Abstract
We examined the main factors affecting the intention of physicians to use teledermatology using a modified Technology Acceptance Model (TAM). The investigation was carried out during a teledermatology pilot study conducted in Spain. A total of 276 questionnaires were sent to physicians by email and 171 responded (62%). Cronbach's alpha was acceptably high for all constructs. Theoretical variables were well correlated with each other and with the dependent variable (Intention to Use). Logistic regression indicated that the original TAM model was good at predicting physicians' intention to use teledermatology and that the variables Perceived Usefulness and Perceived Ease of Use were both significant (odds ratios of 8.4 and 7.4, respectively). When other theoretical variables were added, the model was still significant and it also became more powerful. However, the only significant predictor in the modified model was Facilitators with an odds ratio of 9.9. Thus the TAM was good at predicting physicians' intention to use teledermatology. However, the most important variable was the perception of Facilitators to using the technology (e.g. infrastructure, training and support).
Introduction
Little is known about the factors affecting the adoption of teledermatology. To the best of our knowledge, there have been only two previous studies. In the first study, Al-Qirim investigated the factors affecting the adoption and diffusion of a teledermatology system in a publicly funded health organization serving the district of Waikato in New Zealand. 1 Roger's model of diffusion of innovation 2 was complemented with other dimensions borrowed from the literature and used as a framework for interviews with the stakeholders involved in the adoption of telermatology in the health-care organization. Adoption of teledermatology was influenced by the following factors: clinical and administrative applications, clinical practice, user involvement and acceptance, and the presence of product champions and clinical leaders. The main determinants of the diffusion were the economic benefits and the effectiveness of teledermatology as a diagnostic tool.
In the second study, Stronge and colleagues evaluated the facilitators and barriers to the use of store-and-forward teledermatology in the US military. 3 Three user groups (case managers, primary care managers and dermatologists) were interviewed to investigate the human factors regarding the success or failure of a teledermatology system. System support, speed, personal benefits and increased education and experience were identified by all the user groups as facilitators, whereas usability problems and insufficient training were considered to be barriers.
We have examined the main factors affecting the intention of physicians to use teledermatology using a modified Technology Acceptance Model (TAM). The investigation was carried out during a teledermatology pilot study conducted in Spain. A store-and-forward teledermatology consult system was set up between the dermatology department of the Galdakao-Usansolo Hospital and the primary health centre of Landako. The pilot study lasted nearly 15 months and 225 patients with a total of 254 skin lesions were diagnosed.
Theoretical framework
The TAM was proposed by Davis in 1989 based on the Theory of Reasoned Action. 4,5 It was developed to understand user acceptance of information technology. In its original version, the TAM considered intention as the direct antecedent of behaviour, while attitude and social norms were regarded as the predictors of intention. 4 The TAM decomposes the attitudinal construct of previous models into two distinct factors, Perceived Ease of Use (PEU) and Perceived Usefulness (PU). In an effort to achieve a more parsimonious model, the attitudinal and normative components were dropped from the TAM model, leaving PEU and PU as the predictors of intention. 6 However, there is increasing evidence that the constructs significantly influence behavioural intention and should be retained in the model. 7
The TAM has been tested for the prediction of adoption behaviour for various technologies, including the adoption of telemedicine by health-care professionals. 8,9 The TAM has proved to be suitable for both genders, various age groups, most cultures and for individuals of all levels of information technology competency. 10 It can also predict technology acceptance in both obligatory and voluntary usage settings. 10 In addition, the reliability and the robustness of the TAM 10 as well as the validity of the model's constructs has been demonstrated. 11
Despite the aforementioned advantages, the TAM has some limitations. First, it does not consider the social environment in which the technology is introduced. Consequently, some authors have questioned its applicability to study the behaviour of health-care professionals. 8 Second, the TAM does not take into account the influence of external variables and barriers to technology acceptance. 10 External variables 12 as well as barriers to technology adoption 13 influence behavioural intention to accept the technology and cannot be ignored. Various efforts have been made to extend the TAM by either introducing variables from other theoretical models or by examining antecedents and moderators of PEU and PU. More recently, the Unified Theory of Acceptance and Use of Technology (UTAUT) has been proposed, based on the theoretical framework of behaviour adoption. 6 The UTAUT integrates four core determinants of intention and usage: performance expectancy, effort expectancy, social influence and facilitating conditions. These core determinants were therefore taken into account in our adapted model.
