From the perspective of prescribing physicians, drug coverage decisions seldom deal adequately with idiosyncratic needs. Prescribers would prefer that all medicines be fully covered, regardless of cost, so that discretion can be exercised on a case-by-case basis for individual patients. From the perspective of drug benefit policy makers, who face rapidly increasing demands with limited resources, open formularies are no longer an option. Policies relating to drug coverage must set limits. The mechanism by which limits are set is critical to balancing access to necessary medications, program sustainability, and the autonomy of both prescribers and consumers. A best practice for drug coverage policy focuses on paying for proven health outcomes.
In this column we describe how decision makers in British Columbia, Canada, allocate the public drug subsidy in accordance with scientific evidence of comparative health benefit. The outcomes-based approach to drug coverage that is used in British Columbia can be characterized as the consistent restriction of the public subsidy until manufacturers provide published scientific evidence of a comparative mortality or morbidity benefit. This approach appears to control expenditures while avoiding the financial inequities and adverse health effects often associated with indiscriminate limits on drug coverage.
Origins of the policy
Outcomes-based coverage was introduced into British Columbia's tax-financed provincial drug benefit program, PharmaCare, during the early 1990s. To address the rapid escalation in drug costs and a dearth of credible information about the comparative value of drug products on which funding decisions could be based, PharmaCare formed a partnership with academic researchers at the University of British Columbia, referred to as the Therapeutics Initiative. Funded through renewable five-year grants from British Columbia's Ministry of Health, the initiative is a university-based group comprising a small staff of paid employees (three faculty and one support staff) and a network of approximately 20 family physicians, specialists, and academic researchers who serve on drug review committees. The initiative produces systematic reviews of evidence submitted by manufacturers seeking coverage under the public drug plans and provides physicians and pharmacists with evidence-based information about optimal drug therapy through newsletters and continuing-education activities.
Working with the Therapeutics Initiative, PharmaCare established a decision-making framework for drug coverage decisions based on the tenets of evidence-based medicine. First, only evidence from published clinical trials is considered, under the belief that data supplied on a commercially confidential basis is inimical to good science. The transparency and accountability of the process would also be served by basing decisions on data readily available to the public, including critics of the decision-making process. Second, evidence should come from trials involving the strongest possible study design (ideally, a blind, randomized controlled trial), an appropriate spectrum of participants and practice settings, and comparators with similar properties and clinical end points. These criteria establish standards for the scientific process. For outcomes, standards are established by a third requirement: that trials measure clinically valid outcomes, such as morbidity and mortality. Greater weight would be placed on evidence of reduction in mortality and morbidity than on evidence of intermediate or "surrogate" effects—for example, changes in cholesterol level, bone density, or blood pressure.
PharmaCare's outcomes-based approach to coverage was operationalized in 1995, when PharmaCare subjected selected drug classes to a reference-based subsidy under its reference drug program. PharmaCare based its public subsidy on low-cost options within each drug class, with generous allowances for variations in dosage. Patients who had clinical requirements for specific products received a full subsidy through a special authority process, and any patient could receive a nonreferenced product by paying the difference between the reference-based subsidy and the cost of his or her preferred brand. Manufacturers that wished to have their drug subsidized at a higher rate than therapeutic alternatives would have to show that their product had a scientifically established health outcome advantage. "Convenience" factors were acknowledged to be of value to consumers—who were free to pay the extra cost for their preferred brand—but, in the absence of scientific data showing that convenience factors produced superior health outcomes, PharmaCare would not rationalize a higher public subsidy than comparable alternatives.
Over time, PharmaCare would apply the outcomes-based approach to determining the subsidy for all new products for which manufacturers sought coverage. In many cases, the decisions were based on clear, relatively uncontested applications of the outcomes-based approach to allocating the subsidy. For example, neuroleptic antipsychotic drugs, such as haloperidol and chlorpromazine, have been available and fully subsidized for many years. Established second-generation neuroleptics, such as clozapine and risperidone, are also fully subsidized. Newer second-generation antipsychotics are therefore judged on the basis of comparative clinical impact and cost. Quetiapine, launched four years later, was also listed as a full-benefit drug, because its cost was similar to that of risperidone. By contrast, olanzapine was listed only for patients who did not respond to one of the fully covered first-line treatments, because its higher cost was not justified by evidence of comparative benefit.
PharmaCare has generally provided a public subsidy for as many drug products as have other provincial drug plans in Canada. However, several funding decisions based on the outcomes-based approach have caused some conflict between the views of health policy makers and clinical practitioners. One such example is the case of donepezil and similar products used to treat Alzheimer's disease. The Therapeutics Initiative's initial review of donepezil, and all subsequent systematic reviews, concluded that its impact on a cognitive scale (two to three points on a 70-point scale) had not been correlated to any clinically apparent patient outcomes or health care costs. Insufficient evidence was found to support claims of differences in disability or functions of daily living, and there was no evidence from randomized controlled trials of delayed institutionalization. Furthermore, patients who received donepezil experienced more adverse events than those who received a placebo and were more likely to withdraw from studies prematurely because of adverse effects. The initiative concluded that the available evidence did not establish that the benefits of treatment outweighed the harmful effects. As a result, British Columbia did not fund donepezil, in contrast with every other Canadian province.
Individual physicians, patients' families, and the Alzheimer Society of British Columbia argued that although average drug effects were small, some patients benefited substantially. They also noted that donepezil and related drugs (also not funded by PharmaCare) were the only drug therapies available for this devastating disease. British Columbia's Ministry of Health met with clinicians and researchers to discuss how a trial could be designed to allow for unbiased individual outcome assessment. Consistent with the evidence-based approach to coverage, plans for an "N-of-1" trial for coverage policy were developed for donepezil-related drugs. With this type of trial, each patient would receive one of the drugs and placebo in a randomized sequence of stages to which patients and prescribers would be blinded. PharmaCare would continue to pay for the medicine provided that, according to preset criteria, a patient did significantly better in the drug treatment stages than in the placebo stages. Thus patients who were shown to benefit from the drug through this trial would continue to receive public subsidy. Despite the appeal of this type of trial for coverage policy, it was never implemented because of lack of support from manufacturers and clinical specialists.
In July 2004, the first long-term double-blind randomized trial to assess the effects on institutionalization rates, disability, and health care costs found that donepezil therapy was associated with mild differences in cognitive scores but no difference in disability or institutionalization rate (
1). PharmaCare has consequently maintained its decision not to fund the drug or its competitors. It is not clear at this point whether manufacturers, specialists, and the policy makers would reconsider the N-of-1 trial policy.
Discussion
Although limitations on drug benefits may appear strict, rapid growth in drug benefit costs has made such limitations a necessity. What makes the approach to allocating scarce drug benefits in British Columbia unique is the consistent allocation of the public subsidy on the basis of evidence of a comparative mortality or morbidity benefit. It may be perceived that government funding biases the expert reviews of evidence on which this allocation mechanism is based. However, were cost cutting the government's primary agenda, evidence-based coverage would not be the policy of choice. PharmaCare could far more easily increase copayments, deductibles, or eligibility requirements to reduce the costs of public drug plans. It chose not to do that, because the government pays not only for drug programs but also for all medical and hospital care: the downstream costs of unduly restrictive drug coverage policies. Furthermore, the decision-making framework and evidentiary standards established through ongoing dialogue between the Therapeutics Initiative and PharmaCare were designed to minimize biases.
Notwithstanding the fact that PharmaCare's decision making has been based on the tenets of evidence-based medicine, even evidence-based processes for allocating subsidies create conflict between physicians who are dealing with individual patients and policy makers who are attempting to maximize the benefits accruing from scarce resources. The presentation of evidence and rationales for funding decisions seldom appeases clinicians who are fighting for access for individual patients. For this reason, British Columbia's PharmaCare policies have been devised with special exemptions and options for patients to use their own funds to purchase products that do not meet the criteria for outcomes-based coverage. It has been the experience in British Columbia that, for a majority of patients, physicians change practice patterns by prescribing the fully subsidized drug options within a drug class or treatment category. Furthermore, knowing that rigorous external assessment was appropriate to a policy based on scientific evidence, PharmaCare has provided data access to academic groups that seek to evaluate its outcomes-based policies. Several such groups have received funding, including research teams based at Harvard, the University of Washington, McMaster University, the University of Toronto, and the University of British Columbia. Most of these researchers have sought to evaluate both the financial and health-related impacts of PharmaCare's coverage decisions. Collectively, these researchers suggest that studied policies have saved PharmaCare approximately 14 percent of program costs without generating the adverse effects often associated with typical cost-cutting strategies, such as indiscriminate copayments and deductibles (
2).