|Home | About | Journals | Submit | Contact Us | Français|
Anecdotal reports suggest bariatric surgery may increase the risk of alcohol use disorders (AUD), but prospective data are lacking.
Determine the prevalence of pre- and postoperative AUD, and independent predictors of postoperative AUD.
Longitudinal Assessment of Bariatric Surgery-2 is a prospective cohort study of adults who underwent bariatric surgery at 10 U.S. hospitals. Of 2458 participants, 1945 (78.8% female, 87.0% white, median age=47 years, median body mass index=45.8 kg/m2) completed pre- and postoperative (1 and/or 2 years) assessments between 2006–2011.
Past year AUD symptoms determined with the Alcohol Use Disorders Identification Test (indication of ‘alcohol-related harm,’ ‘alcohol dependence symptoms,’ or score ≥8).
The prevalence of AUD did not significantly differ from pre- to 1 year postoperative (7.6% vs. 7.3%; p=.98), but was significantly higher at 2 years (9.6%; p<.01). Male sex (adjusted odds ratio [OR], 2.1; 95% confidence interval [CI], 1.5–3.0; p<.0001), younger age (OR, 1.3; CI, 1.03–1.7 per 10 years younger with pre-op AUD; p=.03; OR, 2.0; CI, 1.7–2.3 per 10 years younger without pre-op AUD; p<.0001), smoking (OR, 2.6; CI, 1.2–5.6; p=.02), regular alcohol consumption (OR, 6.4; CI, 4.2–9.7; p<.0001), AUD (e.g., OR, 11.1; CI, 7.7–16.1 at age 45; p<.0001), recreational drug use (OR, 2.4; CI, 1.4–4.1; p<.01) and lower ‘belonging’ interpersonal support (OR, 1.1; CI, 1.04–1.2; p<.01) preoperatively, and undergoing a Roux-en-Y gastric bypass (OR, 2.1; CI, 1.4–3.1; p<.001; reference=adjustable gastric band) were independently related to an increased odds of AUD postoperatively.
In this cohort, the prevalence of AUD was greater in the second postoperative year than preoperatively or in the first postoperative year, and was associated with preoperative AUD and regular alcohol consumption, and undergoing Roux-en-Y gastric bypass.
As the prevalence of severe obesity (defined as body mass index [BMI] ≥ 35 kg/m2) rises in the United States,1 it is becoming increasingly common for health care providers and their patients to consider bariatric surgery, the most effective and durable treatment for severe obesity.2 Although bariatric surgery may reduce long-term mortality,3–4 and carries a low risk of short-term serious adverse outcomes,5 safety concerns remain. Anecdotal reports suggest that bariatric surgery may increase the risk for alcohol use disorder ([AUD]; alcohol abuse and alcohol dependence).6 However, only three studies have examined AUD pre- and post- bariatric surgery. Mitchell et al.7 attempted to contact 100 Roux-en-Y gastric bypass (RYGB) patients 13–15 years postoperatively. Of 8 deaths, one was attributed to alcoholism complications. Of 78 patients who agreed to a diagnostic interview, 10 (12.8%) vs. 6 (7.7%) reported AUD, pre- vs. postoperatively, respectively. Ertelt et al.8 mailed a survey to 250 patients 6–10 years following RYGB. Of respondents (n=70), the number of participants who had AUD symptoms pre- (n=6; 8.6%) versus post- (n=7; 10.0%) RYGB was similar. Importantly, the majority (71.4%) with AUD symptoms postoperatively had AUD symptoms preoperatively. Suzuki et al.9 recruited 51 of 530 targeted patients two or more years following RYGB or laparoscopic adjustable gastric banding (LAGB). Preoperative AUD was determined by retrospectively examining clinical psychiatric evaluations. Postoperative AUD was determined by diagnostic interview. No participants met criteria for current AUD preoperatively, and no LAGB patients (n=23) reported current AUD postoperatively. However, 6 of the 28 (21.4%) RYGB patients did, 83.3% of whom had a history of preoperative AUD with remission at surgery.9 Given the limitations of these studies (low participation rate,8;9 small sample size,7–9 retrospective assessments of AUD,7–9 and different time frames7–9 and assessment methods9 for pre- and postoperative periods) it remains unclear whether bariatric surgery influences risk of AUD.
