From mechanisms to systems: Reconceptualising supervised workplace learning in postgraduate medical education

Abstract Background Why do the same supervisory mechanisms (feedback, observation, entrustment) produce different learning experiences across clinical settings? Current research treats them as discrete behaviours rather than examining how they interact within local contexts. This study uses realist theory to reframe supervised workplace learning as contextually configured systems, examining how key mechanisms (the processes through which attending physicians and residents interact) couple differently across postgraduate clinical settings. Methods We conducted a realist‐informed multiple case study across four clinical departments in Ireland (Geriatric Medicine, Surgery, Paediatrics and Infectious Diseases). Semi‐structured interviews with residents and attending physicians explored everyday clinical work, learning and supervisory dynamics. Analysis proceeded in two phases: (1) within‐case, theory‐informed inductive analysis using a realist theory on supervised workplace learning as a sensitising concept; (2) cross‐case configurational analysis tracing how six mechanisms (entrustment, support seeking, monitoring, modelling, meaning making and feedback) patterned differently by context. Results Fifty participants (38 residents, 12 attending physicians) were interviewed. All six mechanisms appeared in every setting but were organised into distinct patterns shaped by local conditions. Geriatric medicine: continuous monitoring and narrow entrustment produced high oversight with limited autonomy progression (shaped by patient frailty, strong continuity, containment culture). Surgery: tightly coupled entrustment‐monitoring‐feedback drove staged procedural progression (shaped by observable performance, resource scarcity, competitive selection). Paediatrics: fluid, overlapping mechanisms enabled relational negotiation (shaped by family‐facing work, small teams and psychological safety). Infectious diseases: fragmented, reactive mechanisms produced discontinuous supervision (shaped by severe discontinuity, high workload, frequent roster changes). What distinguished settings was not mechanism presence but mechanism coupling (how mechanisms reinforced or constrained one another), producing systematically different outcomes. Discussion Supervision operates as a contextually adapted configuration of interdependent mechanisms, not a set of discrete behaviours. The same mechanisms organise into different patterns depending on local conditions (continuity, patient complexity, resources, culture), producing different learning experiences. These context‐dependent patterns (not specialty templates) challenge policies that audit mechanism presence rather than mechanism function, positioning supervision as an emergent practice that requires both enabling structural conditions and skilful pedagogical enactment within those conditions.