In 1980, a letter to the editor of Archives of Dermatology described the apparent effectiveness of dapsone in two cases of disseminated granuloma annulare. Since that time, a few similar cases have been reported, each referring back to the original report.
There has never been a controlled trial using this potentially toxic agent in this condition. Nobody has really described a feasible mechanism by which dapsone might work in granuloma annulare. Yet in a 2009 eMedicine entry for this disease, dapsone is still listed, although not necessarily recommended, as an alternative therapy for this disorder.
At the risk of sounding dogmatic, I can assure you that the lesions of granuloma annulare may improve while the patient is taking dapsone, but the drug itself is probably of no value.
I use this particular article as an example of a problem that is extremely common in dermatology. Since we have relatively few drugs in our therapeutic arsenal and care for hundreds of skin diseases, those few drugs tend to get utilized in situations far afield from their original intended use.
Anecdotal data is published to validate what is often inappropriate therapy, and once in the literature or in the public domain after a lecture or seminar, it is almost impossible to expunge the spurious information unless there is some horrible side effect. Almost never will the author or lecturer follow up his initial observation with a statement that his initial observation was incorrect and that the previous recommendations should be ignored.
It is very easy to employ magical thinking to convince oneself that a treatment is the direct cause of a miraculous improvement, even after subsequent data does not substantiate this. About 15 years ago, there was a wave of enthusiasm for the use of cimetidine for warts. I vividly recall a 14-year-old boy with at least 50 warts whose lesions resolved while on high-dose therapy for eight weeks. Based on this one case, I was positive that cimetidine was the next great advance in wart therapy.
A few years later, a well-designed placebo-controlled trial of this drug in verrucae was published. It demonstrated conclusively that the medication worked no better than a placebo. To my dying day, my emotional self will believe that the drug worked, although that notion is highly doubtful considering subsequent real data. Case reports continue to be published describing the benefits of this H2 blocker. The authors of these reports describe "mixed results" in previous studies. No, there are no mixed results. There is not a single controlled trial that shows improvement that is statistically superior to a placebo, particularly in adults.