There is evidence that some bariatric surgical procedures (i.e. RYGB and sleeve gastrectomy) alter alcohol pharmacokinetics. Given a standardized quantity of alcohol, postoperative patients reach a higher peak alcohol level compared to case controls10–11 or their preoperative values.12–13 In addition, some studies have shown that postoperative patients reach peak alcohol level more quickly11 or take more time to return to a sober state.10;12;13 Patient surveys have revealed similar changes in alcohol sensitivity following RYGB (feeling intoxicated more rapidly, after drinking less, for longer8;14), as well as more difficulty controlling postoperative alcohol intake.14 Alcohol sensitivity studies have not been done in LAGB patients, most likely because the anatomical and physiological changes from LAGB are less likely to impact alcohol absorption and metabolism. To address limitations in the literature, this study aimed to determine whether the prevalence of AUD changed following bariatric surgery in a large multicenter observational study, comparing reported AUD in the year prior to surgery to the first and second years after surgery. In addition, this study aimed to identify independent predictors of postoperative AUD. We hypothesized that preoperative AUD and undergoing RYGB would increase the likelihood of postoperative AUD. In addition, we hypothesized that many of the factors associated with AUD in the general population15 would be associated with increased odds of postoperative AUD.
The Longitudinal Assessment of Bariatric Surgery-2 (LABS-2) is an observational study designed to assess the risks and benefits of bariatric surgery.16 Patients at least 18 years old seeking a first bariatric surgical procedure by participating surgeons at ten centers throughout the United States were recruited between February, 2006 and February, 2009. All participating centers had institutional review board approval and all participants provided written informed consent. By study enrollment closure (April 2009), 2458 participants attended a preoperative research visit, which occurred after the surgery approval process was complete and within 30 days of their scheduled surgery date, and underwent a bariatric surgical procedure (RYGB, LAGB, sleeve gastrectomy, biliopancreatic diversion with duodenal switch, or banded gastric bypass) as part of clinical care (see Figure 1). The study, #NCT00465829, is registered at www.clinicaltrials.gov.
Measures were collected independently of the surgery approval process and clinical care. Participants were informed that their responses were confidential, although the informed consent document specified that investigators could take steps to prevent serious harm (e.g., if suicidal ideation was reported).
The Alcohol Use Disorders Identification Test (AUDIT)17 is a 10 item test developed by the World Health Organization to assess alcohol use and consequences in the prior 12 months, with well-established validity and reliability.18 A total score (0–40) is calculated using all ten items (see eTable 1), a higher score reflecting greater severity of AUD. Additionally, subsets of items indicate whether respondents are positive for ‘consumption at a hazardous level’ (typically consuming at least 3 drinks per occasion or ever having 6 or more drinks on one occasion), ‘symptoms of alcohol dependence’ (not being able to stop drinking once started, needing a drink in the morning to get going, or failing to meet normal expectations because of drinking), and ‘alcohol-related harm’ (not being able to remember, feeling guilt, injuring someone, or eliciting concern due to drinking). For this analysis participants were categorized as having AUD symptoms (referred to as “AUD” throughout) if their total AUDIT score was at least 817 or if they were positive for ‘symptoms of alcohol dependence’ or ‘alcohol-related harm.’
Anthropometric measurements were made using standardized protocols.19 Sociodemographics were self-reported; race was set to missing for participants who did not self-report their race as one or more of the investigator-defined categories (i.e., white/Caucasian, black/African-American, Asian, American Indian/Alaska Native, Native Hawaiian/other Pacific Islander). For analysis, race categories other than white and black were combined as “other” race due to low representation (see Table 1). Physical and mental health were measured using the norm-based physical component and mental component scores from the Medical Outcomes Study 36-item Short-Form Health Survey (SF-36), higher scores (range 0–100) indicating better functioning.20 Depressive symptomatology was measured with the Beck Depression Inventory (BDI) version 1, a higher score (range 0–63) indicating greater severity.21 Past-year use of recreational drugs, smoking status, and binge eating disorder were assessed with the LABS-2 Behavioral form,16 which includes questions to assess all five criteria for binge eating disorder.30 Treatment for psychiatric or emotional problems (hospitalization or outpatient treatment in the past 12 months or current medication use) was assessed with the LABS-2 Psychiatric and Emotional Test Survey.16 Perceived social support was measured using three domain scores (appraisal, belonging and tangible) from the 12-item Interpersonal Support Evaluation List (ISEL-12), higher scores (range 4–16) indicating greater support availability.23 Several LABS-2 forms (Subsequent Surgery form, Surgeon’s Questionnaire, surgical procedure-specific forms, and Health Care Utilization form) were used to collect information on surgical revisions, reversals and new bariatric procedures that occurred after the initial bariatric procedure and before the 2 year assessment.
