1	How Much Benefit is Good Enough? Muriel R. Gillick, MD Department of Ambulatory Care and Prevention Harvard Medical School/Harvard Pilgrim
2	Beyond the Randomized Trial Randomized trial may establish efficacy of treatment Trial data must be translated into policy Policy decisions are value-laden
3	Is the Intervention Good Enough? To make treatment widely available Develop new centers, train physicians To substitute new treatment for existing approach Change formularies; modify guidelines To pay for new intervention Third party payers (Medicare, health plans)
4	Current Decision-Making Process Practice guidelines Food and Drug Administration approval Centers for Medicare and Medicaid process for reimbursing new technology
5	Basis for Policy Decisions Today Utilitarian approach If on average patients who get new rx are better off, make rx available Acceptable to make some people worse off if there is net improvement
6	Values Implicit in Current Approach Some people may be harmed by intervention, but that’s all right as long as treated group benefits on average If there is a risk of an adverse effect, intervention may be withheld from all Cost is not a relevant consideration Resource allocation, justice unimportant
7	Example: CMS Coverage Decisions CMS mandate: provide services that are reasonable and necessary For its 42 million beneficiaries Repeated efforts to define criteria have failed
8	CMS Approach Many decisions made by local carriers Controversial decisions referred to CMS for national decision Manufacturers, professional organizations can request policy decision Requesters submit outcomes data
9	CMS Decision-Making Exhaustive review of scientific evidence by Coverage and Analysis Group Technology assessment May be commissioned by AHRQ Comprehensive review by Medicare Coverage Advisory Committee
10	CMS Issues Decision Memorandum To inform public of decision Specify reasons for decision Outline process followed Summarize evidence considered May specify conditions for use May demand further clinical trials
11	Medicare Coverage Advisory Committee Chartered 1998 Advisory role to CMS Composed of 100 representatives from science, medicine, economics, law Subcommittee chosen for each decision Includes non-voting representatives of public, industry Indicates whether reported benefits translate into improved net health outcomes
12	How CMS Approach Plays Out: Left Ventricular Assist Device Involves surgical implantation of prosthetic pump LVAD accepted rx for patients awaiting heart transplant New indication CMS asked to pay for: definitive rx for older, sicker patients
13	LVAD as “Destination Therapy” 68 patients assigned to device; 61 controls 48% decrease in all-cause mortality with rx Survival rate with device 52% at 1 year Survival rate with device 23% at 2 years Quality of life improved High morbidity: infections (28%), bleeding (42%) Rose et al; NEJM 2001
14	CMS Evaluation Process CMS receives request to reimburse for new indication Medicare Coverage Advisory Committee input (votes 6:1 to approve subject to constraints on eligibility) CMS repeatedly postpones decision CMS agrees to provide coverage CMS sets reimbursement at $70,000
15	Actual Cost of LVAD Device alone: $65,000 Hospitalization total: $200,000 Total cost/year to Medicare program: $350 million (5000 cases) $7 billion (100,000 cases)
16	Beyond Outcomes Data What if we consider cost? What if we factor in values?
17	Suppose Cost is Included in Decision-Making Cost-effectiveness is technique for selecting among competing options when resources are limited Cost-effectiveness ratio= Cost (new)- Cost (current) Effectiveness (new)- Effectiveness (current)
18	Quality-of-Life Considerations QALY=life expectancy x quality of remaining years Single score to evaluate health state based on mobility, pain, anxiety… Assigns utilities to health states Year of perfect health=1 Death=0 States may be worse than death (negative)
19	Standards for C-E Analysis Use societal perspective Include all health effects, costs that flow from intervention Include benefits and harms Use CE = Cost(new)- Cost (old) QALYs (new)-QALYs (old)
20	Incorporating Values into Policy Give preference to patients in poor health Give preference to vulnerable populations Give preference to interventions that are of large magnitude (even if few benefit)
21	What Might LVAD Decision-Making Look Like with Values, Cost? Start with scientific data Use cost-effectiveness threshold Vary C-E threshold depending on special circumstances
22	Cost-Effectiveness of LVAD CE ratio/QALY: best est. $802,700/QALY Low: $500,000/QALY High: $1.4 million/QALY Benchmark $50,000-$100,000/QALY Blue Cross/Blue Shield with Southern Permanente Health Plan and Kaiser Foundation
23	Policy Conclusions Using New Approach: Case of LVAD Start with scientific data; conclude small net benefit with many side-effects; consider rx Does intervention meet cost effectiveness threshold of $100,000/QALY?--NO Should threshold be modified to give preference to patients who are extremely ill?—YES Does intervention meet higher threshold, eg $200,000/QALY?--NO
24	From Observational Studies to Clinical Trials to Public Policy Start with science Add in cost-effectiveness Temper with explicit, previously agreed upon values