|Home | About | Journals | Submit | Contact Us | Français|
To document changes in Axis I psychiatric disorders following bariatric surgery and examine their relationship with post-surgery weight loss.
As part of a 3-site sub-study of the Longitudinal Assessment of Bariatric Surgery (LABS) Research Consortium, 199 patients completed the Structured Clinical Interview for DSM-IV prior to Roux-en-Y Gastric Bypass (RYGB) or Laparoscopic Adjustable Gastric Band (LAGB). At 2 or 3 years post-surgery, 165 (83%) patients completed a follow-up assessment [pre-surgery median body mass index (BMI) 44.8 kg/m2, median age 46 years, 92.7% white, 81.1% female]. Linear mixed modeling was used to test change in prevalence of psychiatric disorders over time, report remission and incidence, and examine associations between psychiatric disorders and weight loss.
Compared with status pre-surgery, the prevalence of any Axis I psychiatric disorder was significantly lower at 2 and 3 years post-surgery [30.2% vs. 16.8% (P=.003) and 18.4% (P=.012), respectively]. Adjusting for site, age, sex, race, pre-surgery BMI, and surgical procedure, pre-surgery mood, anxiety, eating or substance use disorders (lifetime or current) were not related to weight change, nor were post-surgery mood or anxiety disorders (P for all >.05). However, having a post-surgery eating disorder was independently associated with less weight loss at 2 or 3 years (β=6.7%; P=.035).
Bariatric surgery was associated with decreases in psychiatric disorders through 3 years after surgery. Post-surgical eating disorders were associated with less weight loss after surgery, adding to the literature suggesting that disordered eating after surgery is related to suboptimal weight loss.
Several studies have found that rates of lifetime and current psychiatric disorders are substantial among candidates for bariatric surgery (1, 2). Studies employing structured clinical interviews have revealed high lifetime rates of mood disorders (range: 22.0 – 54.8%) and anxiety disorders (range: 15.5 – 37.5%) prior to surgery. Most also suggest high lifetime rates for alcohol abuse and dependence, but documented rates vary widely (range: 0.7 – 33.2%) (3–7). Although bariatric surgery is the most effective treatment for severe obesity (8, 9), post-surgery weight loss is highly variable, even among patients undergoing the same procedure. For example, a recent report from the Longitudinal Assessment of Bariatric Surgery-2 (LABS-2) study found that among participants who underwent Roux-en-Y gastric bypass (RYGB), one of the most common procedures, three year weight loss was more than 32% of initial weight for the top quartile of participants and less than 16% for the bottom quartile of participants (10). Psychopathology may account for some of the variability in post-surgery weight loss (11–15), but there are few methodologically rigorous studies.
Only two studies have used structured diagnostic instruments, considered the gold standard for the assessment of psychiatric disorders, to examine the relationship of pre-surgery Axis I psychiatric disorders with post-surgery outcomes. Kalarchian et al. found that lifetime psychiatric disorder at the time of surgery, particularly mood or anxiety, predicted modestly poorer weight loss at 6 months after surgery after adjusting for the initial body mass index (BMI), sex, race, and age (13). A separate investigation by de Zwaan et al. documented that pre-surgery lifetime history of anxiety disorders, or the presence of both pre-surgery anxiety and depressive disorders (lifetime or current), but not depressive disorders alone, were negatively associated with post-surgery weight loss (11). This study also investigated associations between post-surgery psychiatric disorders and weight change, revealing that post-surgery mood disorder, but not anxiety disorder, was associated with significantly less weight loss at 24–36 months after surgery.
The current study adds to the literature by reporting on the largest well-characterized sample to date. We report on psychiatric disorders assessed prior to bariatric surgery, and again at 2 or 3 years following surgery using the Structured Clinical Interview for DSM-IV (16), administered independently of routine clinical care. The primary aims were to document changes in psychiatric disorders at 2 and 3 years post-surgery and examine the relationship between psychiatric disorders and post-surgery weight loss.
