There are indeed likely "practice effects". That paper does not convincingly show that such effects would wash out the difference in results between a placebo and an effective treatment. And in fact there's rational reason to think the effect would be greater for an effectively treated population. After all, cognitive improvement would also aid in better learning.
Practice effects in AD have not been discussed often.
Great points on practice effects. Its clearly a weakness of having an open trial to not be able to filter these out. I think AVXL tried to address this to a degree with their slide 27 from CTAD where they compared to their Cogstate results with treated patients in past trials (Lim YY et al (2013) Arch Clin Neuropsychol. Jun;28(4):320--30 - I have not pulled this reference so my comments only relate to what is on their slide). A2-73 seems to do better than their reference data in some measures. However, the reference data shows that for the measures in which 2-73 did best, highest reference performance was at the latest date (52 weeks) suggesting possible practice effects. This is especially true because we know that standard cognitive measures would have fallen during that time. OTOH, A2-73 data is best at the earlier testing points. All quite muddled I'd say but I certainly wouldn't hang my hat on the Cogstate testing.
Other aspects like frequency of testing are great points and can only be solved by a placebo control.
Market Data 
Markets 
AI
