One of the basic tenets of diffusion theory for the last 50 years has been the premise that opinion leaders are key to the dissemination of ideas and the rate of the adoption process. All this has been called into question by the research carried out by Duncan Watts at Coumbia University (and interestingly I instinctively wrote about my issues with opinion leaders not working well in my book "Undressing the elephant; why good practice doesn't spread in healthcare).
Duncan's work suggests that although the influencer hypothesis has some merit under most of the conditions tested the influencers had less influence than expected. In fact what he found was that it appeared adoption was helped by a critical mass of easily influenced individuals - turning the perspective around.
I did a brief and incomplete literature search of publications with "opinion leader" in the context of healthcare for the last 2 years. I found 18. Interesting to note that of these half suggested their experiment of using what they called opinion leaders worked in disseminating information and half said it didn't work.
So I guess the jury is still out. Though after 50 years it is good to see some new thinking.