On Tue, 7 Mar 2006, [EMAIL PROTECTED] went:
Their only evidence that it worked was that the number of migraine days during a four week period before randomization (approx 1 per week) was about 50% greater than at the end of treatment, 22-26 weeks later.
And I'm not so sure that even that statement is true. If I'm reading their flowchart correctly, their dropout rate was at least 20% (perhaps greater, depending who's counted as a dropout). How did they handle all those missing data in their intent-to-treat analysis? Like this: "Missing data points (which appeared only at 13 weeks)...were replaced according to the principle of the last observation carried forward." So patients who dropped out on week 13 of a 26-week study were assigned 13 more weeks of data with no variance? And the degrees of freedom for the analyses were calculated as if the sample size wasn't shrinking? That doesn't seem like a sound basis for assessing change over time. --David Epstein [EMAIL PROTECTED] --- You are currently subscribed to tips as: [email protected] To unsubscribe send a blank email to [EMAIL PROTECTED]
