Impact of laparoscopic adjustable gastric banding on type 2 diabetes


John Dixon, Monash University, Bldg 1, 270 Ferntree Gully Road, Melbourne, Vic. 3168, Australia. E-mail:


Bariatric surgery is becoming an accepted option for obese people with type 2 diabetes. Our aim was to assess the impact of laparoscopic adjustable gastric banding (LAGB) through a systematic review of the literature. Data was sourced from Scopus, MEDLINE and EMBASE published from 2000 through May 2011, and five unpublished studies that were performed by industry for regulatory approval were also included. Studies were selected on the basis that they provide some detail of diabetes status before and after LAGB. There were 35 studies meeting the inclusion criteria. There was considerable heterogeneity in study design, sample size, length of follow-up, attrition rates and classification of diabetes status. Weight loss was progressive over the first 2 years with a weighted average of 47% excess weight loss at 2 years. Remission or improvement in diabetes varied from 53% to 70% over different time periods. Results were broadly consistent, demonstrating clinically relevant improvements in diabetes outcomes with sustained weight loss in obese people with type 2 diabetes following LAGB surgery. However, there were significant shortcomings in the reviewed literature with few high-quality studies, inconsistent reporting of diabetes outcomes and high attrition rates. Long-term studies that address these limitations are needed.


Obesity has become a global epidemic. Currently, an estimated 1.1 billion people worldwide are overweight or obese (1). Obesity prevalence rates in several countries now exceed 30% (2). The obesity epidemic is of particular concern because it increases risk for several adverse health conditions. One of the primary health hazards of obesity is diabetes. Approximately half of those diagnosed with type 2 diabetes are obese, and among the obese, risk for developing diabetes increases dramatically with increasing weight (3,4).

Although obesity is the primary cause of diabetes, research has shown that losing as little as 5–10% of body weight can prevent the onset of diabetes or result in improvements or even resolution post occurrence. Moreover, larger decreases in weight generate even greater benefits (5). However, large weight losses among the severely obese population (those with a body mass index [BMI; kg/m2] over 35) have proven difficult to sustain through lifestyle modification. Bariatric surgery, which includes gastric bypass surgery and gastric banding, has been shown to be far more effective than medications and/or lifestyle interventions at sustaining weight loss.

A meta-analysis by Buchwald et al. (6) showed that persons with diabetes undergoing gastric bypass surgery lost an average of 60% of excess body weight (EWL) and that 80% of diabetes cases fully resolved. Although much of the improvements in diabetes resolution are due to the weight loss, there is likely to be an additional metabolic explanation given that many patients had their glucose levels improve even before the weight loss began to accrue (7,8). Gastric banding generates smaller weight losses than gastric bypass and the improvements in diabetes status are also slower to materialize. Yet Buchwald et al. (6) reported that 57% of diabetes cases fully resolved with average weight losses of 46% EWL.

Many review articles have focused on the health improvements resulting from bariatric surgery, including diabetes resolution rates (6,9–14). However, to date, no review article has specifically focused on diabetes resolution rates for obese individuals with diabetes who undergo laparoscopic adjustable gastric banding (LAGB), which is the current standard for band placement. This article fills that gap. We review information from prior studies and review articles specifically related to improvements in diabetes status among obese individuals with diabetes who undergo LAGB.

A current review specifically targeting improvements in diabetes status resulting from LAGB is important given that many obese individuals with diabetes (and many payers) may prefer LAGB over gastric bypass surgery as long as the potential for improvements in diabetes status is relatively high, even if not as high as gastric bypass surgery. This follows because banding is reversible and has lower complication rates than gastric bypass surgery (14–16). This study provides a single source for accessing that information.


Literature search and inclusion criteria

We conducted a comprehensive review of recent literature and controlled clinical studies to ascertain the effect of LAGB on type 2 diabetes. We searched for articles in Scopus, MEDLINE and EMBASE published from 2000 through May 2011 that included the terms ‘laparoscopic band’‘lap-band’‘laparoscopic adjustable gastric banding’ or ‘gastric band’ in combination with ‘diabetes’. The initial search revealed 119 studies. Bibliographic citations were reviewed to identify additional literature for consideration. Exclusion criteria consisted of articles that did not present data on improvements in diabetes status post banding, articles that only discussed gastric bypass surgery, articles that did not include the follow-up period for which diabetes improvement was reported and those in a language other than English. Case reports and abstracts were not included in order to avoid duplication of results with published studies.

