Health consumers’ social media adoption behaviours in Australia

Abstract

This research identifies the underlying drivers impacting on health consumers’ social media usage and acceptance behaviours using technology acceptance model (TAM) as the theoretical lens. A cross-sectional survey of 265 health consumers was conducted through a mall intercept technique. Participants in the survey were over the age of 18 and had access to a public or private healthcare service provider. The data were analysed using structural equation modelling (SEM). The major findings show that perceived ease of use, privacy threat, information quality, social influence and self-efficacy influence health consumers’ social media adoption behaviours. Perceived usefulness was not found to affect health consumers’ social media adoption behaviours. The moderation analysis showed that influences of privacy threats are non-significant for mature age respondents and non-frequent users of social media. This study’s findings have important implications for designing social media strategies for the healthcare industry. The drivers that positively impact on health consumers’ social media usages can be integrated into meaningful strategies to capture the attention of potential consumers. They need to be educated, informed and engaged as health consumers so that they employ social media effectively to their advantage.

Keywords

social media healthcare online health communication technology adoption model patient engagement patient empowerment

Introduction

In recent times, social media has garnered increasing attention from health consumers, health professionals and service providers.^1–3 Social networking applications such as Facebook, PatientsLikeMe or Hello Health have created new levels of patient participation and present unique and unprecedented opportunities for engaging patients with the healthcare centres, health professionals and researchers.^4,5 The healthcare industry around the world is embracing and recognising the power of these social media platforms to serve their client base. Millions of people use social media each day, to seek health information, do online research related to their medical conditions, share stories, write blogs, watch videos and so much more.⁶ Patients can make more informed decisions regarding their health management process.⁷

In hindsight, exposure to too much information may confuse medical help-seeking consumers who might not be familiar with medical knowledge,² which may limit people’s desire to search for health information. Moreover, persisting complex privacy concerns may discourage the usability of social media applications.⁸ Additionally, senior adults who are not up-to-date with the latest technology may find it difficult to use the various and potentially useful apps and services.⁹ Therefore, despite the rising use of social media in healthcare, many stakeholders such as health professionals and consumers, and administrators are sceptical about social media’s relevance and these applications’ potential in their day-to-day activities.¹⁰ These states of affairs are critically important to better understand the application of social media in the healthcare industry and provide benefits for the community.

Social media in the Australian healthcare industry

Research concerning the usage of social media in Australia revealed mixed results. Although health professionals and health consumers are using social media more frequently in recent years, Australian health professionals are still dubious about its usefulness in a healthcare setting. For example, a survey of 187 doctors revealed that 25.7% of doctors do not use social media at all and only 30.5% had communicated with patients by email.¹¹ Further, 65.8% of the doctors are reluctant to engage in social media due to privacy and legal concerns. A survey of 935 Australian allied health professionals revealed that 71.3% of health professionals did not use social media at all and among social media users, only 9.5% of health professionals claimed to utilise social media for professional purposes. However, the majority of respondents recognise the importance of integrating social media into healthcare care, as it has been found that 55.8% of respondents are willing to undertake social media-related training.¹² Both studies^11,12 are characterised by a lack of theoretical grounding and the absence of consumer perception concerning the usage of social media in healthcare.

The perspectives of Australian health consumers are yet to documented in extant literature. A Google search returned only three studies that applied TAM-based frameworks to understand predictors of adoption behaviours.^13–15 McGowan and colleagues (2012) incorporated six factors (perceived usefulness, ease of use, barriers, personal innovativeness, peer access and motivation) when examining the social media adoption behaviour of physicians. The above factors demonstrated a 57% variance in social media usage frequency.¹³ Hanson et al.¹⁴ combined four factors (attitude, behavioural intention, behavioural control and subjective norm) to predict the social media adoption behaviours of healthcare centre patients. Finally, Jo et al.¹⁵ incorporated seven factors (reliability, barriers, health consciousness, social norm, quality accreditation system, perceived ease of use and perceived usefulness) in their investigation of consumers’ acceptance of health information. It is evident that the underlying factors influencing the usage of social media by Australian health consumers are yet to investigated.

To the best of our knowledge research has not yet explored the social media adoption behaviour of Australian health consumers, which implies there is a paucity of empirical research on this topic. The adoption behaviour includes pre-existing perceptions through to continuing adoption of social media. Meanwhile some prior studies addressed the social media usage patterns of health professionals, consumers and healthcare providers.^16,17 Thus, the primary objective of this study is to identify and examine the factors affecting Australian health consumers’ social media adoption behaviours using a newly developed theoretical framework.The extant technological acceptance frameworks are relatively untested in a healthcare social media context.¹³ The proposed theoretical framework may help in understanding the underlying drivers of consumers’ social media usage within the healthcare domain. In turn this can assist healthcare providers deliver patient-centred communication.

Theoretical framework

The proposed framework is grounded under the premise of the basic Technology Acceptance Model¹⁸ (TAM). It denotes the antecedents of technology usage through beliefs about two factors: the perceived usefulness (PU) and perceived ease of use (PEOU) of a technology.¹⁹ Perceived usefulness (PU) explains the users’ perceptions regarding the outcome of using a new technology or information system; that using a new system/technology would result in enhanced performance. Perceived ease of use (PEOU) explains the degree of ease or effort required to use a new system.¹⁹ Thus, PU and PEOU affect the individual’s current and future intention to use a system or technology. Replication of the original TAM study across different settings, for example, cultures, systems and persons confirms the applicability and the TAM model’s robustness and universal applicability.^20,21 However, TAM was developed based on the assumption that information systems are utilised in organisational settings with a goal to improve their efficiency. TAM fundamentally emphasises the design of system characteristics but does not address some salient dimensions of social media.²² TAM does not consider the role of information quality and privacy threats in influencing an individual’s attitude towards social media, and consequently the usage behaviour. Consequently, past research indicated the need for integrating additional variables to improve the explanatory power of the Technology Acceptance Models.^23,24 In addition to the original constructs information quality,²⁵ privacy threats,^26,27 social influence²⁸ and self-efficacy²⁹ are also shown to affect the behavioural intention of people to access a new system. Moreover, age, gender and usage frequency have significant moderating effects on user adoption.²⁸ Based on the construct measurement and validation procedures of MacKenzie et al.,³⁰ the constructs and items were formally defined from the extant theoretical/empirical literature.^26–31 The 7-point Likert scale survey was used to collect data from health consumers regarding their social media usage and continuance intention in healthcare. The proposed framework – Healthcare social media usage model – is presented in Figure 1 below.

Figure 1.

Healthcare social media usage framework.

Perceived usefulness (PU) and perceived ease of use (PEOU)

Past studies used PU and PEOU to explain the behavioural intentions of health professionals to employ different medical technologies or systems.^31–33 Regarding healthcare social media’s uses, prior studies explained and tested the potential benefits and impacts including patient engagement, social support and enhanced community outreach, health message implications and improved quality of patient care in diverse research settings.^2,7,34 Nonetheless, there is no common agreement among scholars regarding the effect of PU and PEOU on social media usage behaviour. Further, none of the studies specifically investigated the impact of PU and PEOU on social media usage behaviour in healthcare settings. Thus, it would be interesting to investigate the perception of usefulness and ease of use within the new, Australian research context and proposing the following two hypotheses for testing:

H1: Perceived usefulness positively affects social media usage behaviour.

H2: Perceived ease of use of positively affects social media usage behaviour.

Information quality and privacy threats

Prior studies indicated that social media can alleviate many challenges pertaining to the practical, social and emotional aspects of health service delivery.^4,35–37 In terms of barriers to using social media in the healthcare sector, a repeated theme addressed by many researchers was the lack of quality and reliability of information.^38–41 Healthcare worldwide is now continuously confronted with the spread of inaccurate, or false health information regarding various diseases and medical conditions.^42–44 Recent research indicated that poor quality information is a significant risk and no clear mechanism exists for auditing information quality and credibility of posted contents. This suggests that information quality is a significant factor for social media use behaviour, yet this assertion has rarely been tested empirically in the healthcare context. The literature also addressed concerns about privacy and confidentiality, data security and the potential damage caused by privacy violations.^45–47 According to Hader and Brown,⁴⁸ health professionals must be aware of patients’ privacy rights. Unintentional disclosure of patient-identifying information results in breaches of confidentiality and posting work-related information online heightens the risk of privacy and confidentiality breaches. There have been several lawsuits against physicians whose patients accused them of violating their privacy.^35,46,49,50 This implies that patients’ privacy threats preclude social media usage. Based on the above discussion, the following hypotheses are proposed for testing:

H3: Information quality significantly influences social media usage behaviour.

