Articles
Multidisciplinary diabetes care with and without bariatric surgery in overweight people: a randomised controlled trial John M Wentworth, Julie Playfair, Cheryl Laurie, Matthew E Ritchie, Wendy A Brown, Paul Burton, Jonathan E Shaw, Paul E O’Brien
Summary
Background Bariatric surgery improves glycaemia in obese people with type 2 diabetes, but its effects are uncertain in overweight people with this disease. We aimed to identify whether laparoscopic adjustable gastric band surgery can improve glucose control in people with type 2 diabetes who were overweight but not obese. Methods We did an open-label, parallel-group, randomised controlled trial between Nov 1, 2009, and June 30, 2013, at one centre in Melbourne, Australia. Patients aged 18–65 years with type 2 diabetes and a BMI between 25 and 30 kg/m² were randomly assigned (1:1), by computer-generated random sequence, to receive either multidisciplinary diabetes care plus laparoscopic adjustable gastric band surgery or multidisciplinary diabetes care alone. The primary outcome was diabetes remission 2 years after randomisation, defined as glucose concentrations of less than 7·0 mmol/L when fasting and less than 11·1 mmol/L 2 h after 75 g oral glucose, at least two days after stopping glucose-lowering drugs. Analysis was by intention to treat. This trial is registered with the Australian New Zealand Clinical Trials Registry, number ACTRN12609000286246. Findings 51 patients were randomised to the multidisciplinary care plus gastric band group (n=25) or the multidisciplinary care only group (n=26), of whom 23 participants and 25 participants, respectively, completed followup to 2 years. 12 (52%) participants in the multidisciplinary care plus gastric band group and two (8%) participants in the multidisciplinary care only group achieved diabetes remission (difference in proportions 0·44, 95% CI 0·17–0·71; p=0·0012). One (4%) participant in the gastric band group needed revisional surgery and four others (17%) had a total of five episodes of food intolerance due to excessive adjustment of the band. Interpretation When added to multidisciplinary care, laparoscopic adjustable gastric band surgery for overweight people with type 2 diabetes improves glycaemic control with an acceptable adverse event profile. Laparoscopic adjustable gastric band surgery is a reasonable treatment option for this population. Funding Monash University Centre for Obesity Research and Education and Allergan.
Introduction Excess body weight is a dominant driver of type 2 diabetes, a disease that places a huge burden on individuals through reduced wellbeing and productivity, and on society through increased health costs.1 Management of type 2 diabetes is generally framed within a chronic disease model, with an expectation of progressive deterioration in glucose control over time.2 This management approach has been investigated in randomised trials of lifestyle change3 or bariatric surgery,4-7 which show that weight loss in obese people with type 2 diabetes reduces drug burden and frequently achieves diabetes remission within 2 years. The effect of substantial weight loss on outcomes in overweight but not obese people (BMI between 25 and 30 kg/m²) with type 2 diabetes is less well established. Investigators of the Look AHEAD trial showed that, after 1 year, modest weight loss of 8·6% bodyweight after lifestyle intervention was associated with a diabetes remission rate of roughly 10% in people whose starting BMI was less than 32·5 kg/m².3 Reports of the effect of bariatric surgery in overweight people with type 2 diabetes are limited to observational studies of gastric bypass, intestinal exclusion, or interposition surgery,8–12
and describe improved glycaemic control after surgery but serious perioperative complications in up to 7% of patients. To date, no randomised trials of bariatric surgery in overweight but not obese people have been reported, and its true benefit, if any, remains uncertain. Laparoscopic adjustable gastric banding is a common outpatient operation for obese people that has a favourable safety profile compared with other bariatric operations13,14 and provides a durable effect, with substantial weight loss sustained for at least 15 years.15 Our previous randomised trial4 comparing laparoscopic adjustable gastric band surgery with medical weight management in moderately obese people with type 2 diabetes (BMI 30–40 kg/m²) showed that gastric banding induced diabetes remission more often than did medical therapy after 2 years of follow-up. We did this trial to establish whether laparoscopic adjustable gastric banding had a similar effect on glucose control in people with type 2 diabetes who were overweight but not obese (BMI 25–30 kg/m²). We postulated that laparoscopic adjustable gastric banding in addition to multidisciplinary diabetes care would improve glycaemic control with acceptable comparative costs and safety compared with multidisciplinary care alone.
www.thelancet.com/diabetes-endocrinology Published online April 8, 2014 http://dx.doi.org/10.1016/S2213-8587(14)70066-X
Lancet Diabetes Endocrinol 2014 Published Online April 8, 2014 http://dx.doi.org/10.1016/ S2213-8587(14)70066-X This online publication has been corrected. The corrected version first appeared at thelancet.com/ diabetes-endocrinology on April 11, 2014 See Online/Comment http://dx.doi.org/10.1016/ S2213-8587(14)70019-1 See Online for a podcast interview with John Wentworth and Paul O’Brien Centre for Obesity Research and Education (CORE), Monash University, Melbourne, Australia (J M Wentworth PhD, J Playfair BHSc, C Laurie BHSc, W A Brown PhD, P Burton PhD, Prof P E O’Brien MD); Molecular Medicine Division, The Walter and Eliza Hall Institute of Medical Research, Parkville, Melbourne, Australia (J M Wentworth, M E Ritchie PhD); and Baker IDI Heart and Diabetes Institute, Melbourne, Australia (J E Shaw PhD) Correspondence to: Prof Paul E O’Brien, Centre for Obesity Research and Education, Monash University, Melbourne, Australia [email protected]
1
Articles
Methods
Study design and participants
For the protocol see http://www. core.monash.org
We undertook this single centre, open-label, parallelgroup, randomised controlled trial in Melbourne, Australia, between Nov 1, 2009, and June 30, 2013. Inclusion criteria were age between 18 and 65 years, BMI between 25 and 30 kg/m², diabetes duration less than 5 years, willingness to be randomised to either study group, and ability to comply with the treatment protocol. Exclusion criteria were positive glutamic acid decarboxylase autoantibody titre, pancreatic disease, previous bariatric surgery, or contraindication to laparoscopic adjustable gastric banding (including previous upper gastrointestinal surgery, hypothalamic disease, pregnancy, history of psychosis, or myocardial infarct in the preceding 6 months). If HbA1c was less than 6·5% (48 mmol/mol), diabetes was confirmed by oral glucose tolerance test. The study was approved by the Human Ethics committees of Monash University and The Avenue Hospital and was registered with the Australasian Clinical Trials Registry (ACTRN12609000286246). The protocol is available online.
Randomisation and masking We randomised patients in a 1:1 ratio, via a computergenerated random sequence (without blocking or stratification), to receive laparoscopic adjustable gastric banding plus multidisciplinary diabetes care or multidisciplinary diabetes care alone. A third party assigned treatment groups with numbered envelopes, which were opened by the participant immediately after providing written consent to join the study.