The proposed theoretical framework was adapted from Chau and Hu's model of telemedicine acceptance and comprised three dimensions: the individual context, the technological context and the organizational context. 14 The individual context encompassed the variables Attitude, which can be defined as the perception by an individual of the positive or negative consequences related to adopting the technology, and Compatibility, which refers to the degree of correspondence between an innovation and existing values, past experiences and needs of potential adopters. 2 Compatibility was added to the model following previous research on telemedicine adoption. 14 The second dimension of the model, the technological context, included the variables PU and PEU from the TAM, as well as Habit. Habit was proposed by Triandis in his Theory of Interpersonal Behaviour (TIB) and refers to behaviour that has become automatized. 15 As regards the organizational context, the variables Subjective Norm and Facilitators were also added to our theoretical model. Subjective Norm originates from the Theory of Reasoned Action 4,5 and assesses the extent to which individuals believe that people who are important to them will approve of their adopting a particular behaviour. The variable Facilitators (or facilitating conditions) originates from the TIB 15 and refers to the degree to which an individual believes that an organizational and technical infrastructure exists to support the use of the system. The adapted theoretical model is shown in Figure 1.
Modified TAM
Methods
A technology acceptance questionnaire was developed following an adaptation of the original TAM to our study. A panel of experts in technology assessment evaluated the face and content validity of the instrument. A total of 276 web-based questionnaires were sent via email to 49 dermatologists, 185 family physicians and 42 paediatricians. These were all physicians who participated in the pilot study. The study questionnaire (see online supplementary file) consisted of 33 items grouped into eight theoretical dimensions (see Table 1). Respondents were asked to rate items on a seven-point Likert scale ranging from strongly disagree to totally agree. An overall score was computed from the mean of all items related to each theoretical dimension. Information about age, gender, medical speciality, clinical centre in which the physician practices, number of years in clinical practice and highest educational grade obtained was also collected. The internal consistency of the instrument was measured by calculating Cronbach's alpha for each theoretical variable. The construct validity of the model was evaluated using inter-item correlation.
Items used for measuring theoretical dimensions
Statistical analysis
The distribution of all variables and the correlations between theoretical variables were investigated. Theoretical variables were not normally distributed due to the presence of a small number of outliers among the 50 centres participating in the pilot study. Centres where the mean score for Intention to Use was below five were categorized as outliers. This resulted in nine centres being categorized as outliers (data not shown).
Because of the non-normal distribution of the dependent variable (Intention to Use), a logistic regression analysis was carried out after dichotomising this variable. The values for this variable were 0 = low or moderate intention and 1 = high intention. The median was chosen as the cut-off point between the two groups. Thus, the respondents having a mean score for intention to use of 5.7 and higher were categorized as high intenders and those with a mean score for intention to use lower than 5.7 were categorized as low or moderate intenders.
The model comprised the TAM variables PU and PEU in a first block, followed by additional theoretical variables (Subjective Norm, Facilitators, Compatibility and Habit) in a second block, and the control variable for outlier centres. Other control variables (age, gender, medical speciality, clinical centre, number of years in clinical practice and highest grade obtained) were also tested. The analyses were performed using a standard package (SPSS version 15.0).
Results
Of the 276 physicians who received the questionnaire, 171 responded (a response rate of 62%). Nearly 58% of respondents were women (see Table 2). Most respondents (76%) were aged 40–60 years. Almost 74% of the respondents were general practitioners, 15% were dermatologists and the remainder were paediatricians. Participating physicians had on average 18 years in clinical practice. Nearly 5% of the physicians had, in addition to their medical degree, a master's degree and 6% had a PhD.
Demographic characteristics of the health professionals (n = 171)
Theoretical variables of the model
Cronbach's alpha was acceptably high (≥0.7) for all variables, see Table 3. The theoretical constructs were well correlated with each other and with the dependent variable (Intention to use). However, multicollinearity was present between the variables Attitude and PU (Spearman correlation of 0.90). This finding is not surprising since PU captures the same construct as the Attitude from the Theory of Reasoned Action. Attitude was thus dropped from the final model.
Variables of the model (mean scores on a scale from 1 to 7)
Logistic regression
The TAM model alone was good at predicting intention to use teledermatology (Nagelkerke R 2 = 0.71) and the variables PU and PEU were both significant (OR 8.4, 95% CI 3.4–21.0 and OR 7.4, 95% CI 2.9–19.0 respectively), see Table 4. This means that for every one unit increase in the PU score an 8.4 increase is expected in the odds of having a high intention of use, holding all other variables constant. Similarly, with every one unit increase in the PEU score, a 7.4 increase is expected in the odds of having a high intention to use teledermatology.
Results of the logistic regression: Original and Modified Technology Acceptance Model
When other theoretical variables were entered in the model, the model was still significant and a little more powerful (Nagelkerke R2 = 0.78). However, the PU and PEU variables from the TAM became non-significant (OR 2.5, 95% CI 0.8–7.9 and OR 2.5, 95% CI 0.7–8.9 respectively) and the only significant predictor was the variable Facilitators (OR 9.9, 95% CI 2.8–34.9). This means that the variable Facilitators accounted for most of the odds of having a high intention to use teledermatology. The effect of other variables, such as the PU and the PEU, was mediated through the perception of facilitators that ease the adoption of teledermatology. Moreover, the effect of belonging to an outlier centre was not significant (OR 2.2, 95% CI 0.4–2.0). This also indicates that the influence of the organizational context is probably also captured by the perception of facilitators to adopting teledermatology.
A discriminant analysis was carried out to identify whether the items in the Facilitators variable differed significantly between low to moderate intenders and high intenders. The results showed that the two groups differed significantly on the three items of the Facilitators variable. The three items measuring facilitators were: (1) ‘I think that my centre has the necessary infrastructure to support my use of teledermatology’; (2) ‘I would use teledermatology if I received adequate training’; and (3) ‘I would use teledermatology if I received technical assistance when I needed it’.
Discussion
Our findings suggest that the TAM is a good predictive model of physicians' intention to use teledermatology. However, the perception of facilitators is the most important variable to consider for increasing physicians' intention to use teledermatology. Although the majority of physicians perceived teledermatology as beneficial, a positive attitude was not sufficient for ensuring their adoption of the new technology. Furthermore, the results demonstrate that intention to use teledermatology differed significantly between centres, indicating that the organizational context is very important. Thus future strategies should focus on providing adequate training to physicians, ensuring that their organization has the necessary infrastructure to support teledermatology and making technical support available to users.
The fact that physicians from some centres tended to have a lower intention to use teledermatology than those in other centres suggests the importance of considering variables at the organizational level in future studies of adoption. A multilevel analysis, that combines individual determinants of intention to use telemedicine technologies and organizational characteristics that might influence them, would appear to be a promising approach.
Limitations to the study
The present study had some limitations. First, although the response rate was acceptable with respect to other similar studies, 14,16 it was not possible to document differences between respondents and non-respondents. However, the fact that some physicians with very low intention to use teledermatology participated in the study indicates that participants were not only those with a positive perception of the technology. A second limitation is that for convenience and time reasons, we used a questionnaire adapted from previous work that was only face- and content-validated by experts. More validation and test-retest reliability should be performed in future studies. Finally, a third limitation is the fact that the results are only valid for the specific (experimental) teledermatology service under study. Thus, it would be interesting to replicate this study in other settings with different teledermatology service organizations.
Conclusion
Teledermatology is often viewed as an efficient and cost-effective means for increasing access to high-quality dermatology services. However, health-care provider acceptance is a key factor in the implementation of teledermatology. The present study showed that physicians held positive perceptions about using teledermatology in their practice. However, resistance was also noted in some health-care organizations. Overall, the most important factor that seemed to influence physicians' acceptance of teledermatology was the perception of appropriate organizational infrastructure, training and support.
Footnotes
Acknowledgements
The study was carried out as part of the Quality Plan of the Spanish Health System. The project was funded by the Health Institute Carlos III (Spanish Ministry of Health) and the Department of Health and Consumer Affairs of the Basque Government. MPG holds a New Investigator career grant from the Canadian Institutes of Health Research (grant 200609MSH-167016-HAS-CFBA-111141). We thank Hugo Pollender for his help in preparing the modified TAM figure.