1. The Cost-Effectiveness of Interventions in Health and Medicine William H. Herman, M.D., M.P.H. University of Michigan

2. Rationale for conducting cost- effectiveness analyses How is cost-effectiveness assessed? What is the cost-effectiveness of diabetes prevention?

3. Barriers to diffusion of new medical treatments Patient Provider System

4. Patient level barriers Demographic (age, gender, race) Socioeconomic position (education, income) Health status (including depression) Self-efficacy Cost

5. Provider level barriers Demographics (age, training, CME, experience) Knowledge of guidelines and critical pathways Attitudes to innovation Opinions of key opinion leaders Peer practices Cost

6. System level barriers Practice structure and organization Information systems Time barriers Cost

7. Why perform CEAs? Resources are limited Choices must be made Choices should consider costs and outcomes

8. Value for Money

9. Essential Elements of Economic Analyses of Health-Care Programs Type of analysis Perspective Type and definition of costs Description and valuation of outcomes Choice of comparator Modeling Discounting Sensitivity analyses

10. Types of Economic Analyses Descriptive cost analysis Cost-benefit Cost-effectiveness Cost-utility

11. Perspective of Economic Analyses Payer Society

12. Type of Costs Direct medical Direct nonmedical Indirect

13. Definition of Direct Medical Costs Cost of intervention Cost of side-effects of intervention Cost of outcomes

14. Description and Valuation of Outcomes Beneficial outcomes produced Adverse outcomes averted

15. Outcomes Clinical Years of life Quality-adjusted life-years

16. QALY Quality-Adjusted Life-Year adjusts length of life for quality of life

17. Quality-Adjusted Life-Year time in health state xquality of life in health state where quality of life = health utility 1.0 = excellent health 0 = death

18. Calculation of QALYs 20 years of life/excellent health 20 x 1.0 = 20 QALYs 20 years of life/10 excellent health 10 with blindness (10 x 1.0) + (10 x 0.51) = 15.1 QALYs

19. Approaches to Measuring Health Utilities Standard gamble Multiattribute utility models Rating scales

20. Multiattribute Utility Models Health Utilities Index (HUI) Quality of Well-Being Index (QWB) EuroQol (EQ-5D)

21. 21

22. 22

23. Choice of Comparator New therapy vs. ? all relevant alternatives? ? usual therapy? ? substandard therapy? ? placebo?

24. Choice of Comparator Failure to compare a new therapy with a strong alternative will result in a deceptively favorable cost- effectiveness picture.

25. Modeling When direct empirical data are not available, methods of imputation and extrapolation are used to estimate outcomes No model generates new data, it merely combines existing information within an explicit framework

26. Discounting Even in a world of zero inflation, there are advantages to receiving benefits earlier and incurring costs later. Discounting adjusts future costs and benefits to current value.

27. Sensitivity Analyses The values of one or more of the key parameters are varied singly or simultaneously to evaluate the robustness of the results to the underlying assumptions.

28. What is the cost- effectiveness of diabetes prevention?

29. Interventions Proven to Delay or Prevent the Development of Type 2 Diabetes Intervention% Risk Reduction Lifestyle (4 trials)29-58% Metformin (2 trials) 26-31% Lifestyle & Metformin (1 trial)28% Acarbose (1 trial) 25% Troglitazone (1 trial)55% Rosiglitazone (1 trial)60%

31. DPP Study Population 3,234 subjects with impaired glucose tolerance (IGT) –Fasting plasma glucose 95 - 125 mg/dl –2 hour plasma glucose 140 - 199 mg/dl Age > 25 years (mean 51 years) BMI > 24 kg/m 2 (mean 34 kg/m 2 ) 68% women 45% minorities

32. DPP Interventions Lifestyle –healthy, low-calorie, low-fat diet & physical activity of moderate intensity (brisk walking for  150 min/week) to achieve and maintain  7% loss of body weight –16 session core curriculum over 6 months then monthly follow-up Metformin –850 mg daily increasing to 850 mg twice daily –standard lifestyle recommendations –quarterly follow-up Placebo –standard lifestyle recommendations

