CSPH Works-in-Progress

1. CSPH Works-in-Progress Cost-Effectiveness Analysis of Thromboprophylaxis for the Prevention of Venous Thromboembolism Associated with Major Urologic Cancer Surgery Ye Wang, PhD Center for Surgery and Public Health August, 2014

2. Presentation Overview • Background • PhD Dissertation Project • Current Project at the Center • Q & A

3. Background – Disease Burden in the US • Venous thromboembolism (VTE): • Deep vein thrombosis (DVT) • Pulmonary embolism (PE) • Annual incidence: • > 250,000 clinically evident cases • ≈ 25,000 deaths per year • Annual VTE-associated health care expenses: • $1.9 to 4.2 billion • > $ 5.0 billion in patients with cancers Prevention of post-surgical VTE in patients with cancers 1/3deaths occur in patients undergoing invasive procedures ≤4cm Heit et al., Arch Intern Med 2008 Spyropoulos et al., J Manag Care Pharm 2007

4. Thromboprophylaxis for VTE • Thromboprophylaxis advocated for VTE according to the ACCP: • ACCP = American College of Chest Physicians • Mechanical device • i.e., intermittent pneumatic compression • Pharmacological agents • i.e., injectable anticoagulants ≤4cm Gould et al., Chest 2012

5. Cost-Effectiveness of Thromboprophylaxis ≤4cm Thromboprophylaxis Complications

6. Cost-Effectiveness Analysis The economic, clinical and humanistic outcomes model: Any disease management should aim to achieve balanced outcomes so that gains in one outcome would not sacrifice the opportunity gains in other outcomes and that the overall gains can be maximized and optimized. ≤4cm Gunter, Am J Manag Care 1999

7. Cost-Effectiveness Analysis (continued) Costs (e.g., resource consumption) Consequences (e.g., clinical or humanistic outcomes) Gunter, Am J Manag Care 1999

8. Cost-Effectiveness Analysis (continued) • Four types of cost-effectiveness analysis: • Cost-minimization analysis • Cost-effectiveness analysis • Cost-benefit analysis • Cost-utility analysis • Recommended: • Quality-adjusted life years (QALYs) • Comparisons across various diseases and health interventions Siegel et al., JAMA 1996

9. Incremental Cost-Effectiveness Ratio (ICER) ICER= Non-Societal Costs Societal Costs Benefits Direct Costs Indirect Costs Procedures Hospitalization Follow-up Visits/Tests Complications Lost Wages Lost Productivity Caregiving Physical Health Mental Health $ (CostsIntervention B – CostsIntervention A) (EffectivenessIntervention B – EffectivenessIntervention A) QALYs (calculated by utilities) Siegel et al., JAMA 1996

10. Utility Quality of Life (Utility) Best possible health state 1 0.8 Disability after Hip Fracture 0.6 Severe Congestive Heart Failure 0.4 Symptomatic Metastatic Prostate Cancer 0.2 Above the Knee Amputation 0 Death Worst possible health state

11. QALYs Quality of Life (utilities) 0.5 One QALY 2 years

12. QALYs (continued) Intervention B QALYs Gained Quality of Life (utilities) Intervention A Death Death Time

13. Cost-Effectiveness Analysis Cost-Effectiveness Plane Increased Cost ✔ ? Costs Money Worsens Health Cost Money Improves Health QALYs Gained QALYs Lost Saves Money Worsens Health Saves Money Improves Health ? ✗ Decreased Cost

14. Willingness-to-Pay (WTP) Thresholds Interpreting ICER (US Perspective) Less than $50,000 per QALY gained Good Value $50,000 to $100,000 per QALY gained Sometimes Good Value Greater than $100,000 per QALY gained Rarely Good Value

15. Decision Tree No Event Well Fatal Dead Anticoagulation Embolus Non-Fatal Disabled Fatal Dead Bleed Non-Fatal Disabled Recursive with multiple recurrences Useful to short-term simulations Difficult to assign utilities

16. Markov Modeling Markov States WELL DISABLED DEAD t (Cycle 0) WELL DISABLED DEAD t+1 (Cycle 1)

17. Markov Modeling (continued) Markov States WELL DISABLED DEAD t (Cycle 0) WELL DISABLED DEAD t+1 (Cycle 1)

18. Markov Modeling (continued) Markov States WELL DISABLED DEAD t WELL DISABLED DEAD t+1

19. Markov Modeling (continued) Markov States WELL DISABLED DEAD t WELL DISABLED DEAD t+1

20. Markov Modeling (continued) Markov States WELL DISABLED DEAD Events can recur Simulate over a lifetime horizon Utilities dependent on the cycle length

21. Markov Modeling (continued) Markov States WELL DISABLED DEAD Markovian Assumption “memory-less”

22. PhD Dissertation Project • Part I. Patient-Reported Outcomes of Anticoagulants • Psychometric properties (validation) of a medication • adherence scale • Evaluation of patients’ knowledge, satisfaction, and barriers • to anticoagulant therapy • … • Part II. Pharmacoeconomics of Anticoagulants • Utility evaluation for anticoagulant-related outcomes • Cost-effectiveness of oral anticoagulants for stroke • prevention in patients with atrial fibrillation

23. Study 1 • Utility evaluation for anticoagulant-related outcomes

24. Study Design • Study design: • Cross-sectional patient survey • Sample size: • 100 patients • Inclusion criteria • ≥ 21 years old • Taking warfarin • Able to comprehend English or Chinese • Utility elicitation methods: • Standard gamble technique

