Medical stakeholder perspectives on implementing a computerized battery to identify neurocognitive impairments among youth in Botswana

Abstract

HIV infection and in utero exposure, common in Sub-Saharan Africa, are associated with pediatric neurocognitive impairment. Cognitive screening can identify impairments, but it is rarely used in this setting. The Penn Computerized Neurocognitive Battery (PennCNB), an evidence-based cognitive screening tool, was adapted for use in Botswana. To facilitate future implementation, 20 semi-structured interviews were conducted to elicit key stakeholders’ perspectives on factors likely to be related to successful uptake of the PennCNB in clinical settings. An integrated analytic approach combining constructs from the Consolidated Framework for Implementation Research and modified grounded theory was used. Results underscore the need for cognitive screening in Botswana and the acceptability of the PennCNB. Implementation barriers include limited time and resources, whereas facilitators include standard procedures for introducing new tools into medical settings and for training implementers. Recommended implementation strategies include integrating screening into the existing workflow, implementing the tool in the medical and educational sectors, and targeting selection of children for assessment. This research addresses the research-to-practice gap by engaging in pre-implementation inquiry and designing for implementation. Results will inform the development of strategies to maximize the likelihood of successful implementation of the PennCNB to identify neurocognitive impairment in children in this high-need setting.

Introduction

The human immunodeficiency virus (HIV) contributes to significant morbidity and mortality in Sub-Saharan Africa (UNAIDS, 2018). In Botswana, in utero HIV exposures occur in roughly a quarter of all births (Slogrove et al., 2020). Due to the availability of free antiretroviral therapy in Botswana, survival to secondary school ages and beyond has become the norm for children born with HIV in this setting (Farahani et al., 2014; World Health Organization), but survival is not without HIV-related sequelae. Children living with HIV (HIV+) and HIV-exposed-uninfected (HEU) children demonstrate greater neurocognitive deficits than their HIV-uninfected peers, specifically in the domains of executive functioning, attention, episodic memory, psychomotor functioning, and information processing speed (Le Doare K, 2012; Ruel et al., 2012; Smith et al., 2012). To support the educational and functional attainment of HIV-affected children and adolescents, early detection of neurocognitive impairments is crucial. However, systematic cognitive screening does not exist in this setting. Current screening approaches are time-consuming and require several hours of a psychologist’s time, including manual calculation of scores using pencil-and-paper tests.

To increase neurocognitive screening in Botswana, the Penn Computerized Neurocognitive Battery (PennCNB) was culturally adapted, translated, and validated for use among youth (age 7–17 years) in Botswana (Scott et al., 2020; Van Pelt et al., 2021). Composed of “neurobehavioral probes” (Gur et al., 1992) validated by functional neuroimaging, the PennCNB streamlines neurocognitive assessment by evaluating performance on major cognitive domains and offering advantages including increased standardization and reliability, automated scoring and data generation, and ease of administration (Gur et al., 2010; Moore et al., 2015). The Botswana version of the PennCNB includes specific modules that measure cognitive domains impacted by HIV. Researchers have successfully utilized the PennCNB in multiple pediatric populations (Hartung et al., 2016; Yi et al., 2016). The PennCNB is freely available, except for start-up costs (e.g., computers). Thus, the PennCNB is a relatively low-cost tool for the screening of neurocognitive deficits among children and adolescents in Botswana. The next step in making the PennCNB widely available is to understand how best to integrate the tool into routine practice, taking into account the resource-limited nature of this setting (Yapa & Barnighausen, 2018).

To effectively facilitate the implementation of PennCNB-based screening, an understanding of the context in which the PennCNB may be implemented, the barriers and facilitators to its use, and suggested implementation strategies are needed. The Consolidated Framework for Implementation Research (CFIR) provides a framework for the systematic evaluation of factors that may influence implementation. The CFIR includes the following five domains: intervention characteristics (e.g., features of the PennCNB), outer setting (e.g., sociopolitical context), inner setting (e.g., clinic leadership structure), characteristics of individuals (e.g., clinicians’ beliefs about PennCNB), and implementation process (Damschroder et al., 2009; Keith, Crosson, O’Malley, Cromp, & Taylor, 2017). Few studies have applied CFIR in the pre-implementation phase of research in low- and middle-income countries (LMICs) (i.e., preparation for implementation of a tool) (Means et al., 2020), thus this novel application offers a generalizable process for other tools to be implemented in medical settings in LMICs. This research aimed to elicit key stakeholders’ perspectives to identify factors likely to be related to successful future implementation of the PennCNB into clinical settings in Botswana.

Methods

Participants

Most HIV+ children in Botswana are enrolled in care, so clinical settings offer an ideal location for cognitive screening. Participants were initially recruited from two stakeholder groups: clinical implementers and clinical leadership. Implementers were defined as individuals involved in the direct care of patients (e.g., nurses), and leadership were defined as individuals involved in the decision-making for medical services (e.g., clinic director). After conducting several interviews and reflecting upon preliminary themes, sampling in the clinical implementer group was enriched for perspectives from both mental health and non-mental health clinicians, thus resulting in three final stakeholder groups: mental health clinicians, non-mental health clinicians, and clinical leadership. Sampling focused on stakeholders in the public sector to increase the generalizability of the findings to other LMICs, for public medical centers have fewer resources and treat more patients from low-income communities compared to facilities in the private sector. Purposive sampling followed by snowball sampling began with three primary institutions: The Ministry of Health and Wellness (MoHW), the University of Botswana, and the Botswana-Baylor Children’s Clinical Centre of Excellence (CoE). The MoHW leads funding and decision-making for all public medical services in the country. The University of Botswana houses the only medical school and public health training program in the country and guides the MoHW on introducing care innovations in government medical centers. The CoE, a partnership between the Government of Botswana and the Baylor International Pediatric AIDS Initiative, specializes in the care of children living with HIV in and around Gaborone, Botswana. Since some participants described a history of both leadership and clinical responsibilities, stakeholders were categorized based on their responsibilities at the time of enrollment. For example, practicing clinicians with previous leadership responsibilities were considered part of the clinical implementer group. Participants were recruited by email and phone. Written informed consent and demographic data were obtained from all participants. The intentions of the research team were shared during this process. All procedures were approved by the Institutional Review Board at the University of Pennsylvania and the Health Research and Development Committee within the Botswana MoHW. No participants withdrew from the study.

Procedure

The CFIR, which facilitates evaluation of implementation determinants across ecological levels, guided the development of the semi-structured interview guide (Supplementary File 1). All interview materials were piloted among a subset of a stakeholder group established for the cultural adaptation of the PennCNB in Botswana (Scott et al., 2020).

Prior to conducting the interviews, the interviewer facilitated an introduction to the PennCNB that included a description of the tool accompanied by images of the 13 PennCNB modules and a demonstration of a test for individuals who needed additional explanation. Participants were encouraged to ask questions about the PennCNB before transitioning to the interview. All interviews were conducted in English by an investigator at the University of Pennsylvania (A.E.V.P.) trained in qualitative research, and most of the discussions occurred in-person in a private space in the participant’s workplace. Due to the SARS-CoV-2 pandemic, some interviews were conducted via video conference. The interviewer had no previous relationships with the participants. The interviewer completed a preliminary review of the emergent themes at the conclusion of each interview to assess thematic saturation for sufficient participant recruitment.

All interviews were audio-recorded and transcribed by an external company, and the transcripts were uploaded into NVivo Qualitative Data Analysis Software (QSR International Pty Ltd. Version 12, 2018) for data management and analysis. Transcripts were not provided to participants for review. One participant reviewed draft versions of this manuscript and provided feedback on the interpretation of findings. Field notes were stored in a secure folder. The Consolidated Criteria for Reporting Qualitative Research guided the reporting of this research (Tong, Sainsbury, & Craig, 2007).

Data Analysis

An integrated analysis combining deductive and inductive approaches was conducted (Bradley, Curry, & Devers, 2007). Constructs from CFIR were identified a priori, and iterative coding to ascertain recurrent themes based on modified grounded theory was performed to develop an initial codebook (Saldana, 2009). To further define strategies and aid in the development of future implementation, suggested implementation strategies were sub-coded using the Expert Recommendations for Implementing Change (ERIC) taxonomy (Powell et al., 2015; Waltz et al., 2015). The ERIC taxonomy consists of 73 discrete implementation strategies grouped into nine clusters. An initial coding scheme was applied to two transcripts by two independent coders (McHugh, 2012; Miles et al., 2014). Disagreements were resolved through discussion with the research team, and the final codebook was applied to all interview transcripts by one investigator. Double coding occurred on 25% of the transcripts (kappa= 0.86). All stakeholder groups were analyzed together.

Results

Participant Characteristics

Twenty individuals were interviewed (Table 1). Sex distribution was equal. Participants included both early-career and more established stakeholders (range of time in current job: 2–31 years; age range: 26–62 years). Most participants were nurses (30%) or physicians (30%) by training. Based on roles and responsibilities at the time of the interview, the sample included nine non-mental health clinicians, seven mental health clinicians (e.g., psychologist), and four leaders (e.g., MoHW officer). Most of the participants (65%) did not have prior exposure to the PennCNB. Interviews lasted approximately 40 minutes (range: 20–69 minutes).

Table 1.

Participant Characteristics

Variable	N (%)
Total	20
Median Age, (range)	40 (26–62)
Sex
Female	10 (50)
Male	10 (50)
Stakeholder Group
Leadership	4 (20)
Mental Health Clinicians†	7 (35)
Non-Mental Health Clinicians‡	9 (45)
Occupation by Training
Nurse	6 (30)
Physician	6 (30)
Psychologist	4 (20)
Social worker	4 (20)
Prior Exposure to CNB
Yes	7 (35)
No	13 (65)

^†2 participants are head of department

^‡2 participants described history of leadership in government and health policy

Themes by CFIR Domain

Table 2 presents the main themes organized by CFIR domain, and Table 3 highlights the barriers and facilitators to implementation irrespective of domain.

Table 2.

Interview Themes by CFIR Domain

CFIR Domain	Supporting Quotes
Outer Setting	“So, at our clinic we [Botswana], unfortunately, don’t have the capacity or the resources to identify any cognitive dysfunction […] We don’t have any measures or any validated tests or anything that would identify any cognitive impairment.” “Basically, if you have cognitive difficulties that equates to insanity or stupidity. Yeah. There’s a lot of stigma-a lot of stigma.” “So, there’s really great need in the country.” “But there are not enough psychologists-we will refer a patient, but the patient will have to wait-will be booked maybe two years from now because those psychologists, they’re seeing inpatients in the hospitals, their families, and they’re also seeing outpatients from clinics. So, it’s quite a long queue.”
Intervention Characteristics	“I think the beauty about it is that you train somebody to be able to administer it, and that person doesn’t need the high level of training in terms of being a neuropsychologist…” “…the fact that it’s computerized, it makes the job easier for us. You just have the child doing it and then you just collect the data and interpret it.” “It’s more like a game-like assessment that they are more used to playing games in the phones, so this time they will be more like playing games in the computer. I think they’ll love and enjoy using it.”
Inner Setting	“Guidelines are generally followed. From experience, you’ll find your healthcare workers refusing to implement something in which they have not seen in writing […] So once it’s in the guidelines level, it should be possible.” “From experience, when you train someone to do something, you train their mentor and then you follow up. That’s the usual structure.” “Maybe in some of the private schools you might be able to find computers, but in a lot of places-like even in the hospitals, there might not be a computer that is readily available to be used by patients.” “Because, as you know, there are few health workers. Sometimes someone is supposed to consult, someone’s supposed to do certain things, then they will end up skipping some things if it takes more time or if there are a lot of patients.”
Characteristics of Individuals	“And I think it will be exciting for people who work in the clinical setting who think they are not resourced to do this; but they do feel that it’s important to be able to assess children.” “I think the perception of the healthcare workers thinking that probably [implementation of the PennCNB will] put so much work into their program. That could be a barrier.” “The other problem would be the patients themselves. If it takes long to administer the test-complaints-so many complaints. They will be complaining that this person has been here for over an hour now.”
Implementation Process	“So we’ve got a school health program which is supposed to be running, and it’s got its strategies, but something like this begins to create a bridge between the schools and the health systems […] So it would work out well from both ends in terms of linking the two systems.” “But generally, when you’re implementing something into care it can -in the HIV setting it can be in the form-it can be part of the guidelines for example […] Through the guidelines it can be possible. So, you can for example, when we are writing the guidelines we have section of neurology, we’ll just say when you’re assessing the child if possible send for neurocognition assessment.” “[…] we already have an integration of the child growth in the HIV care […] So if CNB would be integrated, it would just be one of the tools-okay-one of the kind of screening tool that should be added to the existing ones.” “I would look at the health promotion and the health education team simply because while the tool has clinical components in it, it doesn’t always require a clinician to do it. And this cadre is well placed within the health education space. And just the whole task-sharing bit just makes it easier.” “I think involving stakeholders is the most important thing, because if you just involve your healthcare worker and your client […] once you have this neurocognitive assessment, you might have a flood of patients who need action.” “I think for children, a brief review of the history. If they’re having learning problems, for example, then those would be the ideal candidates, or if they have delayed milestones. But ideally since it takes longer, I don’t think it would be ideal to give it to every child, but so there should be a quick screening of either a few simple questions to determine that this might be an impaired child rather than subjecting every child to this.”

Table 3.

Barriers and Facilitators to Implementation

	Key Themes	Supporting Quotes
Barriers
Time	PennCNB takes too long to complete	“So in a clinic setting where a healthcare auxiliary does other things in the clinic [..] I think having to do four or more assessment a day would take a significant amount of time, and that […] would just weigh down on the operations of the clinic.”
	Too many patients for duration of assessment	“We know that the healthcare system is really overwhelmed […] there’s long waiting lines. People are always waiting for service […] I think the only difficulty that they might be in this being acceptable within practice is people actually having time to complete it or to use it.”
Human resources	Shortage of HCWs	“[…] within nursing and doctors here there’s serious staff shortage.”
	Limited mental health clinicians	“And like I said, psychologists are scarce, they’re rare.”
Physical resources	Limited computer and space availability in some settings	“One of the barriers that I see […] is lack of computers in the-in the clinics or even the primary hospitals […]”
Stakeholder pushback	Reluctance from HCWs due to overwork	“I think the biggest barrier is mindset, the attitude of, is this more work? Why should we be spending time doing this when we could do other things?”
	Patients complain about time	“…If patients see that somebody went into a room and they took one and a half hour, then they might start complaining…”
Facilitators
Physical resources	Computers and space available in some settings	“Well the resources are there, the computers are there, the infrastructure is there, the patients are there.”
Procedures	Follow MoHW guidelines	“[…] if there’s a directive from the Ministry as in it’s an implemented guidelines that when a child presents with a referral about cognitive impairment, then you must assess them using at least this tool, then they have that best practice to-to use. “
	SOPs for new interventions	“But again, when something new comes, the Ministry makes sure that we do workshops [and receive training on] how a particular thing is to be done.”
Acceptability	HCWs satisfaction and high acceptability	“Well, I think it would be a great thing. It would be very, very acceptable, I mean especially to the healthcare workers because it will be something new to help in assessing kids.”
	Desire for cognitive screening	“We might not know the magnitude of the problem in terms of children having those neurocognitive issues […] So I would be very excited to see this going forward and implementing this because definitely is going to help us understand the magnitude of the problem that we may have in that field.”

Outer Setting

Mental health and cognitive issues were described interchangeably by most participants, perhaps due to a lack of a unifying definition of cognitive impairment in the local context. Opinions on healthcare workers’ awareness of cognitive issues varied, but most participants commented that community members do not understand the root cause of mental health and cognitive problems. Instead, underlying neurocognitive deficits are commonly attributed to behavioral issues, or children are labelled with terms describing poor intelligence, which often results in stigma and families’ hesitancy to seek out support. Respondents described cognitive impairment as most commonly identified by poor school performance. Responses related to awareness of the prevalence of neurocognitive impairment among children in the community were variable. All interviewees noted that cognitive and mental health resources in Botswana are scarce, especially the lack of available clinicians with the needed expertise. Existing methods for cognitive screening rely primarily on subjective evaluation (e.g., informal observation of behavior during collection of medical history or discussion with patients and caregivers) or academic performance. Children typically require a referral to receive a psychological assessment, but the limited capacity has resulted in queues lasting months to years to see a psychologist for cognitive and mental health evaluations. Thus, respondents expressed a strong need for the PennCNB.

Intervention Characteristics

Overall, respondents expressed high acceptability of the tool, emphasizing that the PennCNB will help children and provide an important, necessary resource. The discussion of the characteristics of the PennCNB included opinions on the strengths and weaknesses of its delivery via computer. Generally, participants described the “game-like” tests as appealing to children and easy to use for all stakeholders, suggesting implementation feasibility. The automated scoring of the battery was described as an improvement over the manual calculation of scores necessary with current screening methods. Some participants suggested that the tool’s validity made it more likely to be effective at identifying cognitive impairments compared to current methods. The ability to administer the battery without having specialized mental health training was also highlighted as a strength of the tool, especially given a lack of local psychologists. Although most participants considered the computerized aspect as an asset, some respondents expressed concern about computer availability in specific settings and children’s computer literacy. However, most participants perceived children’s computer literacy and familiarity as high due to the integration of computers in schools and prevalent use of smartphones in the community. Some individuals acknowledged the practice PennCNB tests administered to ensure task comprehension before the actual assessment important to minimize the impact of low levels of computer familiarity among children. Finally, the length of the PennCNB (~1.5 hours to complete) could present challenges due to time constraints.

Inner Setting

Participants stated that while the MoHW generally has the authority to approve integration of a new tool in the medical sector, implementation within specific institutions (e.g., CoE) also requires approval from the institutional director. Participants from the MoHW referenced specific committees responsible for approving a tool of this nature. Before beginning a new initiative, potential implementers typically attend training workshops that follow a train-the-trainer approach. There is a high degree of reliance on written guidelines, and standard procedures for incorporating new tools are expected. In some settings, necessary resources (e.g., quiet space for administration and technological support for PennCNB maintenance) may not exist. With regard to clinician access to computers, participants explained that multiple individuals may share a computer for various tasks, thus limiting the availability of a computer for PennCNB assessment. Across all locations, respondents emphasized a shortage of staff of all specialties, limited mental health clinicians, and the higher ratio of nurses to physicians. In some settings, staff turnover is common, and district health management teams play an integral role in implementation.

Characteristics of Individuals

Participants stressed the importance of clinician training on the administration of the tool to allow them to facilitate implementation. Despite recognizing the importance of the neurocognitive screening, participants anticipated that competing responsibilities will make it difficult for them to add this screening to their workload. In addition, respondents described potential user barriers, for patients who already wait a long time to see a provider may oppose waiting even longer for an additional assessment. Thus, children and/or their caregivers may not find the duration of the PennCNB assessment acceptable, particularly if the time it takes for one child to complete the PennCNB results in longer waits for other children to see a provider.

Implementation Process

Stakeholders recommended various strategies for implementing the PennCNB in Botswana.

Setting and target participants.

Participants proposed administering the tool in both the medical sector (e.g., clinics) and the educational sector (e.g., schools). To bridge the gap between the two settings, some individuals highlighted school health programs as a potential venue for implementation. Most respondents described targeted screening for children with the greatest need (e.g., as assessed through school performance, referrals, observation of patients during intake, or high-burden clinics), while some participants suggested universal screening of all children to avoid missing cognitive impairment.

Implementers.

Stakeholders advised leveraging current roles by having clinicians who already interact directly with patients (e.g., nurses and psychologists) administer the assessment; guidance and counselling teachers were recommended for the school setting. Respondents suggested that information technology personnel oversee maintenance of the battery, when available. To increase adherence, participants encouraged integrating the PennCNB assessment into existing guidelines (e.g., HIV treatment guidelines) or standard procedures. Further, some respondents proposed designating a specific day for PennCNB assessments at implementing sites.

ERIC implementation strategies.

Stakeholders recommended multiple implementation strategies that align with the ERIC taxonomy (Table 4). Overall, participants stressed the importance of receiving buy-in from all stakeholders (e.g., healthcare workers, government, and parents). For leadership and clinicians, respondents encouraged communicating the value of the tool and emphasizing its role in helping children. For the community, participants recommended increasing awareness of neurocognitive impairment to encourage individuals to seek out support for cognitive issues. Before integrating the PennCNB into practice, stakeholders expressed the need for education on how to administer the tool and interpret the data, which could occur through train-the-trainer workshops. Some respondents recommended designating a computer for PennCNB administration at implementation sites, and some recommended administering the assessments in a dedicated room.

Table 4.

Recommended Strategies per Expert Recommendations for Implementing Change

Strategy	Stakeholder Recommendations
Use evaluative and iterative strategies	Develop team to monitor ongoing use of PennCNB Improve implementation over time Target selection of initial implementation sites
Provide interactive assistance	Use existing IT support in clinics
Adapt and tailor to context	Leverage existing roles as implementers
Develop stakeholder interrelationships	Achieve buy-in from stakeholders
Train and educate stakeholders	Demonstrate administration of PennCNB Provide written guidelines for implementation Facilitate train-the-trainer workshop Educate stakeholders on value of tool and cognitive impairment (community, implementers, policy makers)
Support clinicians	Revise professional roles to distribute responsibilities to non-specialists via task-shifting
Engage consumers	Utilize media to increase awareness about cognitive impairment
Change infrastructure	Designate specific room for PennCNB administration Designate specific computer for PennCNB Decorate room with kid-friendly colors

Note: The ERIC cluster “Utilize financial strategies” was not articulated.

Discussion

Countries in Sub-Saharan Africa experience high rates of HIV infection and in utero HIV exposure in pediatric populations but lack resources for screening of neurocognitive sequelae. This study employed a novel application of the CFIR in the pre-implementation phase to elicit perspectives on factors likely to impact the successful implementation of the PennCNB into community settings in Botswana. Stakeholders expressed a strong need for cognitive screening and high acceptability of the PennCNB tool, suggesting the added value of the tool in this high-need setting. Barriers and facilitators to implementation, such as limited time and resources and standard procedures for introducing new tools into medical settings, were highlighted. Respondents provided insightful input for the development of strategies for future implementation in this and similar resource-limited settings.

The perceived barriers of insufficient time and resources are consistent with perspectives articulated in other studies (Alatawi M, 2020; Cooke et al., 2019; Geerligs, Rankin, Shepherd, & Butow, 2018; Leonard, de Kock, & Bam, 2020; McNeely et al., 2018). The PennCNB can serve as an alternative to traditional neurocognitive screening tests that require multiple hours to administer via paper and pencil, so implementation of this tool will streamline assessments substantially. Modifications to the tool to shorten the administration time may be possible, but they will require assessments to ensure validity. Previous modification of the PennCNB involved the administration of the battery on a tablet (Basner et al., 2015), so future research could explore similar adaptations to increase accessibility and mitigate concerns over time and shared computer resources. Further, institutional-level support for designated time for implementation (e.g., specific day) or a phased approach to integration may help address the barrier of time (Geerligs et al., 2018). The PennCNB does not require a psychologist for administration, so task-shifting to personnel currently working in the setting may be the most effective strategy. Both the medical stakeholders and literature emphasize integrating interventions into the current workflow (Geerligs et al., 2018; Perry et al., 2019) and task-shifting (Benjamin Wolk et al., 2018; Murray et al., 2014; Patel, 2009). Extensive research has demonstrated success of task-shifting in similar resource-limited settings and LMICs (Mdege, Chindove, & Ali, 2013; Seidman & Atun, 2017). To address the concern about healthcare workers being overworked, incentive-based implementation strategies could be explored. Respondents also recommended both schools and clinics as implementation locations, which may alleviate time constraints and personnel burden in one setting. Finally, integration could include a packaged implementation of the PennCNB with effective support strategies (Van Pelt et al., 2020).

The successful implementation of the PennCNB in Botswana will offer promise for uptake in other resource-limited settings. The government of Botswana invests heavily in its medical system. Historically, this has allowed for the implementation of innovative programs that, subsequently, surrounding countries adopt after demonstration of their effectiveness (e.g., universal HIV testing and prevention of mother-to-child transmission services (Centers for Disease & Prevention, 2004; Creek et al., 2007)). Further, the successful translation of the PennCNB into Setswana (as well as Xhosa (Campbell et al., 2017)) suggests feasibility of translation into other Bantu languages. Thus, results from this research may influence integration in other countries in Sub-Saharan Africa.

This research has limitations. First, the interviews were conducted concurrently with local validation, so participants’ responses reflected their beliefs about implementation rather than experience with implementation. However, some of the respondents (35%) had prior exposure to the PennCNB through research collaboration or participation in the stakeholder group that completed the in-depth cultural adaptation process, and the PennCNB was described in detail before conducting the interviews. Further, this research helps address the research-to-practice gap (Lane-Fall, Curran, & Beidas, 2019) by engaging in pre-implementation inquiry and designing for implementation. Second, social desirability bias and interviewer bias may have existed. The interviewer’s impartiality and lack of involvement in the development of the PennCNB was communicated to encourage honest responses. Third, sampling occurred in the capital city, Gaborone, which may limit the generalizability of the findings. However, initial rollout of the PennCNB is targeted for the greater Gaborone area, and stakeholders were recruited from the public sector. Fourth, the sample comprised few policy makers, but some of the clinician stakeholders had previous government experience.

This is the first effort to understand how to plan for implementation of the PennCNB into practice. Results will inform the development of specific implementation strategies to maximize the likelihood of successful integration of the PennCNB into clinical settings in Botswana, ultimately providing access to cognitive screening for a high-need pediatric population.