Adoption of Mobile Health Applications Based On UTAUT2: A Study of North-Indian States

: Mobile technologies are changing the way people across the globe are achieving their day-to-day tasks. Mobile health applications or mHealth apps are an example of these technologies that are used by healthcare professionals (HCPs) as well as users/patients alike. This study aims to analyze the adoption of mHealth apps which provide the facility of consulting online with an HCP among the population of the north-Indian states using the prior validated measurement scale of Unified Theory of Acceptance and Usage of Technology 2 (UTAUT2). Following convenience and snowball sampling, data was collected through an online survey using Google Forms. Using Smart-PLS 4, descriptive analysis and structural equation modeling were conducted. Among the five constructs and three moderators undertaken for the study, only three constructs, i.e., performance expectancy, social influence and price value were found to have a significant effect on the behavioral intention to adopt mHealth apps in users/patients.


INTRODUCTION
Revolutionizing the entire healthcare industry, mHealth has changed the way health information and services are delivered across the globe.Combined with an increase in the use of smartphone and mobile applications in turn, the world, including India, has witnessed a boom in the development and adoption of mHealth applications (apps) with the total mHealth apps market predicted to exceed 50 billion U.S. dollars in 2025 (Statista, 2020).
With the HCP-patient ratio standing at 1:834 (The Print, 2022), the lack of medical manpower was further highlighted during the COVID-19 pandemic.Owing to the pandemic, the governments across the world started motivating people to avoid visiting healthcare facilities for routine consultations or appointments that could be managed online by using mHealth applications or even government managened online portals.
With features ranging from record maintenance to medication reminders, from appointment booking to sample collection for testing, from online consultation or sharing of records with HCPs, these applications empower users to actively participate in managing their health, enhance patient outcomes, promote preventative care and remote monitoring so as to bridge the gaps in healthcare access by reaching both rural and urban population, surpassing geographical barriers.However, even with such an incredible USP, users face multiple issues in adopting these apps over the traditional system of consultations.
Numerous studies have been conducted across different developing countries to study the factors influencing the adoption of these apps.However, limited research has been conducted in the context of the Indian population.This paper aims to study the determinants of behavioral intention to adopt mHealth apps among the population of India and record the effects, if any, of age, gender and experience on the same.

RESEARCH/THEORETICAL BACKGROUND
The adoption and perception of mHealth applications is an area of research that has received considerable attention by researchers among developed as well as developing nations across the world, moreso after the COVID-19 pandemic.Most of these studies have followed a theoretical framework to determine factors influencing user intention, like, Diffusion of Innovations Theory, TAM (Technology Acceptance Model), UTAUT ((Unified Theory of Acceptance and Use of Technology), TRP (Theory of Planned Behavior), etc.The determinants of behavioral intention have been studied for both the user groups of these apps, which include Healthcare Professionals (HCPs) and patients/ users by a number of researchers.A comparative review of some studies that have used UTAUT (Venkatesh et al., 2003) and its extension UTAUT2 ((Unified Theory of Acceptance and Use of Technology 2) (Venkatesh et al., 2012) to study the determinants of behavioral intention for users/patients is presented in Table 1.It is evident from Table 1 that the researchers have exclusively used Partial Least Square-Structural Equation Modeling (PLS-SEM) based on the quantitative approach by using Smart-PLS software, with respondents of the survey being young as well as the elderly across different studies.A thing to note here is that not all these studies have surveyed only the users of mHealth apps/services.Table 2 provides a comparative analysis of the significant and non-significant variables of the studies included in Table 1.The variation between the results of the studies reviewed in Table 1 exists mainly due to the difference in the specific characteristics of the study as well as due to the additional variables incorporated by them (Duarte and Pinho, 2019).Moreover, it can be deduced from Table 2 that none of the variables consistently have a significant or non-significant impact on the adoption of mHealth apps, which further indicates that among the variables of UTAUT, none are neither necessary nor sufficient in and of themselves to have an impact on the adoption of mobile health, as is also confirmed by Duarte and Pinho (2019) in their study.

METHODOLOGY: Measurement scale
To assess the behavioral intention to use mHealth apps, the questionnaire was set using a prior-validated scale developed by Venkatesh et al (2012), also referred to as UTAUT2.However, for the purpose of this study, among the eight total variables of UTAUT2, habit and hedonic motivation were not included in the proposed framework.Since the mHealth apps under study are used to consult with an HCP, it can be safely concluded that no user uses such apps for the purpose of enjoyment.Moreover, one cannot evaluate the development of a habit towards a specific app in a cross-sectional study (Ashraf et al., 2015).Price value has been included since these apps come with a subscription fee for both the user groups and an individual HCP consultation fee as well for the users, the effect of which on behavioral intention can be studied.The multi-item questionnaire statements (19 items), grouped into 6 factors were slightly modified to fit the mobile health context with the respondents rating their responses on a 7-point Likert scale (1 = strongly disagree; 7 = strongly agree).A demographic profiling section was also added with 4 questions.

