A Nuanced Approach Is Needed to Determine Value of Drugs
Recently, CVS Health announced that it will use the value-based drug pricing system of the Institute for Clinical and Economic Review (ICER) to exclude some high-cost drugs from plans. In a blog post for Health Affairs, Robert W DuBois, MD, PhD, chief science officer of National Pharmaceutical Council, expressed concerns that the approach could restrict patient access to certain medications. We spoke with Dr DuBois about the use of value frameworks to inform formulary decisions and what factors should go into determinations of cost-effectiveness.
To begin, can you briefly explain the new service announced by CVS Health?
As you are aware, health plans and pharmacy benefit managers (PBMs) have formularies, and those formularies are built around tiers based upon preferential treatment that they give one agent vs another. In some cases, generic is the lowest tier, then preferred brand agents, then nonpreferred, and then the specialty drugs fit in their own tier. Historically, those decisions around both what agents are on the formulary and in what tier were built around a one-stage process and sometimes a two-stage process. In a one-stage process, the P&T committee makes purely clinical decisions. Is there evidence to support that this drug is safe and effective and should be available to patients in the environment? It doesn’t have any economics associated with that. In a two-stage process, the group of proposed or clinically approved medications go into a contracting process where rebates and those kinds of things are discussed, and that may then lead to categorization of what tier it might go on. Now, there are some PBMs and payers that do not have it as two separate processes. They may combine those in some fashion where they will think about the clinical and the economic at the same time, but classically it is done in two steps.
What has not been done until the CVS announcement—and then what New York State has done—is to say that a drug will not be on formulary because a cost-effectiveness measurement performed by ICER concluded that it is above a threshold that CVS chose, which is $100,000 per quality adjusted life year (QALY). And that is what is quite new. It is one thing to say this drug may be more expensive than another and we want you to undergo prior authorization [or step care] to make sure it is clinically necessary. With the CVS approach, though, those drugs won’t be available. It’s not a matter of step care. It’s not a matter of prior authorization. It’s not a matter of appeal. They won’t be available. So, that is quite a new and a rather definitive decision that has not previously been done before.
In your blog post for Health Affairs, you raised a concern that basing formulary decisions solely on cost-effectiveness can limit patients’ access to certain medications. Are there particular medications that you think will be unfairly affected by this approach?
Well, I cannot name specific drugs, because I do not know what exactly they are thinking of excluding. As CVS has explained it, it is really removing two elements of value from being considered.
One is that, by using the cost-effectiveness threshold determined by ICER, they are only looking at medical costs and medical cost offsets. So, they are asking how often people end up in the hospital, how often they go to the doctor, how often they need certain scans, and then they add it to the cost of the drug to arrive at a total cost of care. But to do cost-effectiveness properly, you can’t just look at medical costs. You need to look at indirect costs or societal costs. One of the most important elements is productivity. Does a healthier patient go to work more regularly, and are they more functional at work? For most chronic care therapies, that is what you are gaining from much of the treatment. For a lot of diseases, say rheumatoid arthritis as an example, the drugs do not cause you do live longer; they cause you to live better. If one therapy causes you to go to the doctor and you’re nauseous and vomiting and feel horrible for two days, then, and somebody has to stay home and take care of you, but another drug is maybe an oral therapy that requires far less burden of your loved ones to take care of you, you would think that’s part of the value proposition. So, CVS has chosen very specifically not to consider in their cost-effectiveness number things like productivity improvement or caregiver burden.
If you are doing appropriately done cost-effectiveness analyses—and there are standards for what it is supposed to look like, but probably the most recognized one is called the Second Panel on Cost-Effectiveness in Health and Medicine—you need to look not just at medical costs but also at these alternative costs as well. So, I am not maligning cost-effectiveness, but you have to have a complete understanding of cost-effectiveness.
Looking at the critical missing elements I just discussed, think of any disease that might impair your ability to work: migraine headaches, depression, rheumatoid arthritis. And for caregiver burden, think of a lot of the oncologic therapies that might cause you to require somebody to help take care of you quite a bit. Or, if you think about multiple sclerosis, if it is treated well, you need less help at home. If it is not treated as well, you might need more help.
