Are $100,000-plus ICU drugs worth the cost?
The problem with cost modeling is that it is sensitive to input – garbage in means garbage out. Models are hampered by the limitations of what can be accomplished in clinical trials and by the assumptions made to create the model. Is it based on acquisition costs or on charges? How do practice variations within the institution (between the medical ICU and the surgical ICU, for example), or from one institution to the next, affect the assumptions in the model?
Cost comparisons must somehow reconcile data from conflicting studies and multiple treatment options, and they must be able to quantify the effects over time of a given intervention.
Clinical trials that are used to justify cost evaluations are performed in controlled settings. I wish that I could say that all my patients were uniform, but they are not. There is a substantial difference between a 24-year-old with a head injury and a 90-year-old with a head injury, or between a patient with a broken hip and one with end-stage liver disease. How do you begin to compare the effects of drugs across these populations? How do you judge the effect of a specific drug in an era of polypharmacy, when some patients seen in our unit have been on as many as 27 drugs at the same time?
Time can also fade the bloom of a once-promising drug, such as drotrecogin alfa (Xigris). An analysis of the PROWESS study (Crit. Care Med. 2003;31:12-9) appeared to show that drotrecogin alfa was helpful in patients with severe sepsis, and this drug’s alleged efficacy was balanced against its cost, which was estimated at $48,000 per QALY and considered to be in the acceptable range. As we all know, longer experience with this agent found no overall benefit and an association with higher bleeding risk (Cochrane Database Syst. Rev. 2012 Dec 12;12:CD004388). The drug was pulled from the market in late 2011.
Dr. Kane-Gill correctly notes that factor VIIa, when used for its rare indication, offers a benefit and saves money. But many ICUs were using this agent outside of its labeled indication (Ann. Intern. Med. 2011;154:516-22) and found that it caused emboli in patients already at risk for thromboembolic events (J. Trauma. 2010;69:489-500; N. Engl. J. Med. 2008;358:2127-37).
If a drug is in the formulary, and if it is new and more expensive than other therapeutic options, it is human nature to think that new and expensive must mean better than old and cheap. In Europe, ICUs routinely use injectable acetaminophen for nonopioid pain relief. In the United States, the intravenous formulations of this drug are costly – about $10 per dose, compared with about 50 cents for one dose of percocet. Yet the reduction in opioid use with the injectable agent is modest, there is no evidence to suggest reductions in opioid-related adverse events, and there are conflicting data on patient satisfaction (Cochrane Database Syst. Rev. 2011 Oct 5;(10):CD007126).
Economic evaluations of drugs should be viewed with caution because they require knowledge of the literature basis for the model, may exhibit considerable variation in estimates of indirect costs, and are subject to the vagaries of conflicting data as well as changes in data and practice over time. Additionally, estimates of drug cost and benefits can vary by country, region, or facility.
It’s clear that we need much more data about outcomes and the true costs of the care we give.
Dr. Christine C. Toevs is a trauma surgeon and director of the trauma ICU at Allegheny General Hospital in Pittsburgh. She also has a Masters degree in bioethics.
