So long Vioxx

Today Merck announced it is pulling its anti-arthritis drug Vioxx, its #2 seller (after Zocor), off the market. The reason? A study showed that patients who took the drug for more than 18 months had an increased risk of heart attack or other cardiovascular problems.

The irony of this is that the study was designed to show that Vioxx prevents recurrence of colon polyps--which might have led to increased sales. Instead, Merck is going to lose billions in future sales, and will almost certainly have to deal with lawsuits. [That said, I don't know how successful such lawsuits will be, since what I've heard doesn't suggest that Merck knew about this and intentionally withheld the information.]

What bothers me about the situation, though, is why this wasn't caught sooner; after all, the last stage of clinical trials is proving that a drug is both effective *and* safe; in a drug that is intended to treat a chronic condition (such as arthritis), it would need to be shown safe in the long run.

I'll admit I'm not an expert on the subject, but I do know that there is an inherent flaw in the system: the desire for safety requires as extensive testing as possible, while the patent laws give strong incentives to rush drugs to market (since the "clock" on the patent starts essentially from the day the drug is discovered). One potential solution to this problem would be to guarantee a drug company a certain amount of patent protection (say five years or so) from the date of regulatory approval, no matter how long it takes to get to market.

The other problem is that no drug is 100% safe; it's really a question of whether the benefits to a drug outweigh the risks. Clearly, in this case, given that other drugs do essentially the same thing, they don't. But, for a hypothetical example, let's say someone invented a drug for lung cancer that cured 50% of patients, but caused paralysis in 1 in 100,000 patients. Which is more important: maximizing benefit, or minimizing risk? It's something we, as a society, need to discuss, and preferably not with 30-second sound bites.