The use of information and communication technologies (ICT) in healthcare may provide several benefits for the health industry, its stakeholders, and the society at large, including gains in service quality, efficiency, cost savings, and accessibility (Cimperman et al., 2016; Gonçalves et al., 2019; Huang, 2013). Such technological improvements become more critical as challenges associated to resource constraints and aging population increase (Lanseng & Andreassen, 2007; Rahimpour, Lovell, Celler, & McCormick, 2008; Tsai, 2014). Additionally, potential technological improvements in health services can also decrease social inequalities, increasing its need in underdeveloped and emerging countries (Cilliers, Viljoen, & Chinyamurindi, 2018; Fanta & Pretorius, 2018; Mauco et al., 2019).

The definition of telehealth is still subject to debate. Adjekum, Blasimme, and Vayena (2018) describe several terms, including digital health, telehealth, and telecare, which could be interchangeably adopted to describe the use of ICT in health care services to promote general improvements. Velsen et al. (2017) describe telemedicine as health services with significant ICT presence. The main goal of such services is to provide treatment and bridge the distance between patients and healthcare professionals. Telemedicine’s purposes also include the provision of relief and education for patients, and it is just a part of a broader potential use for ICT in health, called telehealth (Caetano et al., 2020; Tsai, 2014). From this broader perspective, telehealth includes practical service applications of ICT in daily health activities, such as teleconsulting, telediagnosis, telemonitoring, teletraining, among others (Caetano et al., 2020). Therefore, teleconsultation is one of the potential ICT services in telehealth. In this regard, the present study adopts the literature that categorizes teleconsultation as medical appointments through ICT means, including video-chat, mobile phone, or any ICT remote solution that enables the communication and interaction between healthcare providers and patients (Caetano et al., 2020; Velsen et al., 2017).

In Brazil, teleconsultation was not allowed by the country’s Federal Council of Medicine (CFM) until April 2020. Before April, CFM’s resolutions required medical professionals to physically examine patients before writing prescriptions. As a consequence of the COVID-19 pandemic, teleconsultation was exceptionally authorized through a presidential sanction, and its validity depends on the duration of the pandemic crisis (Caetano et al., 2020).

Finally, the potential positive impacts of technological solution in health care from a patients’ perspective should be highlighted. Teleconsultation’s positive influence on the lives and wellbeing of patients of distinct pathologies, such as diabetes (Dixon et al., 2020) and asthma (Portnoy, Waller, & Elliot, 2020), has already been proven. For instance, in a recent study with teleconsultation patients in Canada, Ouimet, Wagner, Raymond, and Pare (2020) suggested that teleconsultation’s utilitarian nature, which includes its potential capacity of saving time and of providing valuable and immediate health care access, has already affected positively patients’ perception about its current and future adoption. Therefore, understanding the drivers for the adoption of teleconsultation from a patient perspective becomes one of the critical factors to release its full potential to improve patients’ lives in the coming years.

Innovation acceptance is influenced by several factors. For instance, external factors, such as the urgency for relevant social solutions, increase the pressure for approvals of innovations’ diffusion as well as the need for solutions from potential users (Rogers, 2003). The outbreak of pandemics and other natural disasters had already influenced the approval and introduction of teleconsultation in such circumstances (Baudier et al., 2021). Policy makers are more flexible and more receptive in allowing a broader use of technological solutions, such as teleconsultation, to face challenges imposed by COVID-19 on society and on health providers (Bashshur, Doarn, Frenk, Kvedar, & Woolliscroft, 2020).

