Filmmaker wrote:that means it's actually 30% of 33% added to the 33%... which means 42% but 42% doing what? reducing what exactly? really...

I'm not sure that I follow your math. In the example I noted above for the Avonex study, there were 21.9% of patients progressing in the Avonex treatment group and 34.9% of patients progressing in the placebo treatment group. The relative percentage difference is calculated as follows.

(34.9-21.9)/34.9 = 37.2%

However, the absolute percentage difference between Avonex and placebo regarding disability progression is only 13%. In my opinion, this is the number that really matters.

The use of relative percentage calculations in medical studies has been criticized . Consider the following example which is similar to what I have read about the studies which led to the approval of one of the statins (if I remember correctly, it was this book; "Overdosed America: The Broken Promise of American Medicine" by Dr. John Abramson).

If 4 out of 100 people in the placebo group have a fatal heart attack while only 2 out of 100 people in the statin treatment group have a fatal heart attack, then the pharmas can say that the statin led to a 50% reduction in fatal heart attacks. However, the 50% figure is a relative percentage difference between the two groups. The absolute percentage difference is only 2%.

Here's the best reference on relative percentages that I was able to find via a quick Google search.

Interpretation of Medical Statistics by Leslie E. Daly, Geoffrey Joseph Bourke.

NHE