One of the most important tenets of data analytics is that "the plural of anecdote is not data". As has been mentioned previously in this thread, those that you tend to hear most vocally and frequently from are those that have had some sort of adverse reaction. Those that have had no issues whatsoever, tend to be very quiet as they really don't have anything to report. The frequency of hearing from individuals that have had adverse reactions are being weighted far more than they should (i.e. sample bias).

But here is some specific data that you can look at that should help clarify what the real probabilities of various types of adverse reactions, or side effects, are: