The initial public reactions to smart glasses, notably Google Glass (Figure 1), were quite mixed. Google Glass was both praised for its innovative technologies, such as a see-through screen and hands-free use, and criticized for various reasons. Shortly after the introduction of Google Glass, cautionary stories about “Glassholes” emerged, a term coined to describe individuals who misuse the technology in socially inappropriate ways [1]. Examples of this phenomenon include people who wear Google Glass and ignore the people around them or stare off into space while operating their smart glasses. Google responded with rules to help users avoid these situations [2]. There are several examples of various emotional reactions to smart glasses in the media, and ethical questions are repeatedly raised about wearing smart glasses [3,4]. Although research into smart glasses is continuously expanding, current knowledge and theoretical foundations about the acceptance of smart glasses are still scarce.

Smart glasses have emerged as a growing research topic over the past years in various fields [5,6] and could bring about a revolution in health care [7]. Smart glasses are computerized head-mounted devices worn like eyeglasses, which can collect and provide information to the user and the environment. The majority of smart glasses incorporate a display through which information can be accessed while simultaneously maintaining visibility of the external environment. Furthermore, these devices often provide the advantage of hands-free operation [8]. Pilot studies on the use of smart glasses in health care have been conducted [7]; however, few researchers have addressed the issue of preadoption criteria, and identifying these criteria remains a critical issue [9].

The initial responses to smart glasses raise the question of how the adoption of smart glasses is progressing.

The growing body of research on the adoption of smart glasses was searched in the existing scientific literature. The following keywords were used: “smart glasses,” “head-mounted display*,” “wearable display*,” “augmented reality (smart) glasses,” “adoption,” “acceptance,” and “acceptability.” The search was performed in the following databases on June 12, 2021: Scopus (11 hits), Web of Science (12 hits), ACM Digital Library (96 hits), IEEE Xplore (2 hits), and PubMed (0 hits). All synonyms were connected with the disjunction “OR,” and both “smart glasses” and “adoption” were connected with the conjunction “AND.” Keywords were searched in the title. The identified publications are listed in Multimedia Appendix 1.

Various methodologies have been employed by researchers to investigate the adoption of smart glasses. For an overview of these methods, please refer to Multimedia Appendix 1. The context of the studies varied, with the general context most often consisting of random individuals or students. A small number of studies were found in the context of health care and tourism; therefore, it can be concluded that knowledge about the adoption of smart glasses is still scarce in these fields. Adoption studies with industry, retail, and agriculture respondents are even more limited. Various methods have been applied, with the majority of studies being qualitative in nature. Most of the previous quantitative studies had a relatively low number of respondents. The findings presented in Multimedia Appendix 1 offer foundational inputs for the development of a questionnaire. Furthermore, the social aspects surrounding smart glasses were mentioned multiple times in previous studies, and therefore, it could be assumed that social interaction is a relevant topic regarding the acceptance of smart glasses.

The Unified Theory of Acceptance and Use of Technology (UTAUT) [10], the Technology Acceptance Model (TAM) [11], and the Diffusion of Innovations [12] were found to be the most commonly accepted theories by scholars [13,14]. Some of the findings from prior research in Multimedia Appendix 1 might be related to the aforementioned models such as the influence on perceived usefulness, perceived ease of use, facilitating conditions, image, and perceived enjoyment [10,15]. However, other findings suggest themes regarding both the ethical and social contexts, which can be termed social acceptability [16]. Social acceptability can be divided into conversation partners, such as interpersonal communication, and spectators or third parties, with examples such as disturbance and controversial public use. The theme of privacy is also related to the social context and can be divided into the same categories: conversation partners and third parties. For example, wearing smart glasses is not only about the privacy of the user, but also about other people’s privacy, and this can strongly influence the user’s decision-making [17]. Figure 1 provides an illustrative example of an encounter with a health care professional wearing smart glasses. To summarize the findings of the prior research in Multimedia Appendix 1, it can be concluded that relevant aspects of the preadoption criteria should be informed by the current knowledge in adoption research, such as the TAM, and should also take into account the social context from the perspectives of both caregivers and patients.

