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.
The internet enables consumers to evaluate products before purchase based on feedback submitted by like-minded individuals. Displaying reviews allows customers to assess comparable experiences and encourages trust, increased sales, and brand positivity. Customers use reviews to inform decision making, whereas organizations use reviews to predict future sales. Prior studies have focused on manufactured products, with little attention being paid to health care services. In particular, whether patients prefer to use websites to discuss doctors’ reputation has so far remained unanswered.
This study aims to investigate how patient propensity to post treatment experiences changes based on doctors’ online reputation (medical quality and service attitude) in delivering outpatient care services. Further, this study examines the moderating effects of hospitals’ (organizational) online reputation and disease severity.
Fractional logistic regression was conducted on data collected from 7183 active doctors in a Chinese online health community to obtain empirical results.
Our findings show that patients prefer to share treatment experiences for doctors who have a higher medical quality and service attitude (βservice attitude=.233;
Our research contributes to both theory and practice by extending the current understanding of the impact of individual reputation on consumer behavior. We investigate the moderating effects of organizational reputation and consumer characteristics in online health communities.
In seeking health care provision, patients often face uncertainty regarding the quality of doctors’ services, lacking trustworthy channels for accessing information such as medical quality and bedside manner [
Internet-based media play an important role in providing prepurchase information and informing decisions. These burgeoning new media have been hailed as a democratizing force that enables consumers to discuss products and services online [
Reviews, which are generally agreed to be more effective than traditional advertising [
We argue that reputation, signaled by existing reviews, can predict future reviews. Data were collected from an online health community, which, in recent years, has helped patients find doctors, book outpatient care services, and search for medical information. Unlike extant literature on manufactured products, our study includes both medical quality and service attitude, which are important factors in the health care field, as part of the doctor’s reputation in our model [
In recent years, online health care communities have been developed by patient organizations, medical service providers, and nonprofit organizations to make it easier for patients to find health-related information [
In China, as a result of continued limitations in existing health care provisions, online health communities have been strongly adopted by citizens. China has the world’s largest population and thus represents a huge resource-consumption country. China’s large population generates a variety of unique health care needs and, therefore, exhibits unique behaviors within online health care communities. Health ultimately concerns everyone, and with the emergence of online health care communities, patients now have more channels to find doctor information, whereas doctors have more choices in the way they deliver medical treatment. On the basis of extant literature, we have found few studies that explore the effects of doctors’ reputation on patient propensity to post treatment experiences and the moderating effects of hospitals’ reputation and disease severity. Our research, therefore, aims to fill these gaps.
Hansen [
Expectation-confirmation theory is widely used to explore consumer behavior in both product marketing [
Consumer reviews are an important criterion that impacts consumer behavior. However, existing literature rarely investigates the relationship between doctors’ online reputation and patient propensity to post treatment experiences online. We sought to examine how doctors’ medical quality and service attitude affect their patient propensity to post treatment experiences. Moreover, we attempted to investigate the moderating effects of the hospital’s reputation and disease severity.
Conceptual model.
Nowadays, the internet enables consumers to easily post opinions and express thoughts, feelings, and viewpoints on products and services to the wider online community [
In online health communities, patients hold comparatively high expectations about service quality for doctors with a high reputation. High expectations are less likely to be reached by perceived quality. The degree of expectation would affect consumer satisfaction and their propensity to post about treatment experiences. Higher expectations cause patients to be easily disappointed and dissatisfied after receiving services, which leads to them sharing negative feelings with others online [
An organization’s reputation helps consumers make informed choices when they feel uncertain about a product or service [
On the basis of the theory of psychological choice [
With regard to online health communities, a hospital’s reputation can be treated as an environmental factor. The delivery process of a signal varies among different hospital environments. Thus, the hospital’s reputation can moderate the effect of a doctor’s reputation. In reducing patients’ perceived risks and increasing trust in the doctor’s reputation, a higher hospital reputation can make patients have a higher expectation about doctors’ performance. On the basis of the expectation-confirmation theory [
On the basis of the theory of psychological choice [
In the health care field, patient behavior is also influenced by their characteristics. Disease severity is an important basis for distinguishing between patients. Prior research has indicated that disease severity moderates the doctor’s reputation on the patient’s purchasing behavior [
From a positive perspective, disease severity may influence the patient’s physical and mental health [
From a negative perspective, patients with severe diseases often concentrate less on service attitude [
On the basis of the aforementioned insights, we plan to determine the advantages of these effects in specific contexts. We propose both positive and negative moderating effects of disease severity:
In this section, we describe the research context and data collection process and present the variables and models.