To be included, the studies needed to contain some LAGB patients diagnosed with diabetes prior to surgery. In several cases, although the overall sample was large, the subsample of LAGB patients with diabetes at the time of surgery was small. Moreover, in studies that enrolled patients both with and without diabetes, enrolment, attrition and weight loss outcomes were not always reported separately for each subgroup. Therefore, for each study, we present data on the target population, study design, number of diabetes cases in the sample at baseline, average age, percent female, starting BMI, length of follow-up, attrition rates and excess weight lost from baseline (EWL) for the diabetes sample if available or for the overall sample if not. However, we only present improvements in diabetes status among those patients with diabetes at baseline.

Unpublished clinical studies

In addition to our search of the existing literature, we also included previously unpublished results from five clinical trials conducted by Allergan both in the USA and internationally that were undertaken as part of the regulatory review and approval process for the USA and other countries. The five Allergan trials represent all of their completed LAP-BAND® studies that collected diabetes outcomes data. They are briefly summarized below. We were unable to identify unpublished data from other band manufacturers for inclusion.

  • 1LAGB-001-B was a US-based, multi-centre 1 year study approved by the US Food and Drug Administration (FDA) in 1998 as a continuing and expanded access study to develop and maintain surgical skills at investigational sites and to increase access of the LAP-BAND® for patients. A primary objective was to assess changes in comorbid conditions associated with obesity post surgery. A total of 193 subjects were enrolled at 12 investigational sites, with 15 of these subjects diagnosed with type 2 diabetes prior to implantation.
  • 2LAGB-001-C was a US-based multi-centre 1 year study approved by the FDA in 2000 to allow obesity surgeons to develop LAP-BAND® surgical skills at new investigational sites and to expand access of the device to patients while FDA review was ongoing. An additional goal was to collect data on the band's influence on comorbid conditions. A total of 220 subjects were enrolled at 17 investigational sites, including 22 with type 2 diabetes diagnosed prior to implantation.
  • 3Following FDA approval in 2001, another US-based multi-centre study, LAGB-PM-001-D, was undertaken as a condition of approval to obtain longer term follow-up after implantation for subjects who had been enrolled in the pre-approval studies. A total of 109 subjects were evaluated at year 3, including 17 with type 2 diabetes prior to implantation.
  • 4LAGB-INT-MOB-9802 was a retrospective study at six international sites, including Australia, France, Italy, Mexico and two sites in Belgium. Reduction in obesity-related comorbid conditions was one of the primary outcomes studied. A total of 441 subjects were included in this retrospective study including 46 who had type 2 diabetes prior to surgery. The LAP-BAND® surgeries were performed between November 1993 and November 1998, with primary outcomes, including changes in diabetes status, measured at 1, 2 and 3 years post surgery.
  • 5LAGB-INT-MOB-9801 was a prospective data collection effort from three international investigational sites (Australia, Italy and Mexico) on subjects implanted with the LAP-BAND® System between November 1998 and June 2000. As with the other international study, one of the primary outcomes studied was reduction in comorbid conditions. A total of 225 subjects were enrolled, including 32 who were diagnosed with type 2 diabetes prior to surgery.

Definitions and data presentation

The terminology used to define diabetes improvements, which included terms such as remission, resolution or cured, and ranges considered normal for glycosylated haemoglobin (HbA1c) differed across studies. Therefore, consistent with the terminology adopted by the Diabetes Surgery Summit (International Conference on Gastrointestinal Surgery to Treat Type 2 Diabetes, Rome, Italy, March 29–31, 2007), the term remission was adopted to describe improvements to normal levels among those who were considered to have diabetes at baseline, where normal is as defined in the original manuscript. Using this terminology, diabetes improvements are presented as the percentage of those with diabetes at baseline who were in remission, improved but not in remission, or who showed no change/worsening. In some cases, we were unable to differentiate improvements from remissions. In these cases, we conservatively placed the percentages in the improvements column and a zero in the remission column to ensure the totals added up to 100%. These cases are noted in the table. Using this approach, we then combined results across studies to present weighted average improvements at each time period, with weights based on sample sizes of the included studies. Based on data from those studies that report laboratory values, we also present a separate table that includes changes in HbA1c or fasting plasma glucose (FPG) among the subset of cases with diabetes at baseline. For each table, we also present weighted averages of the variables included in the table with weights based on the number of diabetes cases included in each study. To ease interpretation, we present results in separate tables based on the average length of follow-up post banding: 6 months, roughly 1 year, 15–24 months or 2–5 years.


Thirty-five studies, representing 13 countries, met the initial inclusion criteria. These studies included 23 case series from the published literature and the five Allergan trials that followed LAGB patients post surgery, two retrospective data analysis of LAGB patients, four non-randomized case-control studies that compared LAGB patients to either non-surgical patients or to bariatric surgery patients and one randomized controlled trial. Three types of bands were represented in the data, including LAP-BAND® (Allergan, Santa Barbara, CA, USA), Swedish Band (Ethicon Endo-Surgery, Cincinnati, OH, USA) and Easyband® (Allergan, Lausanne, Switzerland) although the vast majority of studies reported results based on patients who received the LAP-BAND®.

6 months

Table 1 presents results for the three studies with 6-month assessments. All three studies were case series. For these studies, the weighted average baseline age was 46, 76% were female and baseline BMI was 45.0. Remission of diabetes varied considerably across the three studies; the weighted average rate of 62% was largely driven by the large sample size of the Dolan et al. (17) study. Based on the weighted average, an additional 8% showed improvements in diabetes status that fell short of full remission, while 30% saw no change or a worsening of diabetes. Mean EWL was far more consistent, with each study reporting a value between 25% and 31%, leading to the overall weighted mean EWL of 30%.

Table 1.  Baseline and follow-up values for studies with 6 months post-LAGB follow-up
First author (country – year)Target population (surgical system used)Study designNo. of diabetes casesAge% FemaleStarting BMILength of follow-upAttrition (%)Remission (%)Improvement (%)No change/worse (%)Mean EWL lost (%)
  • *

    Based on overall population not diabetes population

  • BMI, body mass index; EWL, excess body weight; T2DM, type 2 diabetes mellitus.

Dolan et al., 2003 (17)– AustraliaT2DM or history of gestational diabetes with at least 6 months follow-up (Lap-Band)Case series8847.57445.06.5 months average (max of 19 months)Not clear but all 49 seem to be followed up for 12 months6543130.60
Spivak et al., 2004 (18)– USAT2DM with BMI >35 (Lap-Band)Case series1440*87*45.3*6 months63*29363525
Weiner et al., 2007 (19)– GermanyT2DM with BMI >35 (Easy Band telemetrically adjustable gastric band)Case series536*81*43.3*6 months3*1000028.40
Weighted average  107 total cases46.07645.1 64.8*6283030

12 months

Ten published studies and four Allergan studies provided results with average follow-up of 12–13 months. The 10 published studies presented in Table 2 included seven case series, two retrospective data analysis and one non-randomized case-control study. For these studies, the weighted average baseline age was 45.4, 72% were female and average baseline BMI was 45.2. The weighted average remission rate for diabetes was 52.3%. An additional 16.8% of diabetes cases improved but were not resolved. At least 50% of participants showed improvements/resolution in each of the published studies except for the one by DeMaria et al. (23). Improvements in EWL from the studies ranged from 22% to 62%, with a weighted average of 34.8%. The results of the published literature were, on average, more favourable than the results of the Allergan trials in terms of diabetes improvements and excess weight lost.

Table 2.  Baseline and follow-up values for studies with 12–13 months post-LAGB follow-up
First author (country – year)Target population (surgical system used)Study designNo. of diabetes casesAge% FemaleStarting BMILength of follow-upAttrition (%)Remission (%)Improvement (%)No change/worse (%)Mean EWL lost (%)
  • *

    Based on overall population not diabetes population.

  • BMI, body mass index; EWL, excess body weight; IGT, impaired glucose tolerance; LAGB, laparoscopic adjustable gastric banding; NR, not recorded; T2DM, type 2 diabetes mellitus; wc, with comorbidity.

Bacci et al., 2002 (20)– ItalyT2DM, BMI >35 with comorbidity (Lap-Band)Case series745 (wc)92 (wc)43.6 (wc)12 months62*57430 
Brancatisano et al., 2008 (21)– AustraliaT2DM, impaired glucose tolerance or metabolic syndrome (Swedish band)Case series7852 (T2DM)58 (T2DM)47.0 (T2DM)12.5 months25% > 6 months*51301937.8 (T2DM)
47 (IGT)83 (IGT)45.0 (IGT)41 (IGT)
Busetto et al., 2008 (22)– ItalyT2DM, age 60 or older (Lap-Band)Retrospective data analysis4664.1*85.2*44.2*12 months7*01000 
DeMaria et al., 2010 (23)– USAT2DM with BMI 30- < 35 (not specified)Retrospective10952.076.6*33.96–12 months6327.5NR62.521.6
Dixon and O'Brien, 2002 (24)– AustraliaT2DM, BMI >35 with comorbidities (Lap-Band)Case series503448.212 months264261038
Gan et al., 2007 (25)– AustraliaT2DM with BMI >35 undergoing LAGB compared to those undergoing Roux-en-Y gastric bypass and laparoscopic sleeve gastrectomy (not specified)Case control- categorized by type of surgery128345.613 months average (max 36 months)117335034.2*
Kasama, et al., 2008 (26)– JapanT2DM with BMI >35 (Lap-Band or Lap-Band VG)Case series443*61.5*37.5*12 months1150252561.7*
LAGB-001-BMulti-centre study of 193 patients (Lap-Band)Prospective1541.5*81.9*46.6*12 months2033.38.358.337.0*
LAGB-001-CMulti-centre study of 220 patients (Lap-Band)Prospective2241.0*85.0*47.4*12 months022.731.845.540.8*
LAGB-INT-MOB-9802Retrospective study at six international sites in Australia, France, Italy, Mexico and Belgium of 441 patients (Lap-Band)Retrospective4636.0*87.0*43.0*12 months2.224.4NR75.646.3*
LAGB-INT-MOB-9801Multi-centre study with sites in Australia, Italy and Mexico of 225 patients (Lap-Band)Prospective3236.0*76.0*45.0*12 months9.472.4NR27.641.0
Nadler et al., 2007 (27)– USAAdolescents with impaired glucose tolerance (Lap-Band or Lap-Band VG)Case series515.8*74.0*47.6*12 months36*1000056.7*
Singhal et al., 2008 (28)– UKT2DM with BMI >35 undergoing banding surgery (Lap-Band VG, Lap-Band VG or Swedish band)Case series10946.766.452.912 months6.636.6 (insulin)61 (insulin)2.4 (insulin)34.3
34.2 (metformin)34.2 (metformin)31.6 (metformin)
Weiner et al., 2003 (29)– GermanyT2DM with morbid obesity undergoing LAGB (Lap-Band for all but two patients who received Swedish band)Case series16137.9*77.9*46.4*12 months2*9208 
Weighted average  696 total cases45.47245.2 16.752.316.829.034.8

15–24 months

Table 3 presents the results of five published case series, two of the unpublished Allergan studies and one published randomized controlled trial (RCT) with follow-up data averaging between 15 and 24 months. The weighted average baseline age for participants in these studies was 39.4 years, 74.2% were female and baseline BMI was 44. Rates of diabetes remission varied between 20% and 100%, although 4 of the 8 studies (including most of the observations) were in the 50–75% range. The weighted average remission rate was 51%, with 4% showing improvements that fell short of remission. Weighted average EWL was 47%, with nearly identical values in both the studies from the literature and those from the unpublished Allergan trials. The RCT published by Dixon et al. (32) presents results that are more favourable than the Allergan trial data or the non-randomized studies.

Table 3.  Baseline and follow-up values for studies with 15–24 months post-LAGB follow-up
First author (country – year)Target population (surgical system used)Study designNo. of diabetes casesAge% FemaleStarting BMILength of follow-upAttritionRemission (%)Improvement (%)No change/worse (%)Mean EWL lost (%)
  • *

    Based on overall population not diabetes population.

  • BMI, body mass index; EWL, excess body weight; LAGB, laparoscopic adjustable gastric banding; NR, not recorded; RCT, randomized controlled trial; T2DM, type 2 diabetes mellitus.

