Get notified

Clinical Reasoning Explained: Models, Errors and Exams

Clinical reasoning explained: how doctors build a diagnosis, the common errors, and how to prepare for multiple-choice, short-answer and short-case exams.

· updated

You’ve read the case, recognised a likely diagnosis and chosen an investigation. Then your supervisor asks why that test matters, or what you’d do if it came back normal, and your explanation stalls. You know the condition, but you haven’t yet made your reasoning visible.

That gap is useful to identify. Clinical reasoning is a trainable process, not a talent you either have or lack. When you can explain the steps between the patient’s story and your decision, a supervisor can give specific feedback and an examiner can assess more than your final answer.

Table of Contents

What Clinical Reasoning Is and Why It Matters

Clinical reasoning is the process of turning a patient’s story, examination findings and test results into a working diagnosis, a management plan and a safety net for when the picture changes.

Think of a navigator updating their position with each new reading. The first fix gives direction, but it isn’t a reason to ignore later information. Similarly, your first diagnostic impression is a starting point, not a commitment.

More than naming the disease

Reasoning includes deciding which history matters, examining with purpose, identifying red flags and ranking competing explanations. It also includes recognising when you haven’t resolved enough uncertainty to send someone home without follow-up.

A useful diagnosis must lead somewhere. If you suspect pneumonia, you still need to consider severity, alternative explanations, the suitability of outpatient care and what deterioration should prompt reassessment.

The process connects information with action.

Why deliberate practice matters

A 2020 Medical Journal of Australia Perspective by Ian Scott and Carmel Crock estimated about 140,000 cases of diagnostic error in Australia each year, including about 21,000 involving serious harm. It reported that more than 80% of diagnostic errors are deemed preventable, and that cognitive factors in clinician decision-making are primary or contributory causes in more than 75%.

The authors name failure to formulate an adequate differential diagnosis and overconfidence in incorrect diagnoses as major contributors, and list biases including premature closure and anchoring. These are not gaps in disease knowledge. They concern how clinicians select, interpret and use what they know.

For students and junior doctors, that distinction matters. Reading another textbook chapter may help when you don’t know a condition, but it won’t automatically correct a habit of ignoring contradictory findings.

Self-check: After a case, state your working diagnosis, your plan and your safety net in a sentence each. If one is difficult to explain, identify what information or decision is missing.

You’re not trying to sound certain. You’re trying to make your decisions understandable, proportionate to the evidence and open to revision.

How Reasoning Moves from Story to Diagnosis

Consider a hypothetical older man with hypertension who describes recent exertional chest tightness. Before calling it angina, establish whether he’s currently symptomatic, whether the pattern is changing and whether he needs immediate escalation.

The conversation should make those decisions understandable to the patient.

A male doctor in a white coat explains a clinical reasoning process chart to a patient.

From narrative to a defensible plan

Use this sequence as a guide, not a rigid script. An unstable patient needs action before a polished summary.

  1. Gather the story. Begin openly, then clarify onset, triggers, duration, associated symptoms and relief. The common slip is either collecting everything without direction or asking only questions that confirm angina.

  2. Examine with purpose. Assess observations, cardiovascular findings and signs of heart failure according to the presentation. Don’t perform a memorised examination without knowing which possibilities the findings would support.

  3. Identify red flags. Ongoing pain, pain at rest, syncope or instability should prompt urgent assessment appropriate to the situation. Don’t postpone escalation until you’ve completed your differential.

  4. Write a problem representation. Summarise an older hypertensive man with recent exertional chest tightness, including whether symptoms are progressive and whether he’s currently stable. A useful summary selects distinguishing features rather than repeating the entire history.

  5. Compare differentials. Consider coronary ischaemia, including whether the presentation suggests an acute coronary syndrome, aortic stenosis and oesophageal reflux. Exertional symptoms support a cardiac explanation; a murmur would strengthen a valvular hypothesis, while meal-related burning would support reflux without excluding cardiac disease.

  6. Choose discriminating investigations. Explain what an electrocardiogram (ECG) could change and whether suspected acute coronary syndrome requires urgent hospital assessment and troponin testing through the appropriate pathway. Don’t use a normal resting ECG to rule out coronary disease or let outpatient testing delay emergency care.

  7. Agree the plan and safety net. Explain the next action, follow-up responsibility and symptoms requiring emergency help. Avoid vague instructions to return if worse.

Context changes the sequence

A two-day international workshop of rural clinician teachers held in Bairnsdale in 2014 (Campbell and colleagues, Rural and Remote Health, 2017) identified the patient’s story, learner’s reasoning, learning context and supervisor’s role as recurring elements in reasoning development. Its account connects narrative and examination with a differential, further investigation and a negotiated plan incorporating evidence and patient preferences.

In rural practice, distance and limited immediate resources make the next safe action especially important. Following patients over time also lets learners see why an initially reasonable decision sometimes needs revision.

For practice, write a problem representation for a patient you assessed recently. Ask your supervisor which omitted detail would most change its meaning.

