Figure 2. Outcomes in a Study of α₂-Adrenergic Agonists vs Stimulants for Preschool-Age Children With Attention-Deficit/Hyperactivity Disorder.

The estimated relative risks and associated 95% CIs for reported medication improvement were obtained by first fitting logistic regression models and then using the approach suggested in Zhang and Yu²³ to convert the estimated odds ratios and their associated 95% CIs to estimated relative risks. The logistic models accounted for clustering within clinicians by site using the approach suggested by LaVange et al²⁴ that applies survey methods, with clinician treated as the primary sampling unit and site as the stratification variable. The adjusted models included age, autism spectrum disorder, and sleep disorder.