Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 33
Assessing and Improving Value in Cancer Care: Workshop Summary 5 Value and the Oncology Market Dr. Jeffrey Lerner of the ECRI Institute (formerly the Emergency Care Research Institute) introduced a panel of expert speakers on pricing for oncology therapies and the economic market in cancer care. The three speakers would discuss the price of medical care, a topic too few discussed in the past, Dr. Lerner said. Instead of thinking about cost and price as secondary issues, perhaps it was time to take them into account through serious science and research, and to count them as part of the value equation. DRUG PRICING AND VALUE IN ONCOLOGY COMPARED TO OTHER AREAS IN MEDICINE Dr. Patricia Danzon of the University of Pennsylvania’s Wharton School began by describing how she and other economists use the term value. To economists, maximizing value from the resources used on all goods and services in the economy, termed economic efficiency, requires that the value gained per dollar spent is equalized across all the ways in which those resources are used. If a bigger “bang for the buck” can be obtained from one use compared to another, there is benefit to be gained by simply transferring resources to the more efficient use, whether in health care versus other economic sectors, oncology versus other medical fields, or certain drugs versus other services within oncology. To achieve this, economists normally rely on incentives to drive efficient resource allocation, but in health care we have to rely on other drivers of value, Dr. Danzon said.
OCR for page 34
Assessing and Improving Value in Cancer Care: Workshop Summary In most markets, the price system, or what Adam Smith called “the invisible hand,” drives efficient allocation of resources. The basic idea is simple: consumers make choices—votes with their dollars for the prices they are willing to pay. Consumer “willingness to pay” conveys to manufacturers the value they place on different resources, and this creates an incentive for manufacturers to align their prices with how much consumers are willing to pay. Market prices assure efficient resource allocation only if consumers are well informed, consumers face the full social costs of services (there are no externalities), and the population distribution of consumer income is appropriate or balanced. In this type of well-functioning price system there would be no need to study cost-effectiveness. With health insurance in place, however, prices are not constrained by consumer willingness to pay. Health insurance exists for a very good reason: to protect us from financial risk that might otherwise be economically catastrophic. In this sense, it is extremely valuable. But insurance undermines patient price sensitivity and leads to “moral hazard” that drives higher volumes and prices of services. Systems of consumer co-payment and cost sharing can mitigate moral hazard, but for many services the cost shares are relatively low as a percent of total cost, and stop-loss measures are in place as an upper limit on a patient’s cost share. Therefore, these mechanisms are limited in their ability to sensitize patients to costs. Less widely studied than moral hazard is the effect of insurance on producer pricing. On the consumer demand curve for medical care with no insurance in place, consumers use fewer services because they pay higher prices. In that situation, a manufacturer would charge a price (P1; see Figure 5-1) based on the intersection of the marginal revenue and marginal cost curves. When consumers are given insurance coverage with a 50 percent coinsurance rate, the manufacturer’s price can be twice as high, as far as the consumer is concerned, because the consumer is only paying half the total price. With this 50 percent coinsurance or co-pay, the manufacturer’s price would increase (P2), and a higher quantity (Q2 compared to Q1) can be sold because consumers pay less of the price. If this is what happens with a 50 percent coinsurance rate, imagine what happens with standard coinsurance products that require consumers to pay a fixed amount or cost percentage on the order of 20 percent or $25. In that situation, the profit-maximizing price would be even higher. To prevent this, most insurance companies include rules and constraints that try to influence this price elasticity of demand—the price sensitivity. From the standpoint of pharmaceutical manufacturers, determining the market price sensitivity can be difficult
OCR for page 35
Assessing and Improving Value in Cancer Care: Workshop Summary FIGURE 5-1 Representative consumer demand curves for medical care with no insurance coverage compared to coverage with 50 percent coinsurance. SOURCE: Danzon presentation, February 9, 2009. because they have essentially three consumers to serve: the patient, the payor, and the physician. Pricing Incentives for Pharmacy- and Physician-Dispensed Drugs Different pricing incentives operate for drugs that are dispensed by physicians, which include most oncology products, compared to those that are dispensed by pharmacies free from any physician stake in dispensing. Pharmacy-dispensed drugs are generally handled through pharmacy benefit managers (PBMs) similar to the Medicare part D prescription drug plans or tiered formularies in private payor systems. In essence, these systems establish drug formularies with the lowest tier and co-pay for generics and higher tiers for branded drugs. By tiering the drugs with differences in co-pays, the PBM has a lever to negotiate discounts with the manufacturer. If a drug is placed on a tier with a lower co-pay, then the manufacturer can expect to get a bigger share of the market, and manufacturers are willing
OCR for page 36
