Diabetes mellitus (DM) is one of the most challenging diseases in the 21st century. It is the sixth leading cause of death globally [1] and continues to increase in prevalence, with macro-and microvascular complications resulting in increased disability and huge health care costs [2].

In Portugal, there were 591-699 new cases of diabetes per 100,000 inhabitants in 2015, representing an expense of 0.7%-0.9% of the Portuguese gross domestic product, and 8%-10% of the total spending on health [3]. The country has one of the highest age-adjusted prevalence of diagnosed type 1 or 2 diabetes in the population aged 20 to 79 years in Europe (9.8%) [4].

Telemedicine includes remote patient monitoring using devices (eg, mobile apps) to remotely collect and send data to health care providers, asynchronous interactions to transmit diagnostic images, vital signs, or video clips, along with patient data for later review, and synchronous live videoconferencing consultation among patients and physicians (eg, teleconsultations) or among physicians and specialist health services [5]. Information and communication technology (ICT) has been increasingly implemented in the care of people with DM to improve patient outcomes in areas such as blood glucose management, diet, medication, and exercise monitoring [6]. Although (remote) teleconsultations at home (TH) have been found to be more effective in inducing HbA_1c reduction when compared with other telemedicine services, such as remote telemonitoring, tele-education, and telecase management, they have been much less adopted than other forms of telemedicine in type 2 DM care [7]. Real-world data show that, before the COVID-19, teleconsultation appointments as a proportion of clinical activity ranged from 2% among a diabetic cohort to 22% among postoperative patients with hepatobiliary cancer [8,9]. In Spain, teleconsultations were used by only 7 (6.9%) of the 102 that used telemedicine in a sample of 1063 patients with type 2 DM, but obtained the highest rate of satisfaction [10]. Studies on teleconsultation in DM are scarce (eg, [11-14]). Given the use level and care potential of synchronous TH, this study investigated the necessary conditions to encourage their use.

Studies on physicians’ (eg, [15]) or patients’ perceptions (eg, [6,16]) about telemedicine have been performed independently of each other. As both groups are essential to the use of these services, this study surveyed the perceptions of the two using an identical data collection instrument and compared the results of the analysis of their responses. Identical questionnaires for both patients and physicians were used because, according to the literature [17], the factors affecting their willingness to adopt teleconsultations were the same. A canonical correlation analysis (CCA) was performed to find associations among a set of predictor constructs derived from the technology adoption literature and the intention of use (IoU) of TH by patients with DM and their physicians. CCA is a multivariate statistical technique used to study the interrelationships among sets of multiple dependent and independent variables [18]. It is an appropriate and powerful multivariate technique to identify the underlying independent relationship between the 2 sets of variables of the studied model because of the high number of variables in each construct.

In summary, this study assesses the perspective of patients with DM and physicians regarding the drivers and barriers inherent to programs that involve patients with DM teleconsulting with their physicians from their homes.

Data were collected from patients with type 1 or 2 diabetes or their caregivers, in case of child patients (75 valid responses) and physicians (68 valid responses) selected by rational choice and snowball sampling (as highly specific populations were at stake) from the north of Portugal during the fourth quarter of 2018. Concerning the patients, 51 questionnaire answers (51/75, 68% of total valid answers) were collected in-person and in paper at primary care centers belonging to the Group of Primary Care Centres of Braga, an organization that coordinates 22 primary care centers; the other were collected on web through DM patients’ associations. For physicians, the answers were collected on web with the collaboration of the same group of primary care centers. This organization sent an email with a link to the questionnaire to their physicians.

The perceptions of both groups were measured using 2 identical questionnaires based on a 7-point concordance Likert scale and a 5-point confidence Likert scale (Multimedia Appendix 2).

CCA was used to analyze the correlation between the set of dependent variables (IoU construct) and the set of predictor constructs. This method is useful when variables have multiple causes and effects, similar to the complex reality of human behavior and cognition. The computations were performed using SPSS (version 24.0; IBM Corp).

The average of the observed values is often used to form the constructs with consequent smoothing of the responses, which can lead to constructs that do not contain the variability expressed in the measurement indicators. CCA examines the relationship between the 2 observed variable sets without having this disadvantage.

Variables with a canonical correlation of 0.45 or above were considered in the final CCA model. The reliability statistics measured by Cronbach α for each construct scale were very good for expected performance (.87 for physicians and .83 for patients), facilitating conditions (.77 for physicians and .74 for patients), attitude (.83 for physicians and .82 for patients), and IoU (.91 for physicians and .78 for patients), and acceptable for expected effort (.60 for physicians and .56 for patients), and social influence (.62 for physicians and .56 for patients).

Survey responses of 75 patients (Table 3 and Multimedia Appendix 2) aged 10-86 years (mean 51, SD 17.1) were obtained. Of the 75 respondents, 6 (8%) had never used computers, smartphones, or tablets (Table 3). The average age of these 6 patients was 73 (SD 8.9, range 61-84) years. On average, patients had 3.1 DM consultations per year (range: 1-12). According to the respondents, DM consultations took 133 minutes on average (including travel and waiting). Of the 75 respondents, 33 (44%) patients or caregivers felt very or extremely confident and 12 (16%), moderately confident using computers or the internet. Moreover, 40% (30/75) were very or extremely confident and 24% (18/75) were moderately confident in the use of real-time video call apps. More than half had heard about telemedicine (43/75, 57% of patients) and 30% (23/75) about teleconsultations, but only 2 had participated in one in real time, and 37% (28/75) had never used an app to make a real-time video call.

In total, 68 valid responses from physicians aged 25-63 years (47/68, 69% of physicians in the interval 26-35 years) were received. Of the 68 respondents, 46 (67%) performed between 10 and 40 consultations per month, with an average duration of 23 minutes. Only 6 (9%) out of 68 physicians had never used a video call app. Moreover, 81% (55/68) felt very or extremely confident and 19% (13/68) moderately confident using computers or the internet, and 54% (37/68) were very or extremely confident. Furthermore, 28% (19/68) were moderately confident in the use of real-time video call apps. Of the 68 respondents, 33 (48.5%) physicians had never heard of TH, and 8 (12%) had already carried out synchronous teleconsultations. Although 56% (38/68) of physicians stated that they intended to use TH in follow-up consultations, 34% (25/68) answered that they would not use TH because they did not consider them a good method for health provision.

The distribution of concordance scores showed a significant variability in both groups. Wilcoxon Mann-Whitney tests identified differences between physicians and patients’ responses. Physicians had higher confidence in ICT use, but they also had higher scores for item E₃—Will only use TH if easy to learn and S₁—if there was technical assistance. Patients had (1) in general, a more favorable attitude toward TH, and higher scores in the perception that (2) TH can invade their privacy (F₃), but (3) be faster (P₁), (4) the medical problem can be correctly understood (E₁) in a TH, and (5) will have TH whenever the counterpart wants to (S₃).

Both patients and physicians considered follow-up to be the best purpose for TH (Figure 2).

The main contribution of this study is the identification of relationships among a set of construct predictors taken from the literature and the intention to use (synchronous) home teleconsultations. Obtaining insights about home teleconsultations from the stakeholders directly involved in the health care interaction—that is, patients and physicians, is particularly important in a time of great pressure for health systems and when an important portion of health care has to be assured at a distance.

TH appear to be safe and effective in appropriate clinical situations [9]. In addition, it should not be forgotten that more vulnerable fringes of the population would not have the resources needed for this type of consultation; for example, in our sample, some older patients had never used computers, smartphones, or tablets. Physicians and patients will likely be supportive of their use if they are offered as supplemental and in support of traditional care models rather than to replace them (most of the 2 samples agreed with this type of teleconsultation; Multimedia Appendix 2, variable A₅). This result is in line with the findings of Gilbert et al [37], from the patient’s perspective, and Greenhalgh et al [9], from that of the physicians.

Health illiteracy and the physical examination ritual (referred by Haig-Ferguson et al [36] in the context of a pediatric chronic fatigue service) may explain why patients see TH only as an extra health care service. Patients should be encouraged and supported for their use.

