201 Smart(phone) Approaches to Mobile App Data Collection in Epidemiologic Studies

Monday, June 5, 2017: 10:00 AM-10:30 AM
Eagle, Boise Centre
Lorene Nelson , Stanford University, Stanford., CA
Ingrid Oakley-Girvan , Cancer Prevention Institute of California, Fremont, CA

Key Objectives:

  • Discuss mobile apps as a means of collecting epidemiologic survey data for a variety of purposes and study designs including:
    • Surveillance studies to determine the prevalence of behavioral risk factors and health outcomes
    • Rapid ascertainment and follow-up after natural disasters, disease outbreaks or terrorist events
    • Intensive assessment of free-living populations (i.e., ecological momentary assessment)
  • Discuss the successes and the challenges we encountered when carrying out a recent SAMHSA-sponsored CSTE-funded survey of college students to assess several key mental health outcomes and substance use behaviors.

Brief Summary:
Mobile health apps have been increasingly developed and used in recent years as a method for providing behavioral interventions to improve mental health outcomes, but have not been employed as often by epidemiologists for the purposes of collecting epidemiologic survey data. Although mobile data collection methods have been used by opinion and market researchers in some limited contexts, it is not known how generalizable their methodologies are for epidemiologic and population health research. We will discuss the successes and the unexpected challenges we encountered when carrying out a recent SAMHSA-sponsored and CSTE-funded survey of college students to assess several key mental health outcomes and substance use behaviors. Our multidisciplinary study team included epidemiologists, survey researchers, behavioral health experts, statisticians, software engineers, user interface experts, and communications researchers. Using a mobile app for epidemiologic data collection had implications at all stages of study design and conduct, including: (1) designating study population and sampling frame (defined or not defined), (2) choosing optimal study design (cross-sectional or prospective), (3) achieving non-biased recruitment of study subjects, (4) maximizing retention of subjects in the prospective component of the study, (5) adapting survey design to fit mobile devices, and (6) analyzing mobile app paradata along with the primary survey data. In addition, there were many decisions to make regarding technical aspects of mobile app development including: type of platform (iOS, Android, both), data security methods, HIPAA compliance, app registration and built-in subject consent, use of gamification, subject tracking dashboards, and subject reminders (push notifications, emails). As epidemiologic researchers move to real-time mobile data collection methods, we will benefit from multidisciplinary teams representing a wide range of relevant expertise and by identifying lessons learned by other disciplines such as opinion survey and market researchers.