Chapter  103:  Pharmacological and Surgical Treatment of Obesity

Appetite Suppressants

Medications have been prescribed for their ability to suppress appetite for over half a century. The first prescription appetite suppressants were the sympathomimetic amphetamine derivatives, so described because they exert their effects by stimulating the sympathetic nervous system. Some of the newer appetite suppressants exert their effects by mimicking the sympathetic nervous system.

Sibutramine is a combined norepinephrine and serotonin reuptake inhibitor. Its putative effect on weight loss is attributed to appetite suppression and increased thermogenesis, secondary to stimulation of brown adipose tissue. Sibutramine was approved for use in conjunction with a low calorie diet as an aid to weight loss in 1998.²³

Fluoxetine is a selective serotonin reuptake inhibitor (SSRI) that was originally approved to treat depression. The original manufacturer submitted an New Drug Application for use of fluoxetine as a weight loss drug in the early 1990s; however, approval was not given, and the application was eventually withdrawn.²⁴

Sertraline, like fluoxetine, is a SSRI. In the early 1990s, it was noted that sertraline administered to laboratory animals resulted in weight loss.^{25, 26}

Phentermine is a sympathomimetic amine of the β-phenethylamine family. It was approved for use by the FDA in 1959 as a short term aid to weight loss in conjunction with a low calorie diet and exercise. Unlike use of sibutramine, use of phentermine leads to the development of tolerance.²⁷

Diethylpropion, like phentermine, is a sympathomimetic agent prescribed for short-term weight loss when used in conjunction with diet and exercise. Diethylpropion is similar in chemical structure to bupropion, which is approved as an antidepressant and as a smoking cessation aid and has also been tested as a weight loss aid.²⁸

Zonisamide was approved by the FDA in 2000 for the treatment of partial (focal) seizures in adults with epilepsy, in conjunction with other anticonvulsants. Although the precise mechanism of action is unknown, it may exert its effects by acting as a sodium or calcium channel blocker. Because one of zonisamide's side effects is appetite suppression, its use as a weight loss drug has been tested.²⁹

Topiramate is also an anticonvulsant, approved in the mid 1990s for the treatment of refractory seizures in conjunction with other anticonvulsants. In the process of testing topiramate for treatment of mood disorders, it was discovered that the agent might mitigate the weight gain often observed with antidepressant treatment³⁰, and a dose-ranging study established it does so in a dose-dependent manner.^{31, 32}

Lipase Inhibitors

Lipase inhibitors putatively aid weight loss by reversibly binding to the active center of the enzyme lipase, preventing the digestion and absorption of some dietary fats. Orlistat was approved in the late 1990s and is currently the only lipase inhibitor approved for weight loss. Orlistat inhibits approximately 30 percent of fat absorption, including the absorption of fat-soluble vitamins.³³

Restrictive Procedures

Gastric banding. Gastric banding achieves weight loss by creating gastric restriction. The uppermost portion of the stomach is encircled by a band to create a gastric pouch with a capacity of approximately 15 to 30 cc. The band consists of an inflatable doughnut-shaped balloon whose diameter can be adjusted in the clinic by adding or removing saline via a reservoir port that is positioned beneath the skin. When the procedure was introduced, the bands were of a fixed size. However, the bands used today are adjustable, which allows the size of the gastric outlet to be modified as needed, depending on the rate of a patient's weight loss. Gastric banding does not produce malabsorption. Currently, almost all of the banding procedures are performed laparoscopically. While it is technically possible to remove the band (e.g., for failed weight loss or complications), doing so will expose the patient to potential risks associated with a second operation and, of course, will necessitate identifying an alternative method for weight loss.

Vertical Banded Gastroplasty (VBG) and other gastroplasty procedures. VBG uses the strategy of mechanical restriction to cause weight loss. The upper part of the stomach is stapled to create a narrow gastric inlet or pouch that remains connected with the remainder of the stomach. In addition, a non-adjustable band is placed around this new inlet in an attempt to prevent future enlargement of the stoma. As a result, patients experience a sense of fullness after eating small meals. Weight loss from this procedure results entirely from eating less: There is no component of malabsorption. VBG was one of the more common surgical procedures for weight loss in the late 1980s and early 19990s but has been superseded since 1995 by adjustable band procedures and procedures that combine mechanical restriction with bypass (see below).

Variations of gastroplasty procedures include horizontal gastroplasty and gastric partitioning without a band. These procedures are no longer performed because they had a high failure rate; thus, they are only of historic interest.

Bypass Procedures

Jejunoileal bypass. Jejunoileal bypass was one of the earliest procedures performed for weight loss. This procedure connected the proximal small intestine to a segment of distal small intestine (located a short distance upstream from the ileocecal junction), which bypassed the majority of the absorptive capacity of the small intestine. Although this procedure produced significant weight loss, patients developed complications such as severe malnutrition, chronic diarrhea, and liver failure. Thus, the procedure was abandoned about 25 years ago and is generally of only historical interest.

Combination Procedures

Roux-en-Y Gastric Bypass (RYGB). RYGB achieves weight loss through a combination of gastric restriction and malabsorption. Reduction of the stomach to a small gastric pouch (30 cc) results in feelings of satiety following even small meals. In addition, because this small pouch is connected to a segment of the jejunum (which is downstream), thus bypassing the duodenum and very proximal small intestine, absorptive function is reduced. The resultant “dumping” of sugar may also aid weight loss via the production of some unpleasant gastrointestinal symptoms following ingestion of high-density carbohydrate-containing foods. Typical symptoms include abdominal pain, cramping, and diarrhea.

The RYGB procedure has been performed regularly since the early 1980s; it was first performed laparoscopically in the early 1990s. RYGB is one of the most common types of weight loss procedures in current use, with almost 50,000 cases performed in 2001.

Biliopancreatic diversion (BPD). BPD, like RYGB, combines both the restrictive and malabsorptive strategies of obtaining weight loss. The stomach is partially resected, but the remaining capacity is generous compared to that achieved with the RYGB. As such, patients eat relatively normal-sized meals and do not need to restrict intake severely. Because the most proximal areas of the small intestine (i.e., the duodenum and jejunum) are bypassed, substantial malabsorption occurs. Although this procedure is not as commonly performed as either banding procedures or RYGB, the approach is strongly favored by some bariatric surgery specialists. The partial biliopancreatic diversion with duodenal switch is a variant of the BPD procedure that, until recently, was performed mostly in Italy and only rarely performed in the United States. Recently, a number of centers in the United States and Canada have begun to perform this procedure, which involves resection of the greater curvature of the stomach, preservation of the pyloric sphincter, and transection of the duodenum above the ampulla of Vater with a duodenoileal anastomosis and a lower ileo-ileal anastomosis.

Additional Sources of Evidence

U.K. University of Aberdeen report. Broom and colleagues prepared a systematic review of the long-term outcomes of treatments for obesity, the implications for health improvement, and the economic consequences, for the National Health Service of the United Kingdom.³⁴ Members of the project team kindly shared with us the draft report October 2002. We ordered all studies referenced therein; using their search terms, we conducted an update of their library search. Search terms are provided in Appendix A. The final report has since been released.

Orlistat review. In 2001, O'Meara and colleagues published a rapid and systematic review of the clinical effectiveness and cost-effectiveness of orlistat in the management of obesity.³⁵ The review was published as a Health Technology Assessment by the United Kingdom National Health Service (NHS). We ordered all studies referenced therein.

Sibutramine review. We identified a high-quality meta-analysis on the effects of sibutramine that was “in press” at the time of our search.³⁶ The first author of this review agreed to allow us to incorporate their results into our evidence report. The authors of that review did not restrict their literature search by language and made extensive efforts to identify unpublished and ongoing trials, including contacting representatives of the pharmaceutical industry.

NICE surgery review. In July 2002, the United Kingdom's National Institute for Clinical Excellence (NICE) published an assessment report³⁷ on the clinical and cost effectiveness of surgery for morbidly obese patients, authored by Clegg and colleagues at the University of Southampton. We ordered all studies referenced therein.

Cochrane surgery review. Our literature search also identified a Cochrane review³⁸ on surgery for morbid obesity, published by the team that authored the NICE surgery review. The Cochrane review was updated in February 2003 during our literature search process. The authors included both randomized controlled trials (RCTs) and non-randomized controlled trials comparing surgery with nonsurgical management for morbid obesity and assessed RCTs comparing different surgical procedures. The review was restricted to adults age 18 years or older with morbid obesity defined as a BMI greater than 40 or a BMI index greater than 35 with serious comorbid disease. In addition to summarizing their results, we obtained copies of studies referenced therein and conducted an update of their library search.

Extraction of Study-Level Variables and Results

We abstracted data from the articles that passed our screening criteria onto a specialized Quality Review Form (See Appendix B). The form contains questions about the study design, the number of patients and comorbidities, dosage, adverse events, the types of outcome measures, and the time from intervention until outcome measurement. With input from the project's TEP, we selected the variables for abstraction. Three reviewers, working in groups of two, extracted data from the same articles and resolved disagreements by consensus. A senior physician resolved any disagreements not resolved by consensus.

To evaluate the quality of the studies, we collected information on the study design, withdrawal/dropout rate, method of random assignment (and blinding), and method for concealment of allocation (the attempt to prevent selection bias by concealing the assignment sequence prior to allocation). We also calculated the percentage of attrition by dividing the number of persons who dropped out of the trial (i.e., the number of people who entered the trial minus the number who completed the trial) by the number of persons entering the trial. The elements of design and execution (randomization, blinding, and withdrawals) have been aggregated into a summary score developed by Jadad.³⁹ The Jadad score rates studies on a 0 to 5 scale, based on the answer to three questions:

1. Was the study randomized?

2. Was the study described as double-blind?

3. Was there a description of withdrawals and dropouts?

One point is awarded for each “yes” answer, and no points are given for a “no” answer. Additional points are awarded if the randomization method and method of blinding were described and were appropriate. A point is deducted if the method is described but is not appropriate. Empirical evidence has shown that studies scoring 2 or fewer points show larger apparent differences between treatment groups than do studies scoring 3 or more.⁴⁰

Choice of Outcomes

The outcome of interest specified by our sponsor was weight loss. Excess weight is associated with other negative health outcomes, such as diabetes mellitus, hypertension, sleep apnea, and osteoarthritis. A loss of 5 percent to 10 percent of body weight by an obese person, followed by long-term weight maintenance, is associated with improved health outcomes.¹⁶

Weight loss can be measured in several ways, most commonly kilograms of weight lost, “excess” weight loss, or percentage of excess weight loss. Excess weight is defined as that above a patient's “ideal body weight” based on height and weight tables. Among these, the most commonly reported outcome, by far, is kilograms (or pounds) of weight lost. Among 111 surgical studies reporting weight loss that we reviewed, 43 reported weight loss only in terms of kilograms or pounds, 17 reported only excess weight loss or some variant, 46 reported both of these outcomes, and 5 reported neither of these. Among the 76 pharmaceutical studies that we assessed, none reported excess weight loss. Therefore, we chose weight loss (in kilograms or pounds) as the principal outcome measure, since this choice allowed us to include the maximum number of studies in our analysis and afforded us the only way of comparing the effectiveness of surgical and pharmaceutical therapies across studies.

Because weight loss achieves health benefits primarily by reducing the incidence or severity of weight-related comorbidities like diabetes, we also endeavored to assess treatments by comparing their effects on these outcomes. Very few of the pharmaceutical studies reported these outcomes; thus, it was not feasible to make cross-study comparisons to address the control of comorbidities. We did search for studies that made within-study comparisons of the control of comorbidities, and we summarize their findings in this report. We also assessed the case series reports of obesity surgery for the control of selected comorbidities and compared these results to those reported in studies containing within-group comparisons.

Selection of Trials for Meta-Analysis

The outcome of interest was weight loss between baseline and followup. Based on clinical considerations, we focused on weight loss measured at six months or later. To make our analyses comparable, we stratified the analysis in the same manner as the “in press” meta-analysis on sibutramine.³⁶ Thus, we defined data collected at “six months” to be data collected at any point between 16 and 24 weeks; likewise, “one year” followup data were collected at any point between 44 and 54 weeks. If a study presented data for two or more time points in an interval, e.g., at 16 and 18 weeks, we chose the longest followup measurement for our analysis.

For some trials, several publications presented the same outcome data. In these cases, we picked the more informative of the duplicates; for example, if one publication was a conference abstract with preliminary data and the second was a full journal article, we chose the latter. The publications dropped for duplicate data do not appear in the evidence table but are noted in the text. We note that multiple citations of the same article were removed at the title screening stage of the project.

For a trial to be included for further analysis, the associated publication(s) had to report on weight loss, provide data prior to the crossover point if the trial was a crossover design, and contain sufficient statistical information for the calculation of a mean difference at six months and/or one year followup as defined below.

Mean Difference

Each trial contained one control or placebo group, which is referred to below simply as the control group. Some trials contained more than one medication group, e.g., at different dose levels. In order not to double-count patients for each trial, we chose the most clinically relevant medication group for our analysis, or in some cases, we combined medication groups. In the discussion below, we assume that a single medication group per study has been determined by choice or defined by combination.

For each trial, we extracted the followup mean weight loss for the control group, the followup mean weight loss for the medication group, and the standard deviation for each group. For studies that included measures for both a six-month and a 12-month followup, we collected those measures separately. If a study did not report a followup mean or if a followup mean could not be calculated from the given data, the study was excluded from analysis. We extracted weight loss as a positive quantity—i.e., greater than zero. For studies that did not report a standard deviation or for which a standard deviation could not be calculated from the given data, we imputed the standard deviation by using those studies and groups that did report a standard deviation and weighting all groups equally.

We converted all means and standard deviations to kilograms. The vast majority of studies measured weight loss in pounds or kilograms. For the few that measured weight loss in terms of BMI, we converted BMI to kilograms by assuming an average height of 5 feet 4 inches. (Past experience has shown that our results—efficacy of weight loss therapies—vary little if any height between 5 feet and 6 feet is chosen to convert BMI to weight.) We then calculated a mean difference for each study, which was the difference between the control group followup mean weight loss and the medication group followup mean weight loss:

mean difference = control followup mean - medication followup mean

We also estimated the standard deviation for that mean difference.⁴² A negative mean difference indicates that the medication group experienced more weight loss at followup than did the control group. For example, if the medication group lost 3 kg on average and the control group lost 1 kg, the mean difference is -2 kg. The mean difference is readily interpretable because it is measured in kilograms.

Sensitivity Analyses

We conducted sensitivity analyses on four study dimensions: Jadad quality score (less than or equal to 2 versus greater than or equal to 3); year of publication (1998 or earlier versus 1999 or later); completion rate (less than 80 percent versus 80 percent or greater; and less than 70 percent versus 70 percent or greater); and dosage (for the study⁴³ of bupropion, the 300 mg arm was used rather than the 400 mg arm; for the fluoxetine analysis, the study⁴⁴ with the 20 mg dose was excluded because all other studies used a 60 mg dose; for the topiramate analysis, the 96 mg study arms were analyzed separately from the 192 mg study arms). We tested for differences between subgroups—e.g., the high-quality versus lower-quality studies—by conducting a meta-regression analysis using a single dichotomous variable to indicate subgroup membership.

Performance of Meta-Analysis

For the 6-month and 12-month analyses, respectively, we estimated a pooled random-effects estimate⁴⁵ of the overall mean difference. The individual trial mean differences are weighted by both within-study variation and between-study variation in this synthesis. We also report the chi-squared test of heterogeneity p-value based on Cochran's Q.⁴⁶ We constructed a forest plot in which each individual trial mean difference is shown as a box whose area is inversely proportional to the estimated variance of the mean difference in that trial. The trial's confidence interval is shown as a horizontal line through the box. The pooled “weighted mean difference” and its confidence interval are shown as a diamond at the bottom of the plot with a dotted vertical line indicating the pooled estimate value. A vertical solid line at zero indicates no effect of medication on weight loss.

Publication Bias

We assessed the possibility of publication bias by evaluating a funnel plot of the trial mean differences for asymmetry, which can result from the nonpublication of small trials with negative results. These funnel plots include a horizontal line at the pooled fixed-effects estimate and pseudo-95% confidence limits.⁴⁷ If bias due to nonpublication exists, the distribution is asymmetric or skewed. Because graphical evaluation can be subjective, we also conducted an adjusted rank correlation test⁴⁸ and a regression asymmetry test⁴⁷ as formal statistical tests for publication bias. The correlation approach tests whether the correlation between the mean differences and their variances is significant, and the regression approach tests whether the intercept of a regression of the mean differences on their precision differs from zero; that is, both approaches formally test for asymmetry in the funnel plot. We acknowledge that other factors, such as differences in trial quality or true study heterogeneity, could produce asymmetry in funnel plots.

Extraction of Adverse-Event Data

Each trial included in the weight loss analysis was examined to determine whether it reported data on adverse events. Adverse events were recorded onto a spreadsheet that identified each trial group, the description of the adverse event as listed in the original article, and the number of subjects in each group. We then abstracted either the number of events or the number of people, depending on how the trial chose to report events. The majority of trials recorded the number of events, rather than the number of unique people who experienced the event. Each event was counted as if it represented a unique individual. Because a single individual might have experienced more than one event, this assumption may have overestimated the number of people having an adverse event. In fact, for some adverse events in a few trials, the number of events reported in a group was greater than the number of individuals in that group. In those cases, we assumed that all individuals in the group experienced the adverse event when we calculated the risk of the event in that group, as described below.

If a report of a trial mentioned a particular type of adverse event in the discussion but did not report data on that adverse event, we did not include that trial in that particular event's analysis. In other words, we did not assume zero events occurred unless the trial report specifically stated that zero events were observed. By taking this approach, we may have overestimated the number of patients for whom a particular adverse event was observed.

