This study integrated an experimental medicine approach and a randomized cross-over clinical trial design following CONSORT recommendations to evaluate a cognitive training (CT) intervention for attention deficit hyperactivity disorder (ADHD). The experimental medicine approach was adopted because of documented pathophysiological heterogeneity within the diagnosis of ADHD. The cross-over design was adopted to provide the intervention for all participants and make maximum use of data.

Children (n = 93, mean age 7.3 +/− 1.1 years) with or sub-threshold for ADHD were randomly assigned to CT exercises over 15 weeks, before or after 15 weeks of treatment-as-usual (TAU). Fifteen dropped out of the CT/TAU group and 12 out of the TAU/CT group, leaving 66 for cross-over analysis. Seven in the CT/TAU group completed CT before dropping out making 73 available for experimental medicine analyses. Attention, response inhibition, and working memory were assessed before and after CT and TAU.

Children were more likely to improve with CT than TAU (27/66 v. 13/66, McNemar p = 0.02). Consistent with the experimental medicine hypotheses, responders improved on all tests of executive function (p = 0.009–0.01) while non-responders improved on none (p = 0.27–0.81). The degree of clinical improvement was predicted by baseline and change scores in focused attention and working memory (p = 0.008). The response rate was higher in inattentive and combined subtypes than hyperactive-impulsive subtype (p = 0.003).

Targeting cognitive dysfunction decreases clinical symptoms in proportion to improvement in cognition. Inattentive and combined subtypes were more likely to respond, consistent with targeted pathology and clinically relevant heterogeneity within ADHD.