Abu-Abeid et al., 2001 (30)– IsraelT2DM, over age 60 with BMI >35 (Lap-Band)Case series763.6*72.2*44.4*21.9 months averageNot clear71290 
Busetto et al., 2004 (31)– ItalyT2DM with BMI >35 (Lap-Band)Case series7337.5*46.6*15.3 months average (12–18 months)21.7*55045 
Dixon et al., 2008 (32)– AustraliaT2DM with BMI 30–40 randomized to LAGB and conventional therapy (Lap-Band)RCT30 to each intervention46.6 (LAGB)50 (LAGB)37.0 (LAGB)24 months8.373NR2762.5*
Kasza et al., 2011 (33)– USAT2DM with BMI >35 (Lap-Band)Case series3143*90.3*45.6*24 months70.2*20126834*
LAGB-INT-MOB-9802Retrospective study at six international sites in Australia, France, Italy, Mexico and Belgium of 441 patients (Lap-Band)Retrospective4636.0*87.0*43.0*24 months6.530.2NR69.850.9*
LAGB-INT-MOB-9801Multi-centre study with sites in Australia, Italy and Mexico of Lap-Band for 225 patients (Lap-Band)Prospective3236.0*76.0*45.0*24 months15.659.3NR40.744.9
Lee et al., 2006 (34)– TaiwanT2DM with BMI >35 (Lap-Band)Case series1631.2*48.4*42.7*24 months0*1000044.0*
Spivak et al., 2005 (35)– USAT2DM with BMI >35 (Lap-Band)Case series1242*45.2*At least 18 months67.4*33333337
Weighted Average  247 total cases39.474.244.0 23.451.03.945.047.0

≥24 months follow-up

Thirteen studies, including eight published case series, two of the Allergan studies and three non-randomized case-control studies provided results with greater than 24 months of follow-up. These are presented in Table 4. The weighted average baseline age for these studies was 43.9 years, 70.8% were female and baseline average BMI was 46.1. For nearly all of the published studies, the rate of diabetes remission was between 40% and 75%. The three exceptions that were well outside of this range had small sample sizes. The Fielding et al. (39) study, which only had four participants with diabetes, reported 100% remission. Rubenstein et al. (44), which had a sample of six people with diabetes only reported diabetes improvements, so no remission information was available. The study by Boza et al. (36) had 11 people with diabetes and reported an improvement rate of 28.6%. The unpublished Allergan trials reported slightly lower rates of diabetes remission with rates of 34–35% though the LAGB-PM-001-D study also reported an 18% improvement rate. Across all of the studies, weighted averages were 37.6% for diabetes remission and 23.1% for improvements. Overall, the weighted average EWL was 44.8%. Figure 1 displays diabetes remission and improvement rates alongside %EWL over time from 6 months to >24 months.

Table 4.  Baseline and follow-up values for studies with >24 months post-LAGB follow-up
First author (country – year)Target Population (surgical system used)Study designNo. of Diabetes CasesAge% FemaleStarting BMILength of follow-upAttrition (%)Remission (%)Improvement (%)No Change/Worse (%)Mean EWL Lost (%)
  • *

    Based on overall population not diabetes population.

  • BMI, body mass index; EWL, excess body weight; IGT, impaired glucose tolerance; LAGB, laparoscopic adjustable gastric banding; NR, not recorded; T2DM, type 2 diabetes mellitus.

Boza et al., 2011 (36)– ChileT2DM (36% Lap-Band, 64% Swedish band)Case series1137.8*70.4%*36.0*48 months (median)57.8*028.6  
Caiazzo et al., 2010 (37)– FranceT2DM choosing to undergoing gastric banding (Swedish band)Case series2344.96545.360 months4.327NR7325
Cottam et al., 2006 (38)– USAT2DM undergoing banding matched with those undergoing gastric bypass surgery (Lap-Band)Case control-categorized by type of surgery4442*80*47.2*Most followed for 36 months77* at 3 years503415.851*
Fielding, 2003 (39)– AustraliaT2DM with BMI >60 undergoing (not specified)Case series439*64.5*69*Up to 60 months831000061*
Frigg et al., 2004 (40)– SwitzerlandT2DM (Lap-Band)Case series5941*79*45*44 months average67581754*
Korenkov et al., 2007 (41)– GermanyT2DM with BMI >34 (Lap-Band)Case series1437.5*72.4*48.5*36–96 months (average 5 years)55743061.9*
LAGB-PM-001-DMulti-centre study of 109 patients from the pre-approval studies (Lap-Band)Prospective1741.2*83.2*47.4*36 months035.317.647.140.0*
LAGB-INT-MOB-9802Retrospective study at six international sites in Australia, France, Italy, Mexico and Belgium of 441 patients (Lap-Band)Retrospective4636.0*87.0*43.0*36 months23.934.3NR65.748.7*
Pontiroli et al., 2002 (42)– ItalyT2DM or IGT with BMI >35 (not specified)Case control- LAGB matched to diet group6642.9*81.1*44.9*36 months17NR8020 
Pontiroli et al., 2005 (43)– ItalyT2DM with BMI >35 undergoing banding matched with those refusing to undergo surgery (Lap-Band)Case control- LAGB matched to those who refused surgery1753.37748.336 months0 (LAGB)45%NR55 
12 (No LAGB)
Rubenstein et al., 2002 (44)– USAT2DM with BMI >35 (Lap-Band)Case series640.8*88.9*48.8*Up to 36 months79* at 3 yearsNR841653.6*
Segato et al., 2011 (45)– ItalyT2DM and morbidly obese (Lap-Band)Case series5247.657.749.136 months (median)30.836.532.730.824.1
Sultan et al., 2010 (46)– USAT2DM with BMI >35 (Lap-Band)Case series10249.35246.360 months14.739.7NR60.345.7
Weighted Average  461 total cases43.970.846.1 23.237.623.139.244.8
Figure 1.

Diabetes remission/improvement rates and %EWL at each time point.

Changes in clinical values

Table 5 reports improvements in clinical biochemical values for the 10 studies that provide this information for those with diabetes at baseline. At 6 months, Weiner et al. (19) showed a clinically and statistically significant improvement from a baseline mean level of 7.5 mmol L−1 to a 6-month mean level of 5.0 mmol L−1. At 12 to 13 months, four studies provided data on lab values related to diabetes improvements. Dixon and O'Brien (24) showed that HbA1c values improved from 7.8% on average at baseline to 6.2% at 12 months. Singhal et al. (28) found smaller improvements of HbA1c from 8.2 to 7.4% and FPG from 9.1 to 6.9 mmol L−1. Brancatisano et al. (21) found an improvement from baseline to 12.5 months of 8.0% to 6.1% and Gan et al. (25) showed an improvement from 8.9% to 7.2% from baseline to 13 months. Thus, each of these studies showed roughly a clinically important 1.5 point improvement in HbA1c. Dixon et al. (32) found that weight loss with LAGB lowered HbA1c from 7.8 to 6.0 at 24 months. Pontiroli et al. (43) reported a decline in HbA1c from 9.4 to 8.0 at 36 months and Korenkov et al. (41) reported a decline in HbA1c from 7.3 to 6.3 with duration lasting between 36 and 96 months post surgery. Finally, Sultan et al. had the longest follow-up of 60 months and found a decline in HbA1c from 7.53 to 6.58 and in fasting glucose from 8.1 to 6.6 mmol L−1.

Table 5.  Changes in diabetes lab values following LAGB
StudyLength of follow-up (months)Lab measureChange in diabetes lab value% change
  • *

    Based on overall population not diabetes population.

  • FPG, fasting plasma glucose; HbA1c, haemoglobin.

Weiner et al. (19)6FPG−3.0 mmol L−1−40.0
Dixon and O'Brien (24)12HbA1c−1.6%−20.5
FPG−3.2 mmol L−1−34.0
Singhal et al. (28)12HbA1c−0.8%−9.8
FPG−2.2 mmol L−1−24.2
Brancatisano et al. (21)12.5HbA1c−1.9%−23.8
FPG−3.9 mmol L−1−40.6
Gan et al. (25)13HbA1c−1.7%−19.1
Busetto et al. (31)15.3FPG−0.6 mmol L−1*−7.2%*
Dixon et al. (32)24HbA1c (surgery)−1.8%−23.1
HbA1c (no surgery)−0.39%−5.1
Pontiroli et al. (43)36HbA1c−1.4%−14.9
Korenkov et al. (41)36–96 (average of 60)HbA1c−1.0%−13.7
Sultan et al. (46)60HbA1c−0.95%−12.6
FPG−1.5 mmol L−1−18.8


