As Robert mentioned, I think the main thing is that the level of competition has increased to an amount of commitment beyond what most people are willing to put in.
I was a coach at the ACM ICPC in 2005 and heard a similar story about the Shanghai Jiao Tong University team practicing all day for the entire previous summer. When I was on the Caltech team, the team members weren't decided until a few months before the contest, and we only practiced a few hours a month, mainly because everyone on the team had other competing demands on theirtime.
In the US, the best students have opportunities to do research with professors that will lead to them basically being guaranteed admission into a top graduate school,
they have easy access to internships at the top companies, and they don't have as much to gain by winning. In other countries, especially with worse economies, there aren't as many good alternative ways to spend time, and doing well can significantly increase everyone's opportunities. The cultures in those particular countries also reinforce math and science and increase the pool of potential contestants.
I think this is mainly true of ICPC - if you look at the IOI (a high school contest) there are a much wider variety of countries represented among the top finishers, though Russia, China, and Poland are still strong there too.