To Örebro University

Background: Randomised controlled trials (RCTs) evaluating ADHD medications often use strict eligibility criteria, potentially limiting generalisability to patients in real-world clinical settings. We aimed to identify the proportion of individuals with ADHD who would be ineligible for medication RCTs and evaluate differences in treatment patterns and clinical and functional outcomes between RCT-eligible and RCT-ineligible individuals.

Methods: We used multiple Swedish national registries to identify individuals with ADHD, aged at least 4 years at the age of diagnosis, initiating pharmacological treatment between Jan 1, 2007, and Dec 31, 2019, with follow-up up to Dec 31, 2020. Hypothetical RCT ineligibility was established using exclusion criteria from the international MED-ADHD dataset, including 164 RCTs of ADHD medications. Cox models evaluated differences in medication switching and discontinuation within 1 year between eligible and ineligible individuals. Quasi-Poisson models compared eligible and ineligible individuals on rates of psychiatric hospitalisations, injuries or accidents, and substance use disorder within 1 year of initiating ADHD medications. People with lived experience of ADHD were not involved in the research and writing process.

Findings: Of 189 699 individuals included in the study cohort (112 153 men and boys [59%] and 77 546 women and girls [41%]; mean age 21·52 years [SD 12·83; range 4–68]) initiating ADHD medication, 53% (76 477 [74%] of 103 023 adults [aged >17 years], 12 658 [35%] of 35 681 adolescents [aged 13–17 years], and 10 643 [21%] of 50 995 children [aged <13 years]) would have been ineligible for RCT participation. Ethnicity data were not available. Ineligible individuals had a higher likelihood of treatment switching (hazard ratio 1·14, 95% CI 1·12–1·16) and a decreased likelihood of medication discontinuation (0·96, 0·94–0·98) compared with eligible individuals. Individuals ineligible for RCTs had significantly higher rates of psychiatric hospitalisations (ncidence rate ratio 9·68, 95% CI 9·57–9·78) and specialist care visits related to substance use disorder (14·78, 14·64–14·91), depression (6·00, 5·94–6·06), and anxiety (11·63, 11·56–11·69).

Interpretation: Individuals ineligible for ADHD medication trials face higher risks of adverse outcomes. This study provides the first empirical evidence for the limited generalisability of ADHD RCTs to real-world clinical populations, by applying eligibility criteria extracted from a comprehensive dataset of RCTs to a large real-world cohort. Triangulating evidence from RCTs and real-world studies is crucial to inform rigorous evidence-based treatment guidelines.