In Reply We appreciate the comments from Mr Richards-Belle and colleagues and Dr Zarkowski regarding our recently published study.¹ Their comments highlight important considerations in pharmacoepidemiological studies using real-world data.

Although the target trial emulation framework helps address several potential problems with traditional pharmacoepidemiological studies using observational data, it does not necessarily eliminate all biases.² We concur with Zarkowski that in real-world settings, the decision to prescribe ADHD medication is made by clinicians based on each patient’s characteristics. Target trial emulation relies on measured covariates to balance differences in patient characteristics across different ADHD treatment strategies. Therefore, residual confounding may occur when the confounders are unmeasured or measured imperfectly.