Cost-Efficacy of Surgically Induced Weight Loss for the Management of Type 2 Diabetes: A randomized controlled trial

A health sector perspective was adopted for the economic evaluation, comprising direct health care costs to government, private insurers, and patients. An incremental cost-effectiveness analysis was undertaken, wherein the net costs and net efficacy of surgical therapy as compared with conventional therapy were calculated and expressed as an incremental cost-effectiveness ratio (ICER). The ICER was expressed as the cost per case of type 2 diabetes remitted.

A 2-year RCT involving 60 obese participants (BMI >30 and <40 kg/m²) was conducted in Australia. It compared surgically induced loss of weight with conventional medical therapy (weight loss and behavior change) for the management of recently diagnosed (<2 years) type 2 diabetes. All patients had previously attempted to lose weight and were involved in a 3-month run-in period where a consulting endocrinologist recommended changes to optimize current diabetes management. Patients were then randomly assigned to the (unblinded) interventions. There were no statistically significant differences in demographics or values contributing to study outcomes between the two groups at baseline.

Conventional therapy reflected the best available medical management. It comprised consultation with a general physician, dietitian, nurse, or diabetes educator at least once every 6 weeks; medical therapies as determined by an experienced endocrinologist specializing in diabetes; and lifestyle modification programs to reduce energy intake and encourage physical activity. In addition to all aspects of conventional therapy, surgical therapy involved the placement of a laparoscopic adjustable gastric band (LAGB) (LAP-BAND System; Allergan Health, Irvine, CA) by an experienced surgeon and associated overnight hospital stay. Patient progress was reviewed by a member of the bariatric surgical team every 4–6 weeks, and adjustments to band volume were made using standard clinical criteria.

One surgical patient withdrew from the trial before surgery, and four medical patients withdrew during the first four months of the trial. Analysis was performed on an intention-to-treat basis. At the end of the 2-year trial period, remission of diabetes (defined as fasting glucose <126 mg/dl and A1C <6.2% while taking no glycemic therapy) was achieved by 22 of 30 (73%) surgical patients and 4 of 30 (13%) conventional therapy patients. Full details of the trial and efficacy results have been published previously (14).

Trial intervention costs were included as follows: LAGB surgery, mitigation of surgical complications, outpatient medical consultations (surgeon/physician [including for lap-band adjustments], general practitioner, endocrinologist, respiratory physician, and dietitian), pathology, medical investigations, weight loss assistance therapies (sibutramine and optifast), and medication (diabetes, antihypertensives, lipids, and other).

LAGB surgery resource use was documented on patient case record forms by hospital clinical staff. Procedures to mitigate surgical complications were recorded in patient case files. Clinical staff recorded dates and descriptions of all outpatient medical consultations, pathology tests, and medical investigations into a trial database. A research nurse interviewed patients every 6 months during the trial to document utilization of weight loss assistance therapies and medication by type, dosage, and frequency. Six-month medication point estimates were assumed to be representative of medication use during the previous 6-month period. Actual pharmaceutical expenditure data held by the Australian government was accessed for all consenting patients (23 surgical and 17 medical) to validate medication costs.

LAGB surgery costs were sourced from a private hospital and private medical specialists to reflect the private provision of surgery, as is most common for LAGB in Australia. Lap-band prosthesis and weight loss assistance therapies were valued based on commercial prices. Unit costs for all other intervention resources were obtained from pricing schedules of the Australian government including the Medicare Benefits Schedule (MBS) (15) and Pharmaceutical Benefits Schedule (PBS) (16). Unit costs were applied to the resource measurements to obtain total costs per intervention arm of the study over the trial period. All costs are reported in 2006 AUD (Table 1). Costs sourced from alternative years were inflated to 2006 values by applying the relevant Australian Institute of Health and Welfare health price deflators (17). The midyear 2006 currency exchange rate was 1 AUD to 0.74 USD according to XE currency exchange (http://www.xe.com).

Diabetes-related outcomes were defined as in the RCT as the number of cases of diabetes remitted at 2-year follow-up. Costs reflect the interventions as implemented in the trial. Analysis of costs and outcomes was performed over a 2-year time horizon and on an intention-to-treat basis. Intervention costs were calculated (for patients who withdrew from the study) up to the point of exit and most recently collected diabetes status was assumed for the remainder of the trial. No discounting was applied to costs and outcomes. All calculations were performed using exact values. Results in the abstract and text were rounded to the nearest 100 AUD, and results in tables were rounded to the nearest 1 AUD.

Table 2 summarizes the mean resource use per patient by the intervention group. A mean resource-use value <1 indicates that the cost was not applicable to all patients (for example, only 4 of 30 patients experienced postoperative complications). Pathology (16 tests) and medication (83 categories) were measured in detail but excluded from the table due to extensive detail. Pharmaceutical expenditure data held by the Australian government for 23 surgical and 16 medical patients were very similar to the self-reported data for the complete samples utilized in the analysis.

