2.1. Telemedicine and Medical teleconsultation

Telemedicine refers to the provision of medical services through interactive data and audiovisual communications, as defined by the World Health Organisation (Albahri et al., 2021). This practice encompasses not only healthcare services but also health education, medical data transfer, diagnosis and treatment. As a result of the combination of medicine and information and communication technologies, telemedicine has proven to be very effective in providing remote healthcare (Nittari et al., 2020). This is especially so in out-of-the-way areas or those with a poor availability of healthcare facilities. The increasing availability of technological infrastructure has driven the expansion of telemedicine, enabling specialists to provide remote care and eliminating the need for patient travel. (Nittari et al., 2020). In this sense, telemedicine is the natural evolution of healthcare in the digital world (Chellaiyan et al., 2019).

An appropriate integration of telemedicine can contribute to the sustainability of healthcare systems through three main channels (Dorsey & Topol, 2016): (1) by facilitating patient access to care while reducing associated costs; (2) by improving the management of chronic treatments, differentiating them from episodic or acute care; and (3) by delivering, through telemedicine, certain services traditionally provided in hospitals and clinics directly to patients’ homes and mobile devices. Despite these potential benefits, much remains to be learned regarding the effective implementation of telemedicine across different healthcare systems. Nevertheless, a growing body of evidence provides consistent support for its economic effects, particularly in terms of cost reduction (Bashshur et al., 2016).

The current practice of telemedicine relies on electronic devices accessible to physicians and patients and inexpensive tools, such as mobile phone cameras, biological sensors and laptops, to collect clinical information easily and without requiring specialised training (Mazzuca et al., 2022). Recent telemedicine has succeeded in reducing the costs and improving the accessibility of medical care, allowing people to access specialists without having to travel and without affecting their daily responsibilities, thus increasing productivity. In addition, healthcare professionals benefit from a decrease in daily workload due to a reduction in the number of missed or cancelled appointments. This, improves revenue, patient care capacity and treatment outcomes (Banning et al., 2022). Therefore, telemedicine has become an alternative for healthcare, bringing healthcare from hospitals and clinics to the home (Lopez et al., 2021).

The widespread implementation of telemedicine has faced several difficulties. Among the most prominent obstacles are a lack of awareness, costs associated with its implementation, the introduction of inefficiencies, difficulty in performing physical examinations, a deficiency of understanding about the benefits of virtual care, a negative financial impact, concerns about legal liability, and regulatory restrictions limiting its use (Makhni et al., 2020).

In the current context, medical teleconsultation has become a fundamental component of telemedicine. Technological advances such as computers, smartphones and tablets have allowed the doctor-patient and doctor-physician relationship to evolve from purely auditory communication to a higher level, where it is possible to send laboratory results, diagnostic images, videos and even show pathologies through video conversations (Liu et al., 2019). The effectiveness of teleconsultation has been demonstrated in multiple studies, such as those of (Augusterfer et al., 2020) and (Mishkin et al., 2022), which analyse its application in psychology and psychiatry and stress its usefulness and development but also emphasise the importance of training for professionals. (Wright & Honey, 2016) study the experience of telemedicine in home care delivery, while (Vural & Ramadan, 2019) analyse its effectiveness in emergencies. Recently, teleconsultation has also been used in the treatment of the aftermath of gender-based violence (Vranda & Cicil, 2022) and in consultations related to the COVID-19 pandemic, with high levels of patient satisfaction (Blanco Terés et al., 2022).