Table 1. Adverse events in medication studies

Table 1. Adverse events in medication studies

Adverse-event category	Descriptives aggregated into this category	Orlistat	Fluoxetine	Bupropion	Topiramate
Anorexia	Anorexia				x
Nausea/vomiting	Nausea	X	X
	Nausea/vomiting	X	X
	Vomiting	X
	Nausea and vomiting		X
	Nausea/vomiting/diarrhea/vomiting		X
Back pain	Back pain	X	X
Bloating/abdominal pain/dyspepsia	Abdominal distention	X
	Abdominal pain	X	X
	Dyspepsia	X	X
	Dyspepsia/hiatal hernia/esophagitis	X
	Gastritis	X
Central nervous system effects	Anxiety			X	X
	Concentration/attention difficulty				X
	Concentration difficulty				X
	Decreased concentration			X
	Depression				X
	Difficulty with concentration				X
	Difficulty with concentration/attention				X
	Difficulty with memory				X
	Dizziness			X	X
	Insomnia			X	X
	Memory difficulty				X
	Mood problems				X
	Nervousness				X
	Somnolence				X
Constipation	Constipation			X	X
Decreased libido/sexual dysfunction	Decreased libido/ sexual dysfunction		X
Depression/mood change	Depression	X	X
	Mood change		X
	Suicide attempt	X
Diarrhea	Diarrhea	X	X	X	X
	Discolored feces	X
	Fatty/ oily stool	X
	Fecal incontinence	X
	Fecal urgency	X
	Increased defecation	X
	Liquid stools	X
	Loose stools	X
	Oily evacuation	X
	Oily spotting	X
	Soft stools	X
	Stool fat	X
	Uncontrolled oily discharge	X
Dry mouth	Dry mouth			X	X
Fatigue/asthenia/hypersomnia/somnolence	Asthenia		X
	Fatigue		X		X
	Hypersomnia		X
	Somnolence		X
Flatulence	Flatulence	X
	Flatus with discharge	X
Gallbladder problems	Cholecystectomy	X
	Cholelithiasis	X
	Gallbladder abnormalities	X
	Gallstones	X
Headache	Headache	X	X	X	X
Hepatic abnormalities	Elevated transaminases	X
	Hepatocellular damage	X
	Liver disorders	X
Hypertension	Hypertension			X
Increased hepatic enzymes	Increased hepatic enzymes				X
Insomnia	Insomnia		X
Nervousness/ sweating/ tremors	Nervousness		X
	Sweating		X
	Tremor		X
Paresthesia	Paresthesia				X
Rhinitis	Rhinitis	X
	Upper respiratory tract infection		X
Seizure	Seizure			X
Taste perversion	Taste perversion				X
Upper abdominal symptoms	Abdominal pain				X
	Dyspepsia				X
	Nausea			X	X
	Nausea or vomiting			X
	Vomiting				x
Upper respiratory problems	Flu-like symptoms			X
	Other respiratory complaints			X
	Pharyngitis			X
	Rhinitis			X	x
	Upper respiratory complaints			X
	Viral infection			X
	Upper respiratory tract infection				X
Urticaria/pruritis/rash	Pruritis		X
	Rash		X
	Skin rashes		X
	Urticaria		X
Vertigo	Vertigo				X

Table 2. Adverse events in surgery studies

Table 2. Adverse events in surgery studies

Adverse-event category	Descriptives used in studies aggregated into this category
Anastomic or gastric outlet stenosis	Anas stenosis, Anastomotic stenosis, Anastomotic stricture, Gastric outlet stenosis, Gastric outlet obstruction, Gastric pouch outlet obstruction, Gastrojejunal anastomotic stenosis, Gastrojejunostomy anastomotic stricture, Narrowing of the communication lumen of 2 parts of stomach, Neostoma stenosis with porch enlargement, Outlet obstruction, Outlet stenoses, Pouch outlet obstruction, Stenosis, Stenosis at gastrojejunostomy, Stenosis of anastomosis, Stenosis of gastroplasty orifice, Stenosis of GJ, Stoma stenosis, Stomal stenosis, Stomal stenosis with endoscopy, Stomal stenosis with reoperation, Stomal stricture, Stricture
Anastomotic, gastric pouch or duodenal leak	Anastomotic failure, Anastomotic leak, Anastomotic leakage, Duodenal leak, Duodenal stomal obstruction, Gastric leak, GJ leak, Leak, Leak (asymptomatic/ contained), Leakage, Leakage/perforation, Loose suture in anastomosis, Surgical leak
Bleeding, all	Bleeding, Bleeding from suture line, Bleeding requiring reoperation or transfusion, Bleeding trocar site, Bleeding unstable BP, Endoluminal bleeding, Gastrointestinal hemorrhage, Gastric bleeding, Gastrointestinal bleeding, GI bleed, GI bleeding, GI hemorrhage, Hematemesis, Hematoma, Hemoperitoneum, Hemorrhage, Intra-abdominal bleeding, Intra-abdominal hemorrhage, Intraperitoneal bleeding, Interoperative hemorrhage, Mesenteric bleeding, Postoperative bleeding surgery required, Pouch hemorrhage, Significant bleeding, Staple line bleed, Staple line hemorrhage, Transfusion stable BP, Upper gastrointestinal bleed, Upper GI hemorrhage
Deep vein thrombosis / pulmonary embolism	Deep vein thrombosis, Deep venous thrombosis, DVT, DVT and PE, DVT/PE, Major pulmonary embolus, Nonfatal pulmonary embolism, Nonfatal pulmonary embolus, PE, Popliteal vein thrombosis, Pulmonary embolism, Thrombosis/ lung embolism
Gastrointestinal symptoms, all	Abdominal pain, Acute cholecystitis, Anorexia, Bypass enteritis, Cholecystectomy, Cholecystolithiasis, Cholelithiasis, Cirrhosis, Colitis, Constipation, Dairy food intolerance, Diarrhea, Difficulty in eating red meat, Dumping, Dumping syndrome, Dysmotility of esophagus, Dyspepsia, Dysphagia, Epigastric pain, Erosive esophatitis, Esophageal dilation, Esophagitis, Food intolerance, Frequent vomiting, Gallstones, Gastritis, Gastritis by endoscopy, Gastritis, esophagitis, Gastroesophageal reflux, GERD, Heartburn, Hepatic failure, Ileus, Intolerable malodorous gas, Involuntary vomiting, Lap cholecystectomy, Liver disease, Liver function abnormal, Lower GI hemorrhage, Mild dumping, Nausea, Nausea and vomiting, Obstipation, Obstructive ileus, Perianal abscess, Postanesthetic jaundice, Prolonged ileus, Prolonged nausea/vomiting, Readmit vomiting, Recurrent vomiting, Reflux esophagitis, Refuse du systeme, Ructus, Severe vomiting, Sigmoiditis, Symptomatic cholelithiasis, Total food intolerance, Transient hepatic dysfunction, Vomiting, Vomiting one or more per week, Vomiting with stapleline intact
Gastrointestinal symptoms, reflux	Dyspepsia, Erosive esophatitis, Esophagitis, Gastritis, esophagitis, Gastroesophageal reflux, GERD, Heartburn, Reflux esophagitis, Refuse du systeme
Gastrointestinal symptoms, vomiting	Frequent vomiting, Involuntary vomiting, Nausea and vomiting, Prolonged nausea/ vomiting, Readmit vomiting, Recurrent vomiting, Severe vomiting, Vomiting, Vomiting one or more per week, Vomiting with stapleline intact
Medical (cardiac, stroke, severe hypertension)	A fib, Angina, Arrhythmia, Cardiac, Cardiovascular complications, CHF, CVA, MI, Severe hypertension, Severe tachycardia
Nutritional and electrolyte abnormalities	Anemia, B12 deficiency, Calcium, Electrolyte imbalance, Electrolytes abnormality, Folate deficiency, Folic acid, Iron deficiency, Iron deficient, Malnutrition, Nutritional deficiencies, Other vitamin deficiencies, Protein calorie malnutrition, Protein-calorie malnutrition, Temporary electrolytes imbalance, Vitamin B12, Vitamin and mineral deficiencies
Reoperations	Adhesiolysis, Band extraction, Band removal or revision, Bleeding requiring reoperation or transfusion, Conversion to duodenal switch, Normal but clinical failure, Open revision, Operation reversed, Patient request, Post-op exploratory laparotomy, Postoperative bleeding surgery required, Reoperation, Reanastomosis, Removal, Removal of band, Reop bowel obstruction, Reoperation for failure of reservoir, Reoperation revision, Reoperation, Reoperation due to malnutrition, Reoperation to reposition band, Revision, Same admission reoperation, Stomal stenosis with reoperation, Take downs, Stroke, Uncontrolled hypertension
Respiratory complications, all	ARDS, Atelectasis, Atelectasis/pneumonitis, Atelectesis/pneumonia, Bronchopneumonia, Chest infections, Effusion, Hemothorax, Hydrothorax, Massive subcutaneous emphysema, Other pulmonary complications, Pleura injury, Pleural effusion, Pneumonia, Pneumothorax, Post-op pneumonia, Postoperative respiratory distress, Prolonged apnea, Pulmonary problems, Pulmonary complications, Resp chest infection, Resp distress, Resp failure, Resp insufficiency, Resp rest, Respiratory, Respiratory complications, Respiratory failure, Respiratory insufficiency, Severe bronchospasm
Respiratory complications, pneumonia	Atelectesis/pneumonia, Bronchopneumonia, Pneumonia, Post op pneumonia, Dehiscence
Surgical, all	Abdominal or pulmonary collections, Abscess, Access port complications, Access port problems, Accidental band perforation, Acute obstruction, Adhesiolysis, Anas stenosis, Anastomotic failure, Anastomotic leak, Anastomotic leakage, Anastomotic stenosis, Anastomotic stricture, Anastomotic ulcers, Aneurysmal deformity of the balloon, Asymptomatic stomach erosions, Band dislocations, Band dislodgement or pouch dilation, Band disruption, Band erosion, Band extraction, Band infection, Band leakage, Band migration, Band penetration, Band rupturing, Band removal or revision, Band rupture, Band slippage, Band slipped out, Bleeding, Bleeding at port, Bleeding from suture line, Bleeding requiring reoperation or transfusion, Bleeding trocar site, Bleeding unstable BP, Bowel obstruction, C dificile colitis, Cervical esophageal mucosal tear from EEA anvil, Conversion, Conversion to duodenal switch, Conversion to open, Convert to open from bleeding, Convert to open gastric perforation, Converted to open, Dehiscence, Dehiscence of the vertical stapleline, Delayed pouch emptying, Dilated proximal pouch, Dilation or slippage, Disconnection, Disconnection of sc port, Distal Roux en Y leak, Duodenal leak, Duodenal stomal obstruction, Early band repositioning, Early pouch dilatation, Efferent limb obstruction,
	Endoluminal bleeding, Enlarged orifice, Enlarged porch, Enlarged pouch with obstructive angulation, Enteric fistula, Enterocutaneous fistular, Erosion, Erosion into stomach, Evisceration, Fascia dehiscence and retrogastric abscess, Fistula, Food impaction, Food obstruction of stoma, Gallbladder puncture, Gastric outlet stenosis, Gastrointestional hemorrhage, Gastric bleeding, Gastric erosion at ring, Gastric fistula, Gastric herniation through band, Gastric leak, Gastric outlet obstruction, Gastric perforation, Gastric pouch outlet obstruction, Gastric slippage, Gastric wall perforation, Gastric wall slippage, Gastric perforation, Gastrogastric fistula, Gastrointestinal bleeding, Gastrointestinal ulcer, Gastrojejunal anastomotic stenosis, Gastrojejunostomy anastomotic stricture, Gastrotomy during lesser curve dissection, GI bleed, GI bleed from erosion, GI bleeding, GI hemorrhage, GJ leak, Hematemesis, Hematoma, Hemoperitoneum, Hemorrhage, Hernia, Hernia/ small bowel obstruction with reoperation, Herniation of band, Herniation of stomach through the band causing obstruction, Hypopharyngeal perforation, Incarcerated hernia, Incisional hernia, Incomplete division of stomach, Infected hematoma, Infected splenic hematoma, Infection around the port, Infection at reservoir, Infected reservoirs, Infection wound major, Internal hernia, Intestinal obstruction, Intestinal obstruction JJ, Intestinal obstruction after 1 year, Intra-abdominal abscess, Intraabdominal bleeding, Intra-abdominal hemorrhage, Intraabdominal sepsis, Intractable ulcer, Intragastric band migration, Intraoperation perforation of stomach, Intraoperative complications, Intraperitoneal abscess, Intraperitoneal bleeding, Intraoperative hemorrhage, Intususception of gastric wall, J-J obstruction, Jejunujejunostomy obstruction, Laceration of liver/spleen, Laceration of reservoir, Lap band balloon leak, Large bowel perforation, Late band repositioning, Late pouch dilatation, Leak, Leak (asympt/ contained), Leak of reservoir, Leakage, Leakage of port from needle perforation 1, Leakage of ring, Leakage/ perforation, Leaking band, Leaking port, Leaks, LGI bleed, Liver hematoma, Liver injury, Loose suture anastomosis, Major infection, Major wound infection, Major wound problems, Marginal ulcer, Marginal ulcers NSAID induced, Mesenteric bleeding, Migration band, Minor dehiscence, Minor infection, Minor splenic injury, Minor wound infection, Minor wound problems, Miscellaneous, Narrowing of the communication lumen of 2 parts of stomach, Neostoma stenosis with porch enlargement, Normal but clinical failure, Obstruction at jejunojejunostomy, Obstruction at roux, Obstructive aneurysmal deformity, Open revision, Operation reversed, Other sepsis, Other wound infections, Outlet obstruction, Outlet stenoses, Pain in left shoulder, Painful port site, Pancreatitis, Partial obstruction, Patient request, Pelviperitonitis, Perforated stress ulcer, Perforation, Perforation of stomach, Perforation of stomach remnant, Perforation or erosion, Perforation pouch, Perioperative complications, Peritonitis, Plugging, Port, Port disconnection, Port infection, Port leakage, Port migration, Port problem, Port-site infection, Postoperative complications, Posterior herniation, Post-op exploratory laparotomy, Postoperative bleeding surgery required, Postoperative comp, Pouch dilation, Pouch hemorrhage, Pouch herniation, Pouch outlet obstruction, Prolapse of stomach, Reoperation, Reanastomosis, Removal, Removal of band, Reop bowel obstruction, Reop for failure of reservoir, Reop revision, Reoperation due to malnutrition, Reoperation to reposition band,
	Repair hiatal hernia, Repair spleen, Repair umbilical hernia, Reservoir infection, Reservoir flipped over, Reservoir infections, Reservoir leak, Reservoir/ tubing breaks, Resp infection, Retained foreign body, Retained sponge, Revision, Rupture of tubing, Same admission reoperation, Small bowel obstruction JJ, Separation of catheter and reservoir, Sepsis, Septicemia, Seroma, Seroma with necrotic fat, Severe wound infection, Significant bleeding, Silicone band migration, Slippage, Slippage of band, Slippage or dilation, Slipping, Small bowel necrosis, Small bowel obstruction, Splenectomy, Splenic artery rupture, Splenic capsule tear, Splenic injury, Splenic laceration, Stable line disruption, Staple breakdown, Staple disruption, Staple line bleed, Staple line disruption, Staple line failure, Staple line fistula, Staple line hemorrhage, Staple line failure, Stapler malfunction, Stapling of nasogastric tube, Stenosis, Stenosis at gastrojejunostomy, Stenosis at mesocolon, Stenosis of anastomosis, Stenosis of gastroplasty orifice, Stenosis of GJ, Stoma dilation, Stoma stenosis, Stoma widening, Stomach injury, Stomach perforation, Stomach slippage, Stomal stenosis, Stomal stenosis with endoscopy, Stomal stenosis with reoperation, Stomal stricture, Stomal ulcer, Stricture, Subhepatic abscess, Subphrenic abscess, Subphrenic abscess, Subphrenic collection, Superficial wound infection, Surgical leak, Surgical wound infection, Symptomatic ulcer disease, Take downs, Technical band problems, Transfusion stable BP, Trocar hernia, Trocar site hernia, Tube dropped off the injection port, UGI bleed, Ulcer, Umbilical hernia, Upper GI hemorrhage, Ventral hernia, Violation of integrity of stomach, Vomiting with stapleline disruption, Wound abscess, Wound dehiscence, Wound hematomas, Wound hernia, Wound infection minor, Wound infection, Wound infection (pus or erythema), Wound infection erythema, Wound infection major, Wound infection minor, Wound infection or dehiscence, Wound infection port, Wound infection pus, Wound infections major, Wound minor, Wound necrotic fat, Wound or port infection, Wound seroma
Surgical: incisional hernia	Hernia, Incarcerated hernia, Incisional hernia, Ventral hernia, Wound hernia
Surgical: internal hernia	Hernia/small bowel obstruction with reoperation, Internal hernia
Surgical: splenic injury	Infected splenic hematoma, Laceration of liver/ spleen, Minor splenic injury, Repair spleen, Splenectomy, Splenic artery rupture, Splenic capsule tear, Splenic injury, Splenic laceration, Deep vein thrombosis
Surgical: wound complications, all	Evisceration, Fascia dehiscence and retrogastric abscess, Hernia, Incarcerated hernia, Incisional hernia, Infection around the port, Infection at reservoir, Infected reservoirs, Infection wound major, Infection wound minor, Major infection, Major wound infection, Major wound problems, Minor dehiscence, Minor infection, Minor wound infection, Minor wound problems, Other wound infections, Port infection, Port-site infection, Reservoir infection, Resp infection, Seroma, Seroma with necrotic fat, Severe wound infection, Superficial wound infection, Surgical wound infection, Ventral hernia, Wound abscess, Wound dehiscence, Wound hematomas, Wound hernia, Wound infection minor, Wound infection, Wound infection (pus or erythema), Wound infection erythema, Wound infection major, Wound infection minor, Wound infection or dehiscence, Wound infection port, Wound infection pus, Wound infections major, Wound minor, Wound necrotic fat, Wound or port infection, Wound seroma
Surgical: wound, infection major	Infection around the port, Infection at reservoir, Infected reservoirs, Infection wound major, Major infection, Major wound infection, Port infection, Port-site infection, Reservoir infection, Severe wound infection, Wound abscess, Wound infection major, Wound infection port, Wound infection pus, Wound infections major
Surgical: wound, infection minor	Infection wound minor, Major infection, Minor infection, Minor wound infection, Superficial wound infection, Wound infection minor, Wound infection erythema, Wound infection minor, Wound minor

For each adverse-event subgroup, we report the number of trials that provided data for any event in the subgroup. We also report the total number of individuals in the medication groups in the relevant trials who were observed to have experienced the event and the total number of patients in the medication groups in those trials. We then report the analogous counts for the control groups in the relevant trials.