25. Health States • Seven long-term health states • Well on warfarin • Well on dabigatran • Well on rivaroxaban • Major ischemic stroke • Minor ischemic stroke • Intracranial hemorrhage (ICH) • Current health state • Four short-term health states • Transient ischemic attack (TIA) • Major extracranial hemorrhage (ECH) • Minor ECH • Myocardial infarction (MI)

26. Health State Descriptions Gage et al., Arch Intern Med 1996 Torrance, J Health Econ 1986 Warrell et al., Oxford Textbook of Medicine 2003

27. Standard Gamble Technique • Methods – SG All health states were considered to be better than death: Choice 1: Staying in the health state under evaluation for the rest of the patient’s life Choice 2: 1 - p p

28. Standard Gamble Technique (continued) All health states were considered to be better than death: p 1 - p

29. Standard Gamble Technique (continued) All health states were considered to be better than death: Indifferent – utility value p 1 - p

30. Results – A Brief Summary • Three best health states (mean ± SD) • Well on rivaroxaban (0.90 ± 0.15) • Well on warfarin (0.86 ± 0.17) • Well on dabigatran (0.83 ± 0.18) • Two health states worse than death (mean ± SD) • ICH (-0.09± 0.51) • Major ischemic stroke (-0.01 ± 0.53) ICH = intracranial hemorrhage; SD = standard deviation.

31. Study 2 • Cost-effectiveness of oral anticoagulants for stroke prevention in patients with atrial fibrillation

32. Methods – Treatment Options • Treatment options: • Dabigatran 150 mg twice daily • Dabigatran 110 mg twice daily • Rivaroxaban once daily • Adjusted-dose warfarin

33. Base case • A hypothetical cohort of patients, who were: • 65 years old • Newly diagnosed with atrial fibrillation • Having no contraindications to anticoagulation

34. Model Information • Model type: • Markov model • Perspective: • The Singapore health care system • Horizon: • Lifetime • Cycle length: • Monthly

35. Model Information • Outcomes: • Direct medical costs • QALYs • ICERs • Willingness-to-pay (WTP) threshold: • Singapore’s 2012 per-capita gross domestic product • (SGD 65,000/QALY) • Software: • TreeAge Pro Suite 2013 (TreeAge Software, Inc., • Williamstown, MA)

36. Model Inputs • Clinical inputs: • Published clinical trials • Utility inputs: • Patient survey • Cost inputs: • Hospital databases

37. Markov Model AF = atrial fibrillation, ECH = extracranial hemorrhage, ICH = intracranial hemorrhage, MI = myocardial infarction, RIND = reversible ischemic neurological deficit, TIA = transient ischemic attack.

38. Markov Model AF = atrial fibrillation, ECH = extracranial hemorrhage, ICH = intracranial hemorrhage, MI = myocardial infarction, RIND = reversible ischemic neurological deficit, TIA = transient ischemic attack.

39. Markov Model (continued) AF = atrial fibrillation, ICH = intracranial hemorrhage, RIND = reversible ischemic neurological deficit.

40. Results – Base-Case Analysis Rivaroxaban versus Warfarin: ICER = SGD 36,231/QALY Dominated Eliminated by extended dominance

41. Results – One-Way Sensitivity Analysis WTP threshold

42. Results – Two-Way Sensitivity Analysis

43. Results – Probabilistic Sensitivity Analysis WTP threshold Rivaroxabanand warfarin were cost-effective in 91.29% and 8.05% of the 10,000 iterations, respectively.

44. Results – A Brief Summary • Base-case analysis • Rivaroxaban was the optimal choice compared to warfarin. • The ICER of dabigatran 150 mg versus warfarin exceeded the WTP threshold. • Dabigatran 110 mg was dominated by warfarin and rivaroxaban. • Probabilistic sensitivity analysis • Using a WTP threshold of SGD 65,000/QALY, rivaroxaban and warfarin were cost-effective in 91.29% and 8.05% of the 10,000 iterations, respectively.

45. Current Project at the CSPH Cost-Effectiveness Analysis of Thromboprophylaxis for the Prevention of Venous Thromboembolism Associated with Major Urologic Cancer Surgery

46. Urologic Cancer in the US ≤4cm National Caner Institute, Surveillance, Epidemiology, and End Results (SEER) Program 2014

47. Effect of VTE in patients with urologic cancer Prevention of post-surgical VTE in patients with urologic cancer ≤4cm Lyman, Cancer 2011

48. Thromboprophylaxis for VTE (continued) Paucity of studies on VTE in the urologic literature The ACCP recommendations for major urologic cancer surgery are extrapolated from General Surgery Ideal use of VTE prophylaxis remains unclear ACCP = American College of Chest Physicians ≤4cm Gould et al., Chest 2012

49. Effectiveness of Thromboprophylaxis for VTE • Study design: • Retrospective data analysis (the Premier) • Major urologic cancer surgery: • Radical prostatectomy • Radical nephrectomy • Partial nephrectomy • Radical cystectomy ≤4cm • Inclusion criteria: • Adults (≥18 years old) • Admitted due to major urologic cancer surgery

50. Research Question Are thromboprophylaxis strategies cost-effective for the prevention of post-surgical VTE in patients with urologic cancer?