Teaching, supervision and feedback in the emergency department

The emergency department is a teaching environment as much as it is a clinical one. The trainee who intubates, the registrar who runs the arrest, the junior who clerks the abdominal pain, and the medical student who presents a case are all learning on every shift, and the quality of that learning is determined by the consultant who teaches, supervises, assesses and feeds back. The Fellowship examination treats this as a distinct and examinable domain — the Scholar and Professional roles — and the candidate is expected to deploy structured bedside-teaching models, deliver feedback that changes behaviour, calibrate supervision to competence, and use the workplace-based assessment tools that sample real clinical performance. The emergency department compounds the difficulty: the patients are undifferentiated, the pace is fast, the shift is short-staffed, and the teaching moment must be seized in minutes between the sick and the sicker. The frameworks below exist precisely to make that teaching efficient, evidence-based and safe.[3][4]

Definition and scope — teaching, supervision, feedback and assessment

Four distinct activities are often conflated, and the candidate must separate them because each has its own method. Teaching is the deliberate facilitation of learning — the transfer, construction and consolidation of knowledge, skill and professional behaviour at the bedside. Supervision is the oversight of the trainee’s clinical work, calibrated to their competence and graduated toward independence. Feedback is, in Ende’s foundational definition, specific information about a trainee’s observed performance that is used to narrow the gap between present and desired performance — information, not judgement alone.[3] Assessment is the systematic judgement of competence against a standard. The four are linked in a loop: the supervisor observes the trainee’s performance, teaches to a gap, feeds back on what was observed, and assesses progress against an entrustment standard that determines the next level of autonomy. The emergency department is the highest-yield environment in medicine for all four, because the trainee performs the full range of clinical tasks on undifferentiated patients under direct consultant supervision — but only if the supervisor is present, observing, and engaged.[10][11]

Miller’s pyramid organises the trainee’s progression and tells the teacher which tier each tool samples. The base is knows (factual knowledge, tested by examination), then knows how (application, tested by case-based discussion), then shows how (performance in a simulated setting, tested by an OSCE), and the apex is does (performance in real practice). Workplace-based assessment is the only family of tools that samples the does tier, which is why it is the gold standard for judging entrustment and competence — and why direct observation, however underused, is irreplaceable. [1]

Bedside teaching — the three high-yield models

Bedside teaching is the most efficient form of clinical teaching because it places the learner, the patient and the teacher at the same point and it samples real performance. The Fellowship candidate must hold three models in working memory and select the one that fits the moment: the one-minute preceptor for a trainee who has just seen a patient and presents a case; SNAPPS for the learner-centred case discussion; and Peyton’s four-stage approach for teaching a practical skill. Each is brief, structured, and designed to be deployed in the minutes the emergency department allows. [1]

The one-minute preceptor — Neher’s five microskills

The one-minute preceptor, described by Neher and colleagues, is the workhorse model for the post-case teaching encounter in the emergency department: the trainee presents a patient, and the supervisor teaches efficiently by walking through five microskills in sequence.[1] The genius of the model is that it takes the unstructured "what do you think, doctor?" exchange and converts it into a diagnostic exercise that reveals the trainee’s reasoning, corrects it at the point of error, and teaches a transferable rule — all in one to two minutes at the bedside.

The cardinal error in the one-minute preceptor is to skip the first two microskills and jump straight to teaching — which delivers a lecture blind to the trainee’s actual reasoning and misses the gap that the model exists to expose. The commitment and the probe are the diagnostic steps that make the teaching targeted.[1]

SNAPPS — the learner-centred case presentation

SNAPPS, developed by Wolpaw and colleagues for outpatient education and now widely adopted across clinical settings, is the inverse of the one-minute preceptor: it places the structure of the teaching encounter in the trainee’s hands rather than the supervisor’s.[2] The trainee is taught to present the case in a format that explicitly surfaces uncertainty, forces a differential, and invites teaching — so the supervisor’s role is to respond to the learner’s stated needs rather than to interrogate. SNAPPS is ideally suited to the emergency department case presentation because emergency medicine is defined by the undifferentiated problem and the working differential, which is exactly what the format demands.

SNAPPS suits the senior trainee who can drive the encounter; the one-minute preceptor suits the junior who needs the supervisor to lead it. The candidate who can name when to deploy each demonstrates the supervisory judgement the examination rewards.[2]

Peyton’s four-stage approach — teaching a practical skill

Peyton’s four-stage approach is the structured method for teaching a practical skill — suturing, central-line insertion, the airway manoeuvre, the splint application — and it appears repeatedly in the Fellowship OSCE "teach a skill" station.[1] The principle is that the learner performs the skill, they do not merely watch it, and the progression from passive observation to active performance is staged so that the motor and cognitive components are separated and rehearsed.

