3

PATIENT PERSPECTIVES OF THE SIGNIFICANCE OF UNDERSTANDING HETEROGENEOUS TREATMENT EFFECTS

As researchers and clinicians search for the best ways to deal with and take advantage of heterogeneous treatment effects (HTE), they must keep in mind the needs and desires of the consumers of the treatments—that is, the patients. Thus, one workshop session was devoted to exploring the patient perspective on HTE, including what heterogeneity means to patients, how they can benefit from it, and how clinicians and researchers can work with patients to realize the promise of HTE. Much of the information offered in this session was relevant to the patient question: How can I use knowledge about HTE to improve the outcomes that are most important to me?

ENGAGING PATIENTS IN DISCUSSIONS ABOUT HETEROGENEOUS TREATMENT EFFECTS

To set the stage for the presentations, Thomas Concannon, Senior Policy Researcher at Tufts University and the RAND Corporation, described his work with the Patient-Centered Outcomes Research Institute (PCORI) to understand patient concerns. PCORI, which funded the workshop, has been working to prepare patients to participate in research. As part of that effort, a team at Tufts University prepared descriptions of the concept of HTE written in a nontechnical manner and then used that material in discussions with three PCORI-funded patient-powered research networks (PPRNs).¹ Each meeting lasted 90 minutes, the last half of which was devoted to structured discussions. During these discussions, participants were given two hypothetical patient scenarios and asked to apply the results of the hypothetical research described in those scenarios, including subgroup analyses, to make decisions regarding their own hypothetical care.

Concannon and his team recorded, transcribed, and then analyzed these discussions. They identified several themes that fell into four major categories: what decisions patients said they made routinely, what kinds of information they seek out in making their decisions, where they go for that information, and how they judge the quality of the information that they discover. Concannon said,

Many of the patients described dealing with concerns and questions in ways that were “clearly not ideal,” Concannon said. One strategy, for instance, was visiting multiple providers until the patient could understand how existing data might be applied in his or her case. Another was treating their own care as an “N-of-1 experiment” until they were able to find a satisfactory treatment; but this can only be done for conditions with symptoms that respond to changes in treatment

___________________

¹ Patient-powered research networks (PPRNs) are networks that are organized and operated by patients and their partners. Each PPRN is dedicated to a particular health condition, such as multiple sclerosis or chronic obstructive pulmonary disease. PPRNs work with researchers to shape the research agenda into the disease of interest and to collect health information from their members that can be shared with researchers. Each PPRN is part of PCORI’s research network, PCORnet.

in a timely fashion. The patients would also delay making a decision when they felt there was not enough information.

Concannon concluded, “Patient-centered care goes well beyond the usual characteristics of patient preferences and questions of access to care and includes questions around data and the patient-centeredness of data.”

Following Concannon’s opening remarks, two presenters and two responders spoke about the patient perspective. A key theme of their presentations was that the traditional relationship patients have had with the health care system will not be sufficient to meet the demands and the opportunities of HTE. The presenters offered a variety of suggestions for the necessary changes.

THE PROBLEM WITH TREATMENTS AIMED AT THE “AVERAGE PATIENT”

“Traditional research failed to help me when I was first diagnosed with MS [multiple sclerosis],” said Seth Morgan, a neurologist and a fellow of the American Academy of Neurology. His experience as a patient in the medical system convinced him of the importance of the current movement to understand HTE and provide patient-centered care.

Fourteen years ago, Morgan said, he was a neurologist in private practice when “I had the unique experience of diagnosing myself with multiple sclerosis.” Following his diagnosis, he continued practicing as a neurologist for a couple years until he felt he was no longer performing adequately. He has since been a patient advocate for the National Multiple Sclerosis Society and for people with disabilities generally.

When he and his doctor, a fellow neurologist, first set out to find a treatment plan, Morgan said, he quickly ran into a conundrum. “The premise of traditional research,” he explained,

When Morgan was first diagnosed with MS, the only available treatments were injectable medications. After researching the options, he and his doctor “by gestalt” determined what they felt was probably the strongest medication on the market. “That was the best we could do—a best guess.” One problem Morgan had with the

medication was his phobia of needles. “I did not like needles.” As his only choice, he stayed on the medication, giving himself an injection every other day for 2.5 years. But he never got used to the needles, and after those 2.5 years, he simply could not do it any longer. “I went to my physician and said, ‘I know the data [are] against us, but I’m stopping this medication. I don’t care if it’s going to cause progression of the disease or not. I just can’t be on injectable medications anymore.’” By chance, GILENYA^®, the first oral medication for MS, was being studied in clinical trials at the time, and he was able to start taking it instead. After about 1 month on the new medication, Morgan said his wife turned to him and said, “You’re back.” What do you mean? he asked her. And she explained, “You were cognitively affected. You were duller than you normally were. And now you’ve cleared up.”

