Ascertainment bias tends (in clinical trials) to refer to biased outcome measurement when the person doing the measuring (like the physician) has knowledge of the treatment a patient received. Imagine a study in dermatology for a new ointment to treat a skin rash vs a placebo. If the dermatologist assessing the improvement knows who received what treatment, it's possible that the placebo group gets lower improvement ratings and the new ointment group gets higher improvement ratings than they otherwise would if the outcome assessor had no knowledge of who received what treatment. (This is one of a few ways this kind of bias could occur.) Long story short, outcomes are systematically shifted from the true values, on average, because someone had knowledge of the treatment. Be careful, though as the definition does depend on context (in some cases being as simple as general sampling bias).