Alcohol and substance abuse, depression and suicide attempts after Roux-en-Y gastric bypass surgery

Original article Alcohol and substance abuse, depression and suicide attempts after Roux-en-Y gastric bypass surgery O. Backman 1,3,D.Stockeld 1, F. Rasmussen 2,E.Näslund 1 and R. Marsk 1 Departments of 1 Clinical Sciences, Danderyd Hospital, and 2 Public Health, Karolinska Institute, Stockholm, and 3 Department of Surgical and Perioperative Science (Hand and Plastic Surgery), Umeå University, Umeå, Sweden Correspondence to: Dr O. Backman, Department of Clinical Sciences, Danderyd Hospital, Karolinska Institute, SE-182 88 Stockholm, Sweden (e-mail: Olof.Backman@ki.se) Background: Small studies suggest that subjects who have undergone bariatric surgery are at increased risk of suicide, alcohol and substance use disorders. This population-based cohort study aimed to assess the incidence of treatment for alcohol and substance use disorders, depression and attempted suicide after primary Roux-en-Y gastric bypass (RYGB). Methods: All patients who underwent primary RYGB in Sweden between 2001 and 2010 were included. Incidence of hospital admission for alcohol and substance use disorders, depression and suicide attempt was measured, along with the number of drugs prescribed. This cohort was compared with a large age-matched, non-obese reference cohort based on the Swedish population. Inpatient care and prescribed drugs registers were used. Results: Before RYGB surgery, women, but not men, were at higher risk of being diagnosed with alcohol and substance use disorder compared with the reference cohort. After surgery, this was the case for both sexes. The risk of being diagnosed and treated for depression remained raised after surgery. Suicide attempts were significantly increased after RYGB. The adjusted hazard ratio for attempted suicide in the RYGB cohort after surgery compared with the general non-obese population was 2 85 (95 per cent c.i. 2 40 to 3 39). Conclusion: Patients who have undergone RYGB are at an increased risk of being diagnosed with alcohol and substance use, with an increased rate of attempted suicide compared with a non-obese general population cohort. Paper accepted 27 May 2016 Published online 28 July 2016 in Wiley Online Library (www.bjs.co.uk). DOI: 10.1002/bjs.10258 Introduction Bariatric surgery has been shown to reduce weight, decrease the risk of early death, and improve cardiovascular and metabolic health 1,2. Despite these improvements, several recent reports have suggested that there is an increased risk of alcohol and substance use disorders after bariatric surgery 3,4. In the USA, patients offered bariatric surgery have a greater lifetime history of substance abuse disorder, but lower ongoing substance and alcohol abuse compared with the general population 5. Another study 6 from the USA showed that 16 per cent of patients seeking bariatric surgery were problem drinkers. A recent large multicentre study 7 from the USA also reported a significantly higher prevalence of hazardous alcohol use, and symptoms of alcohol abuse/dependence, along with a higher risk of fulfilling criteria for alcohol abuse/dependence at 2-year follow-up after Roux-en-Y gastric bypass (RYGB) surgery, compared with the first postoperative year and before surgery. Patients undergoing RYGB surgery seem to be more prone to develop alcohol use disorders than those undergoing restrictive procedures 3,7, perhaps related to the altered uptake and metabolism of alcohol seen after RYGB 8,9. This altered metabolism of alcohol is also seen after sleeve gastrectomy 10, but not after gastric banding 11, although in a rodent model alcohol reward after RYGB was independent of alcohol absorption from the gut 12. Patients who undergo bariatric surgery have a high rate of major depression before surgery. One study 13 reported a lifetime risk before surgery of 38 7per cent. The 10-year follow-up report 14 from the Swedish Obese Subjects (SOS) study found no significant difference between surgical and control groups with regard to depression. At baseline, obese subjects who underwent surgery were more likely to 2016 BJS Society Ltd BJS 2016; 103: 1336 1342