Assessing and Improving Value in Cancer Care: Workshop Summary to give discounts in return for this preferred formulary status. This mechanism of negotiating discounts through the PBM works reasonably well in therapeutic categories containing several drugs that can be substituted for one another and where patients and physicians have no strong preference between them. However, the mechanism works much less well for cancer drugs that cannot easily be substituted for one another and where physicians and patients want greater freedom of choice. In practice, many plans place cancer drugs on a third or fourth tier, and patients are expected to pay a percentage of the cost through coinsurance. Physician-dispensed drugs, including many oncology products, are generally covered by Medicare part B or by private insurance through the patient’s medical benefit. For these drugs, the physician’s reimbursement plays a key role in pricing decisions, and perhaps in dispensing decisions. Before 2005, Medicare reimbursed physicians 95 percent of the Average Wholesale Price (AWP), which is a list price, and pharmaceutical firms discounted this list price to physicians to gain market share while increasing physician margins. In this situation, there was little constraint on the AWP. In 2006, when Medicare began to reimburse based on the Average Sales Price (ASP) plus a 6 percent dispensing fee, no constraint was put on the cost that determined Medicare reimbursement, and the 6 percent fee created perverse incentives to dispense more expensive drugs because these drugs increased physician reimbursement and margin. The absence of a constraint on ASP, combined with the perverse competitive effect of the 6 percent margin, creates an incentive structure that may drive prices up. Given the stop-loss measures in place in Medicare part D and private plans, plus the supplementary coverage (Medigap or Medicaid) to aid patients in paying the 20 percent co-pay for Medicare part B drugs, the degree of constraint on manufacturers’ cancer drug pricing by patient cost sharing is unclear. Medicare part B rules create very little incentive for manufacturers to compete by lowering prices, and the prices set by part B likely spill over into those for part D because manufacturers launching products into both markets tend to make the prices comparable. Overall, this system appears to include few incentives for manufacturers to lower prices, Dr. Danzon said. Comparing Oncology Drug Prices to Other Areas of Medicine: Are They More Expensive? Comparing prices for cancer drugs with those for other diseases requires a common, standardized health outcome measure and consistent
OCR for page 37
Assessing and Improving Value in Cancer Care: Workshop Summary adjustments for relative cost offsets. Cost per quality-adjusted life-year (QALY) gained is the most widely used metric, but cost per QALY is not systematically estimated or published in the United States for medical treatments. Given that this data is not available for drugs in the United States, Dr. Danzon displayed data from an international source, the Canadian Coordinated Drug Review Center (Table 5-1), which is a review body in the Canadian federal government that issues recommendations regarding whether or not drugs should be reimbursed under provincial drug plans based in part on cost-effectiveness. In this dataset the maximum cost per QALY was higher for noncancer drugs, but the median and minimum costs per QALY were higher for the cancer drugs (Table 5-1). For some drugs, data on cost per QALY were not reported. Based on the percentage of these drugs that received a recommendation of “do not list” for reimbursement—60 percent (3 of 5) for cancer and roughly 50 percent (39 of 80) for noncancer—it can tentatively be inferred that a greater percentage of cancer drugs had a higher cost per QALY, either due to higher prices or lower effectiveness, or due to the outright absence of effectiveness data. This is a very small sample and conclusions are therefore very tentative, but taken at face value this provides some evidence, however weak, for higher prices in cancer drugs compared to drugs in other medical fields. Dr. Danzon explained that data on drug costs per QALY are not systematically available in the United States. Without this data on drugs in the TABLE 5-1 Canadian Coordinated Drug Review (CCDR) of Cancer Versus Noncancer Drugs Cancer Treatments Noncancer Treatments Number of drug indications reviewed 10 100 Number of drugs with cost/QALY reported 5 20 Maximum cost/QALY $126,500 $363,516 Mean cost/QALY $73,900 $78,099 Median cost/QALY $71,000 $61,000 Minimum cost/QALY $36,000 $9,225 CCDR recommendation Do not list for reimbursement 2 8 Number of drugs without cost/QALY reported 5 80 CCDR recommendation Do not list for reimbursement 3 39 SOURCES: Danzon presentation, February 9, 2009; Canadian Coordinated Drug Review Center; data from May 2004 through December 2008.
OCR for page 38
Assessing and Improving Value in Cancer Care: Workshop Summary United States, can one be sure the Canadian drug prices given above are reasonably similar to those in the United States? For expensive biologics, including cancer therapies, Dr. Danzon said, price indices for some countries suggest that average prices are similar to or higher than those in the United States. However, the variation in prices across major industrialized countries is not large (Figure 5-2). When the prices are normalized by differences in average per-capita income, again most foreign countries’ prices for biologics are similar to or higher than those in the United States (Figure 5-3). This argues against the idea that we pay disproportionately higher prices in the United States for biologics compared to other countries. However, drugs in categories other than biologics tend to be priced 30 to 40 percent lower in most European countries, Dr. Danzon added. In summary, Dr. Danzon said that data simply does not exist to enable valid comparison of prices for United States cancer drugs relative to prices of drugs for other therapies, and this makes it hard for researchers, payors, patients, and physicians to allocate resources efficiently. Data from the Canadian Coordinated Drug Review Center provides some evidence, however weak, that cancer drugs are higher priced than others. U.S. insurance FIGURE 5-2 Biopharmaceutical drug price indices relative to the United States (100Price Index Relative to U.S. Price percent level), all biologics including cancer drugs. NOTE: Calculations based on IMS Health, Inc., MIDAS data, 2005. * ATC4 = anatomical therapeutic classification, indicates the chemical, therapeutic, and pharmacological subgroup. SOURCES: Danzon presentation, February 9, 2009; Danzon and Furukawa, 2008.
OCR for page 39
Assessing and Improving Value in Cancer Care: Workshop Summary FIGURE 5-3 Biopharmaceutical drug price indices normalized by GDP per capita relative to the United States (100 percent level), all biologics including cancer drugs. NOTE: Calculations based on IMS Health, Inc., MIDAS data, 2005. SOURCES: Danzon presentation, February 9, 2009; Danzon and Furukawa, 2008. reimbursement rules and incentives for physician-dispensed drugs, including many for cancer, may contribute to higher prices. Ultimately, improving the evidence for cost-effectiveness, measured as price per a standardized unit of outcome, will be very important for improving medical care, as well as investment in research and development. INDUSTRY PERSPECTIVE ON PHARMACEUTICAL PRICING IN ONCOLOGY Pharmaceutical manufacturers face many challenges inherent in the valuation of cancer products, said Dr. Greg Rossi of Genentech, and the costs of development of cancer therapies can also be quite high. For Avastin (bevacizumab), the cost associated with the development of this product has been roughly $2.5 billion to generate evidence to understand its potential value. With costs and investments such as these, and the risks of not realizing a return on such investments, valuation of therapies, rewarding innovation, and conceptualizing patient-level and societal-level economics for cancer treatments can all be challenging. In addition, the current payment system, which pays for procedures and products but not for quality or outcomes, may be too blunt an instrument than the complexity of cancer requires. How can a more dynamic, nuanced approach to payment policy be created that enables a greater effect on patient outcomes?
OCR for page 40
Assessing and Improving Value in Cancer Care: Workshop Summary Describing Value from the Manufacturer Perspective There are a large number of factors that influence the manufacturer’s price and value of a therapy, and these factors are differentially weighted. They may be differentially weighted depending on the stakeholder (patient, physician, or payor) or depending upon the region. In the United States cancer pharmaceutical marketplace, there is little price elasticity, so formal price elasticity tends not to be the main factor that Genentech considers to determine the price of a product. What Genentech does weight more heavily are various clinical and economic factors, such as the magnitude of the net health benefits in the initial launch indication, the level of unmet need, whether the drug will be introduced as second- or third-line therapy into patient populations that have limited treatment choices, and the potential value in future indications. In many cases, surrogate endpoint data (e.g., response rate) is often all that is known at the time of initial launch in highly refractory disease settings. Therefore, the clinical benefits known at this time may not reflect the full potential value or benefit of a drug for all of its potential indications; for example, whether the therapy has health benefits as frontline treatment for metastatic disease or as adjuvant therapy for one or more cancer type. Genentech performs a number of economic analyses to assess the impact of the new product introduction from a variety of perspectives (patient, provider, managed care plan, and CMS, among others) to determine its value. These analyses include formal cost-effectiveness analyses, budget-impact analyses (perhaps the most important factor for many private managed care organizations in the United States), and out-of-pocket costs for patients to inform both the assessment of product value and some of the practical aspects of a new product introduction. The factors that influence the value of a treatment vary by stakeholder (Figure 5-4). For regulators such as the FDA, trial data with specific endpoints relative to a highly valid internal control, such as placebo, will be the principal measure of a treatment’s benefit, but costs will not factor into their considerations of risks and benefits. For patients and physicians, out-of-pocket costs, adherence, and trial data are of critical importance, as well as things that are not traditionally measured in registration programs, such as treatment impacts on hope for survival and the level of innovation. In addition, physicians and patients are interested in how trial data for a new treatment compares to standard treatments already used in practice. For commercial payors, net price, cost offsets, and data on the treatment from trials in the real world are especially important (Figure 5-4). How should
OCR for page 41
Assessing and Improving Value in Cancer Care: Workshop Summary FIGURE 5-4 Measures of relative perceived value by stakeholder or user. NOTE: ICER = incremental cost-effectiveness ratio; OOP = out-of-pocket (costs). SOURCE: Rossi presentation, February 9, 2009.
