Exploring the determinants of patients’ continuance intentions in online health communities from the network effects perspective

Abstract

Objectives: Online health communities (OHCs) facilitate patient-physician interaction and the adoption of online health services. However, few studies explored the impact of network effects on patients’ continuance intentions in OHCs. This study aims to explore the determinants affecting OHC patients’ continuance intentions based on the network effects theory and expectation confirmation model (ECM). Methods: An integrated research model and relative hypotheses are proposed. A total of 420 valid responses are collected through an online questionnaire survey to test the research framework using structural equation modeling. Results: The results reveal that direct network effect, cross network effect, and indirect network effect all positively affect perceived ease of use, and the latter two also positively affect perceived usefulness that further affect continuance intention. In addition, other results are consistent with the ECM-based hypotheses and the positive impact of perceived e-health literacy on continuance intention is also explained. Conclusion: Patients’ continuance intention to use OHCs can be improved by network effects through direct, cross, and indirect formats. ECM-based determinants, including confirmation, perceived usefulness, and satisfaction, provide valuable insights for OHC patients’ continuous use. Enhancing e-health literacy helps maintain patients’ intention to continue using OHCs.

Keywords

continuance intention expectation confirmation model network effects online health community perceived e-health literacy

Introduction

With the rise of digital technology and its application in the health field, the scope of medical and health services has gradually formed a closed-loop service ecosystem that combines online and offline medical resources. Online health communities (OHCs) construct broad communication platforms focusing on health-themed interactions and enable patients to continuously obtain medical knowledge and emotional support from their physicians and peer patients.^1–3 Patients can review the physicians’ information and expertise shown on OHCs, adopt numerous valuable information posted on the platforms, and consult with physicians to establish knowledge-based support.^4–6 OHCs such as Haodf (haodf.com) and WeDoctor (wedoctor.com) have achieved rapid development under China’s new policies on “internet plus healthcare.”⁷ For example, 280,000 active physicians has been certified by real name and proof of employment work, and 84 million patients have been cumulatively served on Haodf. More than 260,000 active physicians provide online medical services on WeDoctor, such as health assessment, diagnosis, counseling, and medication guidance.⁸ There are also abundant opportunities for theory development and empirical analyses in OHCs, such as interactions between peer patients and between patients and their physicians.^9,10 Thus, it is evident that the development of OHCs provides patients with opportunities for health-themed interaction and a new model of health management.

However, patients’ negative continuance intention remains an issue limiting the development of OHCs. It has been found that 45% of United States users who adopt or download health-related mobile apps had stopped using them.¹¹ As one of the technical carriers providing services to patients, mHealth is a new paradigm of emerging information technology (such as smartphones, wearable devices, etc.) in the field of health care.¹² Although mHealth can provide convenient medical services and health management through mobile devices, it also faces the threat of user interruption.¹² A recent survey shows new health apps are being developed quicker than demand.¹³ The aforementioned technologies serve as carriers for the development and existence of OHCs, facilitating convenient and flexible interaction and communication between patients and physicians. However, the unsound and unsustainable development of health-related information technology will have a negative impact on OHC users’ continuance intention. Continuance intention refers to one of the primary post-adoption behaviors,¹⁴ which represents a stable performance of user behavior and creates the cornerstone for the survival of information technology products.^15,16 OHCs’ long-term viability and sustainable growth depend more on users’ continuance behavior than their initial adoption decisions. Moreover, continuous usage significantly reduces costs or gains competitive advantages.¹⁷ Accordingly, this study is designed to explore patients’ intention for continuous use of OHCs and rationally apply the mechanism to enhance the process of OHCs’ value creation.

Social interactions are often structured as networks of participants and their relationships and influence consumer behavior through network externalities.¹⁸ Network externalities refer to “the utility that a user derived from the consumption of a good increase with the number of other agents consuming the good,”¹⁹ which affect individuals by affecting their perceived utility of the technology itself and the social benefits of social interactions. With the increasingly complex online social networks, the factors that drive network effects, such as perceived network size or perceived complementarity, namely network externalities, widely exist in the interaction between patients and physicians. For example, the health utility obtained by patients in OHCs not only depends on their e-health literacy and the OHC itself, but also is influenced by the number of physicians on the other side of the platform or the compatible services. It has been demonstrated that network externalities (or network effects) play a significant role in shaping and reshaping users’ attitudes and behaviors toward information technologies.^20,21 However, little attention has been paid to addressing the determinants of OHC users’ continuance intention from an external network platform perspective. Since network effects influence OHC patients’ value creation and capture, the impact of networks on patients’ continuance intention in OHCs should be addressed.

Users’ continuance intention of new technologies is a complex mechanism and the single theory-driven model is insufficient to explain. Thus, it is necessary to establish a complete theoretical model to explain patients’ continuation intention. To explore how patients make continuous use decisions, the expectation confirmation model (ECM) has been widely used to focus on the user’s continuance intention at the post-adoption stage.^22–24 Under the influence of network effects, the value or utility patients obtain from OHC services will be influenced by patients and physicians on both sides of the platform. Combining the ECM with network effects theory can fit the platform characteristics of OHCs and explore the influence mechanism of the determinants on patients’ continuance intentions. Additionally, it is also underlined that electronic health tools cannot function effectively if the target users lack the necessary skills to interact with them.²⁵ Since consumers’ capacity to assess online health information varies according to their level of e-health literacy, users’ actual use effect in adopting OHC services varies. However, there is a gap in the literature investigating the effect of e-health literacy on continuance intention in the context of OHCs. Thus, this study aims to propose an integrated framework to comprehensively explore patients’ continuance intention in OHCs.

Considering the limitations of the existing research, this study focuses on patients’ continuance intention for OHCs’ services from the perspective of network effects and integrates the external factors into the ECM to propose a comprehensive framework. The contributions of this study are as follows. First, we find three types of network effects exist in patient-centered social networks, in which patients’ continuance intention in OHCs is influenced by the patient, physician, and OHC provider. Second, we propose a unified framework of patients’ intention to continue to use OHCs by integrating external key determinants and give a clear explanation of their mechanism. Third, we conduct an empirical analysis and provide reasonable explanations for creating user value and retaining users in OHCs. The findings and implications of the study are elaborated in later sections.

The remainder of this study is organized as follows. This study first introduces the theoretical background and proposes the research model with research hypotheses. Based on the empirical results, this study has fully discussed the findings, implications, limitations, and the conclusion.

Methods

Theoretical framework

According to a systematic literature review about the continuance intention of online technologies,²⁶ expectation confirmation theory (ECT) is one of the most frequently used theories in the field of technology continuance intention. Originating from ECT and TAM,^27,28 Bhattacherjee proposed the original model of ECM for IS continuance intention and explored the effects of post-adoption expectations and subsequent cognitive processes,²⁹ in which the substantive differences between initial adoption and continuance intention were drawn. By combining context-specific factors or theories, the ECM has been extended and applied to information systems and technologies in different fields.^24,30–32 From the user’s internal perspective, satisfaction, perceived usefulness (PU), and perceived ease of use (PEOU) are significant explanatory variables of the continuance intention of OHC services. In order to improve the explanatory and predictive power of the ECM, it is necessary to extend the theoretical framework and practical application. Therefore, this study extracts the basic constructs from the ECM and further explains the influence of external constructs on patients’ informed health decisions and continuance intention while adopting OHCs.

With the construction and development of digital technologies and the platform economy, more and more entities are pouring into the platform, which contributes to forming a more complex multi-sided market. From the patient’s perspective, the OHC has a typical two-sided market structure, that is, same-sided patients to patients and cross-sided patients to physicians. Network effects arise and affect the utility or value that patients derive from the interaction with physicians and the use process of OHC. According to the network effects theory, the number of consumers and complementary services determines how a technology network is created and captured, ultimately affecting the responses and behaviors of its users.^19,33 Network effects and network externalities can be explained from a causal perspective, including network size and complementary services, which are important factors driving network effects.³⁴ Previous studies conceptualized direct and indirect network externalities as referent network size and perceived complementarity.^21,35,36 For example, in online service scenarios, including micro-blogging,³⁷ WeChat,³⁸ MOOC,³⁹ and IoT services,⁴⁰ the more existing users, the easier it is to share information between users and the more extensive connections can be established. In addition, perceived complementarity with the availability or accessibility has been proven to impact the value of a product or service. Meanwhile, researchers focused on the network effects from the same and cross side network in two-sided platform markets.^40–42 The OHCs can be regarded as typical social network platforms, in which various valuable information from the patient-reported and physician-reported posts establishes knowledge-based online support. Based on the characteristics of the OHC platforms and their network structure, we summarize three types of network effects from the patients’ perspective. Figure 1 shows the structure and content of three network effects within OHCs.

Figure 1.

Network effects within OHCs. Notes: Direct network effect (DNE) or cross network effect (CNE) refers to the size and growth of patients are influenced by the value creation and capture processes of the same-sided patients/cross-sided physicians. Indirect network effect (INE) means that the value of OHC services increases with the availability, accessibility, variety and volume of complementary products or services of OHCs.

In line with the analyses above, network effects theory provides potential determinants for this study to explore factors affecting patients’ continuance intention, and the ECM establishes a research framework for continuance intention. By combining network effects theory and the ECM, on the one hand, we provide new insights from the platform structure perspective of OHCs, explaining why the utility of patient acquisition is affected by patients on the same side and physicians on the other side of the platform, as well as platform services. On the other hand, we explain the influence mechanism of the key determinants such as PU and satisfaction on the continuance intention through ECM.

Research hypotheses

The proposed model is presented in Figure 2. To improve the readability of this paper, we provide the abbreviations and definitions of the constructs of the research framework in Table 1.

Figure 2.

Research model. Notes: H1 to H5 are extracted from the original ECM, which illustrate the direct or indirect effect of confirmation, satisfaction, and PU on continuance intentions, respectively. H6 and H7 indicate that PEOU positively influences PU and continuance intention. H8a, b, c and H9a, b, c indicate that NEs are positively associated with PU and PEOU. H10 and H11 indicate that PEHL positively influences confirmation and PU.

Table 1.

Critical constructs of the proposed research model.

Construct	Abbr.	Definition
Confirmation	CON	Matching of user’s expectation with the performance of IS or technology.²⁹
Perceived usefulness	PU	The prospective user’s subjective probability that using a specific application system will increase job performance.²⁷
Perceived ease of use	PEOU	The need for cognitive efforts required to use a particular system.²⁷
Satisfaction	SAT	A positive psychological or affective state that arises from the cognitive appraisal of using technology.²⁹
Continuance intention	CI	User’s intention or behavior to continue using information systems.²⁹
Network effect	NE	The utility or value that users of a product or service derive from the network of other users using similar or compatible products.³⁴
Perceived e-health literacy	PEHL	User’s perceived ability to use a health technology to improve health status.⁴³

The baseline or reference level for users to make evaluations of the good or service is provided by the user’s confirmation of prior expectations.²⁹ Previous experience and judgment determine user’s willingness to continue using and further confirm whether the user’s initial expectations are met.⁴⁴ Confirmation affects both user’s satisfaction and the PU, which will further affect users’ willingness for continuous use.^31,45 Under the basic framework of the ECM, patients’ confirmation of their expectations about the OHCs will influence corresponding satisfaction and PU. For example, if the OHC services exceed the demands of patients’ expectations, the level of satisfaction and PU will be positively impacted. Therefore, we propose that:

Patients’ confirmation with OHC services positively influences PU.

Patients’ confirmation with OHC services positively influences satisfaction.

PU can measure the post-expectations of the performance after the confirmation,⁴⁴ which will lead to positive or negative psychological changes. It is verified that PU directly affects continuance intention or indirectly affects it through satisfaction in different online scenarios,^44,46 especially in the field of health. While OHCs provide users with medical convenience or serve users’ medical purposes well, patients will be willing to continue to seek medical services.

The positive impact of satisfaction on continuance intention is consistently supported by research on various types of digital technologies.²⁶ Previous research demonstrated that satisfaction is a powerful predictor of continuance intention.^44,47 Positive affective state with information systems usage encourages users to use the technology more, while negative experiences may lead to abandonment and avoidance behavior. As for OHCs patients’ usage continuance intention, it is critical that patients feel satisfied with their overall experience of OHCs use, which will directly or indirectly affect the final continuance intention. Therefore, we propose that:

Patients’ PU positively influences satisfaction.

Patients’ PU positively influences continuance intention.

Patients’ satisfaction positively influences continuance intention.

PEOU is a significant factor in influencing users’ perceptions and determining technological continuation intention.^30,48 Patients are more likely to have a favorable experience and believe that OHC services are highly useful when they find them easy to use. In the context of OHCs, if patients perceive the ease of use of OHC services, they are more likely to perceive the usefulness and maintain an intention of continuance usage. Therefore, we propose that:

Patients’ PEOU positively influences PU.

Patients’ PEOU positively influences usage continuance intention.

With the increasing numbers of users utilizing network products or services, existing users are more likely to obtain the network benefits, such as practical value and hedonic benefit.³⁷ For example, patients are more likely to be attracted to OHCs with more active patients and physicians. As more patients and physicians participate in the OHCs, the OHC provider will gather more feedback on the quality of the platform’s services, thereby improving the service quality and usability. Moreover, it has been confirmed that there is a positive relationship between perceived network size and PEOU.³⁷ Therefore, we propose that:

H8a

DNE is positively associated with PU.

H9a

DNE is positively associated with PEOU.

Various valuable information from patient-reported posts can help establish knowledge-based online support through the interaction between patients and physicians. In online service scenarios,^20,37,39 the more existing users, the easier it is to share information between users and the more extensive connections can be established. The network benefits like referent network size and complementary products positively affect user’s usage intention and PU.³⁴ Moreover, based on a wider user base and more resource content, the user’s PEOU will enhance with the expansion of the network size.³⁷ Therefore, we propose that:

H8b

CNE is positively associated with PU.

H9b

CNE is positively associated with PEOU.

INE, such as perceived complementarity, is regarded as a factor influencing the value of a product or service with its availability or accessibility.^34,39 PU and PEOU are closely related to the complementarity or availability of products in the context of OHCs. A wider variety of lower and upper tier “complementary services” support consumers to benefit more from the use of technology networks.⁴² Consistent with the above analysis, DNE and CNE basically stem from the network size of patients and physicians, while INE is closely related to perceived complementarity and perceived compatibility of the OHC services. Therefore, we propose that:

H8c

INE is positively associated with PU.

H9c

INE is positively associated with PEOU.

The knowledge and information of OHCs patients exposed to and actually obtained may be inconsistent due to the influence of various factors, such as the accuracy of the information itself and the e-health literacy of the users.^25,49,50 e-health literacy affects the adoption of health-related information and the outcomes obtained by users in e-health activities. It has been noted that an individual’s motivation or assessment to engage in online health information searching is related to their level of e-health literacy.⁴³ The ability and competencies of users to use e-health tools are closely related to the actual use effect of the users. Those with high levels of e-health literacy are more inclined to assume the role of health information searchers and to carefully consider the authenticity and dependability of their sources of information.⁵¹ The positive effect of e-health literacy on PU, confirmation, and perceived usability has been concluded.⁵² For example, in some cases, using e-health tools will be difficult and inconvenient for users who lack the necessary health literacy, negatively impacting the effectiveness of using OHCs. Therefore, we propose that:

H10

PEHL positively influences the confirmation with OHCs services.

H11

PEHL positively influences PU of the OHC services.

Measurement development

To verify the model and hypotheses proposed in this study, this study originally designed the content of questionnaire survey (See Appendix A of supplemental material for further details). The complete scales of measurement items are consistent with the previous research in the discipline of information system. The scales for DNE, CNE and INE are adopted from the research on network effects^{19,33,35,41,42}; the scales for PEHL are adopted from studies^25,43,53; the scales for PU and PEOU are adopted from studies.^27,54; other scales related to the ECM are adopted from studies.^28,29,55 These multiple-item scales were measured using a seven-point Likert scale anchored from strongly disagree (1) to strongly agree (7). The respondents in our study are native Chinese OHC users, so we made appropriate revisions and fine-tuned terms to fit Chinese medical context. To ensure the reliability and validity of the measurement scales, a pilot-test was conducted on 56 participants using OHCs and minor changes were made to the wording of some questions based on pilot participant feedback.

Data collection and sample

Before starting the questionnaire, participants need to read the informed consent form. After the written informed consent was confirmed, respondents could continue to participate in the questionnaire. Because our research context is in OHCs, the questionnaire set preconditions that only participants with experience in OHC use can participate. The content of the questionnaire survey consists of the basic demographic statistics and the evaluation scales of the research model, in which we also set up an open multiple-choice question to invite participants to answer which OHCs they use.

From August 2022 to October 2022, the empirical data were collected through a questionnaire survey on an online platform (Sojump) from China, and the Sojump-based questionnaire link was sent to participants by email, mobile phone message, or social media. 452 questionnaires were distributed, and 420 valid questionnaires were obtained after discarding incomplete or invalid responses. Among the valid responses, the first three commonly used OHCs are Haodf (haodf.com) with 287 respondents, Ping An Good Doctor (pagd.net) with 274 respondents, and WeDoctor (wedoctor.com) with 271 respondents. The OHCs mentioned above are leading OHCs in China, which provide community-based healthcare and enable patients to continuously obtain medical knowledge and emotional support from their physicians and peer patients.^3,10 For example, Haodf comprises the largest number of high-quality authoritative physicians, and the number of active patients on the platform is also large, which promotes knowledge and emotional interaction between physicians and peer patients.¹⁰

According to the demographic statistics, the proportion of male (54.76%) and female (45.24%) respondents is close. 45% of the respondents are between 18 and 30 years old. And more than half of the respondents have higher education level (bachelor’s degrees and above). Table 2 shows the respondents’ demographic statistics and their OHC usage information.

Table 2.

Demographic statistics and OHCs use information of respondents.

Item	Category	Frequency	Percentage (%)
Gender	Male	230	54.76
Gender	Female	190	45.24
Age	<18	32	7.62
	18–25	76	18.10
	26–30	113	26.90
	31–40	102	24.29
	41–50	65	15.48
	51–60	21	5.00
	>60	11	2.62
Education level	High school or less	86	20.48
	Vocational educat.	80	19.05
	Bachelor’s degree	128	30.48
	Master’s degree	74	17.62
	Doctoral degree	52	12.38
Employment status	Students	52	12.38
	Working professionals	115	27.38
	Public institution staff	69	16.43
	Business staff	99	23.57
	Agricultural laborers	34	8.10
	Retired	44	10.48
	Unemployed	7	1.67
Physical condition	Healthy state	74	17.62
	Sub-healthy state	227	54.05
	Unhealthy state	119	28.33

In this section, we established the research framework and proposed corresponding hypotheses based on ECM and network effects theory. Referring to the existing well-established scales, data were collected through questionnaires and used to analyze the research model.

