Clinicians’ perspectives on thinking about costs of available therapies and making treatment decisions for patients at risk for cardiovascular events.
Deepak L. Bhatt, MD, MPH: I’ll ask Dr Navar the next question. This is, I think, challenging. Obviously, as physicians, we know cost is important. Everybody is always talking about percentage of GDP [gross domestic product] and health care costs. It’s not sustainable in the United States, etc. But how do you personally think about cost when you’re making treatment decisions for that individual patient in front of you? And how does that tie into the issues of accessibility and so forth?
Ann Marie Navar, MD, PhD: Thanks, Dr Bhatt. The payers and health systems use the ICER [incremental cost-effectiveness ratio]-type data to make general decisions around a formulary and try to estimate the population level impact of certain therapies. But as a clinician, when I’m sitting in front of a patient, that is a very different conversation. The reality is that although the cost-effectiveness goes into payer decisions for coverage, what a patient actually pays is very different than what the payers might be calculating in terms of the value. We know from data regarding the early use of PCSK9 [proprotein convertase subtilisin/kexin type 9] inhibitors that the likelihood that a patient picks up a prescription from the pharmacy starts dropping off dramatically at a copay above $20 a month.
So it’s really important that we look at what our patients’ out-of-pocket costs are going to be. We know that those on government insurance programs are ineligible for copay programs, but all of our commercially insured patients should be pointed toward copay assistance programs from the pharmaceutical companies. And for generic medications, we should be looking at things like GoodRx, Inc to find the cheapest pharmacy where they can get therapy. Out-of-pocket cost is super important.
The other thing that’s important to remember is that although these population estimates of cost-effectiveness are really good on a population level, they may miss the actual value for individual patients. There’s a lot of noise in the risk estimate for a certain patient, and they may not have all of the typical risk factors but may have other risk factors that would increase their risk. Or, other things that we don’t think about are things with more long-term value. So a young person with high LDL [low-density lipoprotein] cholesterol who may not be considered very high risk wouldn’t be, as a population, considered cost-effective in 2 years. But this is somebody who has a potential for LDL lowering for several decades, and this person may actually pay off in the long run. So it’s important that we don’t misapply short-term cost-effectiveness analyses when we’re thinking about treating patients.
Deepak L. Bhatt, MD, MPH: Yes, those are very valid, terrific points. Maybe I can turn back to Dr Bress. You convinced Dr Navar that cost-effectiveness is important. OK. But what kind of data do you actually want to deem something as cost-effective? You mentioned some of the numbers in terms of thresholds. What sort of data do you want to make those decisions?
Adam Bress, PharmD, MS: Dr Navar is right. In order to produce high-quality cost-effectiveness estimates, you need high-quality data on the inputs in which the model needs to produce those estimates. I can simplify the inputs into 3 general categories. The first is the effectiveness inputs, which are usually the relative effects of the treatments you’re comparing. So, treatment A versus treatment B on all of the downstream health outcomes we care about, inclusive of both beneficial and harmful outcomes.
The second is cost inputs for both treatment options themselves as well as for the downstream health outcomes that occur over time as a result of the different treatments. And again, inclusive of both cost for acute treatment of those conditions as well as cost of chronic treatment of those conditions over time.
And then the third component is the utility values. These are important for the assessment of the quality-adjusted life years [QALYs]. As different health states occur over time, we need to apply these utility values to create an accurate estimate of QALYs.
In summary, we need good data on the effectiveness of the treatments, on both beneficial and harm outcomes. We need cost inputs for the treatments as well as the events that are incurred or prevented over time. And then, we need utility values to get a good assessment of length and quality of life during the duration of the model. And these inputs, Dr Bhatt, can come from numerous sources that have a range of quality, starting with randomized trials and systematic reviews and meta-analysis of randomized trials, and then observational analyses, health care medication cost data, and qualitative data on health-related quality of life. And ideally, Dr Bhatt, we’d have high-quality data from randomized trials or meta-analysis of randomized trials for all of these inputs, but that’s often not the case. So to Dr Navar’s point, we’re often left to use data from lower-quality designs and sources.