Working memory training in children with borderline intellectual functioning and neuropsychiatric disorders: a triple-blind randomised controlled trial.

Only 72 children were analysed, against the 100 the authors had calculated they needed. Two thirds of the families they approached declined to take part, so the children in this study may not look like the ones in your clinic. Everyone in the study got some form of computerised training and there was no group that got nothing. The improvements seen in both groups could come from practice, from growing up, or simply from taking part in a study, and this design cannot tell those apart. The adaptive group trained for longer in total than the placebo group (about 150 minutes more), so the two groups did not receive the same dose of training. Because the coaches were also blind to who had which programme, coaching was generic and not based on the child's results. This is not how Cogmed is delivered in practice, where a coach follows the child's performance. Training was done on a computer at home or school and was not built into the child's daily routines. Four percent of the data were missing and 15 outlying scores were kept in the analyses.

Funded by MIND Netherlands and the Dutch WKK foundation, two Dutch funding bodies. The authors state the funders had no role in running the study, in analysing or interpreting the data, or in the decision to submit the results. The authors declare no conflicts of interest.

Do not read 'both groups improved' as evidence that working memory training works. Every child in this study received a computerised training programme and there was no untreated group, so time, practice and taking part in a study could account for the gains. The adaptive training was not better than the placebo on the main outcome or on anything else tested.