There is almost no way I can see for programs to manipulate this. The only way to do so is to not graduate someone. If someone graduates enough off cycle, then they can't sit for the exam that year but can the next year. If a resident puts off taking the exam, they are counted on their first attempt no matter when it is. In addition, the total number taking the exam is reported -- you can compare that to 3x the number of categorical residents they take each year, and it should be close (again, some people may be off cycle due to pregnancy leaves, etc).
Interesting, there's some interest by programs to get rid of this report. The idea is that programs would be more comfortable taking residents/students with lower exam scores if they thought they would be good residents but do poorly on exams. This is just vague talk, no one has formally proposed anything. A more reasonable idea would be to report all programs that are >=85% as such, and anyone below with their actual score.