Current empirical evidence suggests that successful adoption of eHealth systems improves maternal health outcomes, yet there are still existing gaps in adopting such systems in Uganda. Service delivery in maternal health is operating in a spectrum of inadequacy, hence eHealth adoption cannot ensue. This study set out to explore the challenges that impede eHealth adoption in women’s routine antenatal care practices in Uganda. A qualitative approach using semi-structured interviews was employed to document challenges. These challenges were classified based on a unified theory of acceptance and use of technology constructs. One hundred and fifteen expectant mothers, aged between 18 and 49 years, who spoke either English or Luganda were included in the study that took place between January to May 2019. Thematic analysis using template analysis was adopted to analyse qualitative responses. Challenges were categorised based on five principal unified theories of acceptance and use of technology constructs namely: performance expectancy, effort expectancy, social influence, facilitating conditions and behavioural intention. Facilitating conditions had more influence on technology acceptance and adoption than the other four constructs. Specifically, the lack of training prior to using the system, technical support, computers and smart phones had a downhill effect on adoption. Subsequently, the cost of data services, internet intermittency, and the lack of systems that bridge the gap between mothers and health providers further hindered technology uptake. In conclusion, strategies such as co-development, training end-users, garnering support at the national and hospital levels should be advocated to improve user acceptance of technology.
This study employed a qualitative design based on UTAUT to collect data on the barriers of eHealth adoption in routine antenatal care. Specific data on expectant mothers was collected using face-to-face semi-structured interviews. The guiding philosophy for conducting this research was interpretive epistemology using inductive reasoning influenced by the ontological underpinning of critical realism. An interpretive approach was used because it identifies the presence of the participant’s perspectives as primary sources of information analysed against cultural and contextual circumstances. 56 Critical realism is a belief that we have a limited understanding of the real world out there, hence knowing the complex reality demands use of multiple perspectives. 57 The purpose of this study was to explore challenges that impede the use of eHealth technologies in routine antenatal care practices among expectant mothers in Uganda. To achieve this, the following questions in Table 1 were used based on the five constructs of the UTAUT. The perceptions of expectant mothers were structured along with the UTAUT constructs of performance expectancy, effort expectancy, social influence, facilitating conditions and behavioural intention. Interview questions following the five constructs of the UTAUT model. The use of the qualitative approach was suitable because this study revolved around getting the participant’s perception and sentiments on the challenges of eHealth adoption, hence there was a general feeling that these thoughts would have been missed if another research design was adopted. Kampala has a total of 1458 health facilities, of these 26 are government-owned, 1371 are private-for-profit and 61 are private-not-for-profit. 58 Of the 26 government-owned facilities, 11 offer antenatal care services to the public. The study focused on six of the 11 health facilities located in Mengo Kisenyi, Kasubi Kawaala, Komamboga, Kitebi, Kiswa and Kisugu, which partly make up the low-income suburbs of Kampala. This area has a total population of 145,020. That is about 1% of Kampala City’s residents who are served by these health facilities. Of these, the composition of the female population is 52.5%. Census data was used to describe the socio-economic profiles of the study participants (Table 2). Nine percent of the females completed ordinary level education, 5% advanced level education and 4% above 18 years are illiterate. Similarly, 3% between 12 and 29 years had given birth, 38% were working, 27% possess mobile phones, 14% used the internet and 15% of the households owned a computer. On average, 0.6% of the participants stayed 5 km or more from a health facility. The data was used as the basis for selecting the study population. Overall profile, statistics, and demographic characteristics by Kampala City’s underserved Parishes, Uganda. 59 Health facilities that provide free basic and comprehensive obstetric antenatal care services to expectant mothers were chosen. Kampala, a heavily populated city was of particular interest because (i) it is a primate city with a wide range of agglomeration of economic and socioeconomic activities; (ii) it is the largest population centre with wide mobile coverage and digital healthcare services, (iii) 30% of its population have access to internet unlike their counterparts that score at 9% and (iv) Kampala is the only town with 4G/LTE. 60 A higher proportion of female individuals in Kampala owned smartphones (18%) compared to male individuals (13%). 61 Kampala is Uganda’s national and commercial capital bordering Lake Victoria with an estimated population of 1,680,800 people covering an area of 3263.3 square miles. It is reported to be among the fastest-growing cities in Africa, with an annual growth rate of 4.03%. Health facilities were purposively selected while study participants were selected based on three broad classifications; (i) proximity to the health facility (within a radius of 15 km), (ii) health measures (gestation, gravida and parity) and (iii) socio-economic status. We used education and literacy as a proxy for socio-economic status. A total of 305 expectant mothers were drawn from Mengo Kisenyi health centre (HC) IV, Kasubi Kawaala HC III, Komamboga HC III, Kitebi HC III, Kiswa HC III and Kisugu HC III during antenatal care clinic days. In all these facilities, ANC services are offered between 9:00 am and 1:00 pm. At the facilities, the research team randomly selected the expectant mothers who were either at the pharmacy, entering or exiting through the main gate, or those exiting the examination room or laboratory. At that point, verbal consent was sought and phone numbers were exchanged, which were later used to schedule interviews, typically within 48 h. Upon receiving verbal consent at the health facility, the research team collected some information on demographics, gravida, gestation, and parity to ascertain the eligibility of the study participants, whilst collecting data to determine their level of eHealth adoption, ranked using a scale of 1 to 5 (1 represented those that had never used eHealth before and 5 represented frequent users). Eligible respondents were telephoned and a meeting was set up to obtain written approval prior to the commencement of the formal interviews. All participants who assented were first asked their language of preference, then presented with a consent form for signing and their rights to