Assessing and Improving Value in Cancer Care: Workshop Summary (2009)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

6 Value in Oncology Practice: Oncologist and Health Insurer Perspectives ONCOLOGISTS’ PERCEPTION OF VALUE Dr. Peter Neumann of Tufts Medical Center explained that his group at the Center for Evaluation of Value and Risk in Health at the Tufts Insti- tute for Clinical Research and Health Policy Studies was interested in how people used—or perhaps did not use—cost-effectiveness information, how it is communicated, and how it is perceived. The cost-effectiveness litera- ture contains numerous studies in oncology. In 2006, roughly 260 studies that examined cost per quality-adjusted life-year (QALY) were published in the cost-utility literature, and about 30 had applications in oncology (CEVR, 2009). In the literature containing cost-per-life-year studies and similar endpoints outside of oncology, the number is orders of magnitude larger. Many of these studies are in leading medical journals that are widely read by physicians. This raises a question: who is using all of these data, and to what extent are oncologists influenced by costs in their treatment recommendations? To answer this question, Dr. Neumann, along with Eric Nadler and Benjamin Eckert (Nadler et al., 2006), published a study of 90 oncologists from two leading academic medical centers who responded to a series of survey prompts regarding their views on costs and cost-effectiveness in cancer treatment. The study found that 77.5 percent of oncologists studied agreed that every patient should have access to effective cancer treatments regardless of their costs. Far fewer, only 30 percent, agreed that the costs of 55 

56 assessing and improving value in cancer care new cancer drugs currently influence their decisions. Eighty-one percent agreed that patient out-of-pocket therapy costs influence their decisions, and a similar majority agreed that the costs of new cancer drugs would impose greater rationing in oncology in the next 5 years (Nadler et al., 2006). These answers seem to reveal a somewhat conflicted relationship with treatment costs among these oncologists, reflecting some of the chal- lenges that oncologists currently face. The same study posed a series of hypothetical scenarios to the oncolo- gists surveyed. They were first asked to “Imagine a new cancer medication for treatment of metastatic lung cancer that on average costs $70,000 more than standard of care. At what minimum improvement in overall survival would you prescribe the new medication instead of the standard of care?” The distribution of responses is shown in Figure 6-1. Even though most surveyed oncologists agreed that costs should not affect care decisions, a majority required a minimum 2–4 months added survival to warrant the $70,000 expense. But there were a few at the extremes, such as one oncolo- gist who would prescribe the medication for just one day of added survival. From the average minimum survival benefit required by these oncologists, Neumann and Nadler derived a mean cost-effectiveness threshold implied by the oncologists’ responses—$318,000 per life-year gained (Nadler et al., 2006). The authors concluded that oncologists’ cost-effectiveness threshold $70,000 50% Proportion of Oncologist Respondents 40% 30% 20% 10% 0% 1 day 1 month 2–4 4–6 6–9 9–12 1 year + months months months months FIGURE 6-1  Distribution of oncologist responses: hypothetical survival benefit needed Minimum Overall Survival Benefit to justify a treatment expense of $70,000. SOURCES: Neumann presentation, February 9, 2009; Nadler et al., 2006. R01506 Figure 4-1 vector, fully editable 

