This study applied a retrospective cohort design based on medical chart review. We collected data on 261 consecutive obese adults (BMI ≥30 kg/m²) who underwent LAGB, RYGB, or SG between 2008 and 2011 in the surgical departments of seven different Korean tertiary medical centers. Patients aged 17 years old or less were excluded. Patients were categorized into three groups depending on the surgical procedure: LAGB (n=72), RYGB (n=73), and SG (n=116). The ethics review boards of the National Evidence-based Healthcare Collaborating Agency and each participating hospital approved the study protocols.

Baseline height, weight, blood pressures, and biochemical data were collected during the physical examination and laboratory assessments performed within a few days of bariatric surgery. Subsequent data were collected when the subjects made follow-up visits to the hospitals. Weight measurements at routine visits were averaged over three months to determine changes in weight at specific time points. For example, if patients measured their weight twice over three months from baseline, the weight at three months was recorded as the average value of the two measurements. Information on the patients' medical history before bariatric surgery was obtained through chart review. To investigate complications related with bariatric surgery, details of each event, onset date, and hospitalization were recorded.

Details on each of the surgical bariatric procedures are provided hereafter. Briefly, for LAGB, a four- or five-port laparoscopic approach was used, depending on the surgeon's preference. The pars flaccida technique was applied for creation of the retrogastric tunnel, and either the LAP-BAND® (Allergan, Irvine, CA, USA) or the Realize Band® (Johnson & Johnson, Neenah, WI, USA) was used. Several gastrogastric fixation sutures with nonabsorbable suture material were applied in all cases. The access port was exteriorized through a subcutaneous tunnel, and was placed by fixation to the rectus abdominis fascia. In all cases, intraoperative band filling was not performed, allowing at least four to eight weeks for capsule maturation around the band.

For RYGB, a six- or seven-port laparoscopic technique was used, and after creation of a small gastric pouch of less than 30 cc, a side-to-side gastrojejunostomy was constructed using a linear stapler. The measured Roux limb length was about 100 cm, and side-to-side jejunojejunostomy was performed using linear staplers. Potential areas for internal hernias were all suture-closed with nonabsorbable suture materials.

Finally, for SG, a five- or six-port laparoscopic technique was used. Omental dissection extended from about 5 cm proximal to the pylorus up to the angle of His; using a bougie with a diameter ranging from 36–40 French or a gastrofiberscope, several linear staples were applied, and a narrow sleeve was achieved. Suture reinforcement was performed or not, depending on the surgeon's preference.

Clinical effectiveness was investigated using four parameters: percentage of weight loss (%WL), percentage of excess weight loss (%EWL), percentage of excess BMI loss (%EBMIL), and absolute weight loss (AWL) from baseline. Percentage of WL was calculated by dividing changes in weight from baseline by baseline weight; %EWL was calculated by dividing weight changes from baseline by excess weight, which was obtained by subtracting ideal body weight from actual baseline weight. Ideal body weight was calculated as the height-adjusted weight depending on sex for a medium frame, according to the Metropolitan Life Insurance Company tables (1983).7 Percentage of EBMIL was calculated by dividing BMI change from baseline by excess BMI, which was obtained by subtracting the ideal BMI (25 kg/m²) from the actual BMI. AWL from baseline was calculated by subtracting weight at a specific time point from weight at baseline.

Remission from comorbidities, including diabetes, hypertension, and dyslipidemia, was investigated over time. In most cases, the management and follow-up of these comorbidities was usually performed by internists or primary care physicians, not by bariatric surgeons. Using the following definitions of concomitant diseases, prevalence before and improvement after bariatric surgery were investigated objectively. Diabetes was defined as the use of antidiabetic medications, fasting blood glucose (FBG) ≥126 mg/dL, or a hemoglobin A1c (HbA1c) level ≥6.5%. Hypertension was defined as the use of antihypertensive medications, systolic blood pressure ≥140 mm Hg, or diastolic blood pressure ≥90 mm Hg. Dyslipidemia was defined as the use of cholesterol-lowering medications, total cholesterol ≥240 mg/dL, triglycerides ≥200 mg/dL, or low-density lipoprotein cholesterol ≥160 mg/dL. Remission was defined as the patient being off medication and having laboratory/measurement values below the diagnostic criteria.

