To explore the anticipated acceptability of a proposed blended learning implementation package, we used the TAM to guide data analysis. TAM is a theoretical model that examines constructs influencing the acceptance or rejection of technologies [26]. The model proposes that acceptability of any technology can be predicted by 2 factors: perceived usefulness and perceived ease of use (Figure 1). Perceived usefulness refers to the belief that a certain technology would increase an individual’s job performance. Perceived ease of use is a belief that the use of the new technology would be effortless [27]. TAM proposes that a populations’ perceptions of usefulness and ease of use of a technology impact the population’s intention to use the technology, which in turn influences the population’s actual usage of the technology. Furthermore, the key constructs of perceived usefulness and perceived ease of use are influenced by external factors, such as pedagogy, digital literacy, infrastructure, and the learning environment. Understanding these external factors is essential for designing a such a package. We used the TAM in this formative sub-study to guide the analysis of the relevant external factors influencing perceived usefulness and perceived ease of use surrounding blended learning in the context of ICT.

This study was conducted at health centers in Balaka and Machinga, two districts in Southeastern Malawi. Malawi is a country in Southeastern Africa with an adult HIV prevalence of 8.9% [1]. Although Malawi has met the second two UNAIDS 95-95-95 targets, it has not yet reached the first 95% HIV testing target, with only 88.3% of adults living with HIV aware of their HIV status at the time of the study [1]. Simultaneously, Malawi has one of the lowest rates of qualified HCWs globally with less than 1 medical doctor per 10,000 people [28], resulting in task shifting of HIV services to lay cadres with less formal training [29]. Particularly relevant to the proposed blended learning strategy, is the fact that Malawi has one of the lowest digital literacy levels (having the necessary skills and knowledge to confidently and effectively use technology) globally ranking 131 out of 134 countries by Network Readiness Index [30].

This formative work was conducted to inform a cluster randomized controlled trial designed to examine the implementation, effectiveness, and cost-effectiveness of a blended learning ICT package [23]. Data for this qualitative substudy were generated through IDIs conducted with lay HCWs between November and December 2021. This formative study purposively selected facilities from both districts (Machinga and Balaka) and different levels of care (dispensaries, health centers, and hospitals) for purposes of generalizability. To be eligible for participation, HCWs had to be at least 18 years old, work full-time at one of the health facilities, and be involved in counseling index clients or tracing contact clients. With engagement from facility supervisors, the research team identified up to 2 eligible participants in each facility. These individuals were actively involved in ICT at the time of the study. A research team member visited all the eligible participants in the 14 facilities to give them more details about the study and to seek their consent. Contact details of all consenting participants were shared with one of the research team members who made arrangements with the participants for the date and time of the interviews.

IDIs were conducted by 3 trained Malawian qualitative interviewers in Chichewa, the local language. All interviews were conducted by 3 female external interviewers. Among them, 1 interviewer had a doctoral degree in social sciences and 2 had bachelor’s degrees in social sciences. The interviewers were trained on techniques of normalization and probing to reduce the risk of socially desirable response bias and had previous training and experience in qualitative data collection and research positions within the organization (postdoctoral fellow, senior research officer, and research officer). Furthermore, interviews were conducted in a private setting at the facilities, with only the interviewers and participants present. Interviews were conducted using a semistructured guide developed for the study through a series of rigorous, iterative discussions among the research team. The questions for this analysis focused on feasibility and acceptability of a blended learning implementation package.

After consenting and before conducting the interview, the research staff introduced themselves and their role, explained the purpose of the interview, and reiterated that they could refuse to answer questions, that all data would be anonymized, and that responses would have no consequences on their employment. Interviewers first asked questions about participants’ views of the standard training. Interviewers then provided participants with a brief description of the planned blended learning approach, explaining that it involved tablet-based asynchronous lessons modeling counseling approaches, as well as interactive synchronous in-person group sessions to practice the skills learned. Finally, they asked about perceived feasibility, probing on potential barriers. Before the start of the IDIs, the interview guides were translated, back translated, and pilot tested with iterative refinement. Interviews were conducted over the phone or face-to-face and lasted approximately an hour long. Repeat interviews were not carried out. Interviewers summarized each IDI immediately after it was conducted. Audio recordings were transcribed verbatim in both Chichewa and translated to English by professional transcribers. Transcriptions were not returned to the participants.

The team took a thematic approach to coding [31]. First, the team read the summaries to identify key themes and observed that they were framed by TAM. The research team then developed a qualitative codebook based on a focused examination of the transcripts, debrief reports, and interview summaries. The codebook consisted of both deductive codes from the interview guides and inductive codes that mapped onto TAM constructs. In addition, 2 trained research analysts (TEM and CJM) applied codes to transcript text using the Dedoose (SocioCultural Research Consultants) web application. As a first step, each analyst independently coded the same transcript and compared code applications to ensure agreements in the coding process. The analysts discussed and resolved any coding disparities, updating the codebook as needed to facilitate a shared understanding of all codes. The analysts then met weekly throughout the coding process, which provided space to review code applications and address any challenges. After coding was completed, analysts reviewed the coded text to extract key themes through the identification of patterns, naming and defining themes and refining the identified themes. Analysis was organized by factors that influence the TAM’s constructs: perceived usefulness (what participants perceived as gained or lost by the blended learning implementation package) and perceived ease of use (what participants perceived as facilitators or barriers to acceptance of the blended learning implementation package). The findings were not shared with the participants for their feedback.

Ethical clearance was provided by the University of North Carolina at Chapel Hill institutional review board (#20‐1810), the Malawi National Health Sciences Research Committee (#20/06/2566) and the Baylor College of Medicine institutional review board (H-48800). Interviewers obtained written informed consent from all participants before starting the IDIs, reminding participants that their participation was voluntary and could be withdrawn at any time. Participants were compensated the equivalent of US $10 per interview. All data were deidentified to protect participant confidentiality.

A total of 14 facilities participated in the IDIs. Among those, 2 of the facilities were district hospitals, 9 were health centers, and 3 were dispensaries. Furthermore, 9 facilities were in Machinga and 5 were in Balaka. In total, 26 HCWs participated in the IDIs. A total of 7 were female and 19 were male. Furthermore, 16 participants were from Machinga and 10 were from Balaka. There were no participants that refused to participate. All the participants were involved in both screening and counselling index clients and conducting contact tracing and had previous exposure to standard ICT training.

Our analysis identified two sets of factors influencing perceived usefulness: (1a) opportunity for self-guided learning through digital portions and (1b) opportunity for social exchange through face-to-face interactions. We identified 5 factors influencing perceived ease of use. In addition, 2 factors; (2a) need for orientation to the digital technology and (2b) accessibility of devices were related to perceived ease of use to the technology itself, while 3 factors (2c) communication around training schedules, (2d) support for logistical arrangements, and (2e) monetary compensation were related to perceived ease of use of the training approach (Figure 2). The following sections develop these themes, and Table 1 shows the illustrative quotes.