Procedures Multidisciplinary diabetes care was based on guidelines from the American Diabetes Association16 and delivered in Melbourne by an endocrinologist (JMW). Biochemistry and physician review was done every 3 months in year 1, and every 6 months in year 2. Consultations with a dietitian and a diabetes educator were arranged within 3 months from enrolment in patients in the multidisciplinary care group or within 6 months in those in the laparoscopic adjustable gastric banding group. Additional consultations with members of the care team were arranged if deemed necessary by either JMW or the participant. If HbA1c was greater than 7% in year 2 of the trial, endocrinology review was scheduled every 3 months. Participants were advised to do at least 150 min of moderate-intensity physical activity each week. HbA1c was targeted to less than 7·0% (54 mmol/mol) with use of drugs available through the Australian Pharmaceutical Benefit Scheme. Metformin was recommended to all participants unless it was not tolerated or if results of an oral glucose tolerance test were normal at 1 year. Sitagliptin or exenatide were commenced if HbA1c was greater than 7·0% despite metformin or sulfonylurea therapy. Insulin was commenced after sitagliptin and 2
exenatide were trialed and shown to be ineffective at lowering HbA1c to less than 7·0%. We were able to combine exenatide with insulin, but usually increased the insulin dose and did not prescribe exenatide if HbA1c was greater than 7·0% despite insulin therapy. Blood pressure was targeted to less than 120/80 mm Hg in participants with albuminuria or a history of cardiovascular disease, and to less than 130/80 mm Hg for others. Antihypertensive drug therapy was intensified if these targets were not achieved after 3 months of lifestyle change. Anti-platelet therapy and statins were prescribed to all participants with a history of cardiovascular disease and to those older than 40 years who had an additional cardiovascular risk factor. Annual ophthalmology review was arranged through the family physician. Drugs were weaned if JMW judged that the participant would continue to meet treatment targets, which were reassessed within 3 months of this change. The dietitian recommended a tailored calorie-restricted diet to all participants. Multivitamin supplements were recommended to all participants in the gastric banding group. The sessions with a diabetes educator focused on diabetes self-management, including blood glucose monitoring and sick-day management (ie, patients are educated about the possibility of hyperglycaemia or hypoglycaemia if they develop an illness such as respiratory tract or bowel infections). Patients in the gastric banding group underwent surgery within 3 months of randomisation. The LAPBAND APS (Allergan Inc, Irvine, CA) was placed via the pars flaccida pathway and sited over the gastric cardia within 1 cm of the oesophago-gastric junction by one of three surgeons (PEO, WAB, PB). The basal fluid within the system at completion of the procedure was usually 4 mL. Incidental hiatal hernia was repaired as part of the procedure. Patient discharge was planned for roughly 3 h after completion of the procedure, and happened for all but one person. Aftercare to adjust the band and optimise eating and exercise behaviours was provided at The Centre for Bariatric Surgery (Melbourne, Australia) according to the management algorithms established by the centre.17 For all patients, clinical biochemistry was done by Melbourne Pathology (Collingwood, Australia). The oral glucose tolerance test to assess diabetes status was done at 2 year follow-up only if HbA1c was less than 6·5% (48 mmol/mol) and fasting blood glucose less than 7·0 mmol/L, irrespective of whether the participant was receiving antidiabetes therapy. Anthropometric data and blood pressure, measured by automated sphygmomanometer (Philips SureSign VS2; MA, USA), were collected by a study nurse. Quality-of-life measures were assessed at baseline and at year 2 with the Short Form-36 questionnaire,18 standardised to Australian norms (Australian Bureau of Statistics Catalogue 4399.0). We recorded all adverse events prospectively, but their severity
www.thelancet.com/diabetes-endocrinology Published online April 8, 2014 http://dx.doi.org/10.1016/S2213-8587(14)70066-X
Articles
was assessed after completion of the trial by the research nurses (JP and CL), verified by contacting the relevant health practitioner or hospital, and adjudicated by the authors. There was no data safety monitoring board.
Outcomes The primary outcome was diabetes remission 2 years after randomisation. Remission was defined as glucose concentrations of less than 7·0 mmol/L and less than 11·1 mmol/L 2 h after oral glucose, at least 2 days after stopping glucose-lowering drugs. Participants could be classified as in remission even if they had been taking glucose-lowering drugs continuously until 2 days before this test. Exploratory secondary outcomes assessed during the trial and after 2 years included weight change, glycaemic control (fasting glucose, HbA1c), blood pressure, lipid profile, drug burden, quality of life, and costs.