Devices worn on the body in general influence the symbolic meaning in the natural environment of the person and the public surroundings and have psychological impacts on wearers and behavioral consequences for wearers [18]. Over the past decades, the TAM [19] had an incredible influence on empirical research. One might also say in general that current adoption models take a snapshot of adoption readiness, and the TAM consistently explains about 40% of the variance in the intention to use IT and the actual usage by individuals [15]. However, these adoption models have limitations [20], in which the current reality regarding smart glasses may not be fully involved. Current models are linear and strongly focused on the user’s perspective alone. Moreover, according to recent studies, social influence, as it is considered in the UTAUT [10], may involve more than just the influence and desire of others in the future use of the one considering it [3]. Several studies have suggested that social context influences usage norms and that public use is controversial [21,22]. The use of smart glasses does not reproduce existing norms and understandings in turn-taking, knowledge, and identity and could reconfigure norms by creating new settings in which activities begin to compete and interfere with each other [23,24]. In addition, privacy is often mentioned to be influential [25]. The privacy of the user is important, and other people’s privacy can also strongly influence the user’s decision-making [17]. Therefore, smart glasses force us to take a broader view of adoption models, and we should not only consider the perspectives of potential users but also look at the surroundings. Hence, social influences might be currently underestimated in the context of the acceptance of smart glasses given the reports from various media and recent publications on the subject.

Turning to a perspective that takes a broader view of users and their surroundings and environment [26], the mediation perspective fits well with the identified gaps regarding the acceptance of smart glasses. The mediation perspective of the philosophy of technology considers the relationship between humans and technology from a phenomenological perspective, and it inspired the beginning of ubiquitous computing [27]. Furthermore, Mann [28] explained the use of smart glasses after over 35 years as a mediated version of reality. Those mediated relationships are reciprocal, and technology is perceived as the mediator between humans and the world. For example, a user looks through glasses rather than at glasses, and this is therefore termed as an embodiment relationship. In addition, smart glasses can represent information on a parallel screen while seeing the world, which is termed as a hermeneutic relationship [29]. A combined perspective of both adoption and mediation could complement each other in understanding the appropriation of smart glasses (Figure 2; derived and adapted from Zuidhof et al [30]). However, knowledge of the mediated experience of smart glasses is still scarce, and recent studies have reported strong influences of the social context in decision-making [17,21]. The theoretical framework of adoption and mediation can reveal both the user’s perspective and the perspective of others who are related to the user, such as a patient.

In the field of the philosophy of technology, mediation is interpreted as determining human-technology relationships and interactions, and anticipating influences on human behavior [29]. Prior studies on the adoption of smart glasses, as illustrated in Multimedia Appendix 1, not only present familiar constructs but also have begun, since 2010, to delve into social themes from various perspectives, including those of the user, conversational partner, and third parties [26]. A large-scale survey in which the anticipated perspective of the health care professional and the anticipated social interaction are united is lacking in current knowledge. Based on previous research results and the theoretical framework of adoption and mediation (as part of a doctoral research project), it seems relevant to explore adoption themes and assess if social interaction influences the intention to use smart glasses.

Figure 2 proposes to adapt the theoretical framework of adoption and mediation to the context of the health care professional and the patient. It is considered that smart glasses for health care professionals are still in an early phase of dissemination, and therefore, it is needed to gather knowledge about anticipating the use of smart glasses. Moreover, health care professionals are expected to anticipate adoption while taking into account social interaction with patients, but they can also take into account the patient’s perspective in social interaction. Therefore, this goes beyond mere adoption for self-interest and may lead to a higher or lower intention to use smart glasses.

Prior research on the adoption of smart glasses has been mainly conducted in general contexts; however, some studies were carried out in the area of tourism or health care. Few of the recent qualitative studies on smart glasses in health care are part of the authors’ doctoral research projects. Furthermore, it appears that there are no large-scale surveys conducted on the adoption of smart glasses and that most of the recent studies have an explorative and qualitative nature. Prior research on the adoption of smart glasses is presented in Multimedia Appendix 1. In addition, the emotional and ethical aspects that can play roles as issues to adoption described above are still involved to a limited extent in recent studies and might be especially relevant in professions where people are vulnerable or interact intensively like in health care.

The objectives of this study were two-fold: (1) to develop an instrument that combines the perspectives of adoption and mediation to measure the intention to use smart glasses by health care professionals, and (2) to obtain insights into how the intention to use is influenced by adoption and social interaction aspects. Regarding the state of the art and the objectives of this study, the following research question was formulated: “To what extent do constructs derived from a mediation perspective contribute to the established constructs of adoption to explain the intention to use smart glasses?”

To achieve the aims of this study, a summated rating scale development approach for social sciences was used [40]. The initial steps were aimed at defining constructs and designing the scale. The resulting scale was evaluated through an online survey among a sample of students.

To develop an item pool, existing literature was consulted, and the state-of-the-art knowledge, as described in the Introduction, was combined with the theoretical framework of adoption and mediation. A core aspect of the framework (Figure 2) is the reciprocal relationship among the user, the technology, and the world. In the next step, the aspects from Figure 2 were substantiated with existing literature and translated into items. See Multimedia Appendix 2 for the item pool. All measurements were recorded using 5-point Likert scales. Back translation was performed by 2 researchers as a validation method for the survey [41] and to develop a robust item pool.

To check for comprehensibility, readability, and internal consistency, 2 pretest moments were organized. First, the questions were reviewed by 2 potential respondents. Next, data collection (as a pretest) was conducted at a health care institution in the Netherlands (n=17). Although the pretest was conducted on a small sample, internal consistency was acceptable. Initially, reversed items were included, but inconsistent response patterns were observed for those items. Respondents reported too much confusion with reversed items, and therefore, these items were removed from the questionnaire.

Respondents (N=450) were recruited in July 2022 on the internet using Amazon’s Mechanical Turk. The questionnaire was made available to respondents who met the following inclusion criteria: (1) language: English; (2) employment industry: health care; and (3) location: Northwest Europe (Austria, Belgium, Denmark, Finland, France, Germany, Iceland, Ireland, Luxembourg, Netherlands, Norway, Sweden, and Switzerland), Canada, United Kingdom, and United States. The questionnaire was closed once the desired number of respondents was reached.

A summary of the demographic information of the respondents is presented in Table 1. The sample pool consisted mainly of male participants, with over three-quarters being 42 years of age or younger. The majority of the respondents were from the United States. With regard to occupation, the participants were distributed as follows: physicians, 192 (42.6%); nurses, 165 (36.7%); and others, 93 (20.7%). Other professions included a diverse group of medical assistants (n=14), medical technical professionals (n=11), managers (n=10), health care professionals (n=9), IT professionals (n=6), and home care professionals (n=4), among other various professions in health care.

Before starting the survey, respondents provided active online consent after reading the opening statement (Multimedia Appendix 3). Furthermore, pictures were used with permission from health care professionals and patients. Respondents were rewarded with US $2 when they completed the survey. Study data were deidentified and anonymously used for analysis. This study was reviewed and approved by the Ethical Review Board of the University of Twente (approval number 211121).

After the opening statement and consent, respondents were asked to watch 2 short videos about smart glasses to get an idea of the technology and its use. The first video showed an augmented reality platform for health care with English-spoken animations [42], and the second video showed an example of Vuzix M400 smart glasses in the context of health care [43]. After watching both videos, respondents were asked to fill out the questionnaire. Demographic information was requested at the end of the questionnaire. Some photos used in the questionnaire are presented in Figure 4.

Before the analyses, data were checked for normality. No significant deviation from normality was found. All skewness and kurtosis values were below 1.96, and values between –2 and +2 are considered acceptable to prove normal distribution [44]. Structural Equation Modeling was performed with maximum likelihood estimation, as advocated by [45,46], to test the conceptual model of the intention to adopt smart glasses by health care professionals. To test if the conceptual model fits the data well, the next absolute and incremental model fit indices were used as suggested by Holbert and Stephensen [47] and Hoyle and Panter [48]. The most common absolute fit index is the χ² goodness-of-fit test, and to minimize the impact of sample size on the model chi-square, the relative/normed chi-square (χ^2/df) was used [49]. The standardized root mean square residual (SRMR) was reported as an absolute fit index. The following incremental fit indices have been reported: Tucker-Lewis index (TLI) and root mean square error of approximation (RMSEA). The practitioner recommendations for model evaluation were followed, suggesting a cutoff value close to 0.95 for the TLI in combination with a cutoff value close to 0.09 for the SRMR to evaluate model fit, and a cutoff value close to 0.06 for the RMSEA [50]. The Fornell and Larcker discriminant validity criterion was used to test discriminant validity. The Fornell and Larcker criterion is satisfied when a construct is more closely related to its own indicators than to other constructs [51].