This review suggests clinically important improvements in diabetes outcomes and sustained weight loss post LAGB. Although rates of diabetes remission/improvement remained high at all time points, diabetes remission rates trended downward over time. The published data and unpublished Allergan studies show that on average, diabetes remission rates decreased from roughly 62% at 6 months to 55% at 12–24 months and finally to 38% beyond 24 months. This downward trend is consistent with data from the Swedish Obese Subjects study (47) and recent reports from gastric bypass surgery (48), and indicative of the progressive nature of type 2 diabetes. Although remission rates trended downward, when combined with diabetes improvements, the longest term studies reveal remission or improvements occurred for over 60% of LAGB patients. The early, unpublished LAGB trials showed lower rates of diabetes remission than the published studies, although, due to their small sample size, combining their data with results from the literature did not have large effects on the overall weighted averages. The smaller remission rates may have been due to more rigorous criteria for defining remission or the fact that physicians had limited LAGB experience. Moreover, some of these studies did not have a classification for diabetes improvements that fell short of remission; patients were coded as ‘in remission’ or ‘not in remission’.

Unlike diabetes remission, EWL consistently increased with length of follow-up for both the published studies and for the Allergan trial data. Combined, EWL increased from 34.8% at 12 months to 47.0% at 24 months before levelling off at 44.8% in studies with data greater than 24 months. This is consistent with other reviews of LAGB weight outcomes for those without diabetes (10). These results are encouraging as those with type 2 diabetes have been reported to have a poorer response to non-surgical weight loss programs (49–51).

The review also revealed significant shortcomings of the current literature. The included studies varied along many dimensions, including study design, duration, reporting measures, outcomes and attrition rates. Concerning attrition, Tables 1–4 reveal it varied from 16.7% on average at 12 months to 23.1% for the longest term studies. Quality varied from a single randomized controlled trial to retrospective cohort audits with no control groups. Two studies included data from individuals with BMIs less than 35 kg m−2. Although these studies are important given the recent FDA approval to expand LAP-BAND® to lower BMI groups, it is unclear whether these results would generalize to those with higher BMIs (52,53). Moreover, the majority of studies provided no biochemical evidence of glycemic control and vague or poorly defined definitions for diabetes improvement or remission. When in doubt, we conservatively assumed no improvements; therefore, our estimates are conservative in this regard. Although the inconsistent reporting may surprise those familiar with managing type 2 diabetes, this is indicative of the broader literature regarding bariatric surgery (6). There is clearly a need for standardized terminology and reporting of relevant outcome measures for those with type 2 diabetes (54).

This review was limited to weight and diabetes outcomes. It did not include short- or long-term complications as this information was not available in nearly all of the studies reviewed. However, safety of the LAGB, albeit not specifically for those with diabetes, has been well described in recent large cohort and registry studies (55–57). These confirm that LAGB surgery provides the lowest post-operative mortality, fewer complications in the first year and the shortest hospital stay of all bariatric surgical procedures. Longer term issues such as proximal pouch dilatation (58), erosion of the band into the stomach (58,59) and port and tubing issues have also been reviewed (60). In large series, reoperation rates for all complications combined are in the order of 10–15% at 5 years, and removal of the band without replacement is roughly 5%, although many early series had higher reoperation rates (61).

Many of the studies also had very small sample sizes for individuals with type 2 diabetes. We account for this by pooling data across studies and calculating weighted averages, although in some cases, data unique to those with diabetes was not available. Moreover, because the longer term studies may suffer from attrition bias, we cannot be sure whether study dropouts might have shown less favourable weight and health outcomes. If so, our estimates would be biased. An additional limitation is that many studies did not report HbA1c values, so our estimates of the effect of LAGB on improvement in HbA1c is represented by only a subset of the included studies. Although longer term studies that address these limitations are needed, the studies to date provide compelling evidence that LAGB leads to both short and longer term clinically relevant improvements in diabetes status primarily attributed to sustained EWL among diabetes patients who undergo the procedure.

Potential conflicts of interest

Dr Dixon, Dr Finkelstein and Mr Segel are consultants to Allergan, and Ms Murphy is an Allergan employee and stockholder.


The authors would like to thank Julie Gilmore of Allergan for assistance with the research and Allergan for financial support.