Different scopes of teleconsultation are used in telemedicine literature. Some authors refer to teleconsultation as remote specialist consultation provided to other health professionals (Augusterfer et al., 2020). Other studies use teleconsultation to denote direct clinical encounters between patients and clinicians conducted remotely (Blanco Terés et al., 2022; Mishkin et al., 2022). According to the current literature, this study adopts a definition of medical teleconsultation as any non‑‑‑ (Morcillo Serra & Aroca Tanarro, 2022). Teleconsultation can be divided into two modalities within this framework: synchronous and asynchronous teleconsultation. Firstly, synchronous teleconsultation involving real-time communication between patients and healthcare providers and necessitating their simultaneous availability (e.g., phone calls or video conferences). Secondly, asynchronous teleconsultation, which involves sending health information for later review by a healthcare provider, without both parties being present simultaneously (e.g., emails and secure messaging). This distinction is relevant from a clinical point of view. Synchronous modalities are better suited for consultations that require real-time assessment, while asynchronous modalities are beneficial for monitoring and managing chronic diseases (Morcillo Serra & Aroca Tanarro, 2022). In this study, participants reported their frequency through different channels: synchronous teleconsultation via phone, videoconference, and chat, and asynchronous teleconsultation via emails and WhatsApps.

Despite its potential benefits, the implementation of teleconsultation within healthcare systems has faced several obstacles and limitations. In this regard, Morillon et al. (2025) provide a literature review identifying the main challenges associated with teleconsultation. Specifically, they emphasise issues related to the type of consultation, the communication medium used, and its intended purpose. In addition, they point to limitations affecting healthcare professionals, stemming from insufficient training or lack of competencies to communicate with patients in a close and empathetic manner through electronic media. Institutional barriers are also identified, linked to characteristics of the healthcare system, service costs and reimbursement mechanisms. However, the majority of the limitations reported in the literature relate to patients’ personal characteristics—such as age, gender, educational level or employment status—as well as their capabilities (including technological skills, experiences and attitudes towards teleconsultation), their interests, medical conditions and social environment.

Feasibility, device size, user-friendly interface, empowering face-to-face interaction, the amount and type of patient data that can be transmitted and portability are key factors that, according to (Vranda & Cicil, 2022), determine the success of consultation tools in telemedicine, whether in asynchronous or real-time mode. For real-time interaction, there are various options such as mobile applications, chat and messaging platforms like WhatsApp or Skype, while emails and faxes are examples of asynchronous means. Although each tool has its advantages and disadvantages, it can be concluded that the medical teleconsultation process is evolving with technological advances, in accordance with (Vural & Ramadan, 2019).

Several studies have segmented telemedicine users according to different areas and analysis criteria. For example, (Cullere et al., 2022) analyse the preferences and evaluation of telemedicine patients, aiming to analyse the causes that justify their motivations for continuing to use the system or not. Also, according to the patient's expectation of the telemedicine provider, (Gabay et al., 2022) identify three different groups of different mentalities, obtaining significant differences between them. In the study by (Dopelt et al., 2021) important differences appear with respect to the segmentation of telemedicine users according to their educational level.

2.2. Segmentation and e-lifestyle

Ensuring effectiveness and sustainability requires health systems to understand the needs of their current and potential users. In this sense, lifestyles are an important element to consider in order to understand the behaviour and use of services (Arenas-Gaitán et al., 2022), such as teleconsultation. From a consumer behaviour point of view, lifestyles help to identify market segments (Wijaya et al., 2020) and have been used in marketing for decades (Lazer, 1963; Plummer, 1974; Singh & Basu, 2023). It is therefore essential for health systems to understand how lifestyles influence the behaviour of their patients in order to adapt their services and marketing strategies accordingly.

Importantly, knowing and understanding lifestyles can be a key factor in developing marketing strategies according to the intended segment. The very way in which segmentation has been carried out has also changed from using demographic variables, such as age or gender, to when looking at how individuals behaved and other variables with a strong psychographic character began to be employed. These psychographic variables showed a greater capacity to explain consumer behaviour (Peral-Peral et al., 2015) than demographic variables. In this way, lifestyles have become the basis for creating groups of consumers with homogeneous behaviours but heterogeneous among them.