Meta-Analysis

For subgroups of events that occurred in two or more trials and occurred at least once in the medication group and at least once in the control group, we performed a meta-analysis to estimate the pooled odds ratio and its associated 95% confidence interval. Given that many of the events were rare, we used exact conditional inference to perform the pooling rather than applying the usual asymptotic methods that assume normality. Asymptotic methods require corrections if zero events are observed; generally, half an event is added to all cells in the outcome-by-treatment (two-by-two) table in order to allow estimation, because these methods are based on assuming continuity. Such corrections can have a major impact on the results when the outcome event is rare. Exact methods do not require such corrections. We conducted the meta-analysis using the statistical software package StatXact.⁴⁹

For interpretability, for any significant pooled odds ratio greater than one (which indicates that the odds of the adverse event being associated with medication is larger than the odds of the event associated with being in the control group), we calculated the RR and number needed to harm (NNH). To perform these calculations, we assume that the expected rate of the adverse event in an untreated population was equal to the observed crude rate among all control patients (the total number of adverse events observed across all control groups divided by the total number of control patients). The number needed to harm is the number of patients who would have to be treated with medication to produce one adverse event, on average, and is clinically more interpretable than the odds ratio. We calculated the risk ratio by making the same assumption about the expected rate among untreated patients.

Power Calculation

We also conducted a power calculation to determine the lowest adverse-event rate that the included medication trials had at least 80 percent power to detect. First, we assumed a sample size equal to the total number of patients receiving each medication across all trials for which we had sample size data, either reported directly in trials we reviewed or from sample size data reported in other systematic reviews. We then determined the lowest detectable adverse-event rate based on a two-sided test of level 0.05. This calculation was performed to assess the statistical power we actually had available to detect adverse events if few or none were observed. Even if no adverse events were observed, we cannot necessarily conclude that the rate is zero, because the available sample size may have been too small to detect a rare event.

Analysis of the Efficacy of Surgical Weight Loss

We first summarized the Cochrane review³⁸ on surgical studies of weight loss. For all surgical studies that we found, we also extracted the mean weight loss and its standard deviation for each study group, generally defined by surgery procedure, at 12 postoperative months and at the maximum followup time greater than or equal to 36 months, as available. We followed the same procedures regarding imputation, conversion to kilograms, and conversion of BMI measurements as we did for the medication meta-analysis.

For randomized controlled trials that reported a within-study comparison of two procedures of interest, a mean difference was calculated (mean weight loss in procedure “1” group minus mean weight loss in procedure “2” group). A positive mean difference indicates that patients in the procedure 1 group lost more weight on average than patients in the procedure 2 group. A negative mean difference indicates that patients in the procedure 1 group lost less weight on average than patients in the procedure 2 group. These mean differences were pooled using a random effects model, and a 95% confidence interval was estimated. For all studies, randomized or not, a pooled mean weight loss for each procedure group was estimated using a random effects model, and an associated 95% confidence interval was constructed.

Analysis of Surgery Mortality

For each group in each study, we recorded the number of deaths observed and the total number of patients in the group. If the study self-identified the deaths as “early” or “postoperative” or if it identified the deaths as within 30 days of the surgery, we termed these “early deaths.” If the deaths were self-identified as “late” or if they were identified as after 30 days, we termed these “late deaths.” If the study was unclear as to the timing of the recorded deaths, we termed these “unclear deaths.” If a study did not report data on death for a group, we recorded zero unclear deaths for that group. That is, we imputed zero for missing data, under the assumption that the authors would have reported a death if there had been one.

For each group of similar procedures (described above in our discussion of the surgery categorization), we conducted analysis for four separate combinations of death definition and type of study: late deaths for RCTs/CCTs; late deaths for CSs; early or unclear deaths for RCTs/CCTs; and early or unclear deaths for CSs. In each setting, we calculated the crude death rate. That is, we divided the total number of deaths observed by the total number of patients in the relevant study groups. This calculation treats all patients from all studies equally and does not take into account any variation across studies in mortality rates, but given the small number of observed deaths, this statistic is simple and easily interpretable.

We were also interested in examining the variation in mortality rates across studies. Thus, for any study with one or more observed deaths, we determined the probability that we would observe as many deaths or more in a sample the size of the study sample, if the population mortality rate is equal to the observed crude death rate. We identified in our results any studies that have a probability less than or equal to 0.05 and, in this way, identified studies with outlying mortality rates compared with the those of the majority of studies. We made this identification by conducting an exact binomial test.

Analysis of Surgery Comorbidity

Three clinicians extracted comorbidity data for all case series studies. Data for diabetes, hypertension, sleep apnea, and lipids were extracted. For each condition, we collected data on the number of people who had the condition at the start of the study. We then collected data on the number of those people whose condition resolved, improved, or was unchanged. A crude proportion was calculated across studies for those who resolved or improved (e.g., the number of people who resolved or improved divided by the number of people with the condition at baseline).

Analysis of Surgery Adverse Events

For selected surgery comparisons (one type of surgery versus another type of surgery) for which RCT/CCT data were available, we estimated a pooled odds ratio and its associated 95% confidence interval using exact methods as described above for the medication adverse-events meta-analysis. We also report the crude adverse-event rate for each RCT/CCT surgery group (total number of affected patients divided by total number of patients at risk). In addition, we report the crude adverse-event rate for each surgery group across all studies (RCT/CCT/CS) combined.

Rating the Body of Evidence

We rated the body of evidence on each topic in terms of quality, quantity, and consistency. The quality of the evidence is the aggregate of the quality of the individual studies and is influenced by study design, execution, and analysis of the results. In terms of study design, our minimum criterion for conclusive evidence for the efficacy of a treatment for a chronic disease is that a study have concurrent comparison groups, i.e., conclusive evidence cannot derive solely from case series or studies with historical controls. Empirical evidence shows that studies without a concurrent comparison group report estimates of effect that are inflated when compared with estimates from studies with a concurrent comparison group.⁵⁰ Numerous examples of interventions—both medical and surgical—can be cited that reported benefits in case series, only to be shown to be less effective or even ineffective when subjected to a study with a concurrent comparison group.^{51, 52} Consequently, in order to judge a treatment as having conclusive evidence of effect, we required statistically and clinically significant within-study comparisons of outcomes. In order to have confidence in making cause-and-effect conclusions between the treatment and outcome when assessing a difference in outcome between two groups receiving different treatments, we need to be confident that the groups were sufficiently similar before the treatment. Random assignment of a large number of patients is the best way to obtain similar groups, but that does not mean that conclusive evidence can come only from randomized trials. Rather, when drawing conclusions, we judge the size of the difference in outcome compared to the possibility that there were pretreatment differences between groups that might explain the outcome differences. If a large number of patients are randomly assigned to groups, the two groups are very likely to be similar at baseline, and we are more confident in drawing conclusions about cause-and-effect relationships, even if the difference in outcome is small. However, even when patients are not assigned randomly to groups, we can still draw conclusions about cause-and-effect if the differences in outcome are very large, so large that we judge it unlikely that measured or unmeasured differences in groups could account for the outcome differences.

The quantity of evidence refers to the number of studies, the sample sizes of the studies, and the sizes of the study effects. In the absence of a single very large study, then the number of studies, their sizes, and the sizes of the effects are taken into consideration when assessing the quantity of evidence.

Peer Review

A draft of this report was prepared in October 2003 and sent to the peer reviewers listed in Appendix D. The comments we received from peer review were used to prepare the final report. Service as a reviewer of this report should not in any way be construed as agreeing with or endorsing the contect of the report.

Sibutramine

Our literature search identified a high-quality meta-analysis that was “in press” at the time of our search and has since been published ⁵³. The authors of this paper agreed to let us incorporate their results into our evidence report. In their literature search, the authors did not restrict their search to English-language publications and made extensive efforts to identify unpublished and ongoing trials, including contacting representatives of the pharmaceutical industry. Inclusion criteria for the review were that the study was a RCT that assessed sibutramine (10 or 20 mg daily), enrolled adults 18 years of age or older who had a BMI of 25 or more, assessed weight loss, and had a treatment duration of at least eight weeks. The primary outcome was mean change in body weight, and data on blood pressure, heart rate, cholesterol, fasting glucose, and glycosylated hemoglobin were abstracted if reported. Studies were analyzed in three strata, based on duration of the trial: 8 to12 weeks, 16 to 24 weeks, and 44 to 54 weeks. Of 1,245 potentially relevant citations, 432 manuscripts and abstracts were reviewed in more detail, which resulted in 44 trials that were considered for inclusion in their analysis. Ten authors provided additional, unpublished data. The meta-analysis authors identified seven trials of 8 to 12 weeks duration that included a total of 546 participants; 12 trials of 16 to 24 weeks duration that included a total of 1,179 participants; and five trials of 44 to 54 weeks duration that included 2,188 participants. The mean age of enrolled patients ranged from 34 to 54 years of age. Adults with known cardiovascular disease were generally excluded from most primary studies. Dietary interventions were a co-intervention in nearly all primary studies, and exercise and behavior modification were each interventions in about one quarter of the studies. Ultimately, twenty-nine studies met all the authors' criteria for inclusion in the analysis. Of these, 23 (79 percent) had a Jadad score of 3 or greater.

Figure 2. Pooled analysis: 8- to 12-week trials of (more...)

   Figure 2. Pooled analysis: 8- to 12-week trials of sibutramine, 10–15 mg daily*

* Weighted mean difference in weight loss at 8 to 12 weeks in kilograms, sibutramine minus placebo. Arterburn DE, Crane PK, Veenstra DL. The efficacy and safety of sibutramine for weight loss: A systematic review.⁵³

CI = confidence interval.

Figure 2

Figure 3. Pooled analysis: 16- to 24-week trials of (more...)

   Figure 3. Pooled analysis: 16- to 24-week trials of sibutramine, 10–15 mg daily*

* Weighted mean difference in weight loss at 8 to 12 weeks in kilograms, sibutramine minus placebo. Subgroup A contains trials that used Last Observation Carried Forward (LOCF) and had greater than 70% follow-up, Subgroup B contains trials that analyzed completers only, and Subgroup C contains trials with follow-up rates less than 70%. See text for details. Arterburn DE, Crane PK, Veenstra DL. The efficacy and safety of sibutramine for weight loss: A systematic review.⁵³

Figure 3

Figure 4. Pooled analysis: 44- to 54-week trials of (more...)

   Figure 4. Pooled analysis: 44- to 54-week trials of sibutramine, 10–15 mg daily*

• Weighted mean difference in weight loss at 8 to 12 weeks in kilograms, sibutramine minus placebo. Arterburn DE, Crane PK, Veenstra DL. The efficacy and safety of sibutramine for weight loss: A systematic review.⁵³

CI = confidence interval.

Figure 4

In a dose-ranging study not included in the meta-analysis, 1,047 patients were given instructions on diet, physical activity, and lifestyle changes and then randomized to received placebo or sibutramine at 1 mg, 5 mg, 10 mg, 15 mg, 20 mg, or 30 mg a day. Weight loss at 24 weeks was dose-dependent, with a mean weight loss above placebo equal to 4.9 percent, 6.2 percent, 7.6 percent, and 8.2 percent for the 10 mg, 15 mg, 20 mg, and 30 mg doses, respectively. Side effects such as dry mouth, insomnia, and nausea also increased in a dose-dependent manner.⁵⁴ Of note, sibutramine was not FDA approved in the 20 mg and 30 mg doses.

Regarding other assessed outcomes, the authors did not identify any evidence that sibutramine reduces mortality or morbidity from obesity-associated diseases. Systolic and diastolic blood pressure outcomes were variable, with some studies reporting small decreases and other studies reporting small increases. Fasting blood glucose and hemoglobin A1c fell slightly in sibutramine-treated patients, but no consistent effect was observed on cholesterol or lipid outcomes. Adverse-event analysis identified no studies in which participants died. The analysis showed that in patients who took sibutramine, heart rate was consistently increased by about four beats per minute.

The conclusions of this systematic review and meta-analysis were that sibutramine with lifestyle modification was more effective than placebo with lifestyle modification in promoting weight loss in overweight and obese adults at all time points that were assessed, with an average of 4.5 kg greater weight loss at one year and a 20 to 30 percent greater likelihood of achieving a weight loss of 5 percent or more (compared to placebo). The authors also concluded that treatment with sibutramine is associated with modest increases in heart rate and blood pressure, very small improvements in glycemic control among diabetics, and (based on the longest- duration and best-quality studies) small improvements in HDL cholesterol and triglycerides. Efficacy and safety beyond two years of treatment are unknown. We calculated that, in aggregate, the RCTs of sibutramine contained sufficient numbers of patients to evaluate adverse outcomes that occurred at a rate of 8 per 10,000 or higher; the existence of outcomes at lower rates is untested.

Orlistat

Our literature search identified 28 studies of orlistat eligible for inclusion in a meta-analysis.^55–82 The average age of patients enrolled in these studies was 48 years old; 73 percent were women; and the average BMI was 36.7. In all 28 studies, diet was a co-intervention in all experimental arms; 39 percent of studies included educational, behavioral, or psychosocial co-interventions; and 18 percent of studies included exercise co-interventions. Consistent with the meta-analysis of sibutramine, we stratified the data according to treatment duration for analysis.

Table 4. Meta-analysis of weight loss in placebo-controlled (more...)

Table 4. Meta-analysis of weight loss in placebo-controlled trials of orlistat at 6 months

Trial	Total n	Mean Difference	95% CI
Broom65	137	-1.80	(-3.15, -0.45)
Deerochangawong67	252	-1.20	(- 2.75, 0.35)
DeRosa82	48	-0.90	(-1.27, -0.53)
Halpern72	280	-1.66	(-4.56, 1.24)
Hauptman56	422	-3.30	(-4.94, -1.66)
Karhunen70	72	-2.50	(-5.41, 0.41)
Muls66	290	-2.78	(-3.73, -1.83)
Micic81	114	-3.41	(-5.72, -1.10)
Naumov78	30	-4.60	(-6.58, -2.62)
Rissanen63	51	-4.10	(7.55, -0.65)
Shi Yi68	428	-3.10	(-4.37, -1.83)
Pooled random-effects estimate		-2.511	(-3.40, -1.63)

1

Chi-squared test of heterogeneity p-value < 0.001.

Figure 5. Pooled analysis: orlistat, 6 months

   Figure 5. Pooled analysis: orlistat, 6 months

Figure 5

Table 5. Meta-analysis of weight loss in placebo-controlled (more...)

Table 5. Meta-analysis of weight loss in placebo-controlled trials of Orlistat at 12 months

Trial	Total n	Mean Difference	95% CI
Bakris74	532	-2.70	(-3.79, -1.61)
Broom66	347	-3.50	(-5.07, -1.93)
Davidson55	591	-2.95	(-4.45, -1.45)
Derosa82	48	-1.00	(-1.49, -0.51)
Gotfredsen79	30	-3.10	(-8.48, 2.28)
Hanefeld75	369	-1.90	(- 2.96, -0.84)
Hauptman56	422	-3.80	(-5.37, -2.23)
Hill57	234	-1.31	(-3.00, 0.38)
Hollander58	321	-1.88	(-3.38, -0.38)
Karhunen70	72	-4.50	(-7.41, -1.59)
Kelley69	265	-2.62	(-3.38, -1.86)
Krempf71	478	-2.90	(-4.71, -1.09)
Lindgarde83	323	-1.30	(-2.51, -0.09)
Lucas73	444	-3.80	(-5.06, -2.54)
Miles76	311	-2.90	(-3.73, -2.07)
Reaven80	247	-2.14	(-3.94, -0.34)
Rissanen280	51	-5.80	(-9.25, -2.35)
Rosenfalck77	12	-4.78	(-11.88, 2.32)
Rossner60	479	-3.00	(-4.17, -1.83)
Sjostrom61	544	-4.20	(-5.26, -3.14)
Vidgren62	75	-4.20	(-7.46, -0.94)
Pooled random-effects estimate		-2.761	(-3.31, -2.20)

1

Chi-squared test of heterogeneity p-value < 0.001.

Figure 6. Pooled analysis: orlistat, 12 months

   Figure 6. Pooled analysis: orlistat, 12 months

Figure 6

Table 6. Meta-analysis of adverse events in RCTs of (more...)

Table 6. Meta-analysis of adverse events in RCTs of orlistat

		Data				Synthesis
Adverse event		Placebo		Intervention groups
	# of studies	# adverse events	sample size	# adverse events	sample size	Pooled OR	95% CI	RR	Number needed to harm (OR>1)
Diarrhea	13	892	3178	4074	4876	54.85	(44.88, 67.48)	3.40	1.48
Flatulence	11	218	2972	1020	4669	3.72	(3.16, 4.39)	3.10	6.49
Bloating/abdominal pain/ dyspepsia	8	111	1388	155	1737	1.55	(1.18, 2.06)	1.48	25.80
Headache	3	29	516	34	519	1.18	(0.68, 2.05)	NC	NC
Nausea/vomiting	3	19	311	18	306	0.95	(0.46, 1.98)	NC	NC
Gallbladder problems	5	13	712	10	1322	0.71	(0.27, 1.82)	NC	NC
Depression/mood change	3	3	153	1	154	0.33	(0.01, 4.15)	NC	NC
Back pain	1	3	114	5	114	NC	NC	NC	NC
Hepatic abnormalities	3	0	498	1	886	NC	NC	NC	NC
Rhinitis	1	6	114	7	114	NC	NC	NC	NC

OR = Odds ratio

RR = Relative risk

NC = Not Calculated

CI = Confidence interval

Note: for the diarrhea group, five trials had more adverse events in the orlistat group than actual number of people. The total number of adverse events was truncated at the total number of people for the OR calculation.

Phentermine

Our literature search identified a recent meta-analysis⁸⁵ that assessed RCTs of the use of phentermine for weight loss in obese individuals. This review identified nine studies published between 1975 and 1999. Our literature review identified no new RCTs of phentermine since this time; consequently, we rely on the existing meta-analysis for our evidence regarding efficacy of phentermine. In this review, six placebo-controlled RCTs contributed data to the pooled analysis. The duration of treatment with phentermine varied from 2 to 24 weeks. More than 80 percent of enrolled individuals were female, and more than 80 percent of participants also received lifestyle modification treatments as co-interventions. The dose of phentermine ranged from 15 to 30 mg a day. In the authors' pooled analysis, subjects treated with phentermine lost an average of 3.6 additional kg of weight compared to placebo, (95% CI, 0.6 to 6.0). In an analysis assessing the effect on maintenance of weight loss, the authors reported that patients treated with phentermine maintained a “fairly large” weight loss compared to placebo (2.43 kg) after discontinuation of the drug. The authors concluded that phentermine use, in addition to lifestyle interventions, resulted in a statistically significant, but modest, increase in weight loss. In this review, no side-effect or adverse-event data were reported. We identified no systematic reports of adverse events with phentermine. However, since phentermine is a sympathomimetic amine, side effects consistent with this class of drugs can be expected, e.g. palpitations, tachycardia, elevation of blood pressure, central nervous system effects, and gastrointestinal effects. Case reports of stroke in persons taking phentermine for weight loss have been reported,^{86, 87} but as with all case-report analyses, a causal relationship cannot be established or assumed. We calculated that in aggregate, the RCTs of phentermine contained sufficient numbers of patients to evaluate outcomes occurring at a rate of 8 per 1,000 or higher; the existence of outcomes at lower rates is untested.