33. Placebo (n=1082) Metformin (n=1073, p<0.001 vs. Placebo) Lifestyle (n=1079, p<0.001 vs. Metformin, p<0.001 vs. Placebo) Incidence of Diabetes Risk reduction 31% by metformin 58% by lifestyle The DPP Research Group, NEJM 346:393-403, 2002

34. Analyses Health system perspective Cost per Quality-Adjusted Life-Year (QALY) Lifetime time horizon Interventions as implemented in the DPP Year 2000 US dollars DPP. Ann Intern Med 142:323, 2005

35. Data Sources Treatment of IGTTreatment of Diabetes CostsDPPCost Model Quality of LifeDPP Quality of Life Model Health OutcomesDPPType 2 Diabetes Model DPP. Ann Intern Med 142:323, 2005

36. Annual Direct Medical Costs in a Man Progressing from IGT to Diabetes with Complications Brandle et al. Diabetes Care 26:2300, 2003.

37. Health Utility Scores in a Man Progressing from IGT to Diabetes with Complications Coffey et al. Diabetes Care 25:2238, 2002.

38. Diabetes Cost-Effectiveness Model Markov model structure Follows a patient cohort from diagnosis of IGT to death IGT transition probabilities based on DPP Diabetes, microvascular and macrovascular transition probabilities based on UKPDS and literature Assumes 10 year interval between DPP onset and UKPDS clinical diagnosis of type 2 diabetes mellitus Tracks costs, QALYs, disease progression, 5 complications, and survival CDC Diabetes Cost-effectiveness Group. JAMA 287:2542, 2002

39. Simulated Cumulative Incidence of Diabetes in the DPP Herman et al. Ann Intern Med 142:323, 2005 20% 8%

40. Simulated Lifetime Clinical Outcomes in the DPP OutcomeLifestyleMetforminPlacebo Diabetes (%)637583 Blindness (%)356 ESRD (%)0.60.81.0 Amputation (%)1.31.61.9 Stroke (%)192121 CHD (%)394142 Life expectancy (yrs)24.724.324.1 Herman et al. Ann Intern Med 142:323, 2005

41. IGT Intervention - Summary Lifetime Outcomes* OutcomeLifestyleMetforminPlacebo Lifetime Costs$51,974$55,261$51,339 Lifetime QALYs10.8910.4510.32  Cost v. Pbo$635$3,922——  QALY v. Pbo0.570.13 ——  Cost/  QALY$1,124$31,286 —— * costs and QALYs discounted at 3% per year Herman et al. Ann Intern Med 142:323, 2005

42. How Attractive Does a New Technology Have to be to Warrant Adoption and Utilization? more costly less costly Increase in QALYsDecrease in QALYs more effective & more costly less effective & less costly less effective & more costly more effective & less costly

43. Distribution of Cost-Effectiveness Ratios for Preventive Measures and Treatments for Existing Conditions Cohen JT. N Engl J Med 2008; 358:661-663

44. Cost-Effectiveness of Selected Interventions in the Medicare Population Intervention Cost-Effectiveness (Cost/QALY) Influenza vaccineCost-saving Beta-blockers after myocardial infarction<$10,000 Mammographic screening$10,000-$25,000 Hypertension medication (DBP >105 mmHg)$10,000-$60,000 Cholesterol management, as secondary prevention $10,000-$50,000 Dialysis for end-stage renal disease$50,000-$100,000 Left ventricular assist devices$500,000-$1.4 million PJ Neumann. N Engl J Med 2005; 353:1516-1522

45. How Attractive Does a New Technology Have to be to Warrant Adoption and Utilization? more costly less costly Increased QALYsDecreased QALYs $100,000/QALY $20,000/QALY

46. Conclusion Interventions for diabetes prevention represent a good value for money in people with IGT.

47. But... An alternative analysis suggested a substantially higher cost per QALY-gained for the lifestyle intervention ($200,000 per QALY- gained). Eddy DM. Ann Intern Med 2005; 143:251-264

48. Purpose To assess the cost-effectiveness of the lifestyle and metformin interventions relative to the placebo intervention with an intent-to-treat analysis spanning the combined 10 years of DPP/DPPOS.

