Thursday, September 17, 2009
Patients who take the powerful new multiple sclerosis drug Tysabri are getting mixed messages about their risk of contracting a life threatening condition called progressive multifocal leukoencephalopathy, or PML. PML is a brain infection associated with the use of Tysabri. It appears that information regarding the level of the risk of PML given to patients on Tysabri depends on who they ask. As Keith Winstein of the Wall Street Journal reports:
Tysabri underscores a central mathematical issue in assessing risks and benefits of medical treatment -- one that has also shown up in calculating the risk of blood clots among heart-stent users, and in figuring out how beneficial chemotherapy is in treating lung cancer.
Simply put, the issue is a matter of whether to adjust for time. In other words, should the chances of contracting a harmful side effect be calculated by figuring out the simple percentage of all those taking the drug who have come down with the side effect? Or should those calculations be adjusted for the duration that patients have been treated?
The former method calculates what's called an "absolute" percentage. The latter, used widely in medical studies and by insurance actuaries, takes into account that risk changes over time: For example, someone who drives a car only one day during his lifetime is less likely to be in a crash than someone who drives for 20 years.
Using the absolute method, the manufactures of Tysabri report that the risk of PML is one in 8,700. The actuarial method results in a risk of one in 1,200 -- much closer to the one-in-1,000 threshold that may cause doctors to become more cautious in prescribing the drug.
The two mathematical methods in calculating risks have led to prior disputes. In 2006, Boston Scientific Corp. and Johnson & Johnson disagreed over whether their models of drug-coated heart stents -- tiny scaffolds that prop open clogged arteries -- caused blood clots years after implantation.
According to this WSJ article, J&J used the absolute method and reported no increase in the percentage of patients with clots in its coated stent. Boston Scientific applied the actuarial method and reported an increase in clots with the use of its stent. This led to an impression that J&J’s stent was safer than Boston Scientific's stent. The New England Journal of Medicine analyzed both stents with the same technique and discovered the same number of clots with both stents.
Similarly in 2005, a Canadian clinical trial inaugurated the use of chemotherapy to treat lung cancer after demonstrating that the treatment reduced mortality by 31%. Half the people in the study were given chemo, and half weren't.
But the results didn't mean that chemo decreased actual lung-cancer deaths -- only that it would extend the life of patients. In fact, after eight years, the study estimated that the same fraction of people would die with treatment as without -- about half. Since people on chemo lived longer, by about 21 months, the study could truthfully report that treatment reduced the average death rate during the study. (emphasis added).
As patients with chronic medical conditions are living much longer lives, it appears that long-term exposure levels as reflected in the actuarial model should be used to quantify the risks associated with treatment options so that both physicians and their patients can make informed treatment choices. [KVT]