H4: Privacy threats negatively influence social media usage behaviour.

Social influence

Social influence emerged as a significant predictor in several technology acceptance studies.^51,52 Social influence is a learning process; and prior to the product acceptance decision, individuals discern their social groups’ successful experiences.⁵³ In general, there was inadequate empirical data available in the literature determining to what extent social influence affects health consumers’ social media adoption decisions. Most discussions focused on the social support aspects of social media.^35,54 However, the direct effect of social influence on social media usage behaviour needs further exploration. In view of the above the following hypothesis is posited:

H5: Social influence significantly influences social media usage behaviour.

Self-efficacy

This research also found that confidence in using social media for healthcare purposes is a significant component of adoption behaviours, which relates to the concept of self-efficacy. The notion of self-efficacy is based on the belief that an individual can efficaciously perform the behaviour required to produce the desired outcomes.⁵⁵ Hocevar et al. introduced the concept of social media self-efficacy, suggesting that that social media self-efficacy stems from an individual’s: (i) perceived social media skills, (ii) confidence in the ability to find information online, (iii) level of social media content production and (iv) level of social media content consumption. Self-efficacious individuals are more likely to engage in social media than those who have limited social media self-efficacy.⁵⁶ Existing literature indicates that an individual’s reluctance to use social media is rooted in the lack of required skills and experience.^35,57–59 Self-efficacious individuals behave more socially and are more likely to engage in online information exchange and be influenced by other people’s input.⁵⁶ Most previous studies primarily addressed the internet self-efficacy of users.^60–62 However, health consumers’ social media self-efficacy is relatively scarce in the contemporary literature. This study is likely to be one the first studies to address this issue. Here the following hypothesis is put forward for testing:

H6: Social media self-efficacy is positively associated with social media usage behaviour.

Social media usage behaviour and continuance intention

The theoretical underpinning of the usage behaviour construct is grounded on the ‘actual system usage’ construct introduced by Burton-Jones and Straub.⁶³ It refers to the degree a user applies one or more system features to complete a particular task.⁶³ Usage behaviour is deemed to be a complex construct because scholars do not agree on its impact on continuance intention and the measurements scales used for this variable. Two common approaches to measuring actual system usage are objective and subjective measures.⁶⁴ Objective measures are based on the system log that includes various aspects of actual use, for example, log-in history or time spent after each login. On the other hand, subjective measures are based on self-reported values that focus on the subjective data such as frequency or intensity of use.⁶⁵ This study adopted the later (subjective) approach, as it is beyond its scope to collect social media usage logs from health consumers.

The theoretical underpinning of continuance intention is an integral part of the Theory of Planned Behaviour, which as mentioned previously is an extension of the Theory of Reasoned Action⁶⁶ and Expectation Confirmation Model (ECM).⁶⁷ According to Bhattacherjee⁶⁷ a system’s initial acceptance is a significant pointer to realising its success. Therefore, based on the work of Bhattacherjee,⁶⁷ this research argues that the long-term viability of social media in healthcare and its eventual success depend on its continued use rather than infrequent usage. Thus, continuance behaviour is a highly relevant construct derived from an empirical perspective because service usage continues well beyond the initial adoption. Based on the above literature the research proposes to test the following hypothesis:

H7: Social media usage behaviour positively influences continuance intention.

Table 1 below summarises the definitions and sources of the constructs used in this study to develop the conceptual framework.

Table 1.

Root of constructs, definition and source.

Construct/sources	Definition	Source
Perceived usefulness (PU)	The users’ perceptions which denotes that using a new system/technology would result in enhanced performance.	Davis et al.¹⁹
Perceived ease of use (PEOU)	Perceived ease of use (PEOU) explains the degree of ease or effort required to use a new system.	Davis et al.¹⁹
Information quality (IQ)	Refers to completeness, ease of understanding, personalisation, relevance and security of web-based content.	Delone and McLean⁶⁸
Privacy threats	The threat that violates the right to determine what personal information about an individual is known to others.	Bhattacherjee and Hikmet⁶⁹
Social influence	An individual’s perception that most people who are important to him advocate he should or should not engage in a particular behaviour.	Fishbein and Ajzen⁷⁰
Self-efficacy	An individual’s beliefs about his/her abilities to perform at a level that exerts influence over events that influence their lives.	Bandura⁵⁵
Usage behaviour	Refers to the degree a user applies one or more system features to complete a particular task.	Burton-Jones and Straub⁶³
Continuance intention	Refers to the degree a user accepts and intend to continue using a certain system features to complete a particular task.	Venkatesh⁷¹

Moderators

The proposed model also considers the influence of three moderators that are expected to influence the six key antecedents of usage behaviour determinants: gender, age and social media usage frequency. According to unified theory of acceptance and use of technology (UTAUT), gender moderates all relationships into intention to use.⁷² The theoretical basis pertaining to the moderation by gender is rooted on gender stereotypes and associated differences between men and women in terms of their beliefs and attitudes regarding technology adoption,^73–75 including health technology adoption. Recent research indicates that gender is strongly associated with the adoption and use of various health technologies.^74,76,77 However, only a few studies have investigated such a relationship from the social media technology perspective, particularly in the context of health consumers. A recent study concerning generic social media adoption highlighted there is no significant difference between males and females on the adoption of social media in Australia.⁷⁸ We expect that such gender stereotyping will give way to widespread social media usage in Australia, where 90% of the total population uses social media sites. Thus, we propose that there will be no gender differences among health consumers in the factors that are proposed in the conceptual model:

H8: The gender effects in the healthcare social media usage framework will not be significant among health consumers.

In a technology adoption setting, the mechanisms concerning the effects of age focus on the notion that older people are relatively reluctant to adopt to new ways of undertaking certain tasks.⁷⁹ We propose that, these propositions are still true among health consumers. In contrast to mature individuals, due to their greater interaction with technologies growing up, younger individuals are likely to feel more comfortable and able to embrace social media. Recent studies indicate that different age groups have specific moderating effects on mobile health services adoption.^80,81 Also, the practical utility of technology adoption stays at a low level among the older generation.^82,83 However, it is not clear whether age has a specific moderating impact on healthcare social media acceptance. This research considered the moderating role of age to examine the differences between the subgroups concerning social media usage behaviour (i.e. young age and mature age). Thus, we propose the following hypothesis:

H9: The age effects in the healthcare social media usage framework will be significant among health consumers.

The principle notion behind usage frequency as a moderator is that frequent use of certain systems reduce anxiety and any problems associated with the new system will dissipate.⁷² Therefore, we argue that usage frequency will reduce the barriers associated with social media adoption and usage and frequent users are more likely to adopt social media. Thus, we propose that usage frequency will have a significant moderating effect in the constructs that are included in the theoretical framework.

H10: The usage frequency effects in the healthcare social media usage framework will be significant among health consumers.