We test our hypotheses using data collected from the WeDoctor website, a leading online health community authorized by the Chinese Health and Family Planning Committee. WeDoctor has become the leading online health community in China, mainly providing appointment booking services for outpatient care. The website helps increase efficiency for both patients and hospitals. Using the WeDoctor website, patients can easily make appointments and save valuable time. By 2020, the community has helped more than 850 million citizens. The WeDoctor website started to provide online written consultation and video consultation services in September 2016. In our proposed model, we do not include written and video consultation services for 2 reasons. First, compared with the outpatient care service appointment function, written and video consultation services are rarely used by patients. Second, our data were collected in the first half of 2016 when only outpatient care appointment services were provided by the website.
More than 7800 hospitals and 260,000 doctors are active in the online community. WeDoctor creates home pages for doctors and their hospitals. Doctors can self-manage their home pages, including modifying schedules for outpatient care services and updating individual information. The website has a formal and comprehensive reputation mechanism, which is important for this study. Patients can post their treatment experiences after receiving outpatient care services in the hospitals. Treatment experiences help future patients make better choices.
We used a crawler to automatically download doctors’ information from the WeDoctor website using the following selection criteria. First, we crawled all active doctors who usually add or modify their outpatient care service information or other individual information (active doctors are recognized by WeDoctor). Second, we selected doctors who treat severe diseases and who treat relatively less severe diseases. Severe diseases include malignant tumors and heart and cerebrovascular diseases. Less severe diseases include endocrine, digestive, and nervous system diseases. The reasons for choosing these disease categories will be explained in detail in the following section. We repeated the collection process in 2 time periods: one week in March 2016 and another week in June 2016. We included in our analyses the doctors who were seen at both collection times, yielding a sample of 7183 doctors. For each doctor, we collected their reviews, reputation, and other relevant information (eg, hospital information). We also collected information on the medical departments with which the doctors were affiliated.
From each doctor’s home page, we collected information posted about patients’ experiences. Each patient can give a score to the doctor’s medical quality and service attitude observed during treatment. Other patients can then read these reviews to make informed decisions.
A doctor’s home page on the WeDoctor website.
A hospital’s home page on the WeDoctor website.
The variables used in this study are in the form of aggregated data at the doctor level, which can help control for the potential influence of patient heterogeneity [
Variable definitions.
Variable | Definition | |
|
||
|
Patient propensity to post treatment experiences | The ratio of the increment of the treatment experience to the increment of outpatient care service demands over 3 months for each doctor. |
|
||
|
Medical quality | Patients give an evaluation score for doctors’ medical quality when patients share treatment experiences. The WeDoctor calculates the mean of medical quality for each doctor based on all the existing treatment experiences posted by patients. The range of values for medical quality is from 0 to 1, with a greater value indicating a higher medical quality. |
|
Service attitude | Patients give an evaluation score for doctors’ service attitude when patients share treatment experiences. The WeDoctor calculates the mean of service attitude for each doctor based on all the existing treatment experiences posted by patients. The range of values for service attitude is from 0 to 1, with a greater value indicating a higher service attitude. |
|
||
|
Hreputation | When patients post experiences, they also give a score on the hospital’s environment and attitude of guide service. The range of values for the hospital’s online reputation is from 0 to 10, with a greater value representing a higher level of satisfaction. |
|
Disease_severity | The severity of disease that patients get. We use one dummy variable to measure it. When the disease is high-risk, the variable is equal to 1. |
|
||
|
Dtitle_dummy1 and Dtitle_dummy2 | Doctors’ medical skills as evaluated by the government, including Chief Doctor, Associate Chief Doctor, and Attending Doctor. Two dummy variables were used. (0, 0) represents Attending Doctor title or below. |
|
Hlevel_dummy | The variable indicating the comprehensive health care quality of doctor |
The dependent variable in our model is patient propensity to post treatment experiences. The variable is the ratio of the increment of the treatment experience to the increment of outpatient care service demands over a certain time period. The dependent variable is defined as follows:
where
The independent variable in our model is the doctor’s online reputation, which is divided into 2 dimensions: medical quality and service attitude. The WeDoctor website calculates the mean of medical quality and mean of service attitude for each doctor based on all the existing treatment experiences posted by patients. The range of values for both medical quality and service attitude is from 0 to 1, with a greater value indicating a higher satisfaction.
