In 2012, 29.1 million people in the United States (approximately 9.3% of the population) were diagnosed with diabetes mellitus, of whom 90% to 95% of adults had type 2 diabetes.1 Although the intensity of glycemic control that is desirable for patients with type 2 diabetes has been a topic of extensive discussion, landmark studies such as the UK Prospective Diabetes Study (UKPDS) and the Action in Diabetes and Vascular Disease (ADVANCE) trial have reported clear benefits of tight glycemic control (mean glycated hemoglobin [HbA1c] <7.0%) in decreasing the risk for diabetes-related complications, such as cardiovascular disease, stroke, and kidney disease.2-4 In the UKPDS, each 1% decrease in HbA1c was associated with a significant reduction in risk for any diabetes-related end point (21% reduction), diabetes-related death (21% reduction), myocardial infarction (14% reduction), and microvascular complications (37% reduction).4 Based on data from studies such as ADVANCE and UKPDS, the American Diabetes Association (ADA) currently recommends an HbA1c goal of <7.0% for most adults with type 2 diabetes.5
In 2012, the ADA estimated that the total economic cost of diagnosed diabetes in the United States was $245 billion, which was 41% higher than the ADA’s estimate of $174 billion in 2007.6 These increasing costs highlight the substantial economic burden that diabetes imposes on patients and on the US healthcare system. One factor is the cost of medications, including prescription medications, insulin, and other antidiabetic agents, which represents 28% of all health expenditures attributed to diabetes.6
To ensure the most effective and efficient use of US healthcare dollars, treatment reimbursement is increasingly being linked to measurable performance indicators or to quality measures.7,8 For type 2 diabetes, HbA1c levels are among the diabetes-related performance indicators that have been incorporated into payer performance measures. Healthcare providers are increasingly required to provide evidence of quality patient care by reporting surrogate markers related to outcomes and quality measures for patients with diabetes, such as blood pressure (ie, the proportion of patients with blood pressure of <140/90 mm Hg) and glycemic control (ie, the proportion of patients achieving a target HbA1c of <7.0%, 8.0%, or <9.0% for those whose HbA1c is poorly controlled).7,8 Value-oriented reimbursement (ie, reimbursement linked to performance indicators and/or to cost reduction) now accounts for 40% of commercial insurance in-network payments, according to the 2014 National Scorecard on Payment Reform.9
The Healthcare Effectiveness Data and Information Set (HEDIS), which was developed by the National Committee for Quality Assurance, is a set of healthcare performance measures that provides benchmarks for quality and value in healthcare.10 HEDIS measures are now used by >90% of US health plans.10 For patients with type 2 diabetes, HbA1c testing was included as a surveillance measure of average glycemic control.7 Several years ago, HbA1c target goals were established for patients with type 2 diabetes, along with a growing support to establish individualized treatment goals. Recent changes to the 2015 HEDIS Comprehensive Diabetes Care measures include the requirement to report the proportion of adults who attain an HbA1c level of <7.0% or <8.0%, based on their age and health status.7
Additional programs, such as the Medicare Shared Savings Program, track the proportion of patients achieving quality measure benchmarks for performance indicators, such as HbA1c and blood pressure.8 The attainment of quality measure goals for parameters such as HbA1c, body mass index, and blood pressure reduction has been associated with significantly lower long-term pharmacy and medical costs in the treatment of patients with diabetes.11,12
Sodium-glucose cotransporter 2 (SGLT2) inhibitors—including dapagliflozin, canagliflozin, and empagliflozin—decrease renal glucose reabsorption, resulting in enhanced urinary glucose excretion and subsequent reductions in plasma glucose and HbA1c concentrations.13 In addition to reducing blood glucose in patients with type 2 diabetes, SGLT2 inhibitors have been shown to have benefits such as body weight and systolic blood pressure reduction.14 The mechanism of action of these agents is independent of insulin.15 The 3 SGLT2 inhibitors can be used in combination with other antihyperglycemic therapies, including (but not limited to) oral antihyperglycemic therapies, such as sulfonylureas, metformin, or insulin analogs.15 An advantage of the unique mode of action of these agents is a low incidence of hypoglycemic episodes when used as monotherapy, although patients receiving SGLT2 inhibitors and background sulfonylurea or insulin are at a higher risk for hypoglycemia compared with patients receiving SGLT2 inhibitor monotherapy.16
The results from placebo-controlled clinical trials of canagliflozin, dapagliflozin, and empagliflozin used either as monotherapy or in combination with 1 (dual therapy) or 2 (triple therapy) other agents demonstrated that all 3 SGLT2 inhibitors significantly reduced HbA1c, body weight, and systolic blood pressure at 24 weeks or 26 weeks.17-19 To date, there have been no head-to-head studies assessing the potential differences in clinical efficacy and safety among these 3 agents in patients with type 2 diabetes. There are, however, published data demonstrating differences in the pharmacodynamic effects of the SGLT2 inhibitors.
