MB: The practice of medicine is based upon scientific data – clinical trials which are reproducible. But the problem is that the primary outcomes in evidenced-based medicine often don’t look at quality-of-life problems or apply quality-of-life issues. You also can’t integrate them with cost, and you can’t look at cost without value. Value-based medicine is the practice of medicine based on the value conferred by a systematic intervention. Value is the ability to measure improvement in both length of life and quality of life. Value-based medicine standardizes parameters used in valuing our interventions, and aims to do it well enough to put it into policy, or at least help physicians know the best way to proceed.

MB: There are a lot of different ways that you could measure quality-of-life changes. Time tradeoff is the one that we use most. There are different methodologies to establish quality-of-life. A function-based methodology focuses on what one is able to do. Instead, we look at a preference-based methodology, in which a patient is trading time for better health. We ask what length of time one might expect to live and how much of that time one would trade to get a particular outcome, such as perfect vision, or perfect ambulation, or perfect gastrointestinal function. When you ask these questions of many patients the confidence intervals become very small, these numbers become very solid, we can compare them across specialties and across different fields and also use them in economic analyses.

MB: One way is called standard gamble, in which the patient would have to determine what percent risk of dying would they accept to receive a certain type of treatment. That’s used quite a bit in Europe, but we did this over a fairly large number of patients and found that people were risk-aversive and don’t want to make that determination. We found that asking patients to make that kind of decision was not nearly as reliable as asking them how many months or years would they trade to get that treatment – the time tradeoff approach. It doesn’t matter which of these preference-based methods you use, as long as they’re used consistently and your database comes from only one source. We’ve been working for years with time tradeoff, so we have many measurements and we are fairly comfortable with them.

A parameter that is sometimes not standardized is who you get the data from. Is it the patient? Is it the doctor? Is it someone out in the community saying what they expect someone else would trade? We believe it should be the patient because, when you ask doctors and community folks the same questions, the numbers come out very different. For example, when you ask physicians about vision loss from age-related macular disease, they give up much less time than a patient who has the disease – showing that they under-value it. If you ask physicians about rehab from a stroke, it’s interesting that they give up more time than the patients do. I don’t know exactly why, but it might be that someone who is actually in rehab sees the differences and has much more hope than the physician.

We standardize the preference-based measurement with patients, and then integrate it with a standardized measurement of cost. We use Medicare’s physician and hospital reimbursement costs, which are consistent and always accessible. We use the standard wholesale price for drugs out of the Red Book, which gives us standardized costs. There are other ways to do it, but if you do it one way every single time, that makes the outputs comparable across all the different analyses you do. Once you’ve evaluated how much something costs per measure of quality-of-life change, then policymakers can determine what is affordable and what isn’t.

MB: You actually start with clinical trials and interventions. You take a look at what the clinical trial says about a particular intervention: whether it works or doesn’t, and what is the end point. You utilize utility analysis to convert that data into value, using the time tradeoff technique. For example, your start out with a clinical trial about cataract extraction where you take someone from a 20/100 vision to a 20/30 vision. Then you convert those numbers – 20/100 or 20/30 – to value. What quality-of-life standard does the patient have with a vision of 20/100 or 20/30? If you ask this over many patients, you’ll get what the utility is of having vision at 20/30 or 20/100. As another example, a particular type of cardiac intervention might involve going from an American Heart Association classification of Class 3 to a Class 1. When you gather data on value, you can then figure out what the quality of life is at a particular level of health. By doing that you can see what the improvement of value is when you do an intervention. We use a decision program that includes not only the value improvement of an intervention, but it also looks at the utilities associated with adverse effects. Once you do that you can establish what your total improvement is, measured in quality-adjusted life years. At that point you can apply the costs to get dollars for quality-adjusted life years.

