To be honest I didn’t read the whole article and methods (just the summary) but it may be possible that increased intelligence just leads to increased masking, which changes the rate of diagnosis and skews the results. However, the results in the paper are quite significant so even if that’s true those effects may just return them to “evens”.