Health economic evaluations and their role in health care decision making. Lieven Annemans Ghent University Lieven.annemans@ugent.be December 204 2 Health expenditure is recognised as growthfriendly expenditure. Cost-effective and efficient health expenditure can increase the quantity and the productivity of labour by increasing healthy life expectancy.. The relatively large share of healthcare spending in total government expenditure, combined with the need for budgetary consolidation across the EU, requires more efficiency and cost-effectiveness to ensure the sustainability of current health system models. Investing in Health, EC, 203 3 What does it mean for innovative technologies/medicines? We need to stimulate and make available innovative technologies/medicines that offer an added therapeutic benefit at an acceptable cost ( are cost-effective), and fill unmet medical needs pricing, reimbursement, recommending usage,. more and more based on health economic evaluations - OECD 2003 - Report of the Belgian EU Presidency, adopted by the EU Council of Ministers of Health in Dec 200 4
Total cost The Fourth Hurdle Cost-effectiveness: basic principles Threshold Intervention Current approach EFFECTIVE SAFE COST-EFFECTIVE PRODUCED IN GOOD QUALITY (GMP) Health effect (life years, QALY) 5 Annemans L. Health economics for non-health economists, academia press 2008 6 QALY = Quality Adjusted Life Years Perfect health INDEX ( utility level ) Most applied method in Europe: EQ5D 5 dimensions recalculated to score Perfect health Example: gain in QALY by avoiding progression/complications INDEX ( utility level ) 0.6 0.5 +2 +5 0.5 0 2.5 Death 0 20 25 TIME Death 0 TIME Annemans L. Health economics for non-health economists, academia press 2008 7 8 2
Example HCV Health-state Utility Source Comments Baseline non-cirrhotic 0.74 Wright et al., 2006 32 Average of mild & moderate utilities assuming 77% mild & 23% moderate EQ-5D Baseline compensated cirrhosis SVR (utility increment) + 0.05 0.55 Wright et al., 2006 32 Wright et al., 200632 EQ-5D Decompensated cirrhosis 0.45 Wright et al., 2006 32 EQ-5D Hepatocellular carcinoma 0.45 Wright et al., 2006 32 EQ-5D Liver transplant 0.45 Wright et al., 2006 32 EQ-5D Post-liver transplant 0.67 Wright et al., 2006 32 EQ-5D PROBLEM: where is the threshold? = what is the value of a QALY?. HISTORICAL +/- 50,000 per QALY: = ICER (incremental cost-effectiveness ratio) of renal dialysis (+/- 200,000 for +/- 4 QALYs) 2. Measure the societal willingness to pay directly (Desaigues et al, 2008): 40,000 3. WHO: GDP per capita (e.g. Belgium = 34000) 4. Threshold is at the discretion of the decision maker, e.g. UK 20,000 per QALY. 0 An eye-opener in 2007: bevacizumab in metastatic colorectal cancer Incremental Cost-Effectiveness Ratio (ICER ) = difference in cost / difference in QALY = 5,654.0 / 0.8 (rounding!) not value for money International examples: league table Treatment Cardiac rehabilitation and prevention program ICER = Cost/QALY gained ( ) dominant Smoking cessation physician counseling 2,000 Statins in secondary prevention vs no statins 3,000 Total Hip replacement vs do nothing 4,000 New HIV treatments vs old 6,000 Latest drugs in multiple sclerosis vs interferon 35,000 Kidney dialysis vs no dialysis 50,000 Annual mammography for women aged 60-69 vs biannual. 70,000 Exercise ECG for all asymptomatic men aged 40 yrs 0,000 Annual CT scan of former heavy smokers to detect lung cancer,00,000 UK HTA report, 2007, Tappenden et al, 2007 Tufts university database plus different sources 2 3
The methodology: decision trees example Chronic diseases: Markov models Cost A (Standard of Care) = 000 Cost Medicine B = 2000 Cost of failure = 0000* 0.7 x 000 + 0.3 x 000 Lifetime projection Time is cut in cycles of 3 months (yr and 2) and cycles of yr (as from yr 3) * Needs separate study to obtain data CALCULTATE EXPECTED RESULT SoC A Success 0.700 000 Treat disease X 4000 Failure 0.300 000 Success New B 0.900 2000 3000 Failure 0.00 2000 3 4 Cost of failure example HCV Validity requirements for models Health state Annual costs Cost year Source Decompensated cirrhosis 6,387 2009 KCE report 57a HCC 8,528 2009 KCE report 57a Liver transplant 93,728 2009 KCE report 57a Post-liver transplant (as from yr 2) 7,047 2009 KCE report 57a Face validity: right model based on clinical experience Technical verification: check for errors Outcome validity: compare results with observed data Convergent validity: compare with other models Predictive validity: prospective tests (can only be done later) PLUS: need for sensitivity analysis! 5 6 4
Uncertainty analysis. Example pifn+rbv+sofx2w vs. pifn+rbvx48w in GT (basecase = 26000 /QALY) Three key problems Uncertainty Budget impact (the 5th hurdle) Medical need San Miguel R, Gimeno-Ballester V, Blázquez A, et al. Gut 204 7 8 The typical Dilemma at Submission INDUSTRY Problem : potential value Give us more evidence that your drug is value for money Allow us first to the market (reimburse/recommend the drug) and then we will be able to show real life evidence PAYER POSSIBLE SOLUTION Accept models/predictions upon submission for reimbursement Re-evaluation later, based on harder evidence More need for registries Risk sharing agreements Velcade in multiple myeloma (UK) Actonel in osteoporosis (USA) 9 20 5
BUT: problems with such schemes! Confounders (e.g. taking other drugs) Case-mix Exceptions Objective measurement of indicator? Workload for physicians Administrative workload Lack of transparency PROBLEM 2. Budget impact Effectiveness and cost-effectiveness are not enough; it must also be affordable Cfr the cornflakes example (Birch and Gafni, 2006) The economic and equity rationale for carrying out budget impact analyses is opportunity cost, or benefits forgone, measured in terms of utility or equitable distribution, by using resources in one way rather than another Need for estimates at population level! Cohen et al (2008) Adamski et al. BMC Health Serv Res. 200 Jun 7;0:53. Coulton et al. 202 2 22 Problems with budget impact! There are only recent guidelines regarding the good conduct of a budget impact analysis. There is no clear reference as to what is an acceptable budgetary impact. Need for health objectives and anticipated budget needs to reach those objectives Problem 3. A QALY is not a QALY. Example: which to choose? Treatment X 0.4 0.2? 0.9 0.7 Treatment Y 0 Disease A Disease B 23 E. Nord. Person trade off method 24 6
Health status Social reference point! (Scitovsky) Belgium: 4 criteria maximal Striving above SRP Pleasure seeking Not necessary No funding Striving towards SRP Necessity depends on severity Accept higher cost/qaly in worst conditions Societal reference point = Acceptable level of health. Added therapeutic value 2. Place of the drug in medical practice (therapeutic and social needs) 3. Cost-effectiveness (cost health insurance / therapeutic value) 4. Budget impact minimal 25 26 Recommendations. Maximize patient s health within the societal limits. Avoid cost myopia (recommendation to hospital managers) Avoid cost ignorance 2. Reimbursement decisions to be based on a balanced assessment of different criteria 3. Promote and monitor the cost-effective use of all technologies (right population, right way) 4. Higher need for follow up data & registries to show real world evidence 5. Change the financial incentives in the system to encourage quality and cost-effectiveness 6. Need for training in health economics 27 Health economic evaluations and their role in health care decision making. December 204 THANK YOU Lieven Annemans Ghent University Lieven.annemans@ugent.be 7