In my mental cabinet of things available to recite on cue, I would like to have an ample supply of facts, statistics and poems. Currently I have none of these (though I do have an embarrassingly fantastic larder of 80s music). But it’s never to late to start, right? So here’s a a statistic that many might find juicy and ripe for the squeezing: In the United States, healthcare accounts for 17.6% of the gross domestic product. And barring any truly radical changes, that number is expected to continue a steady increase in the coming years.
Make of it what you will. For some people—such as David E. Cohn and colleagues of Ohio State University and the University of Alabama—it means figuring out a way to factor cost into the considerations of benefit. Cost-effectiveness analyses are nothing new, of course, but yet there is still no clear method to accurately and fairly weigh the dollars paid per ounce of cure.
With an aim toward making some small step forward in that direction, Cohn et al tried analyzing the cost-effectiveness of Avastin in the treatment of ovarian cancer. (I know I’ve written about Avastin here before, and I don’t mean to pick on this drug, as if it is the richest child on the fancy pharmaceutical playground. This kind of analysis works well here because the high-priced drug acts almost like a caricature for cost considerations in general.)
Here’s a summary of their approach. Keep in mind, this has no bearing on the world right now, is kind of hypothetical, and was done *solely* for the purpose of showing how one might go about looking at the costs of a particular treatment. Still, it makes for a fascinating read (if you’re into this kind of thing).
The researchers used data from an ongoing clinical trial run by the Gynecologic Oncology Group, a government-funded clinical trial cooperative. Results reported in abstract form in 2010 served as the basis for the cost analysis. The study had three treatment arms, with 1,873 women with stage III ovarian cancer randomly assigned to receive: (1) conventional chemotherapy (PC, for paclitaxel/carboplatin); (2) chemotherapy plus Avastin (PCB); or (3) chemotherapy plus Avastin, plus maintenance Avastin (PCB-B). Using actual and estimated costs for the treatments and for associated complications, the researchers calculated the cost of care per “progression-free life-year saved,” or PFLYS.
First: what is PFLYS? Progression-free survival is a key measurement in clinical trials because it describes the amount of time a patient lived before the disease began advancing. In other words, it gives an indication about the amount of treatment-free time granted to a person after undergoing treatment. Once the disease progresses, we know that an individual is likely receiving further treatment, perhaps living less comfortably, and perhaps with a poorer prognosis. That’s PFS. The measurement also gives an early indication of a drug’s efficacy, because why wait to evaluate overall survival time (months or years later) if there are patients who could benefit from the treatment.
PFLYS is looking at how much life without disease was saved on account of the treatment. The authors explain their decision to use PFLYS as their measure of effectiveness: overall survival time can be complicated to model in this way because treatments become increasingly individualized as the disease advances. They acknowledge that using progression-free time likely underestimates the total amount of survival time made possible by a particular intervention, but it does give a sense of a treatment’s overall impact.
But how do you know if a certain benefit is cost-effective? There has to be some guidepost, some dollar figure that people generally agree is a bottom-line measure of cost-effectiveness. That seems like a really hard thing to calculate. Actually, that seems like the thing that needs to be calculated before we can ever factor cost into treatment considerations. And that seems like a really big hurdle to get over. But, lo and behold, this figure has already been determined. Shows how much I know. So here’s another interesting figure for the mental shelves: an intervention is generally considered cost-effective if the ratio comes out to less than (an albeit controversial) $50,000 per quality-adjusted life-year saved. There you have it: a year of life is worth $50,000 in medical bills.
While you stew over that one, here is the result of the analysis: Avastin is not a cost-effective way to treat ovarian cancer. (Is anyone else hearing the “take a seat on the stool” music from American Idol right now? Have I said too much?) The drug would be cost-effective if the price were reduced to 25% of the baseline amount.
Regardless of the conclusion, the numbers alone are meaningful, especially since we don’t normally traffic in these kinds of figures when thinking about drug prices or our own healthcare. PC alone cost $440 per treatment cycle. PCB cost $6,180 per treatment cycle. And the extra B maintenance added an additional $5,740 per treatment cycle. Patients who received PC alone had a PFS of 10.3 months, versus 11.2 months for PCB and 14.1 months for PCB-B. (The costs are estimated using Medicare reimbursement rates.) Importantly, the 3.8 months of additional PFS time were considered to be statistically significant in the original abstract, which speaks in favor of it being added to the roster of treatments for ovarian cancer.
Overall, the model estimated the total cost of PC to be $2.5 million. PCB costs $21.4 million, and PCB-B costs $78.3 million. *—pause for you to read those numbers again —* One small conundrum is that the cost-effectiveness ratio for all of these approaches was way above $50,000 per life-year saved. The PC ratio was $247,616 per PFLYS.
The authors hypothetically extended the survival time from those in the original report, and found that an additional 10 months of survival time would bring the cost-effectiveness ratio of PCB down to $35,976. An additional 18 months of life brought the ratio for PCB-B down to $106,836. Of course, running a much lower price tag through the analysis also had a dramatic effect on the ratios.
There are many more subtleties included in the analysis, which, the authors continually emphasize, is hypothetical and limited. But the approach provides a solid step forward in thinking about how to think about cost. Ovarian cancer is a terrible disease that will be the cause of death for over 13,000 U.S. women in 2010. Most women with ovarian cancer die of the disease. The potential addition of Avastin to the therapeutic armamentarium has been lauded, and no one would argue against having more treatment options. But the stark numbers here — less than six months of additional time without encroaching disease paid for with tens of millions of dollars — are, at the very least, something to mull over. But this study isn’t even about concluding whether or not Avastin should be included among the treatment options for people with ovarian cancer – it’s just showing a way that cost can be considered alongside benefit. It’s a useful demonstration.
(The authors of the analysis disclose no potential conflicts of interest; one author received funding from Genentech, which sells Avastin.)
Medicine cabinet: http://www.flickr.com/photos/the_dan_driscoll/5351353700
$3,500 prescription: http://www.flickr.com/photos/calliope/2989567835/