Published Online: September 07, 2009
Sayeed Ikramuddin, MD; C. David Klingman, PhD; Therese Swan, BA; Michael E. Minshall, MPH
Objective: To assess the cost-effectiveness of Roux-en-Y gastric bypass for treating type 2 diabetes mellitus (T2DM) in the United States compared with standard medical management, using clinical data from a prospective observational study conducted at an academic medical center.
Study Design: Our study used a predictive health economic model (the CORE Diabetes Model) to project the long-term costs and clinical effectiveness of Roux-en-Y gastric bypass as a treatment for T2DM using the prospective observational study as the basis for our clinical effectiveness assumptions.
Methods: The CORE Diabetes Model used Monte Carlo simulation with tracker variables to estimate the lifetime costs and clinical outcomes of Rouxen-Y gastric bypass compared with standard medical management of obese T2DM patients. Sensitivity analyses were performed on key clinical assumptions, discount rates, and shorter time horizons.
Results: The base-case scenario yielded an incremental cost-effectiveness ratio (ICER) of $21,973 per quality-adjusted life-year (QALY) gained. In sensitivity analyses, shortening the time horizon to 5 and 10 years and excluding the negative impact of increased body mass index on the patient’s quality of life had the greatest adverse impact on the ICERs (ie, higher cost per QALY).
Conclusions: Under base-case assumptions, Roux-en-Y gastric bypass is cost-effective in the treatment of T2DM in the United States with an ICER below $50,000 per QALY gained. Sensitivity analyses indicated that bariatric surgery is not cost-effective over shorter time horizons, or if the negative quality-of-life impact of increased body mass index is ignored.
(Am J Manag Care. 2009;15(9):607-615)
A predictive health economic model (the CORE Diabetes Model) was used to project the longterm costs and clinical effectiveness of bariatric surgery as a treatment for type 2 diabetes mellitus.
- Our study demonstrated that compared with medical management, the Roux-en-Y gastric bypass procedure is cost-effective under very conservative assumptions for procedure costs and complication rates/costs, using procedure-related and follow-up data from the University of Minnesota.
- Results for the analysis were robust under most scenarios observed, with the exception of shorter time horizons or when the quality-of-life impact of losing weight was removed.
Obesity is a major independent risk factor for type 2 diabetes mellitus (T2DM), an effect thought to occur through increased insulin resistance of cells.1 Nearly one-third of the nonpregnant adult population (age ≥20 years) in the United States is obese (defined as having a body mass index [BMI] ≥30 kg/m2),2 and about one-quarter of obese persons have T2DM.3 Conversely, about 8% of the US population has T2DM,4 and more than half of the patients with diabetes (both type 1 and type 2) are obese.5 Together, these 2 conditions impose an enormous burden on patients, caregivers, and society as a whole. The total cost of obesity has been estimated at $155 billion, including $60 billion in direct medical costs (adjusted to 2007 price levels from an estimate for 1998)6 and $95 billion in indirect costs (lost productivity, again adjusted to 2007 price levels from an estimate for 1995).7 Similarly, the total cost of T2DM in 2007 was an estimated $174 billion, including $116 billion in direct medical costs and $58 billion in indirect costs.4
In recent years, bariatric surgery—mainly gastric bypass, usually the Roux-en-Y type, but also adjustable gastric banding, and vertical banded gastroplasty—has increasingly become recognized as a highly effective alternative for achieving major weight reduction for obese patients.8,9 Moreover, recent studies have demonstrated that patients with diabetes who have undergone bariatric surgery also experienced major reductions in blood glucose levels, some sufficient to suggest remission of diabetes mellitus.8-10 However, bariatric surgery is not without its own risks, including some potentially serious complications,11 and the procedure itself is expensive12; consequently, candidates for bariatric surgery must be carefully screened. Practice guidelines suggest that bariatric surgery should be considered only for patients with a BMI ≥40 kg/m2 (35 kg/m2 if the patient has major comorbidities such as T2DM) after failure of a 1-year course of well-conducted medical treatment.13,14 Previous studies have demonstrated the cost-effectiveness of bariatric surgery, but most have included nondiabetic obese patients in their study populations.15-17 Our study used an established health economic model and prospective observational data from an academic medical center in the United States to assess the long-term costeffectiveness and clinical outcomes of bariatric surgery compared with standard medical management of T2DM.