Safety assessment was performed by collecting clinical information at two time points: within 30 days and after 30 days from bariatric surgery. Within 30 days, any occurrence of specific complications was noted, including deep vein thrombosis, venous thromboembolism, tracheal reintubation, performance of interventional endoscopic procedure for control of luminal bleeding, dilatation of strictures, air reduction in a kinked gastric tube, tracheostomy placement, percutaneous drain placement, anastomotic complication (e.g., leak, obstruction), bowel obstruction, incisional hernia, band slippage, gastric perforation, port/tubing system complication, pouch dilatation, or pneumonia. After 30 days, any occurrence of gastroesophageal reflux disease, nausea/vomiting, internal hernia, gastric prolapse, band erosion, band removal, anastomotic stenosis, gallbladder stone, or marginal ulcer was also recorded. If patients reported other complications, they were specified in detail.

We summarized the patient characteristics and used analysis of variance (ANOVA) for continuous variables and the chi-square test for categorical variables to investigate differences in baseline characteristics among the three groups. Changes in %WL, %EWL, %EBMIL, and AWL were summarized at 0 (baseline), 3, 6, 9, 12, 15, and 18 months, and 95% confidence intervals were also calculated. ANOVA was used for the statistical testing of outcomes among the three groups at each time point. To reduce data entry error, two people independently input data into the data collection forms, which were developed using Microsoft Access software (2007). Subsequently, the accuracy of the inputted data was checked with SAS and corrected. Statistical data analysis was conducted with SAS software (9.2, SAS Institute, Cary, NC, USA).

Changes in %WL, %EWL, %EBMIL, and AWL are listed in Table 2, and presented graphically in Fig. 1. Patients who underwent RYGB showed a tendency towards greater weight loss than those who underwent SG or LAGB; the difference was statistically significant only at a few time points. The %WL was significant at 9 months (RYGB vs. LAGB), and AWL was significant at 6, 9, and 15 months (RYGB vs. LAGB). The mean %EWL at the last follow-up (18 months) was 52.1% for SG, 61.0% for LAGB, and 69.2% for RYGB.

The prevalence of dyslipidemia was similar among the three groups before surgery. Meanwhile, diabetes and hypertension were more prevalent in the RYGB group, followed by SG. Remission from all concomitant diseases (diabetes, hypertension, and dyslipidemia) was more frequent in those who underwent RYGB (Table 3). While 65.9%, 63.6%, and 100% of patients showed remission from diabetes, hypertension, and dyslipidemia, respectively, in the RYGB group, remission rates were lower in the SG group (30.2%, 14.3%, and 73.7%) and the LAGB group (40%, 34.8%, and 66.7%). Laboratory measurements, including FBG and HbA1c, improved significantly from baseline values in the RYGB and SG groups.

Early complications (within 30 days from surgery) occurred in 26 patients (9.9%), accounting for 12.5% of the LAGB, 12.3% of the RYGB, and 6.9% of the SG, while complications later than 30 days from surgery occurred in 32 patients (12.3%), representing 22.2% of the LAGB, 12.3% of the RYGB, and 6% of the SG (Table 4). The most frequent complication within 30 days from bariatric surgery was surgical site infection, followed by pneumonia/atelectasis. One patient who had undergone RYGB underwent reoperation because of small bowel obstruction. Other complications included leakage, bleeding, kinking of the gastric sleeve, stoma obstruction, rhabdomyolysis, band slippage, access port complications, diarrhea, and reflux symptoms. Complications at more than 30 days after surgery were more common in the LAGB group, although this difference was not statistically significant. Late band complications were mainly due to occurrence of port leak/revision and band erosion/removal. Five bands (6.9%) needed to be removed because of band erosion, access port infection, or band leakage. Other late complications included cholecystitis, internal hernia, intestinal perforation, wound dehiscence, and access port problems. Details on complications by type of surgery are described in Table 5. Some patients had more than one complication. There was one death (1/261=0.38%) due to aspiration pneumonia within 30 days from surgery in a patient who underwent RYGB. The other groups recorded no mortality.