So, that is the first. Then, apart from cost effectiveness, there are other factors that all of these other frameworks consider or incorporate in some fashion. Is it an underserved treatment population, and is that a consideration to think about? If you look at, say, the ICER framework or some of the other frameworks, they take into consideration whether this an area where there are not a lot of treatment options, because if there are not, then a new treatment option has an added value.
Or, is this an area where you can’t really predict who is going to respond and so, the doctor and patient need lots of options? As an example, some people respond to Prozac for their depression and others don’t respond to Prozac but may respond to something else. And that’s just the nature of the disease that one size won’t fit all.
Another factor is the regimen. For example, in the old days of HIV therapy, you had to take 40 pills in a day, a huge burden. People often were not able to adhere and did not do so well. And now you have a new regimen where basically you have one pill. That concept of the difficulty of the regimen is also considered in value frameworks, but it is not incorporated in the cost-effectiveness number.
So, there are a lot of other factors that influence value. Let’s say you have two different therapies, and both allow you on average to live an extra 2 years. But let’s say for one of those therapies, very few people live more than 2.5 years, but for the other therapy, the average is 2 years but 20% of people seem to live 5 or 10 years or longer. The averages do not look different, so the calculations might not look different, but the fact that you could cure a patient is another element that should be considered. So, just pick any disease where you think one of those elements are relevant.
In a counterpoint to your blog post, two representatives of CVS Health argued that manufacturers concerned about their products exceeding the allowed QALY threshold can simply lower the price to ensure access. Do you agree that we’re likely to see this happen as a response to the policy?
Well, it is complicated for a couple of reasons. At some level you could paint a scenario where you would say, everybody knows the calculation, so drug company, just drop your price. But for the industry to say okay, we will just drop our price, presupposes that the calculation is correct and that the threshold is correct. We have already talked about the issues with the calculation. If you say the threshold is: we’re not going to include drugs that have a cost of more than $100,000 per QALY; then question will be, where did the $100,000 come from? In ICER’s assessments, often they use $150,000, and the price that would be acceptable to CVS is going to be a lot different if you set the threshold at $150,000 or $100,000. None of this is written in stone, but one of the ways people determine what a threshold might be is a calculation based upon gross domestic product (GDP) per capita. If the GDP per capita is very high, then you might have more money to spend. So, some people have proposed in some countries a QALY threshold of three times GDP per capita. In the United States GDP per capita is around $50,000 and so, if you triple that, then you get $150,000.
Additionally, one threshold probably doesn’t fit all needs. Let’s take an orphan disease; if you’re only going to have 1000 people, it is hard to do a complete drug development and bring a drug to market and charge a low amount if there’s only 1000 people. If there are 10 million potential people, you can pay for the development at a much lower cost per patient because there are a lot more patients. So, to have a single threshold does not take into account the nuances of different diseases.
In general, your perspective seems to be that the CVS Health approach is too extreme. What would you consider a more reasonable alternative approach to ensuring that manufacturers set reasonable prices for their products?
We have taken a look at the field of value assessment frameworks and identified some common limitations that the field has to solve. One is that you need to incorporate what is important to the patient, such as the burden of taking the regimen, the burden on your loved ones, the ability to go back to work.
The second is you need to take a broad system perspective. When you think about value, there are lots of dollars spent on the health care system. Why is it that we are only doing value assessments for drugs? Shouldn’t we do that for robotic surgery, and PET scans, and proton beam therapy for prostate cancer? So, one needs to be looking very broadly at all areas of expenditure and not just at drugs.
The third is that, in many areas of medicine, things change rapidly. So, if you do a review, you need to make sure that you are doing re-reviews on a timely basis so that if the information changes the value assessment changes.
Another is that the clinical and economic calculations and the modeling that is done has to be transparent. If you’re wondering how they came up with this number, you should be able to look at the model and play with the model to see what is going on.
To a great extent, every framework out there has one or more of these limitations, but they are all getting better. This is not to say we should ignore value determinations, but we have to be cautious because this is a young field. In the meantime, how can it be done better? I would say implement it in a nuanced way. Use all the factors that matter in terms of what patients care about when you assess value. And then, look at all aspects of the health care system as you think carefully about how you want to allocate resources.