The questionnaire on the anticipated adoption and mediation of smart glasses was designed to examine the perceptions of future users regarding smart glasses. It was also designed to get a better understanding of how specific situations in the use of smart glasses influence the intention to use them. The research question was: “To what extent do constructs derived from a mediation perspective contribute to the established constructs of adoption to explain the intention to use smart glasses?”

Participants were queried regarding their familiarity and experience with smart glasses. The constructs “knowledge” and “persuasion” served as constructs in assessing how health care professionals perceive the diffusion of smart glasses. Overall, the respondents reported having substantial knowledge about smart glasses (eg, I know what someone can do with smart glasses; mean score 4.02, SD 0.84) and expressed understanding of the concept of smart glasses (mean score 4.08, SD 0.79). Close to 75% of the respondents (n=339) indicated their willingness to try smart glasses if they were available to them (mean score 3.94, SD 0.93).

To investigate the added value of the observed variables in the conceptual model, a comparison was made between validated TAM2 [52] constructs used to create a baseline and the additional social interaction variables in this study. The constructs “perceived usefulness” and “perceived ease of use” were adapted to the context of smart glasses, and subjective norm and image were included because they represent social interaction. Although the construct “attitude” was not included in the TAM, there was a reason to include this validated construct based on previous studies on the early phase of the diffusion of smart glasses.

Considering the current state of the art and the objectives of this study, we formulated the following research question: “To what extent do constructs derived from a mediation perspective contribute to the established constructs of adoption to explain the intention to use smart glasses?” This study had 2 primary objectives. The first objective was to develop an instrument that combines the perspectives of adoption and mediation to measure the intention to use smart glasses by health care professionals. A conceptual model was created by drawing on previous findings related to smart glass adoption and testing hypotheses about their anticipated effects on attitudes and intentions to use. The instrument developed integrates both adoption and social interaction perspectives, highlighting smart glasses as a mediating factor between individuals, and it was validated with health care professionals. The approach for this instrument builds on prior research and represents a step forward in understanding the role of smart glasses in health care.

The second objective was to obtain insights into how the intention to use is influenced by adoption and social interaction aspects. The structural equation analysis showed that integrating social interaction factors with selected TAM2 constructs led to a higher explained variance in the intention to use smart glasses than TAM2 alone. Key factors influencing the intention to use included attitude, perceived usefulness, intention to use despite social nonacceptance, and passive use of smart glasses. Social presence had an unexpected effect: as social presence increased, attitudes toward smart glasses decreased. Despite social nonacceptance by patients, health care professionals still showed willingness to use smart glasses, suggesting that this factor may be less decisive in health care. Passive use was linked to perceived usefulness and intention to use, indicating that wearing smart glasses on top of the head (as presented in Figure 4) could be convenient to have smart glasses close by, even when not actively in use. These findings suggest that social interaction plays a nuanced role in adoption, and further research is recommended to explore these relationships, especially regarding social presence and passive use.

The second aim of this study was to gain insights into how the intention to use is influenced by adoption and social interaction aspects. The outcomes of the structural equation analysis revealed that our path model (Figure 6) integrating constructs from both TAM2 and social interaction variables demonstrated a higher explained variance in the anticipated intention to use smart glasses, compared to the isolated TAM2 constructs that were selected. This finding indicates that social interaction concerning smart glasses plays a role in anticipating the adoption of smart glasses in health care besides the perceived usefulness and attitude toward smart glasses. More specifically, the intention to use smart glasses is directly influenced by attitude, perceived usefulness, intention to use despite social nonacceptance, and passive use of smart glasses.

The presented model showed that attitude and intention to use both had high squared multiple correlations. This gives a reason to consider whether there is a theoretical difference between these constructs. In the Introduction section, it was also indicated that the attitude toward technology is a debatable construct in adoption studies because the added value in the model would not be substantial [10]. Original adoption models were developed for computer use in the 1980s, but smart glasses are different in several ways: they are worn on the body, are in the field of view, and often contain a camera. In this model, attitude is a mediating variable for subjective norm and the lack of social presence and was found to be important for the intention to use. Follow-up studies may investigate this relationship further.

Intention to use despite social nonacceptance shows that the influence of others is important for the health care professional but that it is not decisive to eventually use smart glasses. This can be understood because care actions can often be unpleasant for the patient but necessary for health purposes. This construct may score very differently in other disciplines where social relationships may be more important. Furthermore, the anticipated intention to use smart glasses is indirectly influenced by the lack of social presence and subjective norm.

