This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Physician rating websites (PRWs) allow patients to rate, comment, and discuss physicians’ quality. The ability of PRWs to influence patient decision making and health care quality is dependent, in part, on sufficient awareness and usage of PRWs. However, previous studies have found relatively low levels of awareness and usage of PRWs, which has raised concerns about the representativeness and validity of information on PRWs.
The objectives of this study were to examine (1) participants’ awareness, use, and contribution of ratings on PRWs and how this compares with other rating websites; (2) factors that predict awareness, use, and contribution of ratings on PRWs; and (3) participants’ attitudes toward PRWs in relation to selecting a physician.
A mailed cross-sectional survey was sent to a random sample (N=1542) from four North German cities (Nordhorn, Hildesheim, Bremen, and Hamburg) between April and July 2016. Survey questions explored respondents’ awareness, use, and contribution of ratings on rating websites for service (physicians, hospitals, and hotels and restaurants) and products (media and technical) in general and the role of PRWs when searching for a new physician.
A total of 280 completed surveys were returned (280/1542, 18.16% response rate), with the following findings: (1) Overall, 72.5% (200/276) of respondents were aware of PRWs. Of the respondents who were aware of PRWs, 43.6% (86/197) had used PRWs. Of the respondents who had used PRWs, 23% (19/83) had rated physicians at least once. Awareness, use, and contribution of ratings on PRWs were significantly lower in comparison with all other rating websites, except for hospital rating websites. (2) Except for the impact of responders’ gender and marital status on the awareness of PRWs and responders’ age on the use of PRWs, no other predictors had a relevant impact. (3) Whereas 31.8% (85/267) of the respondents reported that PRWs were a very important or somewhat important information source when searching for a new physician, respondents significantly more often reported that family, friends and colleagues (259/277, 93.5%), other physicians (219/274, 79.9%), and practice websites (108/266, 40.6%) were important information sources.
Whereas awareness of German PRWs appears to have substantially increased, the use of PRWs and contribution of ratings remains relatively low. Further research is needed to examine the reasons why only a few patients are rating physicians. However, given the information inequality between provider and consumer will always be higher for consumers using the services of physicians, it is possible that people will always rely more on interpersonal recommendations than impersonal public information before selecting a physician.
When searching for a new physician, patients typically want to find a physician who is a
Typically grounded in the assumptions of a theoretical consumer choice model [
The utilization of the comparative quality information by health consumers depends on a range of factors, but at the most basic level, it requires consumers to first be aware of it [
With the most recent previous German studies regarding PRW awareness and usage having been conducted in 2012 and 2013 [
This study was approved by Hannover Medical School’s Research Ethics Committee on January 12, 2016. All participants signed an informed consent form.
A mailed survey was conducted between April and July 2016. A random sample was obtained from the Registry offices of four North German cities of various sizes, under paragraph 34 of the Federal Registry Act (Bundesmeldegesetz) that allows registration authorities to transfer data to other public bodies if certain criteria are met. Inclusion criteria for the random samples were that the person’s place of residency was Nordhorn (53,285 residents, valid December 31, 2015), Hildesheim (101,667 residents, valid December 31, 2015), Bremen (556,326, valid December 1, 2015), or Hamburg (1,787,408 residents, valid December 31, 2015) and that the person was aged between 18 and 85 years. These cities were selected to enable participants from different sized cities to be recruited (small or rural area, medium city, large city, and extra-large city). To ensure participants’ confidentiality and to allow for reminders to be sent to nonresponders, a unique identifier code was assigned to each participant from the random sample before data collection. The document with participants’ identifying information and unique ID was accessible only to the study team, password protected, and stored separately from data documents. Surveys with participants’ unique IDs on them were mailed to a total of 1600 residents in Nordhorn (n=400), Hildesheim (n=400), Bremen (n=400), and Hamburg (n=400). Surveys were sent to residents of Nordhorn, Hildesheim, and Hamburg in the first week of April 2016, with a reminder sent to all nonrespondents 3 weeks later. Due to a late response from the Bremen Registry Office, surveys were sent to residents of Bremen in the first week of June 2016, with a reminder sent to all nonrespondents 3 weeks later. No incentives to participate in the study were provided. Fifty-eight surveys, which were returned because of out-of-date addresses or because participants had died or have a severe disability and are unable to read and write, were excluded from the study, leaving a total of 1542 surveys.