Expected performance factors (time savings, increased productivity or efficiency, better disease management, health improvement, and quality of the clinical examination) were the most important factors for intention to use among physicians, which is in line with the literature (Table 5). On the contrary, except for time savings, patients’ perceptions did not reveal an association between performance variables and intention to use.

Another difference concerns the expected effort needed to use TH. For the patients, explaining and being understood when communicating their medical problems using the computer and the technology being easy to learn are positively associated with intention to use. This type of concern has been identified in the literature [8]. For physicians, these factors are not related to the intention to use, except for the necessary effort to understand the patient’s problem. In their systematic literature review on physicians’ eHealth adoption, Granja et al [44] concluded that the major facilitators of eHealth are the quality of the diagnosis and patient-centered care.

For both groups, the only social influence variable associated with intention to use was the willingness of patients or physicians to participate in TH if their counterpart wants to, which means that each group can encourage the use of the other. The literature has only referred to the importance of physicians’ recommendations for teleconsultation [10].

For both patients and physicians, all facilitating conditions and attitude variables toward using TH were associated with intention to use, which is in accordance with the literature (Tables 4 and 5).

Curiously, contrary to the evidence described in the literature (eg, [53]), this study did not find a statistically significant association between demographic data or confidence in ICT use and attitude for any of the groups, with the exception of confidence in using video calls for patients. The current COVID-19 pandemic has led many people to communicate through videoconferencing. Given that our results point to a positive association between confidence in the use of video calls and attitude, the pandemic situation may have been a booster for patients’ TH adherence. For teleconsultations to be an effective addition to health services beyond COVID-19 they should be considered not only as a technological issue, but also as a complex organizational change problem [54]; from the aspects raised by the authors, we would like to highlight the need for adjusted legal frameworks and reimbursement schemes.

Major barriers to TH use identified were: (1) the inability to correctly understand the medical problem, (2) threats to patient privacy, (3) health data confidentiality, (4) unpleasantness of TH to provide or receive health care, and (5) type of TH use (supplemental care service). On the basis of the perceptions of patients, costs do not seem to be a barrier to TH use, contrary to what has been described in the literature [8]. Probably, this result was observed because, nowadays, technological devices that can be used to make teleconsultations (smartphones, tablets, laptops, etc) are easily available in most of the situations. As digital interaction generally has insignificant costs for patients, cost is highly dependent on the existence of the technology.

The major identified TH drivers were (1) the perception that they facilitate contact, and (2) the fact that the use by each group was highly influenced by the other. Furthermore, physicians are very sensitive to issues related to the performance and quality of service.

The sampling methods limit the generalizability of the results. The composition of the patients’ sample in terms of age and education was similar to that of the general population in northern Portugal. However, the proportion of patients with type 1 (type 2) DM in the sample is higher (lower) than expected in the population [4]. Thus, the patients’ population may be, on average, older and less educated than the sample in this study. As the data concerning the perceptions of physicians were collected on web, the sample may be, on average, more technology favorable than the population. Nevertheless, both samples included individuals ranging from more positive to more skeptic about TH. In addition, the results were compared with and discussed against the findings of related studies.

A CCA revealed a strong association between the predictors and the set of dependent variables, in line with the literature. The data analysis included a joint critical comparison of the perceptions of patients and physicians. To promote the use of home teleconsultations for DM, decision makers should: (1) improve patient health literacy, as the inability to explain medical problems correctly emerges as a barrier to teleconsultation use; (2) explore ICT developments to reduce current limitations of non–face-to-face examination; (3) ensure patient privacy and data confidentiality; and (4) demonstrate the capabilities of home teleconsultations to physicians, namely, in terms of the ability to enhance patient–physician communication and to educate patients and their caregivers toward a better management of the disease.

In the future, it would be interesting that research about teleconsultations acceptance incorporated sustainability related aspects like, for example, fuel consumption, carbon emissions, and loss of work productivity. A recent review [55] concluded that, as patients, health care organizations, and nations continue to look toward video consultations as an alternative, it is essential to continue to theorize in this domain.