surgery, which would ultimately aid in early detection of problematic features and inform treatment development and recommendations. Moreover, individuals with disordered eating may have unique needs in the postoperative period; more research is needed examining the temporal relationships between disordered eating and psychosocial sequalae postsurgery, utilizing tools that are validated for use with bariatric surgery patients. Objective methods of measuring disordered eating, along with complementary self-report measures, may elucidate our understanding of the unique needs of individuals undergoing bariatric surgery. Finally, mixed methods, including qualitative methods, are needed to assess disordered eating following bariatric surgery.