The total intervention cost per group of 30 patients was 401,500 AUD for surgical therapy and 101,900 AUD for medical therapy. Table 2 shows the mean costs per patient by resource category and intervention group. Surgical costs are based on mean hospital admission and operating theater times of 1.28 days and 54 min, respectively. Surgical therapy was significantly more resource intensive than conventional therapy, costing an additional 10,000 AUD per patient over 2 years. LAGB surgery accounted for 85% of this difference. Mean outpatient medical consultation costs were three times higher among surgical patients due to a greater number of consultations (mean number of consultations: 25 for surgical patients and 15 for conventional therapy patients) and higher consultation fees due to lap-band adjustments. Mean medication costs were 1.5 times higher for conventional patients, primarily due to greater use of diabetes medication (mean medication cost: 900 AUD per surgical patient and 1,400 AUD per conventional therapy patient).

During the first 6 months of the trial, mean intervention costs per patient were approximately sevenfold greater for surgical patients (10,500 AUD) than for conventional therapy patients (1,400 AUD). The difference between costs in each intervention group decreased with each subsequent 6-month period until the last 6 months of the trial, when intervention costs were equivalent in both groups (Fig. 1). Over time, mean 6-month medication costs per surgical patient decreased by more than one-half (Fig. 2).

Relative to conventional therapy, surgical therapy facilitated remission of an additional 18 cases of type 2 diabetes at an additional cost of 299,600 AUD, equivalent to (an additional) 16,600 AUD per (additional) case of diabetes remitted.

This economic evaluation of options for managing type 2 diabetes in obese patients was based directly on data drawn from an RCT. The data are therefore less prone to the types of bias and uncertainty that are likely to affect a modeled study. The results suggest that for a decision maker interested in the immediate future (a time horizon of 2 years) an additional 16,600 AUD of direct health care investment is required to remit an additional case of recently diagnosed type 2 diabetes through surgically induced weight loss.

The cost per case of diabetes remitted is not a metric commonly utilized by decision makers. The comparable ICER for conventional therapy relative to no intervention (no costs and no benefits) over the same time horizon is 25,500 AUD per case of diabetes remitted. Therefore, the ICER for surgical therapy (relative to conventional therapy, 16,600 USD) is lower than the comparable figure for conventional therapy (relative to no intervention, 25,500 USD). This suggests that within a 2-year time horizon the surgical therapy program is below the currently accepted cost-effectiveness threshold.

The analysis provides an estimate of just one benefit: the remission of type 2 diabetes. We have not attempted to capture health care savings associated with the reduction of other obesity-related morbidities or the improvements of quality of life associated with weight loss. Furthermore, the outcome measure (cases of diabetes remitted) does not capture the substantial improvements in glycemic control demonstrated by the surgical patients who did not achieve diabetes remission. Therefore, results underestimate the cost-efficacy of surgical therapy.

The key strength of this analysis (the use of observed, within-trial cost and efficacy data) is also one of its limitations. It restricts the analysis to a 2-year time horizon, which may limit the relevance of results to government policy makers interested in long-term intervention costs, health care system savings due to disease prevention, quality-of-life gains, and mortality gains. Our next study uses this cost-efficacy analysis as the basis to model these variables and reports the lifetime cost per quality-adjusted life-year for surgical therapy relative to conventional therapy trial populations. Modeling cost-effectiveness over a longer time period is particularly important for this study because of the likely differential pathways in medium-term weight loss and associated diabetes remission across intervention groups (18,19) and because our analysis indicated that the cost differential between the intervention and control group decreased with time.

The generalizability of results to other populations may be limited due to different intervention effects and complication rates (for example, due to variable expertise of surgical teams) or health care costs (which may vary across countries). As discussed in an editorial of the Journal of the American Medical Association regarding the original RCT (20), the trial was not powered to detect low-probability events including postoperative mortality (21). However, there is evidence to show that the trial results (zero operative mortality) have been replicated under nontrial conditions (22). Data are not available regarding the average resource use of LAGB patients in other parts of the world. The trend toward shorter stays under outpatient LAGB in the U.S. suggests that the Australian costs may be relatively high. In conclusion, from an economic perspective, surgically induced weight loss appears to be a cost-effective intervention for managing recently diagnosed type 2 diabetes in obese patients.

This study was funded by Monash University, Melbourne, Australia, with support from Allergan Health. Allergan Health provides the university with an unrestricted research grant for conducting obesity-related research. A.P. is funded by a VicHealth Research Fellowship.

J.B.D. is a paid consultant for Allergan Inc, the manufacturers of the Lap Band, which was used in this study; he has received research grant support from Allergan Inc. P.O. is director of the Centre for Obesity Research and Education, which receives an unrestricted research grant from Allergan. No other potential conflicts of interest relevant to this article were reported.

We thank Margaret Anderson, who assisted in the retrieval of medical resource use data from the trial database, and Katie Williams for assistance with data extraction from medical records.