Patient motivation as a predictor of digital health intervention effects: A meta-epidemiological study of cancer trials

by Yuqian Yan, Jesús López-Alcalde, Elena Stallings, Elena Jimenez Tejero, Claudia M. Witt, Jürgen Barth

The objective of this meta-epidemiological study was to develop a rating that captures participants’ motivation at the study level in digital health intervention (DHI) randomised controlled trials (RCTs). The rating was used to investigate whether participants’ motivation is associated with the effect estimates in DHI RCTs for cancer patients. The development of the rating was based on a bottom-up approach involving the collection of information that captures participants’ baseline motivation in empirical studies from the Smartphone-RCCT Database. We specified three indicators for rating: indicator 1 captures whether the study team actively selects or enhances the motivation of the potential study participants; indicator 2 captures the study participants’ active engagement before the treatment allocation; and indicator 3 captures the potential bond and trust between the study participants and the person/institution referring to the study. The rating of each indicator and the overall rating varies between high motivation, moderate motivation, and low motivation. We applied the rating across 27 DHI RCTs with cancer patients. We performed meta-regression analysis to examine the effect of patient motivation on quality of life (QoL), psychological outcomes, and attrition. The intraclass correlation coefficient (ICC) indicated moderate to poor inter-rater reliability. The meta-regression showed that cancer patients’ overall motivation before engaging in the intervention was associated with the treatment effect of QoL. Patient motivation was not found to be associated with psychological outcomes or attrition. Subgroup analyses revealed that the clinical effects of DHIs were more prevalent in the high-motivation subgroups, whereas the low-motivation subgroups were unlikely to show intervention benefits. The likelihood of dropouts from DHIs seems to be especially high among the low-bond (indicator 3) subgroup. We suggest using single indicators since they reflect specific content. Better reporting about baseline motivation is required to enable meaningful interpretations in not only primary studies but also in evidence syntheses.