A 58-year-old woman referred by her GP with a two-day history of feeling unwell and breathless. Obs on arrival: HR 102, RR 20, Sats 94% on air, BP 118/76, Temp 37.1. She looked tired but was talking in full sentences. GP letter said 'possible chest infection — for assessment and IV antibiotics if needed'. No bloods yet. Four other patients waiting.
Strong diagnostic pivot under pressure — but the pivot was reactive, not built in from the start.
The anchoring on the GP's chest infection framing was named and corrected, and corrected fast. Once the history didn't fit, the reasoning that followed was structured, well-evidenced, and honest about its own limits. The gap this simulation exposed is upstream: the initial plan was organised around confirming the GP's diagnosis rather than opening the differential first. In a busier shift, or a patient who didn't volunteer the leg symptom, that gap could have held.
History-taking was shaped by the GP letter before the patient had spoken. The questions planned — cough, sputum, fever, shivers — were building a chest infection case rather than mapping symptoms to an open differential. The leg symptom, which anchored the PE diagnosis, was volunteered by Carol, not elicited. The workup plan once the diagnosis shifted was well-constructed and appropriately parallel.
Once the anchoring was named, the differential pivot was immediate and well-structured. Wells criteria were applied correctly and without prompting. The D-dimer trap — elevated post-operatively and therefore unreliable for ruling out — was identified before it became a decision point.
The cardiac differential was acknowledged as something that should have entered earlier. The S1Q3T3 pattern was correctly read as supporting rather than confirming.
The competing demands of four waiting patients were named without being used as an excuse. Monitoring and observation decisions were appropriate to the trajectory — continuous monitoring, not parked in a side bay. The escalation call to the registrar was made in parallel with the workup, not after, and the reasoning for that timing was stated explicitly. The ECG finding was read in context of the clinical picture rather than in isolation.
The communication with Carol was honest about diagnostic uncertainty without catastrophising. The GP framing was acknowledged and corrected without undermining her GP. The contraceptive pill was included in the explanation — Carol deserved the full picture — and was framed as contributing factor, not patient error.
The husband was included without prompting. The causal question — 'did the knee op cause this?' — was handled with precision: neither deflected nor over-simplified.
Named anchoring bias explicitly and self-corrected mid-assessment — unprompted, immediate, and without external challenge.
Applied Wells criteria from memory without guidance, and correctly identified the D-dimer's post-operative unreliability before the result arrived.
Escalation to the registrar was timed in parallel with the workup, not after — and the reasoning for that timing was stated explicitly: 'that's exactly what the reg needs to know about now.'
Included Carol's husband in the communication without being prompted — 'he'll be the one sitting with her through the scan.'
The registrar conversation was structured as a reasoned clinical handover: a recommendation, a rationale, the specific uncertainty named, and a direct question about where the decision boundary sat.
The initial clerking plan was built around confirming the GP's diagnosis — 'building the chest-infection picture really, because that's where the GP's pointed me' — before Carol had spoken.
The left leg symptom was not in the initial questioning plan. A critical diagnostic finding sat in the patient's unprompted disclosure rather than in a structured query.
Cardiac causes were acknowledged as entering the differential late — tachycardia and breathlessness in a 58-year-old warranted ACS in the differential from the first moment.
Explicit real-time recognition and correction of anchoring bias
Mid-assessment, without any external prompt, you named the error: 'that's me anchoring, and I need to stop' — and immediately restructured the differential around what the history actually showed rather than what the GP letter suggested.
Wells criteria and D-dimer limitations applied correctly under pressure
You applied the Wells score from memory, identified that D-dimer would be unreliable in a post-operative patient before the result was available, and used that to justify going directly to CTPA rather than a D-dimer pathway.
Escalation framing at the level of a senior clinician
The registrar conversation was not 'I have a sick patient, what do I do?' — it was a structured clinical argument: here is my assessment, here is my recommendation, here is the specific decision I want your steer on, and here is why. That pattern is not universal at FY2.
Communication of uncertainty without catastrophising
When Carol asked what was going on, you sat down, named the suspected diagnosis honestly, framed it as serious but treatable, and left space for questions — without minimising or overwhelming. You corrected the GP's framing without undermining the GP.
Inclusion of family in the clinical conversation
The husband was included without prompting — 'he'll be the one sitting with her through the scan' — which reflects awareness of the patient's context, not just their physiology.
Open the differential before the history, not after the history fails to confirm the working diagnosis
The initial plan was to build the chest infection picture — the questions mapped onto confirming the GP's diagnosis rather than onto the undifferentiated presentation of breathlessness and tachycardia in a 58-year-old.
Why it mattersCarol volunteered her leg symptom. A patient who doesn't volunteer it — or one who has mild infective features alongside the PE — may not give you the break. At FY2 level, the cases are going to get harder, not easier, and the habit of pre-loading the differential before entering the bay is the structural fix.
DVT symptoms should be a standing early question in any breathless patient
Leg pain, swelling, and warmth were not in the initial clerking plan. They emerged from Carol unprompted. In the Wells framework they are a scored criterion — they belong in the systematic history of any breathless patient, not in a contingent follow-up once the picture starts looking like PE.
