Optimizing and innovating blended interventions and aftercare for addiction (OptiBlend): Integrating data science, methods for behaviour change and a randomized controlled trial

There is a growing tendency to develop and implement blended interventions (digital modules combined with face to face (F2F) sessions) to treat patients with alcohol use disorder with/without other comorbid substance use disorders (A/SUD). The quantity and quality of studies on effectiveness and factors contributing to the effectiveness (e.g. personalized treatment) of blended A/SUD interventions is lagging behind. Studies are therefore urgently needed to test (cost-)effectiveness, explore ways of personalisation (eg a tailored number of F2F sessions) while using state-of-the-art innovative methodologies to innovate and optimize blended A/SUD interventions. The main aims of this mixed-methods project are to innovate a blended A/SUD intervention by adding newly developed sleep/nutrition and aftercare modules, and to evaluate it in a 3-arm randomized controlled trial, in which machine learning is used to personalise and tailor the number of F2F sessions. Our proposed project comprises 4 work packages (WPs). In WP1, we will apply machine learning techniques to our historic large dataset (n>5000) of regular blended care patients at Jellinek to create a prognostic model for the personalisation of the optimal number of F2F sessions which accompany the digital intervention environment in the blended intervention, and we will later on evaluate this model in the proposed RCT. Traditional research methods applied to patient-treatment matching have led to results with limited clinical relevance as only one to a few predictors collected at baseline were found to be statistically associated to patient-treatment matching [1]. However, data science methodologies such as machine learning provide new opportunities to investigate more tentative predictors in one model while controlling the risk of overfitting the data. Second, digital interventions generate many data on intervention usage as a side product as soon as the patient starts interacting with the intervention. These data draw a very detailed picture of how each individual patient interacts with the digital intervention environment. Based on experiences in an ongoing ZonMw project on data science methods for digital A/SUD interventions, we will use this early digital intervention use data in WP1 as well. In WP2, we further innovate and improve blended care for A/SUD patients. Innovation focusses on developing a state-of-the-art aftercare module focusing on relapse prevention, and on the development of transdiagnostic add-on modules to improve sleep quality and eating behaviours/nutrition. Aftercare and transdiagnostic lifestyle modules may help improve the (prolonged) substances use outcomes of blended interventions, in addition to making the patient adapt a more healthy lifestyle and making them less vulnerable to full relapse in case of (near) lapses. Contents of the new modules will be developed guided by the intervention mapping approach, based on the determinants of behaviour in the study population and in close cooperation with patients and other stakeholders in a participatory design. Focus groups will be used to develop and innovate the blended intervention and add-on modules. In WP3, a pragmatic 3-centre, 3-arm randomised controlled trial (RCT) comparing effects and costs of our innovated blended treatment (BL1) vs F2F intensity-personalised (see WP1) innovated blended (BL2) vs F2F treatment as usual (TAU) will be performed. The target group comprises patients (M/F) enrolling voluntarily for A/SUD treatment at one of the three trial sites, age 18+, with a AUD diagnosis and possible other SUD diagnoses. Primary outcome of the RCT is treatment response: no more than 4 alcohol drinking days and no heavy drinking days in the previous month, measured at 12 months follow-up. We will also evaluate reductions in the use of other substances, quality of life and overall mental wellbeing. An economic evaluation is planned alongside the RCT. Nested in the RCT, an Ecological Momentary Assessment (EMA) study to evaluate the innovated add-on modules will be performed. The RCT has a total follow-up duration of 18 months after baseline in order to effectively measure potential effects on relapse / prolonged intervention effects. For WP4, an implementation strategy is developed and implemented in all phases of the project, and through inclusion of relevant stakeholders and use of the RE-AIM framework, this will foster optimal implementation. When the interventions prove to be successful in reducing alcohol use (primary outcome measure) and/or on secondary measures such as substance use, as we anticipate based on previous findings, the implementation plan will be executed in collaboration with patient organisations using a multifaceted tailored strategy to support national implementation.