A randomized, double-blind, placebo-controlled crossover study evaluated the effects of canagliflozin 300 mg on intestinal glucose absorption.20 The results demonstrated that when administered to healthy participants before a meal, canagliflozin 300 mg reduced postprandial glucose excursion via increased urinary glucose excretion, which resulted from renal SGLT2 inhibition and transient intestinal SGLT1 inhibition.20 The first study to directly compare the pharmacodynamic effects of canagliflozin and dapagliflozin at their highest approved therapeutic doses was a randomized, double-blind, crossover study performed in healthy volunteers that assessed urinary glucose excretion, renal threshold for glucose, and postprandial glucose excursion.21
Pharmacy and Therapeutics (P&T) committees evaluate and identify medications, such as SGLT2 inhibitors, that are most medically appropriate and cost-effective for formulary placement to best serve the health system’s patient population.22 An organization’s formulary system utilizes evidence-based processes to select and use medications that offer the best therapeutic outcomes, while minimizing the risks and costs for patients. The use of a simple, short-term cost per response analysis for the dose of each of the 3 drugs studied (ie, canagliflozin, dapagliflozin, or empagliflozin) to estimate the drug cost per placebo-adjusted 1% reduction in HbA1c levels in patients with type 2 diabetes can be a useful decision-making tool for managed care payers.22
In the absence of head-to-head trials, practical economic evaluations of drug therapies, such as cost per response studies, can be useful for evaluating the cost-effectiveness of treatments, and they have been performed in a number of therapeutic areas.23,24 Such an approach was adopted for the present analysis, using the cost per placebo-adjusted 1% reduction in HbA1c from baseline for each of the 3 SGLT2 inhibitors that were marketed in the United States at the time of this study.
Methods
Study Measures
This study modeled the costs of single SGLT2 inhibitor treatments, which were calculated as the wholesale acquisition costs (WACs) per unit dose of each SGLT2 inhibitor multiplied by the number of doses received for 26 weeks. The costs reported are single drug costs (ie, only costs associated with 1 of the 3 agents discussed), and do not take into consideration the costs that are associated with background treatments (ie, metformin and/or sulfonylureas), because the regimens used in the trials that were evaluated were reasonably similar (Table).
The primary analysis was based on the cost per response net of placebo group differences. The WACs per unit dose were obtained from the RED BOOK Online25 and were the same—$11.43—for canagliflozin (100 mg or 300 mg), dapagliflozin (5 mg or 10 mg), and empagliflozin (10 mg or 25 mg). The outcome measure used in the analysis was the mean reduction in HbA1c levels from baseline, which is a standard primary end point of antihyperglycemic drug trials.
Response data were extracted from the pivotal registration studies discussed in the US Food and Drug Administration (FDA)-approved prescribing information for each of the 3 SGLT2 inhibitors, which had similar designs and the same primary end points.26-28 Because there are no head-to-head studies of all 3 agents, data from the pivotal registration trials were used, because these trials had similar methodologies. The monotherapy trials for all 3 agents compared single-drug treatment with a placebo, the dual-therapy trials compared the study drug plus metformin versus metformin alone, and the triple-therapy trials compared the study drug plus metformin versus a sulfonylurea with metformin and a sulfonylurea alone.
In addition, we opted for a simple approach of assessing cost per response to SGLT2 inhibitors, because overcomplexity is generally not warranted for a high-level pharmacy formulary evaluation. Our study included the placebo-controlled trials of each of the 3 agents as monotherapy and dual therapy (combined with metformin), which are listed in the prescribing information for each drug.17-19,26-31 Triple-therapy trials (ie, SGLT2 inhibitor combined with metformin and a sulfonylurea) were also included for canagliflozin, empagliflozin, and for dapagliflozin at the higher dose of 10 mg.27,28,32-34 The length of treatment applied was 26 weeks, based on the comparable length of time for the primary end points of clinical trials that were used as input for the analysis.
