Target population

Our study was approved by the Institutional Review Board and the ethics committee at the University of Washington (STUDY00013617). The ethics committee stated no informed consent was needed as individuals were not identified. Local ethical review committee clearance in Kenya was not logistically feasible with eight institutions [9]. Being in a project funded by the NIH, we opted for one, and the institutions we work with are satisfied with the single ethical clearance. This qualitative study was anonymous, and all participants provided informed consent. Individuals invited to participate included those likely to serve as mentors in our overarching goal of increasing capacity in oral and HIV research in Kenya. To increase diversity, our study population included representatives from Nairobi, Kisumu, Eldoret, and Mombasa. Potential mentors included university professors, clinical researchers, and senior administrators with professional experience in HIV and/or oral health (Table 1).

Data collection and analysis

We conducted 17 interviews involving 22 interview subjects. Three of the interviews were group interviews involving eight subjects, with 2–4 persons in each group interview. These interviews were conducted concurrently, with each respondent sharing their views or contributing to the comments of others. The person speaking first varied for each question.

A semi-structured interview instrument was created to gain an understanding of capacity on oral and HIV research, targeting five relevant areas: support & resources (7 questions), mentors (6 questions), mentees (6 questions), and research & healthcare environment (4 questions). Prior consent was gathered from interviewees, and all interviews were recorded. All identifying information was removed from the transcripts of the recordings. Subjects were assigned a unique code number. All interview materials were kept in a secure location, with only study team access. The recorded materials were transcribed, and each interview ranged between 4–7 pages long, yielding 84 total pages of textual responses for analysis [10].

The researchers employed an inductive approach based on Grounded Theory Methodology (GTM), utilizing constant comparative analysis to derive concepts from data.

The researchers choose to apply an open coding method instead of a priori coding for multiple reasons. The subject area has not been widely examined, and no pre-existing codebook was available. When applying a priori codes, researchers run the danger of overlooking insights from data that do not fit the existing framework. Most of the researchers on the team have deep familiarity with the research and healthcare environments in Kenya and hold opinions on existing challenges from a scholarly perspective; the potential for bias would be high if codes were assigned a priori. The goal of this inquiry was to surface the perspectives of practitioners and administrators in their own voices. Accordingly, open coding was determined to be the most appropriate approach for addressing these challenges as part of a Grounded Theory Method [11]. None of the authors participated in the creation and assigning of codes to the transcripts to avoid introducing bias.

Qualitative analysis was conducted by a group of nine independent reviewers (led by MS) with experience in talking with health care professionals and qualitative data analysis. Each reviewer analyzed one transcript while free coding the text. Transcripts were assigned randomly to reviewers. Reviewers were instructed to capture and utilize the terms or phrases generated by the informants themselves. Prior to beginning the free coding, reviewers took part in a training workshop that used sample responses from multiple transcripts. Coding was conducted at the question response level. The free coding phase of the analysis resulted in 457 unique codes. Since each response could have more than one code, there were many duplicates across transcripts.

Axial coding was performed using a team-based approach. This method has been successfully applied in other health research contexts when the goal is to achieve high intercoder consensus among a single team rather than replicability across different teams of multiple projects [12]. Reviewers were organized into small teams that performed an iterative grouping analysis where each code was typed on a virtual “post-it” note in a collaborative online whiteboard environment (MIRO) [13]. Typically, the researchers would conduct this type of exercise in a face-to-face setting, but the need for isolation during pandemic conditions made it challenging to gather physically in a safe manner. It is unclear what effect this may have had on the judgements and decisions or the analysts. One observation is that the practice of zooming in on one segment of the board at a time, rather than continuously scanning the entire “wall” simultaneously resulted in a faster analysis than usual. As notes were shuffled and moved to different categories, the researcher would focus more deeply on each category individually. Through multiple iterations, the reviewers sorted the open codes into groups that appeared to exhibit a “best fit” to the evidence in terms of leaving out the smallest number of uncategorized codes (outliers). Ultimately, 457 open codes were reduced into 16 axial codes organized into four categories: Research, Motivation, Obstacles, and Support. If reviewers were in disagreement, we used the solution that provided a more complete encapsulation of the codes.