Potential selection bias was examined by comparing preoperative characteristics of LABS-2 participants in the analysis sample (n=1945) to those excluded (n=513) for failure to complete the AUDIT preoperatively or during one of the postoperative time points using Pearson’s chi-square test for categorical variables and the Wilcoxon rank-sum test for continuous variables. Descriptive statistics of alcohol-related measures were limited to participants with AUDIT data at all three time points (n=1400). Missing data were assumed to be missing at random (MAR), i.e., the probability of missing depends only on the observed data.24 Differences in distributions of alcohol-related measures over time were tested using generalized linear mixed models (GLMM) using all available observations. Pair-wise comparisons (P<.05) were made between each combination of the three time points.
GLMM was also used to determine predictors of postoperative AUD using all available observations, by modeling AUD as the dependent variable over time (i.e., yes/no at 1 year and yes/no at 2 years). Independent variables were selected according to the alcohol literature: sex,15;25;26 age,15;25;26 race 15;25;26 ethnicity,15;25;26 marital status,15 education,25;26 employment status,25;26 household income,27 BMI,28 SF-36 physical and mental component scores,15 BDI score,29 binge eating,6 treatment for psychiatric or emotional problems,29 ISEL appraisal, belonging, and tangible support scores,30 smoking status,26 regular alcohol consumption (i.e., ≥ 2 times/week), recreational drug use,26 surgical procedure,10,11,13 and percentage weight loss.31 The following model-fitting strategies were adopted. First, preoperative characteristics and surgical procedure were considered, with site as a random effect. Variables that were not significant in the model (i.e., P ≥.05) were removed by using backward elimination. Because backward elimination can lead to biased models and to over-fitting of data, the analysis was confirmed using forward selection. Next, to assess associations between postoperative characteristics (i.e., SF-36 physical and mental component scores, BDI score, binge eating, treatment for psychiatric or emotional problems, ISEL appraisal, belonging, and tangible support scores, smoking status, recreational drug use, and percentage weight loss) and postoperative AUD, a separate GLMM was fit for each postoperative characteristic controlling for its preoperative value (data not shown). Then, postoperative characteristics that were significantly associated with postoperative AUD, and their preoperative values, were added to the best preoperative multivariable GLMM. Again, backward elimination and forward selected were used for model selection; variables that were significantly related to postoperative AUD (P<.05), and their preoperative values were retained. Once independent variables were determined, all potential interactions were evaluated. Sample sizes for GLMM predicting postoperative AUD reflect the exclusion of participants with indeterminate preoperative AUD (n=4), participants who had a reversal of their surgical procedure (n=4) or a new surgical procedure (n=4) before their 1 year assessment, and participants missing covariate data (n=20–97 depending on the model). Participants who had a reversal (n=4) or new procedure (n=5) between their first and second assessment were retained. However, their 2 year data were censored. All tests were two-sided. Adjusted odds ratios [OR] and 95% confidence intervals [CI] are reported. Statistical analyses were performed with SAS software (version 9.2; SAS Institute Inc, Cary, North Carolina).
Of 2458 LABS-2 participants, 2280 completed at least some self-assessment forms at the preoperative visit; 2265 completed the AUDIT. The current analysis is limited to 1945 of these 2265 participants (86%) who completed the AUDIT one (n=1763) or two (n=1582) years postoperatively (see Figure 1). Characteristics of LABS participants included in this analysis are shown in Table 1. Compared to those not included in these analyses, those included were older (median 47 years vs. 42 years; P<.0001), a greater percentage were white (87.0% vs. 82.0%; P<.01), and a smaller percentage were smokers (2.2% vs. 4.1%; P=.02) preoperatively. There were no significant differences between groups with respect to other characteristics.
Table 2 presents select AUDIT item responses and summary measures, as well as alcohol and drug abuse treatment, pre- and 1 and 2 years postoperatively (all AUDIT item responses by time point are shown in eTable 1, available at http://jama.ama-assn.org/). Number of alcoholic drinks consumed on a typical drinking day was significantly higher preoperatively and 2 years postoperatively than in the first postoperative year. ‘Consumption at a hazardous level’ was significantly more common preoperatively than postoperatively. However, there was a significant increase between the first and second postoperative years. Frequency of alcohol consumption and AUD (and all three of its components: dependence symptoms, alcohol-related harm and AUDIT score ≥8,) significantly increased in the second postoperative year compared to preoperatively or the first postoperative year.
More than half (62.3%; CI, 53.0%–71.5%; 66/106) of those reporting AUD at the preoperative assessment continued to have, or had recurrent AUD within the first two postoperative years. In contrast, 7.9% (CI, 6.4%–9.4%; 101/1283) of participants not reporting AUD at the preoperative assessment had postoperative AUD. Nonetheless, more than half (60.5%; CI, 53.1%–67.9%;101/167) of postoperative AUD was reported by those not reporting AUD at the preoperative assessment.
Male sex, younger age, smoking, regular alcohol consumption, AUD, recreational drug use and lower ‘belonging’ interpersonal support preoperatively, and undergoing a Roux-en-Y gastric bypass were independently related to an increased likelihood of AUD postoperatively (Table 3). The adjusted odds of AUD in the second compared to the first postoperative year was 1.6 (CI, 1.3–2.0; P<.0001). There was an interaction between preoperative AUD and age such that the odds ratio for postoperative AUD associated with preoperative AUD increased with age (eFigure 1, available at http://jama.ama-assn.org/). There were no other significant interactions between covariates. Race, ethnicity, marital status, education, employment status, household income, BMI, SF-36 physical and mental component scores, BDI score, binge eating, treatment for psychiatric or emotional problems, and ISEL ‘appraisal’ and ‘tangible’ support scores measured preoperatively were not independently related to postoperative AUD. However, a lower postoperative SF-36 mental component score was independently related to postoperative AUD, as were postoperative smoking, recreational drug use, and treatment for psychiatric or emotional problems (Table 4).
Given the striking relationship between RYGB (in comparison to LAGB) and AUD (Tables 3 and and4),4), we repeated the analysis evaluating alcohol consumption and AUD by time point stratifying by surgical procedure (eTable 2, available at http://jama.ama-assn.org/). Frequency of alcohol consumption significantly increased in the second postoperative year compared to preoperatively or the first postoperative year among participants who underwent RYGB or LAGB. Among participants who underwent RYGB, the number of drinks on a typical drinking day was significantly lower in the first, but not the second postoperative year, compared to preoperatively, and the prevalence of AUD significantly increased in the second postoperative year (i.e., 7.0% (CI, 5.4%–8.6%) preoperatively and 7.9% (CI, 6.3%–9.8%) 1 year postoperatively vs. 10.7% (CI, 8.8%–12.7%) 2 years postoperatively; P<.01), whereas there was no significant difference in number of drinks on a typical drinking day by time among participants who underwent LAGB, nor was there a significant change in prevalence of AUD (i.e., 9.3% (CI, 6.3%–12.3%) preoperatively and 5.6% (CI, 3.2%–8.0%) 1 year postoperatively vs. 7.0% (CI, 4.4%–9.7%) 2 years postoperatively; P=.24).
Despite physician32 and patient6 concerns that bariatric surgery increases risk of AUD, to our knowledge, this is the first prospective investigation of AUD before and after bariatric surgery. This study found a significantly higher prevalence of AUD in the second postoperative year, overall, and specifically post-RYGB, compared to the years immediately before and following surgery. Although the increase in prevalence of AUD from 7.6% preoperatively to 9.6% 2 years postoperatively may seem small, that 2% increase potentially represents more than 2000 additional people with AUD in the U.S. each year33 with accompanying personal, financial, and societal costs.34
Participants were categorized as having AUD if they endorsed at least one symptom of alcohol-related harm or alcohol dependence, likely identifying some participants who would not meet DSM-IV criteria for AUD.22 Thus, comparisons with prevalence rates of AUD should be made with caution. Nonetheless, in contrast to previous studies in which preoperative prevalence of AUD has been very low (0–2%),35–39 whether assessed as part of the surgical evaluation35–37 or independently for research,38;39 the preoperative prevalence of AUD identified in this study was high (7.6%), but similar to the prevalence of AUD in the general U.S. population (8.5%15; 6.4% adjusted to match the sex distribution of bariatric surgery patients5). Given that current AUD is a contraindication for bariatric surgery,40;41 it is unclear whether the prevalence of AUD preoperatively reflects prior removal of those with current AUD from the surgery pool, or under-reporting. There is some evidence that prevalence of lifetime AUD is higher among bariatric surgery candidates (e.g., 31%38) compared to the general U.S. population (30%15; 24% if adjusted to match the sex distribution of bariatric surgery patients5), although results are conflicting.35–39
The significant increase in prevalence of AUD following RYGB, but not LAGB is consistent with observational9 and pharmacokinetic10;11;13 studies. The prevalence of AUD overall, and specifically among participants who underwent RYGB, did not significantly increase until the second postoperative year, when alcohol consumption in general, and consumption at a hazardous level in particular, increased compared to the first postoperative year among participants who underwent RYGB. Therefore it was likely an increase in alcohol sensitivity following RYGB combined with resumption of higher levels of alcohol consumption in the second postoperative year, which led to the increase in AUD.
