![[ Visit Client Website ]](images/banner.gif)
Theoretical Background and research questions/hypothesis: Data mining has emerged as an impressively powerful means for assessing massive quantities of data effectively and succinctly with remarkable speed. As a marketing tool, it identifies target populations and consumer profiles. As a means for studying healthcare technology, it identifies anything from trends in new media usage to the deficiencies caused by the digital divide on remote, poor, aged, and/or rural populations. This study applies data mining specifically to the daily electronic interactions between physicians and patients and provides evidence as to how and why this method of communication can be effectively used to improve health. Until recently what we have learned about these messages as been very functional: Patients use emails to make appointments, get prescriptions, report daily food intake, or monitor their blood sugars. Physicians feel compelled to comply despite personal resistance for fear of being over-burdened and under-paid for a largely unexplored medium of open access with their patients. The American Medical Informatics Association (AMIA, 1998) has provided the initial ground rules for ethical standards in communicating; the Institute of Medicine (IOM, 2002) has endorsed the medium as a necessary option and right for all patients; and the HITECH Act of 2009 has mandated full use of Electronic Health Records (EHRs) by 2014 with built-in “meaningful use” incentives for all those who comply. (In fact, current discussion of Stage II compliance criteria necessitates provisions for open email access between physicians and patients.) Despite this flurry of research and government attention, the fact remains that to date no one has examined the content of these messages, the words used, the patterns of communication that suggest how and not just what nuances take place within actual physician/patient transactions. This exploratory study combines the power of data mining programming with the logistics of linguistic content analysis in an effort to determine how patients and physicians actually do communicate within the secured portals of their healthcare facilities.
Methods: Over 40,000 electronic messages between 40 physicians and over 3000 patients within a four-year period (2006-2010) were de-identified and analyzed through a newly-designed computer program. With permission from the IRB and the University of Pittsburgh Medical Center’s General Internal Medicine Clinic in Montefiore Hospital, these messages were then studied through a systematic analysis process.
Results: Words were coded into thematic dictionaries, frequency differences were run on subject demographics, and changing patterns of usage were identified over time. (Please note this is an on-going study.)
Conclusions: This exploratory study marks the first of its kind and promises to be a launching pad for further, more in-depth analysis of the coordinated management of meaning between physicians and patients.
Implications for research and/or practice: It suggests that as physician/patient teams progress through a medical relationship, meaning becomes shared, and a clear pattern emerges from conflict/problem to resolution/healing. Finally, this research suggests clear-cut possibilities for creating evidence-based standards for training physicians and patients alike in an effort to achieve the best possible healthcare environment for all those who choose to use this increasingly more popular healthcare medium.