OCR for page 42
Assessing and Improving Value in Cancer Care: Workshop Summary pharmaceutical manufacturers weigh the factors important to these different customers to determine the value of a new treatment at launch? Dr. Rossi made a few comments on costs. The incremental cost-effectiveness ratio (ICER) is not often used to discuss cost in the United States, he said. Furthermore, when physicians are surveyed on the costs of certain therapies, even common ones, there is high variability in the responses given, perhaps because many physicians do not know the precise costs of their therapies. Differences in patient out-of-pocket costs do not necessarily correlate with differences in the actual price of potential drugs to treat a particular condition because of benefit design. These gaps in knowledge of drug prices further complicate communication between physicians and patients around costs. From these complications around costs and the complexity around drug benefits to these various stakeholders, can the true costs and benefits of a treatment be precisely identified to determine its value? For total costs, Dr. Rossi said that this may be possible. However, cost data collected in randomized controlled trials (RCTs) may not be predictive of costs in the real world, though they do allow for reasonable cost approximation at either the patient or societal level. For treatment benefits, standard, agreed-upon methods for adjustment of quantity of life are needed to account for quality of that extended life. Cancer trials have focused on the endpoint of quantity of life gained, but we are seeing that overall survival will be more and more challenging to routinely use as an endpoint in many large randomized phase III trials because of a number of factors, most importantly the complexity of follow-on treatments and crossover trial designs. Figure 5-5 shows the results of an informal analysis of studies from Dr. Rossi’s organization of treatments for breast, colorectal, and lung cancer over the last 10 years. This analysis examines the relationship between progression-free survival and overall survival in these studies. Trials of treatments for metastatic or advanced disease between 1998 and 2008 with at least 100 patients were included in the analysis. Single-arm, single-center, and diagnostic studies were excluded. The 20 largest studies were selected for each of the three cancer types, and the differences in median progression-free survival between the treatment and control arms for each trial were plotted against the median differences in overall survival. The charts show that, compared to control groups in these studies, effects on progression free survival in the intervention groups had a poor predictive value for effects on overall survival in recent breast cancer trials, but had a better predictive value in studies of colon cancer and non-small-cell
OCR for page 43
Assessing and Improving Value in Cancer Care: Workshop Summary FIGURE 5-5 Relationship between differences in median progression-free survival and overall survival in treatment versus control arms from studies of breast, colorectal, and lung cancer treatments. NOTE: Each plot includes the 20 largest studies meeting inclusion and exclusion criteria for each tumor type between 1998 and 2008. Inclusion criteria: trials with progression-free survival and overall survival as endpoints, English language studies between 1998 and 2008, and 100 or more patients with metastatic or advanced disease. Exclusion criteria: single-arm, single-center, and diagnostic studies. NOTE: OS = overall survival; PFS = progression-free survival. Differences in region, year trial was conducted, crossover design, population risk factors, subsequent therapies, and quality of the trials need to be considered but are not captured by the graphs. SOURCES: Rossi presentation, February 9, 2009; Genentech, Inc. study data, 1998–2008.
OCR for page 44
Assessing and Improving Value in Cancer Care: Workshop Summary lung cancer trials (Figure 5-5). A challenge to the research and clinical communities trying to understand value, Dr. Rossi explained, is finding other surrogate endpoints to better approximate definitive clinical endpoints and what outcome heterogeneity exists across the patient subgroups within a certain patient population. Dr. Rossi proposed that there may be different, nonsurvival outcomes worth considering. For first-line therapy, certain outcomes could be characterized based on symptomatology, toxicity or tolerability, convenience, other health-related quality of life, emotional domains, physical function, activities of daily living, productivity, or economic costs. Some studies have successfully characterized such outcomes as these for a particular line of therapy, such as for Tarceva in lung cancer or Sutent in renal cell carcinoma gathered by Pfizer and their clinical collaborators. However, beyond changes in these domains as the disease progresses and treatment changes (i.e., differences in pre- and post-progression health states), what other endpoints can be incorporated to characterize the value of a treatment to patients? The data on endpoints after progression are sparse but deserve greater attention. Patient-reported outcomes are also often excluded from clinical trials in cancer. Michael Woolley and colleagues examined all industry-sponsored studies listed at www.clinicaltrials.gov in the cancer domain for whether or not they included patient-reported primary or secondary endpoints, as well as whether or not patient-reported outcomes and symptoms were included in the specific labels of new cancer therapies (Gondek et al., 2007). Overall, of 2,704 oncology trials listed, only 322 (12 percent) included a patient-reported outcome measure, and only 6 (9 percent) of 70 FDA new or revised labels included patient-reported data. Dr. Rossi saw this as a call to action for patient-reported outcomes, saying, “If we believe them to be valuable, we had better start measuring them, and we had better be able to talk about them.” Dynamic Value and Pricing for Oncology Drugs In the pharmaceutical industry, the dynamic value of a drug depends upon its different value propositions for different lines of therapy in different diseases. Dynamic value is particularly important to consider in cancer therapy because cancer products are built with multiple indications—front line, second line, breast cancer, lung cancer, adjuvant therapy, and so on— each with different implications for a drug’s value. In the case of Herceptin (trastuzumab), the drug was launched into a specific patient population in
