Digital health interventions — delivered via technologies including text messages, apps and websites — are increasingly common for supporting chronic disease self-care. Although studies have focused on improving patient engagement with these interventions, limited research has addressed the association between engagement and clinical outcomes.
Lyndsay Nelson, PhD, Andrew Spieker, PhD, and colleagues evaluated multiple statistical approaches for understanding this association including moderation, mediation and a modified instrumental variable analysis. They applied the approaches to two recent randomized controlled trials that used text messages to support medication adherence in patients with Type 2 diabetes or hypertension.
The researchers report on the strengths, assumptions and limitations of each approach in the Journal of the American Medical Informatics Association. They encourage researchers studying digital engagement to incorporate their newly developed modified instrumental variable approach to understand how engagement supports change in clinical outcomes. Determining an effective level of engagement for an intervention can help ensure users receive the intended benefit.
Other authors of the JAMIA paper include Lindsay Mayberry, PhD, Candace McNaughton, MD, PhD, MPH, and Robert Greevy, PhD. This research was supported by grants from the National Institutes of Health (DK100694, DK092986, HL137943).
