Very much depends on the audience of the website. For American users, there is a good statistics
here (about 55% for August). On the Internet it seems that the figures are less because of the greater prevalence of bannerrotator and paid traffic There are still riastats.com but there the sample is statistically limping. Almost 80% of the data from one site.
\r
Empirically, these data can be corrected by share of alternative browsers. If the proportion of use Firefox and chrome among your visitors higher than the average for the hospital, and silverlite will be better in about the same degree. For example, on one bourgeois site for webmasters, the share of silverlite about 80% but IE's hanging in third place after Fox and chrome.
\r
The number of failures depends on how the user needs your website. If this is not some unique service, I think, in the elegant case, silver will put one in five, the rest will go. To do on SL (flash too) the entire site, I would not recommend it.