Additionally, innovation adoption rate may be influenced not only by its objective characteristics but also by its subjective ones, in addition to adopters’ individual aspects (Rogers, 2003). TAM (Davis, 1989) is considered the basis for technology acceptance’s concept, and it is currently adopted in several studies (Ferreira, Rocha, & Silva, 2014). According to Davis’ (1989) original model, technology acceptance refers to a consumer’s behavioral intention (BI) to use a specific technology, which will lead to its actual use. Behavioral intention is directly influenced by consumers’ attitude (ATT) toward using the technology. Attitude determinants are the cognitive responses to external stimulus based on consumers’ perceptions of a technology’s ease of use (PEOU) and usefulness (PU). Usefulness is also directly influenced by ease of use (Davis, 1989; Davis, Bagozzi, & Warshaw, 1989), as shown in Figure 1.

Figure 1
Illustration of technology acceptance model (TAM).

Nonetheless, TAM’s parsimony generated the need to further understand eventual additional factors that could influence user acceptance (Venkatesh, 2000). Venkatesh and Davis (1996) proposed the integration of the antecedents of TAM’s cognitive factors in the model. Following works searched for inclusion of broader perspectives on technology acceptance, proposing the addition of affective (Kulviwat, Bruner, Kumar, Nasco, & Clark, 2007; Reinares-Lara, Olarte-Pascual, Pelegrín-Borondo, & Pino, 2016) and other non-tangible emotional aspects on consumer’s technology acceptance process (Ramírez-Correa, Grandón, Ramírez-Santana, & Órdenes, 2019).

Regarding health services technology, the inherent and complex nature of such services demanded deeper comprehension of factors involved in patients’ technology acceptance. Hirani et al. (2017) highlighted TAM’s limitations and suggested incorporating items that could address areas of telehealth use by the patient, such as accessibility, comfort, usability, security, and confidentiality. Yun and Park (2010), for instance, proposed the inclusion of two constructs, ‘health consciousness’ and ‘perceived health risk,’ into TAM’s original model to evaluate consumers’ search and use of disease information on internet. Similarly, Cimperman, Brencic, and Trkman (2016) introduced health-related variables (‘doctor’s opinion’, ‘computer anxiety,’ and ‘perceived security’) to investigate elderly users’ acceptance of home telehealth services. Finally, literature review papers focused on technology acceptance in health and healthcare suggest the need to include variables in TAM, depending on the nature of the study and its purpose (Holden & Karsh, 2010; Or & Karsh, 2009; Tao et al., 2020).

This study proposes a more comprehensive model to investigate patients’ technology acceptance in the specific case of teleconsultation, including the complexity and specificities of health services and technologies. Therefore, our study’s model includes: (a) two constructs with voluntary behavioral responses from patients, attitude (ATT) and intention to use (BI) (Davis, 1989; 1993; Davis et al., 1989); (b) two cognition reasoning constructs, perceived ease of use (PEOU) and usefulness (PU), usually associated with the acceptance of any new technological innovation (Davis, 1989; 1993; Davis et al., 1989); (c) and three other potential antecedents (Venkatesh, 2000) of such cognitive constructs. In line with the literature, these constructs include context-specific variables, i.e., constructs that hold significant relationship with health services - trust (TRU) and self-efficacy (SE) -, as well as a construct, technology readiness (TR), directly related to user’s potential acceptance of technological innovations of any kind. This comprehensive set of selected constructs intents to comply with the study’s purpose and to contribute with users’ acceptance literature.

Technology readiness (TR) refers to an individual’s mental and psychological state when using new technologies. Parasuraman (2000) defines TR as “people’s propensity to embrace and use new technologies for accomplishing goals in home life and work” (Parasuraman, 2000, p. 308). Consequently, it encompasses beliefs and feelings that drive individuals’ (Ferreira et al., 2014) or communities’ (Jennett et al., 2003) predisposition to adopt technologies.

In measuring the level of individuals’ TR, Parasuraman (2000) and Parasuraman and Colby (2015)observed that consumers simultaneously presented positive and negative views about technologies. Positive views act as motivators, contributing positively to TR, whilst negative views conceive the inhibitors that diminish consumers’ TR. The views were categorized in four dimensions: optimism and innovativeness as TR motivators, and discomfort and insecurity as TR inhibitors (Yeo et al., 2019). These dimensions reflect relative distinct behaviors that can lead, in some cases, even to paradoxical consolidated results, such as concomitant strong positive and negative feelings toward a technology by the same consumer (Parasuraman & Colby, 2015). Table 1 depicts the clear distinction among the unique characteristics of TR’s dimensions.

