284 9 Chapter 9 which the eHealth intervention is implemented. Consequently, RCT results may have limited applicability to patient outcomes once the trial has ended (43, 44). Another challenge is the multiple component nature of eHealth interventions, which rely on an interplay between technology, human characteristics (e.g., patient behaviour, engagement with technology) and socioeconomic factors (e.g., reimbursement schemes). This complexity makes it difficult to attribute observed effects within an RCT to a specific component of the intervention. This is important to determine what works for whom to derive the effective components. In the case of the smart asthma inhaler programme (chapter 5) we indeed cannot solely deduce from results of our primary endpoint (medication adherence over 12 months follow up) which components of the programme (e.g., reminders and symptom tracker, patient portal) contributed to the potential beneficial effect. The multi-component nature of eHealth interventions also presents comparability issues in meta-analysis. Meta-analyses are based on the premises that interventions are comparable (e.g., drug ‘A’ versus drug ‘B’). Hence, problems arise when interventions are not comparable as is often the case with complex health interventions such as eHealth and the integrated disease management programs studied in chapter 6 (40). While we attempted to address this issue in chapter 6 by performing subgroup analyses based on the dominant intervention component, the substantial heterogeneity observed in some subgroup comparisons indicates that comparison issues may persist, and no definitive conclusions can be drawn regarding the most effective components of an integrated disease management program. Moreover, interpreting the overall estimate of effect can be challenged by the interaction of the components with each other. Alternative designs and recommendations Alternative study designs, such as stepped-wedge trials or hybrid designs incorporating process outcomes, may be better suited to evaluate the effectiveness of eHealth interventions. These designs provide flexibility and adaptability, accommodating the complexities inherent to eHealth interventions (45). Factorial design and realist reviews should be considered as means to identify working components of eHealth technologies (46, 47) and more advanced meta-analytical techniques like meta regression (48), Network Meta Analysis (47) or Individual patient data meta-analysis can be deployed as alternative to traditional meta-analysis to identify working mechanisms of complex health interventions using pooled data. Pragmatic trials (Chapter 5), performed in a real-world clinical practice setting, overcome some limitations of traditional RCTs. They resemble routine care as closely as possible and incorporate the natural variation observed among patients, including heterogeneity in study samples and co-morbidities. Consequently, the results are more applicable to the target population of the intervention (44). As with all study designs, the research process is a balancing act between maintaining internal validity while maximizing external validity (44, 49). To facilitate the design of a pragmatic trial and ensure alignment with the intended goals and purposes, researchers can
RkJQdWJsaXNoZXIy MTk4NDMw