Economic analysis Trial intervention costs comprised costs of laparoscopic adjustable gastric banding surgery, treatment of surgical adverse events, and outpatient consultations (physician, dietitian, diabetes educator, and surgeon). Surgery costs were sourced from The Avenue Hospital (VIC, Australia) and from relevant medical specialists. The cost of the gastric banding prosthesis was based on commercial prices. Drug costs and unit costs for medical and allied health consultations were obtained from pricing schedules of the Australian Government Pharmaceutical Benefits Scheme and the Medicare Benefits Schedule. We did not include other relevant health costs (eg, other hospital admissions, time off work) in the economic analysis. Costs are reported in Australian dollars (AUD) at 2013 values.
non-normally distributed according to the ShapiroWilks test, the appendix provides data as median (IQR) with p values from a non-parametric Mann-Whitney U test. For changes in HbA1c and weight loss over time, the area enclosed by the abscissa and the curve was calculated for each patient with the trapezoid rule. Logistic regression was done with the glm function from the stats package in R, with the significance of each variable established via analysis of deviance with a χ² test. Diabetes status was the outcome and the input variables were a combination of factors that were either known (baseline BMI, percent weight loss at 2 years, diabetes duration, insulin use, HbA1c, fasting glucose, insulin, and C-peptide19–21) or hypothesised (baseline age, sex, number of diabetes drugs, waist circumference, blood pressure, lipid concentrations, and presence of albuminuria) to associate with diabetes remission after bariatric surgery. In all analyses, p values reported are unadjusted and based on a two-sided test when appropriate.
For R statistical free software see http://www.r-project.org
12 079 participants received letters of invitation
1231 responded and were telephoned
1077 ineligible at telephone interview because of BMI >30 kg/m2, no diabetes, or unwillingness to be randomised
154 assessed by clinician
103 ineligible 28 did not have diabetes 35 had BMI >30 kg/m2 40 unwilling to be randomised
Statistical analysis The power calculation was based on our previous trial4 and assumed that 60% of patients in the laparoscopic adjustable gastric band group and 15% of those in the multidisciplinary care group would achieve diabetes remission. With an allowance of five dropouts in each group, an enrolment of 25 participants in each group afforded 93% power to detect the assumed difference of 45% in diabetes remission between the groups (p
Multidisciplinary diabetes care with and without bariatric surgery in overweight people: a randomised controlled trial John M Wentworth, Julie Playfair, Cheryl Laurie, Matthew E Ritchie, Wendy A Brown, Paul Burton, Jonathan E Shaw, Paul E O’Brien
Summary
Background Bariatric surgery improves glycaemia in obese people with type 2 diabetes, but its effects are uncertain in overweight people with this disease. We aimed to identify whether laparoscopic adjustable gastric band surgery can improve glucose control in people with type 2 diabetes who were overweight but not obese. Methods We did an open-label, parallel-group, randomised controlled trial between Nov 1, 2009, and June 30, 2013, at one centre in Melbourne, Australia. Patients aged 18–65 years with type 2 diabetes and a BMI between 25 and 30 kg/m² were randomly assigned (1:1), by computer-generated random sequence, to receive either multidisciplinary diabetes care plus laparoscopic adjustable gastric band surgery or multidisciplinary diabetes care alone. The primary outcome was diabetes remission 2 years after randomisation, defined as glucose concentrations of less than 7·0 mmol/L when fasting and less than 11·1 mmol/L 2 h after 75 g oral glucose, at least two days after stopping glucose-lowering drugs. Analysis was by intention to treat. This trial is registered with the Australian New Zealand Clinical Trials Registry, number ACTRN12609000286246. Findings 51 patients were randomised to the multidisciplinary care plus gastric band group (n=25) or the multidisciplinary care only group (n=26), of whom 23 participants and 25 participants, respectively, completed followup to 2 years. 12 (52%) participants in the multidisciplinary care plus gastric band group and two (8%) participants in the multidisciplinary care only group achieved diabetes remission (difference in proportions 0·44, 95% CI 0·17–0·71; p=0·0012). One (4%) participant in the gastric band group needed revisional surgery and four others (17%) had a total of five episodes of food intolerance due to excessive adjustment of the band. Interpretation When added to multidisciplinary care, laparoscopic adjustable gastric band surgery for overweight people with type 2 diabetes improves glycaemic control with an acceptable adverse event profile. Laparoscopic adjustable gastric band surgery is a reasonable treatment option for this population. Funding Monash University Centre for Obesity Research and Education and Allergan.