Methods

In order to answer the study’s hypotheses a dataset of 259 Australian social media users was subjected to structured equation modelling (SEM). Potential respondents were randomly chosen through a mall intercept survey in Canberra. Most of the survey was conducted outside a pharmacy or clinic in a shopping mall in Canberra over a period of 2 weeks. The mall intercept survey is a method frequently employed for consumer research and considered ideal for gaining access to a representative sample.⁸⁴ However, the mall intercept method may also lead to selection bias which may result from obvious challenges in obtaining a representative sample of the target population.⁸⁵

Regarding the selection criteria for the respondents, participants had to be over the age of 18 with access to a public or private healthcare service provider. In order to ensure a representative sample, data was collected at different times and dates during the week. For example, days of the week were grouped into morning and afternoon sessions. In 1 week, the survey was carried on Monday, Tuesday and Wednesday and in the subsequent week it was on Wednesday, Thursday and Friday. Moreover, the opening day session of the first week was chosen on a random start basis and altered thereafter. A sample selection rule was initiated by which the number of customers was counted as they entered the clinic or pharmacy and every nth person was approached to participate in survey. The number of people to be skipped was set according to a predetermined measure of shopping traffic in the designated locations. Following the recommendations of Nunnally and Bernstein,⁸⁶ t-test was conducted to verify the responses received during the morning and afternoon sessions and two different weeks. The verification process revealed no significant difference between the two groups of respondents. Altogether, the survey questionnaire was administered to 835 adult respondents. In total 265 complete responses were received with a 32% response rate. No incentive was provided for survey participation, and it took between 15 and 20 min to complete.

Univariate outliers were examined using the descriptive analysis with standardised residuals saved and then screened for values above 3.29.⁸⁷ Six multivariate outliers were detected through the residual analysis with the Mahalanobis distance and case-wise diagnostics. After these six cases were deleted, the descriptive analysis revealed that data skewness and kurtosis were within the acceptable value of ±3.29; hence, this does not require a remedy through data transformation.⁸⁸ These identified six cases were verified one by one to understand and clarify why these were outlier cases and justify whether their exclusion was necessary or limits generalisability. The study further applied the assumption of multivariate normality (MVN) and verified sample sensitivity using alternative estimation procedures.^89,90 First, based on Byrne’s⁹¹ recommendations, observations farthest from the centroid Mahalanobis distance under confirmatory factor analysis (CFA) revealed Mahalanobis d-squared value for those six outlier cases were highly significant and demonstrated significant impact on the initial CFA model fit statistics. Second, the outlier values are re-included, and subsequent rerun of the CFA model reaffirms that those outlier cases are sensitive to fit indices. The detailed results of centroid Mahalanobis distance and CFA fit indices (included and excluded six cases) are shown in Table 2. Such verification processes is a common feature of MVN distribution in SEM analyses.⁹² Nevertheless, the exclusion or inclusion of these cases have not affected the results of the path analysis or hypothesis testing. Thus, the relationship between the variables will remain unchanged. As only six cases were identified which is not a significant number compared to ratio of the variables,⁹³ exclusion of those six responses may not be an issue in limiting the generalisability of the current findings. On exclusion of these cases, finally a - total of 259 responses satisfied the requirement of case-to-variable ratio (eight) for CFA and SEM analyses.^87,93

Table 2.

Observations from the centroid Mahalanobis distance.

Case number	Mahalanobis d-squared	p-Value
108	144.594	0.000
26	106.224	0.000
113	100.233	0.000
84	98.087	0.000
28	87.031	0.000
68	79.902	0.000

Model fit (including six outlier cases): CMIN = 553.994; χ²/df = 2.473, IFI = 0.925, TLI = 0.907, CFI = 0.924, RMSEA = 0.075.

Model fit (Excluding six outlier cases): CMIN = 489.827; χ²/df = 1.884; IFI = 0.954; TLI = 0.942; CFI = 0.954; and RMSEA = 0.06.

Improvement of model fit (differences): χ²/df (2.473–1.884) = 0.559; IFI (0.954–0.925) = 0.029; TLI (0.942–0.907) = 0.035; CFI (0.954–0.924) = 0.03 and RMSEA (0.075–0.06) = 0.015.

Survey instrument

For this study, this research considered social media from a holistic perspective rather than being confined to one specific platform or application. Social media use is complex and it is a source of entertainment, connection or information.⁹⁴ Past studies have examined the usage of social media from the perspective of a single application such as health professionals’ use of Facebook, Twitter or the impact of YouTube-based user-generated content on health communities. However, in contemporary social setting, individuals are likely to use more than one application and most social media applications combine different applications on one platform. Therefore, it is not feasible to investigate a single application separately, considering the relationships with other social media applications.⁹⁵ Consequently, this research argues that confining the scope of the study to a single application (for instance, Facebook, Twitter or YouTube) would not accurately portray the usage behaviour of health professionals. To accomplish the goal of this research, the study investigates health consumers’ social media adoption from a holistic perspective.

The survey instrument consisted of three sections: (i) questions related to the demographic profile; (ii) current use of social media, where the respondents were asked to specify the type of social media platforms they use; and (iii) the respondents’ perceptions regarding factors that affect their social media adoption behaviour in healthcare context. These factors are perceived usefulness, perceived ease of use, information quality, privacy threats, social influence, self-efficacy and usage behaviour. All constructs are measured on a seven-point scale, where one represented ‘strongly disagree’ and seven represented ‘strongly agree’. The following definition was provided at the beginning to ensure the definition of social media was well understood. Social media is defined as internet-based applications which allow for the creation and exchange of user-generated content and includes services such as social networking platforms, professional online communities, wikis, blogs and microblogging.⁹⁶

The source of the constructs measures is documented in Table 3.

Table 3.

Construct measures.

Construct	Source	Indicators	Std. Est
Perceived usefulness (α = 0.92, AVE = 0.89)	New measures based on Moorhead et al.,³² Hu et al.⁹⁷ and Chismar and Wiley-Patton⁹⁸	Enable better health communication	0.85
		Enhance health knowledge	0.94
		Increase responsibility	0.89
		Improve quality of healthcare	0.89
		Increase knowledge sharing	0.88
Perceived ease of use (α = 0.86, AVE = 0.87)	Measures based on Davis¹⁸	Learning to find health information is easy	0.85
		Easy to find relevant health information	0.94
		Easy to communicate with health providers	0.89
		Social media applications are easy to use	0.89
Information quality (α = 0.83, AVE = 0.77)	Adapted measures from DeLone and McLean²⁵	Health information is presented well	0.73
		Information provided by social media sites is reliable	0.84
		Easy to understand	0.78
		Constructive	0.75
Privacy threats (α = 0.86, AVE = 0.79)	New measures based on Ajzen,⁶⁹ Cheng et al.⁹⁹ and Duyck et al.¹⁰⁰	Hard to maintain confidentiality	0.77
		Might compromise privacy	0.82
		There are legal implications	0.85
		Concerned about the privacy settings	0.84
Social influence (α = 0.84, AVE = 0.84)	Adapted measures from Venkatesh et al.²⁸	Important people in my life influence me to use social media	0.74
		People influencing my behaviour encourage me to use social media to seek health-related information	0.92
		Sharing of knowledge benefits the health community	0.82
		Influenced by peers	0.87
		My peers use social media for seeking health information	0.79
Self-efficacy (α = 0.88, AVE = 0.83)	Measures based on Compeau and Higgins¹⁰¹	Confident about using social media	0.78
		Possesses required skills to use social media	0.89
		I have no difficulty in using social media	0.82
		I am interested in using social media	0.87
Usage behaviour (α = 0.88, AVE = 0.84)	Measure adapted from Davis¹⁸	Use social media to share medical knowledge with other patients	0.87
		Share medical information with friends and family	0.85
		Seek specific information about health-related issues	0.81
		Use social media to learn about health	0.84
		Use social media to promote health information	0.82
Continuance intention (α = 0.88, AVE = 0.89)	Measure adapted from Bhattacherjee⁶⁷	I will continue using social media for health-related matters	0.93
		I will increase my use of social media in the future	0.90
		I intend to increase my contribution to health communities in the future	0.88
		Increased use of social media will improve healthcare service	0.87

CFA is run with all variables included in the conceptual model to assess model fit, and in the second stage, structural model testing is carried out. Factor loading of scale items was examined. Generally, factor loadings below 0.4 are considered low and items with a low loading should be discarded. In this study the recommended universal cut-off factor loading of 0.50 was used to ensure that all variables had practical significance.⁸⁸ All retained items loaded appropriately as expected with factor loadings ranging between 0.74 and 0.94 and all alpha reliability values were above the minimum acceptable benchmark of 0.70,¹⁰² indicating the acceptable internal consistency of the scale items in a construct. The correlation matrix confirmed that no correlation between any two items was above 0.8. It can therefore be stated that bivariate high collinearity did not exist in this dataset. Also, the average variance extracted (AVE) for each construct exceeds the recommended limit of 0.50.¹⁰³ As shown in Table 4, all square roots of the AVEs (diagonal cells) are higher than the correlations between constructs, and that confirms the constructs’ discriminant validity.