The CORE Diabetes Model
The CORE Diabetes Model (CDM) has been described and validated in 2 separate publications and is consistent with recently published American Diabetes Association modeling guidelines and principles.18-20 The CDM is designed to predict the development and progression of type 1 or type 2 diabetes over long time horizons (≥5 years) using the best published clinical and epidemiologic data available. The model has a standard Markov structure, combined with Monte Carlo simulation and tracker variables, which allows for the development and progression of multiple complications within an individual patient, while at the same time overcoming the memory-free properties of basic Markov models. The CDM design includes 16 submodels that simulate diabetes-related complications and nonspecific mortality.19 Transition probabilities and risk adjustments for the CDM were derived from published resources and have been detailed in previous reports.19 The CDM was validated through 66 separate analyses, including both second-order and third-order validation exercises and is fully detailed elsewhere.20
Effectiveness of Bariatric Surgery
Data on the effectiveness of bariatric surgery were drawn from a prospective observational study conducted at an academic medical center in the United States (Minnesota cohort; unpublished data, University of Minnesota Medical Center, Minneapolis). From January 2001 through May 2007, 2223 consecutive patients underwent Roux-en-Y gastric bypass surgery, of whom 567 (25.5%) had previously been diagnosed with either T2DM or prediabetes. Several clinical end points were measured at varying times before and after surgery (minimum duration of follow-up was 11 months, maximum was 80 months). These included BMI, glycosylated hemoglobin (A1C), lipid parameters (total cholesterol, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, and triglycerides), systolic blood pressure, and use of medications for diabetes (both oral and insulin), hypertension (angiotensin-converting enzyme inhibitors and angiotensin receptor blockers), and dyslipidemia (statins), as well as aspirin for prevention of cardiovascular disease (CVD). Of the 567 bariatric surgery patients with T2DM, 204 (36.0%) had complete data on all end points needed for simulation of future clinical and economic outcomes (unpublished data, University of Minnesota Medical Center). We performed additional analyses for those with follow-up data available (n = 204) compared with those Roux-en-Y gastric bypass patients without end-point follow-up data (n = 363). Compared with the patients who had follow-up data available, the patients without follow-up end-point data (n = 363) were younger (46.6 vs 50.1 years, P <.001), had higher baseline low-density lipoprotein cholesterol (107.2 vs 96.9 mg/dL, P <.001), and had a shorter preoperative duration of T2DM (72.5 vs 104.5 months, P <.001) (unpublished data, University of Minnesota Medical Center). The sex, preoperative BMI and weight, A1C, and systolic blood pressure were similar between the 2 groups of patients with T2DM or prediabetes who had Rouxen-Y gastric bypass (unpublished data, University of Minnesota Medical Center).
The Minnesota cohort demographic and baseline clinical characteristics, plus other data used in the standard “US medical management of diabetes” arm of the CDM (drawn from the literature on diabetes and obesity, as summarized elsewhere19), were used in both the bariatric surgery cohort and the medical management cohort (Table 1). Changes from before and after surgery on the clinical measures listed above constituted the impact of surgery in the bariatric surgery cohort (eAppendix A available at www.ajmc.com). Changes in the control cohort over time on those same measures were based on standard algorithms derived from the literature on diabetes and its complications, and summarized elsewhere.19 For modeling purposes, we made no assumption with respect to weight gain after Roux-en-Y gastric bypass because of lack of data and assumed this value to be zero. We assumed that A1C and systolic blood pressure values would increase based on values obtained from the United Kingdom Prospective Diabetes Study (UKPDS),19 while lipid values would increase along the same curve observed in the Framingham Heart Study.19
Patients in the bariatric surgery observational study were relatively young (mean age 50.1 years, with only 11.8% over age 60 years), predominantly female (77.9%), and Caucasian (83.5%) (Table 1). On average, they had been diagnosed with T2DM or prediabetes for 104 months (8.7 years) and were followed for a mean of 27.6 months (2.3 years). Their baseline BMI scores were high (mean 48.4, with only 12.8% having a BMI <40), but their baseline scores on other key end points were not as high as those of patients in other studies of T2DM (mean A1C of 7.7%, mean systolic blood pressure of 137.2 mm Hg, and mean low-density lipoprotein cholesterol of 96.9 mg/dL).
Adverse Effects of Bariatric Surgery and Complications of Diabetes Although some data on the incidence of various adverse effects of bariatric surgery were available in these observational data, more comprehensive data were available from previously published studies of bariatric surgery (eAppendix B available at www.ajmc.com).11,29 Although the patient populations in those studies included nondiabetic obese patients, the complications of bariatric surgery may not differ substantially between diabetic and nondiabetic patients. Costs and Utilities for Bariatric Surgery and Medical Management of Diabetes The analysis assumed a third-party payer perspective. Direct medical costs (Table 2) and health-state utilities (eAppendix C available at www.ajmc.com) for estimating quality-adjusted life-years (QALYs) for events and states associated with medical management of diabetes in the CDM were drawn from T2DM studies previously detailed elsewhere.19 The increase in overall utility for decreasing BMI associated with bariatric surgery for T2DM patients came from the Cost of Diabetes in Europe–type 2 (CODE-2) study.39 The coefficient for each unit decrease in BMI increases the overall utility score by 0.003813 and was based on the CODE-2 study from Europe.39
PDF is available on the last page.