A notable result of the lack of social presence is the decline in attitude. If the lack of social presence decreases (with an increase in social presence), the attitude toward smart glasses also decreases. This might indicate that respondents do not expect that smart glasses will ever provide more social presence and would not use them for that reason. One explanation that supports this finding is the positive regression coefficient with intention to use despite social nonacceptance by patients. This suggests that professionals would use smart glasses despite social nonacceptance by the patient.

The results of this study also indicate that there is a complex relationship between the passive use of smart glasses and 4 variables: a direct relationship with intention to use and an indirect relationship through perceived usefulness, attitude, and intention to use despite social nonacceptance. This can have 4 explanations, namely that passive use has a negative relationship with the intention to use, which indicates, for example, that health care professionals would rather not passively use smart glasses. However, this seems to contradict the finding of a weaker relationship with perceived usefulness, which could mean, for example, that it is convenient to have the glasses quickly at hand by wearing them on top of the head, just as a doctor places a stethoscope on the neck. Passive use also leads to an increased positive attitude toward glasses, which in turn leads to increased intention. In addition, there was a significant negative regression between passive use and intention to use despite social nonacceptance by the patient. This could indicate, for example, that the health care professional is willing not to use technology or, in this case, is willing to remove it if the smart glasses are not actively used and the patient is not comfortable with them.

The second objective was to gain insights into how the intention to use smart glasses is influenced by the aspects of both adoption and social interaction. Based on our findings, it can be asserted that the presented model provides a higher explained variance for the intention to use smart glasses. The majority of the hypotheses were confirmed. However, some results were unexpected, such as the influence of social presence on attitudes, as discussed previously. Contrary to expectations drawn from prior research [31,53], we found that a lack of social presence did not result in a negative attitude or diminished intention to use. Therefore, we recommend further exploration in subsequent research, potentially using methodologies such as in-depth interviews and quasiexperimental studies. The same applies to the construct of social acceptance during the passive use of smart glasses. The model indicated that this factor also impacts perceived usefulness, suggesting that passive use of smart glasses could be beneficial in a work context and could consequently enhance the intention to use. Further research should shed more light on the underlying reasons for this finding.

This study focused on the combination of adoption and social interaction. A limitation of this study lies in the comparison with the full TAM2 model. A priori–selected TAM2 constructs based on relevance and the aim of the study were included, namely to investigate social interaction and assess to what extent themes from the mediation perspective add something to current models. A comparison with the full TAM2 model is recommended for a follow-up study.

Second, in this study, social acceptance was examined through scenarios and not directly as a variable. The reason for this was to tie in with the previous findings and become more concrete in the specific context. This ignores social acceptance as an independent construct, and it is advisable to do so now that it appears that social interaction plays a substantial role in the model.

Third, as discussed in the conclusion, attitude and intention to use as variables score high on the explained variance, and it is questionable whether they are distinctive enough. Nevertheless, this difference is based on previous literature. Attitude was assumed as a preliminary state in the conceptual model, and it also fits the final model. It is recommended to investigate this further in a subsequent study, and it might be related to the last limitation.

The sample was limited to individuals who have access to MTurk. The distribution of respondents was mainly centered in the United States, and the questionnaire may have been completed mainly by people who were enthusiastic about the topic. As other researchers have addressed potential issues with sample representativeness and quality of the data [54,55], the results are more difficult to generalize and are not necessarily representative of the population of health care professionals. Therefore, it is important to revalidate the results with a reliable and representative dataset.

Furthermore, participants were presented with 2 videos to watch before proceeding further with the questionnaire. These videos, accessible to the public, were specifically chosen to provide a balanced depiction of what smart glasses are and their capabilities. It is worth noting, however, that as commercial videos, they may be designed to persuade viewers of the utility of smart glasses. The researchers selected these videos because they provided examples that were realistic and applicable in real-life work contexts at the time of the study and represented an attempt to prevent unrealistic imaging from smart glasses.

Although this model has been created based on previous literature and has remained as close as possible to the conceptual model, it is not a parsimonious model. From Cutting [56], a series of criteria can be derived to discuss the model evaluation, which is reflected by the accuracy, scope, and simplicity of the model. Parameter counting is widely used to indicate simplicity: if 2 models are both equally valid, the one with the fewest parameters is the best [56]. Several nonsignificant regression coefficients were found in the results of this study, and in combination with the previous recommendations, a subsequent simpler model can be developed.