Table 4.

Bivariate correlations.

	1	2	3	4	5	6	7	8
Perceived usefulness	0.89
Perceived ease of use	0.313**	0.87
Social influence	0.641**	0.359**	0.84
Privacy threats	−0.154	−0.258**	−0.134	0.79
Information quality	0.438**	0.400**	0.505**	−0.086	0.77
Self-efficacy	0.370**	0.452**	0.349**	−0.094	0.416**	0.83
Usage behaviour	0.557**	0.164	0.739**	−0.213*	0.437**	0.327**	0.84
Continuance intention	0.668**	0.286**	0.719**	−0.162	0.577**	0.557**	0.638**	0.89

Correlations are significant at *p<0.05 and **p<0.01 level. The bold face values indicate the square root of AVE.

Analysis of the CFA resulted in the following model fit indices: CMIN = 489.827 with df = 260 and CMIN/df ratio = 1.884; IFI = 0.954; TLI = 0.942; CFI = 0.954; and RMSEA = 0.06. The fit indices are within the acceptable assessment criteria of the model fit.⁹¹ Thus, the statistics provide adequate evidence supporting a good model fit of the first order CFA.

Results

Demographic profile

In this study, to identify a moderation level for age, the dataset was divided to form two sets, each representing individuals who belong to a particular age group. An analysis of age distribution (See Table 5) demonstrates that two major age groups emerged from the survey: Young adults (18–39) and Mature adults (40–59). Only a minority of the respondents (10 respondents) reported to be outside of these age brackets. Representatives of such age brackets may be fundamentally different in terms of various characteristics, perceptions and behaviours. Therefore, selecting a cut-off point of 39 years old during the survey creates two distinct categories of respondents who may potentially exhibit significant moderating affect. However, as it was not possible to extract substantial data from respondents aged 60+, it is apparent that age distribution of respondents is skewed towards 18–59 in this study which can be attributed to the known limitation of mall intercept method in obtaining a representative sample of the target population.¹⁰⁴

Table 5.

Demographic profile.

Categories	Criteria	Frequency	Percentage
Age	18–39	136	52.5
	40–59	113	43.6
	60+	10	3.9
Gender	Male	101	39
Gender	Female	158	61
Social media usage frequency	Non-frequent user	92	35.5
Social media usage frequency	Frequent user	167	64.5

Gender was recorded as male or female, and this naturally forms two moderator levels. Gender distribution of respondents revealed 39% were male and the rest 61% were female. To specify a moderation level for frequency, once-a-week users were considered to be frequent users, whereas once-a-month users were deemed to be infrequent users. In terms of the frequency of social media usage pattern, 64.5% respondents use at least one social media platform frequently (at least once a week), whereas 35.5% are non-frequent users of social media (once in a month).

Social media usage pattern

Among the consumer group, 92.2% used Facebook followed by 88.4% respondents who used YouTube, while 74.4% used LinkedIn and 44.8% used Instagram. Furthermore, 29.3 % used Twitter and 15.8% used WIKI, while 21.50% of participants accessed ‘other’ social media applications not listed on the questionnaire. 29.7% of respondents reported using health-related social networking sites, and 22% are users of blogging platforms. Table 6 below presents the participants’ social media usage pattern.

Table 6.

Consumers’ social media usage patterns.

Type	Frequency	Percentage
Facebook	239	92.2
Twitter	76	29.3
Google Plus	32	12.3
Instagram	116	44.8
YouTube	229	88.4
Blogs	61	23.5
LinkedIn	198	74.4
WIKI	41	15.8
Health-related sites	77	29.7
Others	57	22

SEM results

The following indices served to measure the overall goodness-of-fit of the model: CMIN = 483.211 with df = 229 and CMIN/df ratio = 2.110; IFI = 0.946; TLI = 0.935; CFI = 0.946; and RMSEA = 0.06. Results indicated that all the factors had significant loadings greater than 0.50 (p < 0.001) on their respective constructs. These values were as expected and very close to the minimum cut-off points, indicating an acceptable model fit. Table 7 provides an overview of the hypotheses’ results that were performed in SEM.

Table 7.

SEM results.

Path			Estimate	S.E.	C.R.	p	Result
Social media usefulness	→	Usage behaviour	0.067	0.063	1.067	0.286	Not supported
Perceived ease of use	→	Usage behaviour	0.163	0.055	2.944	0.003	Supported
Privacy threat	→	Usage behaviour	−0.144	0.053	−2.714	0.007	Supported
Information quality	→	Usage behaviour	0.259	0.084	3.086	0.002	Supported
Social influence	→	Usage behaviour	0.284	0.063	4.522	***	Supported
Self-efficacy	→	Usage behaviour	0.300	0.051	5.853	***	Supported
Usage behaviour	→	Continuance intention	1.124	0.125	8.977	***	Supported

Significant at ***p<0.001.

Discussion

Discussion of the results of this research centres around the hypotheses presented for testing. The results demonstrate that all hypothesised relationships are supported except hypothesis 1. It was found that perceived usefulness of social media is not a predictor of social media usage behaviour (β = 0.067, p > 0.05). However, the relationship between perceived ease of use and usage behaviour was statistically significant (β = 0.163, p < 0.01). This outcome contradicts prior research in a similar vein,^105–107 where perceived usefulness of a new technology or system was shown to be a significant predictor of intention to use information technology.

Privacy threats negatively impact on social media usage behaviours of health consumers (β = -0.144, p < 0.01). This finding is consistent with prior studies indicating that privacy and confidentiality are a common area of concern for both health professionals and health consumers.^37,108–110 Conversely, information quality positively influences usage behaviour (β=.259, p < 0.01). The findings are consistent with the extant research. Recent research during the COVID-19 pandemic also indicates that the overwhelming volumes of information generated by social media platforms which are proliferated by a diverse group of people make it difficult for the public to find valid and reliable health information.^{109,111–113} This in turn may compromise public safety at a time when awareness and appropriate preventive actions are paramount. Although social media platforms offer unique opportunities for healthcare providers to connect with patients or potential patients, attributes related to privacy and information quality have not been operationalised within a theoretical framework in prior studies concerning social media adoption.^13,114 Thus, this study complements the existing literature by examining the impacts of privacy threats and information quality on health consumers’ social media usage behaviour.

The Beta coefficient of social influence is positive and significant (β = 0.284, p < 0.001), suggesting that the influence of peers has a significant effect on social media usage behaviour. These findings are consistent with several previous analyses^28,115,116 which demonstrates that social influence plays a significant role in users’ technology adoption. Thus, health consumers’ might be more or less likely to accept and use social media, based on the extent to which people close to them encourage them to use it. Therefore, social influence can be perceived as a potential inhibitor of social media usage for health-related reasons.

The result indicates that self-efficacy is a significant predictor (β = 0.300, < 0.001), of usage behaviour. This finding corroborates many healthcare-based studies.^36,117 Like these studies, we accept that health consumers with a higher level of self-efficacy are more likely to use and adopt social media. The association between usage behaviour and continuance intention was positive and significant (p < 0.001) and reinforces previous findings.^{67,75,118–120}

Moderating effect

As recommended by Steenkamp and Baumgartner,¹²¹ we have conducted measurement invariance tests to verify the statistical concern and validated the proposed theoretical constructs and their suitability for different (cross) groups equivalence such as Gender (Male-Female), Age (Young adults – Mature adults), and usage frequency (Frequent user – Non-frequent user). Furthermore, structural invariance including χ² difference tests have been conducted to verify the moderating effect of these criteria variables. To analyse measurement invariance in across groups or categories, it is essential to assess configural and metric invariances.¹²² To analyse the moderating effect of groups or categories, it is recommended to assess structural invariance. Configural invariance test is a vital statistical method intimating that the same construct is being assessed across some specified datasets or groups.