Value in Oncology Practice 57 for new cancer therapies seem to be high, at least compared to conventional health care cost-effectiveness thresholds. At the same time, oncologists were not sure whether expensive new therapies offer good value. After selecting their minimum survival benefit for an incremental cost of $70,000, the oncologists were asked “Do you think that reflects good value?” Most said it did not. Dr. Neumann explained that he and his colleagues are completing a f ­ ollow-up study to determine the extent to which oncologists nationwide believe that costs influence their prescribing behavior, whether they discuss costs with patients, and how they feel about various reimbursement policies. Dr. Neumann concluded that costs seem to influence oncologists’ treatment recommendations, oncologists do not seem to communicate frequently about costs with their patients, oncologists seem to favor cost- effectiveness information while wanting to remain in charge of their own decisions. There also appears to be some enthusiasm among oncologists for price controls, but most did not favor more cost sharing on the part of patients. Clearly, costs are a central concern in health care today and there is a need for better communication between physicians and patients. PAYING FOR NEW CANCER TREATMENTS: RIGHTS AND RESPONSIBILITIES OF HEALTH INSURERS Ever since Benjamin Franklin opened the first fire insurance company, explained Dr. Newcomer, the responsibility of the insurer has been to spread risk among a group of people and save them from catastrophic financial distress. Payors in health care do the same thing: they pool funds among groups of people so that, when an individual member gets sick, he or she can pay to treat the illness. Insurers also have a responsibility to maintain a capital pool that has fiscal integrity. During the 1980s, insurers gained a third responsibility—that of negotiating on behalf of their members to find reasonable prices for therapies considered effective. However, the entire health industry has failed in this third responsibility, with premiums for a California family of four increasing from only 15 percent of U.S. minimum wage earnings in 1970 to 101 percent in 2005 (GAO, 1975). The system  Figures reflect monthly Federal Employees Health Benefits (FEHBP) total premiums for the government-wide Blue Cross/Blue Shield options for non-postal workers and mini- mum wage earnings for full time work of 173.33 hours per month (2,080 hours per year/12) in California. 

58 assessing and improving value in cancer care is broken, and we are going to need to do something fairly radical to change it, Dr. Newcomer said. Health insurers, or payors, look at value very differently than one would for FDA studies or phase III randomized controlled trials (RCTs), as these trials’ results are limited to a very tightly defined group of patients in controlled treatments. These trials are not the real world. In the real world, there is what Dr. Newcomer called “the cascade of chaos” during the dissemination of any new therapy to the nation’s community health care providers. First, the treatment is approved and promptly used by providers, often with many errors because they lack experience with the treatment. Eventually the errors subside as providers become more familiar with the treatment; they then begin tweaking the indications, and this ultimately leads to large indication extrapolation and off-label usage. The introduc- tion of trastuzumab (Herceptin) for HER2-positive breast cancer was no different. Reddy and colleagues showed that inexperienced community labs misclassified 15 percent of those told they overexpressed the HER2 gene when they actually underexpressed the gene and had no chance of responding to trastuzumab. Conversely, 10 percent of those the community labs said underexpressed the gene actually were overexpressers who should have been treated but were not (Reddy et al., 2006). To make matters worse, when oncologists were asked to document HER2 overexpression in patients receiving trastuzumab, 8 percent of patients being treated showed no evidence of overexpression, and 4 percent of patients on trastuzumab treatment had no genetic test results at all. Combining the laboratory errors and physician misinterpretation of results, the study found over one-third of patients were receiving the wrong treatment due to inexperience. These error rates were caused in large part by inexperience reading and reporting the test results. Before long, oncologists began tweaking the indications for trastuzumab, bending the limits to treat patients whose overexpression tests stained at “2+” strength, rather than strictly those whose result was a “3+” stain. Finally, physicians began to move to off-label uses, such as continuing trastuzumab treatment through multiple relapses and metastases, despite no evidence of efficacy to support these uses (Pusztai and Esteva, 2006). The cascade of chaos has clearly lowered the value of trastuzumab and other therapies like it in the real world. Inconsistency in Cancer Treatment Inconsistency in cancer treatments also reduces value in cancer care even for well-established therapies. Dr. Newcomer showed a series of pictograph 

Value in Oncology Practice 59 grids representing the drug, dose, and schedule regimens of patients with metastatic breast cancer from practices reimbursed by ­UnitedHealthcare. The one thing they had in common was that none of the treatment regi- mens resembled one another, showing that there was no uniform treatment of metastatic breast cancer in these community settings. Jack ­Wennberg also showed considerable treatment inconsistency when he studied a Medicare data set to determine health care resources used for patients in the 6 months before they died, comparing the data across the 77 top hospitals listed in U.S. News and World Report (Wennberg et al., 2004). There was consider- able geographic variation in the number of hospital days and physician visits per patient (Figures 6-2 and 6-3). These data imply that assessments of value in cancer care using clinical evidence from Palo Alto will be markedly dif- ferent from the same assessment using clinical evidence from Manhattan. As a payor, the geographic area one covers will distort the value obtained for a given service or treatment. Dr. Newcomer then showed data on inconsistency in cetuximab and panitumumab treatment for metastatic colon cancer among patients with UnitedHealthcare coverage. Inexplicably, patients with identical conditions receiving the same drug had an average of either 5.3 treatment cycles at $4,428 per cycle at outpatient facilities or 9 cycles at $2,693 per cycle when treated in the physician’s office. There was no correlation between physician fee schedule and use of the outpatient facilities. Clearly, there is a great deal of inconsistency in adult oncology that hampers assessment of its value. 28.0 NYU Medical Center 27.1 24.0 Mount Sinai Hospital 22.8 New York Presbyterian Hospitals 21.6 Cedars-Sinai Medical Center 21.3 20.0 Days 16.0 UCLA Medical Center 16.1 12.0 UCSF Medical Center 11.5 Stanford University Hospital 10.1 8.0 FIGURE 6-2  Days in hospital during the last 6 months of life. SOURCES: Newcomer presentation, February 9, 2009; Wennberg et al., 2004. R01506 Figure 4-2 

