You have to remember that TPRS (the method of CI that I use the most) is intended ONLY to get structure and basic vocabulary into people's heads. So when students have acquired the basic structure (plus whatever vocabulary that was wrapped in), TPRS is no longer as useful.
The leap from "initial fluency" to "full fluency" (as I call it) is somewhat like the jump from BIC ("basic interpersonal competency") to CALPS (comprehensive academic language proficiency -- I think -- I forget the first letter). IF you take the time to have students acquire all the words they would need for the college exam at the same time that you're pumping structure into their heads, it would take too long, because before students have acquired the structure, it takes many more repetitions to acquire a word, and the language they are given as input has to be "more comprehensible" (=100% comprehensible).
After they have the structure in their heads, they can handle more unknowns, just like native speakers can. We are able to unconsciously "know" what part of speech things are, and how they work, because we're not learning new grammar rules for our native language, just expanding vocabulary.
There aren't any lists of CI items for Chinese, either. What we do is simply to take the language we want them to have at the end of the year (easier to do it year by year -- doing it starting from the test and working backwards would be possible but onerous!), isolate the verbs, nouns, and other words, recombine them (so "da3" and "dian4hua4" end up together, though "da3" will still appear with other nouns when it has another sense in the language) into "chunks" of language that logically go together, rate each one as to whether it's crucial, somewhat important or put into the curriculum/textbook by some idiot who had no clue about what was high frequency, and then start in providing dense CI on the crucial items, then the somewhat important ones (or using the somewhat important ones as differentiation for the faster processors and just making it comprehensible for the others). I do this with every Chinese course I teach, since so far there isn't a comprehensive CI based Chinese materials set available (our bad... )
The whole idea of a proficiency exam is that it's based on grasp of the structure of a language and knowledge of a pretty good whack of vocabulary but NOT on any specific vocabulary. It's supposed to measure how well you know the language as a whole. Discrete-item tests depend on a list or knowledge of specific vocabulary and don't really get to how much language a person commands. The college exam is still, IMO, mostly discrete-item. A better question for me would be how long would it take to get a student who can handle the IELTS exam through CI (but that's not the standard that most people in Taiwan care about, certainly not parents of buxiban-age kids.)