Alcohol and substance abuse after Roux-en-Y gastric bypass surgery 1337 be depressed, but significant improvement occurred after surgery. In a recent publication 15 of 7-year follow-up data of patients undergoing bariatric surgery compared with an obese control group, there was a transient improvement in symptoms of depression. RYGB has also been associated with an increased risk of suicide and accidents compared with that in obese control subjects 16,17. Over a 10-year period, a suicide rate of 13 7 per10000formenand5 2 per 10 000 for women was reported from the USA after bariatric surgery, with most suicides occurring within the first 3 years after surgery. The age- and sex-matched suicide rates from the general population presented in the same study were significantly lower at 2 4 and 0 7 per 10 000 respectively. It seems that suicide attempts increase with BMI, but there is an inverse relationship between suicide mortality and BMI in men 18. Most previous studies that have examined these issues are small and come from specialized surgical centres. The present study examined the prevalence of diagnosis and treatment for alcohol and substance use disorders, depression and suicide attempts before and after RYGB in a nationwide cohort study in Sweden, compared with that in a large unselected population cohort (who had not undergone bariatric surgery) using several national databases. Methods This was a register-based nationwide and population-based cohort study of men and women, aged 18 years or more, who underwent a primary RYGB between 1 January 2001 and 31 December 2010, using the national personal identity number for each subject. This allowed identification and individual linkage between different registers. Data sources Data were used from the National Patient Register (NPR), Total Population Register, emigration records, Causes of Death Register, Population and Housing Census, Longitudinal Integration Database for Health Insurance and Labour Market Studies (LISA), and the Prescribed Drug Register (PDR). Inpatient hospital care data were retrieved from the NPR, a nationwide register to which all hospitals are obliged to report. The NPR attained national coverage in 1987 and covers nearly 100 per cent of all hospital admissions. The Total Population Register provides data on dates of death of each individual in Sweden. This register also detects emigration from Sweden; it is complete and continually updated by Statistics Sweden. The Causes of Death Register contains causes of death for 99 5 per cent of people who have died in Sweden since 1961. The quality of the register is considered good, but the register has about a 2-year delay before all data are entered. The Swedish nationwide PDR was established in 2005 and includes all dispensed prescription drugs classified according to the WHO Anatomical Therapeutic Chemical (ATC) classification system. The Population and Housing Census includes complete nationwide data on educational level from 1970 and 1990, and occupational socioeconomic status for every 5 years between 1970 and 1990. In addition, educational level was available from the LISA. Cohorts By using the Swedish version of the Classification of Surgical Procedures (NOMESKO) codes relating to gastric bypass (JDF10, JDF11) with a confirmatory diagnosis of obesity from ICD-9 and ICD-10 (E66 0, E66 1, E66 8 or E66 9), all patients who underwent a primary gastric bypass procedure in the nationwide NPR were identified. Each individual who had surgery was assigned ten ageand sex-matched controls, selected randomly from the general population through linkage with the Swedish Multigeneration Register. Each member of the reference group was assigned a pseudosurgery date corresponding to that of the case participant, giving the same before and after observation periods for the matched individuals. Outcomes Inpatient data, according to ICD-8, ICD-9 and ICD-10, for the following diagnoses were obtained from the NPR: alcohol use disorders (ICD-8 and -9, 303; ICD-10, F10), substance use disorders (excluding alcohol use disorders) (ICD-8 and -9, 304; ICD-10, F11 F16), depression (ICD-8 and -9, 300, 790; ICD-10, F32 F39) and attempted suicide (ICD-8 and -9, E95, E98; ICD-10, X60 X84 and Y10 Y34). In a subgroup of patients, prescribed drug dispensation data were collected from the PDR for the following ATC group of drugs: alcohol dependence (N07BB), benzodiazepines (N05BA), hypnotics and sedatives (N05C), and antidepressants (N06A). This subgroup included all patients from the main cohort who had surgery or a pseudosurgery date between July 2006 and March 2008. The regional ethics committee of Stockholm, Sweden, approved the study.

1338 O. Backman, D. Stockeld, F. Rasmussen, E. Näslund and R. Marsk Statistical analysis Preoperative and postoperative incidence rate ratios (IRRs) were calculated and presented with 95 per cent confidence intervals (c.i.). The preoperative person-time at risk was calculated from their 18-year birthday to the date of surgery/pseudosurgery or until the person had an event for the first time. The person was considered to have an event when the diagnosis in question was detected in the NPR. After surgery, all study subjects were followed up regarding inpatient care diagnoses in the NPR. The follow-up time was from date of surgery/pseudosurgery until date of death, emigration or the end of the study, whichever came first. In the analyses of postoperative morbidity, the patient was considered to have an event the first time the diagnosis was detected in the NPR; person-time at risk was calculated accordingly. Postsurgery hazard ratios (HRs) were calculated with the Cox regression model, adjusting for age, sex, preoperative diagnosis (alcohol use disorder, substance use disorder, depression, suicide attempt), socioeconomic status and educational level. HRs are presented with 95 per cent c.i. Socioeconomic status was classified as: unskilled worker, skilled worker and non-manual worker. Educational level was categorized as: less than 10 years, 10 12 years, or 13 years or more. Incidence rate (IR) and IRR for the studied outcomes for subjects who underwent RYGB and control subjects, before and after surgery or pseudosurgery dates, were also assessed. For the analysis of drug prescriptions, for each drug the percentage of persons who had filled per year was calculated. Owing to the limited statistical power in this subset, results were adjusted by sex and are not presented as stratified. A Poisson regression with generalized estimating equations to account for within-individual correlation due to repeated measurements to estimate relative risks (proportions), confidence intervals and P values was performed. Wald χ 2 tests were used for analysis. STATA software (StataCorp, College Station, Texas, USA) was used for all statistical analyses. Results Two cohorts were obtained. The first cohort consisted of 16 755 patients who had undergone RYGB and 167 550 reference subjects from the general population with data from the NPR; the median follow-up was 1 9 (i.q.r. 0 83 3 76) years. The second cohort comprised 3139 patients who had undergone RYGB and who had 1-year preoperative data and 4-year postoperative data from the PDR, and 31 390 reference subjects from the general population (Table 1). Women in the RYGB cohort had a significantly increased risk of inpatient care for alcohol and other substance use disorders, depression and attempted suicide before surgery compared with those in the reference cohort using the pseudosurgery dates. Men in the RYGB cohort had an increased risk of inpatient care for depression and suicide attempt before surgery, but not of inpatient care for substance use disorders. Men in the RYGB cohort actually had a significantly lower risk of inpatient care for alcohol Table 1 General characteristics of the main gastric bypass cohort, the prescription analysis cohort and their respective reference cohorts (general population, matched for age and sex) Main cohort Prescription cohort RYGB cohort (n = 16 755) Reference cohort (n = 167 550) RYGB cohort (n = 3139) Reference cohort (n = 31 390) Age (years) 18 39 7521 (44 9) 75 210 (44 9) 1499 (47 8) 14 990 (47 8) 40 49 5295 (31 6) 52 950 (31 6) 1004 (32 0) 10 040 (32 0) 50 3939 (23 5) 39 390 (23 5) 636 (20 3) 6360 (20 3) Sex ratio (F : M) 12 723 : 4032 127 230 : 40 320 2383 : 756 23 830 : 7560 Socioeconomic position Non-manual worker 3299 (19 7) 42 895 (25 6) 619 (19 7) 8143 (25 9) Skilled worker 2498 (14 9) 22 461 (13 4) 480 (15 3) 4311 (13 7) Unskilled worker 10 099 (60 3) 84 498 (50 4) 1912 (60 9) 15 576 (49 6) Missing 859 (5 1) 17 696 (10 6) 128 (4 1) 3360 (10 7) Education (years) 10 3003 (17 9) 20 096 (12 0) 529 (16 9) 3695 (11 8) 10 13 10 163 (60 7) 77 424 (46 2) 1942 (61 9) 14 512 (46 2) 13 3530 (21 1) 66 775 (39 9) 653 (20 8) 12 538 (39 9) Missing 59 (0 4) 3255 (1 9) 15 (0 5) 645 (2 1) Values in parentheses are percentages. RYGB, Roux-en-Y gastric bypass.

Alcohol and substance abuse after Roux-en-Y gastric bypass surgery 1339 Table 2 Preoperative incidence rate per 10 000 person-years, incidence rate ratio and postoperative hazard ratio for psychiatric diagnoses in the total gastric bypass surgery cohort compared with the general population, stratified by sex Preoperative IR Preoperative IRR Postoperative HR RYGB cohort Reference cohort Total Men Women Total Men Women Alcohol abuse 8 34 7 40 1 13 (1 00, 1 27) 0 70 (0 57, 0 85) 1 39 (1 20, 1 61) 2 73 (2 36, 3 15) 2 90 (2 30, 3 67) 2 59 (2 15, 3 12) Substance abuse 6 75 3 76 1 80 (1 56, 2 06) 0 98 (0 74, 1 28) 2 23 (1 90, 2 61) 3 17 (2 52, 3 99) 2 38 (1 47, 3 85) 3 48 (2 67, 4 53) Depression 29 50 12 24 2 41 (2 25, 2 58) 1 27 (1 06, 1 51) 2 81 (2 61, 3 03) 3 20 (2 81, 3 65) 3 35 (2 50, 4 48) 3 17 (2 74, 3 68) Suicide attempt 21 18 10 61 2 00 (1 84, 2 16) 1 21 (1 00, 1 46) 2 27 (2 08, 2 47) 2 85 (2 40, 3 39) 3 12 (2 11, 4 60) 2 76 (2 27, 3 35) Values in parentheses are 95 per cent confidence intervals. IR, incidence rate; IRR, incidence rate ratio; HR, hazard ratio; RYGB, Roux-en-Y gastric bypass. Table 3 Incidence rate per 10 000 person-years before and after surgery for psychiatric diagnoses in the reference cohort and the gastric bypass surgery cohort, and incidence rate ratio comparing pre- and postoperative incident rates, stratified by sex Preoperative IR Postoperative IR IRR Total Men Women Total Men Women Total Men Women Reference cohort Alcohol use disorders 7 40 16 64 5 20 16 64 28 59 12 93 2 25 (2 06, 2 45) 1 72 (1 50, 1 97) 2 49 (2 22, 2 79) Other substance use disorders 3 76 6 63 3 07 5 06 6 45 4 62 1 35 (1 15, 1 56) 0 97 (0 73, 1 27) 1 51 (1 25, 1 81) Depression 12 24 12 54 12 17 19 08 16 51 19 88 1 56 (1 44, 1 69) 1 32 (1 10, 1 57) 1 63 (1 50, 1 78) Suicide attempt 10 61 10 99 10 52 10 20 9 42 10 44 0 96 (0 87, 1 06) 0 86 (0 68, 1 07) 0 99 (0 88, 1 11) RYGB cohort Alcohol use disorders 8 34 11 57 7 25 64 78 105 36 52 34 7 77 (6 59, 9 15) 9 11 (6 91, 11 99) 7 22 (5 86, 8 89) Other substance use disorders 6 75 6 49 6 84 31 21 27 21 32 45 4 63 (3 72, 5 72) 4 19 (2 57, 6 66) 4 75 (3 71, 6 05) Depression 29 50 15 88 34 22 97 72 74 25 105 02 3 31 (2 95, 3 71) 4 68 (3 49, 6 21) 3 07 (2 70, 3 48) Suicide attempt 21 18 13 31 23 89 55 39 41 40 59 75 2 62 (2 25, 3 03) 3 11 (2 14, 4 45) 2 50 (2 12, 2 94) Values in parentheses are 95 per cent confidence intervals. IR, incidence rate; IRR, incidence rate ratio; RYGB, Roux-en-Y gastric bypass. use disorder before surgery compared with the reference cohort (Table 2). After RYGB surgery there was an increased risk of inpatient care for alcohol and substance use disorders, depression and attempted suicide in the entire surgery cohort compared with the reference cohort, affecting both men and women (Table 2). A significantly increased risk of inpatient care related to alcohol and substance use disorders after surgery was therefore seen only in men. In the reference cohort, IRRs were no different before and after pseudosurgery dates for substance use disorder in men, and for suicide attempts in both men and women. In the surgical cohort, the risks of all studied diagnoses were significantly higher after surgery than before (Table 3). Suicide attempts were about twice as common in the RYGB cohort before surgery than in the non-obese general population (Table 2). When comparing the IR values for attempted suicide before versus after surgery/pseudosurgery, the IR did not increase in the reference cohort. In the RYGB cohort the IR for attempted suicide was significantly higher after surgery than before (Table 3). The adjusted HR for attempted suicide in the RYGB cohort after surgery compared with that in the general non-obese population was 2 85 (95 per cent c.i. 2 40 to 3 39). Use of prescription drugs In the year before surgery, the surgical cohort was not at higher risk of being treated with medication for alcohol use disorders compared with the reference cohort. Over the 4-year study interval after RYGB, a greater proportion of patients in the surgery cohort had drugs prescribed for alcohol use disorders compared with the reference cohort. The percentage of patients treated for alcohol use disorders increased from 0 4 to2 6 per cent after 4 years. In the reference group, the percentage receiving treatment was significantly lower and stable over time, with a slight rise from 0 2 to0 3 per cent during the study interval (Fig. 1). In the year before surgery, 9 6 per cent of patients in the reference cohort were treated with antidepressant medication versus 27 0 per cent in the surgical cohort. Over the 4-year study period after RYGB there was a slight increase in the proportion of patients treated for depression in the surgical cohort, with 30 6 per cent receiving treatment 4 years after surgery, compared with 10 4 per cent in the reference cohort (Fig. 1).

1340 O. Backman, D. Stockeld, F. Rasmussen, E. Näslund and R. Marsk 3 2 1 RYGB Control 30 20 10 a Alcohol dependence b Antidepressants 10 5 30 20 10 c Benzodiazepines d Hypnotics and sedatives Fig. 1 Percentage of patients having Roux-en-Y gastric bypass (RYGB) surgery and control subjects who received at least one prescription each year for a alcohol dependence, b antidepressants, c benzodiazepines, and d hypnotic and sedatives in the year before surgery (year 0) and during a 4-year follow-up interval after surgery. *P 0 050 versus control group, P 0 050 before versus after RYGB (Wald χ 2 test) Patients in the surgical cohort were more likely to have of benzodiazepines in the year before surgery compared with the reference cohort (7 0 versus 3 0 per cent respectively), and there was a progressive increase during the 4-year follow-up interval, so that 10 0 per cent received in year 4 compared with 3 5 per cent in the control group (Fig. 1). Patients in the surgical cohort were more likely to have of hypnotics and sedatives before surgery than those in the reference cohort (16 2 versus 6 1 per cent), and there was a progressive increase during the 4-year follow-up period so that 24 5 per cent had at least one prescription in year 4 compared with 7 5 per cent in the control group (Fig. 1). Discussion After RYGB surgery, men are at a higher risk of being cared for in hospital with alcohol and substance use disorders compared with an age- and sex-matched, non-obese reference cohort from the Swedish population, controlling for presurgery treatment history. More patients after RYGB were treated with medications for alcohol use disorders than the reference cohort. Patients who underwent RYGB also had an increased risk of attempted suicide after surgery compared with before. There was no such increase in the reference cohort. The risk of depression did not decrease after RYGB surgery. This study lends further support to the observation that RYGB increases the risk of alcohol and substance

Alcohol and substance abuse after Roux-en-Y gastric bypass surgery 1341 use disorders. Previous studies based on interviews and questionnaires 7,19,20 demonstrated increases in alcohol and substance use disorders 2 years after bariatric surgery, particularly after RYGB. Patients who have undergone RYGB have also been shown 3 to have a twofold increased risk of a diagnosis of alcohol use disorder than patients who have undergone restrictive surgery. This may be related to the physiological alterations induced by RYGB. After RYGB there is a rapid increase in plasma alcohol concentration when patients consume one glass of wine, not seen after restrictive surgery 9. Results from the SOS study 4 also demonstrated that the risk of alcohol abuse and alcohol problems was increased after bariatric surgery. Conversely, it has been reported that high-risk drinking may be reduced after bariatric surgery, although the same study 21 found that 7 per cent of patients became new high-risk drinkers. The present data showed that prescriptions for the treatment of alcohol dependence increased over the study period. There is a significant positive association between depression and obesity in the general population, that is more marked in women 22. It is not surprising, therefore, that the RYGB surgery cohort had higher rates of depression before surgery. There was no improvement in medical treatment of depression after RYGB. The Longitudinal Assessment of Bariatric Surgery (LABS) study 23 reported that 35 2 per cent of patients were treated pharmacologically with antidepressants at baseline, decreasing to 27 5 per cent at 3-year follow-up. In a retrospective study 24 of patients undergoing RYGB, it was reported that 16 per cent had a decrease or discontinuation of antidepressants, although there was no change in antidepressant treatment for the majority after surgery. The SOS study 14 demonstrated initial improvements in depression with deterioration as follow-up lengthened, although after 10 years there was still an improvement compared with baseline values. It is not clear why the present data differ, although the SOS study involved mainly patients having restrictive procedures. An increased risk of suicide and suicide attempts has been reported previously 16, mostly within 3 years of surgery 18. In the present study there was an increased risk of suicide attempts resulting in hospitalization both before and after RYGB, compared with the reference cohort. There was no difference in the risk of suicide attempts before and after the pseudosurgery dates in the reference cohort, but in the surgical cohort there was a significantly increased risk of attempted suicide after surgery. A number of reasons for this have been suggested, including genetic susceptibility and changes in the release of gut peptides 25.Thereisa strong association between depression and suicide 26, so one explanation could be the lack of improvement in depression in the present cohort. A strength of the present study is that it was nationwide and based on national databases with high quality and completeness 27. The reference cohort represented the general population of Sweden with the same age and sex distributions as the RYGB cohort, so it was possible to make unbiased and relevant comparisons. By using two different outcome registries, with data on inpatient care and prescription drugs, it was possible to examine the consistency of the findings between these sources. The potential bias owing to patients lost to follow-up was avoided by the design of the present study. There were, nevertheless, some limitations. The control cohort was not matched for BMI, as information on weight loss was not available, and follow-up was short. The prescription drug data indicated only the drug prescribed and not the underlying disorder. This may have been altered after surgery. Some of the pathologies studied are known to be more common in the obese population, although the Scandinavian Obesity Surgery Registry observed, in its annual report for 2014 28, that at 5-year follow-up the mean BMI had decreased from 43 to 31 kg/m 2 for patients undergoing primary RYGB, with significant improvement in many of the co-morbid diseases associated with obesity. Many of the present outcome measures have been correlated to the degree of weight loss 29. It is possible that some of the differences observed in the present study reflect the increased medical attention paid to individuals in the surgical cohort and the fact that the surgical cohort had a lower socioeconomic position than the control cohort, even though this issue was taken into account in the Cox regression analysis. The relationships between alcohol and substance abuse, suicide and depression, and the weight loss and effects on physical co-morbidities induced by RYGB merit evaluation in prospective longitudinal studies. Acknowledgements This study was funded by Stockholm County Council and NovoNordisk Foundation. Disclosure: The authors declare no conflict of interest. References 1 Sjöström L. Review of the key results from the Swedish Obese Subjects (SOS) trial a prospective controlled intervention study of bariatric surgery. J Intern Med 2013; 273: 219 234. 2 Gloy VL, Briel M, Bhatt DL, Kashyap SR, Schauer PR, Mingrone G et al. Bariatric surgery versus non-surgical treatment for obesity: a systematic review and meta-analysis of randomised controlled trials. BMJ 2013; 347: f5934.

1342 O. Backman, D. Stockeld, F. Rasmussen, E. Näslund and R. Marsk 3 Ostlund MP, Backman O, Marsk R, Stockeld D, Lagergren J, Rasmussen F et al. Increased admission for alcohol dependence after gastric bypass surgery compared with restrictive bariatric surgery. JAMA Surg 2013; 148: 374 377. 4 Svensson PA, Anveden Å, Romeo S, Peltonen M, Ahlin S, Burza MA et al. Alcohol consumption and alcohol problems after bariatric surgery in the Swedish obese subjects study. Obesity (Silver Spring) 2013; 21: 2444 2451. 5 Heinberg LJ, Ashton K, Coughlin J. Alcohol and bariatric surgery: review and suggested recommendations for assessment and management. Surg Obes Relat Dis 2012; 8: 357 363. 6 Kudsi OY, Huskey K, Grove S, Blackburn G, Jones DB, Wee CC. Prevalence of preoperative alcohol abuse among patients seeking weight-loss surgery. Surg Endosc 2013; 27: 1093 1097. 7 King WC, Chen JY, Mitchell JE, Kalarchian MA, Steffen KJ, Engel SG et al. Prevalence of alcohol use disorders before and after bariatric surgery. JAMA 2012; 307: 2516 2525. 8 Klockhoff H, Näslund I, Jones AW. Faster absorption of ethanol and higher peak concentration in women after gastric bypass surgery. Br J Clin Pharmacol 2002; 54: 587 591. 9 Woodard GA, Downey J, Hernandez-Boussard T, Morton JM. Impaired alcohol metabolism after gastric bypass surgery: a case-crossover trial. J Am Coll Surg 2011; 212: 209 214. 10 Maluenda F, Csendes A, De Aretxabala X, Poniachik J, Salvo K, Delgado I et al. Alcohol absorption modification after a laparoscopic sleeve gastrectomy due to obesity. Obes Surg 2010; 20: 744 748. 11 Changchien EM, Woodard GA, Hernandez-Boussard T, Morton JM. Normal alcohol metabolism after gastric banding and sleeve gastrectomy: a case-cross-over trial. J Am Coll Surg 2012; 215: 475 479. 12 Polston JE, Pritchett CE, Tomasko JM, Rogers AM, Leggio L, Thanos PK et al. Roux-en-Y gastric bypass increases intravenous ethanol self-administration in dietary obese rats. PLoS One 2013; 8: e83741. 13 Mitchell JE, Selzer F, Kalarchian MA, Devlin MJ, Strain GW, Elder KA et al. Psychopathology before surgery in the longitudinal assessment of bariatric surgery-3 (LABS-3) psychosocial study. Surg Obes Relat Dis 2012; 8: 533 541. 14 Karlsson J, Taft C, Ryden A, Sjostrom L, Sullivan M. Ten-year trends in health-related quality of life after surgical and conventional treatment for severe obesity: the SOS intervention study. Int J Obes (Lond) 2007; 31: 1248 1261. 15 Booth H, Khan O, Prevost AT, Reddy M, Charlton J, Gulliford MC; Kings Bariatric Surgery Study Group. Impact of bariatric surgery on clinical depression. Interrupted time series study with matched controls. J Affect Disord 2015; 174: 644 649. 16 Adams TD, Gress RE, Smith SC, Halverson RC, Simper SC, Rosamond WD et al. Long-term mortality after gastric bypass surgery. NEnglJMed2007; 357: 753 761. 17 Adams TD, Mehta TS, Davidson LE, Hunt SC. All-cause and cause-specific mortality associated with bariatric surgery: a review. Curr Atheroscler Rep 2015; 17: 74. 18 Tindle HA, Omalu B, Courcoulas A, Marcus M, Hammers J, Kuller LH. Risk of suicide after long-term follow-up from bariatric surgery. Am J Med 2010; 123: 1036 1042. 19 Conason A, Teixeira J, Hsu CH, Puma L, Knafo D, Geliebter A. Substance use following bariatric weight loss surgery. JAMA Surg 2013; 148: 145 150. 20 Mitchell JE, Steffen K, Engel S, King WC, Chen JY, Winters K et al. Addictive disorders after Roux-en-Y gastric bypass. Surg Obes Relat Dis 2015; 11: 897 905. 21 Wee CC, Mukamal KJ, Huskey KW, Davis RB, Colten ME, Bolcic-Jankovic D et al. High-risk alcohol use after weight loss surgery. Surg Obes Relat Dis 2014; 10: 508 513. 22 de Wit L, Luppino F, van Straten A, Penninx B, Zitman F, Cuijpers P. Depression and obesity: a meta-analysis of community-based studies. Psychiatry Res 2010; 178: 230 235. 23 Mitchell JE, King WC, Chen JY, Devlin MJ, Flum D, Garcia L et al. Course of depressive symptoms and treatment in the Longitudinal Assessment of Bariatric Surgery (LABS-2) study. Obesity (Silver Spring) 2014; 22: 1799 1806. 24 Cunningham JL, Merrell CC, Sarr M, Somers KJ, McAlpine D, Reese M et al. Investigation of antidepressant medication usage after bariatric surgery. Obes Surg 2012; 22: 530 535. 25 Mitchell JE, Crosby R, de Zwaan M, Engel S, Roerig J, Steffen K et al. Possible risk factors for increased suicide following bariatric surgery. Obesity (Silver Spring) 2013; 21: 665 672. 26 Bostwick JM, Pankratz VS. Affective disorders and suicide risk: a reexamination. Am J Psychiatry 2000; 157: 1925 1932. 27 Ludvigsson JF, Andersson E, Ekbom A, Feychting M, Kim JL, Reuterwall C et al. External review and validation of the Swedish national inpatient register. BMC Public Health 2011; 11: 450. 28 Scandinavian Obesity Surgery Registry. Årsrapporter 2014. http://www.ucr.uu.se/soreg/index.php/arsrapporter [accessed 7 June 2016]. 29 de Zwaan M, Hilbert A, Swan-Kremeier L, Simonich H, Lancaster K, Howell LM et al. Comprehensive interview assessment of eating behavior 18 35 months after gastric bypass surgery for morbid obesity. Surg Obes Relat Dis 2010; 6: 79 85.