The error is to stop at stage 2 and then hand the skill to the trainee: the trainee who has only watched and heard has not yet supplied the cognitive script themselves, and the first unsupervised attempt is the first real test. Stage 4 — the trainee performing while narrating — is the step that confirms the skill is learned, and it is the step most often omitted.[1]

Feedback — the principles and the models

Feedback is the single highest-yield teaching intervention, and the evidence is now meta-analytic. Wisniewski, Zierer and Hattie’s synthesis of 435 studies with over 61,000 learners found that feedback produces a medium effect on learning (d = 0.48), placing it among the most powerful single interventions in education — but the heterogeneity was large, meaning that how feedback is given determines whether it helps, does nothing, or harms.[8] The principles that separate effective from ineffective feedback, established in Ende’s seminal paper and confirmed across the modern literature, are that feedback must be specific (tied to an observed behaviour, not a global judgement), actionable (the trainee can act on it), timely (delivered soon after the observation, while the behaviour is fresh), behaviour-focused (about what was done, not who the trainee is), and balanced (reinforcing and corrective).[3][4]

Pendleton’s rules

Pendleton’s rules, set out in the consultation-teaching tradition, structure the feedback conversation into four sequential moves designed to build receptivity before criticism.[1] The model opens with the trainee and forces the supervisor to listen before speaking, which is its principal pedagogical value and its principal weakness — the rigid positive-first structure can feel formulaic and can bury the correction the trainee most needs to hear.

R2C2 — the facilitated reflective model

R2C2, developed by Sargeant and colleagues and refined for in-the-moment use by Lockyer and colleagues, is the modern, evidence-based feedback model that supersedes the rigid tell-after structure of Pendleton.[5][6] It is built on the recognition that feedback is effective only when the recipient engages with it, accepts it, and acts on it — and that engagement depends on relationship, on the management of the emotional reaction to being assessed, and on coaching for change rather than pronouncing judgement. R2C2 is the model the Fellowship candidate should reach for when the feedback is substantial, the stakes are higher, or the trainee has reacted defensively to feedback before.[7]

The feedback sandwich — and why it is deprecated

The feedback sandwich — a positive, then the correction, then a positive — was once the dominant model and is still widely used, but it is now deprecated in the medical-education literature.[4] Its problems are well characterised. It primes the trainee to dismiss the correction as the predictable filling between two pieces of praise, so the message that most needs to land is the one least attended to. It erodes credibility, because the trainee learns that the positives are a delivery vehicle rather than genuine observation. And it encourages vague positives — the requirement for a positive at each end pushes the supervisor toward the non-specific "good job" that violates every principle of effective feedback. The modern recommendation is to deliver the specific, behaviour-focused correction directly, surrounded by genuine, specific reinforcement where it is earned — not by a mandatory structural positive.[4][8]

Supervision — graded autonomy, observation and entrustment

Supervision is the oversight of the trainee’s clinical work, and its central principle is that the level of supervision is matched to competence, not to seniority or to convenience. The emergency department is the environment in which supervision is most directly patient-facing: the trainee who is under-supervised on a procedure, or over-supervised on a case they could manage, represents a safety failure and a learning failure respectively. Gingerich and colleagues reframed the old binary of "hands-on" versus "hands-off" supervision into a richer account of supervisory approaches — from the most directive (the supervisor does it with the trainee) through intermediate forms of co-management and monitoring, to entrustment — and they showed that the entrustment decision (the supervisor’s judgement that the trainee may perform a task with a defined level of oversight) is the operative event in every supervision encounter.[10]

Direct observation is the supervisor watching the trainee perform — the airway attempt, the history, the procedural-sedation consent — and it is the only source of valid information about how the trainee actually practises. Indirect supervision is the supervisor being available (in the department, on the end of a phone) while the trainee performs independently, then reviewing the work afterward. The modern evidence is unequivocal that direct observation is underused — trainees are far more often indirectly supervised and retrospectively reviewed than directly observed — and that this gap is the principal reason competence is misjudged and entrustment is mis-calibrated.[10][12]

Entrustable professional activities and the entrustment decision

The entrustable professional activity (EPA), formalised by ten Cate and now embedded in competency-based medical education across the Anglophone world, is the unit of clinical work that a trainee can be entrusted to perform unsupervised once a defined level of competence is reached.[9] An EPA — "manage the undifferentiated febrile adult", "perform a focused ultrasound", "lead a resuscitation" — is described with the knowledge, skills and attitudes it requires, and the entrustment decision assigns a supervision level to it. The entrustment scale runs from level 1 to level 5, and the move up the scale is the trainee’s documented progression toward independence.