In the field of consumer behaviour, lifestyles refer to a set of behaviours manifested by individuals motivated by their physical and psychological environment. This concept brings together values, attitudes, beliefs and patterns of consumer behaviour (Koshksaray et al., 2015). We refer to the idea of how consumers live their lives and spend their resources of time and money. In other words, it is a question of patterns of behaviour that differentiate people and help us understand the reasons why each person acts in a certain way (Hassan et al., 2015). In this sense, lifestyles are consistent over time and are a good predictor of consumer behaviour towards certain products or services.

Since the 1970s, several studies have attempted to measure the lifestyle of individuals. One of the first attempts to measure lifestyle is the AIO scale (Wells & Tigert, 1971), which captures the activities, interests and opinions of individuals. This scale defines activities as actual observable behaviours of individuals, interests as items to which an individual pays attention on an ongoing basis, and opinions as responses provided by individuals to specific events. Subsequently, the Stanford Research Institute developed the VALS scale (Mitchell, 1983) with 800 questions that collected demographic information, financial information, habits, activities, attitudes, and beliefs, among others. The VALS2 scale (Richie, 1989) is a revision of VALS and includes only 39 questions, including those that are demographic and psychographic.

In recent years, lifestyle scales related to some specific domains have appeared, such as food (Grunert, 1993), convenience products (Buckley et al., 2007), home (Thøgersen, 2017), wine (Bruwer & Li, 2017), fashion (Dahana et al., 2019) and the Internet (C. Yu, 2011). Especially in the case of Internet-related lifestyles, it makes sense to measure them given that individuals assume a dual role: on the one hand, as users of a technology (Internet) developed through computers; and on the other hand, as consumers of products who opt for this channel to satisfy their needs in one of the different phases of the purchasing process (Pandey & Chawla, 2018). Therefore, the online lifestyle may differ significantly from the general lifestyle presented by an individual (García-Fernández et al., 2020; Kim et al., 2002). In this way, e-lifestyles can be seen as individual forms of behaviour in the digital environment that reflect consumers' values, activities, interests and opinions (C. Yu, 2011).

In research on Internet-related lifestyles, several scales have been proposed. For example, a first approach is the work of (Lee et al., 2009). These authors conducted a study analysing the relationship between lifestyle and the selection of technological products. They identified four lifestyles in this area: fashion awareness, leisure orientation, Internet participation and e-shopping preferences. Their results show that these four lifestyles are direct or indirect antecedents of the tendency to adopt high-tech products. However, the most comprehensive and rigorous tool for measuring Internet-related lifestyle is the e-lifestyle scale, developed by (Yu, 2011).

E-Lifestyles refer to what consumers want, are interested in, and how they feel about the Internet (Ferreira-Barbosa et al., 2023). Yu (2011) developed a study, using exploratory and confirmatory factor analysis, to create an e-lifestyle scale. This scale adapts the activities, interests and opinions of AIO (Wells & Tigert, 1971) and the values of VALS (Mitchell, 1983; Richie, 1989) to the context of the Internet. The 39 items of the scale are grouped into seven factors, reflecting needs, interests, entertainment, sociability, perceived importance, disinterest or concern, and novelty.

The e-lifestyle scale has been widely used in previous studies in various areas. For example, Pandey and Chawla (2014), founded on Swinyard and Smith's (2003) web-based lifestyle scale and supported by qualitative research, modified and adapted it to the Indian market context. Theyidentified six factors: e-enjoyment, e-distrust, e-self-efficacy, e-logistical concerns, e-negative beliefs and e-offerings. Koshksaray et al. (2015) found a relationship between e-lifestyle dimensions and individuals' avoidance of Internet advertising. Their results indicate that lifestyles act as antecedents of individuals' behaviour towards Internet advertising. Yu (2015) applied e-lifestyle in the field of mobile banking, identifying five groups of users with differentiated behaviours: technology laggards, traditional banking lovers, digital followers, digital caregivers and digital seekers. In another paper, Hassan et al. (2015) showed that the e-lifestyle developed by Generation Y individuals was significantly related to the satisfaction they showed as consumers of their mobile phones. Yin et al. (2019) found three e-lifestyle segments which they called: digital technology enthusiasts, digital technology laggards and digital technology neutrals. In their results they found significant differences in the intention to share information about food safety risks between the groups, most notably between enthusiasts and laggards. Recently, Hassan et al. (2020) focused on young Muslims for whom the Internet is an important part of their lives and found that the e-lifestyle helps explain emotional, social and intrapersonal well-being.