OCR for page 45
Assessing and Improving Value in Cancer Care: Workshop Summary 1998 as frontline therapy for metastatic breast cancer. It showed a significant median overall survival improvement of 4.8 months, and its cost per QALY was over $100,000 (Elkin et al., 2004). This was significantly higher than the cost-effectiveness thresholds established in many countries. In the adjuvant phase, trastuzumab has subsequently been shown to have a much lower cost per QALY of $26,400 (Garrison et al., 2007). But the opportunity to test trastuzumab in the adjuvant breast cancer setting was predicated on its approval for use in the metastatic phase. If Genentech were to price trastuzumab according to a cost-effectiveness threshold, one might argue that the price should be discounted significantly for the treatment of metastatic breast cancer and increased substantially for adjuvant treatment. The case of Avastin (bevacizumab) was even more complicated. At the time of launch, Dr. Rossi explained, bevacizumab had failed a phase III trial for treatment of breast cancer, shown significant impact in colorectal cancer, and further studies were underway for treatment of lung cancer. What should the price have been based on? The first indication? The average effect over the life cycle? Weighting for endpoint or disease severity? It was decided that pricing of bevacizumab would be based on its benefits and costs compared to analogues for the same indications, such as Erbitux, irinotecan, and oxaliplatin, which had also recently launched. When it became clear through phase III studies in breast and lung cancer indications that bevacizumab optimally benefited patients at a higher dose in these settings, the Avastin patient-assistance program was put into place so that patients who required more than 10,000 milligrams of the drug could subsequently receive it at no further cost. There have been no price increases since that time. These examples raise important questions. As we think about value-based pricing of pharmaceuticals, how can the complexity of a product’s indications and life cycle be taken into account for initial pricing? How willing are we to pay for innovations, especially for “first-in-class” therapy innovations that can pave the way for other therapies and biosimilars, generate evidence of success in therapeutic approach, further knowledge of appropriate patient populations through biomarker and subgroup identification, and may lead to reduced costs of subsequent therapies in their class through the development of biosimilars? In closing, Dr. Rossi said that he expected better outcomes in cancer care to be achieved through combining targeted therapies in specific patient subgroups. How patient benefit is defined and measured to determine value will be critical to understand and incorporate ahead of time in the development of future therapeutics and companion diagnostics studies. Managing costs while enhancing value will include both pricing-related and nonpricing-related steps. Nonpricing steps could include using com-
OCR for page 46
Assessing and Improving Value in Cancer Care: Workshop Summary panion diagnostics to identify responders to treatment, considering patient-reported outcomes in measures of treatment benefit, integrating economic evaluations into study design, enhancing patient-assistance programs, and generating evidence after launch to understand real-world outcomes. Pricing steps could include expenditure caps and patient risk sharing. Ultimately, we will have to be innovative both in terms of advances in cancer treatments and in terms of oncology payment policy. BUILDING THE EVIDENCE BASE FOR VALUE OF NEW TREATMENTS: COST-EFFECTIVENESS ANALYSES ALONGSIDE CANCER CLINICAL TRIALS Cancer care providers include a wide variety of health care professionals in collaboration, particularly in the area of cancer clinical research. As a clinician and a clinical researcher, Dr. Deborah Schrag of the Dana-Farber Cancer Institute recognized the obligation those in her position had to promote value from a number of different perspectives. Patient needs—survival, hope, trust, compassion, access, and recognition of personhood—had to be reconciled with research obligations to society at large and considerations of cost-effectiveness, cost-utility, efficiency, and equity in the distribution of resources. These sets of obligations are layered onto the unique concerns of oncology providers themselves, including their interest in respect, professionalism, and security. Economic security in particular is often difficult for providers to talk about. For clinical researchers, balancing multiple perspectives and the obligations that accompany them can be challenging. Creating Value and Understanding Costs in Clinical Trials Individual physicians make hundreds of risk and benefit calculations for individual patients on a daily basis, taking into account their knowledge of the literature, their previous experience with other patients, their judgment about the individuals they treat, and the preferences they and their patients hold. With all of these factors to consider, how can medical oncologists systematically determine treatment value with respect to cost in the context of clinical trials? A systematic approach that is relatively little-used in the United States is cost-effectiveness analysis. Cost-effectiveness analyses are comparisons that can take the form of Cost-minimization analysis to determine which treatment costs more: treatment A or treatment B;
OCR for page 47
