(C-104) Demographic Insights into Participant Retention and Adherence in Longitudinal Digital Health Studies

Introduction:: Longitudinal digital health studies are becoming increasingly common due to the ubiquity of smartphones and access to the internet. These studies combine passively collected information from digital devices, such as commercial wearable devices, and actively contributed data, such as surveys, from participants. However, low retention and adherence to study tasks pose a challenge in collecting representative data. As a result, there is a need to identify factors that affect participant engagement in digital health studies. In this study, we explore participant retention and survey adherence in two large-scale digital health studies and develop a methodology to identify these factors.

Materials and Methods:: To better understand participant engagement, we defined metrics related to adherence in longitudinal digital health studies. We also developed models to infer how demographic factors may be associated with these metrics. Specifically, we defined retention as the time until a participant drops out of the study and three metrics of adherence: 1) Total Number of Surveys Completed, 2) Participation Regularity (consistency of filling out surveys), and 3) Time of Activity (i.e., participant filled out more surveys at the beginning of the study period vs. later in the study).

We used data from two large-scale, longitudinal digital health studies conducted by two separate institutions to assess these metrics and explore potential differences by age, sex, race, and ethnicity. We performed unsupervised clustering and survival analysis, which are well-established methods in this field, and recurrent event analysis on survey data, which is a novel application of this method to remote digital health data. Our goal was to develop a methodology that can be widely employed to better understand participant retention and adherence in digital health studies.


Results, Conclusions, and Discussions:: The study consistently observed age to be associated with retention and adherence, with younger participants dropping out sooner and having lower adherence than older participants. Participants could be sorted into groups based on their survey filling patterns, and there were statistically discernible differences between the groups based on demographic factors. Participants in the age ranges of “18-29” were more likely to exit the study than other age groups.

The methods developed in this study can be widely employed to better understand participant retention and adherence in digital health studies. Age was found to be the most significant factor associated with participant retention and survey adherence. This study's findings can be utilized to improve retention and adherence to support the development of equitable digital health technologies.


Acknowledgements (Optional): : We would like to thank Paul Louisell and Chris Barker for their helpful feedback on analysis approaches via the American Statistical Association’s Statistical Consulting Section Community Discussion online forum. We would also like to thank the Predictive Intelligence in Pandemic Prevention (PIPP) for funding.

References (Optional): :