In the field of health, there is a lack of studies linking lifestyles and health-related digital tools. Along these lines, the work of Weber et al. (2020) focuses on the over-60s in Germany, their general lifestyles (values, activities, and interests) and their use of the Internet as a source of health information. They find three major lifestyles: social adventurer, average family person, and uninterested inactive. The authors explain that the lifestyles of older people are a good predictor of their behaviour, above other demographic criteria such as age. It is important to know how patients relate to the Internet, since it is one of the main sources of health information (Weber et al., 2020), and one of the main avenues for the introduction of new services, such as teleconsultation. In our case, developing an e-lifestyle based segmentation enables us to better understand the current and future behaviour of users of digital health services. This knowledge should serve as a basis for decision-making by health system managers.

In summary, the e-lifestyle scale has proven to be robust and adaptable to different study contexts. However, there is a lack of studies developing this type of segmentation in the field of health-related services. As a result of the above literature review, we propose the following hypothesis:

2.3. Teleconsultation adoption

The Unified Theory of Acceptance and Use of Technology (UTAUT) was introduced by Venkatesh et al. (2003) to elucidate the adoption of information technologies. According to this framework, the intention to use information technologies is determined by four variables: perceived ease of use, perceived usefulness, social influence, and facilitating conditions. Perceived ease of use evaluates the user's belief in the simplicity and usability of the technology. Perceived usefulness examines the user's perception of the benefits derived from using the information technology. Social influence addresses the impact of influential figures on the decision to embrace information technology. And facilitating conditions consider the presence of an environment conducive to effective information technology utilisation. A decade later, the same author proposed UTAUT2 as a framework for evaluating the use of consumer technologies, incorporating the variables of habit, price value, and hedonic motivation (Venkatesh et al., 2012). In this model, habit refers to the extent to which people tend to perform behaviours automatically because of learning, while price value encompasses the individuals’ cognitive trade-off between the benefits of the applications and the monetary cost of using them. Additionally, hedonic motivation has to do with the fun or pleasure derived from using technology.

Several studies have adopted the UTAUT models as a framework for investigating telemedicine. For instance, Cimperman et al. (2013) indicate that, among other variables, perceived usefulness, effort expectancy, social influence, and facilitating conditions predict the usage of home telemedicine services among older adults in Slovenia. Similarly, Kamal, Shafiq, and Kakria (2020) report that perceived ease of use, perceived usefulness, social influence, and facilitating conditions positively influence behavioural intentions towards the acceptance of telemedicine services among the rural population of Pakistan. Lastly, Ramirez-Rivas et al. (2020) established that perceived usefulness explains attitudes towards telemedicine in a sample of Brazilians. Specifically, in elucidating the usage of teleconsultation, a recent study conducted a large-scale survey and data analysis across China, France, Italy, and the United Kingdom based on the UTAUT conceptual framework (Baudier et al., 2021). The results of this study highlight performance expectancy as a significant antecedent in explaining the utilisation of teleconsultation services.

Another line of research related to the UTAUT models involves the exploration of user segmentation using this conceptual framework. For example, Ramírez-Correa et al. (2020) focus on segmenting users of video games, while Arenas-Gaitán et al. (2019) delve into the segmentation of elderly Internet users. Furthermore, some studies have connected e-lifestyles with information technology acceptance models. A study by Mailangkay (2019) identified a positive relationship between e-lifestyles and attitudes towards using online tourist booking systems in Indonesia. Also, Arora and Sahney (2019) established that a dimension of the e-lifestyles (e-distrust) is an antecedent to the intention to research products online before visiting a physical store to purchase. Additionally, the findings from García-Fernández et al. (2020) support the association between e-lifestyles and perceived ease of use in mobile fitness applications among Spanish consumers.