Introduction Excess body weight is a dominant driver of type 2 diabetes, a disease that places a huge burden on individuals through reduced wellbeing and productivity, and on society through increased health costs.1 Management of type 2 diabetes is generally framed within a chronic disease model, with an expectation of progressive deterioration in glucose control over time.2 This management approach has been investigated in randomised trials of lifestyle change3 or bariatric surgery,4-7 which show that weight loss in obese people with type 2 diabetes reduces drug burden and frequently achieves diabetes remission within 2 years. The effect of substantial weight loss on outcomes in overweight but not obese people (BMI between 25 and 30 kg/m²) with type 2 diabetes is less well established. Investigators of the Look AHEAD trial showed that, after 1 year, modest weight loss of 8·6% bodyweight after lifestyle intervention was associated with a diabetes remission rate of roughly 10% in people whose starting BMI was less than 32·5 kg/m².3 Reports of the effect of bariatric surgery in overweight people with type 2 diabetes are limited to observational studies of gastric bypass, intestinal exclusion, or interposition surgery,8–12
and describe improved glycaemic control after surgery but serious perioperative complications in up to 7% of patients. To date, no randomised trials of bariatric surgery in overweight but not obese people have been reported, and its true benefit, if any, remains uncertain. Laparoscopic adjustable gastric banding is a common outpatient operation for obese people that has a favourable safety profile compared with other bariatric operations13,14 and provides a durable effect, with substantial weight loss sustained for at least 15 years.15 Our previous randomised trial4 comparing laparoscopic adjustable gastric band surgery with medical weight management in moderately obese people with type 2 diabetes (BMI 30–40 kg/m²) showed that gastric banding induced diabetes remission more often than did medical therapy after 2 years of follow-up. We did this trial to establish whether laparoscopic adjustable gastric banding had a similar effect on glucose control in people with type 2 diabetes who were overweight but not obese (BMI 25–30 kg/m²). We postulated that laparoscopic adjustable gastric banding in addition to multidisciplinary diabetes care would improve glycaemic control with acceptable comparative costs and safety compared with multidisciplinary care alone.
www.thelancet.com/diabetes-endocrinology Published online April 8, 2014 http://dx.doi.org/10.1016/S2213-8587(14)70066-X
Lancet Diabetes Endocrinol 2014 Published Online April 8, 2014 http://dx.doi.org/10.1016/ S2213-8587(14)70066-X This online publication has been corrected. The corrected version first appeared at thelancet.com/ diabetes-endocrinology on April 11, 2014 See Online/Comment http://dx.doi.org/10.1016/ S2213-8587(14)70019-1 See Online for a podcast interview with John Wentworth and Paul O’Brien Centre for Obesity Research and Education (CORE), Monash University, Melbourne, Australia (J M Wentworth PhD, J Playfair BHSc, C Laurie BHSc, W A Brown PhD, P Burton PhD, Prof P E O’Brien MD); Molecular Medicine Division, The Walter and Eliza Hall Institute of Medical Research, Parkville, Melbourne, Australia (J M Wentworth, M E Ritchie PhD); and Baker IDI Heart and Diabetes Institute, Melbourne, Australia (J E Shaw PhD) Correspondence to: Prof Paul E O’Brien, Centre for Obesity Research and Education, Monash University, Melbourne, Australia [email protected]
1
Articles
Methods
Study design and participants
For the protocol see http://www. core.monash.org
We undertook this single centre, open-label, parallelgroup, randomised controlled trial in Melbourne, Australia, between Nov 1, 2009, and June 30, 2013. Inclusion criteria were age between 18 and 65 years, BMI between 25 and 30 kg/m², diabetes duration less than 5 years, willingness to be randomised to either study group, and ability to comply with the treatment protocol. Exclusion criteria were positive glutamic acid decarboxylase autoantibody titre, pancreatic disease, previous bariatric surgery, or contraindication to laparoscopic adjustable gastric banding (including previous upper gastrointestinal surgery, hypothalamic disease, pregnancy, history of psychosis, or myocardial infarct in the preceding 6 months). If HbA1c was less than 6·5% (48 mmol/mol), diabetes was confirmed by oral glucose tolerance test. The study was approved by the Human Ethics committees of Monash University and The Avenue Hospital and was registered with the Australasian Clinical Trials Registry (ACTRN12609000286246). The protocol is available online.