Data collection
Since it was not possible to obtain the list of all the users registered on mHealth apps, the survey was uploaded on Google Forms and the link of the same was then shared and distributed through convenience sampling, followed by snowball sampling to the residents of north-Indian states.Except for two questionnaire items, one each from Facilitating conditions and Price value, all factor loadings exceeded the 0.70 thresholds in the pilot study of 40 respondents.These two items were removed from the questionnaire and the questionnaire link was shared again with the appropriate sample population across India.A total of 278 responses were collected, out of which 63 did not have any prior experience using a mHealth app which rendered their responses hypothetical, 12 had no experience using smartphones and 3 respondents had left at least 1 questionnaire item unanswered.The final sample included a total of 200 respondents, out of which 63% were females and 60% users were post-graduate or higher degree holders.
The demographic profile also shows 54.5% of the users are between the ages of 20-29 years which suggests that the younger generation is more inclined to the use of technology with 81.5% having the experience of using smartphones for more than 5 years.

DATA ANALYSIS AND RESULTS
The descriptive analysis was conducted through Google Sheets, followed by PLS-SEM analysis using Smart-PLS 4.0, which aligns with the reviewed studies (Table 1).The analysis included assessing the measurement model and the structural model.
Assessment of the measurement model includes determining the validity and reliability of the measures included in the questionnaire.Table 3 depicts the factor loadings (all.0.7), the reliability indicators, i.e., the Cronbach's alpha and the composite reliability (all >0.The HTMT (Heterotrait-Monotrait) Ratio of Correlations and the Fornell and Larcker criterion were applied to further assess the discriminant validity (Hair et al., 2011).For the HTMT ratio to show that the constructs measure distinct concepts and are discriminant, the values obtained must be less than 0.90 (Henseler et al., 2015).Furthermore, in the Fornell and Larcker criterion, the square root of the AVE of each construct should be higher than its correlation with any other construct.Pertaining to this analysis, both the criteria to ascertain discriminant validity were met, as is depicted in Table 4, thus, ensuring all the constructs are valid and reliable.Additionally, the cross-loadings also confirmed the discriminant validity of constructs since the results showed that factor loadings of each construct were higher than other construct loadings (Table 5) (Hair et al., 2011).Once the measurement model was assessed for reliability and validity measures, the structural model was analyzed.This was done through bootstrapping using a re-sample of 5000 (Hair et al., 2011).Path coefficients, t statistics and significance levels are depicted in Table 6.This study follows the UTAUT2 model to determine the behavioral intention of users to adopt mHealth apps for consulting with a healthcare professional.Results of Table 6 indicate that performance expectancy, social influence and price value have a significant impact on the user intention to adopt mHealth apps with path coefficients greater than 0.10 and p < 0.05, approving H1, H3 and H5.However, for effort expectancy and facilitating conditions neither of the two criteria are met, with both having path coefficients less than 0.10 and p values > 0.05.Thus, H2 and H4 are not supported.In addition to these variables, the hypotheses pertaining to the three moderators of UTAUT2, i.e., H6, H7 and H8 based on age, gender and experience respectively, are found to have insignificant moderating effects on behavioral intention.

CONCLUSION AND IMPLICATIONS
This study aimed at determining the factors that affect the behavioral intention of a user towards using mHealth applications.Going by the R2 value of 0.644, it can be concluded that the model moderately explains the variance in measuring behavioral intention of users to adopt mHealth apps (Hair et al., 2011).As per Table 6, social influence (H3) followed by price value (H5) and performance expectancy (H1) are the significant determinants of behavioral intention based on their path coefficient values.
These results imply that while choosing to adopt these applications, users are often influenced by the opinions of the people in their social circles, getting their money's worth through online consultation as well as the perceived advantages of using such systems over the traditional ones in their day-to-day lives.
In order to promote the adoption of these apps more and more, developers can focus on issues like data security, record maintenance, appointments and/or medicine reminders, user-friendly interface and inclusion of interactive features within the apps.Additionally, users will be attracted towards such platforms if they are able to access them on their current devices and do not face issues in figuring out how to navigate through an app, thus minimizing effort and facilitating the usage of the same.
Taking into consideration the results of moderation analysis, based on age, gender and experience, all of which were insignificant, it can be further concluded that with regards to the Indian population, these moderators have no effect on the behavioral intention to adopt mHealth apps.This can be explained due to the growing familiarity and access to smartphones among people in India, irrespective of age and gender which was estimated to reach over 1 billion in 2023 (Statista, 2023) and also, to the usage of mobile apps in general.

LIMITATIONS AND FUTURE RESEARCH
This study has a number of limitations.The study only analyzed the effect of the constructs of UTAUT2 on behavioral intention.However, the effect of behavioral intention on actual usage was not studied.
Researchers can also follow the same model for a longitudinal study to examine and compare results across different time periods.Future research can also focus on and build upon the existing factors/constructs of UTAUT2 by including more constructs that can affect behavioral intention to adopt these apps.This study has also focused on mHealth apps that provide the facility of consulting online with an HCP through audio/video chat and text messaging.Other categories of mHealth apps like mHealth apps based on privately owned firms and mHealth apps of established hospitals like Apollo available on the app stores can also be studied and compared in terms of adoption among people.Additionally, researchers can also gather similar data from different countries for a comparative and thus, more constructive analysis.

7 )
and the convergent validity indicator, i.e., the average variance extracted values (all >0.5) (Hair et al., 2011).The findings of the measurement model indicate that the construct reliability and convergent validity is confirmed since all the values are above the threshold mentioned by Hair et al. (2011).

Table
. Review of Studies Based on UTAUT and UTAUT2

Table 3 . Measurement Model
Note: The bold and italic values show the square root of AVE for the Fornell and Larcker criterion.