Neither Morgan nor his doctor noticed that the injectable medication caused subtle cognitive issues—a side effect that Morgan suspects might have been anticipated if less attention had been paid to the “average patient” and more paid to him as an individual. He was an older individual who developed MS and also had strong family histories of Alzheimer’s disease in both parents. If the proper studies had been conducted, researchers might have recognized that this particular drug could cause cognitive issues in members of a vulnerable subpopulation—and he might have avoided 2.5 years of dulled cognitive function.

Today, he noted, there are multiple treatments for MS. Yet, there is still no way to provide doctors and patients any direction on which medication should be tried first or which medication is most promising for a particular individual patient. A big part of the problem, he said, is that it is impossible to compare the outcomes of different studies “because of the different parameters and specifications that were delineated by the individual pieces of research.” Ideally, studies would be conducted with uniform selection criteria and comparable treatment regimes and analyses to enable valid comparisons among treatments.

Even better, Morgan continued, would be trials that compared the performance of different drugs in various subgroups of patients. For each subgroup, the trial would ask, Which medication has the least likelihood of causing side effects and the greatest likelihood of providing benefits? “Of course,” he said, “the trouble is that you’re not going to get drug companies to fund those because they don’t want to have head-to-head studies against a different drug in case they happen to come out on the short end, and they have no control over the way in which patients are selected.”

Still, he believes the situation is improving.

“It’s a real big problem,” he concluded, “and I think it’s important that we continue this movement toward patient-oriented treatment because we need to understand the situation of each individual as best we can and figure out what subgroup of individuals is most likely to benefit or not benefit from a given treatment.”

TAKING PATIENT PREFERENCES INTO ACCOUNT

In the next presentation, Christine Stake, Research Operation Manager of the Ann & Robert H. Lurie Children’s Hospital of Chicago, described two of her experiences with the health care system that indicated the importance of doctors communicating with their patients and understanding what is important to them in a treatment.

The first experience she described was with severe osteoarthritis in her hip. As a child, she had hip dysplasia. And by the time she was in her early 30s, she was in severe pain—to the point that she had to crawl up her stairs. When Stake went to a doctor, she was told she needed a hip replacement, but she was “too young.” Her only other options were physical therapy and pain medication, both of which stopped working within 6 months. So, again, she went to the doctor, only to hear same thing:“You need a hip replacement, you’re too young.”

When she asked for further explanation and for the specific contraindications, Stake was told that, because of her young age, she would likely need yet another hip replacement in her lifetime. When she asked for evidence, however, her doctor could not provide any—there were no studies of hip replacements in young people because such operations were rare. Consequently, there was no way for her—or her doctors—to know how long a hip replacement was likely to last in a young person like herself.

Throughout her search, she saw several doctors and heard the same answer from each: No hip replacement because you are too young. Stake was frustrated for a couple reasons. First, the doctors were making a decision without any research to guide that decision. Related research had shown that, in the elderly population, hip replacements are successful, she said. “For the average patient, the outcomes were incredibly high; but this surgery was not being allowed

to be offered to me.” Her other problem was she felt the doctors were acting paternalistically. She understood, however, they thought the likelihood of her having to replace her hip twice was too high. In short, they were basing the decision on what they thought was the most important factor: risk of a repeat surgery. Yet, they never communicated with her about what she thought was the most important factor or about her concerns about quality of life. Was she willing to make the trade-off of someone in her 30s trying to have a life versus the risk of possibly needing another hip replacement later in life? No one had that conversation with her. Finally, Stake said, she found a doctor who talked with her about a hip replacement in terms of these trade-offs. Jointly, she and the doctor decided that she would indeed get a hip replacement.

“As a patient,” Stake explained, “I don’t walk in and expect a doctor to say, ‘Based on you as an individual, this is the right decision for you.’ But I do expect a conversation—to say, ‘These are your risks, these are your benefits, let’s make the best educated decision we can.’ … We really have to talk with patients about what’s important to them.” For her, Stake said, the hip replacement was the right choice. “I had that surgery. The minute I woke up from surgery, my joint pain was gone.” Afterward, she returned to school and completed her doctorate, for which her research focused on patient decision making.

That experience reinforced the lessons she learned nearly a decade earlier when, at 24 years old, doctors found a lump in her throat. They did not expect it to be cancer, she said. Indeed, the doctors told her it was highly unlikely to be cancer—but when the tests came back, it was in fact cancer. She needed her thyroid removed. The question for her was, What sort of follow-up radiation treatment would she choose? She could choose to have either a high dose or low dose of radioactive iodine to clear up any remaining cancer cells in her throat. Normally, she was told, a patient would be given the high dose, but there was a chance that the high dose might affect her fertility. Which did she want?

