Barriers and Facilitators of eHealth Adoption among Patients in Uganda – A Quantitative Study

The adoption of eHealth has not made great strides in Uganda especially among patients despite its potential in improving patient outcomes through access to care, patient engagement and its ability to reduce unnecessary hospital visits. Previous studies have focused on barriers and facilitators of eHealth in general. None has examined the adoption of eHealth among patients. Therefore, this study set out to investigate the barriers and facilitators of eHealth adoption among patients in Uganda. A cross-sectional survey was conducted in four districts across the country. A total of 292 patients of 18 years and above participated in the study and their selection was through simple random sampling. The bivariate analysis results revealed that education level (χ2 = 14.9, ρ<0.05), gender (χ2 = 4.95, ρ<0.05) and location (χ2 = 85.9, ρ<0.05) have a statistical significant relationship with eHealth adoption. The logistic regression model further revealed that male patients (OR=2.662), those with master’s degree and above (OR=2.2797) and those residing in Kampala (OR=.012) were more likely to use eHealth systems than their counterparts. The success of eHealth requires players in the health sector to ardently focus on the socio-demographic factors of the users, technological and hospital conditions if eHealth adoption is to ensue.


Introduction
Communities constantly face health-related issues yet healthcare is still a huge public health concern in developing countries. With majority of the population affected by all sorts of illnesses (communicable and non-communicable), coupled with accessibility challenges especially in the rural communities, the adoption of information and communication technologies (ICT's) has been seen as an alternative to realize efficiency and effectiveness in healthcare service provision [1]. ICT's in the health sector are generally termed as eHealth or digital health technologies. World Health Organization defines eHealth as the cost-effective and secure use of ICT in support of health and health-related fields, including health care services, health surveillance, health literature, health education, knowledge and research [2]. eHealth is an umbrella term that covers a wide range of health and care services delivered through information and communication technologies, such as electronic health records (EHRs), health information systems, remote monitoring and consultation services (e.g. telehealth, telemedicine, telecare), tools for self-management, and health data analytics [3]. eHealth tools [mobile and fixed phones, voice over internet protocol, text and multimedia messaging] encourage communication between healthcare providers and their clients, sharing of information and knowledge among healthcare providers and establishing of better healthcare for patients [4].
Despite all the tremendous investments in the ICT infrastructure by the Government of Uganda and the private sector to support eHealth [6], use of digital technologies is still very low especially among patients. Previous studies have focused on barriers and facilitators of eHealth in general. None has examined the adoption of eHealth among patients in Uganda, and to the best of our knowledge, this is the first of its kind. This paper will contribute to understanding the factors that influence the successful adoption of eHealth among patients for those seeking to implement patient-centered systems, and will be a pedestal in enhancing the national eHealth strategy of Uganda. Investigating barriers and facilitators for successful eHealth adoption among patients is vital for informing policy and relevant stakeholders investing in the sector.

Study setting
The study was conducted in central, southwestern, eastern and northern Uganda. Data was collected from health facilities located in; Kampala central division, Mbarara municipal council, Jinja central division and Mbale municipality. The study aimed at investigating the barriers and facilitators of ehealth adoption among patients in Uganda. The choice of these districts was because they rank in the top twenty largest districts by population size, have moderate levels of internet penetration and have a good mix of urban and peri-urban population [34], [35]. Inclusion criteria included i) recovering patients and outpatients above 18 years of age who sought medical services from national and regional referral hospitals, health centre II, III, IV, and clinics. Sixty-eight health facilities were visited.

Study Design
The study employed a cross-sectional design using a quantitative data collection approach covering a period of October 2020 -January 2021. A structured survey questionnaire formulated in English with three main themes (demographic data, barriers and facilitators) was used to collect data. Barriers and facilitators used in the questionnaire were drawn from existing studies [6], [19], [23], [28], [37]- [39], [21], [27], [29], [16], [22], [32], [30], [40], [41], [20], [33], [42], [10]. All authors designed the questionnaire, however it was specifically tailored to fit the scope of this study. A thorough scrutiny of the barriers and facilitators was done where patterns were identified and factors clustered into three major themes (hospital, technological and individual factors). To ensure consistency and clarity, two independent researchers validated the questionnaire. A pre-test was conducted in October 2020 with 20 outpatients at Corsu Rehabilitation hospital, Entebbe and Rubaga Hospital. The responses on the questionnaires were measured using a five point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). However, at the advent of the CoVID-19 pandemic, there was a mandatory requirement to observe the Ministry of Health CoVID-19 standard operating procedures especially when in public. Thus, data was collected in three different ways, i) using the Open Data Kit (ODK), ii) a google form, and iii) a physical questionnaire. Respondents were free to choose any one method of their convenience. For those that chose to fill a google form, a consent form was sent through email, and once filled, a google link was subsequently sent. For participants who opted for the ODK tool and physical questionnaires, the research team first sought written consent from them before collecting data.

