When I was an undergraduate, I began working with a neuroradiologist at Yale, looking at intercranial aneurysms. We had a conundrum: up to 3–7% of the general population have intracranial aneurysms and aren’t aware of them unless they get a CTA of the head. Sometimes they’re completely asymptomatic and the aneurysms are often small (2–3 mm).
When these aneurysms are found, the patients may be referred to neurosurgery or neurointerventionalists and are told they have an aneurysm that could rupture at any time. Do they want to treat it?
Treatment is not without risks either. Physicians must go through vessels delicately to coil the aneurysm, which can cause complications, or the patient may undergo open surgery to have the aneurysms clipped. Alternatively, a patient could elect to undergo surveillance of the aneurysms with yearly imaging to track the size or changes in morphology. If so, how often?
It’s a complicated question, and a difficult decision for patients and physicians alike, given all the variables involved. During this time, I came across literature regarding cost-effectiveness analysis and felt it could be a useful model technique to tackle this problem. I took a class at the Yale School of Public Health, purchased related software and began using it—and discovered a powerful tool to help improve both treatment planning and patient consenting.
What are cost-effectiveness analyses?
Many people think cost-effectiveness analysis is a complicated mathematical concept, but it’s actually quite simple and straightforward. In essence, you’re comparing multiple strategies.
For ease of illustration let’s say you are looking at two strategies, A and B, to determine their costs and effectiveness. You pick strategy A as a reference strategy and determine if strategy B has achieved an additional benefit for the patient. Then you look at the cost difference. If that added benefit costs $1,000 more, is it worth it? What about $1 million more? Basically, cost-effectiveness is trying to answer that question.
There are multiple ways to calculate cost and effectiveness—and that’s where the complexity can come in. Typical methods include using a cohort study, wherein you look at patient cohorts, survival rates and/or quality of life measurements, and the total medical expenditure of each cohort. Then you average that out in cohort A and cohort B and compare the costs and differences in outcome.
I have more experience in using a modeling-based strategy to calculate the cost and effectiveness. This model is important when it comes to making treatment decisions for patients who have multiple treatment options available—which is common, given that we have so many advances in the field of IR that are effective alternates or improvements on surgical procedures. Cost-effectiveness encapsulates all the possibilities to provide an overall assessment of what is the optimal treatment for these patients.
The benefits
It can be very difficult to do head-to-head comparisons for some treatments. Randomized controlled trials are very expensive, and some diseases might be rare enough that it’s hard to gather enough patients. In addition, there may be ethical challenges in doing true randomized controlled trials if you know one strategy may be better than another. Cost-effectiveness modeling enables you to simulate a large cohort of patients—even multiple parallel cohorts of patients—going through the different pathways and determining their cost and outcome. I think that’s powerful.
This model cannot replace a randomized controlled trial or large, prospective study, but when gold-standard research is inaccessible, we can consider cost-effectiveness or modeling as a supplement to the topic.
Personalizing the analysis
In general, when a cost-effectiveness analysis is published, it has an overarching conclusion. But I don’t think one should just look at the conclusion. Most papers also include sensitivity analyses, including probabilistic sensitivity and deterministic.
Probabilistic sensitivity analysis basically tells you how confident the model or the authors are about this conclusion and how robust it is against all the input variables. If the range is small, we can feel confident that strategy A has a high likelihood of being the better strategy for most patients.
Then one should look at the deterministic sensitivity analysis, where we vary one variable across a wide range to see how it affects the conclusion. This is usually a key variable that can be very patient-, practice- or experience-dependent from the physician. Using this information, IRs can look at their own practice and see, in their experience, where their patients land and what the approximate value of that key variable is. This enables IRs to utilize their own demographics and see how they trend, and then determine the value of a strategy in their own practice.
Recent application
We recently used the cost-effectiveness analysis model on benign prostatic hyperplasia (BPH) and published our studies in the Journal of Vascular and Interventional Radiology. We compared prostate artery embolization (PAE) and transurethral resection of the prostate (TURP). TURP has long been regarded as the gold standard for benign prostate hyperplasia, but it is a surgery and there are risks. Meanwhile, PAE is an up-and-coming treatment that offers a minimally invasive option.
We compared the two strategies for treatment of BPH that does not respond adequately to medication alone and concluded that PAE is more cost-effective for this patient. Not only is it more cost-effective, but it achieved better effectiveness at a lower cost. We call this the dominant strategy, because it provides patients with better outcomes while saving money in the healthcare system. We then varied all variables for PAE, and even with extreme variable values that were outside the reality of a clinical setting, PAE was still more cost-effective.
This kind of analysis and data can be very important, both for creating a treatment plan, but also for showing the value of various procedures to patients and administrations.
To learn more about the PAE study and application of this cost-effectiveness model, read the JVIR article, “Prostatic artery embolization versus transurethral resection of the prostate for benign prostatic hyperplasia: A cost-effectiveness analysis.”
