JMIR Publications

ABSTRACT

Background:

Testing digital health interventions through decentralized (i.e., fully remote) clinical trials (DCTs) offers advantages for collecting ecologically valid data at scale but poses challenges for recruitment, retention, and data quality. Likewise, the logistical challenge of collecting behavioral or biological assays outside the laboratory can limit the rigor and precision of data collected in the field. These challenges may be amplified in DCTs that incorporate remotely collected behavioral and biological assays, particularly in DCTs testing the efficacy of digital health interventions in populations selected for elevated levels of psychological distress, such as elevated depressive symptoms.

Objective:

We developed a protocol for the Behavior, Biology, and Well-being (BeWell) Study—a large-scale, three-arm, longitudinal, randomized DCT—to rigorously evaluate the efficacy of a mobile health well-being intervention (Healthy Minds Program). The developed protocol involved the remote collection of a multimodal battery of self-report (e.g., questionnaires), behavioral (e.g., elicited affect), and biological (e.g., blood and stool assays) measures. The objective of the present study is to describe the BeWell Study protocol, characterize performance across recruitment, retention, and data completion and quality metrics, and examine predictors of completion and quality to inform future digital health DCTs.

Methods:

Descriptive statistics were used to describe the sources of participant recruitment, retention, and data quality. Binary logistic regressions were conducted to describe the relationships between recruitment sources, demographics, and depression level on data completion and quality

Results:

Participants with elevated depressive symptoms (PHQ-9≥5) were enrolled from all 50 US states (n=1,157, Mage=38.69, SD=12.13; women=77.20%, non-Hispanic White = 68.91%). Participants were predominantly enrolled from Craigslist (60.09%) and university email messages (11.08%). Retention in the DCT was high (84.01%), with excellent data completion and quality across measures, demonstrating the feasibility of remote collection within a 4-month DCT of blood, stool, and behavioral measures typically obtained within laboratory settings. Recruitment sources and baseline depressive symptoms were not predictive of post-intervention data completion or quality. Older age and having a college degree predicted higher rates of data completion post-intervention.

Conclusions:

The BeWell Study protocol offers one approach to incorporating objective measures of well-being, such as biological and behavioral measures, into DCTs testing digital health interventions. Implications and limitations of the developed study protocol are further discussed. Clinical Trial: NCT05183867