Saturday, April 09, 2005

Use the Right Number!
What do Tom DeLay and Bextra Have in Common?

Recently, Bextra was taken off the market because there is evidence that it could increase the risk of vascular disease.  That sounds like a sensible thing.  Vascular disease is bad.  Nobody wants that.  And I will trust that the number crunchers at FDA crunched the right numbers to come up with the correct recommendation. 

However, finding the right numbers to crunch is not always a straightforward process. 

Image that I told you that a widely-used drug is likely to increase the risk of death, not only from vascular disease, but also cancer, pneumonia, HIV, and meteor strikes.  It increases the risk that the patient will die from sticking a knife into a toaster to dislodge a stuck piece of bread.  At first glance, it would seem that such a drug should be removed from the market, immediately.  But in fact, there are many such drugs on the market, and rightly so. 

You see, everyone has to die from something.  So if a drug reduces the risk of death from one cause, it automatically increases the risk of death from all other causes*.  If you do not die from heart disease, you then are more likely to die from cancer.  Therefore, in assessing the safety of a drug, it is not meaningful to say that the drug increases the risk of death from some particular cause: any drug that reduces the risk of death from one particular cause will have such an effect.  The absolute risk simply is not the right number to use in this context. 

In the recent controversy about Tom DeLay, it has been revealed that his wife and daughter have received a total of ~ $500,000 from his various campaign funds.  The Left is all up in arms about that, because $500,000 is a big number.  The Right is dismissive.  Having done the math, they say that the half-million actually amounts to only about 40 to 50 thousand dollars per person-year . 

Neither side is using the right number.  What matters is not the total amount, nor the amount per person per year.  What matters is this: how much were they making per hour of legitimate work? If they made $50,000 per year for a 40-hour workweek, 48 weeks per year, that would be reasonable.  In fact, it might indicate that they were underpaid.  But if they only did a few hours of real work per year, then the same amount could indicate something really fishy. 

In this case, neither the total amount, nor the amount per year, is the right number.

How much do the Ms. DeLay's make per hour?  Is fifty thousand dollars chicken feed to them, or real money?  Did they actually do anything substantive to earn the money?  What did they do with the money?  Donate it to charity, or buy a yacht? 

Believe me, I would love to see Mr. DeLay loose his job over an ethics scandal.  After all, he once tried to con me into donating $300 to his campaign.  (I told the staffer who called me that I would pay him $300 if he could guarantee that he would never call me again.)  However, the fact is, we cannot tell -- from what has been reported so far -- whether the half-million dollars was inappropriate.  Nobody is using the right number. 

What we really want to know is this: what would it take for the FDA to declare Tom DeLay unsafe and ineffective?
* To be obsessively accurate, this statement is true only if all causes of death are fully independent of each other, which is not really the case, but the main point still stands.

UPDATE Here's another example of the wrong number, in this case, concerning pollution. A reporter does not understand the distinction between emissions and emissions intensity.