Once again, the problem was, as she noted, the lack of evidence to support whichever decision she made. All she knew was the low dose meant a higher risk of her cancer returning, while a high dose was more of a threat to her fertility. She chose the lower dose. “I made the decision based more on quality of life because I didn’t have the evidence to make that decision,” she explained.

One lesson that Stake drew from her experiences is the importance of getting information that helps outlier patients make decisions about treatment. She said,

Stake concluded by challenging those in the audience to observe things differently. “How do we assess patients on that continuum and not always ignore the two standard deviations or those outliers? And how can we maximize those opportunities to have those conversations with patients?”

PROVIDING PATIENTS WITH DECISION-MAKING TOOLS

In her response to the presentations, Karina Davidson, Professor of Behavioral Medicine at Columbia University, focused on the need to provide patients with the proper decision-making tools to help them figure out how to deal with information about treatment heterogeneity.

“I agree that heterogeneity is a big barrier—or opportunity—for the practice of our clinical work in the next generation,” she began. Indeed, she said, as a clinical psychologist she is used to dealing with patients who tend to need personalized solutions. The depressed, the anxious, the schizophrenic, the addicted, the obese, the smokers who are trying to quit—for none of these patients is there any sort of “magic pill” that has been clinically shown to help everyone with that problem.

“My whole life,” she said, “has been thinking about the heterogeneity of treatment effect and the fact that we ask patients to deal with this completely weird counterfactual: What will your life be like on this lifelong drug that is going to cause you palsy, cognitive decline, and all these side effects but clear up reality-versus-not? And how can they possibly imagine that?” Furthermore, the evidence available to clinical psychologists may or may not be relevant to a particular patient, and its applicability will only become known after the treatment is done.

With that, she moved to what she described as her main point—that evidence-based medicine as originally described by David Sackett is actually evidence-based practice, not medicine. “He abhorred the idea of recipe medicine,” she said, “and implored us to think through, empirically and with evidence, how we incorporate patient preferences, values, and their preferred outcome into the art of the practice.

But that piece was hard, so we worked on—I think, importantly—the randomized controlled trial data.” But the result of that, she said, is that neither patients nor practitioners ended up getting the evidence they needed in the way they needed it in order to apply it quickly and efficiently.

“We could do better than that in heterogeneity of treatment effect,” she said. If the research is done, but the needs of the practitioners and patients are not taken into account, then the job is only half done, she said. It is important, she said, that medical researchers and others in the health care community have a conversation about the sorts of dissemination and implementation tools that will help practitioners and patients understand and apply what is being learned about HTE. “If we can come up with ways that are innovative and new and join up with the people in dissemination and implementation,” she said, “we might have some really great ways that we can start … figuring out the best practices for informing patients and practitioners about when heterogeneity of treatment effect is expected and what the best course is for a patient.”

INSURERS AND HETEROGENEITY

It is not just clinicians and patients who will need to learn to deal effectively with treatment heterogeneity, said Robert Dubois, Chief Science Officer and Executive Vice President of the National Pharmaceutical Council. Health insurance companies and other payers must adapt, as well. In his response to the panel presentations, he spoke about the challenge that HTE poses to health insurance companies.

The development of new medical treatments depends on a virtuous cycle, he said: There is a need, a drug is developed that meets that need, the drug gets used, it gets paid for, and then some of those dollars are used to pay for developing new treatments. For the virtuous cycle to succeed, Dubois said, the relevant patients have to perceive that a treatment is beneficial to them, doctors have to recognize it as the preferred therapy for those patients, and, crucially, insurance companies have to pay for that therapy.

How do insurance companies decide which treatments to cover? Increasingly, Dubois said, their decisions depend on various sorts of value frameworks, and for the most part those value frameworks base their determination of value on the average patient. There are some exceptions, he noted. When a genetic test can predict whether a patient will respond well to a particular treatment, for example, insurance companies may value the treatment differently for different drugs.

The vast majority of decisions, however, do not take into account many of the factors that are important to patients, he said. Consider the choice of a cancer

therapy. Some regimens are very complex for the patient to deal with, while others are simple and straightforward. Some require injections, others do not. Some have an effect on one’s quality of life, others affect quantity of life. And different patients value these things differently. One patient may want a straightforward regimen that will not ruin his quality of life, and he will accept that he may not live quite as long as he could; while another may be willing to put up with any sort of regimen, any sort of side effects, because she really wants to live as long as possible. Yet, when an insurance company does a value determination for this sort of cancer drug, it gets a single answer: Its value is X dollars.

