Re: Harnessing the cloud of patient experience: using social media to detect poor quality healthcare
January 31, 2013
To the editors:
We were pleased to read the recent article by Greaves et al.1 outlining new methodological techniques to analyze patients' online ratings of care. We agree with the authors that social media websites represent a wealth of first-hand patient experiences with health and healthcare, but have largely remained untapped by biomedical researchers - especially to gain new insights into how to improve clinical care. We concur that "big data" techniques such as machine learning and natural language processing can be extremely powerful to synthesize the large amount of textual data on these sites.
However, our previous work has also suggested the importance of traditional research methods applied to social media content. In particular, qualitative analysis adds perspective to patients' online dialogue where big data mining techniques perhaps cannot. In a qualitative examination of primary care provider ratings on Yelp,2 we analyzed 712 reviews of 455 doctors in four large urban areas (Chicago, New York, Atlanta and San Francisco). We found that these provider ratings often reflected the entire visit experience (i.e., parking, wait times, front desk staff) rather than focusing solely on the clinical encounter with the provider. Similarly, we recently qualitatively coded over 450 Twitter messages about cancer screening,3 and found that miscellaneous tweets such as jokes or popular culture references could be distinguished from the rich information about personal patient experiences with pap smears or mammograms. In both instances, these nuances in the online content may have been missed by applying data mining or natural language processing alone.
Therefore, we advocate using mixed methods approaches to analyzing social media content about health and healthcare experiences, as these techniques are inherently complementary to one another. Big data approaches allow researchers to examine millions of messages to uncover trends and overall sentiment in the online content, as well as the potential to rank the prevalence of specific discussion topics on social media sites. However, in combination with qualitative analysis of a carefully selected subsample of online content, the textual data can be interpreted in light of additional context - allowing researchers both breadth and depth in their work.
Moreover, not only should we aim to understand patient values and preferences from the large amounts of publicly available dialogue on social media, but we should also look to online social media as a means to directly engage in this dialogue with patients. Because of the ease of use and the speed of information dissemination, online social media channels have become a cornerstone of everyday life, transforming the ways that society shares ideas and beliefs, news, and information about products and services among individuals and organizations. To be truly patient-centered, healthcare providers and systems should play an active role in communicating important health and healthcare messages through the channels in which growing numbers of patients are already engaged.
Courtney R. Lyles, PhD & Urmimala Sarkar, MD, MPH University of California San Francisco, Division of General Internal Medicine
References 1. Greaves F, Ramirez-Cano D, Millett C, Darzi A, Donaldson L. Harnessing the cloud of patient experience: using social media to detect poor quality healthcare. BMJ quality & safety 2013. 2. Lopez A, Detz A, Ratanawongsa N, Sarkar U. What patients say about their doctors online: a qualitative content analysis. Journal of general internal medicine 2012;27(6):685-92. 3. Lyles CR, Lopez A, Pasick R, Sarkar U. "5 Mins of Uncomfyness Is Better than Dealing with Cancer 4 a Lifetime": an Exploratory Qualitative Analysis of Cervical and Breast Cancer Screening Dialogue on Twitter. Journal of cancer education : the official journal of the American Association for Cancer Education 2012.
Conflict of Interest: