I met someone who worked on the admin side of NHS trials. They said exactly what you posted, anything noteworthy that happens should be reported. If you take the placebo and get hit by a bus outside the hospital, it should be reported. That way if 10% of patients in the study get hit by a bus they have more data.
Obviously a silly example but if things aren't reported because they don't appear to be connected to the study, we don't have the data to analyse at a higher level.