Table 1
Technology readiness dimensions, according to Parasuraman and Colby (2015).

TR influences adoption rates of new technological services and products (Kuo, 2011; Lin & Chang, 2011). Health studies suggest that understanding TR could be critical to increase the success rates of new technology (Quan et al., 2020), to avoid wasting resources, and to decrease the chances of frustrating stakeholders (Jennett et al., 2003; Yosser et al., 2020). However, literature review suggests that there is still little attention and production of empirical studies for this area and low use of reliable TR frameworks to assess end users’ readiness for new health technology services (Kayser et al., 2019; Mauco et al., 2019).

Although TR refers to general beliefs regarding technology, there are intuitive relationships between TR and TAM’s constructs (Ratchford & Barnhart, 2012). Parasuraman and Colby (2015) suggest that higher TR leads to higher PEOU and BI, and propose further studies to understand the relationships between their TR scale (Technology Readiness Index 2.0 - TRI 2.0) and TAM. The theoretical association between TR and TAM constructs was tested with various technological products and services. As suggested by Venkatesh (2000), PEOU and PU are significantly influenced by TR, as a global measure (Ferreira et al., 2014; Lin, Shih, & Sher, 2007) or split by its four dimensions (Lam, Chiang, & Parasuraman, 2008).

Results suggest that individuals with high levels of TR motivators (innovativeness and optimism) tend to perceive technology usefulness (PU) more easily (Lin et al., 2007).

Regarding PEOU, people with high levels of technology readiness present lower levels of discomfort and insecurity about a given technology, and can understand more easily how such technology works and how to use it (Ferreira et al., 2014).

In the health sector, a study that analyzed the adoption of mobile electronic medical records by nurses suggested that high levels of discomfort and insecurity negatively affected PEOU (Kuo, Liu, & Ma, 2013). Lanseng and Andreassen (2007) evaluated users’ readiness and attitude to predict the adoption of self-service technology in health diagnosis, and measured users’ TR to select the study group, composed only of ‘pioneers’ (users with higher propensity for technology acceptance). The pioneers’ PU and PEOU were later evaluated and the model to predict the pioneers’ behavioral intention was validated. The researchers assumed, therefore, that users with higher TR would be more prone to adopt new technologies. Therefore, this study proposes:

Hypothesis 1a: Technology readiness will have a direct positive impact on patients’ perceived ease of use of a technological health service (teleconsultation).

Hypothesis 1b: Technology readiness will have a direct positive impact on patients’ perceived usefulness of a technological health service (teleconsultation).

From a social cognitive perspective, self-efficacy (SE) can be interpreted as a person’s abstract judgments about his/her capabilities to perform a task (Tsai, 2014). From a technological perspective, SE refers to a person’s self-evaluation of his/her capability of efficiently using a new technology (Compeau & Higgins, 1995) whereas in health studies, SE represents a patient’s judgment of his/her capabilities to use a telehealth system or solution (Rahimpour et al., 2008).

Venkatesh and Davis (1996) observed conceptual similarities between SE and PEOU. The authors tested their relationship empirically and found that SE can act as predictor of PEOU. Moreover, SE has more influence on PEOU when systems or technologies have not been experienced by users, since they tend to base their decisions on more abstract criteria (Venkatesh, 2000; Venkatesh & Davis, 1996). This is the case in our study, considering most respondents had never experienced teleconsultation.

The link between SE and PEOU was suggested by various authors, such as Venkatesh and Davis (1996) in their study about users’ perceptions of computer programs. Khalifa and Shen (2008) also found a significant relationship between SE and PEOU in a B2C mobile commerce study, and Abdullah, Ward, and Ahmed (2016) found similar results while investigating UK students’ intention to adopt e-portfolios. Similar results, showing a positive and significant relationship between SE and PEOU, were found for diverse telehealth services too, such as home telecare (Rahimpour et al., 2008) and a national telehealth system (Tsai, 2014).

In this study, self-efficacy refers to patients’ judgment of their ability to use teleconsultation services. Additionally, we assume that users tend to believe a new technology is easier to use depending on their judgment on their capabilities to use such technology. Thus, this study hypothesizes that:

Hypothesis 2: Self-efficacy will have a direct positive impact on patients’ perceived ease of use of a technological health service (teleconsultation).

Trust

Trust reduces uncertainty (Malaquias & Hwang, 2016) and incorporates the idea of vulnerability. People accept voluntarily to become vulnerable to others’ actions in exchange for relevant personal gain, without being able to monitor or control others’ actions and behaviors.

On the other hand, such trustor’s willingness to be vulnerable is impacted by his/her perceptions on trustee’s ability, benevolence, and integrity and is directly associated with a willingness to take risk (Mayer, Davis, & Schoorman, 1995; Schoorman, Mayer, & Davis, 2007; Velsen et al., 2017). In healthcare, the relationship between patients and healthcare providers creates an unbalance, generating the need for patients to have high levels of trust in the providers, which increases confidence risks in case of services failures (Lanseng & Andreassen, 2007).

Velsen et al. (2017) observe that patients’ willingness to rely on health services is due to health services’ peculiar characteristics (e.g., high importance for the patient, sensitive personal data) and is determined by a multidimensional perspective. In the case of health technology-related services, patients’ trust is the sum of trust in all factors associated with these services. For instance, data sensitivity (Cimperman et al., 2016), the complexity of the interplay between enablers and impediments (Adjekum et al., 2018), and high expectations from patients (Tsai, 2014) denote the intuitive and natural issue of trustworthiness involved in health technology-related services.

Trust can be related to PU through trustor’s uncertainty regarding expected gains. The higher the consumer’s trust in the vendor or the technology, the higher the perceived usefulness of such vendor or technology should be (Geffen, Karahanna, & Straub, 2003). Tsai (2014) observed that elderly individuals’ trust on healthcare providers’ ability, benevolence, and integrity directly affected the expected outcomes of telehealth systems, while Lanseng and Andreassen (2007) suggested that users’ trust on e-healthcare providers affected positively users’ perceptions of the potential gains from using such technology. Therefore:

Hypothesis 3: Trust will have a direct positive impact on patients’ perceived usefulness of a technological health service (teleconsultation).

Perceived usefulness (PU) refers to an individual’s subjective judgment and expectation about how a certain technology might improve his/her performance or outcomes for a given task, which might influence the individual’s tendency to adopt it (Davis et al., 1989). In healthcare, PU is associated with patients’ perceptions of benefits that the adoption of a health-related technology, such as teleconsultation, can bring to their daily lives and health conditions (Cimperman et al., 2016).

Since one’s attitudes (ATT) toward a behavior are generated by strong beliefs (Davis et al., 1989; Fishbein & Ajzen, 1975), PU, as one of these beliefs, directly and positively affects ATT. A relationship between PU and ATT was empirically tested and validated in studies with diverse technologies, including nanoimplants (Reinares-Lara et al., 2016) and e-book readers (Ferreira et al., 2014). Similar empirical evidences were found in health technology-related services (Tao et al., 2020), such as a telecare system (Huang, 2013) and internet disease-seeking information (Yun & Park, 2010). Consequently, this study assumes that:

Hypothesis 4a: Perceived usefulness will have a direct positive impact on patients’ attitude in relation to adoption of a technological health service (teleconsultation).

PU can also potentially influence consumers’ intention to adopt (BI) a technology, including health-related technologies (Tavares & Oliveira, 2016). Kulviwat, Bruner, Kumar, Nasco, and Clark (2007) suggest it is possible to obtain distinct or contradictory results in consumers’ attitude and intentions that can be found in some technologies. Use dependency may overcome eventual negative consumer’s attitude, such as mobile phones and health services. Tao et al.’s (2020) evaluation of fifty-one trials from health empirical studies on technology adoption also suggests significant influence on patients’ BI. Hence,

Hypothesis 4b: Perceived usefulness will have a direct positive impact on patients’ behavioral intention to use a technological health service (teleconsultation).

While PU refers to consumers’ subjective appraisal of a technology performance, perceived ease of use (PEOU) includes the effort side of such an appraisal, encompassing the perceptions of how easy or effortless the use of a given technology may be (Davis, 1989; Davis et al., 1989; Venkatesh, 2000). PEOU influences attitude toward technology adoption both directly and indirectly, through PU. The sense of efficacy and instrumentality are two PEOU mechanisms influencing attitude (Davis et al., 1989). Bandura (1982) points to existing links between efficacy and self-motivation/affect, which can be found in the relationship between PEOU and ATT. Evidences of such positive impact were shown in several studies in the consumer context (Davis, 1993; Davis et al., 1989), including health technology services (Tao et al., 2020).

Regarding indirect impact on ATT, Davis, Bagozzi, and Warshaw (1989) posit that easier to use technologies save consumers’ efforts that are redeployed to the actual use of the technology, leading to higher results from this given technology with the same level of efforts. Hence, all the rest being equal, the easier a particular technology may be perceived, the higher the perceptions of its performance or usefulness will be (Kulviwat et al., 2007). Ease of use plays an important role for technology adoption in health services (Klaassen, Beijnum, & Hermens, 2016). In the health patient context, many studies found positive influence of PEOU on PU and ATT, including the willingness to continue using a fitness app (Beldad & Hegner, 2018) and the acceptance of telehealth services by older users (Cimperman et al., 2016). Consequently, this study theorizes that:

Hypothesis 5a: Perceived ease of use will have a direct positive impact on patients’ attitude in relation to adoption of a technological health service (teleconsultation).

Hypothesis 5b: Perceived ease of use will have a direct positive impact on patients’ perceived usefulness of a technological health service (teleconsultation).

Attitude encompasses the evaluative affect from an individual related to his/her willingness to perform a target behavior (Fishbein & Ajzen, 1975) and is determined by this individual’s salient beliefs (Davis et al., 1989). Hence, an individual will tend to adopt a new technology whenever his/her evaluation process is dominantly positive. The intensity of someone’s intention to perform a specific behavior (BI) is influenced not only by the affective (positive or negative) feelings, represented by ATT, but also by cognitive expectations (PU), since performance improvement can be relevant for achieving rewards related to specific consumer activities (Davis et al., 1989).

Kulviwat et al. (2007) highlight the relevance of ATT to understand technology acceptance in consumer context. Evidences of the positive ATT-BI relationship are found in studies about health technological services acceptance as well (Huang, 2013; Lanseng & Andreassen, 2007; Yun & Park, 2010). Moreover, despite the ongoing debate about ATT’s real relevance to predict BI, the evaluation of 22 trials in health studies suggests that ATT can still be a critical predicting factor to technology acceptance models (Tao et al., 2020).

Finally, due to its reliability to reflect actual use, behavioral intention (BI) has been efficiently adopted as the sole construct to measure technology adoption in several studies (Holden & Karsh, 2010; Ferreira et al., 2014). In addition, as teleconsultation is still in its incipient introductory phase in Brazil, most study’s respondents were still not using this service by the time the data was collected. In such situations, BI gains further reliability as the closest construct to reflect TAM’s actual usage (Davis, 1989). Therefore, we utilized BI as the final users’ technology acceptance measure in this study.

Hypothesis 6: Attitude toward adoption will have a direct positive impact on patients’ behavioral intention to use a technological health service (teleconsultation).

Figure 2 shows the study’s proposed model and all its hypotheses.

Figure 2
Proposed model with study’s hypotheses.

The selection of scales was based on their adherence to the survey’s constructs, relevance (scales already developed and tested), and previous empirical application within our scope (health and technology). Therefore, the study used the following scales: Parasuraman and Colby’s (2015) TRI 2.0 scale of 16 items for TR; a five-item scale for trust (Velsen et al., 2017); a three-item scale for SE (Sun, Wang, Guo, & Peng, 2013); a five-item scale for PU (Huang, 2010); a four-item scale for PEOU (Huang, 2010); and two separate three-item scales for ATT (Lee, 2005) and BI (Khalifa & Shen, 2008). All construct-related items were measured by five-point Likert scales. Since all scales were originally developed in English, the questionnaire was translated into Portuguese and translated back into English, guaranteeing the scales’ integrity. Thirteen additional questions were included to cover demographic characteristics and other general areas of interest. All items were evaluated by marketing or innovation diffusion researchers as well as by a health specialist, both before and after the pre-test process.

The pre-tests included video and phone interviews with potential research respondents aimed at evaluating the clarity and fluidity of the survey’s questions. Ten respondents were included in the first pre-test and six additional ones participated in the second pre-test. After each pre-test, the study’s researchers assessed together interviewees’ inputs and refined the instrument in a consensus-search mode.

At the end, the final instrument, consisting of 52 questions (39 related to the seven constructs’ scales and 13 to demographic and other general information), received only minor wording and formatting changes in two demographic and one TR items. All survey’s items from each research construct are presented in Appendix A.

Since teleconsultation is a general service that could be adopted by any person, the study’s population could be the entire society. In line with Tao et al.’s (2020) findings, general consumers and specific patients are both important for studies of technology acceptance in health. Moreover, as teleconsultation was in its primary implementation phase in Brazil, drivers of technology acceptance could come from users’ expectations, instead of actual use. Therefore, to enhance the study’s final findings, our research used a broad scope for sampling. The main guiding selection principle was the inclusion of respondents who could be responsible for their own choice of the preferred medical care’s channels and providers. Thus, the ages of the 415 valid respondents varied from 17 years old (one respondent) to 80 years old (two respondents).

The questionnaires were self-administered and completed electronically. They included a short introductory explanation about the research’s purpose and teleconsultation. Respondents’ recruiting was performed initially by open and general invitations in social media (e.g., LinkedIn). Only voluntary participation was applied in this research and respondents’ anonymity was guaranteed. The survey took place in September 2020.

In total, 527 answers were received. Conditions for validation included the need of 100% response rate for the 39-construct items as well as other consistency checks, as recommended; 415 responses were considered valid for further statistical analyses.

According to valid answers, only 26.7% of respondents could be classified as regular users of teleconsultation (having experienced the service more than once). From the remaining 73.3%, a total of 19.0% believed that they could have used teleconsultation once, and the remaining 54.3% said that they had never used such service before; however, the great majority (80.0%) had already heard about it. Part of the respondents had already been advised by their doctors (20.0%) or their health plans (27.2%) to use the service. Most respondents believed that teleconsultation should be implemented in Brazil, be it with a limited scope regarding medical specialties (79.3%) or only for limited phases of the health treatment (83.5%). Answers came from 17 out of 27 Brazilian states. Additional information about the sample group can be found in Table 2.

Table 2
Sample population’s characteristics and other info.

Comparing the demographic profiles of the respondents who might have used teleconsultation at least once (45.7%) to the ones that had never experienced teleconsultation before (54.3%), Table 3 shows that teleconsultation’s penetration had only reached the majority of the sample population in four sub-groups of respondents: (a) women (53.5%); (b) holders of private health plans with reimbursement conditions (51.4%); (c) respondents who search for health services in the city of São Paulo (51.5%); (d) or in other Brazilian state capitals (56.6%), except Rio de Janeiro. On the other hand, respondents who (a) pay directly for their health services (23.5%), (b) are over 65 years old (38.9%), (c) are men (38.7%), and (d) do not live in Brazilian state capitals (27.4%) seem to be more resistant to experiencing teleconsultation so far. Such results suggest that, as predicted in this study’s scope, teleconsultation is still at its earlier stages of implementation in Brazil.

Table 3
Use of teleconsultation according to the sample’s demographic characteristics.

Using information for independent and dependent variables from a single source (the respondents) could generate biases, resulting in common method variance issues. The study applied Harman’s one-factor test to investigate such risk (Podsakoff & Organ, 1986), resulting in eight factors with eigenvalues greater than 1.0 and no single factor being able to explain most of the variance. The factor that better explained the variance corresponded to 28.5% of the total. Results suggest that common method variance is not a problem for the collected data.

To test the scales’ validity, dimensionality, and reliability, the study employed a confirmatory factor analysis (CFA). Final results, after several iterations for model refining, reached satisfactory fit indexes while maintaining adequate numbers of items per construct, including the maintenance of at least three items per construct. The final 32-item measurement model presented the following indexes: χ₂ = 981,617 and df = 428 (p < 0,001) IFI = 0.934; TLI = 0.932; CFI = 0.933; RMSEA = 0.056; and χ₂/df = 2.293, which suggest adequate fit of the study’s proposed model to the collected data (Hair, Black, Babin, & Anderson, 2009; Hu & Bentler, 1999).

Model’s face validity was reached by: (a) adopting scales already developed, tested, and used in similar areas of research (health and technology); (b) observing best practices from literature in our translation process; and (c) executing and evaluating two pre-tests before releasing the final instrument. Previous literature review studies in consumer marketing, technology, and health technology predicted positive relationships among constructs.

Table 5 shows positive correlations in all cases, demonstrating nomological validity for the model (Hair et al., 2009). In order to measure variables’ fit to model’s proposal, convergent validity and reliability were further examined for each construct. Table 4 shows results for composite reliability (CR), average variance extracted (AVE), Cronbach’s alpha coefficients, and descriptive statistics (mean and standard deviation).

Table 4
Reliability indexes and descriptive statistics of study’s constructs.

Table 5
Correlations and discriminant validity matrix.

Note. AVE on diagonal; correlations above and squared correlations below main diagonal.

Reliability tests for each construct reached the minimum required thresholds (CR and alphas higher than 0.7) indicating consistency and reliability among items within each construct (Hair et al., 2009). Regarding convergent validity, all constructs reached the minimal AVE value of 0.5, suggested by Fornell and Larcker (1981).

Latent constructs’ discriminant validity was also assessed through the measurement model results. Discriminant validity, referring to how different a construct really is from the others (Hair et al., 2009), was evaluated by analyzing the factor loads of each item, which must be greater in relation to the constructs they are supposed to measure than in relation to other constructs present in the model. The relative importance and significance of the factor loads for each item were interpreted according to the suggestions of Hair, Black, Babin, and Anderson (2009), who state that factor loads greater than 0.3 can be considered significant, greater than 0.4 can considered important, and greater than 0.5 can be considered very significant. All items presented factor loads greater than 0.6 in relation to their respective constructs, suggesting satisfactory discriminant validity. Nevertheless, further testing was carried out, since, according to Fornell and Larcker (1981), constructs achieve discriminant validity when all AVEs are greater than the corresponding squared inter-construct correlations with the other constructs present in the model. Table 5 presents the AVEs for each construct in the main diagonal and the squared correlation coefficients between each pair of constructs in the cells below the main diagonal.