As shown in Table 8, Model 1, the configural invariance model for several data groups in terms of gender, age and usage frequency represents acceptable level of fit which suggests that these groups are equivalent. Moreover, metric invariance evaluates whether the factor loadings are equal for each of the scale items across groups data. This implies that factor loadings are equal across groups (gender, age and usage frequency), however, the intercepts can differ between groups. If items are metrically invariant and if the factor variances are across-groups invariant, the items are equally reliable across sample (data) groups. Therefore, we constrained all the factor loadings to be equal between the gender (male and female), usage frequency (frequent and non-frequent user) and age (young adults – mature adults) samples. However, we allowed error variances, item intercepts and factor covariance to vary freely across the (groups) samples. We further forced the variances of the constructs to be invariant across the (groups) samples. In Table 8, Model 2 shows a very good fit to the data and non-significant χ² difference provides further evidence and confirmation of metric invariance in the measures. These verifications suggest the measures are consistent and similar across categories to assess the theoretical model.

Table 8.

Invariance measures – configural and metric.

Gender	χ² value	df	χ² difference	RMSEA	IFI	TLI	CFI	χ2/DF
Model 1: Configural model	744.47	448	–	0.051	0.939	0.923	0.938	1.662
Model 2: Metric invariance	762.74	472	CMIN = 18.27	0.049	0.940	0.928	0.939	1.616
Model 2: Metric invariance	762.74	472	Df = 24, p < 0.790	0.049	0.940	0.928	0.939	1.616
Usage frequency- Frequent and non-frequent User
Model 1: Configural model	717.32	448	–	0.048	0.944	0.930	0.943	1.601
Model 2: Metric invariance	741.37	464	CMIN = 24.05	0.048	0.942	0.930	0.941	1.598
Model 2: Metric invariance	741.37	464	Df = 16, p < 0.09	0.048	0.942	0.930	0.941	1.598
Age- Young adults – mature age
Model 1: Configural model	709.51	448	–	0.048	0.946	0.932	0.945	1.584
Model 2: Metric invariance	733.68	464	CMIN = 24.17	0.048	0.944	0.932	0.943	1.581
Model 2: Metric invariance	733.68	464	Df = 16, p < 0.09	0.048	0.944	0.932	0.943	1.581

Additionally, structural invariance examines whether the causal invariances are identical across groups (gender, age and frequent and non-frequent user samples). Following the same procedure, we have conducted a series of χ² difference tests for the constrained and unconstrained structural model. Initially, we estimated the unrestricted models in which the path coefficients were able to vary freely. We also estimated a fully restricted models where all path coefficients were constrained across consumer (user) group samples. Table 9, Model 3 demonstrates a reasonable fit and provides evidence that there is no significant difference between gender group samples. However, the other two groups – usage frequency and age of consumers – highlighted significant differences. It is evident that the path coefficients are invariant across gender, but they are variant across frequent and non-frequent user (usage frequency) and young adults – mature adults (age) consumers. The difference in these groups and categories do moderate the perceived uses behaviour of consumers. Furthermore, we have conducted path-by-path analyses for all hypothesised paths to reveal the hypothesised path level differences. Although the applied moderating effects testing approach has been highly recommended and cited in the literature,^90,91 this technique might not be immune to criticism because it is facilitated by theorised dichotomous variables which may also be losing explanatory power of the model. Nonetheless, as suggested in the literature, small explanatory power difference between constrained and unconstrained models are not inevitably important in analysing the moderating effect of categorical variables.¹²³ The current study shows that explanatory power differences for the moderating effect of identified categorical variables are less than 0.10 and do not recommend that this value: firstly, could be meaningful; and secondly, should be used in explaining the power of categorical moderation in this research. Moreover, estimating latent variable interactions could be become more intuitive and insightful for any future research.

Table 9.

Moderating effect: structural invariance.

Gender moderation	χ² value	df	χ² difference	RMSEA	IFI	TLI	CFI	χ2/DF
Model 3: Structural invariance (Constrained – unconstrained)	815.08–800.32	476–460	CMIN = 14.76	0.054	0.930	0.914	0.928	1.74
	815.08–800.32	476–460	Df = 16 p = 0.542	0.054	0.930	0.914	0.928	1.74
Usage frequency moderation: frequent and non-frequent
Model 3: Structural invariance (Constrained – unconstrained)	800.57–769.61	476–460	CMIN = 30.96	0.051	0.936	0.921	0.934	1.67
	800.57–769.61	476–460	Df = 16 p = 0.014	0.051	0.936	0.921	0.934	1.67
Age Moderation: young-old user
Model 3: Structural invariance (Constrained – unconstrained)	794.79–760.79	476–460	CMIN = 34	0.051	0.935	0.921	0.934	1.68
	794.79–760.79	476–460	Df = 16 p = 0.005	0.051	0.935	0.921	0.934	1.68

As it can be seen from Table 8, there is no significant difference between the male and female groups. However, age moderates the relationship between usage behaviour and its antecedents. Interestingly, the impact of perceived threat on usage behaviour (p > 0.05) and ease of use on usage behaviour (p > 0.05) is non-significant in the case of mature age health consumers. This is an interesting finding and is consistent with a recent study conducted on mobile healthcare (mHealth),⁸⁴ which shows that privacy threats negatively influence the behavioural intentions of young mobile health (mHealth) consumers, but do not affect the elderly. Other studies have also demonstrated that technologically savvy populations, such as teenagers and undergraduates, tend to be concerned about disclosing private information on social networking sites.^124,125 It can therefore be deduced that Australian mature adults seem to be less concerned with privacy threats and ease of use in comparison to their younger counterparts.

In terms of usage frequency, the path-by-path analysis reveals a non-significant impact of information quality (p > 0.05) and privacy threats (p > 0.05) on usage behaviour in the case of non-frequent social media users. However, they are significant for frequent users (p < 0.05). One possible explanation is that non-frequent users of social media do not spend as much time on the platforms as much as frequent users. Non-frequent users are likely to be less concerned about information quality and privacy-related matters.

Implications, contributions and conclusions

The research adds to the existing literature in several ways and explains some insightful contributions. These contributions are explained in more detail below.

Contributions to theory refinement

This research lays theoretical foundations for deepening our understanding of the possible predictors of social media adoption behaviour of health consumers in an Australian setting. The research framework used in our research integrated several critical factors that are rationalised using evidence from the extant literature. These drivers of social media usage behaviour of health consumers are empirically tested and validated with user data collected exclusively in Australia. To the best of our knowledge, this is one of the first quantitative research study in the domain of healthcare social media that takes into account the perceptions of general health consumers of Australia. The construct measurement scales derived from existing literature are deemed reliable given the solid composite reliabilities of the constructs. It confirms that these instruments can be safely applied in future research within the social media healthcare domain.

The findings suggest that consumers’ social media usage behaviour is a function of ease of use, social influence, information quality, privacy threats and self-efficacy. These factors have substantial predictive power in explaining consumers’ social media usage behaviour. Surprisingly, usefulness of social media has not emerged as a determinant of usage behaviour. This finding apparently contradicts the orthodox notion that utilitarian orientations of perceived usefulness constitute a central tenet of technology use.^126,127 Research indicates perceived usefulness is a multi-dimensional construct and different dimensions which have different implications for specific use patterns.¹²⁶ This particular outcome adds to our knowledge in that degree of usefulness of social media may vary according to what this platform is being utilised for. Therefore, healthcare providers intending to promote social media as a health communication tool should find out methods of enhancing the level of its usage through promotional campaign. What can be highlighted are the benefits of using this tool especially for consumers of health services in Australia and beyond.

To sum up, the main theoretical contributions of this research lie in the development and validation of a novel integrative research framework with the potential for predicting social media adoption mechanisms by healthcare consumers in an Australian setting. This research documents factors that affect consumers’ intention to use and adopt social media in healthcare by converging various perspectives including ease of use, privacy threats, information quality, social influence and self-efficacy. It is expected to provide a sound theoretical underpinning for future research on social media usage behaviour in healthcare.

