Initial results from HSAs are impressive. I have published articles summarizing these in HR journals hoping to "open" the minds of those who buy our health insurance for us. Having said that, let's be careful here. What we really need to study HSAs is a large longitudinal data set of medical claims before and after selection of HSA plans. This would allow for examination of how much utilization change they cause, their impact on overhead costs and whether they are generating any signficant adverse selection. Getting this data up till now has been impossible but professors like me will continue the hunt!
Do read this paper! A good antidote for all the gloom and doom. An added bonus is that it has a handy summary of the faults in the HSA isn't workig literature.