The current study was part of the LABS Consortium of studies, which have been described in detail (17). The initial LABS study, (LABS-1) involved 5,050 patients and examined the short-term safety of bariatric surgery (18). LABS-2 examined long-term efficacy of bariatric surgery in 2,458 patients(19). The current study, (LABS-3 Psychosocial), was conducted in a smaller set of participants (n=202), many of whom also participated in LABS-2, at three of the six LABS-2 clinical centers: The Neuropsychiatric Research Institute, Fargo, ND; Columbia-Weill Cornell Medical Centers in New York, NY; and the University of Pittsburgh in PA. The study was approved by the University of North Dakota Institutional Review Board, the Columbia-Weill Cornell Medical Center Institutional Review Board, the University of Pittsburgh Institutional Review Board, and the Duquesne University Institutional Review Board. All participants provided informed consent prior to data collection, which occurred from January 2007 to September 2014.
The 199 LABS-3 patients who underwent RYGB or Laparoscopic Adjustable Gastric Band (LAGB) were considered for the current analysis. Of those, 165 had pre-surgery data plus at least 1 of 2 follow-up assessments at years 2 or 3 (154 at year 2 and 129 at year 3) and were included in the analysis sample. Of the 34 participants excluded, 27 refused or were unable to schedule an assessment, 4 could not be reached, and 3 had died.
LABS-3 participants were interviewed using the Structured Clinical Interview for DSM-IV Patient Version to assess Axis I psychiatric disorders (16). Interviews were conducted independently of the routine pre-surgery screening process. Methods and baseline data on psychiatric disorders have been reported previously (5). Of note, the SCID screener for alcohol use disorder required reporting having drunk five or more drinks on one occasion, a threshold that might be too high after bariatric surgery given evidence that alcohol pharmacokinetics may be altered by some procedures, thereby decreasing tolerance (20–23). Lifetime history was assessed at the pre-surgery assessment. Current status, based on symptoms within the past month, was rated at all assessments. Pre-surgery psychiatric status was categorized as having the disorder at time of surgery, having a history of the disorder but not at time of surgery, and no history of the disorder. Participants with a current psychiatric disorder pre-surgery were considered to be in remission if they did not currently have the disorder post-surgery. Incident psychiatric disorder was defined as not having the disorder currently pre-surgery, but meeting diagnostic criteria at a post-surgery assessment.
Participants recorded the names and frequency of use of all prescribed medications taken currently using a study-specific medication form (17). Antidepressant medication use was defined as taking a medication classified as an antidepressant at least daily; antianxiety medication use was defined as taking a medication classified as an antianxiety or benzodiazepine at least daily; and psychiatric medication use was defined as use of antidepressant or antianxiety medication or a medication classified as norepinephrine reuptake inhibitor, antimanic, CNS stimulant, dibenzodiazepine, dibenzothiazepine, or phenothiazine at least daily.
Participants’ weight was measured to the nearest pound at the baseline and annual assessments with bare feet and in light-weight clothing. Self-reported weight, which has been validated in this population (24), was utilized when an in-person follow-up assessment could not be performed. Including all baseline and follow-up assessments, 8.4% (41/487) were self-reported. The weights of women in their 2nd or 3rd trimester of pregnancy or within 6 months post-partum were censored. Weight change was calculated as (follow-up weight minus baseline weight) divided by baseline weight.
Analyses were conducted using SAS version 9.3 (SAS Institute, Cary, NC, USA). Statistical significance was set at p<.05; tests were two-sided. Descriptive statistics (frequencies and percentages for categorical variables; medians, 25th-75th percentiles, and ranges for continuous variables) were used to summarize pre-surgery characteristics. Those excluded from the analysis sample (n=34) were compared to those in the analysis sample (n=165) using the chi square test for categorical variables and the Wilcoxon rank sum test for continuous variables.
Data were assumed to be missing at random. Generalized linear mixed models were used to identify pre-surgery characteristics related to missing follow-up assessments among the analysis sample. Age was the only significant factor, with younger adults missing more assessments (data not shown). Thus, all longitudinal analyses controlled for age.
