First, try this:
http://tinyurl.com/zsvop
There are several researches on this topic by IPI, european groups and testing run by myself. There are some speculations on the possible changes in the formulation of KRST (search for the old pure-silver list) but I now think the real reason is elsewhere.
There are a number of reports from early 1980s and earlier, proving dilute KRST treatment was completely effective with microfilm of that time. But later 1980s the same treatment was not as reliable as the older tests, again, with the emulsion of late 1980s and 90s.
There are two differences I see in these studies. One is that the standard condition for accelerated testing became harsher and harsher. Many early tests were run at 100ppm or 500ppm peroxide. Recent ones are run with 1000ppm or sometimes 3000ppm or even higher. Samples are usually incubated in a constant humidity and temperature for several hours to days. So, one possible difference is that, something that was effective against 100ppm peroxide is now ineffective against 1000ppm peroxide.
Another difference is in the way emulsion is made. During 1980s, commercial emulsion production methods saw a lot of technological advancement. Microfilm emulsions, being slow, high contrast and requires high resolution, were the first product category to apply a lot of new technologies, such as tabular grain emulsion. The frist group of t-grain products were released around 1983, including Kodacolor VX-1000, medical X-ray film, and microfilm (all Eastman products).
I think both of these difference gave rise to apparently contradicting results from early and late 80s to 90s.
I have run some tests with AGFA MCC, AGFA MCP and Forte Fortezo papers using 3000ppm peroxide concentration. Polysulfide treated prints were most resistant to peroxide attacks across the density range, from light to dark.
Some published tests included enlarging papers. One was the work at Smithsonian, another was the work of Harald Sorgen at University of Applied Sciences Koln. Maybe there are more if I look up my file. The reason why microfilm was studied is because of the availability of research funding and perhaps other factors.
Anyway, it's not that selenium is not effective at all. It is somewhat effective but not as effective as polysulfide.