A lot of the medical studies that get published—even in respected journals—have faulty conclusions that are later proved to be misinformed or flat-out wrong. John Ioannidis, an associate professor at a Greek medical school, wrote a compelling paper about this phenomenon a couple of months ago in the Journal of the Public Library of Science. (As an example, Gelf found faults with a study that blames Hollywood for normalizing poor public health behaviors.) Gelf emailed anesthesiologist Joe Stirt about a recent posting on the subject he wrote for a section of his blog he calls "Behind the Medspeak."

Stirt, who practices at the Richmond Community Hospital in Virginia, was replying to a question from a reader who had heard that redheads need 20% more anesthesia than other patients. He took issue with the study that concluded this, noting the small sample size and the conflicting results found in a follow-up study. Stirt concluded, "The variation within any group, redheads included, is tremendous, way larger than the variation within a subgroup such as those with a particular hair color."

Gelf emailed Stirt with some follow-up questions. Here's the exchange, edited for clarity.

Gelf Magazine: What do you make of the fact that the doctors who did the first study got such a significant P-value for their results? [Editor's note: A P-value means, roughly, the probability that the results observed in a study could have occurred by chance.]

Joe Stirt: Nothing. Really small P-values with small samples have small meaning.

GM: Given that you treat all your patients as an N of 1, would you automatically disregard all attempts to generalize about anesthesia patients?

JS: Yes. Of 100 people randomly chosen, 1 cc IV [one cubic centimeter administered intravenously] of a non-depolarizing muscle relaxant has been shown to have no apparent affect on 10% of people, no noticeable effect on 25%, and noticeable effect on the other 65%. Of that 65%, one person will stop breathing and die unless supported until the drug wears off. Which one? You won't know by looking. You have to do the experiment. N = 1.

GM: Do you think that doctors who do such studies are wasting time and resources?

JS: No. Such studies are very worth doing because they do provide small steps from which big leaps, in the end, may be more easily made.

GM: Why is there such great variation between patients in terms of anesthesia needed?

JS: That's the best question of all and I haven't a clue.