Assessing and Improving Value in Cancer Care: Workshop Summary Cost-effectiveness analysis with units of life-years gained; or Cost-utility analysis with units of quality-adjusted or quality-discounted life-years, which can also be expressed in terms of the incremental cost-effectiveness ratio (ICER). The latter is the optimal approach, but estimation of the ICER requires that one treatment strategy be compared to an alternative. Sometimes the alternative is no treatment, but more often it is the default treatment standard. Cost-effectiveness analyses are useful when both the efficacy and the cost of a treatment are greater than those of the standard of care. In trials, cost-effectiveness analyses are most needed when the treatment benefit is small but affects a large population, when a new treatment is particularly costly compared to standard treatment, or when there is simply a high degree of uncertainty regarding the economic effect of a new treatment. Typically these analyses are considered in later-stage (i.e., phase III) trials once the probability of a drug reaching market is higher. Given the resources required to estimate cost effectiveness, and that comparison of two alternative strategies is essential, cost-effectiveness analysis does not make sense in the context of phase I or II studies. Ideally, cost-effectiveness analyses are built into the trial with clear specification of data-collection and analytic methods, and data is gathered prospectively. However, preliminary data (e.g., power calculations for cost-effectiveness analyses) to inform data-collection methods may be absent, meaning that post hoc analytic plans are often applied to prospective data. Retrospective data assembly may be used once a treatment has shown clinical efficacy, tallying resource utilization post hoc from either the original study population or a comparable one. Given the complexity and intensive resources required to integrate cost-effectiveness analyses into phase III clinical trials, back-of-the-envelope calculations are often used to identify key cost drivers (see Box 5-1). Such calculations are fast, inexpensive, and potentially misleading, particularly when they rely on assembly of retrospective data that were not collected for this purpose. Often key cost drivers are unavailable. Challenges in Conducting Cost-Effectiveness Analyses in Cancer Trials There are immense challenges in conducting cost-effectiveness analyses in cancer clinical trials. To illustrate this point, Dr. Schrag presented data
OCR for page 48
Assessing and Improving Value in Cancer Care: Workshop Summary BOX 5-1 Back-of-the-Envelope Cost-Effectiveness Analysis: What Is the Incremental Benefit of Adding Erlotinib to Gemcitabine for Pancreatic Cancer? Overall survival benefit 12.8 days Quality-adjusted survival assuming mild erlotinib toxicities—most frequently diarrhea, rash 9.4 days Quality-adjusted survival assuming severe erlotinib toxicity—most frequently diarrhea, rash 8 days Lifetime incremental costs per patient: Costs of erlotinib $10,300 Costs of adverse events $780 Costs of extra survival time $4,100 Total costs $15,200 Costs per life-year (costs of erlotinib plus costs of extra survival time all divided by years of overall survival gained) $410,000 Costs per QALY (adjusted for mild to severe symptoms) $430,000 (mild) to $510,000 (severe) SOURCE: Schrag presentation, February 9, 2009; adapted from Miksad et al., 2007. from a cost-effectiveness analysis, the COST study (Nelson et al., 2004), performed as part of a phase III cooperative group RCT in which the two study arms—laparoscopic-assisted colectomy versus open colectomy to treat resectable colon cancer—showed similar patient quality of life, cancer recurrence rate, survival, and complications (Clinical Outcomes of Surgical Therapy Study Group, 2004; Weeks et al., 2002). With such similar outcomes, it made sense to look at differences in cost, and a cost-minimization analysis was performed to answer the question of whether laparoscopic-assisted colectomy was less expensive than open colectomy. The study took the perspective of a third-party payor, and an intention-to-treat analysis was performed, as would be the case in any clinical trial. The time horizon used has important implications for any cost-effectiveness analysis, and the study investigators chose a 2-month postoperative horizon after making sure that there were no treatment differences in late outcomes. Hospital costs vary geographically and by hospital, so standard cost-effectiveness analysis methodology is to measure
OCR for page 49
Assessing and Improving Value in Cancer Care: Workshop Summary resource utilization, itemize all the services and equipment likely to differ between the study arms, and then convert this to standard unit costs. In this study, patients who had open colectomy stayed in the hospital 1.2 days longer, but the laparoscopy-assisted colectomy patients spent an average of 57 more minutes in the operating room where a greater number of surgical equipment cartridges were used (see Table 5-2). Unit costs were difficult to obtain and varied between academic and community hospitals. At academic centers, a hospital day costs almost 50 percent more than at community hospitals, but the cost of operating room professionals was less. Therefore, the result of the cost comparison between the two arms depends on whether academic or community hospital unit costs apply (see Table 5-3). As a result of running TABLE 5-2 Resources Used by Colectomy Method—Laparoscopic-Assisted vs. Open Resource Category Colectomy Method Laparoscopic-Assisted Open P-Value Mean length of stay, days 5.5 6.7 < .001 Mean operating room time, minutes 166 109 < .001 Equipment cartridges used per patient 3.4 2.5 < .001 SOURCES: Schrag presentation, February 9, 2009; Clinical Outcomes of Surgical Therapy Study Group, 2004. TABLE 5-3 Cost of Resources Used by Colectomy Method, at Academic and Community Hospitals Resource Category Cost (2007 $US) Academic Center Community Hospital Hospital day 1,426 925 Laparoscopic-assisted colectomy: professional component of surgery 1,676 2,105 Open colectomy: professional component of surgery 1,653 2,065 Laparoscopic-assisted colectomy: technical component of surgery plus fixed operating room supplies 3,454 5,472 Open colectomy: technical component of surgery plus fixed operating room supplies 3,204 3,738 SOURCES: Schrag presentation, February 9, 2009; Clinical Outcomes of Surgical Therapy Study Group, 2004.