Diethylpropion

Our literature search identified a recent meta-analysis⁸⁵ that assessed RCTs of the use of diethylpropion for weight loss in obese individuals. This review identified 13 studies published between 1965 and 1983. Our literature review identified no new RCTs of diethylpropion since this time; consequently, we rely on the existing meta-analysis for our evidence regarding efficacy of diethylpropion. In this review, nine placebo-controlled RCTs contributed data to the pooled analysis. The duration of treatment with diethylpropion varied from 6 to 52 weeks. More than 80 percent of enrolled individuals were female, and 100 percent of participants received lifestyle modification treatments as co-interventions. The dose of diethylpropion was 75 mg a day. In the authors' pooled analysis, subjects treated with diethylpropion lost an average of 3.0 additional kg of weight over placebo (95% CI, -1.6, 11.5). The authors concluded that diethylpropion use, in combination with lifestyle interventions, was associated with a modest increase in weight loss of borderline statistical significance. In this review, no side-effect or adverse-event data were reported. We calculated that, in aggregate, the RCTs of diethylpropion contained sufficient numbers of patients to evaluate outcomes occurring at a rate of 8 per 1,000 or higher; the existence of outcomes at lower rates is untested.

Fluoxetine

Our literature search identified nine studies of fluoxetine treatment that reported weight loss outcomes (fluoxetine is usually indicated for treatment of depression, obsessive compulsive disorder, and bulimia).^{44, 88–95} Of note is that the doses used for weight loss were higher (60 mg) than those used for depression (20 mg). The average age of patients enrolled in these studies was 48 years old; 69 percent were women; and the average BMI was 35.5. In 78 percent of the studies (7 of 9), diet was a co-intervention; 33 percent of studies included an educational, behavioral, or psychosocial co-intervention; and 12 percent of the studies included exercise as a co-intervention. Consistent with the analyses of sibutramine and orlistat, we stratified the data according to treatment duration for analysis.

Table 7. Meta-analysis of weight loss in placebo-controlled (more...)

Table 7. Meta-analysis of weight loss in placebo-controlled trials of fluoxetine at 6 months

Trial	Total n	Mean Difference	95% CI
Connolly93	24	-3.90	(-4.83, -2.97)
Goldstein89	303	-2.70	(-4.09, -1.31)
Gray95	36	-7.40	(-9.13, -5.67)
Marcus90	31	-9.10	(-13.32, -4.88)
Mendoza94	60	-4.57	(-6.58, -2.56)
Michelson44	192	-0.90	(-1.87, 0.07)
O'Kane92	18	-6.50	(-7.35, -5.65)
Pooled random-effects estimate		-4.751	(-6.72, -2.77)

1

Chi-squared test of heterogeneity p-value < 0.001.

Figure 7. Pooled analysis: fluoxetine, 6 months

   Figure 7. Pooled analysis: fluoxetine, 6 months

Figure 7

Table 8. Meta-analysis of weight loss in placebo-controlled (more...)

Table 8. Meta-analysis of weight loss in placebo-controlled trials of fluoxetine at 12 months

Trial	Total n	Mean difference	95% CI
Breum88	29	-0.70	(-8.53, 7.13)
Darga91	30	-3.60	(-4.87, -2.33)
Goldstein89	212	0.40	(-1.70, 2.50)
Marcus90	21	-14.50	(-22.62, -6.38)
Michelson44	78	-0.20	(-2.59, 2.19)
O'Kane92	16	-5.80	(-7.57, -4.03)
Pooled random-effects estimate		-3.151	(-5.82, -0.48)

1

Chi-squared test of heterogeneity p-value < 0.00.

Figure 8. Pooled analysis: fluoxetine, 12 months

   Figure 8. Pooled analysis: fluoxetine, 12 months

Figure 8

Table 9. Meta-analysis of adverse events in RCTs of (more...)

Table 9. Meta-analysis of adverse events in RCTs of fluoxetine for obesity

		Data				Synthesis
Adverse event		Placebo		Intervention groups
	# of studies	# adverse events	Sample size	# adverse events	sample size	Pooled OR	95% CI	RR	Number needed to harm (OR>1)
Nervousness/sweating/tremors	4	10	294	63	297	7.85	(3.87,17.63)	6.37	5.48
Nausea/vomiting	6	31	322	82	349	3.27	(1.94, 5.67)	2.68	6.17
Fatigue/asthenia/hypersomnia/somnolence	6	37	336	87	363	2.83	(1.82, 4.45)	2.36	6.70
Insomnia	3	14	270	31	296	2.19	(1.10, 4.58)	2.06	18.15
Diarrhea	6	26	336	46	363	1.86	(1.10, 3.23)	1.74	17.37
Urticaria/pruritis/rash	4	7	74	11	76	1.67	(0.53, 5.65)	NC	NC
Headache	5	61	314	79	340	1.35	(0.91, 2.03)	NC	NC
Rhinitis	4	72	294	77	297	1.08	(0.73, 1.60)	NC	NC
Depression/mood change	3	6	68	6	70	0.96	(0.24, 3.82)	NC	NC
Back pain	2	5	46	6	47	NC	NC	NC	NC
Bloating/abdominal pain/dyspepsia	2	4	46	5	47	NC	NC	NC	NC
Decreased libido/sexual dysfunction	2	1	46	4	47	NC	NC	NC	NC

OR = Odds ratio

RR = Relative risk

NC = Not Calculated

CI = Confidence interval

Note: for the nausea/vomiting group, one trial had more adverse events in the fluoxetine group than actual number of people. The total number of adverse events was truncated at the total number of people for the OR calculation.

Sertraline

Our literature search identified one study of sertraline.⁹⁶ This study assessed the effect of sertraline in maintaining weight loss in 53 women (out of a total of 68) who had completed a 26-week weight reduction program that combined a very low calorie diet and behavior therapy. To be eligible for inclusion in the sertraline trial, subjects had to have lost at least 10 percent of their initial weight; they were then randomly assigned to receive sertraline beginning at 50 mg per day, titrating upwards to 200 mg per day, or placebo. All patients also attended a 54-week relapse prevention program that addressed skills required to maintain weight loss. Patients had a mean age of 42 years and a mean BMI of 30, having lost approximately 23 kg in the prior six-month study. At the end of the 54-week period of evaluation, sertraline-treated patients had regained an average of 17.7 kg, while placebo-treated patients had regained an average of 11.8 kg, a difference the authors did not report as statistically significant. However, in the first 10 weeks of the study, sertraline-treated patients regained significantly less weight than did the placebo-treated patients.

More sertraline-treated patients than placebo-treated patients reported fatigue, nausea, difficulty concentrating, and problems urinating. Three sertraline-treated subjects discontinued the study because of adverse events, compared with no adverse events were associated with the placebo. Of note is that insomnia, headache, nausea, and fatigue were reported by approximately half of all sertraline-treated patients.

Bupropion

Table 10. Meta-analysis of weight loss in placebo controlled (more...)

Table 10. Meta-analysis of weight loss in placebo controlled trials of bupropion at 6–12 months

Trial	Total n	Mean Difference	95% CI
Anderson 43	217	-4.91	( -6.78, -3.05)
Croft 98	423	-1.17	( -2.25, -0.09)
Jain 99	384	-2.70	( -3.57, -1.83)
Pooled random-effects estimate		-2.771	( -4.50, -1.05)

1

Chi-squared test of heterogeneity p-value < 0.001.

Figure 9. Pooled analysis: bupropion, 6–12 (more...)

   Figure 9. Pooled analysis: bupropion, 6–12 months

Figure 9

Table 11. Meta-analysis of adverse events in RCTs of (more...)

Table 11. Meta-analysis of adverse events in RCTs of bupropion

		Data				Synthesis
Adverse event		Placebo		Bupropion
	# of trials	# adverse events	Sample size	# adverse events	sample size	Pooled OR	95% CI	RR	Number needed to harm (OR> 1)
Dry mouth	2	13	321	58	428	3.26	(1.71, 6.64)	2.99	12.43
Diarrhea	1	7	112	18	215	1.37	(0.52, 4.01)	1.34	47.19
Constipation	2	20	321	39	428	1.31	(0.72, 2.44)	1.29	56.28
Upper respiratory problems	2	104	321	171	428	1.22	(0.88, 1.69)	1.14	22.23
Headaches	2	42	321	66	428	0.99	(0.63, 1.57)	NC	NC
Central nervous system effects	2	31	321	41	428	0.98	(0.58, 1.66)	NC	NC
Upper abdominal symptoms	2	24	321	30	428	0.81	(0.44, 1.50)	NC	NC
Hypertension	1	0	112	1	215	NC	NC	NC	NC
Seizure	1	0	209	1	213	NC	NC	NC	NC

NC = Not Calculated.

Topiramate

Our literature search identified nine studies that assessed the efficacy of the drug topiramate for weight loss. One study,¹⁰⁰ which did not include a placebo group, is excluded from review. One study¹⁰¹ was dropped because it duplicated data in another included study.³¹ Two articles reported data on the same trial;^{102, 103} however, one had a larger sample size, so only it was included, leaving six studies for analysis. ^{31, 103–107} These six were judged sufficiently clinically similar to support a pooled analysis. All but one of these studies were published only as abstracts at the time of our analysis. Of note, all these studies reported their data only as percent weight loss, so the outcome for this analysis was percent weight loss. Many of the studies assessed multiple doses, the most common being 96 mg/day and 192 mg/day. We determined that the higher dose produced significantly more weight loss than the lower dose (by 1.75 percent) over the duration of the study, so we present data only on the higher dose. In these studies, the average age of subjects was 47; 68 percent were female, and the baseline weight was 102 kg. Four of the six studies had as co-interventions diet, exercise, education, and behavioral theraphies.

Table 12. Meta-analysis of weight loss in placebo controlled (more...)

Table 12. Meta-analysis of weight loss in placebo controlled trials of topiramate at 6 months

Trial	Total n	Mean Difference	95% CI
Bray 31	151	-3.70	(-5.23, -2.17)
Caterson 107	195	-8.20	(-9.55, -6.85)
Prud' homme105	46	-6.00	(-9.19, -2.81)
Rissanen 104	226	-9.10	(-10.36, -7.84)
Stenlof 103	272	-7.00	(-8.14, -5.86)
Tonstad 106	109	-4.60	(-6.40, -2.80)
Pooled random-effects estimate		-6.511	(-8.25, -4.77)

1

Chi-squared test of heterogeneity p-value < 0.001.

Figure 10. Pooled analysis: topiramate

   Figure 10. Pooled analysis: topiramate

Figure 10

Table 13. Meta-analysis of adverse events in RCTs of (more...)

Table 13. Meta-analysis of adverse events in RCTs of topiramate

		Data				Synthesis
Adverse event		Placebo		Topiramate 192 mg
	# of trials	# adverse events	sample size	# adverse events	sample size	Pooled OR	95% CI	RR	Number needed to harm (OR>1)
Parasthesia	6	101	624	524	643	20.18	(13.99, 29.67)	4.92	1.58
Taste perversion	5	11	527	100	545	11.14	(5.80, 23.57)	9.19	5.85
Central nervous system effects	6	194	624	376	643	3.97	(2.90, 5.49)	2.06	3.02
Constipation	2	9	212	31	211	3.96	(1.77, 9.77)	3.52	9.36
Dry mouth	3	13	349	37	346	3.13	(1.59, 6.55)	2.90	14.13
Upper abdominal symptoms	5	73	589	123	610	1.76	(1.27, 2.47)	1.61	13.26
Fatigue	6	160	624	204	643	1.36	(1.03, 1.80)	1.25	15.91
Upper respiratory problems	5	184	487	199	508	1.32	(0.87, 1.99)	1.18	14.90
Diarrhea	3	47	275	55	297	1.08	(0.68, 1.71)	1.07	89.42
Vertigo	1	4	137	5	135	NC	NC	NC	NC
Headaches	1	16	177	20	178	NC	NC	NC	NC
Increased hepatic enzymes	1	0	75	0	76	NC	NC	NC	NC

NC = Not Calculated

Note: For the central nervous system effects and upper respiratory problems groups, one trial in each had more adverse events in the topiramate group than actual number of people.

For the paresthesia group, two trials had more adverse events in the topiramate group than actual number of people.

The total number of adverse events was truncated at the total number of people for the OR calculation.

Zonisamide

Our literature search identified one eligible study that assessed the efficacy of the drug zonisamide for weight loss.¹⁰⁸ This study was a double-blind, RCT that enrolled 60 patients with a mean age of 37 years. Approximately 90 percent were women, and their mean BMI was 36. Patients were randomized to begin receiving placebo or zonisamide at 100 mg per day; daily doses were increased to a maximum of 600 mg per day, based on response. Patients in both groups were also instructed to follow an individualized diet that was devised to reduce their daily energy intakes to 500 kilocalories per day below maintenance level. Increased physical activity was also encouraged for participants in both groups. Patients were followed for 16 weeks in the double-blind portion of the study, with an additional 16-week single-blind extension available. Of the 60 patients, 51 (85 percent) completed the 16-week phase. Using a “last observation carried forward” analysis for dropouts, the researchers found that patients in the zonisamide group lost an average of 6.0 percent of baseline body weight, compared to 1.0 percent for placebo patients (p <.001). In the extension phase of the study, 37 patients (20 in the zonisamide group, 17 in the placebo group) continued, and 36 completed the study at week 32. Ten of the 19 zonisamide patients who completed the study had lost at least 10 percent of initial body weight at week 32 (compared with none of the placebo patients) (p <.001). Heart rate decreased by an average of approximately two beats per minute in the overall sample, and there were no differences between groups. The authors reported that systolic and diastolic blood pressure readings did not change over time. The authors also reported that the total numbers of adverse effects over the study period were 2.1 and 1.6, for those assigned to zonisamide and placebo, respectively, a difference that did not reach statistical significance. Among individual adverse effects, the only statistically significant differences observed were in fatigue: ten patients in the zonisamide group reported fatigue, compared with one in the placebo group (p <.006 by the Fisher exact test). The mean serum creatinine increased from 0.79 milligrams per deciliter to 0.92 milligrams per deciliter for zonisamide-treated patients and from 0.76 milligrams to 0.79 milligrams per deciliter in placebo-treated patients (p <.001).

Summary of Medication Studies

Table 14. Summary of findings regarding medications (more...)

Table 14. Summary of findings regarding medications for weight loss

Medication	Source of data	Weight loss assessed at:	Mean weight loss in treated patients compared to placebo (95% CI)
Sibutramine	Existing meta-analysis of 24 RCTs	52 weeks	-4.45 kg (-5.29, -3.62)
Orlistat	Our meta-analysis of 23 RCTs	52 weeks	-2.76 kg (-3.31, -2.20)
Fluoxetine	Our meta-analysis of 9 RCTs	52 weeks	-3.15 kg (-5.82, -0.48)
Phentermine	Existing meta-analysis of 9 RCTs	2 to 24 weeks	-3.6 kg (-6.0, -0.6)
Diethylpropion	Existing meta-analysis of 13 RCTs	6 to 52 weeks	-3.0 kg (-11.5, 1.6)
Bupropion	Our meta-analysis of 3 RCTs	24 to 52 weeks	-2.77 kg (-4.5, -1.0)
Zonisamide	1 RCT	16 weeks	Additional 5% of pre-treatment weight loss
Topiramate	Our meta-analysis of 6 RCTS	24 weeks	Additional 6.5% of pre-treatment weight loss (95% CI 4.8%, 8.2%)

Benefits

Weight loss and maintenance. We identified two RCTs that compared bariatric surgery to a nonsurgically treated control group. The first is the RCT that compared horizontal gastroplasty and diet to diet alone^{120, 134} and was analyzed in the Cochrane Review. This RCT generated two articles that reported net weight loss at 6 months¹²¹ and 24 months¹²⁰ followup. At 6 months, weight loss was not different between the two groups, but at 24 months followup, the net weight change from baseline greatly favored surgical therapy (net weight change of 30.5 kg versus 8.0 kg for surgical and nonsurgical therapy, respectively), although only about half the original patients contributed data at 24 months. We also identified another randomized trial that compared jejunoileal bypass to “medical treatment” (not otherwise specified) in 186 patients. Again at 24 months followup, the mean difference in weight loss greatly favored surgical therapy (mean difference = 37 kg). Of note, these studies were conducted more than 20 years ago, and the surgical procedures assessed are not considered relevant to modern bariatric surgery, in that improvements in procedures and technique have been associated with significantly greater long-term weight loss, as compared to horizontal gastroplasty, and fewer major complications as compared to the jejunoileal bypass.

In addition to the two RCTs, we identified numerous reports from an observational study, the Swedish Obese Subjects (SOS) study.^109–116 In the intervention portion of this study ( a cross-sectional registry portion was also included), obese adults (BMI ≥ 34 for men and ≥ 38 for women) were assessed in two groups: those who voluntarily underwent bariatric surgery (most of whom were treated with vertical banded gastroplasty) and a group of matched controls treated medically. Matching was done on 18 variables, including gender, age, height, and weight. The average age of enrolled subjects was 47, about two-thirds were women, and average BMI at baseline was about 41. At eight years of followup, among 251 surgically treated patients, the average weight loss was 20 kg (or 16 percent of body weight), whereas among 232 medically treated patients, the average weight did not change. Patients treated with RYGB lost more weight than those treated with vertical banded gastroplasty or banding procedures.¹¹⁰ Based on this latter finding and on RCT evidence summarized in the text to follow, had all patients in the SOS trial been treated with RYGB the difference in weight loss between surgical and medical therapy would likely have been even greater—probably on the order of 10 kg more. Even though the SOS study was not randomized, patients were well matched in both groups, and the magnitude of the observed differences is so large that it is very unlikely that unmeasured variables could account for these differences. Thus, we believe this study provides conclusive evidence of the superiority of surgical treatment for the patients enrolled in the study (middle-aged adults with a BMI of about 41). Also contributing to the strength of this study is the extended duration of followup, documenting sustained weight loss and improved health up to eight years after treatment. The SOS study recently reported sustained improvements in weight loss at 10 years followup, compared to controls (J. Torgerson, presentation at the 2003 annual meeting of the American Society for Bariatric Surgery).

