Yes, pretty much - they're being filtered into more appropriate (possibly could be argued by those taking them as more interesting) qualifications that aren't A Levels, leaving only the strongly academically inclined, such as those who are very able in Maths, to take Maths.
There's also the aspect that achieving top grades in Maths requires a literacy level able to decode the questions to then remember/understand the Mathematical processes and apply these to answer them correctly - so again, this means that the smaller number of boys with literacy levels able to understand the questions are already higher ability in multiple areas.
It's like saying in a 1983 comprehensive of 200 kids 'Boys are better than girls at O Level' when the boys less likely to get the very highest grades and could well have difficulty reading had all been filtered off into CSEs along with the very lowest ability/dyslexic/persistently absent/ill girls.
CSE 1-4: 95 boys
CSE 1-4: 10 girls
O Level A: The 5 highest ability boys. So 100% of boys that took the exam achieved the highest grade.
O Level A: 4 girls. Only 4% of girls achieved the highest grade.
O Level ABCDENUX: 90 girls of all but the very lowest ability.
And yes, it is mixing the percentage calculations as well. But it makes the boys look better whether you're looking at just those five or the proportion of the cohort (5%), doesn't it?