Poisson mixed models (PMM) with robust error variance were used to compare rates of psychiatric disorders and related medications between assessments (pre-surgery and 2 and 3 years post-surgery). Simulation was used to adjust for multiple comparisons (25). For rare outcomes modeling was not possible; only observed data are presented. PMM also were used to estimate the remission and incidence rates of the more common psychiatric disorders at 2 and 3 years post-surgery and to determine factors related to having any mood disorder and any anxiety disorder at 2 and 3 years post-surgery; other disorders were too rare post-surgery to model. Based on prior research (15, 18, 26, 27), pre-surgery psychiatric status, age, sex, race, pre-surgery BMI, surgical procedure, and percent weight change were included as independent variables, with control for study site to adjust for correlations among patients from the same site. Interactions between main effects were tested and included if they reached statistical significance.
Linear mixed models were used to test and estimate associations between any mood disorder, any anxiety disorder, any substance use disorder and any eating disorder, and post-surgery weight change, controlling for site and factors previously related to surgical weight loss (10, 28–30), including sex, age, race, pre-surgery BMI and surgical procedure. In the first model pre-surgery psychiatric status was examined. In the second model post-surgery psychiatric status (current), controlling for pre-surgery psychiatric status, was examined. Interactions between main effects were tested and included if they reached statistical significance.
To determine whether it was appropriate to report results separately for men and women, a sensitivity analysis was performed in which we investigated whether change in psychiatric disorders, predictors of post-surgery psychiatric disorders, and predictors of weight change differed by sex.
Characteristics of the analysis sample are shown by sex in Table 1. Subsequent analysis is reported among the entire sample because sensitivity analysis revealed that change in psychiatric disorders over time, predictors of post-surgery psychiatric disorders, and predictors of weight change did not differ by sex (data not shown).
Those included (n=165) vs. excluded from analysis due to missing follow-up (n=34) did not significantly differ with respect to pre-surgery characteristics shown in Table 1 or pre-surgery psychiatric disorder status (see Table S1, Supplemental Digital Content 1). Approximately one fifth of the sample (18.2%; n=30) was male. The majority (59.4%; n=98) underwent RYGB, and 40.6% (n=67) underwent LAGB. Median (25th, 75th percentiles) percentage weight change was −36.8% (−41.6, −29.3) at 2 years and −33.7% (−38.4, −25.2) at 3 years for RYGB, and −13.2% (−19.0, −7.0) at 2 years and −12.2% (−19.2, −3.2) at 3 years for LAGB.
Observed rates of psychiatric disorders and medication use pre-surgery and 2 and 3 years post-surgery are shown in Table 2. Modeled data for the more common disorders and medications are shown in Table 3. Given the sample size and low prevalence of some disorders and infrequent use of some types of medications, we had limited statistical power to detect differences between time points. However, prevalence of any psychiatric disorder was significantly lower at both 2 and 3 years post-surgery vs. pre-surgery. Additionally, the rate of any anxiety disorder was significantly lower at year 3 vs. pre-surgery (8.2% vs. 17.1%; p=.037), and the rate of any eating disorder was lower at year 2 vs. pre-surgery (1.3% vs. 7.3%; p=.021). Prevalence of use of any psychiatric medication, and specifically any anti-depressant medication or any anti-anxiety medication, was stable over time.
Table 4 provides observed and modeled remission and incidence of psychiatric disorders at 2 and 3 years post-surgery based on pre-surgery status. For all classes of disorder, rates of remission far exceeded incidence after surgery. Only one patient experienced the post-surgery incidence of an eating disorder.
Risk of mood and anxiety disorders at 2 and 3 years post-surgery are provided in Table S2, in Supplemental Digital Content 1. Adjusting for site, pre-surgery BMI, sex, age, race, surgical procedure, and percent weight loss, having a mood disorder or an anxiety disorder at time of surgery greatly increased the risk of having the same disorder post-surgery (P for all <.01). Having a lifetime history of, but not current, mood disorder at time of surgery also increased the risk of a post-surgery mood disorder. However, having a lifetime history of, but not current, anxiety disorder pre-surgery was not related to risk of a post-surgery anxiety disorder. There were significant interactions between age and pre-surgery mood disorder (P=.015) and anxiety disorder status (P=.007), such that the risk associated with pre-surgery status was greater among younger versus older individuals. Additionally, older age was a risk factor for post-surgery mood disorder only among those with no history of mood disorder, while younger age was a risk factor for post-surgery anxiety only among those with current anxiety pre-surgery (Table S2, Supplemental Digital Content 1). White race was significantly associated with lower risk of a post-surgery mood disorder [RR=0.40 (95% CI: 0.17–0.95), P=.037]. No other factors (e.g., pre-surgery BMI, sex, surgical procedure or percent weight loss) were significantly related to having a mood disorder or an anxiety disorder post-surgery (Table S2, Supplemental Digital Content 1).