Sampling and data collection
Three hundred and twenty patients from 68 health facilities located across the four districts received questionnaires, and 292 were successfully returned, contributing 91.2% to the response rate. The number of health facilities were determined using Yamane's formula of determining sample size [43]. Using this formula, 1+ ( ) 2 where n is the sample size, N is the population size and e is the level of precision (assumed to be 10% for this study), we were able to determine our sample (68 health facilities) using a population of 209 health facilities. After determining our sample size, we then purposively selected the health facilities. At the health facilities, study participants were randomly selected using simple random sampling and this exercise lasted approximately four months. Study participants were recruited from the eye and dental clinics, cancer and heart institute, maternity and orthopedic wards, and others were selected while entering or exiting the hospital gates. For all selected participants (patients), telephone contacts were exchanged and follow-up was done to ensure that the questionnaire was filled and appointment to have it picked set.
Ten research assistants (RA's) together with the authors participated in the distribution of the questionnaires. All RA's were graduate students who, despite their experience in data collection had to first be trained on the primary objective of the research, research ethics, code of conduct and communication skills.

Data Analysis
All data was coded, processed and analyzed using IBM SPSS Statistics Version 21 (New York, USA). Phase one of the analysis started with a descriptive bivariate analysis to understand the demographic composition of the study participants and provide basic information about the dataset. This was achieved using the mean and standard deviation, to ascertain how spread out the responses were; Pearson's Chi-square test (χ 2 ) to test the independence between different variables; and cross tabulation to summarize the relationships between different variables. Phase two involved using multivariate analysis achieved through using logistic regression because of its ability to estimate the probability that a patient will use eHealth systems and determining which socio-demographic or socio-economic factor significantly influences eHealth adoption among patients. The logistic regression model used "eHealth system use" as the dependent variable. The independent variables were education level, age, gender, type of patient, location, employment status and type of health facility.

Ethical approval.
A multi-layered approval process was adopted in this research. First, approval was sought from the ethical review committee of school of public health, Makerere University, which was followed by Uganda National Council of Science and Technology under registration number SS945ES. Subsequent approval was sought from health facilities and from study participants. Consent was both verbal and written, however respondents were first asked for their verbal approval, which was later accompanied by a written one. All human subjects consented and were informed of their rights to withdraw at any point of the study.    The need for fast execution of processes will motivate users to use eHealth systems 3.77 1.115

Results
If the users of the system trust the service, they will be obliged to use it 3.59 .980 In my opinion, if the system facilitates research and development, adoption will increase 3.66 .975 The lack of ownership by users bars adoption of eHealth systems 3.70 .839 The lack of developer support affects eHealth adoption Consequently, the participant's opinions on the individual barriers of eHealth were strongly inclined to the lack of acceptance among users (µ=3.92±1.123), the technophobia (µ=3.90±2.812) and the digital illiteracy (µ=3.90±.855).

Logistic Regression Analysis
The odds of whether or not to use eHealth technologies is computed using the formulae, Where Ŷ is the predicted probability coded with 1 (use of eHealth technologies) rather than 0 (do not use eHealth technologies), 1-Ŷ is the predicted probability of the other decision, and X is the predictor variable, gender. Given the results in table 4 (A), 76% did not use eHealth systems.
The intercept-only model is,   The variables in the equation output in Table 8 shows that the regression equation is; ln( ) = −1.537 + .621 This model predicted the odds that a subject of a given gender will use eHealth technologies to access health services. The odds prediction equation is; If the patient is a woman (gender=0), then they are only 0.21 as likely to use eHealth technologies as she is not to use them. If a patient is a man (gender=1), they are only 0.4 times more likely to use eHealth technologies than not to.
Converting the odds to probabilities, using the formulae below, the model predicts that 17% of women and 29% of men will use eHealth technologies.

Ŷ = 1 +
The odds ratio predicted by the model in table 8, using the formulae .621 , implies that the model predicts that the odds of using eHealth technologies are 1.861 times higher for men than they are for women.
The variables were denoted as follows.
Use of eHealth systems/devices with the highest value of 1 (have ever used) and 0 (have never used any eHealth).
Gender was a variable denoting the sex of the patients, with 1 for males and 0 for females Education was the education attainment level of the patients, with 1 representing those with completed masters degrees and above, 2 representing those with completed bachelor's degree, 3 diploma and 4 ordinary certificate.
Location was the districts of residence of the patients, with 1 representing Kampala, 2 representing Jinja, 3 Mbale and 4 Mbarara.
Type of patient was denoted with 1 for outpatients and 0 for recovering patients  Both models (Pearson Chi-Square model and the logistic regression model) indicate that education level, gender and location of the patients are strong determinants of eHealth adoption.

