I'd say that reading assessments vary widely by district. For instance, our school district uses three different assessments, each up to three times a year: MAP testing, DIBELS testing, and Running Reading Records. MAP testing asks comprehension questions and vocabulary questions, but also requires students to know the meaning of words like "synonym," "antonym," "homonym," and "homophone" when answering comprehension questions. DIBELS testing, as I understand it, is primarily about fluency and accuracy, and is done with a student reading directly to a proctor of some sort. And Running Reading Records involve the student reading directly to a teacher with the teacher, subjectively, recording accuracy and the student's ability to answer questions about the test.
So, IME, a student's reading level as tested in school can be an inaccurate measure of what a child is actually reading, understanding, and enjoying outside of school. I know this definitely happened to a couple of my kids. And, btw, I remember hearing somewhere that's it's a pretty common thing for advanced readers to substitute similar words or to skip over some less important words when reading for pleasure. Perhaps for school purposes it's better to be 100% accurate and fast, but in real life, maybe not quite as much.
