Increasing retention in a large-scale, remote wearable device study: learnings from the COVID-RED trial

Background: Remote digital health trials can reach large populations quickly and collect sizable amounts of real-world data. Yet, many studies suffer from high drop-out rates (in some settings over 40%) and poor retention, negatively impacting the results’ precision and generalizability.

Objectives: To explore targeted, innovative data-driven retention strategies applied to a large fully remote, single-blind, digital trial (“COVID-RED”) that evaluated the performance of an algorithm ingesting data from a wearable device and self-reported symptoms to detect pre-symptomatic SARS-CoV-2 infections.

Methods: The COVID-RED trial collected data from 17,825 participants February through November 2021. Participants wore the Ava bracelet while sleeping and synced it upon waking with a custom mobile application. In addition to reporting symptoms, health status and diagnostic SARS-CoV-2 test results in app, they completed baseline questionnaires, bi-weekly surveys, and COVID-19 antibody tests. Besides common retention activities such as general e-mail reminders, our retention plan included data-driven and targeted activities based on behavioral theory to boost retention and compliance throughout the 9-month follow-up.

Results: Of the 17,825 participants, 14,105 (79.1%) remained participating by the last 3 months and 12,021 (67.4%) by the end of the trial. In total, 5,649 (31.7%) participants discontinued. Common retention activities applied in the COVID-RED trial included newsletters, social media posts, website updates and a helpdesk. Innovative, data-driven activities included personalized e-mail reminders with gamification based on participant compliance and the development of a multivariable prediction model for discontinuation in the remainder of the trial. This model identified three groups as high-risk for discontinuation in the last 3 months with sufficient sample size: students who are single and live alone (n=90; OR = 1.104 [1.007, 1.209]); vaccinated pensioned participants (n=1,019; OR = 1.087 [1.031, 1.145]); and vaccinated participants who have a steady relationship and live alone (n=1,358; OR = 1.049 [1.015, 1.085]). In line with behavioral theory, no extra activities were planned for the first group, while phone calls were made to the other groups about their activities. Exploratory analyses into the effect of these strategies are underway.

Conclusions: This study describes how innovative and targeted data-driven retention strategies can be applied on top of standard methods to improve compliance and retention in large, remote digital health trials. While further research is necessary, the applied exploratory retention interventions showed a lower dropout rate than standard methods.