The examination of the Table 5 shows that almost all shared variances are lower than the variance extracted by the items measuring the constructs, with the exception of the pairs self-efficacy/ease of use, usefulness/attitude, usefulness/intention, and attitude/intention. Given those exceptions to Fornell and Larcker's (1981) criteria, we performed an additional test to ensure discriminant validity as indicated by Anderson and Gerbing (1988): for each problematic pair of constructs (SELF/PEOU, PU/ATT, PU/BI, and ATT/BI), we calculated the difference between the original model with correlations allowed to be free and one constrained model that fixed each problematic correlation to unity (i.e., perfectly correlated). Results for all cases showed that statistically significant and lower χ₂ values were found for the unconstrained models, providing evidence that discriminant validity was achieved for self-efficacy, ease of use, usefulness, attitude, and behavioral intention.

In sum, after proceeding with validation tests, the model can be considered reliable and ready for study’s hypotheses evaluation.

Structural equation modeling (SEM) through the use of the software AMOS was employed to test the proposed model and the study’s hypotheses. SEM verifies the fit of the model to the data and the levels of significance for the hypothesized relationships between the study’s constructs, attesting their potential veracity or not (Byrne, 2010). The estimation of the model generated indices indicating a good fit to the data: χ₂ = 1,214,744 with df = 440, χ₂/df ratio of 2.76, IFI = 0.91, TLI = 0.90, CFI = 0.91, and RMSEA = 0.065 (Hair et al., 2009; Hu & Bentler, 1999).

The authors then verified the path coefficients to test the study’s hypotheses in terms of significance, magnitude, and direction (Byrne, 2010). Figure 3 and Table 6 show significance in all hypothesized relationships. Finally, the proportions of the variance explained for each of the four dependent variables of the model were: 47.2% for usefulness; 67.7% for behavioral intention, 75.3% for easy of use; and 84.4% for attitude.

Figure 3
Standardized path coefficients for the study’s proposed model (* p < 0.05, ** p < 0.01).

Table 6
Hypotheses, standardized path coefficients, and significances.

Although respondents’ average characteristics may represent a relevant target audience for teleconsultation implementation in Brazil (higher social classes and private health plan consumers), the non-probabilistic process implemented during data collection makes generalization unfeasible. Moreover, only 9.5% of the sample’s population is 60 years old or more, a fact that can be a concern for any health service implementation, given the need to service a growing senior population.

Nonetheless, future research should try to include a longitudinal understanding of the COVID-19 pandemic impact on consumers’ acceptance of teleconsultation. This could add value to the debate about the impact of “exceptional times and situations” on the acceptance and diffusion of innovations, mainly from a user perspective. Additionally, other telehealth services could be included in future studies, mainly from countries like Brazil, where telehealth is still in its incipient implementation.

Our research’s results also showed the relevance of including ‘technological’ factors in the studies that aim at evaluating the adoption of telehealth solutions from a patient perspective. Thus, future studies can evaluate the inclusion of other potential technological factors in the search for antecedents of patients’ propensity to adopt teleconsultation and other telehealth solutions.

In addition, a deeper understanding of the profiles of teleconsultation’s users and potential users (e.g., age, income, and other sociodemographic aspects) could help policy makers and providers in directing efficiently their efforts for this service adoption. Moderation studies should investigate the differences that might exist among groups differing in sociodemographic aspects as well as evaluate if groups with different levels of experience with teleconsultation (those who have already used it or those who have only heard of it, for instance) present distinct significances or magnitudes for the relationships proposed in the model. Furthermore, studies evaluating the indirect effects of the proposed antecedents on behavioral intention and the mediating role of attitude and perceived usefulness, while testing out several competing models, could also provide interesting insights on the relationships between the different constructs and patients’ intentions to adopt new technologies for health services.

Finally, as tech health services’ complexity involves significant participation of diverse stakeholders, acceptance of teleconsultation in Brazil should be observed also by other domains, mainly from the clinicians’ point of view, given their primordial role in ICT success adoption in health services.

* Corresponding Author