A very quick look at this makes me wonder about a number of areas. No data sharing, of course, which is a shame.
What did that letter by Biggs say? They're talking about sample size and type 1 errors a lot, but/so why are they not reporting effect sizes as well as statistical significance, to back up their arguments (if the effect sizes do indeed back up their arguments)?
This is an interesting sentence:
The differences between our findings and the previous GIDS study re change in psychological function may relate simply to sample size
From what I can see, the previous study (which presumably showed improvements in psychological functioning) used a clinician-reported measure (i.e., clinicians giving ratings based on their judgments). The measures in the current study are based on parent- and self-report.
The follow-up rate at 24 months isn't great, is it? Less than a third for most measures?