Discussion
The study revealed that training patients to use the technology, communicating the benefits of eHealth to users, security of patient data and the ease associated with using eHealth systems facilitate adoption. On the other hand, digital illiteracy, technophobic nature of users, lack of system's acceptance among users and system's that do not meet the needs of the users negatively affect adoption. Further analysis revealed that education level (master's degree and above), location (residing in Kampala) and gender (being male) significantly influenced eHealth adoption.
This study revealed that training was a facilitator of eHealth adoption. Training has been identified as one of the key success factors in technology acceptance. Commensurate to our findings, receiving necessary training prior to using the system has been encouraged in several other studies [6], [38], [44], [19], [28], [21], [40], [30], [33]. Training equips users with the knowledge of the system [20], gives a chance to acclimate to the new processes and in the long run boosts confidence to use the system. Effective user training will ensure that users have an optimal starting point for working with the new information system [45], facilitate optimal IT use and acceptance [46] and ensure users with differing levels of IT skills become comfortable with the software [47]. With inadequate training, the system operates, but does not fulfill its desired expectations whilst non-trained users will resist the change [47]. Although some studies emphasize that usability eliminates the need for training [48], many studies have referred to training and support in relation to acceptance of eHealth systems [49], [50]. In some studies, training boosted peer and management support, which was a catalyst for system learning and use [33]. Some studies, however stress that training can only be effective if systematic processes are properly followed [46].
Communicating the benefits of eHealth systems to the users ranked second in this study. Systems are as good as the users knowing their benefits, and training has been found to fulfill this. Reminding users of the usefulness of the systems increases the chances of adoption as alluded by several other studies [40], [30]. In some studies, communicating anticipated benefits was reported to increase user acceptance of the eHealth system [51], [5], [52]. Communication tightens the loose ends between the patients and the healthcare providers, but most importantly, makes the users aware of the system. User resistance and low adoption of eHealth has largely been attributed to the lack of awareness of the potential benefits of these systems. In some studies, naïve optimism, as a result of lack of communication, has created pockets of resistance even before implementation [53].
This study also revealed that the security of patient data is very crucial in accelerating adoption of eHealth. Securing patient data involves protecting confidential medical information and once security is compromised, it creates a sense of fear and resistance among users. Similar to this study, concerns over privacy and security being compromised have been raised in several other studies [6], [44], [39], [38], [10], [54] as barriers of eHealth adoption. In a study conducted by Chang [42], participants expressed concerns about the confidentiality and the security of patient data with smartphones, with specific concern on multimedia capabilities that were perceived as having the potential of abuse. When patients exude fear in the system, its use and adoption will be far from being attained. The more robust and secure the system is, the less likelihood of attack, hence adoption.
The ease associated with using systems was a factor revealed in this study that was critical for successful adoption of eHealth among patients. There has been wide debate on whether ease of use can be ascribed to technology acceptance. But Kassim [55] study stresses that the ease of use is associated with increase in user satisfaction and trust in the system. In a study conducted in Ghana [56], ease of use and perceived usefulness had the strongest influence on eHealth adoption than any other factors. Likewise, other studies [57] have underscored the relative importance of ease of use in influencing eHealth adoption.
This study revealed that digital illiteracy is a barrier to eHealth adoption among patients. This can largely be attributed to lack of training and no user involvement at the time of design and implementation. Lacking digital skills to operate a system can be aggravated by the little or no formal education. As reported in other studies [20], [24], [37], [40] the lack of ICT skills to operate digital technologies was a very big impediment to adoption. Whereas a study conducted in Finland indicated that digital literacy does not have a direct impact on adoption [58], research conducted in Uganda, found out that expectant mothers did not use digital health technologies in their routine antenatal care practices because they lacked technical skills to operate the internet, computers and smartphones [29]. Because of this problem, many users become technophobic -the fear to use technology. This study revealed a strong correlation between digital illiteracy and technophobia. This technology fear exhibited by users is partly due to little or no exposure to technology or digital tools and the fear to be ridiculed [59], and as a result, many shun using eHealth systems as reported in other similar studies [19], [44]. If not dealt with at an early stage, it may result into one being cyberphobic [59] which is an abnormal fear detrimental to users.
Lack of system's acceptance among users was another barrier of eHealth adoption cited in this study. System's acceptance among users could be caused by no user involvement [33], when the system does not address the needs of the users [60], not communicating the benefits of the system [40] and to a smaller extent, attitude towards technology [23]. Like this study, several other studies [13], [26], [24], [61] reported user acceptability of the systems to be a very big challenge to eHealth adoption. In a study conducted by [62], they recommended ensuring user acceptance to fully realize the potential of digital health technologies. In another study that was conducted in Iganga district hospital involving nursing mothers, the NeMo system was successful because of acceptability among mothers [13]. When users do not have a sense of ownership of the system, acceptance will be hard which increases system rejection.
The study further revealed that once the system does not address the needs of the users, then ehealth adoption cannot ensue. Usefulness is the knowledge the users have of the system and the benefits that accrue from its use. Many scholars [38], [28], [23], [30] [13], [60], [63] have equally reported on perceived usefulness in facilitating or impeding the successful adoption of systems in general. Once users do not perceive the system as useful, acceptability will be very low. However, some studies have recommended training users [6] and active user participation in the system evaluation process [54], [14] to enhance user's knowledge of the system. Behind a successful eHealth systems is the ability to satisfy the needs of the users [57], [64].
This study revealed that demographic factors such as education level, location and gender could influence eHealth adoption. Both models indicated that the gender of the patient can influence adoption, however the logistic regression model further revealed that male patients were 2.662 times more likely to use eHealth systems than the females. The findings in this study can be corroborated with [56], [18], who equally reported gender to be a determinant of eHealth adoption, although, in their study, females were more likely to adopt than the males. However, other studies have equally reported the males enjoying higher levels of eHealth adoption than the females [65], [66], [18]. Conversely, the study revealed that a patient with a master's degree or higher was 2.297 times more likely to use eHealth technologies than the rest of the participants. Education influences eHealth adoption and many scholars have equally underscored the importance of education in accelerating eHealth technology acceptance and adoption [19], [41], [31], [67]. In a study conducted in Ghana, it was revealed that participants having a higher education used eHealth devices more often than their counterparts [56]. Education shapes attitude and perception, and it has been reported to improve self-efficacy [68], [69]. Lastly, this study revealed that location as a factor strongly influences eHealth adoption. Specifically, the odds were in favor of the participants residing in Kampala than the rest of the districts. Unlike Kampala, these locations have poor network coverage, intermittent internet connectivity and poor telecommunication infrastructure, which, most times disrupts connectivity [22], [10], [16]. There's little literature to support location as a determinant of eHealth adoption, however, in some study, though not necessarily related to eHealth adoption, it was reported that location affected the adoption of commercial internet [70]. Similarly, a study by [71] found a correlation between technology adoption and geographical distance.