The entrustment decision is driven by the supervisor’s judgement of the trainee’s competence on the specific task, the context (acuity, complexity, time of day, backup available), and the trainee’s self-assessment. The examiner expects the candidate to articulate why a given trainee is at a given level — the observed evidence, the workplace-based assessment data, the context — and to distinguish the trainee’s level on a simple case from their level on a complex one. Entrustment is task-specific and context-specific, never global.[9][10]

Assessment — workplace-based assessment in the emergency department

Workplace-based assessment (WPBA) is the family of tools that samples the does tier of Miller’s pyramid — real clinical performance, observed in the real environment. The Fellowship candidate must distinguish the three high-yield instruments, because each samples a different facet of competence and each is examiner-fair game.[11][12]

The mini-CEX (mini Clinical Evaluation Exercise), validated by Norcini and colleagues, is a focused 15- to 20-minute observation of a single clinical encounter — a history, an examination, a clinical decision and a communication — followed by immediate structured feedback and a global rating.[11] It samples clinical reasoning, communication and professionalism in the real encounter. The DOPS (Direct Observation of Procedural Skills) is the analogous instrument for a practical procedure: the supervisor observes the trainee perform a defined procedure against a structured checklist, and provides immediate feedback on the technical and non-technical components. The CBD (Case-Based Discussion, sometimes ACAT — Assessment of Clinical Assessment and Treatment) samples clinical reasoning through a structured discussion of a case the trainee managed, probing the decision-making, the diagnostic logic and the synthesis.[12]

The crucial finding of Lörwald and colleagues' qualitative synthesis is that the educational impact of mini-CEX and DOPS depends not on the rating but on the quality of the dialogue and feedback that follows the observation — the tool assesses nothing and teaches nothing if the observation is not followed by the conversation.[12] The four themes they identified — context (time, usability), users (supervisor and trainee knowledge and attitudes), implementation (observation and feedback), and outcome (educational impact and the trainee’s appraisal of the feedback) — explain why the same instrument produces rich learning in one department and a tick-box exercise in another. The take-home for the candidate is operational: WPBA works only when it is observed, immediate, conversational and specific.[12]

Management — graded supervision, entrustment and the bedside teaching encounter

The applied skill the Fellowship examination tests is the integration of these frameworks into a single supervisory encounter in the emergency department. The effective supervisor moves up and down the entrustment ladder by task and by context: the registrar at EPA level 3 on a simple laceration is dropped to level 1 on the difficult airway; the trainee at level 4 on a stable chest infection is dropped to level 2 when the same patient deteriorates. Over-supervision of the competent trainee stalls their progression and frustrates them; under-supervision of the novice exposes the patient to harm. The match is to competence and context, recalibrated on every shift, and the evidence for the match is the direct observation that feeds the entrustment decision.[10][9]

The effective supervisor is explicit about the entrustment level they are applying and why — they tell the trainee "I am going to watch you do this one, because I have not seen you do it before" (level 1, near-total supervision) or "you run this, I am in the department, call me if you need me" (level 3, reactive) — because the named level makes the supervision legible, lets the trainee push for the next level, and prevents the silent drift toward either over- or under-supervision. The move from level 1 (around 100% oversight) toward level 5 (independent, 0% direct supervision) is the documented arc of training, and the supervisor’s job is to advance the trainee deliberately, on evidence, never by default.[9][10]

Differential — good versus poor teaching

When a trainee fails to progress, or reports a poor learning experience, or makes an error traceable to a supervision gap, the supervisor must work through the differential of teaching failure, because each pattern carries a different countermeasure. The patterns below are the high-yield diagnoses the Fellowship candidate must distinguish.[4][10][12]