Randomisation and masking We randomised patients in a 1:1 ratio, via a computergenerated random sequence (without blocking or stratification), to receive laparoscopic adjustable gastric banding plus multidisciplinary diabetes care or multidisciplinary diabetes care alone. A third party assigned treatment groups with numbered envelopes, which were opened by the participant immediately after providing written consent to join the study.
Procedures Multidisciplinary diabetes care was based on guidelines from the American Diabetes Association16 and delivered in Melbourne by an endocrinologist (JMW). Biochemistry and physician review was done every 3 months in year 1, and every 6 months in year 2. Consultations with a dietitian and a diabetes educator were arranged within 3 months from enrolment in patients in the multidisciplinary care group or within 6 months in those in the laparoscopic adjustable gastric banding group. Additional consultations with members of the care team were arranged if deemed necessary by either JMW or the participant. If HbA1c was greater than 7% in year 2 of the trial, endocrinology review was scheduled every 3 months. Participants were advised to do at least 150 min of moderate-intensity physical activity each week. HbA1c was targeted to less than 7·0% (54 mmol/mol) with use of drugs available through the Australian Pharmaceutical Benefit Scheme. Metformin was recommended to all participants unless it was not tolerated or if results of an oral glucose tolerance test were normal at 1 year. Sitagliptin or exenatide were commenced if HbA1c was greater than 7·0% despite metformin or sulfonylurea therapy. Insulin was commenced after sitagliptin and 2
exenatide were trialed and shown to be ineffective at lowering HbA1c to less than 7·0%. We were able to combine exenatide with insulin, but usually increased the insulin dose and did not prescribe exenatide if HbA1c was greater than 7·0% despite insulin therapy. Blood pressure was targeted to less than 120/80 mm Hg in participants with albuminuria or a history of cardiovascular disease, and to less than 130/80 mm Hg for others. Antihypertensive drug therapy was intensified if these targets were not achieved after 3 months of lifestyle change. Anti-platelet therapy and statins were prescribed to all participants with a history of cardiovascular disease and to those older than 40 years who had an additional cardiovascular risk factor. Annual ophthalmology review was arranged through the family physician. Drugs were weaned if JMW judged that the participant would continue to meet treatment targets, which were reassessed within 3 months of this change. The dietitian recommended a tailored calorie-restricted diet to all participants. Multivitamin supplements were recommended to all participants in the gastric banding group. The sessions with a diabetes educator focused on diabetes self-management, including blood glucose monitoring and sick-day management (ie, patients are educated about the possibility of hyperglycaemia or hypoglycaemia if they develop an illness such as respiratory tract or bowel infections). Patients in the gastric banding group underwent surgery within 3 months of randomisation. The LAPBAND APS (Allergan Inc, Irvine, CA) was placed via the pars flaccida pathway and sited over the gastric cardia within 1 cm of the oesophago-gastric junction by one of three surgeons (PEO, WAB, PB). The basal fluid within the system at completion of the procedure was usually 4 mL. Incidental hiatal hernia was repaired as part of the procedure. Patient discharge was planned for roughly 3 h after completion of the procedure, and happened for all but one person. Aftercare to adjust the band and optimise eating and exercise behaviours was provided at The Centre for Bariatric Surgery (Melbourne, Australia) according to the management algorithms established by the centre.17 For all patients, clinical biochemistry was done by Melbourne Pathology (Collingwood, Australia). The oral glucose tolerance test to assess diabetes status was done at 2 year follow-up only if HbA1c was less than 6·5% (48 mmol/mol) and fasting blood glucose less than 7·0 mmol/L, irrespective of whether the participant was receiving antidiabetes therapy. Anthropometric data and blood pressure, measured by automated sphygmomanometer (Philips SureSign VS2; MA, USA), were collected by a study nurse. Quality-of-life measures were assessed at baseline and at year 2 with the Short Form-36 questionnaire,18 standardised to Australian norms (Australian Bureau of Statistics Catalogue 4399.0). We recorded all adverse events prospectively, but their severity
www.thelancet.com/diabetes-endocrinology Published online April 8, 2014 http://dx.doi.org/10.1016/S2213-8587(14)70066-X
Articles
was assessed after completion of the trial by the research nurses (JP and CL), verified by contacting the relevant health practitioner or hospital, and adjudicated by the authors. There was no data safety monitoring board.