Contributions to filling the knowledge gap on the role of moderating variables

This research to some extent fills in the knowledge gap in the literature on the moderating effects of health service users’ demographics such as their level of physical progression (age), gender and their usage frequency of social media. The study findings may provide an opportunity for healthcare providers and policy-makers to understand and take necessary actions with regard to the current status of users’ perceptions of the potential benefits and challenges of using social media in healthcare. To sum up, adequate knowledge of the moderating effects of the issues discussed above may provide guidelines for the healthcare sector. Appropriate strategies that address these issues and improve the current status of social media usage amongst Australian consumers should be designed.

Contributions to generalisation of the drivers of social media usage behaviour

Generalisation of the drivers of social media usage in the healthcare sector in an Australian context is a complex issue in that social media is a dynamic creature that makes it difficult to generalise over time and space. However, we can argue with some reservations that our research has unearthed some critical evidence regarding the predictors of social media usage behaviour of Australian consumers, paving the way for a fuller explanation of the findings down the track. Furthermore, although any generalisation has a long way to go, this research has, at least, structurally modelled the impacts of the possible drivers. It has taken into account the complexities associated with the moderating variables complicating the straightforward nexus between consumer perception and practice of social media in Australia. In this sense, our research advances the previous research on the topic and contributes to this rapidly growing field of study.

Limitations of the study and avenues for further research

Every research study has its limitations and the current one is no exception. The study had limitations in terms of the generalisability of the findings due to employing the mall intercept method and small sample size. The majority of research participants are within the 18–59 age bracket. Thus, this research does not adequately capture the perceptions of health consumers who are 60 years or older. Future research should also include more participants covering the 60-plus age brackets to document the perceptions/opinions of elderly people regarding social media usage in healthcare. Towards overcoming this demographic issue, future research can employ purposive sampling frame to accommodate more elderly respondents to document the perceptions/opinions of elderly people regarding social media usage in healthcare. Furthermore, this research only investigates generic social media users’ perceptions regarding in the healthcare domain. Future research should investigate the perceptions of the sample who use social media specifically for health-related reasons.

This research has been undertaken in an Australian context which tends to limit the generalisation of the findings internationally. Future research can overcome this limitation by using large samples drawn from other Western countries with similar social and economic characteristics such as the USA, Canada, New Zealand and the UK. At present, limited information is available regarding the quality of health information shared in health-related sites. It is also not known as to how much of it is evidence-based. Additionally, not much is known about patients’ reactions to health information. It was also unclear to what extent consumers use this information for better medical outcomes.

Theoretically, the research investigates the concept of social media holistically instead of defining specific social media tools such as Facebook, Twitter or health blogs. Although this was very useful in obtaining a holistic understanding of the critical factors affecting social media adoption, such an approach limits in-depth analysis of specific, elements and features in particular. Future research endeavours could investigate specific social media tools such as Facebook, Twitter, health-related blogs and so on, and especially their implications for medical services. Future research endeavours may consider focusing on a more specific platform of social media (for instance, Facebook or Twitter) and its impact on specific patient groups. Finally, future research can emphasise the relationship between perceived usefulness and social media usage behaviour over space and time to see if any change in the nexus between these two phenomena has occurred.

Footnotes

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) received no financial support for the research, authorship and/or publication of this article.

ORCID iD

Irfanuzzaman Khan

References

Fernández-Luque

Bau

Health and social media: perfect storm of information. Healthc Inform Res 2015; 21(2): 67–73.

Gamache-OLeary

Grant

. Social media in healthcare. In: Proceedings of the 50th Hawaii international conference on system sciences University of Hawai’i at Manoa. 2017.

Merchant

RM.

Evaluating the potential role of social media in preventive health care. JAMA 2020; 323(5): 411–412.

Chretien

Kind

Social media and clinical care: ethical, professional, and social implications. Circulation 2013; 127(13): 1413–1421.

De Martino

D’Apolito

McLawhorn

, et al. Social media for patients: benefits and drawbacks. Curr Rev Musculoskelet Med 2017; 10(1): 141–145.

Ray

In sickness and in health: clinical research and social media. Perspect Clin Res 2017; 8(2): 73–78.

Saraogi

NLPA

. How social media is transforming pharma and healthcare. Appl Clin Trials 2018; 27(2).

Chakraborty

Vishik

Rao

HR.

Privacy preserving actions of older adults on social media: exploring the behavior of opting out of information sharing. Decis Support Syst 2013; 55(4): 948–956.

Cho

User perceptions of social media: a comparative study of perceived characteristics and user profiles by social media. Online J Commun Media Technol 2013; 3(4): 149.

10.

Grajales

III Sheps

, et al. Social media: a review and tutorial of applications in medicine and health care. J Med Internet Res 2014; 16(2): e13.

11.

Brown

Ryan

Harris

How doctors view and use social media: a national survey. J Med Internet Res 2014; 16(12): e267.

12.

Usher

Woods

Casella

, et al. Australian health professions student use of social media. Collegian 2014; 21(2): 95–101.

13.

McGowan

Wasko

Vartabedian

, et al. Understanding the factors that influence the adoption and meaningful use of social media by physicians to share medical information. J Med Internet Res 2012; 14(5): e117.

14.

Hanson

West

Thackeray

, et al. Understanding and predicting social media use among community health center patients: a cross-sectional survey. J Med Internet Res 2014; 16(11): e270.

15.

Song

T-M

Kim

BG.

Analysis of the factors affecting consumer acceptance of accredited online health information. J Korean Med Sci 2017; 32(11): 1757–1763.

16.

Surani

Hirani

Elias

, et al. Social media usage among health care providers. BMC Res Notes 2017; 10(1): 654.

17.

Panahi

Watson

Partridge

Social media and physicians: exploring the benefits and challenges. Health Informatics J 2016; 22(2): 99–112.

18.

Davis

FD.

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

19.

Davis

Bagozzi

Warshaw

PR.

User acceptance of computer technology: a comparison of two theoretical models. Manage Sci 1989; 35(8): 982–1003.

20.

Yousafzai

Foxall

Pallister

JG.

Technology acceptance: a meta-analysis of the TAM: Part 1. J Model Manage 2007; 2(3): 251–280.

21.

King

A meta-analysis of the technology acceptance model. Inf Manage 2006; 43(6): 740–755.

22.

Rauniar

Rawski

Yang

, et al. Technology acceptance model (TAM) and social media usage: an empirical study on Facebook. J Enterp Inf Manage 2014; 27(1): 6–30.

23.

J-H

Wang

S-C.

What drives mobile commerce? An empirical evaluation of the revised technology acceptance model. Inf Manage 2005; 42(5): 719–729.

24.

Venkatesh

Thong

Consumer acceptance and use of information technology: extending the unified theory of acceptance and use of technology. MIS Q 2012; 36: 157–178.

25.

DeLone

McLean

ER.

Information systems success: the quest for the dependent variable. Inf Syst Res 1992; 3(1): 60–95.

26.

Al-Muhtadi

Shahzad

Saleem

, et al. Cybersecurity and privacy issues for socially integrated mobile healthcare applications operating in a multi-cloud environment. Health Informatics J 2019; 25(2): 315–329.

27.

Bansal

Gefen

The impact of personal dispositions on information sensitivity, privacy concern and trust in disclosing health information online. Decis Support Syst 2010; 49(2): 138–150.

28.

Venkatesh

Morris

Davis

, et al. User acceptance of information technology: toward a unified view. MIS Q 2003; 27: 425–478.

29.

Chen

Yen

DC.

Dimensions of self-efficacy in the study of smart phone acceptance. Comput Stand Interfaces 2011; 33(4): 422–431.

30.

MacKenzie

Podsakoff

NP.

Construct measurement and validation procedures in MIS and behavioral research: integrating new and existing techniques. MIS Q 2011; 35(2): 293–334.

31.

Mun

Jackson

Park

, et al. Understanding information technology acceptance by individual professionals: toward an integrative view. Inf Manage 2006; 43(3): 350–363.

32.

Tung

F-C

Chang

S-C

Chou

C-M.

An extension of trust and TAM model with IDT in the adoption of the electronic logistics information system in HIS in the medical industry. Int J Med Inform 2008; 77(5): 324–335.