Clinical response data (ie, placebo-adjusted 1% reduction in HbA1c levels from baseline values) that were used as input for the analysis are listed in the Table.17-19,26-34 The trial study populations included in the analysis differed slightly in different respects, such as mean age and duration of diabetes. Most notably, the trials for empagliflozin differed from the other 2 agents because they included a larger proportion of Asian patients; however, overall, the studies of the 3 agents present largely comparable evaluations of the individual drugs versus placebo, particularly when stratified by the number of background treatments (Table).
Data Analysis
The average cost per response, which was defined as the cost per placebo-adjusted 1% reduction in HbA1c levels, was calculated with an Excel-based tool using the following equation:
WAC per unit dose × number of doses weekly × Difference in HbA1c reduction from placebo (adjusted mean) |
The recommended dose for each agent is once daily; therefore, the number of weekly doses was 7 across all 3 agents.
Calculations were performed for each of the 2 FDA-approved doses for each drug and for the 3 treatment regimens (monotherapy, dual therapy, and triple therapy) that were evaluated per drug. The cost per response was also calculated for the absolute reduction of HbA1c that was unadjusted for the placebo effect (see Appendix at the end of this article).
Results
The results of the costs per response analysis by agent, dose, and regimen type at 26 weeks are shown in the Figure.
For canagliflozin 100 mg, the costs per placebo-adjusted 1% reduction in HbA1c were $2286, $3355, and $2930, respectively, for monotherapy, dual therapy, and triple therapy (Figure, Part A). For canagliflozin 300 mg, the costs were $1793, $2702, and $2261, respectively, for monotherapy, dual therapy, and triple therapy.
For dapagliflozin 5 mg, the costs per placebo-adjusted 1% reduction in HbA1c were $4161 and $5201, respectively, for monotherapy and dual therapy (Figure, Part B). For dapagliflozin 10 mg, the costs were $2972, $4161, and $3015, respectively, for monotherapy, dual therapy, and triple therapy.
For empagliflozin 10 mg, the costs per placebo-adjusted 1% reduction in HbA1c were $2972 for monotherapy and $3467 for dual and triple therapies (Figure, Part C). For empagliflozin 25 mg, the costs were $2311 for monotherapy and $3467 for dual and triple therapies. The placebo-adjusted changes in HbA1c levels for the dual and triple therapies of empagliflozin (both doses) were exactly the same, and thus resulted in identical cost per response values.
In addition, the results of the analysis that was conducted on absolute change in HbA1c levels and was unadjusted for the placebo effect are reported in the Appendix at the end of this article.
Discussion
Our analysis presents the single drug costs per placebo-adjusted 1% reduction in HbA1c levels with multiple doses and treatment regimens for each of the 3 SGLT2 inhibitors (ie, canagliflozin, dapagliflozin, and empagliflozin) that are available in the United States. For monotherapy, dual-therapy, and triple-therapy regimens with canagliflozin 300 mg, the drug costs per placebo-adjusted 1% reduction in HbA1c levels were $1793, $2702, and $2261, respectively, which were the lowest costs among the 3 agents. These results are based on the WAC per unit dose that was the same across all 3 agents and doses, at $11.43. The results also indicated that the point estimate for the HbA1c reductions, as reported in previously published placebo-controlled clinical trials provided in the prescribing information of each drug, was largest after treatment with canagliflozin 300 mg.35,36 Of note, the efficacy data assumptions used herein come from different placebo-controlled, randomized, controlled registration trials rather than from direct head-to-head comparative studies of each drug against the other.
The cost per response analysis offers a simple, transparent, and easily understandable method for calculating the drug costs per patient, and attains a quality indicator that can be tracked by organizations measuring performance in healthcare. The cost per response is a well-established method for demonstrating the drug costs that are associated with a treatment in a variety of therapeutic areas.23,24,37,38 Our analysis uses a universally accepted primary end point (ie, the measure of mean reduction in 1% HbA1c from baseline)—which was also the primary end point for the clinical trials evaluated here—as the denominator used to calculate the cost per response.
Because the healthcare costs associated with type 2 diabetes are increasing, it is important to consider the drug costs per patient and the clinical end point of reduction in HbA1c when making patient treatment choices. The prescription drug costs (ie, WAC) of the 3 agents in our study were identical, yet our analysis demonstrated differences in cost per response among these agents when placebo-adjusted 1% reduction in HbA1c levels from baseline values was used as the end point at 26 weeks.
Limitations
One major limitation of our study is the lack of availability of published real-world evidence of the effectiveness and cost-effectiveness of SGLT2 inhibitors; we, therefore, extrapolated efficacy data from individual placebo-controlled trials as measures of effectiveness. Furthermore, the efficacy assumptions used in our evaluation were obtained from each drug’s respective package insert and reflected the outcomes observed in the corresponding landmark clinical trials. The baseline population characteristics, designs, and subsequent analyses of these trials were often similar when stratified by treatment regimens (ie, monotherapy, dual therapy, or triple therapy; Table).17-19,26-34
However, despite the general similarities, differences were noted across the trials in some of the patient baseline demographic characteristics, notably in race (trials of empagliflozin enrolled relatively more Asian patients). These differences affect our ability to reliably assess the treatments across the trials. Therefore, the results of the present analysis should be interpreted with due caution.
Nevertheless, our reliance on the primary analysis of the results net of placebo effect is designed to mitigate some of these potential differences across the trials.
An additional limitation is the assumption regarding the time points analyzed (24 weeks vs 26 weeks). The empagliflozin and dapagliflozin trials used for efficacy input reported the results at 24 weeks, and these results were conservatively assumed to be maintained at 26 weeks; the canagliflozin trial results were reported at 26 weeks. In addition, the results of our study are based on point estimates that did not account for uncertainty.
Some of these limitations can be addressed, in part, by a formal but more complex Bayesian network meta-analysis, including the labeled therapies of interest. Such an analysis was beyond the scope of the present analysis, given the goal of providing a transparent evaluation to high-level pharmacy formulary decision makers in the United States. Such analyses will be addressed in the future, alongside analyses that will focus on additional end points, such as the proportion of patients achieving their HbA1c goals, or the 52-week data when additional clinical trial data become available for the 3 SGLT2 inhibitors in monotherapy, dual-therapy, and triple-therapy regimens.
Finally, the analysis pertaining to SGLT2 inhibitor monotherapy was provided for the sake of completeness, because it is not expected that these agents are routinely used as monotherapies.
Conclusion
The total US economic cost of diabetes is increasing, and drug cost accounted for 28% of all diabetes-related health costs in 2012. P&T committees assess clinical efficacy and cost-effectiveness when evaluating drugs (eg, SGLT2 inhibitors) for formulary addition. A simple, transparent, and increasingly used method for estimating the differences in drug costs and the treatment effects between drugs that have not been directly compared in clinical trials is cost per response. Our analysis may provide additional information for formulary committees that are considering the inclusion of SGLT2 inhibitors in their formularies for the treatment of patients with type 2 diabetes mellitus.
Acknowledgments
Writing support was provided by Excerpta Medica. Conceptual support was contributed by Marcia F.T. Rupnow, PhD. Editorial support was provided by Bilge Yoruk, PhD, and John Andrade, MSc, of Excerpta Medica.
Funding Source
This analysis was funded by Janssen Scientific Affairs, LLC.
Author Disclosure Statement
Dr Lopez and Dr Macomson own stock in Johnson & Johnson. Mr Ektare and Dr Patel are consultants to Janssen. Mr Botteman is a shareholder of Pharmerit International and a consultant to Janssen.
Dr Lopez is Associate Director, Translational Science Team, Health Economics and Outcomes Research, Janssen Scientific Affairs, Raritan, NJ; Dr Macomson is Associate Director, Translational Science Team, Health Economics and Outcomes Research, Janssen Scientific Affairs, Raritan, NJ; Mr Ektare is Lead Outcomes Research Analyst, Pharmerit International, Bethesda, MD; Dr Patel is Director of Health Economics and Outcomes Research, Pharmerit International, Bethesda, MD; Mr Botteman is Partner, Pharmerit International, Bethesda, MD.
References
1. Centers for Disease Control and Prevention. National diabetes statistics report: estimates of diabetes and its burden in the United States, 2014. 2014. www.cdc.gov/diabetes/pubs/statsreport14/national-diabetes-report-web.pdf. Accessed September 11, 2014.
2. Patel A, MacMahon S, Chalmers J, et al; for the ADVANCE Collaborative Group. Intensive blood glucose control and vascular outcomes in patients with type 2 diabetes. N Engl J Med. 2008;358:2560-2572.
3. Holman RR, Paul SK, Bethel MA, et al. 10-year follow-up of intensive glucose control in type 2 diabetes. N Engl J Med. 2008;359:1577-1589.
4. Stratton IM, Adler AI, Neil HAW, et al. Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes (UKPDS 35): prospective observational study. BMJ. 2000;321:405-412.
5. American Diabetes Association. Glycemic targets. Diabetes Care. 2015;38(suppl 1):S33-S40.
6. American Diabetes Association. Economic costs of diabetes in the U.S. in 2012. Diabetes Care. 2013;36:1033-1046. Erratum in: Diabetes Care. 2013;36:1797.
7. National Committee for Quality Assurance. Proposed changes to existing measure for HEDIS1 2015: Comprehensive Diabetes Care (CDC). 2014. www.ncqa.org/Portals/0/HomePage/CDC.pdf. Accessed December 1, 2014.
8. Centers for Medicare & Medicaid Services. Medicare Shared Savings Program quality measure benchmarks for the 2015 reporting year. www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/sharedsavingsprogram/Downloads/MSSP-QM-Benchmarks-2015.pdf. Accessed August 17, 2015.
9. Catalyst for Payment Reform. National Scorecard on Payment Reform. 2014. www.catalyzepaymentreform.org/images/documents/nationalscorecard2014.pdf. Accessed November 27, 2014.
10. National Committee for Quality Assurance. HEDIS & performance measurement: measuring performance. www.ncqa.org/tabid/59/Default.aspx. Accessed August 17, 2015.
11. Lafeuille M-H, Grittner AM, Gravel J, et al. Quality measure attainment in patients with type 2 diabetes mellitus. Am J Manag Care. 2014;20(1 suppl):S5-S15.
12. Stephens JM, Botteman MF, Hay JW. Economic impact of antidiabetic medications and glycemic control on managed care organizations: a review of the literature. J Manag Care Pharm. 2006;12:130-142.
13. Gerich JE. Role of the kidney in normal glucose homeostasis and in the hyperglycaemia of diabetes mellitus: therapeutic implications. Diabet Med. 2010;27:136-142.
14. DeFronzo RA, Davidson JA, Del Prato S. The role of the kidneys in glucose homeostasis: a new path towards normalizing glycaemia. Diabetes Obes Metab. 2012;14:5-14.
15. Berhan A, Barker A. Sodium glucose co-transport 2 inhibitors in the treatment of type 2 diabetes mellitus: a meta-analysis of randomized double-blind controlled trials. BMC Endocr Disord. 2013;13:58.
16. Vasilakou D, Karagiannis T, Athanasiadou E, et al. Sodium–glucose cotransporter 2 inhibitors for type 2 diabetes: a systematic review and meta-analysis. Ann Intern Med. 2013;159:262-274.
17. Ferrannini E, Ramos SJ, Salsali A, et al. Dapagliflozin monotherapy in type 2 diabetic patients with inadequate glycemic control by diet and exercise: a randomized, double-blind, placebo-controlled, phase 3 trial. Diabetes Care. 2010;33:2217-2224.
18. Roden M, Weng J, Eilbracht J, et al; for the EMPA-REG MONO Trial Investigators. Empagliflozin monotherapy with sitagliptin as an active comparator in patients with type 2 diabetes: a randomised, double-blind, placebo-controlled, phase 3 trial. Lancet Diabetes Endocrinol. 2013;1:208-219.
19. Stenlöf K, Cefalu WT, Kim K-A, et al. Efficacy and safety of canagliflozin monotherapy in subjects with type 2 diabetes mellitus inadequately controlled with diet and exercise. Diabetes Obes Metab. 2013;15:372-382.
20. Polidori D, Sha S, Mudaliar S, et al. Canagliflozin lowers postprandial glucose and insulin by delaying intestinal glucose absorption in addition to increasing urinary glucose excretion: results of a randomized, placebo-controlled study. Diabetes Care. 2013;36:2154-2161.
