Gray BM, Vandergrift JL, Gao G(, McCullough JS, Lipner RS. Website Ratings of Physicians and Their Quality of Care. JAMA Intern Med. 2015;175(2):291-293. doi:10.1001/jamainternmed.2014.6291
One-third of consumers in the United States who consulted physician website ratings reported selecting and/or avoiding physicians because of these ratings.1 However, little is known about the validity of these ratings. Available studies have focused mostly on hospital website ratings or non-US website ratings.2,3 We partially address this gap by measuring the association between US physician website ratings and traditional quality measures (QMs) of clinical and patient experience.
We used a sample of 1299 physicians who completed an American Board of Internal Medicine Hypertension or Diabetes Practice Improvement Module between July 1, 2011, and November 30, 2012. Quality measures were drawn from about 25 Practice Improvement Module medical record abstractions and patient survey responses (59% response rate per physician). From medical record abstractions, we computed overall, intermediate outcome, and clinical process-of-care composites based on an expert panel’s assessment of quality.4 We also computed 2 QMs each for clinical and patience experience.
Website physician rating measures were drawn from Internet searches in which each physician’s name, specialty, and city were entered into the Google search engine. We extracted information from 8 free publicly available leading health-based websites. This information included physician rating, the number of patient ratings per physician, and search ranking. We normalized physician website ratings by dividing each rating by the website’s maximum score (eg, 4 of 5 stars equals 80%).
Physician-level QMs were regressed on website ratings. Regressions were estimated separately for each QM and controlled for physician, patient, and website characteristics. We estimated a model using all 8 websites and another model limited to the website with the highest Google search result as it may be more representative of what consumers view.
The project was reviewed by the University of Maryland College Park institutional review board and was determined to be exempt from institutional review board review. All data applied in this analysis were analyzed anonymously. Physicians who enroll in an American Board of Internal Medicine certification program enter into a business associates agreement that permits the American Board of Internal Medicine use of their de-identified data at an aggregate level for research purposes (www.abim.org/privacy.aspx). After linking publicly available rating data from the Internet, we were blinded to the physicians’ identities and the data were de-identified and were viewed and analyzed in aggregate. Patient survey data did not include any protected health information or other identifiers and were collected by physicians participating in the American Board of Internal Medicine’s maintenance of certification program; therefore, informed consent could not be collected. Furthermore, no names of patients were recorded.
Physician website ratings existed for 61.0% of physicians, with 5.6 patient ratings per physician and a mean normalized rating of 81.6% (Table 1).
The associations between physician website ratings and clinical QMs were small and statistically insignificant (≤0.3 percentage point change associated with a 20 percentage point rating change; P > .41) (Table 2). For patient experience QMs, associations were also small but were statistically significant (≤1.7 percentage point change associated with a 20 percentage point rating change; P < .05). For example, regression results indicate that a physician with a website rating of just 1 of 5 stars had 79% of their patients rate overall quality of their care as very good or excellent vs 82% for a physician who had a perfect 5 out of 5 rating. Overall, results were similar across website samples.
We found no evidence that physician website ratings were associated with clinical QMs. We did find a statistically significant, but small, association between physician website ratings and 2 Practice Improvement Module measures of patient experience. Overall, the weak associations we found could have resulted from the low number of website ratings per physician or because patients whose ratings are reported on websites are not typical of the overall population of patients treated by the physicians in our sample. Alternatively, weak associations may have been due to inherent limitations of our Practice Improvement Module data related to their content, sample structure, and chronic condition focus. Finally, we did not examine associations with patients’ narratives on the website.
Notably, a study of ratings for websites in the United Kingdom reported stronger associations with practice-level QMs than we found.2 More generally, patient experience measures, properly constructed and administered, have been shown to be associated with clinical outcomes.5 Regardless of whether US physician website ratings are related to the QMs we examined, these measures of patient experience may capture other important aspects of quality of care. Furthermore, the high prevalence and use of physician website ratings by US consumers indicates that either there might be a benefit not reflected in our QMs or consumers have no other easily accessible source of information.1,6 Overall, our study provides valuable information to consumers considering the usefulness of physician website ratings.
Corresponding Author: Bradley M. Gray, PhD, American Board of Internal Medicine, 510 Walnut St, Ste 1700, Philadelphia, PA 19106 (firstname.lastname@example.org).
Published Online: December 1, 2014. doi:10.1001/jamainternmed.2014.6291.
Author Contributions: Dr Gray had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Gray, Gao, McCullough, Lipner.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Gray, Vandergrift, McCullough, Lipner.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Gray, Vandergrift, McCullough, Lipner.
Administrative, technical, or material support: Vandergrift, Gao, Lipner.
Study supervision: Gray, Lipner.
Conflict of Interest Disclosures: Drs Gray and Lipner and Mr Vandergrift are paid employees at the American Board of Internal Medicine.
Funding/Support: The American Board of Internal Medicine provided financial and material support for this study. This study was partially supported by National Science Foundation Career Award 1254021 (Dr Gao).
Role of the Funder/Sponsor: The American Board of Internal Medicine had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.