In multivariable analysis controlling for age, sex, race, pre-surgery BMI, surgical procedure and site, neither pre-surgery current or lifetime mood disorder (P=.71), any anxiety disorder (P=.19), nor any eating disorder (P=.99), were significantly related to percentage weight change at 2 or 3 years (Table S3, in Supplemental Digital Content 1). Pre-surgery lifetime history of substance use disorder, with a prevalence of 34.8% (n=57), also was not related to weight change (P=.16); current substance use disorder pre-surgery was too rare (i.e., 1.2%: n=2) to evaluate.
Having a current mood disorder or anxiety disorder was not related to weight change at 2 or 3 years post-surgery (Table 5; P=.91 and P=.12, respectively). However, having a current eating disorder was associated with less weight loss at 2 and 3 years post-surgery [6.7% (95% CI: 0.5–13.0; P=.035].
This study documented that the prevalence of psychiatric disorders were reduced through 3 years after bariatric surgery among the largest, carefully-assessed sample to date. Pre-surgery, 30% of the patients in our study met diagnostic criteria for an Axis I disorder, dropping to 17% and 18% at 2 and 3 years after surgery, respectively. Having a current or lifetime mood or anxiety disorder at the time of surgery, or at post-surgery follow-up, was unrelated to post-surgery weight loss. Having a lifetime substance use disorder, or a current or lifetime eating disorder at the time of surgery also was not related to post-surgery weight loss. However, the very small subset of patients who had eating disorders at post-surgery follow-up (1.3% and 2.7% at 2 and 3 years, respectively) lost, on average, 6.7% less weight.
In the present study, anxiety disorders were significantly lower at year 3 vs. pre-surgery (8.2% vs. 17.1%, respectively), but mood disorders were unchanged (12.1% vs. 12.2%, respectively). These findings may be compared to the report by de Zwaan and colleagues (11) which documented a significant decrease in mood disorders at 24–36 months relative to pre-surgery (14.3% vs. 32.7%, respectively), but no change in anxiety disorders (14.3% vs. 16.8%, respectively). Nonetheless, both studies suggested that mood and anxiety disorders are stable or reduced after bariatric surgery and are predicted by having the same disorder prior to surgery. Further research is required to identify the causal pathways for pre- to post-surgery change in mood and anxiety disorders, including investigation of a wider range of factors such as treatment expectations. Moreover, the two studies differed in results on how mood and anxiety disorders affect weight loss. Although the present study documented that mood and anxiety disorders did not have prognostic significance, de Zwaan and colleagues found that a past history of anxiety disorders pre-surgery, as well as mood disorders post-surgery, were both associated with less weight loss. Therefore, further research is needed to clarify the impact of mood and anxiety disorders on weight loss.
Our finding that post-surgery eating disorder was related to less weight loss at 2 and 3 years after surgery is consistent with a growing body of literature (31). However, it should be noted that only 2 participants at year 2 and 4 participants at year 3 reported eating disorders (all binge eating disorder). Because binge eating may not be possible due to the size of the pouch after bariatric surgery, many reports have examined loss of control (LOC) eating, without the requirement of consumption of an objectively large amount of food. In the present study, LOC eating or other disordered eating that did not meet DSM-IV diagnostic criteria were not assessed with the SCID. More prospective research is needed on the relationship between disordered eating and weight loss after surgery, including assessments of LOC eating.
Lifetime history of substance use disorders pre-surgery, reported by over one third of participants, was not predictive of post-surgery weight loss. Very few participants met diagnostic criteria for a current substance use disorder pre- or post-surgery, providing insufficient power to test change over time or association with weight loss. However, our previous report from the LABS-2 study showed that alcohol use disorders ascertained by the Alcohol Use Disorders Identification Test (AUDIT) increased significantly in the second year after RYGB and were associated with pre-surgery consumption and alcohol use disorders (32), suggesting that a more fine grained analysis of outcomes among patients with a history of or risk factors for alcohol use problems is warranted.
The present study also examined changes in use of psychotropic medications, although it is not known if these medications were prescribed specifically for psychiatric disorders. Interestingly, rates of any psychiatric medication usage, as well as any antidepressant medication or any antianxiety medication, remained relatively stable during the period of investigation, despite the reduced number of individuals meeting diagnostic criteria for any psychiatric disorder. Relatively little is known about clinical pharmacology after bariatric surgery, but two studies show evidence of reduced bioavailability of antidepressants after RYGB (33, 34). Developing a better understanding of pharmacokinetics of psychotropic medications after surgery will inform clinical management of this patient population.
Strengths of the present investigation include use of standardized assessment of psychiatric disorders administered separately from clinical approval for surgery among a large cohort. Although the analysis sample was predominantly white and female, this is typical of the bariatric surgery patient population. Limitations include potential bias associated with self-selection into the study or dropout from the study, as well as limited power for some analyses. Also, disordered eating may be underestimated because the application of DSM-IV diagnostic criteria for eating disorders (anorexia nervosa, bulimia nervosa and binge eating disorder) does not include subclinical disorders such as LOC eating. Additionally, all postoperative assessments of psychiatric disorders were limited to current disorder, and it is possible that patients could have met criteria for a disorder after operation, but not at the 2 or 3 year assessment time point. Finally, the sample is restricted to patients who underwent RYGB and LAGB, and does not include gastric sleeve, which is increasing in popularity and acceptance (35).
In summary, prevalence of having any Axis I psychiatric disorders decreased following surgery, and pre-surgery history of the most common disorders (i.e., mood, anxiety, substance use, and eating disorders) was unrelated to weight loss through 3 years following bariatric surgery. Although very uncommon, binge eating disorder after surgery was associated with modestly less weight loss. Future hypothesis-driven studies should include measures of LOC eating as well as diverse samples of patients undergoing different types of procedures.
Source of Funding: This clinical study was a cooperative agreement funded by the National Institute of Diabetes and Digestive and Kidney diseases (grant DSS-U01DK066557; Columbia, U01-DK66667 [in collaboration with Cornell University Medical Center CTSC, grant UL1-RR024996]; Neuropsychiatric Research Institute, grant U01-DK66471; University of Pittsburgh Medical Center, grant U01-DK66585 [in collaboration with CTRC, grant UL1-RR0241531]). Additionally funded by the National Institute of Diabetes and Digestive and Kidney diseases grant R01 DK084979.
Dr. Kalarchian reports receiving funding for research in bariatric surgery from The Obesity Society (TOS)/Nutrisystem and the American Society for Metabolic and Bariatric Surgery (ASMBS). Dr. Devlin reports royalties from Wolters Kluver Health. Dr. Marcus reports receiving honoraria from the Scientific Advisory Board and Weight Watchers International. Dr. Mitchell reports royalties for books from American Psychiatric Press Inc., Guilford, and Routledge.
Conflict of Interest: Drs. King, Garcia, Chen and Yanovski report no conflicts of interest.
Melissa A. Kalarchian, Duquesne University School of Nursing.
Wendy C. King, Department of Epidemiology, University of Pittsburgh, Graduate School of Public Health.
Michael J. Devlin, Columbia University College of Physicians and Surgeons / New York State Psychiatric Institute.
Marsha D. Marcus, Department of Psychiatry, University of Pittsburgh School of Medicine.
Luis Garcia, Sanford Health System, and the University of North Dakota School of Medicine and Health Sciences.
Jia-Yuh Chen, University of Pittsburgh, Graduate School of Public Health.
Susan Z. Yanovski, Division of Digestive Diseases and Nutrition, National Institute of Diabetes and Digestive and Kidney Diseases.
James E. Mitchell, Neuropsychiatric Research Institute, and the University of North Dakota School of Medicine and Health Sciences.