OCR for page 50
Assessing and Improving Value in Cancer Care: Workshop Summary the operating room longer and with higher personnel costs at the community hospitals, laparoscopic-assisted colectomy cost $2,454 more at community hospitals compared to open colectomy while saving $62 at academic centers (see Table 5-4). What were the lessons learned from this study? Economically, the choice between laparoscopic-assisted and open colectomy consists of a trade-off between higher operative costs and shorter hospital length of stay. Laparoscopic-assisted colectomy is relatively less expensive at institutions with higher rooming but lower operative personnel costs. Taking the true opportunity cost of community surgeons’ time into account, however, this operative approach will actually be more expensive in most settings. Companion cost-effectiveness analyses such as these are challenging. If the primary study fails to meet its clinical endpoint, the cost-effectiveness analysis is rendered moot and uninteresting. For instance, one cost-effectiveness companion to an RCT comparing gemcitabine plus bevacizumab to gemcitabine plus placebo for pancreatic cancer was stopped after 2 years of intensive data collection when bevacizumab was found to provide no additional benefit. Another example, an ongoing companion to a large RCT for colorectal cancer, has been modified multiple times to keep up with study design changes to reflect new first-line therapies and the addition of KRAS biomarker testing for patients on cetuximab. Ultimately, these studies can be useful, but they have all the risks and caveats of TABLE 5-4 Incremental Unit Costs by Colectomy Method at Academic and Community Hospitals Incremental Cost of Laparoscopic-Assisted Colectomy (2007 $US) Unit Costs from Academic Center Unit Costs from Community Hospital Hospital stay cost −1,665 −1,083 Operating room total cost 1,142 3,275 Anesthesia total cost 89 140 Recovery 10 −16 Intensive care unit days 659 333 Reoperation −2 −1 Rehospitalization −293 −189 TOTAL −62 2,454 SOURCES: Schrag presentation, February 9, 2009; data obtained courtesy of Dr. Jane Weeks; Clinical Outcomes of Surgical Therapy Study Group, 2004.
OCR for page 51
Assessing and Improving Value in Cancer Care: Workshop Summary the trials they accompany. It is also exceedingly difficult to secure funding for cost-effectiveness analyses in the United States. Cost-effectiveness data cannot be explicitly considered when coverage decisions are made and this undoubtedly helps to explain the difficulty obtaining funding to support these studies. Patient Concerns Over the Cost of Prescription Drugs Even if oncologists were able to understand a treatment’s cost-effective-ness, would it matter to patients? Or are cancer patients insensitive to treatment costs when making treatment decisions? How much do patients worry about the costs of their treatments? Do they discuss these concerns with providers on their oncology team? Dr. Schrag presented preliminary results from a survey she and her colleagues performed involving 409 patients who were asked these questions. Thirty-nine percent were not worried, 31 percent were a little worried, and only 10 percent were very worried about the costs of treatment. Among those who were very worried about the costs of treatment, 77 percent of them had not discussed these concerns with their doctors. Clearly, most patients are much more worried about their cancer and their symptoms than they are about costs, Dr. Schrag said. But in cases where patients have significant cost concerns, much more discussion is needed with doctors. To summarize, Dr. Schrag listed barriers to integration of cost-effectiveness analyses in evaluation of cancer treatment. These barriers include Substantial data-collection efforts are required to obtain reliable data, Lack of data systems architecture in place to support cost-effectiveness analyses, Reluctance of institutions to share cost data, Investigator suspicion of the validity of cost-effectiveness analyses, Underdeveloped cost-effectiveness analytic methods, Competing study priorities and limited funding, Occasional irrelevance of the cost-effectiveness analyses’ results, Political and regulatory hurdles mean the information cannot be used in regulatory decisions, and Cultural preferences to avoid cost-effectiveness analyses for fear of “rationing.”