Limitations of the Study
The limitations of this study that were both in breadth and accessibility of study sites. At a hospital level, there were many restrictions to access study participants due to the COVID-19 pandemic. Some facilities that had initially approved our study later backed out in a bid to curb the COVID-19 virus spread. At a country level, the two nation-wide lockdowns affected both public and private transport. Inter and intra-district movements were limited, at a certain point, the study had to be halted because it was no longer possible to get travel permits from the relevant government organs. At the participant level, patients were very jittery to interact with our research assistants because at that time, Uganda was at the peak of the second wave, hence many participants declined to participate. Another limitation of the study was that no data was collected on the level of exposure and effectiveness of eHealth technologies as a selection criterion, which could have affected the perception of the participants. Rather, the objective of the study was to get empirical data on the barriers and facilitators, which will inform a more rigorous study.

Conclusion
This study showed that gender, education and location have a significant influence on eHealth adoption. The study also revealed that hospital, technological and individual characteristics had a positive influence on eHealth. Specifically, in order of score, it was revealed that training patients, communicating eHealth benefits to the users, user involvement in the preliminary implementation phase were hospital factors that influenced eHealth adoption among patients. Subsequently, technological factors such as security of patient data and ease of use had an uphill influence on eHealth adoption. Lastly, the study revealed that individual factors such as lack of acceptance among users, technophobia, digital illiteracy and eHealth systems designs that do not meet the patient's needs had a negative influence on eHealth adoption. Whereas the other factors under hospital, technological and individual barriers/facilitators showed influence of eHealth, their average score was relatively low.
With respect to implications, the findings of this study should be used by Ministry of Health of the Republic of Uganda to enforce technology inclusion at every point of care in health facilities, and embark on advocacy and training programs to enhance digital skills of patients.
Similarly, this study divulges key barriers and determinants of eHealth adoption among patients, which, when ardently addressed by the ministry or other partnering agencies, can change the face of eHealth adoption.
The key success factor of any technology is having stakeholders work collaboratively; hence, players in the sector should embark on programs that create synergy between patients and the healthcare providers in order to accelerate eHealth adoption.

Author Contributions:
The authors' contribution towards this study were as follows; Conceptualization, Hasifah K. Namatovu; methodology, both authors; data collection, both authors; validation, both authors; formal analysis, Hasifah K. Namatovu; investigation, both authors; resources, Hasifah K. Namatovu; data curation, both authors; writing-original draft preparation, Hasifah K. Namatovu; writing-review and editing, both authors; supervision, both authors; project administration, both authors; funding acquisition, Hasifah K. Namatovu. All authors have read and agreed to the published version of the manuscript.