21. Sha S, Polidori D, Farrell K, et al. Pharmacodynamic differences between canagliflozin and dapagliflozin: results of a randomized, double-blind, crossover study. Diabetes Obes Metab. 2015;17:188-197.
22. Tyler LS, Cole SW, May JR, et al; for the ASHP Expert Panel on Formulary Management. ASHP Guidelines on the Pharmacy and Therapeutics Committee and the Formulary System. Am J Health Syst Pharm. 2008;65:1272-1283.
23. Belozeroff V, Lee A, Tseng S, et al. Cost per responder analysis in patients with secondary hyperparathyroidism on dialysis treated with cinacalcet. J Med Econ. 2013;16:1154-1162.
24. Liu Y, Wu EQ, Bensimon AG, et al. Cost per responder associated with biologic therapies for Crohn’s disease, psoriasis, and rheumatoid arthritis. Adv Ther. 2012;29:620-634.
25. RED BOOK Online. Truven Health Analytics; 2015. http://sites.truvenhealth.com/redbook/online/. Accessed February 8, 2015.
26. Farxiga (dapagliflozin) tablets [prescribing information]. Wilmington, DE: AstraZeneca Pharmaceuticals LP; March 2015.
27. Invokana (canagliflozin) tablets [prescribing information]. Titusville, NJ: Janssen Pharmaceuticals, Inc; March 2015.
28. Jardiance (empagliflozin) tablets [prescribing information]. Ridgefield, CT: Boehringer Ingelheim Pharmaceuticals, Inc; June 2015.
29. Bailey CJ, Gross JL, Pieters A, et al. Effect of dapagliflozin in patients with type 2 diabetes who have inadequate glycaemic control with metformin: a randomised, double-blind, placebo-controlled trial. Lancet. 2010;375:2223-2233.
30. Häring H-U, Merker L, Seewaldt-Becker E, et al; for the EMPA-REG MET Trial Investigators. Empagliflozin as add-on to metformin in patients with type 2 diabetes: a 24-week, randomized, double-blind, placebo-controlled trial. Diabetes Care. 2014;37:1650-1659.
31. Lavalle-González FJ, Januszewicz A, Davidson J, et al. Efficacy and safety of canagliflozin compared with placebo and sitagliptin in patients with type 2 diabetes on background metformin monotherapy: a randomised trial. Diabetologia. 2013;56:2582-2592.
32. Häring H-U, Merker L, Seewaldt-Becker E, et al; for the EMPA-REG METSU Trial Investigators. Empagliflozin as add-on to metformin plus sulfonylurea in patients with type 2 diabetes: a 24-week, randomized, double-blind, placebo-controlled trial. Diabetes Care. 2013;36:3396-3404.
33. Matthaei S, Bowering K, Rohwedder K, et al; for the Study 05 Group. Dapagliflozin improves glycemic control and reduces body weight as add-on therapy to metformin plus sulfonylurea: a 24-week randomized, double-blind clinical trial. Diabetes Care. 2015;38:365-372.
34. Wilding JPH, Charpentier G, Hollander P, et al. Efficacy and safety of canagliflozin in patients with type 2 diabetes mellitus inadequately controlled with metformin and sulphonylurea: a randomised trial. Int J Clin Pract. 2013;67:1267-1282.
35. Pacou M, Taieb V, Abrams KR, et al. Bayesian network meta-analysis to assess relative efficacy and safety of canagliflozin in patients with type 2 diabetes mellitus (T2DM) inadequately controlled with metformin. Value Health. 2013;16. Abstract PRM193.
36. Pacou M, Taieb V, Abrams KR, et al. Bayesian network meta-analysis to assess the relative efficacy and safety of canagliflozin in patients with type 2 diabetes mellitus (T2DM) inadequately controlled on metformin and sulphonylurea (MET+SU). Value Health. 2013;16. Abstract PDB2.
37. Kirson NY, Rao S, Birnbaum HG, et al. Matching-adjusted indirect comparison of adalimumab vs etanercept and infliximab for the treatment of psoriatic arthritis. J Med Econ. 2013;16:479-489.
38. Schechter CB, Cohen HW, Shmukler C, Walker EA. Intervention costs and cost-effectiveness of a successful telephonic intervention to promote diabetes control. Diabetes Care. 2012;35:2156-2160.