- Click here for explanation of classification scheme.

• To review literature to estimate the effectiveness of bariatric surgery relative to non-surgery therapy
• To compare outcomes of surgical techniques.

• Surgery with concurrent comparison group
• RCT
• Matched cohort studies
• Controlled clinical trials
• Case series with 10 or more patients
• Mean weight loss and standard deviations reported
• Follow-up time: 12 months and maximum greater than or equal to 36 months.  

• Mean weight loss not reported
• Follow-up time not long enough
• Duplicate study (data already used)
• Mean weight loss not reported by surgery type.

Search Procedures

Initial search began with an electronic MEDLINE search, followed by EMBASE and subsequent periodic search updates. 

Study Quality Assessed

Yes.

Blinding Used

N/A.

Intervention Investigated

• Intervention: surgical therapy and procedures for obesity
• Main Outcomes: weight loss, mortality, complication rates and control of obesity-related co-morbid conditions
• Population: obese individuals.

 Statistical Analysis

• Mean weights and standard deviations
• For comparative studies mean differences were calculated
• Mean differences pooled using a random effects model  
• 95% confidence intervals estimated. 

 

Information Abstracted from Articles

• Number of patients and co-morbid conditions
• Adverse events
• Types of outcome measures and time from intervention until outcome
• Characteristics of study samples: median age, percentage of women, median baseline weight (in kg or BMI), percentage of patients with co-morbid conditions, percentage of improvement or resolution of the conditions and median follow-up time
• Whether the case series studies reported on consecutive patients. 

 Combining Information

The most commonly reported measurements of weight loss at baseline and at follow-up in kilograms or BMI.

Analytic Methods Used

• Means and standard deviations at 12 months and maximum follow-up time greater than or equal to 36 months
• For comparative studies, mean differences were pooled using a random effects model at 95% CI.

 

 

Number (N) of Articles Included

N=167; 159 surgery articles reported on weight loss; an additional eight reported on complications.  

Number of Articles Identified

N=1103; 1064 screened.

Final N for Weight Loss

N=89 reviewed and contributed to the weight loss analysis.

Number and Type of Studies Reviewed

• Two RCTs compared bariatric surgery with non-surgical procedure. One used horizontal gastroplasty plus diet and diet alone. This study generated three articles. The second used jejuno-ileal bypass with medical treatment (studies were more than 20 years old).
• Numerous reports on observational Swedish Obese Subjects study (comparative intervention) with voluntary bariatric surgery vs. matched controls treated medically. Matching was done on 18 variables (age, height, weight, sex). For this study the intervention was RYGB, vertical banded gastroplasty, or banding procedures.
• A third RCT (abstract only) compared surgery with medical treatment.

Sample Size of Studies and Characteristics of Study Participants

• Article one (first RCT) with horizontal gastroplasty: 
  • Initial N=30, 26 completed the study; VLCD; Initial N=30, 29 completed the study 
  • 88% women; average age 34 years 
• Article two with horizontal gastroplasty:
  • Initial N=22, 20 completed the study; VBG: Initial N=23, 21 completed study
  • 81% women; average age 34 years
• Another RCT jejuno-ileal bypass:
  •  Initial N=32, 18 final compared to medical treatment (Initial N=27, 22 final);
  • 61% women; average age 33 years 
• SOS study: N=251 in surgical group compared to N=232 in medically treated group; average age 47 years; 66% women
• SOS: 10-year follow-up data N=1703, surgery included gastric bypass, banding procedures or VBG (mainly women, age not reported)
• Abstract RCT, N=79 patients randomly assigned to laparoscopic adjustible gastric banding or medical treatment (VLCD, pharmotherapy and exercise). Characteristics not available.  

 

Pooled Results for Weight Loss Trials

Number of Trials and Surgery	Weight Loss at 12 Months (kg)	Mean Diffrence at 12 Months (95% CI), kg	Weight Loss at 36 Months (kg)	Mean Difference (95% CI), kg
2 Trials Comparing All RYGB vs. All VBG	42.43kg RYGBP  (N=114) vs. 34.45kg VBG (N=117)	7.97 (2.99 to 12.96)	39.73kg for n=103  RYGBP vs. 30.65kg for N=96 VBG	9.29 (161 to 16.96)
1 Trial Open RYGBP vs. Lap RYGBP	34.35kg RYGBP open (N=21) vs. 37 kg RYGBP lap (N=30)	-2.64 (-11.28 to 6.00)	No trial	No trial
2 Trials All VBG vs. All Adjustable Band	38.58kg all VBG (N=71) vs. 24.20kg adjustable band (N=76)	14.41 (9.39 to 19.42)	35.51kg for N=64 all VBG vs. 32.97kg for N=60 for adjustable band	2.79 (-16.63 to 22.21)

Main Results of the Review

• In the first two RCT trials, net weight loss was not significantly different at six months between the two groups. However, at 24 months follow-up, the net weight change from baseline "greatly favored" surgery (30.5kg loss) over non-surgical methods (8kg).
• Next RCT at 24 months follow-up, mean difference in weight loss of 37kg "favored surgical procedure" over medical
• Swedish (SOS) average weight loss for surgically treated group was 20kg at eight year follow-up; no change was observed for medically treated group. Patients treated with RYGB lost more weight than VBG or banding procedures.
• In the 10-year follow-up found significantly greater weight loss with surgically treated group than controls (16% vs. 1.6%; P<0.001). Patients with gastric bypass lost more weight than those treated with VBG or banded procedures.
• SOS also reported that results were inconclusive for individuals with BMI of 40kg per m² or higher compared to individuals with a lower BMI.
• From the abstract, weight loss outcomes at two years (24 months) were reported as excess weight loss of 71.5% for the surgically treated group vs. 21.4% for the medically treated group (P<0.001). There was not enough data in abstract for generalization assessment.   

• Surgical treatment was more effective than non-surgical procedures for weight loss and control of some co-morbid conditions in patients with a BMI of 40kg per m² or higher
• For patients with a BMI of 35kg per m² to 39kg per m², data is inconclusive because information was derived from case series without concurrent camparison group
• More data is needed to confirm or refute relative efficacy of surgery for less severely obese persons
• Benefits seem to lie with the surgical procedure (e.g., RYGB vs. laparoscopic adjustable band) which seems to provide greater benefit especially for patient types such as age, sex, BMI or co-morbidity profile
• Controlled trials or well matched observational studies are needed to address procedures' effectiveness, comparable ability to generate sustainable weight loss, complication rate reduction in co-morbid conditions and improvement in quality of life.

Government:

Agency for Healthcare Research and Quality

Strengths

• The review provided us with valuable insight into available data up to 2005
• There was an attempt to utilize RCTs, clinical controlled trials, and cohort studies
• Findings on the Swedish studies were significant, however the matching of the controls may be an issue.

Limitations

• Use of old RCT studies (more than 20 years old) and case studies in which results were pooled
• Sample sizes were small
• Only one large cohort was used. 

Concerns

• The abstract data was included in the results so validity could not be verified
• The summary tables provided more detailed characteristics than the report
• More information on methodology is necessary.