Results

To analyze the hypotheses proposed in the research model, SPSS 20.0 and Amos 22.0 software were employed to test the measurement and structural model. The use of structural equation modeling (SEM) will confirm the robustness of the findings as it is based on the maximum likelihood algorithm, which takes into account the error terms in establishing loadings, correlations and other relevant measurements.⁵⁶

Common method bias

Since this questionnaire survey was perception-based and all the measurement scales of this study were self-reported, there were potential threats in terms of common rater effects, item characteristic effects and measurement context effects.⁵⁷ Thus, we need to rule out the potential for common method bias (CMB). We used two techniques to test the CMB. First, we conducted Harman’s single-factor test.⁵⁸ The exploratory factor analysis extracted nine factors with the first factor accounting for 36.648% of the variance, which is lower than the critical threshold (40%).⁵⁷ Second, we adopted a common-method factor test introduced by Liang et al.⁵⁹ We introduced the common-method factor by creating a latent variable that directly correlates with each construct’s indicators. The result demonstrates that the average variance substantively explained by the principal constructs and the common-method factor are 0.681 and 0.011, respectively. And the ratio between them is about 62:1, which has sufficient explanatory and persuasive power. Moreover, most indicators’ loading on the common-method factor is not significant. Thus, the CMB is unlikely to be a serious concern for this study. Detailed results are provided in Appendix B of the supplemental material.

Measurement model analysis

Before the assessment of the quality of the measurement, we examined the normality of the data through the skewness and kurtosis test. The results demonstrated that the absolute values of skewness and kurtosis for all constructs are less than 2, indicating these values were within the criterion and satisfied further structural equation analysis.⁶⁰

Reliability analysis is mainly used to examine the stability and consistency of the results measured by the scales in the questionnaire. Considering the accuracy of the results and the ease of operation, this study adopts Cronbach’s alpha to examine the reliability of the constructs and indicators, which is the most commonly used indicator in current research.^16,61,62 Cronbach’s alpha values of all constructs are larger than 0.8 and all the loadings of each indicator are above 0.7, indicating that the measures have good reliability.^63,64 For the convergent validity,^64,65 all average variance extracted (AVE) of each construct exceeds 0.5, such that the latent variable explains more than half of the variance observed in its indicators. And the composite reliability (CR) exceeds 0.7, suggesting a good convergent validity. The detailed results are shown in Table 3.

Table 3.

Standardized item loading, cronbach’s alpha, CR and AVE values.

Construct	Item	Factor loading	Cronbach’s alpha	CR	AVE
Direct network effect (DNE)	DNE1	0.862	0.910	0.911	0.719
	DNE2	0.819
	DNE3	0.847
	DNE4	0.862
Cross network effect (CNE)	CNE1	0.829	0.897	0.897	0.686
	CNE2	0.818
	CNE3	0.811
	CNE4	0.854
Indirect network effect (INE)	INE1	0.773	0.893	0.893	0.676
	INE2	0.837
	INE3	0.839
	INE4	0.837
Perceived E-health literacy (PEHL)	PEHL1	0.826	0.933	0.933	0.666
	PEHL2	0.781
	PEHL3	0.834
	PEHL4	0.822
	PEHL5	0.829
	PEHL6	0.818
	PEHL7	0.801
Perceived usefulness (PU)	PU1	0.821	0.888	0.888	0.665
	PU2	0.790
	PU3	0.836
	PU4	0.815
Perceived ease of use (PEOU)	PEOU1	0.829	0.894	0.894	0.678
	PEOU2	0.827
	PEOU3	0.833
	PEOU4	0.805
Confirmation (CON)	CON1	0.822	0.896	0.896	0.683
	CON2	0.848
	CON3	0.818
	CON4	0.817
Satisfaction (SAT)	SAT1	0.890	0.899	0.898	0.689
	SAT2	0.836
	SAT3	0.812
	SAT4	0.778
Continuance intention (CI)	CI1	0.886	0.901	0.901	0.696
	CI2	0.806
	CI3	0.798
	CI4	0.843

As demonstrated in Table 4, the square roots of AVEs for each construct (diagonal elements) are higher than the correlation between each pair of constructs (nondiagonal elements), thereby indicating a good discriminant validity.⁶⁵ Hence, all the criteria are met.

Table 4.

Discriminant validity.

Construct	DNE	CNE	INE	PEHL	PU	PEOU	CON	SAT	CI
DNE	0.848
CNE	0.439	0.828
INE	0.430	0.502	0.822
PEHL	0.470	0.491	0.436	0.816
PU	0.386	0.417	0.433	0.416	0.816
PEOU	0.437	0.395	0.463	0.403	0.448	0.824
CON	0.423	0.369	0.365	0.375	0.484	0.466	0.826
SAT	0.495	0.434	0.433	0.488	0.436	0.510	0.487	0.830
CI	0.579	0.519	0.530	0.534	0.530	0.557	0.517	0.579	0.834

Notes: Diagonal elements (in bold) represent square root of AVE value for the corresponding construct while other nondiagonal values represent the correlations.

According to the aforementioned results, all nine constructs of the measurement model have strong internal consistency, reliability, convergent validity, and discriminant validity, which confirm the validity of the questionnaire and support this study to obtain reliable results.

Structural model analysis

To examine the goodness-of-fit (GOF) of the research model, we measured the fit and assessment indices. As shown in Table 5, various GOF indices of the research model indicated a reasonably good model fit.^56,66 In order to further illustrate the robustness of the proposed model, we added age, gender, education level and employment status as control variables to test their effects. Cohen f² was adopted to evaluate the effects of control variables and the multivariate coefficient of determination (R²) was applied to calculate Cohen f ².^67,68 As shown in Table 6, it can be concluded that the control variables had insignificant or limited effects on our research model.

Table 5.

Model fit summary.

Goodness of fit measures	Parameter estimates	Goodness of fit ranges	References
Chi square/Degrees of freedom (CMIN/DF)	1.561	1–3	Kline⁶⁰; Gefen et al.⁶⁹; Hu and Bentler⁷⁰
Goodness of fit index (GFI)	0.890	>0.9
Root mean square error of approximation (RMSEA)	0.037	<0.05
Comparative fit index (CFI)	0.967	>0.9
Tucker-lewis index (TLI)	0.964	>0.9
Incremental fit index (IFI)	0.967	>0.9
Normed fit index (NFI)	0.913	>0.9

Table 6.

Multivariate coefficient of determination (R²) results.

Variables	R²		Control variables effects
Variables	With control variables	Without control variables	ΔR^2a	f ^2b	Effects
CON	0.128	0.119	0.009	0.010	Insignificant
PU	0.307	0.305	0.002	0.003	Insignificant
PEOU	0.276	0.251	0.025	0.035	Small
SAT	0.245	0.235	0.010	0.013	Insignificant
CI	0.421	0.406	0.015	0.026	Small

Notes: ^aΔR²: R² with control variables - R² without control variables; ^bf²: Cohen f².

Figure 3 shows the standardized path coefficients and corresponding level of significance.

Figure 3.

The results of the research model. Notes: *p < .05; **p < .01; ***p < .001. 14 of the 15 hypothesized associations were significant at p < .001 or p < .05, with H8a being insignificant.

The detailed path coefficients and the corresponding statistics are shown in Table 7. From the perspective of the ECM, the effect of confirmation on PU is significant (H1, β = 0.250 and p < .001). Confirmation (H2, β = 0.361 and p < .001) and PU (H3, β = 0.287 and p < .001) significantly impact satisfaction. PU (H4, β = 0.285 and p < .001) and satisfaction (H5, β = 0.322 and p < .001) significantly impact continuance intention. In addition, PEOU (H6, β = 0.150 and p < .001; H7, β = 0.308 and p < .001) significantly affects PU and continuance intention.

Table 7.

The results of hypothesis testing.

Hypothesis	Relationship	Std.Beta	Std.Error	t-value	p-value	Result
H1	CON —> PU	0.250	0.054	4.776	0.000	Supported
H2	CON —> SAT	0.361	0.06	6.650	0.000	Supported
H3	PU —> SAT	0.287	0.058	5.331	0.000	Supported
H4	PU —> CI	0.285	0.06	5.255	0.000	Supported
H5	SAT —> CI	0.322	0.05	6.502	0.000	Supported
H6	PEOU —> PU	0.150	0.059	2.589	0.010	Supported
H7	PEOU —> CI	0.308	0.055	6.284	0.000	Supported
H8a	DNE —> PU	0.069	0.057	1.168	0.243	Not supported
H8b	CNE —> PU	0.125	0.061	2.049	0.040	Supported
H8c	INE —> PU	0.155	0.059	2.528	0.011	Supported
H9a	DNE —> PEOU	0.263	0.053	4.676	0.000	Supported
H9b	CNE —> PEOU	0.150	0.058	2.540	0.011	Supported
H9c	INE —> PEOU	0.286	0.056	4.775	0.000	Supported
H10	PEHL —> CON	0.404	0.052	7.619	0.000	Supported
H11	PEHL —> PU	0.128	0.062	2.055	0.040	Supported

From the perspective of network effects, we examined three types of network effects on PU and PEOU. DNE is not statistically significant in explaining PU, and H8a is not confirmed. CNE (H8b, β = 0.125 and p < .05) and INE (H8c, β = 0.155 and p < .05) significantly affects PU. DNE (H9a, β = 0.263 and p < .001), CNE (H9b, β = 0.150 and p < .05) and INE (H9c, β = 0.286 and p < .001) significantly affect PEOU.

Focusing on the PEHL, PEHL (H10, β = 0.404 and p < .001; H11, β = 0.128 and p < .05) is statistically significant in explaining confirmation and PU. In conclusion, 14 of the 15 hypothesized associations were significant at p < .001 or p < .05, with H8a being insignificant.

In this section, the test of CMB was conducted to rule out the potential threat of inaccurate results. On the one hand, we estimated the reflective measurement model by validating the reliability, convergent validity, and discriminant validity; on the other hand, we analyzed the various paths of the structural model to provide scientific support for the conclusions drawn.

Discussion

Principal findings

This study identifies three aspects of the key findings. First, this study demonstrates that network effects act on patients’ continuance intention to use OHCs through direct, cross, and indirect formats. Direct network effect (DNE), cross network effect (CNE) and indirect network effect (INE) all play roles in continuance intention through PEOU. CNE and INE also have significant associations with perceived usefulness PU, while DNE only has a positive association with PEOU. According to the empirical results, it can be explained that the more physicians adopt and use the OHCs, the more convenient the complementary and compatible services of the OHCs, and thus the more significant the impact on patients’ PU. The path coefficients in the analysis results indicate that INE has a greater positive impact on PU than CNE. In particular, the positive effects of CNE, DNE and INE on PEOU increase in turn. INE is related to the availability, variety and quantity of complementary and compatible products or services, for example, patients’ PEOU will rise in line with the diversification and convenience of an OHC’s payment options.

It is worth noting that there is no substantial association between DNE and PU. Neither the number of active patients nor the use effect of patients from the same side of the two-sided platforms (OHCs) affect the user’s perception of usefulness. It can be explained that the degree of usefulness to patients of OHCs is not directly related to the number of patients used, but to the number and quality of physicians and additional complementary services of OHCs. Since both PU and PEOU have direct or indirect positive effects on continuance intention, such findings provide valuable insights into explaining patients’ intention to use from a network effects perspective.

Second, all hypotheses related to the ECM framework are supported in this study, which are consistent with previous research in various IS/IT adoption contexts.^22,31,45,46 According to the findings, if patients’ expectations for OHCs use are met or the experience is more positive than expected, patients will identify the high performance of OHCs and recognize the positive effect of the OHCs services at the same time. Patients will have a high sense of satisfaction and ultimately lead to a high continuance intention. Moreover, the positive effect of PU on continuance intention indicates that patients with higher PU are more likely to feel satisfied and continue using OHCs compared to patients with lower PU. Besides, it has been demonstrated that satisfaction is a necessary antecedent in explaining patients’ continuance intention.

Finally, this study also provides valuable insight into why PEHL is a crucial factor in this model. PEHL reflects the user’s technical ability to obtain health information, which ultimately positively affects patients’ continuance intention towards OHCs. The higher the patient’s PEHL, the more able to confirm the expectation from OHCs and perceive the effect after using OHCs. In addition, this study also demonstrated the positive association of the PEOU with the patients’ continuance intention, the results are in line with previous studies.^30,71

Practice implications

This study offers practitioners the basic mechanism of explaining why and how patients are motivated to continue using OHC services. First, platform operators are encouraged to incentivize the positive network effects of OHCs. According to the three forms of network effects (DNE, CNE, and INE), platform operators should mobilize the positive effects of patients on continuance intention from the perspectives of patients, physicians, and platform functions. Take the CNE as an example, the higher the number of active physicians, the better the service quality, and the stronger the CNE, which will directly improve patient’s PU and PEOU and indirectly strengthen patient users’ continuance intention.