33.

Gonzalez

Mitra

Turel

Motivational impacts on intent to use health-related social media. J Comput Inf Syst 2020; 60(2): 136–145.

34.

Rolls

Hansen

Jackson

, et al. How health care professionals use social media to create virtual communities: an integrative review. J Med Internet Res 2016; 18(6): e166.

35.

Antheunis

Tates

Nieboer

TE.

Patients’ and health professionals’ use of social media in health care: motives, barriers and expectations. Patient Educ Couns 2013; 92(3): 426–431.

36.

Naslund

Aschbrenner

Marsch

, et al. The future of mental health care: peer-to-peer support and social media. Epidemiol Psychiatr Sci 2016; 25(2): 113–122.

37.

Moorhead

Hazlett

Harrison

, et al. A new dimension of health care: systematic review of the uses, benefits, and limitations of social media for health communication. J Med Internet Res 2013; 15(4): e85.

38.

Arazy

Nov

Patterson

, et al. Information quality in Wikipedia: the effects of group composition and task conflict. J Manag Inf Syst 2011; 27(4): 71–98.

39.

Eysenbach

Infodemiology and infoveillance: framework for an emerging set of public health informatics methods to analyze search, communication and publication behavior on the Internet. J Med Internet Res 2009; 11(1): e11.

40.

Hawn

Take two aspirin and tweet me in the morning: how Twitter, Facebook, and other social media are reshaping health care. Health Aff 2009; 28(2): 361–368.

41.

Pirraglia

Kravitz

RL.

Social media: new opportunities, new ethical concerns. J Gen Intern Med 2013; 28(2): 165–166.

42.

Southwell

Thorson

Sheble

The persistence and peril of misinformation. Am Sci 2017; 105(6): 372–375.

43.

Tasnim

Hossain

Mazumder

Impact of rumors or misinformation on coronavirus disease (COVID-19) in social media. Journal of Preventive Medicine. Epub ahead of print 29 March 2020. DOI: 10.31235/osf.io/uf3zn.

44.

Loeb

Taylor

Borin

, et al. Fake news: spread of misinformation about urological conditions on social media. Eur Urol Focus 2020; 6(3): 437–439.

45.

Harris

Mueller

Snider

Social media adoption in local health departments nationwide. Am J Public Health 2013; 103(9): 1700–1707.

46.

Ventola

CL.

Social media and health care professionals: benefits, risks, and best practices. Pharm Ther 2014; 39(7): 491–520.

47.

Hamm

Chisholm

Shulhan

, et al. Social media use among patients and caregivers: a scoping review. BMJ Open 2013; 3(5): e002819.

48.

Hader

Brown

. Patient privacy and social media. AANA J 2010; 78(4): 270–274.

49.

Thompson

Dawson

Ferdig

, et al. The intersection of online social networking with medical professionalism. J Gen Intern Med 2008; 23(7): 954–957.

50.

Anderson

Puckrin

Social network use: a test of self-regulation. J Nurs Regul 2011; 2(1): 36–41.

51.

Vanneste

Vermeulen

Declercq

Healthcare professionals’ acceptance of BelRAI, a web-based system enabling person-centred recording and data sharing across care settings with interRAI instruments: a UTAUT analysis. BMC Med Inform Decis Mak 2013; 13(1): 129.

52.

Phichitchaisopa

Naenna

Factors affecting the adoption of healthcare information technology. EXCLI J 2013; 12: 413–436.

53.

Lee

Shi

Cheung

, et al. Consumer’s decision to shop online: the moderating role of positive informational social influence. Inf Manage 2011; 48(6): 185–191.

54.

Thielst

CB.

Social media in healthcare. Chicago, IL: Health Administration Press, 2010.

55.

Bandura

Self-efficacy: toward a unifying theory of behavioral change. Psychol Rev 1977; 84(2): 191.

56.

Hocevar

Flanagin

Metzger

MJ.

Social media self-efficacy and information evaluation online. Comput Human Behav 2014; 39: 254–262.

57.

Adzharuddin

Ramly

NM.

Nourishing healthcare information over Facebook. Procedia Soc Behav Sci 2015; 172: 383–389.

58.

Chou

W-yS

Prestin

Lyons

, et al. Web 2.0 for health promotion: reviewing the current evidence. Am J Public Health 2013; 103(1): e9–e18.

59.

Eschenbrenner

Nah

FF-H

. Social media in health care. In: International conference on HCI in business: 2015, August 2–7, 2015, pp.76–85. Los Angeles, CA, USA, August 2–7, 2015, Springer.

60.

Eastin

LaRose

Internet self-efficacy and the psychology of the digital divide. J Comput Mediat Commun 2000; 6(1): JCMC611.

61.

Hsu

M-H

Chiu

C-M.

Internet self-efficacy and electronic service acceptance. Decis Support Syst 2004; 38(3): 369–381.

62.

Joo

Y-J

Bong

Choi

H-J.

Self-efficacy for self-regulated learning, academic self-efficacy, and Internet self-efficacy in Web-based instruction. Educ Technol Res Dev 2000; 48(2): 5–17.

63.

Burton-Jones

Straub

Jr.

Reconceptualizing system usage: an approach and empirical test. Inf Syst Res 2006; 17(3): 228–246.

64.

Straub

Limayem

Karahanna-Evaristo

Measuring system usage: implications for IS theory testing. Manage Sci 1995; 41(8): 1328–1342.

65.

Turner

Kitchenham

Brereton

, et al. Does the technology acceptance model predict actual use? A systematic literature review. Inf Softw Technol 2010; 52(5): 463–479.

66.

Ajzen

The theory of planned behavior. Organ Behav Hum Decis Process 1991; 50(2): 179–211.

67.

Bhattacherjee

Understanding information systems continuance: an expectation-confirmation model. MIS Q 2001; 25: 351–370.

68.

Delone

McLean

ER.

The DeLone and McLean model of information systems success: a ten-year update. J Manag Inf Syst 2003; 19(4): 9–30.

69.

Bhattacherjee

Hikmet

Physicians’ resistance toward healthcare information technology: a theoretical model and empirical test. Eur J Inf Syst 2007; 16(6): 725–737.

70.

Fishbein

Ajzen

Belief, attitude, intention, and behaviour: an introduction to theory and research. Reading, MA: Addison-Wesley, 1975.

71.

Venkatesh

Determinants of perceived ease of use: integrating control, intrinsic motivation, and emotion into the technology acceptance model. Inf Syst Res 2000; 11(4): 342–365.

72.

Venkatesh

Davis

A theoretical extension of the technology acceptance model: four longitudinal field studies. Manage Sci 2000; 46(2): 186–204.

73.

Venkatesh

Sykes

Zhang

. ‘Just what the doctor ordered’: a revised UTAUT for EMR system adoption and use by doctors. In: 2011 44th Hawaii international conference on system sciences: 2011, 2011, pp.1–10. New York, NY: IEEE.

74.

Zhang

Guo

Lai

K-H

, et al. Understanding gender differences in m-health adoption: a modified theory of reasoned action model. Telemed J E Health 2014; 20(1): 39–46.

75.

Kijsanayotin

Pannarunothai

Speedie

SM.

Factors influencing health information technology adoption in Thailand’s community health centers: applying the UTAUT model. Int J Med Inform 2009; 78(6): 404–416.

76.

Hoque

MR.

An empirical study of mHealth adoption in a developing country: the moderating effect of gender concern. BMC Med Inform Decis Mak 2016; 16(1): 51.

77.

Hoque

Bao

Sorwar

Investigating factors influencing the adoption of e-Health in developing countries: a patient’s perspective. Inform Health Soc Care 2017; 42(1): 1–17.

78.

Dzandu

Boateng

Agyemang

, et al. Social media adoption among university students: the role of gender, perceived usefulness and perceived ease of use. Int J Soc Media Interact Learn Environ 2016; 4(2): 124–136.

79.

Salthouse

Mediation of adult age differences in cognition by reductions in working memory and speed of processing. Psychol Sci 1991; 2(3): 179–183.

80.

