Measuring the Rise of Mobile and Online Care: Promises and Challenges in Big Data
Web server logs, shared medical record
Background/Aims: Engaging patients and families is a key aspect of meaningful use of electronic medical records. Although several populations are less likely to adopt and use patient websites to engage in shared electronic medical records, the increasing availability of mobile devices and applications has potential to extend patient engagement in its use. No single data source, however, currently tracks individual patient shared record use across devices and software applications. Our objective is to describe our experience aggregating data sources for measuring enrollee use of the shared electronic medical record across mobile and desktop devices.
Methods: Between January 2010 and August 2014, we merged data from 358,415 Group Health enrollees from Web server activity logs (including mobile application activity) with Epic/Clarity electronic medical record data. Web server logs were scanned for string combinations to identify devices used to access shared medical record (SMR) services (including medication refills, medical test results, secure messaging encounters, after-visit summaries, appointment requests, medical problem lists, allergies and immunizations). Logs were matched to Epic/Clarity SMR page views within a time window.
Results: In August 2014, approximately 5.5 million of 150 million Web server log entries (<4%) were extracted for device information. After excluding redundancies, 245,425 SMR entries were retained. Of these, 70% (170,705) with device information was linked with Epic/Clarity patient page views. Of unmatched SMR activity, 29% (71,578) was on server logs alone and 1% (3,142) was on Epic/Clarity alone. Standard Web browsers accounted for 88% (187,262) of overall Web activity (76% [187,262] desktop, 12% [29,964] mobile device) and mobile applications accounted for 8% (18,904); 4% (9,295) was unable to be linked. Among 94,303 enrollees accessing the SMR with device information, 92% (83,081) used standard Web browsers (78% [70,484] desktop, 14% [12,597] mobile device) and 21% (18,965) used mobile applications.
Discussion: Web server logs and Epic/Clarity can be combined to describe enrollee use of mobile and desktop services. Development and maintenance of data requires regular monitoring for consistency and content. Extraction logic must be dynamic to accommodate device market and health care system changes. Understanding adoption and use of online services across devices will be essential to successfully engaging patients and families in care.
Jordan LL, Chang E, Kriekenbeck G, Ralston JD. Measuring the Rise of Mobile and Online Care: Promises and Challenges in Big Data. J Patient Cent Res Rev 2015;2:115. http://dx.doi.org/10.17294/2330-0698.1138