“The problem,” Dubois said, “is that patients do not fit into simple dichotomous categories; and if you have 3, 4, 5, 10 types of outcomes that are important, then you have to take those into account. And, unfortunately, we’re increasingly seeing value frameworks that cannot handle that.”

This is not just a theoretical problem, he said. New York State just carried out a review of a new cystic fibrosis drug based on cost per outcome, and it decided that the drug did not meet its threshold and would not be paid for. It was a one-size-fits-all assessment of that cystic fibrosis drug. However, Dubois said, it is very likely that because different patients would have different preferences and experience different outcomes with that drug, some patients would find that particular drug to be very valuable for them, perhaps more valuable than any other drugs that are on the market. “Those patients will not have access to the drug,” he said. “And we go from a virtuous cycle to a vicious cycle because if drugs come to market and they’re not being valued correctly for different subgroups of patients, patients won’t get it, it won’t be paid for, and then new drugs won’t get developed.”

The take-away message, Dubois concluded, is that in the era of recognizing HTE, it is not just clinicians and patients who will need to come up with new ways to make decisions. The payers, too, will need to develop new ways of determining value and making decisions about which treatments will be paid for and which will not.

DISCUSSION

In the discussion session following the presentations, Richard Willke, Chief Science Officer of the International Society for Pharmacoeconomics and Outcomes Research (ISPOR), spoke about how U.S. insurance companies might deal with treatment effect heterogeneity. Generally speaking, he said, most U.S. insurance companies do not have a one-size-fits-all approach to paying for medical treatment. Most have “utilization management,” which attempts to draw

distinctions among patients and make reimbursement decisions based on some individual factors. The insurance companies are an important audience for the topics being discussed at the workshop, he said. “They’d like to know better. We maybe don’t give them enough credit for trying to do the right thing.” Their actuaries are pretty sophisticated, he said, but they need the right sorts of evidence to make those decisions. “We ought to think about that audience as well.”

Dubois agreed, and he said that it may not be as difficult as one might first think for insurance companies to make decisions that take both HTE and patient preferences into account. Although, in theory, patients could have a wide variety of sets of preferences that insurance companies would have to deal with, Dubois said he believes there may be just a few groupings that would cover most patients. One group of patients, for example, puts more focus on quality of life, while another wants treatments that will extend life as much as possible, and there are perhaps two or three other subgroups. Once the subgroups and their preferences are determined, it will be necessary to come up with utility scores for different options, with each subgroup having different utility scores. Then by combining the utility scores with data on the likelihood of various outcomes, it would be possible to calculate the expected benefit of various treatment options for each of the subgroups. It has not been done before, Dubois noted, but it could be a workable approach.

Naomi Aronson, Executive Director of Clinical Evaluation, Innovation, and Policy of Blue Cross Blue Shield, also addressed the question of the role that value plays in determining what insurance companies will pay for treatments. “There is a lot of discussion of value now,” she said. However, she added, insurance companies are contractually obligated to determine payments according to a medical necessity model, not a value model. Many clinicians are concerned with how that model is skewing payment decisions, she said, because “low-value cancer care is driving out very high-value treatments for, say, inflammatory arthritis.” In response, there is a push toward adopting value-based payments, with insurance companies paying less for procedures and individual services and more for episodes of care. “There is a commitment to drive payment in that direction,” she said. “And it is happening.” It probably will not be easy, however. In order to move toward value-based payment, insurance companies must determine who will benefit from procedures and who will not, “which is exactly the heterogeneity we’re talking about here.”

Turning to a different topic, Frank Davidoff, Editor-in-Chief (Emeritus) of the Annals of Internal Medicine, offered a comment about medical care being a combination of both a production model and a service model. He spoke of a

movement he described as “being developed largely under the leadership of Paul Batalden” of the Institute for Healthcare Improvement.

“Basically, the model they have begun looking at is the model in business of what they call the production model and comparing that with the service model,” Davidoff said. “In a production model, you produce a product and you send it out into the marketplace, and it gets used,” he explained. It is, in a sense, a one-way transaction. By contrast, the service model is “a reciprocal, repetitive, cyclic, ongoing kind of process which is continuously adapting and changing.” Examples of the service model include everything from house painting to lawyers to social work.

“Medicine has the disadvantage, it seems to me, of sharing the qualities of both,” Davidoff said. But in some respects, he suggested, we are trying to fit a service industry in a production model. This leads to much of this discomfort and tension in medicine that had been talked about at the workshop, he suggested.

This page intentionally left blank.