Quantifying medical research in QALYs
Posted to the personalized-medicine researchers’ list during the early days of MetaMed, the personalized-medicine consulting startup that Alyssa co-founded with Michael Vassar. The note walks new researchers through the standard cost-benefit framework for evaluating diagnostic tests and treatments: list the expected gains and losses in quality-adjusted life years (QALYs), and try to put numbers on each one even when the numbers are rough.
Some initial thoughts on quantifying research. These are just my thoughts, and don’t represent official policy.
Classically, the value of a medical intervention is measured in quality-adjusted life years, or QALYs . E.g., if an intervention gives you another year of healthy life, that’s a benefit of 1 QALY. If it gives you another year of life, but makes you bedridden and in the hospital half the time, that may only be a benefit of 0.5 QALYs.
When you recommend a test or a treatment, it’s important to consider all the costs and benefits of that test/treatment, and to quantify them if possible. It’s difficult to get numbers that are exact, since the QALY is somewhat subjective. However, that shouldn’t stop us from quantifying; there are many cases where a decision is obvious (one way or the other) by many orders of magnitude. The important thing is to get down some number, even if the number is imprecise. I’ll go through some of the relevant factors, using colonoscopy and some made-up numbers as an example.
On the benefit side of the calculation, the important things to consider are (a) how common the disease is to begin with, (b) how often the test catches the disease, and (c) to what extent catching the disease leads to beneficial treatment. Suppose (making up numbers) that 5% of people get colon cancer. Regular colonoscopy catches colon cancer 50% of the time, and this early detection (on average) extends healthy lifespan by two years, by enabling more effective treatments. The expected benefit of regular colonoscopy is then 2 × 0.5 × 0.05 = 0.05 QALYs.
On the cost side, the largest factor to consider is often the time cost of the test. E.g., suppose that a colonoscopy takes a full day of being in the hospital, and that it’s recommended every five years, starting at age 50. The average 50-year-old lives to about 85, so that’s a total of eight tests. 8 × 1 / 365 = 0.022, so that’s a cost of about 0.02 QALYs.
Another cost factor to consider is preparation and recovery time, and risks, if any. Some forms of chemotherapy, for example, require extensive recovery lasting weeks or months, during which quality of life is reduced. If a chemotherapy treatment has a 1% chance of killing the patient, and the average cancer patient has five expected healthy years remaining, then this is also a cost, of 0.01 × 5 = 0.05 QALYs. For colonoscopy, suppose that it has a week-long recovery period, and that this recovery reduces life quality by 10%, due to pain etc. The cost for a lifetime of tests is then 8 × 0.1 × 7 / 365 = about 0.015 QALYs.
Finally, it’s important to consider the financial cost of medical intervention. A lot of things are covered by insurance, so they might have little cost to the patient. On the other hand, it’s common for, e.g., some cancer treatments to cost hundreds of thousands or even millions of dollars out of pocket. Financial cost can be approximately mapped to QALYs by looking at hourly wages, which average about $20 for an American adult, but which for our clientele are probably more like $50. In this case, suppose that colonoscopy has an average cost, after insurance etc., of $1,000. Eight tests are then $8,000, which equates to 160 hours = ~10 days / 365 = 0.03 QALYs.
In this scenario (again emphasizing that the numbers are made up), the total expected benefit is 0.05 QALYs, while the total expected cost is 0.02 + 0.03 + 0.015 = 0.065 QALYs. The expected costs and benefits are close, neither clearly exceeding the other, so one might say that the net benefits of the test are questionable or ambiguous or something like that.