Survey questions were primarily adapted from previous surveys conducted in Germany [
For the questions concerning the awareness, use, and personal contribution of a rating on rating websites, there exists a cascade of questions where an inconsistent answer pattern could arise if a respondent answered “yes” to a question following a question they had answered “no” to. This inconsistency was solved by only including the first answer. Descriptive statistics included medians and means for continuous variables and percentages for categorical variables. Questions that used 4-point Likert response scales were dichotomized at the midpoint because sample sizes for some cells were often too small to be analyzed. Pearson chi-squared tests were used to analyze awareness, use, and contribution of ratings on rating sites, the role of PWRs when searching for a new physician, characteristics of respondents, and patterns of nonresponse. To test response rates between measures of the same subject (eg, awareness of PRW vs awareness of rating websites for hotels), we used the McNemar test. When comparing the percentages of participants using rating websites between two types of websites (PRWs and another), only those respondents who were aware of both types of websites were included. Similarly, when comparing the percentages of participants contributing a rating on two types of websites (PRWs and another), only those respondents who had used both types of websites were included. To assess potential predictors of the three outcomes—(1) awareness of PRWs, (2) use of PRWs, and (3) previously rated a physician—nine candidate predictors were preselected based on theoretical considerations and previous findings [
Overall, a total of 280 completed surveys were returned, corresponding to an 18.16% (280/1542) response rate. Seventy-five of the completed surveys came from Nordhorn, 72 from Hildesheim, 62 from Bremen, and 71 from Hamburg. In addition, 169 formal refusals to participate in the study were received, of which 103 provided a reason for nonparticipation. Key reasons given for nonparticipation included no interest in the topic (n=21), no computer and/or Internet (n=10), not aware or do not use PRWs (n=9), health reasons (n=9), time reasons (n=5), and age reasons (n=5). Overall, 15.7% (44/280) of respondents were aged 30 years and less, 29.3% (82/280) were aged between 30 and 50 years, 38.6% (108/280) of respondents were aged between 50 and 70 years, and 16.4% (46/280) were 70 years and older. Furthermore, 55.0% (154/280) of respondents were female, 58.9% (165/280) were married or in a civil partnership, 78.4% (218/278) had never been employed in health care, 81.4% (227/279) had public health insurance, 35.1% (98/279) suffered from a chronic illness, 29.7% (83/279) had changed their place of residence in the last 10 years 1 to 2 times, and 10.0% (28/279) had changed 3 or more times. Nonresponder analysis comparing all responders (n=280) with all nonresponders (n=1320) showed that gender composition did not significantly differ between responders and nonresponders (χ21=1.9,
Overall, 72.5% (200/276) of the respondents were aware of PRWs (see
Of the respondents who were aware of PRWs, 43.6% (86/197) had used PRWs (see
Of the respondents who had used PRWs, 23% (19/83) had rated physicians at least once (see
Proportion of respondents who were aware of rating websites.
Proportion of respondents who had used rating websites.
Proportion of respondents who had personally rated on rating websites.
Standardized coefficients of all predictors across all 3 outcomes were generally very low, suggesting that except for the impact of responders’ gender (higher for females than for males) and marital status (higher if married or in a civil partnership than if not) on awareness of PRWs and responders’ age on use of PRWs (higher for younger), no other predictors had a relevant impact. Complete results of the tests assessing potential predictors for the outcomes “already aware of PRWs,” “previously used a PRW,” and “previously rated a physician on a PRW” are shown in
Selecting a new physician.
Question | n (%) | |
Family, friends, and colleagues | 259 (93.5) | |
Other physicians | 219 (79.9) | |
Practice websites | 108 (40.6) | |
Physician rating sites | 85 (31.8) | |
Business directories | 51 (19.0) | |
No experience | 10 (12) | |
Very useful | 9 (11) | |
Somewhat useful | 28 (33) | |
Less useful | 34 (40) | |
Not at all useful | 4 (5) | |
No experience | 9 (10) | |
Very strongly | 7 (8) | |
Somewhat strongly | 22 (26) | |
Less strongly | 33 (38) | |
Not at all | 15 (17) |
aPRWs: physician rating websites.
Whereas 31.8% (85/267) of the respondents reported that PRWs were a very important or somewhat important information source when searching for a new physician, they significantly more often reported that other factors were very important or somewhat important information sources, with 93.5% (259/277) endorsing family, friends, and colleagues, 79.9% (219/274) other physicians, and 40.6% (108/266) practice websites (
This study has resulted in three key findings. First, awareness of German PRWs was found to be high (72.5%), though the usage of PRWs (43.6%) and especially the contribution of ratings (23%) remain relatively low. Awareness, use, and contribution of ratings on PRWs were also significantly lower in comparison with all other rating websites, except for hospital rating websites. Second, respondents’ age was the only relevant predictor of use of PRWs, with younger respondents more likely to use PRWs. Third, when selecting a new physician, the importance of factors such as family and friends as information sources were endorsed more frequently than PRWs.