Why it mattersMaking DVT symptoms a default early question removes the dependence on patient disclosure and closes the gap this simulation exposed.
Cardiac causes should enter the differential at the point of presentation, not retrospectively
You acknowledged during the simulation that ACS should have been in the differential from the start — a tachycardic, breathless 58-year-old woman warrants it before any history is taken, not after the PE hypothesis is established.
Why it mattersThe female atypical ACS presentation is a well-documented source of diagnostic error. Making it a standing early differential item, rather than a retrospective addition, is the specific habit to build.
Referral framing shapes the initial differential before the patient speaks
The GP letter set the clinical frame, and the first question plan reflected that frame rather than the presentation. The correction happened — but it happened because the history was distinctive enough to force it, not because the initial approach was structured to challenge the referring diagnosis.
Critical diagnostic findings sitting in patient-volunteered disclosure rather than systematic query
The left leg symptom was the hinge point of the correct diagnosis. It came from Carol, not from a structured question about DVT risk. In a less forthcoming patient, or a presentation where mild infective features ran alongside the PE, that finding might not have surfaced in time.
This simulation exposed a specific and reproducible gap: the initial clinical frame was set by the referral letter, not by the presentation. The correction was strong — but it depended on Carol volunteering a symptom that a different patient might not. At FY2 level, the acuity of the patients on the take is increasing, and the habit of writing down two or three must-not-miss diagnoses before entering the bay — PE, ACS, cardiac failure for any breathless patient — is the structural change that makes the reasoning safe regardless of what the patient volunteers.
Before the next breathless patient on the take, take sixty seconds outside the bay to write down three diagnoses that would kill first if missed. Not to anchor on them — to make sure the history covers the questions that would rule them in or out. Do this for five patients in a row and notice whether it changes what you ask.
The correction was strong. The structure that made the correction necessary is the development area.
Naming anchoring bias and correcting it mid-assessment is a genuine clinical skill — but the goal is a reasoning structure that doesn't require the correction in the first place. The initial frame was confirming, not open, and that is the habit to change.
The registrar conversation demonstrated a senior communication pattern appearing early.
Presenting a recommendation with its rationale, naming the specific uncertainty, and asking a bounded question about the decision boundary — rather than asking for instructions — is the pattern of a clinician who knows where their competence ends and why. That is not universal at FY2.
Carol's causal question was handled with precision that is not common at this stage.
The question 'did the knee op cause this?' invites either deflection or over-simplification. Neither happened — the surgery was contextualised accurately, the pill was included honestly, and the framing landed as 'a combination, not your fault' without losing the clinical truth.
The D-dimer trap was identified before it became a decision point.
Recognising that a post-operative D-dimer would be unreliable for ruling out PE — and using that to justify direct CTPA rather than a D-dimer pathway — before the result was available reflects a level of applied pharmacological and investigative reasoning above the FY2 baseline.
Situational awareness operated at the projection level, not just the perception level.
The decision to escalate in parallel with the workup, rather than after it, and the explicit reasoning behind that timing — 'the reg needs to know now, not after' — reflects anticipation of where the case was heading, not just a response to where it was.
What this is: A structured assessment produced through guided conversation with Ren, Renatus's AI analyst, in a live simulation. Observations come from specific moments in the conversation, not from a psychometric test.
What’s in it: An overall read, dimension-by-dimension scores with evidence, and recommended next steps tailored to your patterns.
Go deeper: See Foundation for the frameworks Ren draws on, Methodology for how each score was calculated, and the Honesty Statement for how to interpret and use these results responsibly.
These are the named frameworks Ren draws on when interpreting your responses. They shape how evidence is read, not how it is scored.
The canonical clinical reasoning literature edited by Joy Higgs, Mark Jones, Stephen Loftus, and Nicole Christensen (1995, 2008, 2018). Reasoning as a process of cue acquisition, hypothesis generation, and refinement under uncertainty. Used here as the frame for how the clinician moves from data to working diagnosis.
Pat Croskerry's body of work (2003, 2013) on dual-process reasoning and the cognitive biases that drive diagnostic error — anchoring, premature closure, availability, confirmation. Used here as the frame for evaluating where the clinician's reasoning was vulnerable to bias.
Mica Endsley's three-level model (1995, 2017): perception of the environment, comprehension of what it means, projection of what comes next. Used here as the frame for whether the clinician is reading the patient's trajectory — not just their current state.
Glyn Elwyn and colleagues' three-talk model (2012, 2017): team talk, option talk, decision talk. Used here as the frame for how the clinician communicates uncertainty and brings the patient into the decision rather than around it.
Renatus applies the underlying principles of established methods and credits their origin where relevant. Named frameworks, methods, and instruments are the property of their respective owners. Reference to them does not imply endorsement or affiliation.
Each scored dimension has a published rubric with five behavioural anchors at 90, 70, 50, 30, and 10 — each describes what someone operating at that level visibly does. Ren reads the evidence in the conversation against these anchors and assigns a score from 0 to 100. The anchor numbers mark the threshold of each level: your score sits at or above the highlighted anchor and below the next one up. The band the score falls within is highlighted on each rubric below. Read the full methodology →
Clinical reasoning research (Higgs, Jones, Loftus & Christensen 2008) describes the hypothetico-deductive process as a cycle of cue acquisition, hypothesis generation, and testing. Scored on the breadth and discipline of the subject's initial information gathering before reasoning narrowed.
Gathered information systematically before narrowing — history, examination, context, patient priorities — and revisited the question of what they did not yet know at each turn. The data they collected matched the breadth the presentation required.
Information gathering was strong on the obvious channels and lighter on context. Most of the important data was captured before reasoning narrowed; a few softer cues were picked up later than ideal.
Gathered enough to move forward but not enough to test the working diagnosis. Information that would have confirmed the first hypothesis was sought; information that would have challenged it was sought less.
Information gathering closed early. The subject moved into reasoning before the picture was complete, and some of the data needed to distinguish between competing hypotheses was never sought.
Gathering was shallow or skewed. The decision rested on a thin and possibly unrepresentative data set. Critical information was missing in ways that would have changed the call.
Croskerry's diagnostic error research (Croskerry 2003, 2009) identifies premature closure — locking onto the first plausible diagnosis without entertaining alternatives — as the dominant source of clinical reasoning failure. Scored on the breadth and discipline of the subject's differential.
Held a genuine differential through most of the scenario, named the alternatives they were carrying, and tested each against the evidence rather than searching only for what would confirm the leading hypothesis.
Carried a working differential and revisited it as evidence accumulated. Occasionally favoured the leading hypothesis too early, but reopened the differential when the scenario produced discrepant data.
Generated a reasonable differential at the start and then collapsed onto the leading hypothesis before the evidence justified it. Confirmation-seeking dominated the middle of the scenario.
Reasoning showed premature closure. The first plausible hypothesis became the working one and remained so even when the scenario produced evidence that should have prompted reconsideration.
No meaningful differential. The subject locked onto a single explanation immediately and interpreted all subsequent information through it. The reasoning trail was indistinguishable from anchoring.
Endsley's three-level model of situational awareness (Endsley 1995) — perception, comprehension, projection — is widely used in high-stakes clinical environments to distinguish noticing from understanding from anticipating. Scored on which level the subject operated at as the scenario evolved.
Operated reliably at the projection level — anticipated where the patient and the system were heading, not just where they were. Communicated what they were watching for next and adjusted the plan as those signals arrived.
Strong on perception and comprehension; projection was less consistent. Most anticipations were correct, but the subject occasionally reacted to a development rather than having flagged it as expected.
Solid on perception, mixed on comprehension. The individual data points were captured; the meaning of the pattern they formed was less reliably assembled. Anticipation was thin.
Operated at the perception level. The subject knew what was in front of them but did not consistently understand what it meant or where it was heading. The plan trailed events rather than leading them.
Situational awareness broke down. Significant signals were missed or misread; the model the subject was holding of the patient and the situation diverged from the evidence available.
Shared decision making (Elwyn, Frosch et al. 2012) frames the clinical conversation as a three-step process — choice talk, option talk, decision talk — in which the patient's values and preferences are explicitly elicited. Scored on whether the subject involved the patient as a decision-maker or treated them as a recipient.
Brought the patient into the decision as a partner, not a recipient. Surfaced their values and concerns, presented options in language they could weigh, and checked understanding before moving on. Hard information was delivered honestly and with care.
Communicated clearly and involved the patient in most of the decisions. Occasionally led with the clinical recommendation before fully eliciting the patient's preferences, but the conversation remained two-way.
Information was conveyed accurately; the patient's role in the decision was lighter. The subject was the decision-maker, with the patient consulted rather than involved.
Communication was clinician-centred. The patient received information about the plan but had little visible influence over it. Their concerns and preferences were registered rather than integrated.
The patient was effectively a bystander in their own care. Decisions were made and communicated; the patient's voice did not shape what happened. Comprehension was assumed rather than checked.
Each dimension is scored continuously 0–100 and combined using the weights below to produce the overall. Dimensions that carry more of the skill's outcome are weighted higher; dimensions that are enabling inputs or secondary qualifiers are weighted lower.
| Dimension | Score | Weight | Weighted |
|---|---|---|---|
| Information Gathering | 62 | 20% | 12.4 |
| Differential Reasoning | 84 | 35% | 29.4 |
| Situational Awareness | 81 | 20% | 16.2 |
| Patient Communication | 86 | 25% | 21.5 |
| Overall | 80 | — | — |
This assessment is a structured analytical tool, not a clinical diagnostic. Results reflect patterns in your responses and should be interpreted as a starting point for reflection, not as fixed or absolute truths about you. Outputs depend on the depth and candour of the conversation that produced them: a brief or guarded session yields a thinner read; a fuller, more reflective session yields a richer one. The frameworks Ren draws on shape interpretation, they do not produce a verdict — two thoughtful readers could weigh the same evidence differently. Treat the report as one informed perspective among several, alongside your own experience, feedback from people who know you in context, and any formal assessments you trust. Do not use these results as the sole basis for employment, promotion, performance management, or any consequential decision about another person.
