Thursday, December 07, 2006

More on Lexapro Vs. Cymbalta

Another press release on this study (see yesterday’s post for more details) indicated that the difference between medications on the MADRS was 2.42 points, with an associated p-value of .04. That answers my question as to whether the mean difference was significant – it was. Super. According to the press release, there were 137 patients in the Lexapro group and 133 in the Cymbalta group. With the above mean difference and p-value, that yields an approximate effect size difference of (drum roll, please) .25, which is a small effect size difference. Of course, maybe the good folks at Forest Labs used an alternative method to calculating effect size, such as used by AstraZeneca in the BOLDER II study. That seems to be a good way to inflate the apparent effects of a treatment.

Additionally, I see exceedingly few antidepressant trials that use one measure. I am willing to bet there was another measure of depression, such as the Hamilton Rating Scale for Depression. The Clinical Global Impressions Scale or some other more general scale was also likely used. Note that in the press releases (here and here), there is no discussion of other measures, meaning that either this was a strange study that opted to only use one rating scale, or that the data are being selectively reported in the press, so that by the time the data hit a journal, the medical community and many potential patients already perceive that Lexapro is a superior medication. I’ll be willing to eat a huge slice of humble pie if I’m wrong on this.

Oh, one more thing. According to today’s press release, Lexapro patients had flexible dosing whereas Cymbalta patients did not. Hmmm. .. I wonder if that might make a difference in results? It absolutely pains me to defend Cymbalta. Really. But when study results trickle out to the press in such a selective manner, something needs to be said. We have the likely selective reporting of measures, getting excited over a small difference, and employing different dosing strategies between groups -- all the hallmarks of marketing trumping science.

No comments: