Give Your Treatment Effect a Meaning: Applying the ICH E9 (R1) Estimand Framework to Internet-based Interventions

Randomized controlled trials are the gold standard for evaluating internet-based interventions (IBIs). However, their value depends on whether the estimated treatment effect accurately reflects its intended meaning. Unfortunately, this is not always the case. Decisions addressing so-called intercurrent events, such as treatment discontinuation, can shift effect interpretations. For instance, imputing data for those who discontinued the IBI instead of collecting follow-up data can shift the meaning from "the effect of simply providing individuals access to the IBI" to "the effect under the hypothetical scenario in which all individuals complete the IBI." To control the substantive meaning of the effect estimate, it is essential to clearly define the estimand, i.e., to provide a detailed and systematic description of the intended effect. A well-defined estimand justifies trial design decisions, including assessment and data-analytic strategies. Though widely adopted in pharmaceutical research, the ICH E9(R1) Addendum on Estimands and Sensitivity Analysis remains unrecognized in IBI research. Yet, its principles can improve research quality and reporting of trials studying the effects of IBIs. This manuscript introduces the addendum to IBI researchers by (1) explaining estimands and their five defining attributes, (2) emphasizing intercurrent events' role in the interpretation of treatment effects and strategies for handling them, (3) relating estimands to concepts like intention-to-treat and per protocol analyses, and (4) illustrating their application in two exemplary trials. We discuss how estimands inform study design, analysis, and reporting, offering practical recommendations for IBI research. We conclude that estimands are essential for investigating treatment effects in IBIs.