In this case, they went for a placebo set-up because they were also trying to grab a little data to show whether 25 mg was significantly distinct from 10mg. The comparison to placebo was not needed, but is winning them a little extra publicity.
Note also that the study probably include some regular visits to the hospital. Some people may eat more healthy food a few days before a blood exam so the doctor gets happier. Some people may change other habits, like no binge drinking. So you need to measure the results in a group that follow the same explicit and implicit rules.
Does the report of the symptoms use the same criteria in the study than in general population? What about dizziness? Does this produce dizziness? Does the general population report dizziness to their family doctor? Replace dizziness with chest pain, or constipation, or whatever symptom you are interested.
Also, people lie to their doctor because they try to be nice. It's not a voluntary lie, but if there is a study about curing something, people want to be nice and give optimistic feedback. You must have a double blind control group so you can compensate this.