One of the task's of the U.S. Food and Drug Administration (FDA) is to approve novel drugs. It has often been mentioned that drugs are not compared to each other head-to-head, as would be desirable for comparative effectiveness. Mostly, drugs are still compared to placebo.
However, the FDA is also responsible for pulling unsafe drugs off the markets. And in doing so, the FDA has one means of comparing the risks of particular drug to competing drugs.
Example: Recently, the FDA has investigated whether taking the diabetes drug Avandia(R) (rosaglitazone) is associated with an increased risk for myocardial infaction (heart attack). After an FDA panel recommended to pull the drug (here's a Kaiser Health News summary for advisory board meeting), the agency still decided to keep it on the market. What's interesting, however, is that the committee was flipping back and forth over the decision amidst the various aspects: efficacy, safety, and also competing drugs. A blog post by the NY Time's Precription blog chronicles the course of the meeting. Most interestingly, in the 1:17 pm entry, a Dr. Robert Temple stated that the FDA is supposed to pull drugs if "it’s nearly identical with another drug that doesn’t have such a serious safety problem". An example cited was Seldane(R) (terfenadine), an anti-histamine that was pulled because it may arrhythmias and there were competitor products without that safety issue on the market.
So if adverse events are part of the effectiveness considerations, this means that the FDA actually does have a way of assessing if not the benefits then at least the risks.