OCR for page 52
Assessing and Improving Value in Cancer Care: Workshop Summary Dr. Schrag concluded that a variety of strategies to build the evidence base are needed to increase value in oncology. More and better information is always needed on what works, how clinicians treat, and the consequences. We will need to accept limitations and coverage restrictions that will curb the use of expensive treatment technologies in circumstances where there really is no supporting evidence, she said. On the horizon of a new era of personalized medicine, what is going to be funded—cost-effectiveness analyses or the development of more biomarkers? Can we fund both? DISCUSSION Dr. Lee Newcomer of UnitedHealthcare asked Dr. Rossi why the Avastin patient-assistance program appeared to be underutilized. Dr. Rossi replied that the underuse was caused by the program not being actively promoted by the company. Dr. Bhadrasain Vikram of the National Cancer Institute asked Dr. Rossi what Genentech’s strategy was for collecting real-world outcomes. Dr. Rossi replied that, while RCTs remain the cornerstone of evidence development, Genentech had just begun a disease-based registry in breast cancer to look at patterns of care, health economics, patient-reported outcomes, and clinical outcomes in the real world. Dr. Allen Lichter of the American Society of Clinical Oncology asked Dr. Schrag whether there were ways to anticipate at the outset the study population size needed to definitely achieve a cost-effectiveness trial result—similar to power calculations for trials of clinical benefit. She said that techniques such as adaptive trial design, dynamic trial design, and value-of-information theory could be used to calculate such a threshold ahead of time, but the crucial component is also to understand key cost drivers, which can be difficult before the trial has begun. Dr. Thomas Smith of the Virginia Commonwealth University Massey Cancer Center pointed out two examples of competitor drugs increasing the price of all the drugs in a market after they were introduced—aromatase inhibitors in breast cancer and thalidomide derivatives. How can market forces improve the drug prices that patients pay? Dr. Rossi responded by saying that this was an area for comparative effectiveness research to show true drug equivalency and support direct competition. He also noted markets where the introduction of competitor drugs have reduced prices. Dr. Smith commented from his experience that it was especially hard for patients in safety net health centers to afford their anticancer treatments. Often enrollment in pharmaceutical manufacturers’ indigent care
OCR for page 53
Assessing and Improving Value in Cancer Care: Workshop Summary programs is spectacularly difficult and complex. These barriers prevented many disadvantaged patients from being able to benefit from modern chemotherapeutic drugs they needed. Dr. Jane Perlmutter of the Gemini Group asked why the value of a patient’s time had not been factored in the cost-effectiveness analysis example Dr. Schrag showed in detail. Dr. Schrag acknowledged that this could be an important consideration in some studies and that significant literature exists on ways to collect information from patients on their time costs. Dr. Lerner asked whether the current economic situation had put a greater pressure on patients and doctors to discuss price, since more patients are in economic distress. Dr. Schrag thought that this was the case. Patients do not have employment security anymore, she said, and employers in this economy can be forced to lay off even employees with new cancer diagnoses. Physicians have come to rely on social workers and legal assistance for patients more often and earlier for people in jeopardy. Some patients want to talk about costs very directly, but they are a small minority. Most decide not to think about it. Just like the mortgage market was unsustainable, this avoidance of the cost issue is equally unsustainable. REFERENCES Clinical Outcomes of Surgical Therapy Study Group. 2004. A comparison of laparoscopically assisted and open colectomy for colon cancer. New England Journal of Medicine 350(20):2050–2059. Danzon, P. M., and M. F. Furukawa. 2008. International prices and availability of pharmaceuticals in 2005. Health Affairs 27(1):221–233. Elkin, E. B., M. C. Weinstein, E. P. Winer, K. M. Kunitz, S. J. Schnitt, and J. C. Weeks. 2004. HER-2 testing and trastuzumab therapy for metastatic breast cancer: A cost-effectiveness analysis. Journal of Clinical Oncology 22(5):854–863. Garrison, L. P., D. Lubeck, D. Lalla, V. Paton, A. Dueck, and E. A. Perez. 2007. Cost-effectiveness analysis of trastuzumab in the adjuvant setting for treatment of HER2-positive breast cancer. Cancer 110(3):489–498. Gondek, K., P.-P. Sagnier, K. Gilchrist, and J. M. Woolley. 2007. Current status of patient-reported outcomes in industry-sponsored oncology trials and product labels. Journal of Clinical Oncology 25(32):5087–5093. Miksad, R. A., L. Schnipper, and M. Goldstein. 2007. Does a statistically significant survival benefit of erlotinib plus gemcitabine for advanced pancreatic cancer translate into clinical significance and value? Journal of Clinical Oncology 25(28):4506–4507. Weeks, J. C., H. Nelson, S. Gelber, D. Sargent, G. Schroeder, and for the Clinical Outccomes of Surgical Therapy (COST) Study Group. 2002. Short-term quality-of-life outcomes following laparoscopic-assisted colectomy vs. open colectomy for colon cancer. Journal of the American Medical Association 287(3):321–328.
OCR for page 54
Assessing and Improving Value in Cancer Care: Workshop Summary This page intentionally left blank.