60 assessing and improving value in cancer care 80.0 NYU Medical Center 76.2 70.0 Cedars-Sinai Medical Center 66.2 60.0 Mount Sinai Hospital 53.9 50.0 Visits UCLA Medical Center 43.9 40.0 New York Presbyterian Hospitals 40.3 30.0 UCSF Medical Center 27.2 Stanford University Hospital 22.6 20.0 10.0 FIGURE 6-3  Average number of physician visits per patient during the last 6 months of life. SOURCES: Newcomer presentation, February 9, 2009; Wennberg et al., 2004. R01506 Dr. Newcomer’s Proposals for Improving Value in Cancer Care Figure 4-3 First, Dr. Newcomer said, the United States needs its own National redrawn Institute for Health and Clinical Excellence editable Eventually, a certain vector, fully (NICE). cost per QALY will have to be set as our cost-effectiveness limit for a treat- ment, and beyond that we simply cannot cover it in an insurance package. Second, Dr. Newcomer wondered if a new FDA designation of “scientific approval” could be created to require new drugs with uncertain economic benefits to be covered only if the patients receive them in controlled ways with no off-label uses (while enrolled in registries or trials, for example) to build scientific evidence for the drug. The drug would then be reevaluated after 3 years for a final decision. Third, Dr. Newcomer emphasized that consistent practices across providers and geographies had to be achieved. He described work he had begun with medical oncology groups who had decided on their own to go to standardized therapies of their choosing. These groups are going to be incentivized differently for standardizing treatments, explained Dr. Newcomer—they will be paid for patient services rather than based on the drugs they use. UnitedHealthcare plans to observe these practices and compare outcomes of the standardized treatments they choose as a type of cluster randomized study. It is hoped that this will advance best practices faster than RCTs. 

Value in Oncology Practice 61 In closing, Dr. Newcomer concluded that the way valuation of treat- ments occurs must be changed. Compact clinical trials that are currently used do not represent the actual value of treatments once they reach the community. The underlying system that assesses value must be changed because the one we have today doesn’t work. International PERSPECTIVES ON ASSESSING VALUE FOR ONCOLOGY PRODUCTS European countries, with all of their variation, differ quite a bit in their use of oncology products, said Michael Drummond of the Univer- sity of York. Many eastern European countries, with lower gross domestic product (GDP) per capita, are quickly adopting modern medicines to keep pace with the West. The evidence-based medicine approach is common in northern European countries and gaining ground in southern countries. Most European health care systems are based on national health services or social insurance, and hospitals, along with cancer drugs, are funded through global budgets or case mix-related payments. Health technology assessment (HTA) is growing and there are many clinical practice guidelines. Recently, HTA with cost-effectiveness analysis has been used more and more in reim- bursement and coverage decisions, with the United Kingdom, Netherlands, Hungary, Belgium, Finland, Norway, Portugal, Sweden, Slovakia, Ireland, and Germany leading this trend while Spain, Italy, and France consider whether or not to follow suit. Cancer drugs account for 10–15 percent of total cancer care expen- diture, and their costs are increasing 15–20 percent per year. With respect to expenditure per capita on cancer drugs of different vintages, the United States shows the highest spending overall and considerable spending on newer drugs while the United Kingdom shows less overall spending and very little spending on new drugs (Figure 6-4). Dr. Drummond showed a series of international comparisons of the uptake of cancer drugs (imatinib, trastuzumab, cetuximab, and bevacizumab). In each comparison the United States led all others in uptake and overall usage of the drugs. The National Institute for Health and Clinical Excellence (NICE) was created in 1999 as part of the United Kingdom’s National Health Service (NHS), and it receives funding from the government. Most physicians do not believe it is independent of the government, and NICE is widely seen as a rationing body because of its health technology appraisal program, though it has programs in other areas such as public health interventions, 

62 assessing and improving value in cancer care 40 2003–2004 35 2000–2002 30 1995–1999 <1995 25 20 15 10 5 0 United States France Japan Switzerland Austria Belgium Finland Germany Sweden Spain Italy Denmark Netherlands Canada Norway Australia Hungary United Kingdom Czech Republic New Zealand Poland South Africa SA Insured pop. All countries FIGURE 6-4  Adjusted per capita cancer drug sales (€) in 22 countries, by drug-release year (2005 data). SOURCES: Drummond presentation, February 9, 2009; Jonsson and Wilking, 2007; based on IMS Health, IMS MIDAS Quantum (for South Africa, sales per capita is R01506 presented along with two capita rates for the total population as well as for the insured [18.5%] population). Figure 4-4 vector, fully editable new investigational procedures, and clinical guideline development. Also, the program takes care to appraise both a treatment’s clinical effectiveness and its cost-effectiveness. The health technology appraisal program has two tracks: 1.  ultiple technology appraisal (MTA) is a full systematic review and M economic analysis performed by independent centers (mostly aca- demic) of several drugs in a treatment class or several technologies simultaneously. It requires 54 weeks. 2.  ingle technology appraisal (STA) was introduced as a fast-track S appraisal process. STA is not an independent appraisal but a detailed analysis of a drug company’s submitted analysis without external review. It requires 39 weeks. To date, almost all of the drugs that have gone through STA have been cancer treatments. 

Value in Oncology Practice 63 The health technology appraisal program at NICE has also employed a cost-effectiveness threshold of £20,000 to £30,000 per QALY gained, or roughly $30,000 to $50,000 per QALY (at the time of the workshop), for approval. Above this threshold, drugs are unlikely to be approved for use by the NHS. Dr. Drummond presented his research on NICE guidance regarding new cancer drugs appraised between May 2000 to March 2008. Data were extracted from published NICE technology appraisals and eventual licen- sure in the United Kingdom NICE drug appraisal outcomes were classified for each indication at one of three levels: (1) no restrictions for use in the NHS per the drug’s license, (2) no routine use (the drug is banned from NHS use altogether), or (3) restricted use only under certain circumstance, for certain patients, or in certain clinical indications narrower than the license. Of the 55 treatments appraised, 30 (55 percent) were approved without restrictions on their license, 16 (29 percent) were approved with restrictions, 8 (15 percent) were given no routine use, and one (2 per- cent) was not licensed (Mason and Drummond, 2008). Drugs that were restricted were most often approved for only a particular subset of patients. Alternatively, they were limited to use only in patients who responded while on them, as first- or second-line therapy only, or in patients who had not previously tried them. Reasons for these restrictions varied (Figure 6-5), and included insufficient evidence of effectiveness, methodological issues in economic analyses, uncertainty concerning the evidence submitted for the appraisal, an incremental cost-effectiveness ratio (ICER) that did not clearly meet NICE criteria, or an ICER that was too high. Recent Controversy Surrounding NICE NICE recently appraised several new drugs for treating renal carcinoma and recommended that none of them be used by the NHS because of poor cost-effectiveness. There was uproar as a result, with oncologists and patients outraged that the drugs were not offered in the United Kingdom but were readily available in other countries. While there was speculation that NICE had become overly stringent in its decisions, Dr. Drummond Versus interferon alpha, one drug (sunitinib) cost £71,462 per QALY (£31,185 for 0.44 QALYs gained), a second (bevacizumab) cost £171,301 per QALY (£45,435 for 0.27 QALYs gained), and a third (temsirolimus) cost £94,385 per QALY (£22,272 for 0.24 QALYs gained) (NICE, 2008). 

