Hm, well actually, since someone posted the PDF above, seems you're right. Decaf did quite well (although perhaps noisier due to smaller sample size). I do won

Hm, well actually, since someone posted the PDF above, seems you're right. Decaf did quite well (although perhaps noisier due to smaller sample size). I do wonder about some of those error bars and controls though. For example artificially sweetened coffee decaffinated at 1.5-2.5 cups per day seems to have a higher rate of death (1.11) - but artifically sweetened decaffinated at 3.5-4.5 cups has half the rate of death (0.52). By contrast black decaf at 3.5-4.5 per day does worse than that (0.66). Anyway, if there's any significance to this, you're right that caffeine does not seem to play into it much in their stats.