>he act of qualifying for the trial can confound the results The example you give doesn't confound the result of the trial (provided it's properly randomised) b

>he act of qualifying for the trial can confound the results The example you give doesn't confound the result of the trial (provided it's properly randomised) but it does make it hard to generalise the result to the general population. What you describe is actually common — antidepressant trials typically exclude people with comorbid medical or psychological conditions, despite the fact that the rate of comorbitity in depression runs at >60%.