participate clearly spelled out. After thorough scrutiny of the inclusion criteria coupled with those who declined to participate after being telephoned (citing reasons like privacy, busy schedules, lack of spousal consent), the number dropped from 305 to 115 participants. Selection bias cannot be completely avoided, however, the rigor in the selection process we employed aimed at minimizing this possibility. This whole study took five months between January and May 2019. Inclusion criteria: Expectant mothers aged 18–49 years residing in Kampala and its outskirts who could speak English or Luganda were eligible for the study. Face-to-face semi-structured interviews were conducted with 115 expectant mothers because they gave researchers the flexibility to navigate and probe participants’ viewpoints on the challenges of eHealth adoption. Four research assistants (RA) and the lead researcher (HKN) conducted the in-depth interviews at the homes of the participants. Using research assistants during the exercise was to a certain degree minimize bias issues. The research assistants who were university graduates were trained in qualitative data collection methods, objectives of the research, dialogue management, and how to respond to participants of a different disposition. Similarly, basic training in transcription, coding and analysis was done. The interview guide was in English, but the research team used both English and Luganda (a local dialect that is widely spoken in the central region of Uganda) during discourse to allow those who could not comprehend English to participate with ease. The research team took notes during the interviews; however, all sessions were audio-recorded. The interview guide was composed of questions that explored the challenges that hinder the use of eHealth technologies during ANC and recommendations to improve eHealth adoption among expectant mothers. Each interview took an average of 45 minutes to complete. The majority of the participants preferred having the interviews during mid-morning hours because they had to first finish their house chores. A pretest was done with the expectant mothers of Wakaliga, Rubaga division to determine that the respondents will understand the questions as well as ascertain the validity of these questions. One repeat interview was conducted with seven mothers to help demystify some responses that were not clear in the first interviews. Interviews were conducted at the homes of the respondents to allow participants to freely express themselves without feeling pressured. Interviews followed topical trajectories however, the conversation slightly pivoted away from the main questions to gain deeper insights into the participants’ views. Responses from the qualitative survey were analysed within 2 months. Thematic analysis using template analysis was adopted using preconceived themes based on the UTAUT constructs. Transcripts and researchers’ notes were compiled to identify common patterns, which were later coded and validated by the four RA’s, the lead researcher (HKN), and the second author (TJO). AQUAD 7 software was used for transcription; however, minor improvements were made to make the transcription 100% accurate particularly in cases where participant utterances were in the local dialect. The lead researcher, TJO and the four research assistants participated in the transcription process. For each researchers’ transcriptions, the frequency of responses to a particular question was documented and compared across other researchers’ transcriptions; hence, analysis for each group of respondents was developed. This was done to identify common patterns, similarities, and differences in responses. The lead researcher employed inductive reasoning which begins with a detailed observation of the world followed by empirical generalizations and identification of preliminary relationships. 62 The data was analysed to help explain the current challenges that impede expectant mothers from using eHealth initiatives during their ANC routine practices. Also, it provided an insight into the possible recommendations that could drive adoption. Uploaded transcripts were carefully studied to get insight into the data collected; coded to provide an overview of the disparate data, 63 and clustered to identify patterns and generate themes. A deductive approach to theme identification was employed since we had preconceived themes. Coding was done in two phases; the RA’s, who essentially used descriptive codes to summarize detailed data, did the first level coding and, the second level coding was done by HKN which was interpretive, focusing majorly on pattern codes aimed at giving more inference to data by clustering data into a smaller number of more meaningful units. In the first level coding, the RA’s generated 56 codes, and after thorough scrutiny, overlaps, ambiguous, irrelevant and redundant codes were integrated or discarded leaving 21 codes in the second level coding (pattern codes), which were further collapsed into five preconceived themes. To understand the distribution of the demographic characteristics of mothers and test the independence of the variables, coded data was processed using IBM SPSS statistics. This was further analysed using descriptive statistics and Pearson’s chi-square illustrated in Table 3. Demographics of the respondents. To eliminate researcher bias and prejudice, the second author (TJO) with a good research trajectory coded part of the data (pattern codes), henceforth, discussions of the similarities and differences in coding were held. To ascertain the validity of the findings, the analysis and all coding instances (first level codes and pattern codes) were first given to the study participants and later to two independent researchers (with doctorates in community psychology and ethical sociology) to further audit the findings. Researchers unanimously agreed on the coded data which was later compared against those of the study participants and independent researchers for purpose of building consensus. This rigorous process can ascertain largely the level of authenticity of the analysis process. After transcription, a subset of five interviews were gradually examined until all were added to the template. Using a clustered analysis of five interviews offered a good cross-section of the insights and views covered in the entire dataset. For each interview examined, if it offered new insight, it was compared against the already defined themes in the template, and if it did not fit, then a new theme was created. This iterative process involved defining new themes, modifying existing ones, and to a certain extent, especially where redundancy was registered, some themes were discarded which left the five preconceived themes (PE, EE, SI, FC and BI). The themes were a result of the pattern codes created in the second-level coding. To avoid misrepresentations, data errors and collecting feedback for theme enhancement, the generated data was validated with the seven respondents, and the feedback collected was used to strengthen the themes. The resultant template is shown in Figure 2. Final version template.
N/A
DIMA AI Care