MB: It depends on what you’re looking at and what the model is. That’s why it’s important, from our standpoint, that you have physicians doing this because they understand the clinical trials and they know how to make the models that include the major effects. We have published the book in the hope of guiding some standardization by showing what we believe is probably the best way to standardize the variables. The question is, who does this and how does it get done? I don’t expect a physician in a private practice to do these calculations, but there are centers such as ours or others in the country and internationally that do these. Ultimately, these types of analyses, if done in a standardized fashion and properly, can become the basis for quality standards, and I think that specialty societies would be involved in doing this. A lot of societies now have care guidelines, but most are based on evidence. If you base them on value, you’re adding an additional sophistication to it that I think probably is reflective of better quality of care.

MB: We use a decision analysis computer model called Triage, which is used by many centers. But it’s the model-making that is important. You have to make sure that you put in the appropriate weighted values of the interventions, and the weighted side effects. Take cataract surgery as an example. You start out with a patient who has a cataract and the two first branches of this decision model are going to be: do surgery or not do surgery. When you treat and do surgery the next branches are going to be the possible outcomes of that surgery: the first is no adverse effects, the second might be a 0.5 percent chance of retinal detachment, the third might be inflammation, and so on – all the data coming from clinical trials. What the utility analysis will tell you is what is the quality-of-life of someone who in fact has that condition. You have to be cautious to make sure you have done a wide review of the key clinical trials to develop a model that includes the incidence of the different effects of the particular intervention, and then incorporate an aggregate patient value into each of those.

MB: Yes. You can establish a value for the most probable outcome of an intervention, know what the value would be to patients based on your utility analysis, and look at the cost as dollars for quality-adjusted life years. You can take an intervention that is very effective from a clinical standpoint, has much value and doesn’t cost very much – and that’s going to come out to be very cost effective. You can also have interventions that cost a lot, but also turn out to be quite cost-effective if the value is quite high when you do the calculations. We also had an example of an inexpensive intervention – treating dental procedures with some antibiotics – that turned out to be very expensive when you look at what generally happens and what the value is. There have been cost-effectiveness parameters such as $100,000 per quality-adjusted life year that have been used as standards – I think that was a number that was taken out of the air. Now, many people look at interventions that are $50,000 per quality-adjusted life year or less as being very cost effective. In my view, it really comes down to a particular society and what they want to spend.

MB: What are the values that people want to look at when they look at pay-for-performance reimbursement and value-based purchasing? I think it should be the ability to perform interventions that are offer the best value for the least money. Let’s take the example of proton pump inhibitors: when you look at clinical trials of the these drugs, you can get cost per quality-adjusted life years for all of them, and you’ll find the over-the-counter ones to be the least expensive and the most cost-effective. Another example might be age-related macular degeneration, which is the number one cause of blindness in this country. Over the last few years, and even months, there have been a number of pharmacologic agents that have made huge headway in treating this illness, but very few of these have head-to-head comparisons in clinical trials. But with a value-based assessment, you can take the clinical trial data and understand that a certain drug improves vision a certain degree, and a different drug does it at another level, and you can come up with numbers such as cost-per-quality for all of these different drugs. These numbers are going to be able to tell a physician or an insurance company what, for different types of scenarios, might be the best way to go because they include the side effects as well as the clinical effects.

So how do we use this? Physicians might look carefully to understand how to evaluate these cost-utility numbers and make some specific decisions on how they treat a particular problem. Insurance companies might have five or six or seven drugs in one category, and they can get cost-per-quality numbers that may help them establish different copays depending if someone wants to use, say, the most cost-effective drug that has the same value as another. We always look at value first: those drugs that have the same value to patients, reflective of what it does, how it works, and how it changes the quality of life for that patient. If you have two drugs that have equal value, then you look at the costs, and insurance companies are not doing their jobs if they don’t look at things that are less expensive for the same value. From a patient’s standpoint – and that’s the most important thing – they can delineate those interventions, whether it’s pharmaceuticals or devices or surgery or medicine, that give small effects in improving the quality of life. They can tell you which might be detrimental. They can tell you which ones have large effects. But of those interventions that have similar value and do similar things, this methodology now allows us to take a look at the costs.