Second, while ensuring the level of medical service supply, platform operators should also pay more attention to the convenience, friendliness and timeliness of the using process and experience. Another important aspect of the findings involves the direct impact of platform use satisfaction, PU, and PEOU on continuance intention. For example, improving the accuracy of health information, optimizing the interface for quick search functions, improving service quality to cope with unexpected results such as system crashes, and developing entertainment and social functions.

Third, it is necessary to improve patients’ e-health literacy capabilities and enhance their awareness of OHCs. In the process of continuously adopting OHCs, patients will protect their own health status, physicians will get wider employment opportunities, and OHCs can also achieve long-term and stable development, thereby forming a virtuous circle of the online medical ecosystem.

Limitations and directions for future research

This study suffers from several limitations. First, this study only focuses on the participants in China and the research findings may not be generalizable to OHC users in other countries. Future research can replicate this study and compare the findings in different economic and social environments.

Second, this study conducts a limited investigation of the roles of determinants. Other theories and factors, such as health information consumption behavior, information accessibility and the credibility of information sources, can enrich the influencing factors on user behavior and provide more comprehensive conclusions.

Third, this study simplifies the OHCs into two-sided platforms that connect physicians and patients. However, the participants in the ecological chain related to OHCs are diverse, including administrative departments, pharmaceutical e-commerce, medical insurance companies, etc. The long-term development of OHCs requires the joint efforts of these main parties, stimulating the positive network effects and ultimately achieving value co-creation. Future research needs to be extended to other participants to ensure the entire platform ecosystem’s healthy operation.

In addition, this study is a cross-sectional study, which lacks information about how the dynamic changes are associated with OHC users over time. It cannot adopt and capture the potential changes in variables. Future studies need to consider longitudinal studies with panel data to measure and compare the changes over time.

Finally, based on the structural equation modeling, the findings in this study can only be explained from the causal symmetry relationship. The influence of multiple factors on continuance intention cannot be solved from the configuration perspective. Future research can consider combining qualitative comparative analysis (QCA) method to explore the causal relationship of multiple concurrent results caused by different combinations of dependent variables.

In summary, based on the results of the empirical analysis, we confirmed the mechanism of network effects on patients’ continuance intention in OHCs, and summarized the main theoretical and practical contributions of the study. Meanwhile, it should be noted that this study still has other research opportunities in terms of research topics and research processes.

Conclusions

This study theoretically elaborates on the determinants affecting patients’ decisions on continuance intention of OHC services and empirically generates credible causal evidence about network effects by the actual data analysis. The network effects and ECM are utilized as the theoretical basis to focus on patients’ post-adoption intentions within the OHC context. 14 out of 15 hypotheses were supported in the theoretical framework established by this study. The research results prove that the network effects generated by the interaction mechanism among patients, physicians and the OHC platform exist and affect the continuous usage behavior of patients to varying degrees. Patients’ personal e-health literacy also has a positive impact on the confirmation and PU. The findings of this study make contributions to the literature on patient behavior and lead to potential interests for future research on OHCs usage continuance intention. This study also sheds light on the practical implications about patient retention and value creation through the lens of motivating network effects, which has practical guiding significance for maintaining the competitive advantage and achieving sustainable development of the OHCs.

Supplemental Material

Supplemental Material - Exploring the determinants of patients’ continuance intentions in online health communities from the network effects perspective

Supplemental Material for Exploring the determinants of patients’ continuance intentions in online health communities from the network effects perspective by Aihui Ye, Runtong Zhang and Hongmei Zhao in Health Informatics Journal.

Footnotes

Acknowledgements

The authors appreciate the support of the Beijing Logistics Informatics Research Base.

Author contributions

Aihui Ye: Conceptualization, Methodology, Formal analysis, Writing - original draft, Writing - review & editing. Runtong Zhang: Conceptualization, Funding acquisition, Writing - review & editing, Supervision. Hongmei Zhao: Conceptualization, Methodology, Data curation.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Social Science Foundation of China [grant number 18ZDA086]; and National Natural Science Foundation of China [grant number 62173025].

Ethical statement

ORCID iD

Aihui Ye

Data availability statement

The datasets and underlying research materials analyzed during the current study are available from the corresponding author upon reasonable request.

Supplemental Material

Supplemental material for this article is available online.

References

Deng

Wang

, et al. How online health community participation affects physicians’ performance in hospitals: empirical evidence from China. Inf Manag 2021; 58(6): 103443. DOI: 10.1016/j.im.2021.103443.

Cao

. Understanding knowledge sharing in online health communities: a social cognitive theory perspective. Inf Dev 2023; 39(3): 539–549. DOI: 10.1177/02666669221130552.

Liu

Guo

. The effects of participating in a physician-driven online health community in managing chronic disease: evidence from two natural experiments. MIS Q 2020; 44(1): 391–419. DOI: 10.25300/misq/2020/15102.

Zhang

Han

. Adoption of online follow-up service by patients: an empirical study based on the elaboration likelihood model. Comput Hum Behav 2021; 114: 106581. DOI: 10.1016/j.chb.2020.106581.

Liu

Perera

Wang

, et al. Physician engagement in online medical teams: a multilevel investigation. J Bus Res 2023; 157: 113588. DOI: 10.1016/j.jbusres.2022.113588.

Chen

. A study on the cross-platform influence mechanism of physicians’ live streaming behavior on performance. Internet Res. 2024. DOI: 10.1108/INTR-10-2023-0947.

Liu

Perera

Wang

. Does the physicians’ medical team joining behavior affect their performance on an online healthcare platform? Evidence from two quasi-experiments. Transp Res E Logist Transp Rev 2023; 171: 103040. DOI: 10.1016/j.tre.2023.103040.

Wang

Lei

, et al. The influence of doctors’ online reputation on the sharing of outpatient experiences: empirical study. J Med Internet Res 2020; 22(12): e16691. DOI: 10.2196/16691.

Liu

Gao

. Which voice are you satisfied with? Understanding the physician-patient voice interactions on online health platforms. Decis Support Syst 2022; 157: 113754. DOI: 10.1016/j.dss.2022.113754.

10.

Chen

Guo

, et al. Exploring the online doctor-patient interaction on patient satisfaction based on text mining and empirical analysis. Inf Process Manag 2020; 57(5): 102253. DOI: 10.1016/j.ipm.2020.102253.

