Research Mentorship for Undergraduate Medical Students in Low-Income Settings: A Perspective from Uganda Authors: Pascal Mathew Okorobe Published: 2026 ABSTRACT AbstractResearch mentorship is widely regarded as the single most effective route into research for medical students, yet it remains least accessible in low-income settings, where mentor time, institutional funding, and ethical infrastructure are all constrained. This Perspective, a descriptive account of early implementation experience rather than a formal program evaluation, presents the design and implementation of a phased, student-led research mentorship program for medical students at Makerere University, Uganda, and argues that closing this gap requires redesigning mentorship models rather than importing well-resourced ones. The program sequenced training to begin with evidence synthesis, which required neither funding nor ethical approval, before progressing to primary research, and distributed mentorship across small peer groups rather than one-to-one faculty pairings. Implementation nonetheless surfaced substantial challenges: mentor hesitancy and inconsistent availability, conflicting schedules between mentors and time-pressured students, high attrition, with only 33 of 75 enrolled students (44%) attending all five webinars, and difficulty absorbing students who joined after the program began. Five months in, only 5 of 13 groups had completed the deliverable needed to begin the program’s second phase. Drawing on these findings alongside comparable near-peer and institutionally anchored mentorship models elsewhere, this Perspective argues that mentorship in resource-constrained settings requires deliberate structural design, tiered near-peer mentoring, proactive institutional anchoring, and modular entry points, rather than reliance on volunteer goodwill, and proposes these as concrete considerations for the program’s next phase. Introduction Research is a cornerstone of evidence-based medicine, and early exposure to it during undergraduate training is associated with stronger critical thinking, communication, and scientific competence.1 Structured research training is linked to measurably higher scholarly output among participating students,2 and early research involvement is increasingly viewed as essential to producing the clinician-scientists and health system innovators medicine will need.3 For undergraduate medical students in many low- and middle-income countries (LMICs), however, meaningful research participation remains structurally out of reach. Commonly cited barriers include limited access to structured training, inadequate mentorship, and resource constraints that make even modest research projects difficult to conduct.4 Difficulty identifying and accessing a research mentor is repeatedly reported as among the most significant of these obstacles.1 Conventional research training models compound the problem: they tend to be resource-intensive, dependent on funding, institutional infrastructure, and lengthy ethical review processes that few undergraduate programs in resource-constrained settings can readily support.4 Mentorship is often proposed as the solution, and the evidence supports this: mentorship meaningfully improves students’ research skills, confidence, and professional development.5 But the models through which mentorship is conventionally delivered were not originally designed with resource scarcity in mind, and many still assume available faculty time, institutional funding, and administrative capacity for ethics review, none of which can be taken for granted in many LMIC medical schools.6,7 Some initiatives have adapted mentorship specifically for such settings, most often by changing where mentors are drawn from or how the initiative is institutionally backed, and these offer useful precedent. A near-peer program within the Makerere University Surgery Society, itself a student society, cascaded mentorship from residents to senior students to junior students.8 The International Student Surgical Network’s peer mentorship program similarly drew mentors from within its own trainee membership, selecting them by a research-knowledge screening test rather than recruiting external researchers, and enrolled participants largely from LMICs.9 The Community of Young Research Peers in Tanzania combined vertical mentoring from senior researchers with horizontal peer mentoring, formally backed by a multi-institution consortium from the outset.10 This paper argues that closing this gap requires rethinking not whether mentorship matters, but how it is structured. Drawing on the design and implementation of a phased, low-cost, peer-scaffolded research mentorship program for undergraduate medical students in Uganda, I argue that scalable research capacity-building in resource-constrained settings depends less on replicating well-resourced mentorship models than on redesigning them: sequencing skill acquisition to defer costly and administratively heavy stages, distributing mentorship load across small peer groups rather than concentrating it in scarce faculty time, and building tangible outputs into each stage to sustain engagement. The remainder of this Perspective describes the gap that prompted this approach, the program’s design logic, the challenges encountered in implementation, and the broader lessons these challenges offer for research mentorship in similar settings. Context: The Gap We Set Out to Fill At Makerere University College of Health Sciences, undergraduate medical students are introduced to research through a mandatory two-phase program undertaken after the third year, combining research methods training with Community Based Education, Research and Services (COBERS) program over roughly twelve weeks.11 For many undergraduate students, this program functions more as a curricular checkbox. Most students approach it as a graduation prerequisite rather than a foundation for scholarly work, and its design makes producing a publishable output difficult: before conducting community-based research activities, students operate under introductory letters from the university to community authorities rather than formal institutional review board (IRB) approval. Even a student motivated to publish afterward therefore faces an immediate barrier: the underlying data were never collected under research ethics approval, which most journals require. Separately, formal training in review writing is limited undergraduates, despite evidence synthesis being one of the few research outputs students could realistically produce without funding, infrastructure, or ethical clearance. These gaps are not unique to Makerere; they reflect a broader pattern across LMIC medical education, where time-limited curricula and the absence of structured, output-oriented training keep undergraduate research productivity low.4 What was missing was not enthusiasm for research or awareness of mentorship’s value, but a training model structured around what students in a resource-constrained setting could actually do: begin with an activity that requires no funding and no ethics approval, build genuine competence through it, and only then progress to the more resource-intensive work of primary research. This is the gap our research mentorship program was designed to close. The Program: Design Logic This was a student-initiated program, developed and run as an activity of the Makerere University Medicine Society (MakMS), an undergraduate student society, rather than a formal curricular offering of the College of Health Sciences. At the time of its implementation, I was the chairperson of this society and a lead coordinator of this program. Recruitment and Cohort Interested MakMS members registered through an online form. Seventy-five students registered, spanning all five years of undergraduate study (Year 1: n = 8; Year 2: n = 20; Year 3: n = 10; Year 4: n = 25; Year 5: n = 18). Baseline Competencies A pre-program questionnaire assessing baseline research competencies was administered before Phase 1 began, completed by 42 of the 75 enrolled students (56% response rate); a corresponding post-program questionnaire was planned but had not yet been administered, since no group had completed Phase 1 at the time of writing. Baseline responses indicated substantial unmet training need: 69% of respondents (29/42) had never attended any formal research training before, and 43% (18/42) had never participated in a research project in any capacity. Self-rated confidence was correspondingly low: 67% rated themselves only “slightly” or “not at all” confident searching medical literature databases, and 55% had never heard of a reference manager. Across twelve specific research skill domains, self-rated competence (0–100 scale) averaged 37%, ranging from approximately 28% for reference management and navigating ethical approval to approximately 55–56% for basic descriptive statistics. Group and Mentor Allocation Registered students were organized into 13 working groups of six to eight members, deliberately mixing senior students (years four and five, who had encountered basic research methods through the College’s COBERS program) with junior students who had not. Each group was assigned a mentor: a senior researcher with an established publication record who had agreed to take on the role. Mentors were drawn both from Makerere University faculty and from researchers unaffiliated with the institution, selected pragmatically on who the organizers believed could mentor effectively and who was willing. Fifteen mentors were approached. Program Structure The program’s central design decision was deliberate rather than incidental: the ordering of its two phases. Phase 1 began with evidence synthesis. All enrolled students were trained through five webinars, delivered to all enrolled students over ten weeks, aiming to produce a complete draft review manuscript (narrative, scoping, or systematic). Sessions covered, in order: (1) introduction to research and scientific writing, (2) literature search and reference management, (3) an overview of review types, topic selection, and narrative reviews, (4) scoping reviews, and (5) systematic reviews. Each session carried a defined output produced by each group under mentor guidance, so that progress could be tracked cumulatively: ten topic ideas after session 1, an annotated bibliography after session 2, and, across sessions 3 through 5, a final topic with justification followed by a full draft manuscript. This phase required neither funding nor ethical approval, so it could begin immediately and admit students regardless of institutional research infrastructure. Only in Phase 2, once students had produced a substantive piece of scholarly writing, would the program progress to designing feasible, low-cost primary studies, culminating in a full research proposal and grounding in the ethical approval process. Conventional undergraduate research training usually introduces primary research first, since this is what “doing research” is typically taken to mean. But primary research is precisely the stage at which resource-constrained students are most likely to stall: it demands funding, ethics review, and sustained institutional support that a student-led society could not itself guarantee. By reversing this order, the program let students build genuine competence in literature searching, critical appraisal, and scientific writing, and produce a tangible, near-publishable output, before needing to clear the higher-friction barriers of funding and ethics approval. Defining and Tracking Success For Phase 1, the program’s operational marker of success was a completed draft review manuscript per group, the deliverable required to advance to Phase 2. Attendance at each webinar was recorded via an attendance form, and each group’s per-session output served as a completion marker tracked informally by mentors and organizers rather than through a centralized monitoring database. This approach, while sufficient to identify the attrition and stalling patterns reported below, was not a formal monitoring-and-evaluation framework, a limitation discussed in Lessons and Broader Argument. Additionally, at completion, a post-program questionnaire will be administered to assess the competences developed. Whether this design held up in practice, and what it took to make it work, is the subject of the sections that follow. Implementation: What Happened Mentor Willingness and Availability Of the fifteen mentors approached, thirteen (87%) agreed to take on a group, a response rate that was, on its face, encouraging. But the reasons given by those who declined were revealing: busy schedules, and, more pointedly, prior experience with medical student mentees. One prospective mentor put it directly, warning that “medical students are tricky; they get so overwhelmed with requirements of school hence may not commit”. This was not merely a scheduling problem but a credibility problem: the program was asking senior researchers to invest time against a widely held expectation that medical students would not follow through. Even among mentors who accepted, several groups reported mentors who later stopped responding, suggesting that initial willingness did not reliably translate into sustained engagement. Scheduling Conflicts Mentors repeatedly cancelled sessions because of shifting commitments. On the student side, clinical and academic demands regularly displaced the program. Students described this directly: one reported that “our clinical teaching exceeded the intended time”; another explained having “to attend a group discussion for our upcoming exam” instead of a session; a third described becoming “so overwhelmed for the day. I couldn’t take in anymore. I had to rest”, rather than attend. These accounts suggest that non-attendance was rarely a deliberate withdrawal from the program, but an ordinary casualty of a curriculum competing, without protected time, against clinical and academic obligations that could not be deferred. Attrition These pressures fed directly into the program’s most serious challenge. Attendance records showed that only 33 of the 75 enrolled students (44%) attended all five webinars. Five months into the program, of the 13 groups originally formed, only 5 (38%) remained actively engaged and working toward a finished manuscript; the remaining 8 groups (62%) had effectively stalled, undermined by the combined effects of attrition and breakdowns in mentor–mentee coordination. Phase 1 had been designed to run over ten weeks, with Phase 2 beginning once groups produced a draft manuscript; at five months, no group had yet done so, and Phase 2 had accordingly not begun. This gap between intended timeline and actual progress is itself a finding: it suggests these challenges were not minor friction that resolved with patience, but structural pressures substantial enough to more than double the program’s planned duration for groups still in progress, while removing a majority of groups from active participation altogether. Late Joiners Attrition was compounded by a rolling-enrolment problem the program had not anticipated. As word spread informally within the society, some students who had not registered at the outset developed interest partway through and asked to join. The program accommodated them, but these late entrants had missed both earlier training content and the outputs their groups had already produced, and consistently struggled to catch up with peers present from the start, a direct cost of a curriculum in which each session built cumulatively on the last. Adaptations In response, the program adapted rather than continuing unchanged. Struggling groups were reorganized, consolidating committed members from multiple stalled groups into fewer, more functional ones. Organizers sought additional mentors selected specifically for their flexibility, on the understanding that availability, not willingness in principle, had determined which mentor relationships survived. Discussions also began toward formally integrating the program into one of the College’s established research organizations, intended to give the initiative institutional standing beyond its origins as a student society activity, though this integration had not yet been formalized at the time of writing. Data Limitations These observations were gathered primarily through attendance records, informal mentor reports, and direct informal interactions with students who could be reached after missing sessions, rather than through a structured qualitative evaluation. They are therefore subject to recall bias among students explaining their own non-attendance, and to the possibility that students who disengaged entirely, and were hardest to reach, are underrepresented relative to those who remained loosely connected to the program. Lessons and Broader Argument The challenges encountered in Phase 1 were not, on examination, failures of individual commitment. Most mentors approached said yes; most students who enrolled did so in earnest. What implementation revealed instead was that the program’s structure had not been built to withstand ordinary, predictable behavior from time-pressured people. This distinction matters, because it changes what the lesson is. The lesson is not that mentors need to be more dedicated or students more disciplined; it is that mentorship models in resource-constrained, volunteer-driven settings must be designed on the assumption that mentor time will be inconsistent and student attention interrupted, rather than treating these as exceptions to be managed as they arise. The mentor recruitment experience illustrates this directly. The mentor who described medical students as “tricky” was not expressing an unfounded bias; he was describing, in advance, the attrition pattern the program went on to observe. That this expectation was volunteered unprompted suggests it reflects real prior experience with unstructured mentorship arrangements, not a one-off objection. The deeper problem is that the program asked senior researchers to commit unpaid time to a ten-week relationship with no institutional recognition, no defined minimum touchpoints, and no formal role beyond informal goodwill. Under those conditions, both the initial reluctance of two mentors and the later non-responsiveness of others who had accepted were predictable outcomes of a structure offering mentors no scaffolding beyond their own motivation. Tellingly, the program’s own response was not to appeal for greater mentor commitment but to seek mentors specifically for their flexibility, an implicit acknowledgment that availability, not willingness in principle, had been the actual constraint. This raises a natural question: would a differently structured model have avoided these pressures? Comparable programs elsewhere suggest a partial answer. A near-peer mentorship program recently implemented within the Makerere University Surgery Society, at the same institution, deliberately distributed mentorship across three tiers, residents mentoring senior students, who in turn mentored junior students, rather than relying on a single tier of externally recruited senior researchers.8 This tiering addresses the exact bottleneck this program encountered: mentor availability concentrated in a small pool of already time-pressured senior researchers, where a tiered structure would instead distribute mentorship across a much larger, more available pool of senior students and residents, each mentoring only slightly junior peers rather than an entire group from a position of significant seniority. Similarly, the Community of Young Research Peers programme in Tanzania combined vertical mentoring from senior researchers with horizontal, peer-to-peer mentoring among trainees, and was formally backed from the outset by a consortium of partnering institutions rather than assembled informally by a student society;10 its subsequent evaluation explicitly recommends this dual-mentoring, institutionally anchored structure for other resource-limited settings.10 A recent global evidence synthesis commissioned by the World Health Organization Special Programme for Research and Training in Tropical Diseases reaches a related conclusion, framing research mentorship as a collective institutional responsibility rather than an individual or volunteer undertaking, and arguing that resource-limited institutions can strengthen mentorship not primarily through new funding but by more deliberately leveraging capacity that already exists within them.7 Read against this literature, the difficulties this program encountered look less like an isolated implementation failure than the predictable result of omitting design features that comparable programs have converged on. This remains a plausible reading rather than a demonstrated one: none of these design changes has been tested within the present program, and the settings compared here differ in scale, country, and institutional context, so lessons drawn from one university’s experience, or from a single international student network, may not transfer directly to another without local adaptation. The late-joiner problem points to a related but distinct design flaw. Students who developed interest after the program had begun were not turned away, but consistently struggled to integrate, because strict week-by-week sequencing meant every session depended on the cumulative output of the one before. This was not a failure of initiative; it was a direct structural cost of a design optimized for low barriers to entry at the very start, at the expense of low barriers to entry at any point. In a population whose availability is inherently unpredictable, given rotating clinical schedules and unpredictable examination timing, a linear, single-entry curriculum will systematically exclude precisely the students most affected by that unpredictability. A model intended for resource-constrained, disruption-prone settings should treat modular structure or built-in re-entry points as a core requirement, not a convenience added later. Taken together, these findings support a broader claim beyond this single program. Mentorship is consistently identified as one of the most effective levers for building undergraduate research capacity, and there is little dispute that it works when sustained. The harder, underexamined question, particularly for low-income settings, is what allows mentorship to survive contact with the conditions under which it must actually operate: volunteer mentors with competing professional demands, students with unpredictable clinical and academic loads, and programs run without institutional funding or formal curricular protection. Comparable programs suggest this is answerable in concrete design terms rather than left to willpower: distributing mentorship across tiers of relatively available near-peers rather than a single tier of scarce senior researchers, and securing institutional anchoring proactively rather than in response to attrition already underway. These lessons are not offered purely as retrospective analysis. Because Phase 2 has not yet begun, precisely because so few groups have completed the Phase 1 deliverable meant to trigger it, they represent design decisions the program can still act on before Phase 2 is launched. Introducing a near-peer tier, drawing junior mentors from senior students who have completed Phase 1, rather than relying solely on externally recruited senior researchers, would directly address the mentor availability constraint this experience exposed. Formalizing the institutional integration currently being pursued, rather than treating it as a corrective measure, would give Phase 2, which carries the added burden of ethical review and primary data collection, institutional standing from the start rather than after early signs of strain. Building a modular or staggered entry structure into Phase 2’s design would prevent the same late-joiner exclusion problem from recurring at a stage where the cost of falling behind, in ethics approval timelines and data collection windows, is considerably higher than in Phase 1. This account is necessarily a single-institution, single-program experience, drawn from one cohort’s Phase 1 implementation rather than a formally evaluated comparison of design alternatives; the extent to which these proposed changes resolve the difficulties described here can only be established once they are tried. That limitation does not diminish the argument so much as define its purpose: this Perspective is offered as a design hypothesis grounded in direct implementation experience and supported by comparable models elsewhere, intended to be tested as the program moves, imminently, into its next phase. A further consideration follows from the nature of this account rather than from any gap in it. As a Perspective, this piece is deliberately a first-person account grounded in direct implementation experience. The quantitative figures reported here, attendance, mentor response rates, and group completion, are independent of authorial perspective. Their interpretation, and the design changes proposed in response, are not: both are shaped by the author’s position as the individual who recruited the mentors, observed the attrition, and implemented the subsequent adaptations. A different coordinator, or an external evaluator, might draw different conclusions from the same records. This distinction between record and interpretation has been maintained as far as possible throughout this Perspective, and the proposed design changes should accordingly be read as the reasoned judgment of the program’s designer and coordinator, rather than as findings of an independent evaluation. Conclusion Mentorship remains the intervention most reliably linked to research engagement among undergraduate medical students, yet the models through which it is conventionally delivered assume resources that many low-income institutions cannot guarantee. The MakMS Research Mentorship Program’s early implementation shows what happens when a well-intentioned, low-cost mentorship model meets the ordinary constraints of student and mentor time: not collapse, but persistent friction that substantially outlasted the program’s intended timeline and left most groups unable to progress. Comparable near-peer and institutionally anchored models elsewhere suggest specific design considerations worth pursuing, rather than a simple call for more resources or more committed volunteers: distributing mentorship across tiers of available near-peers, anchoring initiatives institutionally before strain makes this necessary, and building modular entry into program design from the outset. As this program prepares to enter its second, higher-stakes phase, these are offered not as validated solutions but as a design hypothesis and a case study for other student-led initiatives attempting to build undergraduate research mentorship where resources are scarce.