In this context, we aim to explore whether groups of teleconsultation users with relatively similar e-lifestyles exhibit heterogeneous behaviours in adopting this technology. Therefore, we propose the following hypothesis:

3.1. Sample

Within the framework of this research, we worked with an initial sample of 1,500 teleconsultation users over 18 years of age who had made use of some type of non-face-to-face consultation, teleconsultation, during the last year. To guarantee the reliability and validity of the data obtained, a filtering process was carried out on the initial sample, eliminating those questionnaires that were completed in less than three minutes and those with inconsistent responses. As a result, an operational sample of 1,412 questionnaires was obtained. It should be noted that the sample was obtained in Spain between May and November 2022, and that a specialised data collection company was contracted to collect the data. The surveys were carried out using an electronic questionnaire. In this regard, we would like to highlight that Spain has one of the highest levels of Internet penetration in the world (94.9%), above other countries such as Germany (93.3%), the United States (91.8%) or Japan (82.9%) (Statista, 2023).

The Spanish health system follows the Beveridge model. This model is characterised by tax-based financing, universal access, salaried or capitated doctors, a minor role for the private sector and a strong state involvement in management. In the European context, other states that follow this model include Portugal, Italy, the United Kingdom, Ireland, Denmark, Finland, and Sweden. According to the most recent available data (Ministerio de Sanidad, 2025), more than 241 million medical consultations were provided in Spain in 2023, of which 29.5% were conducted through teleconsultation. In comparative European terms, Spain exhibits a medium–high level of teleconsultation penetration, particularly when compared with other countries operating under a Beveridge‑type healthcare model (OECD, 2025).

With regard to the characteristics of the respondents, 64.4% were male and 35.5% were female. Furthermore, 68.2% of the respondents resided in urban areas with a population of more than 50,000 inhabitants, while the remaining 31.8% resided in rural areas. In terms of educational background, only 1.7% had no or only basic education, while 59.3% had secondary education and 39% had university education. The age of respondents ranged from 18 to 74 years, with an average age of 38.6 years. Household size varied between 1 and 10 members, with a mean of 3.6. The number of children under 18 living in the household ranged from 0 to 6, with an average of 1.1, while the number of household members aged over 70 ranged from 0 to 5, with an average of 0.2.

3.2. Questionnaire and Measurement scales

The questionnaire was structured into three sections. The first section collected respondents’ socio‑demographic characteristics. Specifically, respondents were asked about their gender, age, education level and place of residence. In this sense, questions on family size and number of ascendants and descendants were incorporated.

The second section of the questionnaire focused on respondents’ e‑lifestyles. The e-lifestyle measurement scale was adapted from Yu (2011). The scale contains 39 items covering seven components: Needs, Interest, Entertainment, Social, Perceived Importance, Barriers, and Novelty. Moreover, the variable technology anxiety was incorporated using the Meuter et al. (2003) scale.

Finally, the third section examined respondents’ relationship with medical teleconsultation. Individuals were asked about their intention to use teleconsultation and their availability of means, adapting the scale developed by Venkatesh et al. (2012) to measure facilitating conditions, as well as the scale created by Kwon and Wen (2010) to measure usage. We inquired into the use of medical teleconsultation by asking about the frequency of use of the different media: Telephone, Videoconference, Chat, e-mail and WhatsApp. Five-point Likert scales were used. Detailed descriptions of the measurement scales are provided in the Appendix.

3.3. Statistical tools

To develop our research, we have used several statistical tools, beyond descriptive statistics. First, we utilised an exploratory factor analysis (EFA) employing the SPSS 26 software package to reveal the underlying dimensions and to summarise the multiple items collected on the different measurement scales into a few factors (Hair, 2009). In addition, a confirmatory factor analysis (CFA) was conducted using SmartPLS 4 to further validate the results (Hair et al., 2025). Second, we developed a latent class analysis (LCA) using the Latent Gold 6.0 software package. LCA is a technique for exploring unobserved heterogeneity in a sample. This methodology assigns individuals to different segments under the assumption that the data come from several homogeneous groups or segments that are mixed in unknown proportions (McLachlan & Basford, 1988; Vermunt & Magidson, 2005).

LCA assumes that an observed set of data derives from several underlying subpopulations and makes statistical inferences about the properties of these subpopulations having information on the pooled population only. Unlike other clustering techniques, such as hierarchical clustering, that try to find clusters with some arbitrary chosen distance measure, LCA derives clusters utilising a probabilistic approach. To do so, LCA employs maximum likelihood estimation with the expectation-maximisation algorithm, a well-established measure of the goodness of fit of a statistical model (Naldi & Cazzaniga, 2020). Based on the results of the cluster analysis using e-lifestyle-related factors as variables, the demographic and behavioural characteristics of the identified segment are analysed. For this last step, other information related to the personal characteristics of the respondents and their relationship with the medical teleconsultation was taken as covariates.

4.1. Exploratory Factor Analysis (EFA)

First, to group 39 items contained in the e-lifestyle scale (C. Yu, 2011) into dimensions, an EFA (J. Hair, 2009) was employed to maximise the explained variance of each factor. The Kaiser-Meyer-Olkin (KMO) test, Bartlett's test of sphericity and the determinant of the correlation matrix were used to assess the adequacy of the data for factor analysis. This confirmed that the correlation structure was suitable for the extraction of stable and interpretable latent factors (J. Hair et al., 1999). In all the cases, KMO levels above the recommended 0.7 were obtained. Bartlett's tests of sphericity were significant in all the cases (0.000). And the determinants of the correlation matrix were close to zero.

Compared to the expected results, we were surprised by the scale related to Needs. Yu´s 2011) proposal on Needs offered a single dimension containing the nine proposed items. However, our results pointed to two distinct dimensions within the Needs construct (C. Yu, 2011). The first one contained items related to "Internet Needs for Work", and the second dimension contained items on "Internet Needs for Everyday Actions". No differences appeared in the other dimensions.

As a result of this analysis, the 39 items of Yu´s (2011) scale were grouped into 8 factors: Internet Needs for Work, Internet Needs for Daily Actions, Interest, Leisure, Social, Importance, Barriers, and Novelty.

The same analysis was repeated for the variables related to technology acceptance. The results were similar to the previous ones. No differences were found in the expected dimensions. The 30 items related to technology acceptance were grouped into the eight factors expected within the UTAUT2 model: Performance Expectancy, Effort Expectancy, Facilitating Conditions, Hedonic Motivation, Habit, Social Influence, Use, Intention to use, and Technology Anxiety.

To ensure the reliability of the measurement scales, a confirmatory factor analysis (CFA) was conducted following the guidelines proposed by Hair et al. (2025). The results confirm the validity and reliability of all the scales utilised. In most cases, the individual factor loadings of the items exceed the recommended threshold of 0.7. At the construct level, internal consistency reliability was assessed using Cronbach’s alpha, composite reliability and the average variance extracted (AVE). Cronbach’s alpha values range from 0.769 to 0.935, all exceeding the recommended threshold of 0.7. Composite reliability values range between 0.777 and 0.935, also above the recommended minimum. AVE values range from 0.510 to 0.828, surpassing the 0.5 threshold. Finally, discriminant validity was assessed using the HTMT criterion. All the HTMT values were below the recommended threshold, which was further confirmed through a bootstrapping procedure with 5,000 subsamples.

We have created an appendix that includes the measurement scales used, their sources, and the main results of the confirmatory factor analysis.

4.2. Latent Class Analysis (LCA)

We have developed the LCA using as variables the factors calculated in the previous step of the EFA related to e-lifestyles (C. Yu, 2011). The first step of the LCA is to identify the optimal number of segments. This was calculated for an interval between one and nine clusters. We used the Bayesian Information Criterion (BIC) indicator as a criterion for the choice (see Table 1). This is one of the most widely used indicators due to its consistency and parsimony (Neath & Cavanaugh, 2012). The literature on latent class analysis (LCA) (Nylund et al., 2007; Vermunt & Magidson, 2005) considers the BIC to be the preferred initial criterion for determining the optimal number of clusters, in comparison with alternative model selection criteria. As a result, an eight-cluster solution was chosen. The other indicators (LL, AIC and AIC3) offered even larger numbers of clusters. These indicators should be interpreted as complementary diagnostics rather than as decisive criteria. In this respect, entropy and R² act as robustness diagnostics in LCA. High entropy indicates stable and unambiguous class assignment, while R² values show that the latent classes meaningfully explain variation in key indicators. In our case, entropy reaches a value of 0.87 and R² attains 0.83, both close to 1 and well above the commonly accepted threshold values of 0.6 and 0.1, respectively, thus supporting the robustness of the selected solution. Together, they confirm that the selected solution is statistically and substantively robust. Considering the number of clusters, their sizes and the data, the eight-cluster option seems appropriate (Villarejo-Ramos et al., 2019).

In Table 2, we find the profiles of each cluster, with respect to the variables that have been used for their formation. We have used the Wald test to assess the statistical significance of model parameters, confirming that the observed differences between latent classes were not due to random variation (Vermunt & Magidson, 2005). The results show that the eight factors that make up the concept of e-lifestyle (Internet Needs for Work, Internet Needs for Daily Actions, Interest, Leisure, Social, Importance, Barriers, and Novelty) are significant, as shown by the p-value of the Wald statistic (less than 0.05). Therefore, each indicator contributes significantly to the ability to discriminate between clusters. The R² measure of each indicator, defined as the ratio of the between-class variance to the total variance of the variable analysed, shows how much variance of each indicator explains the 8-cluster model (Vermunt & Magidson, 2005). This R² should not be interpreted as a coefficient of determination in a causal sense, but rather as an indicator of explanatory relevance and class discrimination. That is, it reflects the extent to which the variables included are informative for constructing the resulting clusters. It should be noted that R² in LCA is a descriptive measure of discriminatory power and does not imply causal importance.

On the other hand, for a better characterisation of the clusters we have used a set of covariates. Though these covariates are not part of the model for the formation of the clusters, they help us to better understand the characteristics of the individuals that make up each cluster. To this end, we have taken as covariates personal and demographic indicators of the individuals (Table 3), and also indicators related to the use of medical teleconsultation (Table 4). As before, we applied a Wald test to look for statistical differences between the eight clusters.

Table 3 shows that, from a socio-demographic point of view, there are significant differences between clusters that help us to characterise them. These are in the level of education and age of the individuals, as well as in the income of the families. And from a psychographic point of view, we also found differences between clusters with respect to the technological anxiety shown by individuals. On the other hand, we did not find that gender, place of residence, family size or the number of ascendants or descendants help us to characterise the segments. Overall, these results support hypothesis H1.

In addition, we incorporated a set of covariates related to the use and adoption of teleconsultation (Table 4). As a result, we found differences between clusters with respect to having the technological means, facilitating conditions, use and intention to use teleconsultation. In contrast, we found no differences in perceptions of usefulness, ease of use, hedonic motivation, habit, or social influence. Regarding the means of teleconsultation, we found significant differences in the case of telephone consultation, and we noted no differences in other means such as video calls, chats, or others. Finally, the health status perceived by the individuals did not help us to distinguish one group from another. We also found no relationship between the segments and the type of health insurance of the individuals. Overall, our results support hypothesis H2.