AUD in the year prior to surgery substantially increased the odds of AUD in the first two postoperative years, consistent with the chronic and recurrent nature of AUD.15 Regular alcohol consumption preoperatively also independently increased the likelihood of postoperative AUD. A lower sense of ‘belonging’ (i.e., availability of people to do things with) preoperatively was predictive of postoperative AUD, suggesting that interpersonal social support may protect against AUD. Most other predictors of AUD identified in this study (i.e., male sex, younger age, smoking, recreational drug use) have been associated with AUD in the general U.S. population.15;26
We did not find a significant association between preoperative depressive symptoms, binge eating, mental health, or past-year treatment for psychiatric or emotional problems, and postoperative AUD. However, worse mental health and treatment for psychiatric or emotional problems postoperatively were significantly associated with AUD, consistent with cross-sectional studies reporting associations between psychiatric disorders and AUD.10;12;13;15 The direction of these relationships is unclear.
Percentage weight loss was not independently related to postoperative AUD. However, results require replication. Only a few studies, to our knowledge, have investigated the inverse, with various measures of alcohol consumption, alcohol-related problems, and treatment. In a sample of 440 LAGB patients there was a significant positive relationship between preoperative alcohol consumption and percentage excess weight loss at 1 year (R=0.23; P<.005).42 In addition, two studies (n=80 RYGB43 and n=413 RYGB, LAGB and sleeve gastrectomy44) found that a history of preoperative substance use disorder (including AUD and drug abuse/dependence), in remission at time of surgery, predicted better postoperative weight loss, while one study (n=207 RYGB45) reported no significant association. Finally, among 160 patients who reported some weight regain post-RYGB, those who reported someone had expressed concern about their postoperative alcohol or drug use (<10% of participants) had an increased odds (OR=12.7, CI, 1.7–93.8; P=.01) of significant regain.31 Future studies are needed to clarify if and how postoperative weight loss is related to alcohol use and AUD and vice versa.
Although safe levels of alcohol consumption have yet to be established for the postoperative patient, it is concerning that one in eight participants reported consuming at least 3 drinks per typical drinking day and one in six participants reported ‘consumption at a hazardous level’ in the second postoperative year, given the negative impact heavy drinking may have on vitamin and mineral status,46 hepatic function,46 and weight loss.31 It is also noteworthy that alcohol or drug abuse treatment was uncommon both pre- and postoperatively (e.g., 0.5%–0.8%; 7–10% of those with AUD).
Major strengths of this study include the prospective design, large sample from 10 hospitals throughout the U.S., and use of a validated and reliable alcohol screening tool. Some study limitations with respect to interpretation of results should be noted. First, lifetime history of AUD was not assessed. Thus, we were unable to determine whether postoperative AUD was ‘new-onset’ vs. ‘recurrent.’ Second, while research data was collected independently of the surgery approval process and clinical care, some participants may have under-reported their alcohol use due to concerns that their responses would affect their surgery eligibility or social desirability. However, the fact that 7.8% of study participants reported symptoms of AUD at their preoperative research visit indicates that participants who may have withheld symptoms of AUD during their clinical assessment in order to qualify for surgery did not withhold this same information when completing the AUDIT for research purposes. Third, when participants reported having at least 5 drinks on a typical drinking day (per AUDIT item 2) a safety protocol was triggered to assess the need for referral, which may have led to under-reporting of alcohol consumption or problems at future (i.e., postoperative) assessments. Fourth, because this study does not have a control group, we cannot rule out the possibility that reporting of AUD would increase independent of surgery 2 years after baseline assessment. However, the relationship between surgical procedure and postoperative AUD (i.e., higher likelihood with RYGB compared to LAGB) provides evidence that the most common surgical procedure was likely at least partially responsible for the increase in postoperative AUD at 2 years. Finally, we were underpowered to determine if risk of postoperative AUD was related to race or ethnicity, or lower incidence surgical procedures.