The moderating variables were the hospital’s online reputation and disease severity for patients treated in the hospital. The hospital’s online reputation reflects the integral medical quality and integral service attitude delivered by the hospital. The range of values for the hospital’s online reputation is from 0 to 10, with a greater value representing a higher level of satisfaction. We used mortality rates to distinguish the severity of different diseases. The Chinese Health Statistics Yearbook, published in 2019 [
We included both doctors’ titles and hospital levels in our model to control for their popularity offline. In China, each doctor has an offline title that represents their medical skills and level of experience, including Chief Doctor, Associate Chief Doctor, and Attending Doctor. These titles are evaluated and issued by government agencies. We use 2 dummy variables, Dtitle_dummy1 and Dtitle_dummy2, to measure doctors’ titles. Similarly, each hospital in China is assigned a rank, classified as class A, B, or C, with class A being the best quality of hospital. Hospital level, which is also evaluated and issued by government agencies, represents their medical quality and medical technical strength. As the number of class C hospitals in this online health community is very small, we combined it with class B and used 1 dummy variable, Hlevel_dummy, to measure hospital level. The detailed definitions of these dummy variables are as follows:
We use general linear model regression to obtain empirical results. Fractional logistic regression is most suitable for our dependent variable (% of patients posting treatment experiences). On the basis of all the hypotheses, the empirical models are as follows:
where
We use the expectation-confirmation theory in our hypotheses to argue that patients have higher expectations when they choose doctors with high reputations. Patients are likely to feel disconfirmed between expectation and perceived quality of the service and express their feelings online.
Description and correlation (N=7183).
Variables | Mean (SD) | Minimum | Maximum | Patient propensity to post treatment experiences | Dtitle_dummy1 | Dtitle_dummy2 | Hlevel_dummy | Dmedical_quality | Dservice_attitude | Hreputation |
Patient propensity to post treatment experiences | 0.067 (0.108) | 0 | 0.944 | —a | — | — | — | — | — | — |
Dtitle_dummy1 | 0.348 (0.476) | 0 | 1 | 0.229b | — | — | — | — | — | — |
Dtitle_dummy2 | 0.443 (0.496) | 0 | 1 | −0.050b | −0.591b | — | — | — | — | — |
Hlevel_dummy | 0.088 (0.283) | 0 | 1 | −0.180b | −0.016b | 0.037b | — | — | — | — |
Dmedical_quality | 0.338 (0.442) | 0 | 1 | 0.790b | 0.197b | −0.038b | −0.167b | — | — | — |
Dservice_attitude | 0.402 (0.471) | 0 | 1 | 0.860b | 0.208b | −0.044b | −0.175b | 0.889b | — | — |
Hreputation | 5.740 (4.192) | 0 | 10 | 0.477b | 0.082b | −0.055b | −0.328b | 0.447b | 0.486b | — |
Severity_diseases | 0.780 (0.414) | 0 | 1 | −0.329b | 0.080b | −0.022b | −0.587b | 0.280b | 0.309b | 0.537b |
aThis table is symmetrical. The number in the lower left corner is same as the at top right corner.
bCorrelation is significant at the .01 level (2-tailed), significant at .01
The analyses are deemed fit using Stata, a data analysis software. The empirical results are shown in
Hypotheses 1a and 1b concern the impact of the doctor’s reputation on patient propensity to post treatment experiences. From model 4 in
As the results of model 4 show that a hospital’s reputation has no significant impact, we focused on its margin effect, with results demonstrating that a hospital’s reputation has a positive influence on patient propensity to post treatment experiences (β=.001;
Hypotheses 2b and 2c test the moderating effects of organizational reputation on the relationship between individual reputation and consumer behavior. From model 4 in
Hypotheses 3b and 3c examine the moderating effects of consumer characteristics (disease severity) on the relationship between individual reputation and consumer behavior. From model 4 in
To better interpret our results, we use the empirical results for the dependent variable, the increment of outpatient care service demands, and take its log value in the empirical model. The results are shown in
Results for the patient propensity to post treatment experiences. General linear model regression was used to obtain results.