Comorbidities. As mentioned, bariatric surgery is recommended to help control the morbidities associated with excess weight. A series of reports from the SOS study support the superiority of obesity surgery compared to medical therapy in ameliorating or preventing the comorbidities of obesity. At 24 months after surgery, among 845 surgically treated patients and 845 matched controls (two-thirds women, average age of 48, average BMI about 41), the incidence of hypertension, diabetes, and lipid abnormalities was markedly lower in the surgically treated patients (adjusted odds ratios of 0.02 to 0.38, depending on condition).¹¹⁶ At eight years of followup, the effect of surgery on the reduction in diabetes risk was still dramatic (odds ratio = 0.16), while the effect on reduction in risk for hypertension did not persist (odds ratio = 1.01).¹¹⁰ However, significant decreases in both systolic (8.3 mm Hg) and diastolic (6.7 mm Hg) blood pressure persisted in the small (6 percent) subset of patients who underwent a gastric bypass and lost significantly more weight than the 94 percent of patients who underwent a vertical banded gastroplasty or gastric banding.⁷² Additional reports from the SOS study support a substantial benefit of surgery in reducing sleep apnea,¹¹¹ symptoms of dyspnea and chest pain,¹¹¹ and improving quality of life.¹¹⁴ The latter study assessed health-related quality of life in four domains: health perception, mental well being/mood disorders, psychosocial functioning, and self-assessment of eating behavior. The researchers matched 487 surgically-treated patients to an equivalent number of control patients; at two years followup, data were available for 98 percent of surgical patients and 82 percent of control patients. Improvements in all domains in surgical patients compared to control patients were greatest at 6 months after surgery and diminished slightly at 24 months. Differences between groups were substantial in nearly all domains: in general, one-half to two-thirds of an effect size. The differences were related to the degree of weight loss, meaning that patients who lost a greater amount of weight had greater improvements in quality of life. The SOS study is the only one we identified that compares comorbidities between surgically treated patients and a concurrent control group receiving nonsurgical treatment.

We assessed reports of surgery case series for data on the control of four comorbidities: diabetes, hypertension, sleep apnea, and hyperlipidemia. Of the 114 case series publications, 21 papers reported quantitative information on the control of diabetes. The proportion of patients with preoperative diabetes who showed improvement or resolution of their diabetes after surgery ranged from 69 percent to 100 percent, with a median reported value of 100 percent. For control of hypertension, 18 papers reported results that ranged from 25 percent to 100 percent of patients showing improvement or resolution of hypertension following surgery and a median reported improvement of 89 percent. Fourteen studies reported results for sleep apnea: the range of improvement was 95 percent to 100 percent of patients, with a median of 100 percent of patients reporting improvement or resolution of sleep apnea. Ten studies reported on hyperlipidemia following surgery, with 60 percent to 100 percent of patients reporting improvement or resolution of hyperlipidemia following surgery with a median of 88 percent. These reported improvements in comorbidities are substantial and suggest that bariatric surgery is helping to relieve the burden of these comorbidities in severely obese individuals. However, a cause-and-effect relationship cannot be conclusively proven from case series data alone. Still, these results are consistent with the statistically significant improvement reported by the SOS study for diabetes, hypertension (in the RYGB subset), and sleep apnea, although the magnitude of benefit reported in SOS was smaller than that reported in the case series.

Although not assessed in this report, improvements in cardiac dysfunction,^{152, 227–231} gastroesophageal reflux,^232–239 pseudotumor cerebri,^{240, 241} polycystic ovary syndrome,²⁴² complications of pregnancy,^243–247 stress urinary incontinence,²⁴⁸ degenerative joint disease,^249–252 nonalcoholic steatohepatitis,²⁵³ severe venous stasis disease,^254–257 and overall quality of life^{137, 180, 216, 258–267} have been reported in some case series of obesity surgery. As mentioned above, a cause-and-effect relationship cannot be conclusively proven from case series data alone.

Table 16. Weight loss following bariatric surgery by (more...)

Table 16. Weight loss following bariatric surgery by procedure (at 12 months followup), pooled results by study type

Controlled trials (includes controlled trials comparing one procedure to another)			All studies (includes all controlled trials and case series)
Procedure	Weight loss (kg) at 12 months	# trials # patients	Procedure	Weight Loss (kg) at 12 months (95% CI)	# Studies/Patients
RYGB (all) versus VBG (all)	δ 42.43 kg	2 trials) n=114	RYGB (all) versus VBG (all)	Δ 43.46 kg (41.24, 43.46)	l=32 / n=2937
	δ 34.45 kg	n=117		Δ 32.16 kg (29.92, 34.41)	l=21 / n=2080
	Mean difference = 7.97 kg CI (2.99, 12.96)
RYGB (all) versus Adj Band (all)	Not reported	(0 trials)	RYGB (all) versus Adj Band (all)	Δ 43.46 kg (41.24, 45.68)	l=32 / n=2937
				Δ 30.19 kg (27.95, 32.42)	l=27 / n=5562
RYGB (open) versus RYGB (lap)	δ 34.35 kg	(1 trial) n=21	RYGB (open) versus RYGB (lap)	Δ 43.89 kg (41.09, 46.69)	l=25 / n=2074
	δ 37.00 kg	n=30		Δ 42.17 kg (38.95, 45.38)	l=10 / n=863
	Mean difference = -2.64 kg CI (-11.28, 6.00)
VBG (all) versus Adj Band (all)	δ 38.58 kg	(2 trials) n=71	VBG (all) versus Adj Band (all)	Δ 32.16 kg (29.92, 34.41)	l=21 / n=2080
	δ 24.20 kg	n=76		Δ 30.19 kg (27.95, 32.42)	l=27 / n=5562
	Mean difference = 14.41 kg CI (9.39, 19.42)
RYGB (all) versus BPD (all)	Not reported	(0 Trials)	RYGB (all) versus BPD (all)	Δ 43.46 kg (41.24, 45.68)	l=32 / n=2937
				Δ 51.93 kg (45.10, 58.75)	l=3 / n=735

Table 17. Weight loss following bariatric surgery by (more...)

Table 17. Weight loss following bariatric surgery by procedure (at ≥ 36 months followup), pooled results by study type

Controlled trials (includes controlled trials comparing one procedure to another)			All studies (includes all controlled trials and case series)
Procedure	Weight loss (kg) at ≥ 36 months followup (95% CI)	# Trials # Patients	Procedure	Weight loss (kg) at ≥ 36 months followup (95% CI)	# Studies/patients
RYGB (all) versus VBG (all)	Δ 39.73 kg	(2 trials) n=103	RYGB (all) versus VBG(all)	Δ 41.46 kg (37.36, 45.56)	l=21 / n=1281
	Δ 30.65 kg	n=96		Δ 32.03 kg (27.67, 36.38)	l=18 / n=1877
	Mean difference = 9.29 kg CI (1.61, 16.96)
RYGB (all) versus Adj Band (all)	Not reported	(0 trials)	RYGB (all) versus Adj Band (all)	Δ 41.46 kg (37.36, 45.56)	l=21 / n=1281
				Δ 34.77 kg (29.47, 40.07)	l=17 / n=3076
RYGB (open) versus RYGB (lap)	Not reported	(0 trials)	RYGB (open) versus RYGB (lap)	Δ 41.58 kg (37.38, 45.78)	l=20 / n=1266
				Δ 38.32 kg (28.04, 48.60)	l=1 / n=15
VBG (all) versus Adj Band (all)	Δ 35.51 kg	(2 trials) n=64	VBG (all) versus Adj Band (all)	Δ 32.03 kg (27.67, 36.38)	l=18 / n=1877
	Δ 32.97 kg	n=60		Δ 34.77 kg (29.47, 40.07)	l=17 / n=3076
	Mean difference = 2.79 kg CI (-16.63, 22.21)
RYGB (all) versus BPD (all)	Not reported	(0 trials)	RYGB (all) versus BPD (all)	Δ 41.46 kg (37.36, 45.56)	l=21 / n=1281
				Δ 53.10 kg (47.36, 58.84)	l=1 / n=50

Five RCTs were identified that compared surgical procedures and reported data sufficient for pooling; that is, the studies compared similar surgical procedures and reported weight loss data in sufficient detail. In two studies comparing RYGB procedures to VBG,^{128, 129} including 231 patients in total, pooled weight loss outcomes for both procedures were substantial (at least 30 kg at 36 months for both) and favored RYGB at both 12 and 36 months (8–9 kg more weight loss from RYGB). These results are supported by the pooled results from all studies combined (both RCTs and case series), which report data on approximately 2,000 patients for each procedure. These combined data show that RYGB patients reported about 10 kg more weight loss than patients treated with VBG, at both 12 and 36 months.

Several additional randomized trials compared RYGB and other gastric bypass procedures with VBG or other gastric partitioning procedures,^{125, 130, 131, 140–142} but the results could not be included in our pooled analysis because either they did not report their results in terms of kilograms of weight lost or they did not report the results in sufficient statistical detail. Nevertheless, the results of all these studies support the conclusion that gastric bypass produces superior weight loss to gastroplasty procedures

In two RCTs, the weight lost using VGB, compared to laparoscopic adjustable gastric banding, was 14 kg more at 12 months followup but only about 3 kg more at 36 months followup. No difference in net weight loss was seen in the pooled results from all studies combined.

Finally, one RCT compared open RYGB with that performed laparoscopically.¹³⁹ Again, the weight loss for both approaches was substantial, but no significant differences between the two were found (greater than 30 kg for both at 12 months); a result that was supported by the “all studies” pooled analysis at both 12 months and out to 36 months. Because the final anatomic reconfiguration is the same for laparoscopic and open RYGB, weight loss and comorbidity outcomes should be identical. However, these procedures involve very different technical approaches that result in different types and rates of complications.

Summary of Benefit Data

The data we identified support that surgical treatment results in greater weight loss than does medical treatment in obese individuals (BMI ≥ 40), resulting in 20 to 30 kg of weight loss, maintained up to eight years, that is accompanied by significant improvements in several comorbidities. For patients with a BMI between 35 and 40, the data strongly support the superiority of surgical therapy but cannot be considered conclusive in the absence of a study with a concurrent comparison group. Similarly, the evidence supports but does not prove an effect of surgical treatment for obesity on improvement in a large number of weight-related comorbidities for this population.

Further supporting the superiority of surgical therapy in patients with a BMI of 40 or greater is the observation that the weight loss reported in surgical studies is an order of magnitude greater than weight loss reported in pharmaceutical or diet studies of obesity (weight losses of 20–40 kg at one or two years in surgical studies versus 2 to5 kg in pharmaceutical studies), although direct comparisons cannot be made across studies because the patient populations are clearly different: The surgical studies enrolled only patients who are severely obese, whereas the average BMI in the medical weight loss studies was about 33. Additionally, many surgical studies report sustained weight loss (i.e. at 24 months or longer), whereas studies of medical weight loss therapies that report data beyond 12 months are rare, and those that do tend to report regain of most initial weight loss.

Both RCT data and observational data demonstrate clearly that RYGB results in greater weight loss than vertical banded gastroplasty. All three procedures for which we found data—RYGB; VBG; and laparoscopic adjustable band procedures—report substantial long-term weight loss.

Risks

We divided the risks of surgery or adverse events associated with it, into mortality and morbidity and further divided morbidity risks into four primary and six secondary categories, based on research findings. Our analyses were stratified by study design: Randomized trials have strong internal validity but are frequently of small sample size and limited generalizability; case series frequently contain data on many more patients, but comparisons of outcomes across procedures in different publications may not be warranted. Randomized and controlled clinical trials (CCTs) were considered to have sufficient internal validity for statistical comparisons of outcomes between groups within individual studies to be made, and—where the number of studies comparing the same surgical procedures was sufficient (two or more)—we pooled the results. Data from case series were used (along with data from RCTs/CCTs) to calculate the simple proportion of outcomes by procedure, but no statistical comparisons were made between proportions across different procedures, as we did not judge these comparisons to have sufficient internal validity to justify such comparisons: Patient selection and other factors may vary greatly across studies. Nevertheless, we present these findings alongside our findings from RCTs and CCTs, to allow the reader to compare and contrast the findings from the two study designs.

Table 18. Mortality analysis, surgical procedures

Table 18. Mortality analysis, surgical procedures

	Early or time unspecified deaths				Late deaths
Procedure	Controlled trials		Case series		Controlled trials		Case series
RYGB	1.0% (0.5, 1.9)	l=15/ n=907	0.3% (0.2, 0.4)	l=50/ n=11290	1.1% (0.4, 2.5)	l=9/ n=524	0.6% (0.4, 0.8)	l=24/ n=5411
BPD	Not Reported	l=0/ n=0	0.9% (0.5, 1.3)	l=7/ n=2808	Not Reported	l=0/ n=0	0.3% (0.01, 0.6)	l=4/ n=2362
Band	0.4% (0.01, 2.1)	l=6/ n=268	0.02% (0, 0.78)	l=35/ n=9222	Not Reported	l=0/ n=0	0.1% (0.02, 0.2)	l=11/ n=3975
VBG	0.2% (0, 1.4)	l=11/ n=401	0.3% (0.1, 0.5)	l=33/ n=4091	0.0% (0, 16.8)*	l=1/ n=20	0.6% (0.4, 1.0)	l=20/ n=2638

* one-sided, 97.5% confidence interval

Early = ≤ 30 days from procedure, or designated “early” in the original report.

Late = > 30 days from procedure, or designated “late” in the original report

These data on mortality came from selected patient populations, that is, a specific clinic or surgeon performing the procedures on patients enrolled in a research study. The first assessment of 30-day mortality in unselected patients was reported by Dr. David Flum at the 2003 Clinical Congress of the American College of Surgeons.²⁶⁸ Among more than 62,000 procedures performed in the state of Washington between 1987 and 2001, the 30-day mortality as assessed using administrative data was 1.9 percent. Furthermore, a strong association was observed between a surgeon's experience and mortality: Surgeons who had performed 20 or fewer procedures during the period of the study had an almost 5 percent rate of 30-day mortality.

Table 19. Postoperative adverse events by bariatric (more...)

Table 19. Postoperative adverse events by bariatric procedure (RYGB versus VBG, Band, or BPD), results pooled by study type

	Controlled trials (includes controlled trials comparing one procedure to another)		All studies (includes all controlled trials and case series)
Category of adverse event	Adverse event (%) by procedure	(#trials) # patients	Adverse event (%) by procedure	# Studies/patients
1. Gastrointestinal symptoms (including reflux, vomiting, dysphagia, dumping syndrome, etc.)	RYGB = 18.3% versus VBG = 15.2%	(3 trials) n=169	RYGB = 16.9% versus VBG = 17.5%	l=34 / n=7374
	OR=1.29 (0.70, 2.42)	n=178		l=21 / n=1692
	RYGB versus Band	(0 trials)	RYGB = 16.9% versus Band = 7.0%	l=34 / n=7374
	Not reported			l=17 / n=3400
	RYGB versus BPD	(0 trials)	RYGB = 16.9% versus BPD = 37.7%	l=34 / n=7374
	Not reported			l=1 / n=305
1a. Reflux	RYGB versus VBG	(0 trials)	RYGB = 10.9% versus VBG = 2.2%	l=3 / n=727
	Not reported			l=7 / n=823
	RYGB versus Band	(0 trials)	RYGB = 10.9% versus Band = 4.7%	l=3 / n=727
	Not reported			l=4 / n=485
	RYGB versus BPD	(0 trials)	RYGB = 10.9% versus BPD = NR	l=3 / n=727
	Not reported			l=0 / n=0
1. Gastrointestinal symptoms (cont) 1b. Vomiting	RYGB versus VBG	(0 trials)	RYGB = 15.7% versus VBG = 18.4%	l=8 / n=1324
	Not reported			l=10 / n=1177
	RYGB versus Band	(0 trials)	RYGB = 15.7% versus Band = 2.5%	l=8 / n=1324
	Not reported			l=4 / n=562
	RYGB versus BPD	(0 trials)	RYGB = 15.7% versus BPD = 5.9%	l=8 / n=1324
	Not reported			l=1 / n=305
2. Nutritional and electrolyte abnormalities (including mineral, vitamin, protein deficiencies, etc.)	RYGB versus VBG	(0 trials)	RYGB = 16.9% versus VBG = 2.5%	l=10 / n=2088
	Not reported			l=4 / n=397
	RYGB versus Band	(0 trials)	RYGB = 16.9% versus Band = Not Reported	l=10 / n=2088
	Not reported			l=0 / n=0
	RYGB versus BPD	(0 trials)	RYGB = 16.9% versus BPD = Not Reported	l=9 / n=2088
	Not reported			l= 0 / n=0
3. Surgical, preventable and not preventable issues (including anastomotic or stoma-related, bleeding, reoperations, wound, etc.)	RYGB = 20.3% versus VBG = 15.1%	(5 trials) n=241	RYGB = 18.7% versus VBG = 23.7%	l=49 / n=10088
	OR=1.48 (0.88, 2.49)	n=252		l=34 / n=3247
	RYGB versus Band	(0 trials)	RYGB = 18.7% versus Band = 13.2%	l=49 / n=10088
	Not reported			l=34 / n=8846
	RYGB versus BPD	(0 trials)	RYGB = 18.7% versus BPD = 5.9%	l=49 / n=10088
	Not reported			l=5 / n=2663
3a. Anastomotic, gastric pouch or duodenal leak	RYGB = 1.4% versus VBG = 2.8%	(2 trials) n=70	RYGB = 2.2% versus VBG = 1.0%	l=30 / n=5645
	OR=0.49 (0.01, 9.74)	n=72		l=14 / n=1456
	RYGB versus Band	(0 trials)	RYGB = 2.2% versus Band = 3.3%	l=30 / n=5645
	Not reported			l=2 / n=180
	RYGB versus BPD	(0 trials)	RYGB = 2.2% versus BPD = 1.8%	l= 30/ n=5645
	Not reported			l=4 / n=2358
3. Surgical, preventable and not preventable issues (cont.) 3b. Anastomotic or stomal stenosis	RYGB = 6.9 versus VBG = 14.9	(2 trials) n=72	RYGB = 4.6% versus VBG = 6.0%	l=27 / n=6078
	OR=0.51 (0.13, 1.72)	n=74		l=16 / n=1696
	RYGB versus Band	(0 trials)	RYGB = 4.6% versus Band = Not reported	l=27 / n=6078
	Not reported			l=0 / n=0
	RYGB versus BPD	(0 trials)	RYGB = 4.6% versus BPD = Not reported	l=27/ n=6078
	Not reported			l=0 / n=0
3c. Bleeding	RYGB = 1.0% versus VBG = 0.0%	(1 trial) n= 99	RYGB = 2.0% versus VBG = 0.7%	l=19 / n=5026
	OR = Not estimable	n= 106		l=6 / n=1027
	RYGB versus Band	(0 trials)	RYGB = 2.0% versus Band = 0.3%	l=19 / n=5026
	Not reported			l=6 / n=2844
	RYGB versus BPD	(0 trials)	RYGB = 2.0% versus BPD = 0.2%	l=19 / n=5026
	Not reported			l=2 / n=1617
3d. Reoperations (including those related to anastomosis, band, bleeding, revisions, etc.)	RYGB = 0% versus VBG = 3.7%	(1 trial) n= 52	RYGB = 1.6% versus VBG = 11.3%	l=9 / n=4356
	OR=0 (0, 5.73)	n= 54		l=7 / n=520
	RYGB versus Band	(0 trials)	RYGB = 1.6% versus Band = 7.7%	l=9 / n=4536
	Not reported			l=11 / n=2140
	RYGB versus BPD	(0 trials)	RYGB = 1.6% versus BPD = 4.2%	l=9 / n=4536
	Not reported			l=2 / n=1101
4. Medical (cardiac, stroke, or severe HTN)	RYGB versus VBG	(0 trials)	RYGB = 4.8% versus VBG = 4.7%	l=5 / n=2161
	Not reported			l=2 / n=473
	RYGB versus Band	(0 trials)	RYGB = 4.8% versus Band = 0.7%	l=5 / n=2161
	Not reported			l=1 / n=150
	RYGB versus BPD	(0 trials)	RYGB = 4.8% versus BPD = Not Reported	l=5 / n=2161
	Not reported			l=0/ n=0