vector, fully editable landscape above, scaled for portait below 64 assessing and improving value in cancer care 100% 90% 80% Percent of drug evaluations 70% 60% 50% 40% 30% 20% 10% 0% Evidence ICER explicit ICER: lack Methodological Insufficient Uncertainty ICER reason for of evidence issues evidence of issues too high restriction effectiveness FIGURE 6-5  Reasons for NICE restrictions by percent of drug evaluations, May 2000 through March 2008 (n = 24). SOURCES: Drummond presentation, February 9, 2009; Mason and Drummond, 2008. explained that NICE had not changed its criteria at all. Instead, the cost of drugs submitted for appraisal had increased, and a greater number exceeded the cost-effectiveness threshold (Mason and Drummond, 2008). These controversies surrounding NICE pose important questions: 1.  ow did NICE arrive at the threshold, and is the threshold at the H correct cost-effectiveness level? 2.  re oncology drugs special in some way that should exempt them A from NICE rules? Regarding the first question, Dr. Drummond explained that the thresh- old is not based on research. Analysis of current cancer care expenditures in the NHS has found that the health service spends about £13,000 for every QALY gained—much less than the threshold (Martin et al., 2007). Ongoing research in the United Kingdom population is examining what the public thinks a QALY is worth and whether it is worth the same amount in different circumstances. Regarding whether cancer drugs should be exempt from NICE rules, there has been a major development. Treatments can now qualify for what is called “supplementary guidance for end-of-life therapies” only if they are indicated for a small patient population with a life expectancy less than 24 months, if no equivalent active therapy exists, and if they would add at least three months to patients’ life expectancy. For drugs recommended 

Value in Oncology Practice 65 for approval in this manner, the QALYs gained assume full quality of life in the added months. Also, NICE’s Appraisal Committee can take the view that, when adjusted this way, the QALYs gained at the end of life as a result of the treatment should be weighted highly enough for the therapy to be considered cost-effective when judged against the institute’s existing cost-effectiveness threshold for approval. This compromise is based on the understanding that QALYs are worth more to those who have less time to live, or those who are very unfortunate health-wise. The application of this new guidance has led to approval of the drugs for renal cancer that were initially declined, although there were also additional negotiations between NICE and the manufacturers that contributed to this change. Another approach has been the establishment of performance-related contracts, such as that for Velcade (bortezomib). During Velcade’s tech- nology appraisal, which appeared not to be going as well as hoped, the manufacturer approached NICE and offered to provide credit to the NHS for those patients who did not show a clinical response to the drug. Other manufacturers have done the same to try to reduce concerns among payors. Dr. Drummond concluded that there is considerable variation across Europe both in access to drugs and the extent to which they are assessed for value. However, assessments like those performed by NICE are widely used in drug formulary listings and the assessments do lead to restrictions in the use of medications. There are two key issues: what is considered good value for the money, and whether cancer should be treated differently from other diseases. DISCUSSION Dr. Bhadrasain Vikram of the National Cancer Institute asked Dr. Newcomer why, with the huge data he had available to him at United­ Healthcare, he did not simply perform his own NICE-like scientific approval designation. Dr. Newcomer explained that the insurance regula- tions in the United States vary from state to state, and this introduces many barriers to implementing uniform coverage decisions. He had found it very difficult to mandate that UnitedHealthcare not cover a drug because of expense or equivocal evidence. Dr. John Mendelsohn of the M.D. Anderson Cancer Center asked whether NICE’s guidance was only for the NHS or whether it also applied to the United Kingdom’s private sector as well. Dr. Drummond explained 