Variables | Model 1, coefficient (SD) | Model 2, coefficient (SD) | Model 3, coefficient (SD) | Model 4, coefficient (SD) | Model 5, coefficient (SD) |
Constant | 0.054a (0.006) | 0.012b (0.003) | 0.026a (0.003) | 0.012a (0.004) | 0.012a (0.004) |
Dtitle_dummy1 | 0.031a (0.015) | −0.012a (0.008) | −0.011a (0.008) | −0.010a (0.008) | −0.010a (0.008) |
Dtitle_dummy2 | 0.015a (0.015) | −0.006a (0.008) | −0.005a (0.008) | −0.004a (0.008) | −0.004a (0.008) |
Hlevel_dummy | −0.041a (0.025) | 0.004c (0.014) | −0.012c (0.015) | −0.013a (0.015) | −0.013a (0.015) |
Dmedical_quality | N/Ac | 0.013a (0.015) | 0.013a (0.015) | 0.076a (0.022) | 0.052a (0.012) |
Dservice_attitude | N/A | 0.157a (0.014) | 0.158a (0.014) | 0.222a (0.019) | 0.233a (0.019) |
Hreputation | N/A | N/A | 0.001a (0.001) | 0.001 (0.001) | 0.001 (0.001) |
Severity_diseases | N/A | N/A | −0.024a (0.011) | −0.004d (0.001) | −0.004d (0.001) |
Dmedical_quality×Hreputation | N/A | N/A | N/A | 0.002d (0.001) | 0.008a (0.002) |
Dservice_attitude×Hreputation | N/A | N/A | N/A | −0.004d (0.002) | −0.007a (0.002) |
Dmedical_quality×Severity_diseases | N/A | N/A | N/A | −0.036a (0.014) | −0.039a (0.014) |
Dservice_attitude×Severity_diseases | N/A | N/A | N/A | −0.044a (0.012) | −0.063a (0.012) |
Dmedical_quality×Hreputation×Severity_diseases | N/A | N/A | N/A | N/A | −0.009 (0.014) |
Dservice_attitude×Hreputation×Severity_diseases | N/A | N/A | N/A | N/A | 0.004 (0.012) |
Log likelihood | −4790.53 | −4531.46 | −4233.40 | −3610.23 | −3577.21 |
Pseudo- |
0.014 | 0.015 | 0.018 | 0.020 | 0.021 |
aSignificant at .001.
bSignificant at .05.
cN/A: not applicable.
dSignificant at .01.
Results for the increment of outpatient care service demands. Ordinary least squares regression was used to obtain results.
Variables | Model 1 |
Constant | −0.188a (0.018) |
Dtitle_dummy1 | 0.399a (0.015) |
Dtitle_dummy2 | 0.143a (0.014) |
Hlevel_dummy | 0.039b (0.023) |
Dmedical_quality | 0.341c (0.110) |
Dservice_attitude | 1.859a (0.098) |
Hreputation | 0.018a (0.002) |
Severity_diseases | 0.029 (0.019) |
Dmedical_quality×Hreputation | 0.051c (0.020) |
Dservice_attitude×Hreputation | 0.050c (0.018) |
Dmedical_quality×Severity_diseases | 0.319b (0.148) |
Dservice_attitude×Severity_diseases | −0.325b (0.131) |
Adjusted |
0.785 |
aSignificant at .001.
bSignificant at .05.
cSignificant at .01.
In our study, it was found that many doctors did not receive any reviews from patients, which may have caused bias in our findings. A small increment in treatment experiences will not change the doctor’s reputation too much [
Robustness check results. General linear model regression was used to obtain results.
Variables | The increment of treatment experiences ≥1; n=4461 | The increment of treatment experiences ≥5; n=2462 | The increment of treatment experiences ≥10; n=1651 |
Constant | 0.351a (0.014) | −0.055 (0.085) | −0.319 (0.275) |
Dtitle_dummy1 | −0.029a (0.002) | −0.037a (0.003) | −0.043a (0.003) |
Dtitle_dummy2 | −0.017a (0.002) | −0.028a (0.003) | −0.033a (0.003) |
Hlevel_dummy | −0.106a (0.006) | −0.178a (0.010) | −0.179a (0.012) |
Dmedical_quality | 0.014b (0.010) | 0.108a (0.066) | 0.489a (0.154) |
Dservice_attitude | 0.028b (0.016) | 0.299a (0.155) | 0.718c (0.288) |
Hreputation | −0.012a (0.001) | −0.018 (0.008) | 0.112 (0.027) |
Severity_diseases | −0.121a (0.016) | −0.252a (0.061) | −0.170c (0.030) |
Dmedical_quality×Hreputation | 0.007c (0.002) | 0.001c (0.000) | 0.055b (0.023) |
Dservice_attitude×Hreputation | −0.006b (0.002) | −0.017c (0.002) | −0.032c (0.018) |
Dmedical_quality×Severity_diseases | −0.021c (0.014) | −0.147c (0.080) | −0.234a (0.063) |
Dservice_attitude×Severity_diseases | −0.004a (0.001) | −0.414a (0.101) | −0.206c (0.103) |
Log likelihood | −4021.50 | −5112.12 | −5825.11 |
aSignificant at .001.
bSignificant at .01.
cSignificant at .05.
This study provides valuable insights into the impact factors of sharing patient reviews in online health care communities. We study the impact of individual reputation, organizational reputation, and consumer characteristics on patient propensity to post treatment experiences and the moderating effects of organizational reputation and patient characteristics. From our results, most of the hypotheses are supported.
Our findings suggest that both medical quality and service attitude positively impact patient propensity to post treatment experiences, which is consistent with the expectation-confirmation theory [
Our results provide further evidence for the theory of psychological choice [
The moderating effects of hospitals’ reputation and disease severity. PPPTE: patient propensity to post treatment experiences.
In this study, we examine the theory of psychological choice [
Our study makes several contributions to the literature. First, this is one of the earliest in-depth studies to analyze the role of reputation in patient propensity to post treatment experiences. Prior studies have focused on the relationship between reputation and sellers’ sales in both product fields [
Second, this study contributes to the existing literature on reputation by researching the role of individual reputation, organizational reputation, and interaction effects. Prior studies have only considered reputation at one level, either individual [
Third, we enrich the existing literature on the impacts of consumer characteristics on consumer behavior. Consumer characteristics have been recognized by researchers as one of the most influential factors for different consumer behaviors [
This study also has significant practical implications. First, our results show that when patients decide whether to post treatment experience reviews, service attitude works more effectively than medical quality. Our findings also suggest that doctors need to pay more attention to their service attitude than ever before. When people have diseases, they become vulnerable and seek emotional support from doctors. Moreover, contradictions and disputes between doctors and patients have intensified, which has reached an unprecedented level in recent years, requiring doctors to improve their service attitudes. Second, not only do we find that doctors’ reputation has a positive impact on the number of reviews posted but also the hospital’s reputation; thus, to encourage more patients to post reviews online, doctors must take the impact of the hospital’s reputation into consideration. For example, doctors can move to other hospitals with higher reputation. Third, disease severity mitigates the relationship between doctor reputation and patient propensity to post treatment experiences. Compared with doctors who treat severe diseases, doctors who treat less severe diseases should pay closer attention to their online reputation. As their online reputation increases, doctors who treat less severe diseases receive a greater number of patient reviews than those who treat severe diseases.
Our study has several limitations. First, we include one online service, the WeDoctor website. Although improving the internal validity, this choice may reduce the generalizability of our findings. Other contexts should be examined in future studies. Second, we did not collect patient-level data; because of this, we could not measure demographic characteristics and specific disease severity for each patient. Future research can improve our findings by collecting data at the patient level. Third, we did not analyze the content of treatment experience reviews. These new treatment experiences may reflect different feelings and play different roles and should be investigated in future studies. Last but not least, future studies should adopt a longitudinal approach to improve our findings by addressing potential endogeneity issues and dynamic effects.
This work was supported by the National Natural Science Foundation of China grant number 72001087.
None declared.