Respondents’ reported awareness of PRWs was substantially higher than the awareness of PRWs reported by previous German studies, which found 29% awareness in 2012 and 32% in 2013 [
However, despite this increase in PRW awareness, the percentage of those who had used PRWs (among the responders who were aware of them) was substantially lower than previous German studies. For instance, whereas Emmert et al only reported that 25% of all respondents had used PRWs, their data show that among those who were aware of PRWs, 79% had used PRWs [
Respondents’ age was found to be the only relevant predictor of use of PRWs, with younger respondents more likely to use PRWs. This finding supports previous research [
Among the responders who had used PRWs, the percentage of those who had rated a physician was also much lower than previous German studies. For instance, whereas Emmert et al reported that only 11% of all respondents had rated a physician, their data show that among those who had used PRWs, 44% had rated a physician [
However, it is also useful to consider the reasons why participants’ awareness and usage of PRWs were all significantly lower in comparison with the other non–health care-related rating websites, which is consistent with the results of the 2 Hanauer et al studies [
These findings suggest that the awareness and usage of PRWs may, in fact, always be lower in comparison with the other non-health care–related rating websites because the information inequality between provider and consumer will always be higher for consumers using the services of physicians (and hospitals) than it is for consumers using other services such as hotels and restaurants, and even more so for consumers buying products such as media and technical products. Whereas changes to the way in which PRWs are currently designed may make them more useful [
Nevertheless, there is currently limited research examining the reasons why patients are not rating their physicians on PRWs, and more research is needed regarding this issue to identify barriers that may be addressed. A recently published study by Patel et al explored patients’ views regarding rating general practitioners on PRWs, within the context of other feedback methods available in England [
This study has a number of limitations that should be taken into account when interpreting the results. Responder bias may have influenced the results; however, as those who responded to our survey are likely to be generally more interested in the issue, the relatively low use of PRWs and low contribution of ratings should be taken seriously. Additionally, with a response rate of 18.1%, a generalization of the quantitative results to all the inhabitants of the four North German cities is likely not possible. Differences may also exist between other regions in Germany with respect to rating sites. However, the survey was mailed to a random sample of an average population from four North German cities of different sizes. The nonresponder analysis also found no significant difference in gender, and whereas responders were slightly older than nonresponders on average, the effect size was small. One of the reasons nonparticipants provided for not participating in the survey was that they were not aware of PRWs or do not use PRWs. If nonparticipants did not participate simply because they were not aware of PWRs, the average proportion of participants that is aware of PRWs is likely to be overestimated. However, out of the 169 refusals to participate in the study that were received, only 9 of them reported that they did not participate because they are not aware of PRWs or do not use PRWs. Furthermore, as there is a cascade of questions (awareness > use > contribution of ratings), there need not be an overestimation in relation to the use of PRWs and rating of a physician on a PRW at least once, since only participants who are actually aware of PWRs were included for these questions. We therefore do not think that this issue has significantly impacted our results. Responses were self-reported, and therefore we do not know the actual use of PRWs or what PRW participants visited. Additional research involving a larger sample would be desirable.
This study indicates that awareness of PRWs in Germany has substantially increased in recent years. This is a positive development and suggests that a lack of awareness is no longer a key barrier for PRW usage in Germany. Nevertheless, the level of usage of PRWs remains relatively low, and moving forward, the focus should be on a better understanding of the reasons why patients are not rating their physicians on PRWs, so that barriers that may be addressed can be identified. However, given the fact that the information inequality between provider and consumer will always be higher for consumers using the services of physicians, the awareness and usage of PRWs may, in fact, remain lower in comparison with the other non-health care–related rating websites. While changes to the way in which PRWs are currently designed may make them more useful, it is possible that people will always rely more on interpersonal recommendations than impersonal public information for selecting a physician.
Survey.
Awareness and use of rating websites.
Factors predicting awareness and use of physicians rating websites.
Model accuracy for multiple logistic regression model and the lasso based on 10-fold repeated cross-validation.
Allgemeine Ortskrankenkasse
Hochschulinterne Leistungsförderung
multiple logistic regression
Physician rating websites
area under the receiver operating characteristics curves
Techniker Krankenkasse
The authors would like to thank Dr Sabine Bossert for her helpful suggestions regarding the development of the survey and Brita Sander for administrative assistance. The authors also thank the reviewers of the manuscript for their helpful comments and suggestions. This project was funded by the Hannover Medical School’s Hochschulinterne Leistungsförderung (HiLF).
None declared.