What is a significant reduction in CAM usage?

In science, we generally decide in advance that an experimental observation is significant when it is unlikely to occur by mere chance less than one time in twenty observations.  Thus, the P<0.05 rule that frustrates us when P=0.051, and the next repeat of an experiment shifts it to P=0.1.  The next experiment will probably push P to 0.2, because the first couple of repetitions were total flukes.

In clinical trials, the same rule applies, but we also have to consider if a finding is clinically significant. A different approach might be significantly better than another, but how many patients would actually benefit, and how much of a difference would it make to their lives? A statistically significant increase in life expectancy of 1 hour is hardly worth taking a drug worth £5 per dose, three times a day for thirty years.  Some drugs cost tens of thousands of pounds per year, and can extend lives by as little as a month on average.  Poor old NICE has to make these decisions about cost:benefit ratios.

So, what kind of change in public perceptions of quackery is significant? I'd guess that if 75% of people agreed that something as stupid as homeopathy is in fact stupid, you'd be onto a win.  Recent experiences with my supposedly clever students (via @DrAust_PhD) suggest that this is about the best that can be achieved, even with fairly intensive, didactic instruction.

So, if 45% of people have resorted to CAM, and of them only 18% of people have used homeopathy, hasn't the battle on that front pretty much been won?

Or, should we be aiming for a reduction of total CAM usage down to 25% or less? 

Feel free to argue the toss.


See the Wellcome Monitor, again, for the numbers.