I haven't read the paper itself, but that text, to me, does not give a "clear message [...] that we need to improve instruction in long division and fractions".
The text talks about correlation between the two, but does not give any arguments for causation. Judging from the title of the paper ("Early Predictors of High School Mathematics Achievement"), neither does that paper.
People who derailed in 5th grade are probably unlikely to get back on track because education in the higher classes stands on the shoulders of the previous classes.
If it not a causation, this may also mean that "non motivated students" never care about math. Either in 5th grade or in high school or later.
Or maybe it is a combination of both. The article only shows correlation. But how can one study causation without destroying the futures of students?
Make two groups; give one of them some extra and/or supposedly improved math schooling. Check their math grades and, years later, their success in higher education. See how much they correlate.
The text talks about correlation between the two, but does not give any arguments for causation. Judging from the title of the paper ("Early Predictors of High School Mathematics Achievement"), neither does that paper.