In yesterday s Dispatch, we discussed a study finding that silymarin, a widely used extract of milk thistle, has no effect on those who take it to relieve hepatitis C symptoms. While doubtful of silymarin s therapeutic properties to begin with, we noted the study s limitations specifically its inability to be widely generalized. Now, however, we stand slightly corrected.
In yesterday s Dispatch, we discussed a study finding that silymarin, a widely used extract of milk thistle, has no effect on those who take it to relieve hepatitis C symptoms. While doubtful of silymarin s therapeutic properties to begin with, we noted the study s limitations specifically its inability to be widely generalized. Now, however, we stand slightly corrected.
One of our readers, Hoover Institution Fellow and former ACSH trustee Dr. Henry Miller, has written to emend our assessment of the study:
The issue isn't whether you can "generalize" the results; rather, it's how confident you can be about the conclusion that there's no effect. Assuming that the study design was otherwise valid, the small sample size would have detected efficacy only if the therapeutic effect was large. In other words, you don't have much statistical power with a small study.Consider, for instance, that the synthetic version of the human growth hormone in the 1980s contained data on only 28 patients. In that case, there was sufficient statistical power because all the patients responded and one was indeed able to "generalize" the results.
It looks like we can maintain our doubts about silymarin after all.
