Qing Zeng*, The University of Utah, Salt Lake City, United States Rebecca Morris, University of Utah, Salt Lake, United States Catherine Smith, University of Wisconsin, Madison, United States Katherine Doyon, University of Utah, Salt Lake, United States Dawn Sweeney, University of Utah, Salt Lake, United States Jorie M Butler, University of Utah, Salt Lake, United States Track: Research Presentation Topic: Collaborative biomedical research, academic / scholarly communication, publishing and peer review Presentation Type: Poster presentation Submission Type: Single Presentation Last modified: 2014-09-19

If you are the presenter of this abstract (or if you cite this abstract in a talk or on a poster), please show the QR code in your slide or poster (QR code contains this URL).

Background: Patient satisfaction has frequently been used as an indicator of health care quality, often determined through surveys. New health care reforms place an increased emphasis on health care quality, and many are concerned if patient satisfaction surveys can accurately depict this. Satisfaction scores across the nation appear to be inconsistently reported, and suffer from response biases that inflate patient satisfaction ratings. This suggests that more research needs to be done to better understand what satisfaction means, and why the scores are inconsistent.

Objective: In this study we took a different approach and analyzed social media data containing patient-authored reviews of health care systems, hospitals, and clinics. The goal of the analysis was to assess patient satisfaction through open-ended survey questions, as we hypothesized that some factors that contribute to patient satisfaction are not captured by the more frequently used rating scale and multiple choice questionnaires.

Methods: We collected Google+ reviews of 142 medical health care facilities located in the following 9 regions of the US: East North Central, East South Central, Mountain, New England, Middle Atlantic, Pacific, South Atlantic, West North Central, and West South Central. One city was randomly selected from each region, and all reviews of all health care facilities in that city were collected from Google+. This allowed us to collect a national sample without regional bias. These facilities ranged from large academic centers to single provider offices, and were staffed by a variety of physicians from general practitioners to specialists. Google+ was used as it provided better national coverage of health care facilities compared to other sites. The reviews underwent a thematic analysis. We developed our initial codes using measures found in the Health Resources and Services Administration (HRSA) Patient Satisfaction Survey, but found that many topics frequently discussed in the reviews were not covered by this survey. In order to address this, a customized codebook was created jointly by the staff members tasked with reading and coding the reviews. NVivo 10 software was used for the coding and analysis of the reviews. In addition, an automated topic modeling analysis was conducted.

Results: Overall 729 open-ended online reviews were collected for analysis. A unique codebook was created to look at topics that contributed to patient satisfaction or dissatisfaction. Of the first 153 reviews that were fully analyzed, 43.1% were negative and 56.9% were positive reviews. About half of the reviews (51.6%) were self-identified as patients. Reviewers frequently commented about the staff (82.4%), getting care (66.7%), and the facility itself (37.3%). Topic modeling generated 41 stable topics using the 729 reviews, with 21 topics associated with positive reviews and 20 topics associated with negative reviews.

Conclusions: Compared to multiple choice and rating scale questionnaires, online reviews allow users to provide more detailed and comprehensive information on satisfaction and dissatisfaction. Our analysis of the data identified a number of additional topics that contribute to patient satisfaction, not covered by the HRSA Patient Satisfaction Survey.