It's called Bad Medicine, by David Wotton - he has a website, which I shall now fail to link to (still mastering lj), at Bad Medicine (http://www.badmedicine.co.uk), which includes a link to the TLS review whence I picked up on its existence. As I said, it's depressing about how doctors (just like everybody else) prefer their own firmly held beliefs to statistically valid evidence - there's an early study that showed how incredibly effective bleeding was at speeding recovery from pneumonia... at least, that is, until you realise that the patients bled earliest and heaviest were also the youngest and most healthy.