49. Background The DPPOS followed participants for an additional 7 years during which time those in the lifestyle and metformin interventions were encouraged to continue those interventions. During DPPOS, lifestyle participants received extra lifestyle support and all participants were offered a 16 session group lifestyle intervention and 4 healthy lifestyle program sessions per year. A recent intent-to-treat analysis demonstrated a persistent benefit of the lifestyle and metformin interventions on the incidence of type 2 diabetes for at least 10 years after randomization.

50. Cumulative Incidence of Diabetes during DPP/DPPOS DPP Research Group. Lancet. 2009; 374:1677-1686 52% 47% 42% Risk reduction vs Placebo DPP – 3 years Lifestyle 58% Metformin 31% Risk reduction vs Placebo DPP/DPPOS – 10 years Lifestyle 31% Metformin 19% 10-year incidence

51. Methods Data on resource utilization, cost, and quality-of-life were collected prospectively during DPP and DPPOS. To estimate the cost of lifestyle if it had been administered in a group format rather than individually as it was during DPP, we recalculated costs assuming that the core curriculum and monthly follow-up visits with the lifestyle case managers were conducted as group sessions with ten participants Economic analyses were performed from a health system perspective that considered direct medical costs.

52. Cumulative, Undiscounted, Per-participant, Direct Medical Costs of the DPP/DPPOS Interventions by Intervention Group and Year

53. Cumulative, Undiscounted, Per-participant, Direct Medical Costs of Medical Care Received Outside the DPP/DPPOS by Intervention Group and Year

54. Undiscounted, Per-participant, 10-year Direct Costs of Medical Care Received Outside the DPP/DPPOS by Intervention Group and Type Costs ($) by categoryLifestyleMetforminPlacebo Outpatient visits6,8457,1457,325 Inpatient care5,6315,8176,856 ER visits1,9411,6901,825 Urgent care visits1,6971,9451,811 Calls to physicians712742712 Prescription medications6,4906,6196,959 Self monitoring supplies and laboratory tests*1,2481,6281,978 TOTAL24,56325,61527,468 * diabetic participants only

55. Cumulative, Undiscounted, Per-participant, Total Direct Medical Costs of the DPP/DPPOS Interventions and Medical Care Received Outside the DPP/DPPOS by Intervention Group and Year

56. Cumulative, Undiscounted, Per- participant, Quality-Adjusted Life-Years Gained by Intervention Group and Year

57. Incremental Cost-Effectiveness Ratios over 10 Years by Intervention Group – Health System Perspective Differences in costs (  cost ) Lifestyle vs placebo DPP group lifestyle vs placebo Metformin vs placebo Health system perspective 1 Undiscounted 928 -650 -321 Discounted 2 1,226 -323 -159 Differences in QALYs (  QALY) Undiscounted 0.14 0.02 Discounted 0.12 0.02 Incremental cost-effectiveness ratios (  Cost /  QALY) Health system perspective 1 Undiscounted 6,651Cost-saving Discounted 2 10,037Cost-saving 1 Includes total direct medical costs 2 Both costs and QALYs discounted at 3%

58. Summary Over 10 years, from a payer perspective: The lifestyle intervention was cost-effective and the group lifestyle intervention was cost-saving compared to the placebo intervention –The group lifestyle intervention was approximately 1/3 less expensive than the lifestyle intervention –The increased cost of the lifestyle intervention relative to the placebo intervention was largely offset by the reduced costs of non-intervention-related medical care –The lifestyle intervention was associated with better quality-of-life than the placebo intervention The metformin intervention was cost-saving or at least, cost-neutral compared to the placebo intervention –The increased cost of the metformin intervention relative to the placebo intervention was entirely offset by the reduced costs of non-intervention-related medical care

59. Conclusion Health policy and societal policy in the United States should support the funding of intensive lifestyle and metformin interventions for diabetes prevention