Zhao

Zhou

What factors influence the mobile health service adoption? A meta-analysis and the moderating role of age. Int J Inf Manage 2018; 43: 342–350.

81.

Wang

ES-T

. Perceived control and gender difference on the relationship between trialability and intent to play new online games. Comput Human Behav 2014; 30: 315–320.

82.

Guo

Sun

Wang

, et al. The dark side of elderly acceptance of preventive mobile health services in China. Electron Mark 2013; 23(1): 49–61.

83.

Guo

Zhang

Sun

The privacy–personalization paradox in mHealth services acceptance of different age groups. Electron Commer Res Appl 2016; 16: 55–65.

84.

Bush

Hair

Jr.

An assessment of the mall intercept as a data collection method. J Mark Res 1985; 22: 158–167.

85.

Murry

Lastovicka

Bhalla

Demographic and lifestyle selection error in mall-intercept data. J Advert Res 1989; 29(1): 46–53.

86.

Nunnally

Bernstein

Psychometric theory. New York, NY: McGraw-Hill, 1978.

87.

Tabachnick

Fidell

LS.

Using multivariate statistics. Boston, MA: Allyn & Bacon/Pearson Education, 2007.

88.

Hair

Black

Babin

, et al. Multivariate data analysis. 6th ed. Upper Saddle River, NJ: Pearson Prentice Hall, 2006.

89.

Bentler

PM.

Simultaneous equation systems as moment structure models: with an introduction to latent variable models. J Econom 1983; 22(1–2): 13–42.

90.

Browne

MW.

Covariance structures. In: Hawkins

(ed.) Topics in applied multivariate analysis. London: Cambridge University Press, 1982, pp.72–141.

91.

Byrne

BM.

Structural equation modeling: perspectives on the present and the future. Int J Test 2001; 1(3–4): 327–334.

92.

Hulland

Chow

Lam

Use of causal models in marketing research: a review. Int J Res Mark 1996; 13(2): 181–197.

93.

Hair

Anderson

Tatham

, et al. Multivariate data analysis. 5th ed. Upper Saddle River, NJ: Prentice-Hall International, Inc., 1998.

94.

Wiederhold

BK.

Using social media to our advantage: alleviating anxiety during a pandemic. Cyberpsychol Behav Soc Netw 2020; 23(4): 197–198.

95.

Panahi

Social media and tacit knowledge sharing: physicians’ perspectives and experiences. PhD Thesis, Queensland University of Technology, 2014.

96.

Osatuyi

Information Seeking on Social Media Sites: An Exploratory Study. AMCIS 2012 Proceedings. 15. http://aisel.aisnet.org/amcis2012/proceedings/VirtualCommunities/1 (2012, accessed 2 July 2018).

97.

PJ-H

Chau

Sheng

ORL

. Adoption of telemedicine technology by health care organizations: an exploratory study. J Organ Comput Electron Commer 2002; 12(3): 197–221.

98.

Chismar

Wiley-Patton

. Test of the technology acceptance model for the internet in pediatrics. In: Proceedings of the AMIA symposium: 2002, 2002, pp. 155–159. American Medical Informatics Association.

99.

Cheng

Lam

Yeung

AC.

Adoption of internet banking: an empirical study in Hong Kong. Decis Support Syst 2006; 42(3): 1558–1572.

100.

Duyck

Pynoo

Devolder

, et al. User acceptance of a picture archiving and communication system. Methods Inf Med 2008; 47(2): 149–156.

101.

Compeau

Higgins

CA.

Computer self-efficacy: development of a measure and initial test. MIS Q 1995; 19:189–211.

102.

Charter

Feldt

LS.

The relationship between two methods of evaluating an examinee’s difference scores. J Psychoeduc Assess 2000; 18(2): 125–142.

103.

Fornell

Larcker

DF.

Structural equation models with unobservable variables and measurement error: algebra and statistics. J Mark Res 1981; 18: 382–388.

104.

Bruwer

JDW

Haydam

Lin

Reducing bias in shopping mall-intercept surveys: the time-based systematic sampling method. S Afr J Bus Manage 1996; 27(1 & 2): 9–16.

105.

Holden

Karsh

B-T.

The technology acceptance model: its past and its future in health care. J Biomed Inform 2010; 43(1): 159–172.

106.

Kuo

Y-F

Yen

S-N.

Towards an understanding of the behavioral intention to use 3G mobile value-added services. Comput Human Behav 2009; 25(1): 103–110.

107.

Liao

C-H

Tsou

C-W

Huang

M-F.

Factors influencing the usage of 3G mobile services in Taiwan. Online Inf Rev 2007; 31(6): 759–774.

108.

Van Der Velden

El Emam

. “Not all my friends need to know”: a qualitative study of teenage patients, privacy, and social media. J Am Med Inform Assoc 2013; 20(1): 16–24.

109.

Cuan-Baltazar

Muñoz-Perez

Robledo-Vega

, et al. Misinformation of COVID-19 on the internet: infodemiology study. JMIR Public Health Surveill 2020; 6(2): e18444.

110.

Wang

Lin

, et al. Seeking and sharing health information on social media: a net valence model and cross-cultural comparison. Technol Forecast Soc Change 2018; 126: 28–40.

111.

Househ

Borycki

Kushniruk

Empowering patients through social media: the benefits and challenges. Health Informatics J 2014; 20(1): 50–58.

112.

Kordzadeh

Warren

Toward a typology of health 2.0 collaboration platforms and websites. Health Technol 2013; 3(1): 37–50.

113.

Brennen

Simon

Howard

, et al. Types, sources, and claims of COVID-19 misinformation. Reuters Institute, 7 April 2020, 3–1.

114.

Dumpit

Fernandez

CJ.

Analysis of the use of social media in Higher Education Institutions (HEIs) using the Technology Acceptance Model. Int J Educ Technol High Educ 2017; 14(1): 5.

115.

Hanson

West

Neiger

, et al. Use and acceptance of social media among health educators. Am J Health Educ 2011; 42(4): 197–204.

116.

J-C

Lee

S-C

Suh

Y-H

: Determinants of behavioral intention to mobile banking. Expert Syst Appl 2009; 36(9): 11605–11616.

117.

Maillet

Mathieu

Sicotte

. Modeling factors explaining the acceptance, actual use and satisfaction of nurses using an Electronic Patient Record in acute care settings: an extension of the UTAUT. Int J Med Inform 2015; 84(1): 36–47.

118.

Lee

M-C.

Factors influencing the adoption of internet banking: an integration of TAM and TPB with perceived risk and perceived benefit. Electron Commer Res Appl 2009; 8(3): 130–141.

119.

Szymanski

Henard

DH.

Customer satisfaction: a meta-analysis of the empirical evidence. J Acad Market Sci 2001; 29(1): 16–35.

120.

Hong

Kang

MS.

An international comparison of technology adoption: testing the UTAUT model. Inf Manage 2011; 48(1): 1–8.

121.

Steenkamp

J-BEM

Baumgartner

. Assessing measurement invariance in cross-national consumer research. J Consum Res 1998; 25(1): 78–90.

122.

Saleh

Quazi

Keating

, et al. Quality and image of banking services: a comparative study of conventional and Islamic banks. Int J Bank Mark 2017; 35: 878–902.

123.

Aguinis

Beaty

Boik

, et al. Effect size and power in assessing moderating effects of categorical variables using multiple regression: a 30-year review. J Appl Psychol 2005; 90(1): 94.

124.

Livingstone

Taking risky opportunities in youthful content creation: teenagers’ use of social networking sites for intimacy, privacy and self-expression. New Media Soc 2008; 10(3): 393–411.

125.

Rosenblum

What anyone can know: the privacy risks of social networking sites. IEEE Secur Priv 2007; 5(3): 40–49.

126.

Yang

C-c

Brown

BB.

Factors involved in associations between Facebook use and college adjustment: social competence, perceived usefulness, and use patterns. Comput Human Behav 2015; 46: 245–253.

127.

Davis

FD.

User acceptance of information technology: system characteristics, user perceptions and behavioral impacts. Int J Man Mach Stud 1993; 38(3): 475–487.