11.

Krebs

Duncan

. Health app use among us mobile phone owners: a national survey. JMIR Mhealth Uhealth 2015; 3(4): e101–e119. DOI: 10.2196/mhealth.4924.

12.

Kim

Lee

, et al. Identification of critical quality dimensions for continuance intention in mHealth services: case study of onecare service. Int J Inf Manag 2019; 46: 187–197. DOI: 10.1016/j.ijinfomgt.2018.12.008.

13.

Grundy

. A review of the quality and impact of mobile health apps. Annu Rev Publ Health 2022; 43: 117–134. DOI: 10.1146/annurev-publhealth-052020-103738.

14.

Shang

. Service quality, perceived value, and citizens’ continuous-use intention regarding e-government: empirical evidence from China. Inf Manag 2020; 57(3): 103197. DOI: 10.1016/j.im.2019.103197.

15.

Elhoushy

Jang

. How to maintain sustainable consumer behaviours: a systematic review and future research agenda. Int J Consum Stud 2023; 47(6): 2181–2211. DOI: 10.1111/ijcs.12905.

16.

Wang

Feng

, et al. A novel user-generated content-driven and Kano model focused framework to explore the impact mechanism of continuance intention to use mobile APPs. Comput Hum Behav 2024; 157: 108252. DOI: 10.1016/j.chb.2024.108252.

17.

Jia

Guo

Barnes

. Enterprise 2.0 post-adoption: extending the information system continuance model based on the technology-organization-environment framework. Comput Hum Behav 2017; 67: 95–105. DOI: 10.1016/j.chb.2016.10.022.

18.

Tavasoli

Fazli

Ardjmand

, et al. Competitive pricing under local network effects. Eur J Oper Res 2023; 311(2): 545–566. DOI: 10.1016/j.ejor.2023.04.039.

19.

Katz

Shapiro

. Network externalities, competition, and compatibility. Am Econ Rev 1985; 75(3): 424–440.

20.

Sarkar

Khare

. Influence of expectation confirmation, network externalities, and flow on use of mobile shopping apps. Int J Hum Comput Interact 2019; 35(16): 1449–1460. DOI: 10.1080/10447318.2018.1540383.

21.

Luo

Wang

, et al. Examining protection motivation and network externality perspective regarding the continued intention to use M-health apps. Int J Environ Res Publ Health 2021; 18(11): 5684. DOI: 10.3390/ijerph18115684.

22.

Tam

Santos

Oliveira

. Exploring the influential factors of continuance intention to use mobile apps: extending the expectation confirmation model. Inf Syst Front 2020; 22(1): 243–257. DOI: 10.1007/s10796-018-9864-5.

23.

Khan

Loh

Hossain

, et al. Cynicism as strength: privacy cynicism, satisfaction and trust among social media users. Comput Hum Behav 2023; 142: 107638. DOI: 10.1016/j.chb.2022.107638.

24.

Ghazali

Mutum

Lun

. Expectations and beyond: the nexus of AI instrumentality and brand credibility in voice assistant retention using extended expectation-confirmation model. J Consum Behav 2023; 23(2): 655–675. DOI: 10.1002/cb.2228.

25.

Norman

Skinner

. eHealth literacy: essential skills for consumer health in a networked world. J Med Internet Res 2006; 8(2): e9. DOI: 10.2196/jmir.8.2.e9.

26.

Yan

Filieri

Gorton

. Continuance intention of online technologies: a systematic literature review. Int J Inf Manag 2021; 58: 102315. DOI: 10.1016/j.ijinfomgt.2021.102315.

27.

Davis

. Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Q 1989; 13(3): 319–340. DOI: 10.2307/249008.

28.

Oliver

. A cognitive model of the antecedents and consequences of satisfaction decisions. J Mar Res 1980; 17(4): 460–469. DOI: 10.2307/3150499.

29.

Bhattacherjee

. Understanding information systems continuance: an expectation-confirmation model. MIS Q 2001; 25(3): 351–370. DOI: 10.2307/3250921.

30.

Liu

Wang

, et al. What influences patients? continuance intention to use AI-powered service robots at hospitals? The role of individual characteristics. Technol Soc 2022; 70: 101996. DOI: 10.1016/j.techsoc.2022.101996.

31.

Bolen

. Exploring the determinants of users’ continuance intention in smartwatches. Technol Soc 2020; 60: 101209. DOI: 10.1016/j.techsoc.2019.101209.

32.

Liu

YLE

Huang

. Exploring the perceptions and continuance intention of AI-based text-to-image technology in supporting design ideation. Int J Hum-comput Int. 2024; 1–13. DOI: 10.1080/10447318.2024.2311975.

33.

Katz

Shapiro

. Technology adoption in the presence of network externalities. J Polit Econ 1986; 94(4): 822–841. DOI: 10.1086/261409.

34.

Lin

Bhattacherjee

. Elucidating individual intention to use interactive information technologies: the role of network externalities. Int J Electron Commer 2008; 13(1): 85–108. DOI: 10.2753/Jec1086-4415130103.

35.

Song

Xue

Rai

, et al. The ecosystem of software platform: a study of asymmetric cross-side network effects and platform governance. MIS Q 2018; 42(1): 121–142. DOI: 10.25300/Misq/2018/13737.

36.

Lyu

Guo

Chen

. Understanding people’s intention to use facial recognition services: the roles of network externality and privacy cynicism. Inf Technol People 2023; 37(3): 1025–1051. DOI: 10.1108/Itp-10-2021-0817.

37.

Zhao

. Enhancing perceived interactivity through network externalities: an empirical study on micro-blogging service satisfaction and continuance intention. Decis Support Syst 2012; 53(4): 825-834. DOI: 10.1016/j.dss.2012.05.019.

38.

Zhang

, et al. How WeChat can retain users: roles of network externalities, social interaction ties, and perceived values in building continuance intention. Comput Hum Behav 2017; 69: 284–293. DOI: 10.1016/j.chb.2016.11.069.

39.

Wang

Tan

. What makes MOOC users persist in completing MOOCs? A perspective from network externalities and human factors. Comput Hum Behav 2018; 85: 385–395. DOI: 10.1016/j.chb.2018.04.028.

40.

Hsu

Lin

JCC

. An empirical examination of consumer adoption of internet of things services: network externalities and concern for information privacy perspectives. Comput Hum Behav 2016; 62: 516–527. DOI: 10.1016/j.chb.2016.04.023.