# trials considered for analysis by procedure

• RYGB - 70 trials

• VBG - 48 trials

• BAND - 41 trials

• BPD - 7 trials

First, we note that most of the RCT/CCT data cells include at most one study. Only four trials were identified, and three of these trials compared RYGB to VBG. In most cases, only a few hundred patients have been studied in each comparison. Second, none of the comparisons of complications between these surgical procedures show statistically significant differences, and the 95% confidence intervals are very wide, that is, we can neither conclude nor exclude that clinically important differences exist. For example, the proportion of patients with reported anastomotic or stomal stenosis was about 5 percent higher (absolute percentage) in the VBG group than in the RYGB group, but these differences are not significant. Third, the absolute rates of some complications are substantial, although many may be minor in their degree of severity. For example, the proportion of subjects receiving VBG who have gastrointestinal complications is 15.2 percent in the RCT/CCT data and 17.8 percent in the case series data; the proportion of subjects receiving RYGB who experience nutritional deficiencies is 26.8 percent in the case series data (many of these nutritional deficiencies were mild); and the proportion of subjects receiving a banding procedure who require reoperation is 7.3 percent in the case series data. Fourth, some differences between procedures in the proportions of subjects with different complications or adverse events are compatible with the anatomic changes caused by the procedure. Thus, for example, gastrointestinal symptoms are reported by almost 18 percent of patients treated with VBG but reported by fewer than half that number of patients treated with banding procedures, although this difference was not statistically significant. At a minimum, these data indicate that the proportion of patients with adverse events or complications may be on the order of 10 percent to 20 percent (although the majority of these may be mild and respond to conservative treatment) and that the occurrence of these complications may differ among procedures in ways that are clinically important.

Table 20. Postoperative adverse events following bariatric (more...)

Table 20. Postoperative adverse events following bariatric procedures differs by operative approach (open versus laparoscopic), pooled results by study type

	Controlled trials (includes controlled trials comparing one procedure to another)		All studies (includes all controlled trials and case series)
Category of adverse event	Adverse event (%) by operative approach	(#trials) # patients	Adverse event (%) by operative approach	# Studies/patients
1. Respiratory (including pneumonia, atelectasis, respiratory insufficiency, etc.)	Open = 3.0%	(2 trials) N=101	Open = 2.4%	l=26 / n=5317
	Lap = 1.9%	N=104	Lap = 1.5%	l=18 / n=5437
	OR=1.54 (0.17, 19.42)
1a. Pneumonia	Not reported	(0 trials)	Open = 0.9%	l=11 / n=3555
			Lap = 0.8%	l=7 / n=2064
2. Medical (cardiac, stroke, severe HTN)	Not reported	(0 trials)	Open = 6.0%	l=5 / n=1360
			Lap = 3.1%	l=5/ n=1463
3. Surgical, preventable and not preventable issues (including wound, hernia, splenic injury, reoperations, anastomotic, etc.)	Open = 31.1%	(3 trials) N=122	Open = 22.1%	l=68 / n=11094
	Lap = 26.1%	N=134	Lap = 12.4%	l=47 / n=13752
	OR=1.32 (0.72, 2.43)
3a. Wound, all	Open = 13.1%	(3 trials) N=122	Open = 11.4%	l=47 / n=8333
	Lap = 0.0%	N=134	Lap = 2.3%	l=27 / n=8320
	OR=Not estimable
I. Wound infection, major	Open = 3.0%	(2 trials) N=101	Open = 4.0%	l=8 / n=907
	Lap = 0.0%	N=104	Lap = 1.6%	l=17 / n=4694
	OR=Not estimable
II. Wound infection, minor	Open = 14.3%	(1 trial) n=21	Open = 10.8%	l=9 / n=739
	Lap = 0.0%	n=30	Lap = 3.3%	l=5 / n=1555
	OR=Not estimable
III. Incisional hernia	Open = 8.2%	(3 trials) n=122	Open = 11.9%	l=29 / n=3686
	Lap = 0.0%	n=134	Lap = 0.4%	l=7 / n=1075
	OR=Note
3b. Internal hernia	Open = 0.0%	(1 trial) n=76	Open = 1.4%	l=4 / n=287
	Lap = 1.3%	n=79	Lap = 2.9%	l=6 / n=3422
	OR=0.00 (0.00, 40.40)
3c. Splenic injury (including splenectomy or repair)	Not reported	(0 trials)	Open = 1.2%	l=18 / n=5065
			Lap = 0.2%	l=4 / n=1541
3d. Reoperations	Open = 0.0%	(1 trial) n=25	Open = 8.1%	l=18 / n=3465
	Lap = 4.0%	n=25	Lap = 2.7%	l=9 / n=5018
	OR=0.00 (0.00, 38.94)
4. DVT and/or PE	Open = 1.0%	(2 trials) n= 97	Open = 1.3%	l=23 / n=4704
	Lap = 0.9%	n= 109	Lap = 0.6%	l=16 / n=3995
	OR=1.22 (0.02, 96.69)

# trials considered for analysis by operative approach:

• Open - 98 trials

• Lap - 53 trials.

Efficacy and Safety of Weight Loss Medication

We identified three controlled trials of medication that reported data specific to adolescents. One study (described in two reports) assessed mazindol, which was not an included drug for this review.^{269, 270} A second study assessed the use of a caffeine/ephedrine mixture, which was also not an included drug for this review.²⁷¹ Elimination of that study left one study for our review.²⁷² This study, which examined the efficacy of sibutramine, was conducted in two six-month phases. Phase I was a double-blind RCT; in Phase II, all participants received the drug. The study enrolled boys and girls ages 13 to 17 years who had a BMI of 32 to 44. Subjects were randomized to receive sibutramine (beginning at 5 mg per day and increasing to 15 mg per day) or matched placebo. All participants received the same comprehensive family-based behavioral weight loss program, which included regular group sessions led by dieticians and psychologists or psychiatrists. Participants in both groups were instructed to consume a 1200- to 1500 Kcal-per-day diet of conventional foods with approximately 30 percent of calories from fat, 15 percent from protein, and the remainder from carbohydrate. Exercise of approximately 120 minutes per week or more was also prescribed. Overall, 82 subjects were enrolled, with a mean age of 14 and a mean BMI of 38. Approximately two-thirds of subjects were female. At 6 months, subjects treated with sibutramine lost a mean of 7.8 kg, which was equal to an 8.5 percent reduction in initial BMI, whereas placebo-treated patients had a significantly smaller, 3.2 kg, weight loss, which was equal to a 5.4 percent reduction in BMI. More than twice as many sibutramine-treated patients achieved a 5 percent to 10 percent reduction in initial BMI than did placebo-treated patients. During Phase II, the patients who had switched from placebo treatment to sibutramine lost an additional 1.3 kg of weight, while those who continued on sibutramine gained 0.8 kg.

Sibutramine-treated subjects had an average increase in heart rate of five to six beats per minute compared to placebo-treated subjects, whereas blood pressure changed minimally between groups. With respect to other adverse events, three of the patients treated with sibutramine experienced a “marked and sustained” increase in blood pressure of greater than or equal to 10mm Hg, which required discontinuation of the medication; no such events were reported in the placebo group. No other differences in adverse events were reported between groups.

Efficacy of Bariatric Surgery

Our literature search identified twelve papers that reported weight loss after bariatric surgery in adolescents. Three of these papers^196–198 reported results following jejunoileal bypass and will not be considered here. Two papers^{273, 274} reported case series results from the same institution; only the latter paper is included to avoid potentially double counting patients.

The first study was a case series of ten adolescents who underwent RYGB surgery.²⁷⁵ These patients ranged in age from 15 to 17 years, 7 of the 10 were female, and their BMIs ranged from 41.4 to 70.5. Most patients had comorbidities, including sleep apnea, hypertension, and vertebral compression fractures. The authors reported no early postoperative complications. Weight loss in excess of 30 kg was observed in 9 of the 10 patients at postsurgical followup times of 8 months to 156 months. The “most serious complication” reported was protein calorie malnutrition and micronutrient deficiency, which developed in one patient approximately one year after the bypass. This patient's recovery was “uneventful” after total parental nutrition was instituted. Two other patients had symptomatic cholelithiasis that required cholecystectomy. A fourth patient experienced small bowel obstruction ten years after gastric bypass surgery, and a fifth patient required repair of an incisional hernia. Iron deficiency anemia occurred in five of the seven girls, and transient folic acid deficiency occurred in three patients.

The second study reviewed the 20-year experience of one bariatric surgery center with adolescents ages 13 to 17.²⁷⁶ Thirty-three subjects received surgery between 1981 and 2001. Of these 33, 19 were female and 14 were male. The mean preoperative BMI was 52, and the mean age was 16 years. Preoperative comorbid illnesses included one case of Type II diabetes, 10 cases of hypertension, 6 cases of sleep apnea, 5 cases of gastroesophageal reflux, and 10 cases of degenerative joint disease. One patient underwent horizontal gastroplasty, two had vertical banded gastroplasties, 17 had standard gastric bypass procedures, 10 had long limb gastric bypass, and three had distal gastric bypass. The authors reported no operative deaths or anastomotic leaks. They reported one case of pulmonary embolism, one major wound infection, four minor wound infections, three stomal stenoses treated with laparoscopic dilation, and four marginal ulcers treated medically. The authors report initial weight loss in all patients, but five patients had regained all or most of their lost weight at five or ten years after surgery. For the other 28 patients, the authors report an average of 77 percent of excess weight lost and a BMI of 29 at five postoperative years, with slight increases over longer durations (to a BMI of 30 and 31 at ten years and 14 years, respectively). Two late deaths were judged unrelated to surgery. Comorbid conditions were resolved in all but two patients with hypertension, two patients with gastroesophageal reflux disease, and 7 patients with joint pain.

The third study²⁷⁷ reported the results of one surgeon's experience performing the procedures on 22 severely obese children between 1983 and 1995 (9 males and 13 females; 3 patients were under the age of 12). During the first 5 years of the study period, VBG was performed; then the procedure was changed to RYGB and BPD. The author stratified his analysis by the diagnosis of preoperative sleep apnea. Among patients without the diagnosis, the BMI fell postoperatively from 56.4 to 35.5, whereas among those with the diagnosis, the BMI fell from 70.3 to 46.5. Of the nine sleep apnea patients who had long-term followup, all had resolution of their sleep apnea. Postoperative complications included one case each of vitamin A and D deficiency, folic acid deficiency, gallstone development, kidney stone, laryngeal edema, and incisional hernia, and three cases of protein deficiency. One patient was found to have a brain stem tumor at the time of operation and subsequently died. The author reported two late deaths, the one just mentioned and one in an eighteen-year-old female who was found dead at home 3.5 years postoperatively.

The fourth study²⁷⁴ reports the results of gastric bypass surgery in 41 children and adolescents, including 11 with Prader-Willi syndrome (a type of developmental disability that is characterized in part by insatiable appetite and resultant weight gain), who were all under age 20 years. Most patients underwent gastric bypass; however, eight patients had gastroplasty. Results reported here are for the 30 patients without Prader-Willi syndrome only. The average preoperative weight was 238 percent of ideal body weight. At three years post-op, average weight decreased to 171 percent of ideal body weight, and at five years post-op, it was 187 percent of ideal body weight. Eleven major complications occurred in the early postoperative period : three wound infections, two of which occurred in conjunction with dehiscence; two cases of stomal obstruction (one which required revision); three cases of atelectasis; two cases of pneumonia; and one case of subphrenic abscess. Two deaths occurred. One death occurred on the third postoperative day, ostensibly due to a suture line leak and overwhelming sepsis or massive pulmonary embolus, although no postmortem was performed. The second death was sudden and unexpected at 36 months postoperatively. Four of the patients ultimately underwent revisions for failure to lose weight satisfactorily.

Two studies reported results for patients who were treated with laparoscopic gastric banding. The first of these²⁷⁸ studies reported on 17 patients under the age of 20 who were operated on by a single surgeon; of these patients, 7 were between the ages of 12 and 17. Among the larger group of 17 patients, the median preoperative BMI was 44.7 which fell postoperatively to 36.1 at 6 months, 32.6 at 12 months, and 30.2 at 24 months. The authors reported no effect on growth or development at followup in patients who underwent the procedure. They also reported complications in two patients, one with band slippage at 11 months, which was corrected laparoscopically, and another that required replacement of a leaking port. The second study²⁷⁹ reported on 11 patients between the ages of 11 and 17 years (8 girls), all of whom received laparoscopic adjustable gastric bands. The mean preoperative BMI was 46.6. One patient had heart failure and pulmonary hypertension, two patients had amenorrhea, and another had gallstones. The author reported no perioperative complications and no late complications. After a mean of 23 months of followup, the mean BMI had fallen to 32.1 with improvement in all comorbid conditions.

The last two studies involved surveys of patients who had undergone bariatric surgery as adolescents. The first of these contacted 14 of 18 patients who had been less than 21 years of age at the time of surgery (performed between 1982 and 1994). These patients (11 females and 3 males) had all undergone VBG. The preoperative BMIs were 45 and 59, respectively, which fell postoperatively, and at the time of the interview were 33 and 35, respectively. The authors reported that one male patient with preoperative sleep apnea had “complete clinical resolution” following weight loss. The second study consisted of an interview of 34 out of 39 patients who had undergone RYGB or VBG between 1979 and 1990. The patients (27 females and 7 males) were all between the ages of 11 and 19 at the time of the surgery. Preoperative BMI averaged 47 and at followup was 32. Complications included a staple line failure in one patient, which was also “suspected in other patients with large weight regains.” The authors reported no major postoperative complications. Five revision procedures and four subsequent cholecystectomy procedures were scheduled or performed.

In summary, a handful of case reports of bariatric surgery in adolescents have appeared, reporting on a total of 172 subjects. These reports document benefits in terms of weight loss and resolution of complications as well as harms in terms of surgical complications. No studies have compared these benefits and harms to those of similar patients who received nonsurgical therapies such as diet or medication.

Publication Bias

Our literature search procedures were extensive and included canvassing experts regarding studies we may have missed. However, we tested for evidence of publication bias only in the medication meta-analysis and found such evidence in one case (orlistat at 12-month followup). We made explicit assumptions about the lack of reporting of mortality and other adverse events and discussed the possible bias that might result. We acknowledge that publication bias may still exist despite our best efforts to conduct a comprehensive search and the lack of statistical evidence for the existence of bias. Publication bias may occur for a variety of reasons, including investigators' loss of interest in the study if “negative” results are obtained or if results are contrary to the interest of the sponsor or investigator.

Study Quality

An important limitation common to systematic reviews is the quality of the original studies. Recent efforts to identify elements of study design and execution that may be associated with bias have, in many cases, been unable to distinguish biased studies or provided results that were not reproducible. Therefore, the current state of the science is to avoid rejecting studies or using quality criteria to adjust the results of meta-analysis. Thus, we made no attempt to assign greater importance to some studies based on “quality.” Because empirical evidence is lacking regarding the relationship of other study characteristics to bias, we did not attempt to use other criteria. Most of the studies of orlistat and fluoxetine had Jadad scores of 3 or greater, a threshold that in other settings has been shown to be associated with less bias. Our sensitivity analyses on these higher-quality RCTs upheld our main result. Also, to be as inclusive as possible for purposes of assessing safety, we did consider nonrandomized and noncontrolled studies in our surgery analysis while noting the limited inferences that can be drawn from such designs.

Heterogeneity

Evidence of heterogeneity was observed for all the medication meta-analyses. We used a pooled random-effects approach to attempt to incorporate any heterogeneity and assessed the results of sensitivity analyses using variables that might account for heterogeneity (quality, completeness of followup, dose, year of publication). However, we were unable to explain most of the heterogeneity. Our pooled results should be interpreted in light of the observed heterogeneity.

Followup Times

For medication studies, we were able to perform meta-analysis only on 6-month and 12-month followup results for medication studies. For the surgical studies, the wide variation in followup times should be kept in mind when considering our pooled results.

Applicability of Findings

The results of the studies we synthesized are directly applicable only to the persons included in those studies. In some cases, enrollment was highly selective to avoid certain comorbidities. Whether the results are applicable to more representative populations is unknown.

Medications

A number of RCTs of weight loss medications have been conducted; nevertheless, significant unanswered questions remain regarding the medications assessed in this report. One of our key questions concerned relative efficacy, a question that cannot be conclusively answered without head-to-head RCTs that compare the different agents. However, the placebo-controlled trial data we reviewed suggest that if any statistically significant differences are seen, they are likely to be clinically small (a difference of a few kilograms at 12 months). Whether it is worth trying to detect such differences is a matter for policymakers. A more relevant question regarding efficacy may be whether combinations of agents promote greater weight loss than individual agents. One study that combined orlistat and sibutramine reported no increase in weight loss over sibutramine alone. Another relevant question is whether use of any of those drugs combined with more aggressive behavioral interventions and diet therapies would be more effective than the results seen in the RCTs to date, where many of the dietary interventions were modest.

