Are physician-rating websites biased? Study examines bias in picking doctors, exaggerating consumer opinions

"Patients need high quality information about the most consequential service that they consume: healthcare," said Ritu Agarwal, professor of information systems and director of the Center for Health Information and Decision Systems at the University of Maryland's Robert H. Smith School of Business. "They want and need to be able to make informed choices."

"The Information Value of Online Physician Ratings" is co-authored by Agarwal, Guodong "Gordon" Gao, assistant professor, and PhD candidate Brad Greenwood, both of the Smith School, along with Jeffrey McCullough of the School of Public Health at the University of Minnesota. The researchers will deliver their findings at INFORMS Healthcare 2011 on June 20 in Montreal, Quebec. The conference takes place June 20-22 at the Hilton Montreal Bonaventure. Agarwal is editor-in-chief of the INFORMS journal Information Systems Research.

The authors investigated potential biases among Internet users rating general practitioners on websites such as Angie's List, HealthGrades, RateMDs, Vitals, and Yelp. They looked specifically at the way patients determine which physicians to rate and the intensity of opinions they express.

Citing a 2010 Pew Internet and Life Project survey, the authors write, "Online physician ratings are gaining popularity among patients. The public's demand for this information is striking: 59% of U.S. adults have looked online for health information, and among them 16% have consulted rankings or reviews online of doctors or other providers. Angie's List also found that 76% of its users welcome physician rating information."

Doctors have expressed a contrasting view. The American Medical Association and many physicians have criticized online physician rating services, charging that consumers lack detailed clinical information and that those who input ratings represent a disgruntled minority rather than average patients. They contend that consumer ratings provide negligible or even misleading information.

The authors addressed these concerns in their research.

On the selection of physicians rated, they asked if patients using the online sites review the full range of doctors (including those viewed positively, neutrally, and negatively); if they are biased toward giving negative reviews to doctors (which the authors characterize as "bad mouthing"); or if they are biased toward giving mostly positive reviews (providing what the authors call "sounds of silence" about poor caregivers).

On the selection of opinions to express, the authors investigated if respondents tend toward hyperbole and largely report their most extreme experiences with doctors.

Data for the study was taken from four data sets: online data from RateMDs, one of the largest American physician rating services; an offline patient survey conducted by the consumer advocacy group Consumers' Checkbook for the cities of Denver, Memphis, and Kansas City; the U.S. Census 2007 Economic Census, which was a source of information about population and median income in the three cities examined; and state medical board websites.

The authors came to two major conclusions that give credence to concerns about bias among respondents to online physicians rating services.

First, they find that physicians with low ratings in offline surveys are less likely to be rated online, therefore supporting the "Sound of Silence" effect in selecting what physicians to rate.

Second, although authors find a strong correlation between the online ratings and offline population opinion, the association is strongest in the lowest quartile of opinions. These results suggest that online ratings are more informative when identifying low-quality physicians, but not as effective in discerning high quality physicians from median ones. The authors also confirm that patients are most likely to provide ratings for their most flagrant or negative experiences with physicians.