41.

Chu

Manchanda

. Quantifying cross and direct network effects in online consumer-to-consumer platforms. Market Sci 2016; 35(6): 870–893. DOI: 10.1287/mksc.2016.0976.

42.

Gong

Cheung

CMK

Zhang

KZK

, et al. Cross-side network effects, brand equity, and consumer loyalty: evidence from mobile payment market. Int J Electron Commer 2020; 24(3): 279–304. DOI: 10.1080/10864415.2020.1767427.

43.

Quinn

Bond

Nugent

. Quantifying health literacy and eHealth literacy using existing instruments and browser-based software for tracking online health information seeking behavior. Comput Hum Behav 2017; 69: 256–267. DOI: 10.1016/j.chb.2016.12.032.

44.

Pereira

Tam

. Impact of enjoyment on the usage continuance intention of video-on-demand services. Inf Manag 2021; 58(7): 103501. DOI: 10.1016/j.im.2021.103501.

45.

Zhang

Yin

Guo

, et al. The role of previous affective responses in the continuance of mHealth monitoring services: a longitudinal investigation. IEEE Trans Eng Manag 2023; 71: 6982–6994. DOI: 10.1109/Tem.2023.3275924.

46.

Mishra

Shukla

Rana

, et al. Re-examining post-acceptance model of information systems continuance: a revised theoretical model using MASEM approach. Int J Inf Manag 2023; 68: 102571. DOI: 10.1016/j.ijinfomgt.2022.102571.

47.

Bhattacherjee

Lin

. A unified model of IT continuance: three complementary perspectives and crossover effects. Eur J Inf Syst 2015; 24(4): 364–373. DOI: 10.1057/ejis.2013.36.

48.

Shin

Yoon

, et al. Comprehensive senior technology acceptance model of daily living assistive technology for older adults with frailty: cross-sectional study. J Med Internet Res 2023; 25: e41935. DOI: 10.2196/41935.

49.

Crocker

Feng

OLV

Duncan

. Performance-based measurement of eHealth literacy: systematic scoping review. J Med Internet Res 2023; 25: e44602. DOI: 10.2196/44602.

50.

Zhang

Yan

Cao

, et al. The role of perceived e-health literacy in users’ continuance intention to use mobile healthcare applications: an exploratory empirical study in China. Inf Technol Dev 2017; 24(2): 198–223. DOI: 10.1080/02681102.2017.1283286.

51.

Neter

Brainin

. eHealth literacy: extending the digital divide to the realm of health information. J Med Internet Res 2012; 14(1): e19. DOI: 10.2196/jmir.1619.

52.

Sun

. Understanding determinants of end-user’s continuance intention on fitness wearable technology. Int J Hum Comput Interact 2022; 40(3): 537–557. DOI: 10.1080/10447318.2022.2124360.

53.

Norman

Skinner

. eHEALS: the eHealth literacy scale. J Med Internet Res 2006; 8(4): e27. DOI: 10.2196/jmir.8.4.e27.

54.

Venkatesh

Davis

. A model of the antecedents of perceived ease of use: development and test. Decis Sci J 1996; 27(3): 451–481. DOI: 10.1111/j.1540-5915.1996.tb00860.x.

55.

Bhattacherjee

Perols

Sanford

. Information technology continuance: a theoretic extension and empirical test. J Comput Inf Syst 2008; 49(1): 17–26. DOI: 10.1080/08874417.2008.11645302.

56.

Dhagarra

Goswami

Kumar

. Impact of trust and privacy concerns on technology acceptance in healthcare: an Indian perspective. Int J Med Inf 2020; 141: 104164. DOI: 10.1016/j.ijmedinf.2020.104164.

57.

Podsakoff

MacKenzie

Lee

, et al. Common method biases in behavioral research: a critical review of the literature and recommended remedies. J Appl Psychol 2003; 88(5): 879–903. DOI: 10.1037/0021-9010.88.5.879.

58.

Harman

. Modern factor analysis. Chicago, IL: University of Chicago Press, 1976.

59.

Liang

Saraf

, et al. Assimilation of enterprise systems: the effect of institutional pressures and the mediating role of top management. MIS Q 2007; 31(1): 59–87. DOI: 10.2307/25148781.

60.

Kline

. Principles and practice of structural equation modeling. 14th ed. New York, NY: Guilford Publications Inc, 2023.

61.

Zhang

Guo

, et al. Examining the role of ICT usage in loneliness perception and mental health of the elderly in China. Technol Soc 2021; 67: 101718. DOI: 10.1016/j.techsoc.2021.101718.

62.

Chiu

Cho

Chi

. Consumers’ continuance intention to use fitness and health apps: an integration of the expectation-confirmation model and investment model. Inf Technol People 2021; 34(3): 978–998. DOI: 10.1108/Itp-09-2019-0463.

63.

Straub

. Validating instruments in mis research. MIS Q 1989; 13(2): 147–169. DOI: 10.2307/248922.

64.

Henseler

Hubona

Ray

. Using PLS path modeling in new technology research: updated guidelines. Ind Manag Data Syst 2016; 116(1): 2–20. DOI: 10.1108/Imds-09-2015-0382.

65.

Fornell

Larcker

. Evaluating structural equation models with unobservable variables and measurement error. J Mar Res 1981; 18(1): 39–50. DOI: 10.2307/3151312.

66.

Jiang

Liu

Chi

. Effect of writing style on social support in online health communities: a theoretical linguistic analysis framework. Inf Manag 2022; 59(6): 103683. DOI: 10.1016/j.im.2022.103683.

67.

Cohen

. Statistical power analysis for the behavioral sciences. London, UK: Routledge: 2013.

68.

Zhang

. Impact of patient information behaviours in online health communities on patient compliance and the mediating role of patients’ perceived empathy. Patient Educ Counsel 2021; 104(1): 186–193. DOI: 10.1016/j.pec.2020.07.001.

69.

Gefen

Straub

Boudreau

. Structural equation modeling and regression: guidelines for research practice. CA 2000; 4(1): 7. DOI: 10.17705/1CAIS.00407.

70.

Bentler

. Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives. Struct Equ Model 1999; 6(1): 1–55. DOI: 10.1080/10705519909540118.

71.

Chung

Joung

, et al. Chatbot e-service and customer satisfaction regarding luxury brands. J Bus Res 2020; 117: 587–595. DOI: 10.1016/j.jbusres.2018.10.004.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.26 MB