To summarize, the prevalence of AUD was greater in the second postoperative year than preoperatively or in the first postoperative year; this finding appears to be driven by RYGB, which accounted for 70% of surgeries and doubled the likelihood of postoperative AUD compared to LAGB. Several factors associated with AUD in the general population, including history of AUD and regular alcohol consumption, also increased the likelihood of postoperative AUD in this bariatric surgery sample. Although preoperative AUD greatly increased the likelihood of postoperative AUD, over half of participants with postoperative AUD did not report AUD in the year prior to surgery.
This study has important implications for the care of pre- and postoperative bariatric surgery patients. Regardless of alcohol history, patients should be educated on the potential effects of bariatric surgery, in particular RYGB, to increase the risk of postoperative AUD. In addition, alcohol screening and, if indicated, referral should be offered as part of routine pre- and postoperative clinical care. Further research should examine the long-term impact of bariatric surgery on AUD, and the relationship of AUD to postoperative weight control.
Funding/Support: LABS-2 was funded by a cooperative agreement by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK). Grant numbers: Data Coordinating Center -U01 DK066557; Columbia-Presbyterian - U01-DK66667 (in collaboration with Cornell University Medical Center CTSC, Grant UL1-RR024996); University of Washington - U01-DK66568 (in collaboration with CTRC, Grant M01RR-00037); Neuropsychiatric Research Institute - U01-DK66471; East Carolina University – U01-DK66526; University of Pittsburgh Medical Center – U01-DK66585 (in collaboration with CTRC, Grant UL1-RR024153); Oregon Health & Science University – U01-DK66555.
Role of the Sponsor: NIDDK scientists contributed to the design and conduct of the study, which included collection, and management of data. The Project Scientist from NIDDK served as a member of the Steering Committee, along with the Principal Investigator from each clinical site and the Data Coordinating Center (DCC). The DCC housed all data during the study and performed data analyses according to a pre-specified plan developed by the DCC biostatistician and approved by the steering committee and independent Data Saftey Monitoring Board. The decision to publish was made by the LABS steering committee, with no restrictions imposed by the sponsor. As a co-author, an NIDDK scientist contributed to the interpretation of the data and preparation, review, or approval of the manuscript.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr. Courcoulas has received research grants from Allergan Pfizer, Covidien, and EndoGastric Solutions, and is on the Scientific Advisory Board of Ethicon J & J Healthcare System. Dr. Pories has received research grants from J & J Ethicon Endosurgery and GlaxoSmithKline.
Author Contributions:Dr. King and Ms. Chen had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: King, Mitchell
Analysis and interpretation of data: Chen, Courcoulas, Engel, Kalarchian, King, Mitchell, Pories, Steffen, Yanovski
Drafting of the manuscript: King
Critical revision of the manuscript for important intellectual content: Chen, Courcoulas, Engel, Kalarchian, King, Mitchell, Pories, Steffen, Yanovski
Statistical analysis: Chen, King
Obtained funding: Courcoulas, Mitchell, Pories
Study supervision: Courcoulas, Mitchell, Pories, Yanovski
Wendy C. King, Epidemiology, GSPH, University of Pittsburgh.
Jia-Yuh Chen, Biostatistics, GSPH, University of Pittsburgh.
James E. Mitchell, Neuropsychiatric Research Institute.
Melissa A. Kalarchian, Psychiatry, University of Pittsburgh School of Medicine.
Kristine J. Steffen, Neuropsychiatric Research Institute.
Scott G. Engel, Neuropsychiatric Research Institute.
Anita P. Courcoulas, Surgery, University of Pittsburgh Medical Center.
Walter J. Pories, Surgery, Brody School of Medicine, East Carolina University.
Susan Z. Yanovski, Division of Digestive Diseases and Nutrition, National Institute of Diabetes and Digestive and Kidney Diseases.