Table 4 presents the frequency with which each identified profile utilises different teleconsultation channels. The interaction mode of the channels can be categorised into two groups: synchronous channels such as telephone, videoconferences, and chats, and asynchronous channels such as emails and WhatsApps. Telephone consultation (a synchronous channel) is the only channel where statistically significant differences between profiles are found (Wald = 18.40, p = 0.01). This suggests that e-lifestyle mainly influences real-time voice interactions, not the use of video, chats, or asynchronous channels. The findings are consistent with the evidence that suggests telephone calls are the most commonly used and accessible method of synchronous teleconsultation, especially among older or less digitally active users (Morcillo Serra & Aroca Tanarro, 2022).

In summary, our results support the two proposed hypotheses. We have identified eight profiles (segments) of individuals based on their Internet-related lifestyle, e-lifestyle (Yu, 2011, 2015). There are three profiles characterised by a pro-Internet e-lifestyle. We have called them "Digital Convinced" (cluster 7), "Social Internauts" (cluster 6) and "Internet Interested" (cluster 8). Together these three profiles account for just over 16% of the sample. Digital Convinced users (cluster 7) are characterised by the highest scores across nearly all the variables used to construct the clusters, with the exception of perceived barriers, which are notably low (Table 2). In addition, this group exhibits high levels of facilitating conditions and resources, as well as a high level of use and intention to continue using medical teleconsultation services (Table 4). Social Internauts (cluster 6) display a profile similar to the previous segment.

The main difference lies in their pattern of Internet use, which is more strongly oriented towards everyday needs rather than work‑related purposes. In this regard, their high level of social use is particularly noteworthy (Table 2). This group also presents high levels of facilitating conditions, as well as high use and intention to continue using teleconsultation services (Table 4). Internet Interested users (cluster 8) likewise share similarities with the two segments described above. For this group, the Internet plays a particularly important role in meeting work‑related needs, although it is also widely used for everyday activities. Their profile is distinguished primarily by a high level of interest in Internet use (Table 2).

We have identified three other profiles with a more neutral behaviour towards e-lifestyle, which we have named: "Conventional Users" (cluster 2), "Utility Seekers" (cluster 4) and "Digital Apaths" (cluster 5). Together, these three profiles account for just over 37% of the sample. Conventional Users (cluster 2) represent one of the largest segments in terms of number of users. Overall, they exhibit a centred profile, with mean values close to zero across both the variables defining their e‑lifestyle (Table 2) and their adoption of medical teleconsultation services (Table 4). Utility Seekers (cluster 4) are characterised by a high level of Internet use for everyday needs and very limited use for work‑related purposes (Table 2). Although this group displays sufficient resources and facilitating conditions, their use of medical teleconsultation remains very low (Table 4). Similar to the previous segments, Digital Apaths (cluster 5) also present a centred behavioural profile, albeit with lower levels than those observed in other centred segments (Table 2). From the perspective of teleconsultation adoption (Table 4), this group is characterised by low facilitating conditions and a relatively higher average age (Table 3).

Finally, we find two large profiles with an anti-Internet lifestyle: "Digital Suspicious" (cluster 1) and "Digital Renegades" (cluster 3). Together they account for more than 47% of the sample. Digital Suspicious users (cluster 1) constitute the largest segment. Their e‑lifestyle profile (Table 2) is characterised by below‑average values across all the variables, with the expected exception of perceived barriers, which are comparatively high. A similar pattern is observed with respect to their adoption of medical teleconsultation (Table 4), reflected in low facilitating conditions, limited use and low intention to use. By contrast, Digital Renegades (cluster 3) are characterised by the lowest scores on e‑lifestyle dimensions and the highest levels of perceived barriers to Internet access (Table 2). This group also shows the lowest facilitating conditions for medical teleconsultation adoption and the lowest intention to use these services (Table 4). A further explanation of each profile is provided in the discussion of the results.