Common errors and pitfalls

The recurring failures are those the frameworks exist to prevent. The non-specific feedback — "good job", "keep it up" — carries no behaviour-specific information and changes nothing; the correction is specific, actionable and tied to an observed behaviour. The feedback sandwich primes the trainee to dismiss the correction and erodes credibility; it is deprecated, and the specific correction is delivered directly. The absent feedback — the observation that is never followed by the conversation, or the feedback deferred to the end-of-term form months later — teaches nothing because the behaviour is long forgotten; feedback is timely, ideally in the same shift. Over-supervision of the competent trainee frustrates and stalls progression; under-supervision of the novice exposes the patient to harm; the level is matched to competence and context, recalibrated on every shift. Observation without feedback assesses nothing and teaches nothing — observation and feedback are a single unit. The lecture instead of the microskill delivers facts blind to the trainee’s reasoning and misses the gap; the commitment and the probe come first. Peyton stopped at stage 2 hands over a skill the trainee has only watched and heard, not narrated and performed; stage 4 confirms the learning. The halo effect — the trainee who is liked, or who is from one’s own alma mater, is rated above the evidence — corrupts the entrustment decision; the level is set on observed performance, not on regard. The public or humiliating correction destroys psychological safety, suppresses the surfacing of uncertainty, and produces the trainee who hides their errors. The undocumented entrustment level leaves the supervision implicit and inconsistent; the named level makes the supervision legible and the progression deliberate.[3][4][10][12]

Evidence and regional guidelines

The evidence base for teaching, supervision and feedback in medical education is grounded in Ende’s foundational definition of feedback, in Neher’s one-minute preceptor and Wolpaw’s SNAPPS for bedside teaching, and in Peyton’s four-stage approach for skill teaching.[3][1][2][1] The feedback-models literature is synthesised in Orsini and colleagues' review of common models for the clinical educator, which catalogues Pendleton’s rules, the feedback sandwich and R2C2 and confirms the sandwich’s deprecation.[4] The R2C2 model is developed and refined in Sargeant and colleagues' original evidence-based formulation and in Lockyer and colleagues' adaptation to in-the-moment feedback and coaching.[5][6][7] The quantitative case for feedback is made in Wisniewski, Zierer and Hattie’s meta-analysis (d = 0.48).[8] The supervision and entrustment literature is anchored in Gingerich and colleagues' account of supervisory approaches and the entrustment decision, and in ten Cate’s entrustable-professional-activity framework.[10][9] The workplace-based assessment evidence rests on Norcini and colleagues' validation of the mini-CEX and on Lörwald and colleagues' qualitative synthesis of the factors that determine the educational impact of mini-CEX and DOPS.[11][12]

Exam pearls

• One-minute preceptor = Neher’s five microskills: get a commitment, probe the supporting evidence, teach a general rule, reinforce what was right, correct errors — in that order; the commitment and the probe are the diagnostic steps.
• SNAPPS = Summarise, Narrow the differential, Analyse the differential, Probe the preceptor, Plan management, Select a self-study issue — learner-centred, suits the senior trainee who can drive the encounter.
• Peyton = four stages: demonstrate (normal speed, silent), deconstruct (slow, narrated), demonstrate-while-trainee-narrates, trainee-performs-while-narrating — the practical-skill method; stage 4 confirms learning and is the step most often skipped.
• Feedback principles = specific, actionable, timely, behaviour-focused, balanced — the non-specific "good job" fails all five.
• Pendleton’s rules = four in order: trainee says what went well, supervisor says what went well, trainee says what to improve, supervisor says what to improve — rigid, positive-first.
• R2C2 = Relationship, Reactions, Content, Coach (Sargeant, Lockyer) — the modern evidence-based model; coaching for change, not pronouncing judgement.
• Feedback sandwich = deprecated: positive-negative-positive primes dismissal, erodes credibility, encourages vague positives; deliver the specific correction directly.
• Wisniewski meta-analysis: feedback effect d = 0.48 (435 studies, over 61,000 learners) — powerful, but the how determines the outcome.
• Supervision = matched to competence and context, not seniority: over-supervision stalls, under-supervision harms; the level is recalibrated every shift.
• Entrustment scale = five levels: 1 observe only, 2 direct proactive, 3 reactive (the usual on-shift independence threshold), 4 independent with post-hoc review, 5 fully independent — entrustment is task- and context-specific, never global.
• WPBA samples the "does" tier of Miller’s pyramid: mini-CEX (encounter, reasoning, communication), DOPS (procedure), CBD (clinical reasoning and synthesis) — each samples a different facet.
• Lörwald: the educational impact of mini-CEX and DOPS depends on the dialogue and feedback, not the rating — observation without feedback assesses and teaches nothing. [1]

Red flags