Statistics, drugs, and hard ethical questions

The New York Times has an article this morning on the FDA and drug approval process and some interesting controversy surrounding experimental cancer treatments. (Did you know you can get the NYT on the Kindle? Cool stuff)The FDA’s lead cancer guy is under attack from both sides of the debate. Some people say that he’s letting unsafe, unproven drugs get through and others say he’s holding back life saving treatments with unnecessary bureaucracy. Strangely, both camps are talking about the same drugs. It’s a pretty good article. It gets to the heart of the matter: Gleevec has obvious, amazing benefits. It goes through FDA review really quickly. Other drugs are subtle. And in that subtlety is the controversy.

Arthur Benjamin did a TED talk where he suggested that high schools forgo calculus in favor of statistics. I tend to agree – calculus is really important for scientists, but they can get it in their freshman year at University. To make sense of this and many other important ethical issues, everyone needs some statistics background. How many people need to suffer as a control group without treatment in order to assess whether a drug is subtly helping?This is a pretty hard statistical question. “Just look in your heart and your conscience will be your guide” just doesn’t cut it for these kinds of questions.

For instance: Aspirin seems to help prevent heart attacks. Out of 22,000 people, 56 per year had heart attacks with aspirin as compared with 96 not on aspirin. That implies that taking aspirin is a good idea, but without thousands of data points, it would be impossible to tell. If you only had 220 people, you get absolutely no conclusion. Look at it this way: if you take aspirin every day and don’t have a heart attack this year, you may be one of the 40 people who aspirin saved, or one of the 21,904 who wouldn’t have had a heart attack anyway! All you know for sure is that you’re not one of the 56 people who had heart attacks.

What we know is not what we think we know or what our gut instinct or common sense might tell us. At 546:1 odds against low-dose aspirin having any effect, it seems stupid to take it except that it’s so cheap it’s almost free, virtually no side effects, and heart attacks are serious as… um… well, they are really serious. If aspirin cost $10 per dose and caused erectile dysfunction, I doubt it would be worth taking. But what if it cost $1 per dose and sometimes (1:10,000) caused permanent deafness? Should your grandmother take it? Let your heart be your guide.