An online software app called WordCloud Generator by Monkey Learn was used to create word clouds and diagrams of common words and phrases extracted from the interview transcripts [14, 15]. Word clouds are a useful means to make a large body of text comprehensible, particularly when paired with other qualitative means of analysis. While the constant comparative analysis identified emergent theme categories to enable the creation of an affinity diagram, word clouds, based on frequency counts of open codes, created graphic representations of significant terms and concepts in the actual words of the interview subjects. These word clouds presented an alternative depiction of the data for communicating with audiences less familiar with interpreting coded qualitative data.

Research

Our analysis represented stakeholders across three different categories including practitioners, researchers, and administrators, all uniformly supported the development of a robust research environment to address the gap in understanding the relationship between oral health and HIV. The research emergent theme consisted of sub themes, including collaboration, interventions, research gap, dissemination, and infrastructure. One term appearing in the research word cloud is KAVI, meaning Kenya AIDS Vaccine Initiative (KAVI) (Figure 2A). KAVI was established in 1998 as a research unit within the Department of Medical Microbiology at the University of Nairobi, with an initial mission to conduct basic research in epidemiology of HIV and to carry out HIV/AIDs vaccine trials. Interviewees with KAVI were included as important stakeholders for our project. The research gap in oral health HIV frequently arose in coded data, one participant shared, “In Kenya, oral health has always been left behind. Participation of oral health researchers has not been active.” Another participant expanded, “We have unmet needs of oral health because of a lack of research. We are therefore unsure of the needs because of the lack of data. Our interventions are not evidence based.” Stakeholders were motivated to write publications and contribute to closing the research gap in oral health HIV. The affinity diagram shows the emergent theme research is connected to emergent themes support and motivation (Figure 1). Interestingly, the research theme is not connected to obstacles. Overall, stakeholders were motivated and passionate about research. They expressed interest and highlighted the need to address the research gap of oral health and HIV. Stakeholders were willing to partake in oral HIV research, given a few obstacles were removed (Figure 2A).

Motivation

The primary motivation for participants to cultivate a research environment examining the intersection of oral health and HIV has three parts. One of these parts was the desire to improve health outcomes by using an evidenced-based approach to interventions. The second one was a desire to extend one’s professional network and collaborate with practitioners and researchers across interdisciplinary lines. The third part was a desire to publish one’s work to share knowledge or advance one’s professional standing. Additionally, stakeholders frequently spoke about their interest in the research gap in oral health and HIV. A large desire exists to improve health outcomes for PWH. Finally, stakeholders want to contribute to evidence-based interventions. One stakeholder expressed the need for research with the following remark, “We have unmet needs of oral health because of a lack of research. We are therefore unsure of the needs because of the lack of data. Our interventions are not evidence-based.” This stakeholder explained the current state of oral health in Kenya, emphasizing a lack of data for evidence-based interventions. Stakeholders recognized the research gap within oral health HIV and strongly wanted to study oral health HIV, write publications, and disseminate their findings to further their careers.

One of the most prominent words on the motivation word cloud was “mentee” (Figure 2B). Stakeholders were interested in leading and mentoring the next generation of researchers. Stakeholders were asked several questions regarding what attributes a good mentor and what attributes a good mentee hold. Many participants had a response similar to the following, “A good mentor is one who nests you in their ongoing program in the area of research and ensures lifelong support until one is able to stand on his or her own as a reputable researcher.” In addition to walking with the person they are mentoring, mentors should be able to give their time to regular meetings (Table 2). We also learned about the strong mentoring community established at Moi University, with a participant sharing, “In Moi University, we have mentor training taking place every year and they are linked to the students and these relationships have gone beyond school. Both quotes shared from interviews express strong mentorship bonds sustained throughout their careers.” Stakeholders made it clear; support is needed to sustain the participation of mentors. Though stakeholders used different terminology or had slightly differing ideas of execution, compensation was a theme. Words such as, “pay”, “financial”, “resources”, and “monetary”, appeared frequently in the motivation word cloud (Figure 2B). Stakeholders felt lack of compensation was an obstacle to mentorship (Figure 2C). Stakeholders expressed recognition would assist mentor retention. One participant shared, “Recognition is one of the incentives to retain the mentors. Ability to see your mentee succeed in their project.” Most of the factors identified to influence mentor participation were already in place, the motivation to conduct oral health HIV research, opportunities to write publications, network, and access to conduct research and recruit mentees at universities. Despite the influential pieces in place, the emergent theme motivation had the second most connections with the obstacles emergent theme (Figure 1). Incentives to research oral health and HIV are needed, particularly regarding funding and time management.

Obstacles

Time commitment and lack of funding were frequently cited as the primary obstacles preventing many practitioners, researchers, and university administrators from participating as a mentor. The words “funding” and “lack” are prominently displayed in the center of the obstacles emergent theme word cloud, indicating the high frequency usage in interviews (Figure 2C). Despite stakeholders’ high motivation, interest in the oral health HIV research gap, and available setting to conduct research at research universities, stakeholders agreed funding was needed for oral health HIV research to be feasible. One stakeholder expressed, “Financial incentive will make people cut down on other hustles and concentrate on mentoring.” Without funding, mentors were unable to prioritize time away from their paying jobs. This stakeholder’s comment also suggests that by funding oral health HIV research, the time management obstacle may also be removed. With stakeholder input, we have narrowed down obstacles to one main barrier: funding.

Outside of the funding obstacle, words such as “demands”, “institutional expectations”, “timelines”, and “constraints” appeared. These words represented the time management obstacle. Stakeholders spoke of competing priorities and advised that oral health has not been a priority. “With the prevalence of oral diseases standing at over 50% among PWH, efforts targeting oral health should be a priority” [1]. Funding is a large motivation for mentor participation for most stakeholders. Unclear expectations were another theme within the obstacles emergent theme. Mentors need directional support or guidance on clear goals and directions of the research projects and the mentor-mentee relationship. Clear expectations would promote empowerment and increase motivation for both mentors and mentees.

Support

Several stakeholders identified the support their institutions or agencies could offer in terms of qualified personnel, mentees or mentors, and contributing to recruitment efforts. The word “university” was recorded several times in relation to supporting research activity (Figure 2D). The study identified academic institutions as sources of support for research activity and mentee recruitment. A participant stated, “As academic staff, we are expected to do clinical and laboratory services and research and do a publication. If we want to start something, we have the personnel to start.” Academic institutions were well positioned to support oral health HIV research by providing and helping to identify mentors and mentees. At Moi University, a mentor training is held each year and mentors are matched to mentees. With the support of Kenyan universities, personnel were available to research oral health HIV.

One set of interview questions was focused on what makes a good mentor. The findings showed that, according to the stakeholders, a good mentor has the following traits: passion for the research, developing potential in others, realistic, and best interest of the mentee at heart (Table 2). The stakeholders shared a passion for oral health HIV research and desired to mentor, however, many cited a lack of time in their schedules as an obstacle. Tangible incentives like compensation or more time allocated for research would incentivize stakeholders to become mentors (Table 3). Barriers to becoming a mentor included common themes of lack of time, competing priorities, lack of funding, and lack of structure to communicate and sustain mentorship relationships. Based on the affinity diagram, one facet of the support emergent theme was recruitment (Figure 1). Incentives are required to conduct recruitment of mentors and mentees due to time management obstacles (Table 3). Although we have identified universities as locations where many potential mentors and mentees are located, funding is needed to allocate time and resources towards recruitment and mentor retention.

In the affinity diagram, support and research are connected by the research infrastructure needed to support digital platforms and partner institutions. One stakeholder informed the interviewer, “It’s about developing platforms and systems that collect information about oral health and HIV.” TABASAMU, a collaboration with the University of Washington and University of Nairobi, needs support to create and operationalize digital platforms with participation from partner institutions.