It's highly likely that participants were unblinded by side effects, a common problem in many placebo-controlled trials that is rarely addressed. If you're a lucid dreamer, and interested enough to participate in a clinical trial, you might be more apt to report the outcome of interest during active drug (= side effect) periods. Note that this need not be a conscious decision, just an unconscious bias.
Many RCT's include some language about asking participants to guess treatment assignment - but this result is rarely reported...