Analysis and Proposal for the Dispensation of Digital Therapeutics in Italy

Proposals document accompanying the talk "Digital therapeutics: state of the art in Italy" (SIFO 2026 National Congress, Rome, 6 November 2026). It is a proposals paper, kept separate from the Supplementary Information of the talk, which contains the data and analyses. Each paragraph is tagged as Data, Analysis or Proposal, so that sourced facts, the author's inferences and the author's opinions can be told apart. The starting point is the "efficacy without use" problem: digital therapeutics (DTx) that work in controlled settings are abandoned early in real life. Trial dropout ranges from about 19% to 48%, real-world abandonment reaches a median of 70% within 100 days, and human contact is the best-documented protective factor, although only as an association. The paper notes that most usage data concern health apps in general, not DTx as defined by ISO/TR 11147:2023, and that we found no post-marketing usage data for DTx. The bill S. 1918, as approved by the Italian Chamber of Deputies on 27 May 2026, defines digital therapies and their assessment, but in our reading it does not mention prescription or dispensation. The step between prescription and intake is therefore undefined. We propose three distinct activities with three actors: prescription (prescriber), dispensation (a health mediator distinct from the prescriber) and intake (patient). We distinguish software and procedural "digital excipients", to be described in HTA dossiers and assessed on use endpoints, separately from the clinical endpoint of the active function. Health mediation comprises five acts: verify, explain, activate, control, and vigilance, delivered at three levels of intensity proportional to risk. We propose the term "dispensazione", with an explicit definition, and discuss its legal risks. As mediator we propose a specialised DTx pharmacist in the pharmaceutical services of the local health authorities, working remote-first from regional or inter-ASL hubs, with in-person dispensation when needed and with other professionals as co-mediators. An illustrative sizing, built on German activation data, suggests that a few dozen full-time pharmacists would suffice in the first years. Evidence for the model is limited: a Danish routine-care programme with psychologist support shows that human mediation is feasible at scale, but not that it improves outcomes. We propose a comparative pilot with retention at 12 weeks as primary endpoint, and state what result would lead to revising or withdrawing the proposal. Conflicts of interest and the use of an AI model (Claude, Anthropic) for analysis and drafting are declared in the document. The Italian version is published on the website of the Associazione TendenzeSaluteSanità.