Outcomes The primary outcome was diabetes remission 2 years after randomisation. Remission was defined as glucose concentrations of less than 7·0 mmol/L and less than 11·1 mmol/L 2 h after oral glucose, at least 2 days after stopping glucose-lowering drugs. Participants could be classified as in remission even if they had been taking glucose-lowering drugs continuously until 2 days before this test. Exploratory secondary outcomes assessed during the trial and after 2 years included weight change, glycaemic control (fasting glucose, HbA1c), blood pressure, lipid profile, drug burden, quality of life, and costs.
Economic analysis Trial intervention costs comprised costs of laparoscopic adjustable gastric banding surgery, treatment of surgical adverse events, and outpatient consultations (physician, dietitian, diabetes educator, and surgeon). Surgery costs were sourced from The Avenue Hospital (VIC, Australia) and from relevant medical specialists. The cost of the gastric banding prosthesis was based on commercial prices. Drug costs and unit costs for medical and allied health consultations were obtained from pricing schedules of the Australian Government Pharmaceutical Benefits Scheme and the Medicare Benefits Schedule. We did not include other relevant health costs (eg, other hospital admissions, time off work) in the economic analysis. Costs are reported in Australian dollars (AUD) at 2013 values.
non-normally distributed according to the ShapiroWilks test, the appendix provides data as median (IQR) with p values from a non-parametric Mann-Whitney U test. For changes in HbA1c and weight loss over time, the area enclosed by the abscissa and the curve was calculated for each patient with the trapezoid rule. Logistic regression was done with the glm function from the stats package in R, with the significance of each variable established via analysis of deviance with a χ² test. Diabetes status was the outcome and the input variables were a combination of factors that were either known (baseline BMI, percent weight loss at 2 years, diabetes duration, insulin use, HbA1c, fasting glucose, insulin, and C-peptide19–21) or hypothesised (baseline age, sex, number of diabetes drugs, waist circumference, blood pressure, lipid concentrations, and presence of albuminuria) to associate with diabetes remission after bariatric surgery. In all analyses, p values reported are unadjusted and based on a two-sided test when appropriate.
For R statistical free software see http://www.r-project.org
12 079 participants received letters of invitation
1231 responded and were telephoned
1077 ineligible at telephone interview because of BMI >30 kg/m2, no diabetes, or unwillingness to be randomised
154 assessed by clinician
103 ineligible 28 did not have diabetes 35 had BMI >30 kg/m2 40 unwilling to be randomised
Statistical analysis The power calculation was based on our previous trial4 and assumed that 60% of patients in the laparoscopic adjustable gastric band group and 15% of those in the multidisciplinary care group would achieve diabetes remission. With an allowance of five dropouts in each group, an enrolment of 25 participants in each group afforded 93% power to detect the assumed difference of 45% in diabetes remission between the groups (p
