yes, you're right about that being a misinterpretation of 'could not reject the null hypothesis'.

i couldn't get the full text of the meta-analysis, but i've read similar studies in the past and one critical effect was that parent's ratings about perceived hyperactivity were heavily affected by whether the parent thought their child received sugar but there was no effect detected based on whether the children actually received sugar.

so it doesn't completely rule out whether sugar has an effect, but it does show that parental 'hyperactivity detectors' are broken and will be biased based on whether they think their kid had sugar (so you and everybody else should look at your anecdotal expereience with suspicion)