Another of our key questions concerned the optimal duration of treatment and whether it varies by age, gender, or race. We found no RCT data to answer this question; therefore, new clinical trials would need to be performed. Some physicians have expressed the opinion that they expect their overweight patients will always need to take diet medications, in essence treating overweight as a chronic disease like hypertension. Given that possibility, information about long-term (i.e., much longer than 12 months) effectiveness and safety is needed.

The question of side effects, particularly the possibility of rare adverse events, remains unanswered for most of these drugs.

Surgery

For patients with BMI ≥ 40, we regard the data as conclusive concerning the superiority of surgical therapy compared to existing pharmaceutical and diet therapy. Significant advances will need to occur in the medical control of obesity or its complications for new comparative studies to be warranted. For patients with a BMI between 35 and 40, we do not regard the existing published data as conclusive, because the data are derived from case series without a concurrent comparison group. Although randomized clinical trials would be welcome, given the widespread adoption of bariatric surgery in adults, it might be difficult to mount a trial comparing surgical and nonsurgical weight loss methods. If RCTs cannot be performed, conclusive data could be obtained from well-conducted observational studies, such as a population-based matched cohort study similar to the SOS study, that would assess the effectiveness of bariatric surgery compared to nonsurgical therapy. Nonsurgical therapies to be evaluated should include high-intensity behavioral interventions in addition to pharmacotherapy. Such a study should address the balance between benefits (in terms of weight loss and comorbidities) and risks in relevant patient subgroups and should consider costs. The criteria used to identify subgroups should be those clinical factors that may be related to increased or decreased benefits or risks of surgery. For example, patient age, weight, and severity of comorbidity all may influence the net benefit of surgery compared to nonsurgical therapy. In this example, patient subgroups might be identified based on BMI; age might be stratified as below or above 55 years old; and comorbidities such as diabetes could be stratified by measures such as hemoglobin A1c or the presence of end organ damage. Only by conducting such a study can it be determined whether the net benefits of surgical therapy compared to nonsurgical therapy apply equally to all patient subgroups. Surprisingly few patients would need to be studied to assess whether or not bariatric surgery provides greatly superior outcomes compared to nonsurgical therapy. We calculate that as few as 215 patients would need to be included in each group to provide 80 percent power to detect an effect of bariatric surgery on reducing diabetes equal to only half that reported in the SOS trial. Given that more than 100,000 bariatric operations are performed yearly, attempting to enroll 0.2 percent of these patients in a clinical study seems feasible.

If the eligibility criteria for bariatric surgery were relaxed (such as allowing inclusion of people with a BMI of 30 to 32), then it becomes more justifiable to require an RCT to assess the relative health benefits and risks of surgical versus nonsurgical treatment prior to widespread adoption of surgery in this patient population. Precedent exists for mounting such studies, even when the surgical therapy is already disseminated in the community. The studies comparing medical and surgical therapy for carotid artery stenosis are good examples.

RCTs would also be useful to compare the effectiveness and safety of various surgical procedures (e.g. adjustable band procedures versus RYGB).

Last, given the increasing rate of obesity in adolescent and pediatric populations, the need is urgent for more data about the relative efficacy of treatments. In our opinion, conducting a randomized trial of bariatric surgery in the adolescent population is still feasible. Such a study would go a long way toward establishing the role of surgery in this patient population.

Reference List

Reference List

1.

Abu-Abeid S, Gavert N, Klausner J M. et al. Bariatric surgery in adolescence. J Pediatr Surg. 2003; 38(9): 1379–82. [PubMed]

2.

Abu-Abeid S, Szold A. Results and complications of laparoscopic adjustable gastric banding: an early and intermediate experience. Obes Surg. 1999; 9(2): 188–90. [PubMed]

3.

Aghahosseini H, Roulet D, Cavin R. [Treatment of morbid obesity with adjustable gastric prosthesis: experience and results at the Riviera Hospital in Montreux]. Rev Med Suisse Romande. 2001; 121(10): 709–12. [PubMed]

4.

Agren G, Naslund I. A prospective randomized comparison of vertical banded gastroplasty (VBG), loop gastric bypass (GBY), and gastric banding (GB). Int J Obes. 1989; 13: 595.

5.

Al-Jiffry B O, Shaffer E A, Saccone G T. et al. Changes in gallbladder motility and gallstone formation following laparoscopic gastric banding for morbid obesity. Can J Gastroenterol. 2003; 17(3): 169–74. [PubMed]

6.

Alden J F. Gastric and jejunoileal bypass. A comparison in the treatment of morbid obesity. Arch Surg. 1977; 112(7): 799–806. [PubMed]

7.

Alper D, Ramadan E, Vishne T. et al. Silastic ring vertical gastroplasty- long-term results and complications. Obes Surg. 2000; 10(3): 250–4. [PubMed]

8.

Andersen T, Backer O G, Astrup A. et al. Horizontal or vertical banded gastroplasty after pretreatment with very- low-calorie formula diet: a randomized trial. Int J Obes. 1987; 11(3): 295–304. [PubMed]

9.

Andersen T, Backer O G, Stokholm K H. et al. [Gastroplasty versus very low calorie diet in morbid obesity. Short-term results of a randomized clinical trial]: <Original> Gastroplastik versus ekstrem lavkalorie-diaet (very-low-calorie diet) ved svaer adipositas. Korttidsresultater af en randomiseret klinisk undersogelse. Ugeskrift for laeger. 1982; 144(6): 390–4. [PubMed]

10.

Andersen T, Backer O G, Stokholm K H. et al. Randomized trial of diet and gastroplasty compared with diet alone in morbid obesity. New England journal of medicine. 1984; 310(6): 352–6. [PubMed]

11.

Andersen T, Stokholm K H, Backer O G. et al. Long-term (5-year) results after either horizontal gastroplasty or very- low-calorie diet for morbid obesity. Int J Obes. 1988; 12(4): 277–84. [PubMed]

12.

Anderson A E, Soper R T, Scott D H. Gastric bypass for morbid obesity in children and adolescents. J Pediatr Surg. 1980; 15(6): 876–81. [PubMed]

13.

Anthone G J, Lord R V N, DeMeester T R. et al. The duodenal switch operation for the treatment of morbid obesity. Ann Surg. 2003; 238(4): 618–628. [PubMed] [Free Full text in PMC]

14.

Balsiger B M, Poggio J L, Mai J. et al. Ten and more years after vertical banded gastroplasty as primary operation for morbid obesity. J Gastrointest Surg. 2000; 4(6): 598–605. [PubMed]

15.

Baltasar A, Bou R, Arlandis F. et al. Vertical banded gastroplasty at more than 5 years. Obes Surg. 1998; 8(1): 29–34. [PubMed]

16.

Belachew M, Belva P H, Desaive C. Long-term results of laparoscopic adjustable gastric banding for the treatment of morbid obesity. Obes Surg. 2002; 12(4): 564–8. [PubMed]

17.

Belachew M, Legrand M, Vincent V. et al. Laparoscopic adjustable gastric banding. World J Surg. 1998; 22(9): 955–63. [PubMed]

18.

Bloomston M, Zervos E E, Camps M A. et al. Outcome following bariatric surgery in super versus morbidly obese patients: does weight matter? Obes Surg. 1997; 7(5): 414–9. [PubMed]

19.

Breaux C W. Obesity Surgery in Children. Obes Surg. 1995; 5(3): 279–284. [PubMed]

20.

Brolin R L, Robertson L B, Kenler H A. et al. Weight loss and dietary intake after vertical banded gastroplasty and Roux-en-Y gastric bypass. Ann Surg. 1994; 220(6): 782–90. [PubMed] [Free Full text in PMC]

21.

Brolin R E, Bradley L J, Wilson A C. et al. Lipid risk profile and weight stability after gastric restrictive operations for morbid obesity. J Gastrointest Surg. 2000; 4(5): 464–9. [PubMed]

22.

Brolin R E, Kenler H A, Gorman J H. et al. Long-limb gastric bypass in the superobese. A prospective randomized study. Annals of surgery. 1992; 215(4): 387–95. [PubMed] [Free Full text in PMC]

23.

Buckwalter J A. A prospective comparison of the jejunoileal and gastric bypass operations for morbid obesity. World J Surg. 1977; 1(6): 757–68. [PubMed]

24.

Capella J F, Capella R F. The weight reduction operation of choice: vertical banded gastroplasty or gastric bypass? Am J Surg. 1996; 171(1): 74–9. [PubMed]

25.

Choban P S, Flancbaum L. The effect of Roux limb lengths on outcome after Roux-en-Y gastric bypass: a prospective, randomized clinical trial. Obes Surg. 2002; 12(4): 540–5. [PubMed]

26.

Choban P S, Onyejekwe J, Burge J C. et al. A health status assessment of the impact of weight loss following Roux- en-Y gastric bypass for clinically severe obesity. J Am Coll Surg. 1999; 188(5): 491–7. [PubMed]

27.

Choi Y, Frizzi J, Foley A. et al. Patient satisfaction and results of vertical banded gastroplasty and gastric bypass. Obes Surg. 1999; 9(1): 33–5. [PubMed]

28.

Cook C M, Edwards C. Success habits of long-term gastric bypass patients. Obes Surg. 1999; 9(1): 80–2. [PubMed]

29.

Courcoulas A, Perry Y, Buenaventura P. et al. Comparing the outcomes after laparoscopic versus open gastric bypass: a matched paired analysis. Obes Surg. 2003; 13(3): 341–6. [PubMed]

30.

Crampton N A, Izvornikov V, Stubbs R S. Silastic ring gastric bypass: results in 64 patients. Obes Surg. 1997; 7(6): 489–94. [PubMed]

31.

Dargent J. Laparoscopic adjustable gastric banding: lessons from the first 500 patients in a single institution. Obes Surg. 1999; 9(5): 446–52. [PubMed]

32.

Das S K, Roberts S B, Kehayias J J. et al. Body composition assessment in extreme obesity and after massive weight loss induced by gastric bypass surgery. Am J Physiol Endocrinol Metab. 2003; 284(6): E1080–8. [PubMed]

33.

Davila-Cervantes A, Ganci-Cerrud G, Gamino R. et al. Open vs. laparoscopic vertical banded gastroplasty: a case control study with a 1-year follow-up. Obes Surg. 2000; 10(5): 409–12. [PubMed]

34.

de Csepel J, Nahouraii R, Gagner M. Laparoscopic gastric bypass as a reoperative bariatric surgery for failed open restrictive procedures. Surg Endosc. 2001; 15(4): 393–7. [PubMed]

35.

De Luca M, de Werra C, Formato A. et al. Laparotomic vs laparoscopic lap-band: 4-year results with early and intermediate complications. Obes Surg. 2000; 10(3): 266–8. [PubMed]

36.

de Wit L T, Mathus-Vliegen L, Hey C. et al. Open versus laparoscopic adjustable silicone gastric banding: a prospective randomized trial for treatment of morbid obesity. Annals of surgery. 1999; 230(6): 800–5. [PubMed] [Free Full text in PMC]

37.

de Zwaan M, Lancaster K L, Mitchell J E. et al. Health-related quality of life in morbidly obese patients: effect of gastric bypass surgery. Obes Surg. 2002; 12(6): 773–80. [PubMed]

38.

DeMaria E J, Sugerman H J. A critical look at laparoscopic adjustable silicone gastric banding for surgical treatment of morbid obesity: does it measure up? Surg Endosc. 2000; 14(8): 697–9. [PubMed]

39.

DeMaria E J, Sugerman H J, Meador J G. et al. High failure rate after laparoscopic adjustable silicone gastric banding for treatment of morbid obesity. Ann Surg. 2001; 233(6): 809–18. [PubMed] [Free Full text in PMC]

40.

Doherty C, Maher J W, Heitshusen D S. Prospective investigation of complications, reoperations, and sustained weight loss with an adjustable gastric banding device for treatment of morbid obesity. J Gastrointest Surg. 1998; 2(1): 102–8. [PubMed]

41.

Doherty C, Maher JW, Heitshusen DS. Long-term data indicate a progressive loss in efficacy of adjustable silicone gastric banding for the surgical treatment of morbid obesity. Surgery 2002;132(4):724–7; discussion 727–8.

42.

Dolan K, Creighton L, Hopkins G. et al. Laparoscopic gastric banding in morbidly obese adolescents. Obes Surg. 2003; 13(1): 101–4. [PubMed]

43.

Doldi S B, Micheletto G, Lattuada E. et al. Adjustable gastric banding: 5-year experience. Obes Surg. 2000; 10(2): 171–3. [PubMed]

44.

Dymek M P, Le Grange D, Neven K. et al. Quality of life after gastric bypass surgery: a cross-sectional study. Obes Res. 2002; 10(11): 1135–42. [PubMed]

45.

Feng J J, Gagner M. Laparoscopic biliopancreatic diversion with duodenal switch. Semin Laparosc Surg. 2002; 9(2): 125–9. [PubMed]

46.

Field R J Jr, Field R J 3rd, Park S Y. Vertical banded gastroplasty: is obesity worth it? J Miss State Med Assoc. 1992; 33(12): 423–32. [PubMed]

47.

Fielding G A, Rhodes M, Nathanson L K. Laparoscopic gastric banding for morbid obesity. Surgical outcome in 335 cases. Surg Endosc. 1999; 13(6): 550–4. [PubMed]

48.

Fobi M A. Vertical Banded Gastroplasty vs Gastric Bypass: 10 years follow-up. Obes Surg. 1993; 3(2): 161–164. [PubMed]

49.

Fobi M A, Lee H, Igwe D Jr. et al. Prospective comparative evaluation of stapled versus transected silastic ring gastric bypass: 6-year follow-up. Obes Surg. 2001; 11(1): 18–24. [PubMed]

50.

Forestieri P, Meucci L, De Luca M. et al. Two years of practice in adjustable silicone gastric banding (LAP- BAND): evaluation of variations of body mass index, percentage ideal body weight and percentage excess body weight. Obes Surg. 1998; 8(1): 49–52. [PubMed]

51.

Forsell P, Hallerback B, Glise H. et al. Complications following Swedish adjustable gastric banding: a long-term follow-up. Obes Surg. 1999; 9(1): 11–6. [PubMed]

52.

Forsell P, Hellers G. The Swedish Adjustable Gastric Banding (SAGB) for morbid obesity: 9 year experience and a 4-year follow-up of patients operated with a new adjustable band. Obes Surg. 1997; 7(4): 345–51. [PubMed]

53.

Freeman J B, Kotlarewsky M, Phoenix C. Weight loss after extended gastric bypass. Obes Surg. 1997; 7(4): 337–44. [PubMed]

54.

Goulding S T, Hovell B C. Anaesthetic experience of vertical banded gastroplasty. Br J Anaesth. 1995; 75(3): 301–6. [PubMed]

55.

Greenstein R J, Rabner J G. Is Adolescent Gastric-Restrictive Antiobesity Surgery Warranted? Obes Surg. 1995; 5(2): 138–144. [PubMed]

56.

Griffen W O Jr, Young V L, Stevenson C C. A prospective comparison of gastric and jejunoileal bypass procedures for morbid obesity. Ann Surg. 1977; 186(4): 500–9. [PubMed] [Free Full text in PMC]

57.

Gustavsson S, Westling A. Laparoscopic adjustable gastric banding: complications and side effects responsible for the poor long-term outcome. Semin Laparosc Surg. 2002; 9(2): 115–24. [PubMed]

58.

Hall J C, Watts J M, O'Brien P E. et al. Gastric surgery for morbid obesity. The Adelaide Study. Ann Surg. 1990; 211(4): 419–27. [PubMed] [Free Full text in PMC]

59.

Hedenbro J L, Frederiksen S G. Fully stapled gastric bypass with isolated pouch and terminal anastomosis: 1–3 year results. Obes Surg. 2002; 12(4): 546–50. [PubMed]

60.

Hess D S, Hess D W. Biliopancreatic diversion with a duodenal switch. Obes Surg. 1998; 8(3): 267–82. [PubMed]

61.

Hesse U J, Berrevoet F, Ceelen W. et al. [Adjustable silicone gastric banding (ASGB, Bioenterics) and the Swedish adjustable gastric banding (SAGB, Obtech) in treatment of morbid obesity]: <Original> Das anpassbare Silicon Gastric Banding (ASGB, Bioenterics) und das Schwedische anpassbare Gastric Banding (SAGB, Obtech) zur Behandlung der morbiden Obesitas. Der Chirurg; Zeitschrift fur alle Gebiete der operativen Medizen. 2001; 72(1): 14–8. [PubMed]

62.

Higa K D, Boone K B, Ho T. Complications of the laparoscopic Roux-en-Y gastric bypass: 1,040 patients--what have we learned? Obes Surg. 2000; 10(6): 509–13. [PubMed]

63.

Higa KD, Boone KB, Ho T, et al . Laparoscopic Roux-en-Y gastric bypass for morbid obesity: technique and preliminary results of our first 400 patients. Arch Surg 2000;135(9):1029–33; discussion 1033–4.

64.

Horchner R, Tuinebreijer W. Improvement of physical functioning of morbidly obese patients who have undergone a Lap-Band operation: one-year study. Obes Surg. 1999; 9(4): 399–402. [PubMed]

65.

Howard L, Malone M, Michalek A. et al. Gastric Bypass and Vertical Banded Gastroplasty- a Prospective Randomized Comparison and 5-Year Follow-up. Obes Surg. 1995; 5(1): 55–60. [PubMed]

66.

Kalfarentzos F, Dimakopoulos A, Kehagias I. et al. Vertical banded gastroplasty versus standard or distal Roux-en-Y gastric bypass based on specific selection criteria in the morbidly obese: preliminary results. Obes Surg. 1999; 9(5): 433–42. [PubMed]

67.

Karason K, Lindroos A K, Stenlof K. et al. Relief of cardiorespiratory symptoms and increased physical activity after surgically induced weight loss: results from the Swedish Obese Subjects study. Arch Intern Med. 2000; 160(12): 1797–802. [PubMed]

68.

Karason K, Molgaard H, Wikstrand J. et al. Heart rate variability in obesity and the effect of weight loss. Am J Cardiol. 1999; 83(8): 1242–7. [PubMed]

69.

Karason K, Wallentin I, Larsson B. et al. Effects of obesity and weight loss on left ventricular mass and relative wall thickness: survey and intervention study. BMJ. 1997; 315( 7113): 912–6. [Free Full text in PMC] [PubMed]

70.