66 assessing and improving value in cancer care that it applied almost entirely to the NHS because few of the technologies NICE has examined impact the largest applications of private health insur- ance in the United Kingdom, such as outpatient surgery, and few private payors followed the NHS’ coverage decisions in the same way private pay- ors in the United States follow decisions of the Centers for Medicare and Medicaid Services (CMS). Dr. Robert Mass of Genentech asked how NICE integrates the value of innovation into its equation when assessing novel drugs. Dr. Drummond said that it was clearly in the guidance for the NICE committee to consider innovation in the drugs they assess, but many decisions revolve around the cost-effectiveness threshold. The French system, he said, has gone further toward pricing drugs based on their level of innovation. Dr. Neumann added that Medicare’s national coverage decisions seem to be moving in the direction of considering innovation but hinge centrally on clinical out- comes. He noted that a treatment can be very innovative without necessarily being good value. Dr. Tunis recalled while at CMS that he had seen data to suggest oncologists prescribe chemotherapy to increase personal income. He asked Dr. Neumann whether a self-report survey such as his could actually provide insights into personal revenue-driven prescribing. Dr. Neumann said that he and his colleagues tried to address this issue a number of ways, asking such questions in the survey as “To what extent have Medicare rules on oral chemotherapy limited your prescribing?” Around 60 percent said the Medicare rules had limited their prescribing. Dr. Neumann said the survey also asked oncologists whether they thought physicians in their profession made too much money. Few thought so. Dr. Neumann reported that he and his colleagues had also tried to identify the subset of respondents in practice settings with greater opportunity to make money by prescribing. Unfortunately, this was hard to tease out. Dr. Newcomer recalled an article in Health Affairs that suggested oncologists were not prescribing just to make money when there was not a reasonable indication, though they were maximizing revenue by choosing the more expensive regimen when they had multiple options (Jacobson et al., 2006). Dr. Sargent was careful to note that Dr. Newcomer’s standardization of practices was not a true cluster randomization design because there was no embedded randomization step. Dr. Sargent suggested that Dr. Newcomer introduce some element of randomization to the process and encour- aged him to do some type of matching so that the study groups could be compared in valid ways. Dr. Newcomer agreed with these points. His first 

Value in Oncology Practice 67 priority, though, was to encourage standardization of treatments because this alone could dramatically improve outcomes. Dr. Mendelsohn pointed out that it should be the patient who decides the therapies she or he will receive. The physician is ethically required to present a balanced recommendation for therapy, but the physician cannot make the ultimate decision. Dr. Neumann agreed and added that more research was needed on how people think about small probabilities of large gains and how this influences their decisions. REFERENCES CEVR (Center for the Evaluation of Value and Risk in Health). 2009. Cost-effectiveness analysis registry. www.cearegistry.org (accessed April 11, 2009). GAO (United States General Accounting Office). 1975. Information on 1976 health insurance premium rate increases for federal employees health benefits program. Washington, DC: United States Office of Personnel Management. Jacobson, M., J. A. O’Malley, C. C. Earle, J. Pakes, P. Gaccione, and J. P. Newhouse. 2006. Does reimbursement influence chemotherapy treatment for cancer patients? Health Affairs 25(2):437–443. Jonsson, B., and N. Wilking. 2007. Market uptake of new oncology drugs. Annals of Oncology 18(Suppl 3):18. Martin, S., N. Rice, and P. C. Smith. 2007. The link between health care spending and health outcomes: Evidence from English programme budgeting data. Research paper 24. York, United Kingdom: Centre for Health Economics, University of York. Mason, A. R., and M. Drummond. 2009. Public fundng of new cancer drugs: Is NICE getting nastier? European Journal of Cancer 45(7):1188–1192. Nadler, E., B. Eckert, and P. J. Neumann. 2006. Do oncologists believe new cancer drugs offer good value? The Oncologist 11(2):90–95. NICE (National Institute for Health and Clinical Excellence). 2008. Renal cell carcinoma— Bevacizumab, sorafenib, sunitinib, and temsirolimus. http://guidance.nice.org.uk/TA/ Wave14/22 (accessed February 9, 2009). Pusztai, L., and F. J. Esteva. 2006. Continued use of trastuzumab (Herceptin) after progression on prior trastuzumab therapy in HER-2-positive metastatic breast cancer. Cancer Investigation 24(2):187–191. Reddy, J. C., J. D. Reimann, S. M. Anderson, and P. M. Klein. 2006. Concordance between central and local laboratory HER2 testing from a community-based clinical study. Clinical Breast Cancer 7(2):153–157. Wennberg, J. E., E. S. Fisher, T. A. Stukel, J. S. Skinner, S. M. Sharp, and K. K. Bronner. 2004. Use of hospitals, physician visits, and hospice care during last six months of life among cohorts loyal to highly respected hospitals in the United States. BMJ 328(7440):607.