Karason K, Wikstrand J, Sjostrom L. et al. Weight loss and progression of early atherosclerosis in the carotid artery: a four-year controlled study of obese subjects. Int J Obes Relat Metab Disord. 1999; 23(9): 948–56. [PubMed]

71.

Karlsson J, Sjostrom L, Sullivan M. Swedish obese subjects (SOS)--an intervention study of obesity. Two-year follow-up of health-related quality of life (HRQL) and eating behavior after gastric surgery for severe obesity. Int J Obes Relat Metab Disord. 1998; 22(2): 113–26. [PubMed]

72.

Kim C H, Sarr M G. Severe reflux esophagitis after vertical banded gastroplasty for treatment of morbid obesity. Mayo Clin Proc. 1992; 67(1): 33–5. [PubMed]

73.

Kothari S N, DeMaria E J, Sugerman H J. et al. Lap-band failures: conversion to gastric bypass and their preliminary outcomes. Surgery. 2002; 131(6): 625–9. [PubMed]

74.

Laws H L, Piantadosi S. Superior gastric reduction procedure for morbid obesity: a prospective, randomized trial. Ann Surg. 1981; 193(3): 334–40. [PubMed] [Free Full text in PMC]

75.

Lechner G W, Elliott D W. Comparison of weight loss after gastric exclusion and partitioning. Arch Surg. 1983; 118(6): 685–92. [PubMed]

76.

Lechner G W, Callender A K. Subtotal gastric exclusion and gastric partitioning: a randomized prospective comparison of one hundred patients. Surgery. 1981; 90(4): 637–44. [PubMed]

77.

Letiexhe M R, Scheen A J, Gerard P L. et al. Post-gastroplasty recovery of ideal body weight normalizes glucose and insulin metabolism in obese women. J Clin Endocrinol Metab. 1995; 80: 334–369. [PubMed]

78.

Lujan J A, Hernandez Q, Frutos M D. et al. Laparoscopic gastric bypass in the treatment of morbid obesity. Preliminary results of a new technique. Surg Endosc. 2002; 16(12): 1658–62. [PubMed]

79.

Lundell L, Forssell H, Jensen J. et al. Measurement of pouch volume and stoma diameter after gastroplasty. Int J Obes. 1987; 11(2): 169–74. [PubMed]

80.

Lundell L, Ruth M, Olbe L. Vertical banded gastroplasty or gastric banding for morbid obesity: effects on gastro-oesophageal reflux. Eur J Surg. 1997; 163(7): 525–31. [PubMed]

81.

MacLean L, Rhode B. Does Genetic Predisposition Influence Surgical Results of Operations for Obesity? Obes Surg. 1996; 6(2): 132–137. [PubMed]

82.

MacLean L D, Rhode B M, Forse R A. Late results of vertical banded gastroplasty for morbid and super obesity. Surgery. 1990; 107(1): 20–7. [PubMed]

83.

MacLean L D, Rhode B M, Nohr C W. Late outcome of isolated gastric bypass. Ann Surg. 2000; 231(4): 524–8. [PubMed] [Free Full text in PMC]

84.

MacLean LD, Rhode BM, Sampalis J, et al. Results of the surgical treatment of obesity. Am J Surg 1993;165(1):155–60; discussion 160–2.

85.

Marceau P, Biron S, Bourque R A. et al. Biliopancreatic Diversion with a New Type of Gastrectomy. Obes Surg. 1993; 3(1): 29–35. [PubMed]

86.

Mason E E. Vertical banded gastroplasty for obesity. Arch Surg. 1982; 117(5): 701–6. [PubMed]

87.

Mason E E, Doherty C, Cullen J J. et al. Vertical gastroplasty: evolution of vertical banded gastroplasty. World J Surg. 1998; 22(9): 919–24. [PubMed]

88.

Matthews B D, Sing R F, DeLegge M H. et al. Initial results with a stapled gastrojejunostomy for the laparoscopic isolated roux-en-Y gastric bypass. Am J Surg. 2000; 179(6): 476–81. [PubMed]

89.

Melissas J, Christodoulakis M, Spyridakis M. et al. Disorders associated with clinically severe obesity: significant improvement after surgical weight reduction. South Med J. 1998; 91(12): 1143–8. [PubMed]

90.

Miller K, Hell E. Laparoscopic adjustable gastric banding: a prospective 4-year follow-up study. Obes Surg. 1999; 9 (2): 183–7. [PubMed]

91.

Mitchell J E, Lancaster K L, Burgard M A. et al. Long-term follow-up of patients' status after gastric bypass. Obes Surg. 2001; 11(4): 464–8. [PubMed]

92.

Mittermair R P, Weiss H, Nehoda H. et al. Laparoscopic Swedish adjustable gastric banding: 6-year follow-up and comparison to other laparoscopic bariatric procedures. Obes Surg. 2003; 13(3): 412–7. [PubMed]

93.

Morino M, Toppino M, Bonnet G. et al. Laproscopic adjustable silicone gastric banding versus vertical banded gastroplasy in morbidly obese patients. Ann Surg. 2003; 238(6): 835–842. [PubMed] [Free Full text in PMC]

94.

Narbro K, Agren G, Jonsson E. et al. Sick leave and disability pension before and after treatment for obesity: a report from the Swedish Obese Subjects (SOS) study. Int J Obes Relat Metab Disord. 1999; 23(6): 619–24. [PubMed]

95.

Naslund I. The size of the gastric outlet and the outcome of surgery for obesity. Acta Chir Scand. 1986; 152: 205–10. [PubMed]

96.

Naslund I. Gastric bypass versus gastroplasty. A prospective study of differences in two surgical procedures for morbid obesity. Acta Chir Scand Suppl. 1987; 536: 1–60. [PubMed]

97.

Naslund I, Beckman K W. Gastric emptying rate after gastric bypass and gastroplasty. Scand J Gastroenterol. 1987; 22 (2): 193–201. [PubMed]

98.

Naslund I, Hallgren P, Sjostrom L. Fat cell weight and number before and after gastric surgery for morbid obesity in women. Int J Obes. 1988; 12(3): 191–7. [PubMed]

99.

Naslund I, Jarnmark I, Andersson H. Dietary intake before and after gastric bypass and gastroplasty for morbid obesity in women. Int J Obes. 1988; 12(6): 503–13. [PubMed]

100.

Naslund I, Wickbom G, Christoffersson E. et al. A prospective randomized comparison of gastric bypass and gastroplasty. Complications and early results. Acta Chir Scand. 1986; 152: 681–9. [PubMed]

101.

Nguyen NT, Goldman C, Rosenquist CJ, et al. Laparoscopic versus open gastric bypass: a randomized study of outcomes, quality of life, and costs. Ann Surg 2001;234(3):279–89; discussion 289–91.

102.

Nguyen NT, Ho HS, Palmer LS, et al. A comparison study of laparoscopic versus open gastric bypass for morbid obesity. J Am Coll Surg 2000;191(2):149–55; discussion 155–7.

103.

Nilsell K, Thorne A, Sjostedt S. et al. Prospective randomised comparison of adjustable gastric banding and vertical banded gastroplasty for morbid obesity. Eur J Surg. 2001; 167(7): 504–9. [PubMed]

104.

Nowara H A. Egyptian experience in laparoscopic adjustable gastric banding (technique, complications and intermediate results). Obes Surg. 2001; 11(1): 70–5. [PubMed]

105.

O'Brien P E, Dixon J B, Brown W. et al. The laparoscopic adjustable gastric band (Lap-Band): a prospective study of medium-term effects on weight, health and quality of life. Obes Surg. 2002; 12 (5): 652–60. [PubMed]

106.

O'Brien P E, Brown W A, Smith A. et al. Prospective study of a laparoscopically placed, adjustable gastric band in the treatment of morbid obesity. Br J Surg. 1999; 86(1): 113–8. [PubMed]

107.

Oh CH, Kim HJ, Oh S. Weight loss following transected gastric bypass with proximal Roux-en-Y. Obes Surg 1997;7(2):142–7; discussion 148.

108.

Olbers T, Lonroth H, Fagevik-Olsen M. et al. Laparoscopic gastric bypass: development of technique, respiratory function, and long-term outcome. Obes Surg. 2003; 13(3): 364–70. [PubMed]

109.

Papasavas P K, Hayetian F D, Caushaj P F. et al. Outcome analysis of laparoscopic Roux-en-Y gastric bypass for morbid obesity. The first 116 cases. Surg Endosc. 2002; 16(12): 1653–7. [PubMed]

110.

Peace K, Dyne J, Russell G. et al. Psychobiological effects of gastric restriction surgery for morbid obesity. N Z Med J. 1989; 102(862): 76–8. [PubMed]

111.

Perugini RA, Mason R, Czerniach DR, et al. Predictors of complication and suboptimal weight loss after laparoscopic Roux-en-Y gastric bypass: a series of 188 patients. Arch Surg 2003;138(5):541–5; discussion 545–6.

112.

Pontiroli A E, Pizzocri P, Librenti M C. et al. Laparoscopic adjustable gastric banding for the treatment of morbid (grade 3) obesity and its metabolic complications: a three-year study. J Clin Endocrinol Metab. 2002; 87(8): 3555–61. [PubMed]

113.

Pories W J, Caro J F, Flickinger E G. et al. The control of diabetes mellitus (NIDDM) in the morbidly obese with the Greenville Gastric Bypass. Ann Surg. 1987; 206(3): 316–23. [PubMed] [Free Full text in PMC]

114.

Pories W J, Flickinger E G, Meelheim D. et al. The effectiveness of gastric bypass over gastric partition in morbid obesity: consequence of distal gastric and duodenal exclusion. Ann Surg. 1982; 196(4): 389–99. [PubMed] [Free Full text in PMC]

115.

Quaade F. Jjunoileal bypass for morbid obesity. A bibliographic study and a randomized clinical trial. Surgical clinics of North America. 1979; 59(6): 1055–69. [PubMed]

116.

Rand C S, Macgregor A M. Adolescents having obesity surgery: a 6-year follow-up. South Med J. 1994; 87(12): 1208–13. [PubMed]

117.

Randolph J G, Weintraub W H, Rigg A. Jejunoileal bypass for morbid obesity in adolescents. J Pediatr Surg. 1974; 9(3): 341–5. [PubMed]

118.

Reddy RM, Riker A, Marra D, et al. Open Roux-en-Y gastric bypass for the morbidly obese in the era of laparoscopy. Am J Surg 2002;184(6):611–5; discussion 615–6.

119.

Reinhold R B. Late results of gastric bypass surgery for morbid obesity. J Am Coll Nutr. 1994; 13(4): 326–31. [PubMed]

120.

Rigg CA. Proceedings: Jejunoileal bypass by morbidly obese adolescent. Acta Paediatr Scand Suppl 1975;(256):62–4.

121.

Schauer P R, Ikramuddin S, Gourash W. et al. Outcomes after laparoscopic Roux-en-Y gastric bypass for morbid obesity. Ann Surg. 2000; 232(4): 515–29. [PubMed] [Free Full text in PMC]

122.

Scopinaro N, Gianetta E, Adami G F. et al. Biliopancreatic diversion for obesity at eighteen years. Surgery. 1996; 119(3): 261–8. [PubMed]

123.

Sjostrom C D, Lissner L, Wedel H. et al. Reduction in incidence of diabetes, hypertension and lipid disturbances after intentional weight loss induced by bariatric surgery: the SOS Intervention Study. Obes Res. 1999; 7(5): 477–84. [PubMed]

124.

Sjostrom C D, Peltonen M, Sjostrom L. Blood pressure and pulse pressure during long-term weight loss in the obese: the Swedish Obese Subjects (SOS) Intervention Study. Obes Res. 2001; 9(3): 188–95. [PubMed]

125.

Sjostrom C D, Peltonen M, Wedel H. et al. Differentiated long-term effects of intentional weight loss on diabetes and hypertension. Hypertension. 2000; 36(1): 20–5. [PubMed]

126.

Smith S C, Edwards C B, Goodman G N. et al. Open vs laproscopic Roux-en-Y gastric bypqass: comparison of operative morbidity and mortality. Obes Surg. 2004; 14: 73–76. [PubMed]

127.

Soper R T, Mason E E, Printen K J. et al. Gastric bypass for morbid obesity in children and adolescents. J Pediatr Surg. 1975; 10(1): 51–8. [PubMed]

128.

Stanford A, Glascock J M, Eid G M. et al. Laparoscopic Roux-en-Y gastric bypass in morbidly obese adolescents. J Pediatr Surg. 2003; 38(3): 430–3. [PubMed]

129.

Stoner J, Stoner P, Sytsma J. 42-month preliminary follow-up of the silastic ring vertical banded gastroplasty. Obes Surg. 1997; 7(6): 513–5. [PubMed]

130.

Strauss R S, Bradley L J, Brolin R E. Gastric bypass surgery in adolescents with morbid obesity. J Pediatr. 2001; 138(4): 499–504. [PubMed]

131.

Sugerman H J, Kellum J M Jr, DeMaria E J. et al. Conversion of failed or complicated vertical banded gastroplasty to gastric bypass in morbid obesity. Am J Surg. 1996; 171(2): 263–9. [PubMed]

132.

Sugerman H J, Londrey G L, Kellum J M. et al. Weight loss with vertical banded gastroplasty and Roux-Y gastric bypass for morbid obesity with selective versus random assignment. Am J Surg. 1989; 157(1): 93–102. [PubMed]

133.

Sugerman H J, Starkey J V, Birkenhauer R. A randomized prospective trial of gastric bypass versus vertical banded gastroplasty for morbid obesity and their effects on sweets versus non- sweets eaters. Ann Surg. 1987; 205(6): 613–24. [PubMed] [Free Full text in PMC]

134.

Sugerman HJ, Sugerman EL, DeMaria EJ, et al. Bariatric surgery for severely obese adolescents. J Gastrointest Surg 2003;7(1):102–7; discussion 107–8.

135.

Suter M, Bettschart V, Giusti V. et al. A 3-year experience with laparoscopic gastric banding for obesity. Surg Endosc. 2000; 14(6): 532–6. [PubMed]

136.

Suter M, Giusti V, Heraief E. et al. Early results of laparoscopic gastric banding compared with open vertical banded gastroplasty. Obes Surg. 1999; 9(4): 374–80. [PubMed]

137.

Suter M, Giusti V, Heraief E. et al. Laparoscopic Roux-en-Y gastric bypass: initial 2-year experience. Surg Endosc. 2003; 17(4): 603–9. [PubMed]

138.

Szold A, Abu-Abeid S. Laparoscopic adjustable silicone gastric banding for morbid obesity: results and complications in 715 patients. Surg Endosc. 2002; 16(2 ): 230–3. [PubMed]

139.

Tacchino R M, Mancini A, Perrelli M. et al. Body composition and energy expenditure: relationship and changes in obese subjects before and after biliopancreatic diversion. Metabolism. 2003; 52(5): 552–8. [PubMed]

140.

The Danish Obesity Project. Randomised trial of jejunoileal bypass versus medical treatment in morbid obesity. Lancet. 1979; 2(8155): 1255–8. [PubMed]

141.

Toppino M, Morino M, Bonnet G. et al. Laparoscopic surgery for morbid obesity: preliminary results from SICE registry (Italian Society of Endoscopic and Minimally Invasive Surgery). Obes Surg. 1999; 9(1): 62–5. [PubMed]

142.

Toppino M, Morino M, Capuzzi P. et al. Outcome of vertical banded gastroplasty. Obes Surg. 1999; 9(1): 51–4. [PubMed]

143.

Torgerson J S, Sjostrom L. The Swedish Obese Subjects (SOS) study--rationale and results. Int J Obes Relat Metab Disord. 2001; 25 Suppl 1: S2–4. [PubMed]

144.

van de Weijgert E J, Ruseler C H, Elte J W. Long-term follow-up after gastric surgery for morbid obesity: preoperative weight loss improves the long-term control of morbid obesity after vertical banded gastroplasty. Obes Surg. 1999; 9(5): 426–32. [PubMed]

145.

van Gemert W G, Adang E M, Greve J W. et al. Quality of life assessment of morbidly obese patients: effect of weight- reducing surgery. Am J Clin Nutr. 1998; 67(2): 197–201. [PubMed]

146.

Van Rij A M. Gastric bypass and gstric partitioning in morbid obesity: results of a controlled trial. Aus NZ J Med. 1984; 14(3 suppl 1): 340.

147.

VanWoert J, Michalek A, Alger S. et al. A prospective randomized trial (PRT) of gastric bypass (GB) versus vertical banded gatroplasty (VBG) for morbid obesity (MO). Clinical Research. 1992; 40(2): A642.

148.

Viddal K O. Intestinal bypass. A randomized, prospective clinical study of end-to- side and end-to-end jejunoileal bypass. Scand J Gastroenterol. 1983; 18(5): 627–34. [PubMed]

149.

Weiner R, Blanco-Engert R, Weiner S. et al. Outcome after laparoscopic adjustable gastric banding - 8 years experience. Obes Surg. 2003; 13(3): 427–34. [PubMed]

150.

Weiner R, Bockhorn H, Rosenthal R. et al. A prospective randomized trial of different laparoscopic gastric banding techniques for morbid obesity. Surgical endoscopy. 2001; 15(1): 63–8. [PubMed]

151.

Westling A, Gustavsson S. Laparoscopic vs open Roux-en-Y gastric bypass: a prospective, randomized trial. Obes Surg. 2001; 11(3): 284–92. [PubMed]

152.

White J J, Cheek D, Haller J A Jr. Small bowel bypass is applicable for adolescents with morbid obesity. Am Surg. 1974; 40(12): 704–8. [PubMed]

153.

Wiesner W, Schob O, Hauser R S. et al. Adjustable laparoscopic gastric banding in patients with morbid obesity: radiographic management, results, and postoperative complications. Radiology. 2000; 216(2): 389–94. [PubMed]

154.

Wittgrove A C, Clark G W. Laparoscopic gastric bypass, Roux-en-Y- 500 patients: technique and results, with 3–60 month follow-up. Obes Surg. 2000; 10(3): 233–9. [PubMed]

155.

Yale C E. Gastric surgery for morbid obesity. Complications and long-term weight control. Arch Surg. 1989; 124(8): 941–6. [PubMed]

156.

Yale C E, Weiler S J. Weight control after vertical banded gastroplasty for morbid obesity. Am J Surg. 1991; 162(1): 13–8. [PubMed]

157.

Yashkov YI, Timoshin AD, Oppel TA. Vertical banded gastroplasty: first experience in Russia. Obes Surg 1997;7(4):317–20; discussion 321.