Cognitive Models Compared and Where Each Breaks

The same chest-pain presentation can be approached through several cognitive models. They describe different ways of thinking, not competing rules that require you to choose a permanent favourite.

With pattern recognition, exertional tightness relieved by rest may bring a coronary disease illness script to mind. An illness script is an organised mental picture of a condition’s typical setting, features and course, rather than a disconnected symptom list.

With hypothetico-deductive reasoning, you deliberately compare coronary disease, valvular disease and gastrointestinal causes. You ask questions and select examination findings to test those possibilities rather than merely accumulate detail.

Cognitive models, how they work and where they break

Model How it works Strongest in Main risk
Pattern recognition and illness scripts Matches the presentation to a stored disease picture Common, classic presentations Misses atypical cases and anchors on the first match
Hypothetico-deductive Forms a few hypotheses early and tests each with history, examination and tests Undifferentiated or complex problems Narrows too soon when the early hypotheses are weak
Dual process (System 1 and System 2) Fast intuitive judgement checked by slower analytic review Recognising when a case feels wrong Relies on the fast system when fatigued or rushed
Probabilistic (Bayesian) reasoning Revises the likelihood of each diagnosis as each finding arrives Ranking tests and weighing findings Hard to apply without prevalence and test accuracy estimates

Dual-process theory offers a practical cue: slow down when something doesn’t fit. If a familiar chest-pain pattern includes unexplained syncope, don’t smooth that detail away to preserve your initial impression.

Bayesian reasoning asks how each finding changes likelihood. A test result doesn’t stand alone, and a negative result may not reduce concern sufficiently when the clinical picture remains worrying.

Recognising findings isn’t the same as integrating them

In a study of 21 experienced general practitioners (Groves and colleagues, Medical Teacher, 2003), more than 20% of misdiagnoses occurred even though no errors were made in identifying and interpreting the relevant clinical information. The difficulty was combining that information into a coherent explanation, rather than retrieving facts.

That’s why a supervisor may ask you to reconcile apparently conflicting findings. Naming each clue correctly isn’t enough if your conclusion ignores how the clues fit together.

A useful explanation connects the findings: this supports my leading diagnosis, this weakens it, and this unresolved feature changes what I must do next.

Experienced clinicians blend these models. The practical skill is recognising when your current approach has stopped helping and deliberately changing how you examine the problem.

Common Reasoning Errors and How to Catch Them

A patient describes burning chest discomfort after meals, then says it also comes on when walking uphill. Calling it reflux and moving on is premature closure. Continuing to favour reflux after hearing about exertional symptoms is anchoring. The distinction matters: one error ends the search too soon; the other keeps the first impression in place despite new evidence.

An unfocused history can sound thorough while leaving the key question unanswered. Asking about every system without establishing whether the pain occurs at rest or with activity gathers details, but not the detail that would change the assessment.

Reasoning errors, what they look like and how to counter them

Error What it looks like Countermeasure
Premature closure Stops at the first plausible diagnosis Ask what else fits, and what does not fit
Anchoring Keeps the first impression despite new findings Restate the problem after each new result
Unfocused data collection Asks broad questions without a hypothesis Name the questions each answer must settle
Poor pattern recognition Misses the typical features of a known condition Compare the case with a named illness script and list the mismatches
Overconfidence Treats a diagnosis as certain without adequate discriminating evidence State your confidence and the result that would change it

Poor pattern recognition may mean failing to connect exertional discomfort and relief with rest to a coronary illness script. Learning the pattern helps; checking for mismatches prevents the pattern from becoming a shortcut.

Overconfidence shows up when you call a diagnosis definite despite missing information or unresolved contradictions. State what remains uncertain, then link that uncertainty to an investigation, review or escalation decision.

Why the sequence deserves feedback

A prospective study of second-year medical students in Western Australia (Gilkes and colleagues, Medical Teacher, 2022) recorded diagnostic reasoning errors in objective structured clinical examinations (OSCEs). At the preclinical OSCE, which examined 235 students, those with unfocused data collection scored 7.7% lower on the history station and those with poor pattern recognition 4.2% lower. Between that OSCE and the end-of-year one, poor pattern recognition became less frequent, while overly narrow reasoning and premature closure increased.

Familiarity with common patterns can make reasoning quicker, but speed can also make alternatives easier to overlook. A plausible diagnosis does not, by itself, show that the reasoning was sound. Feedback should identify whether the learner gathered targeted information, made sense of it and revised the initial view when needed.

For example: “You asked questions that supported your diagnosis, but did not investigate the contradictory symptom.” That names an observable behaviour the learner can change. In self-review, ask: What would I expect to see if my leading diagnosis were wrong?

What Examiners Score in Multiple-Choice, Short-Answer and Short-Case Exams

Knowing the diagnosis doesn’t guarantee that you’ll demonstrate the reasoning an assessment requires. A multiple-choice question asks you to discriminate between supplied options; a clinical station requires you to generate, communicate and act on your own priorities.

The table describes useful preparation targets, not a universal marking scheme. Your examination’s published instructions and scoring criteria remain the reference.

What each format tests and where candidates lose marks

Format What reasoning it tests What a strong answer shows Common gap
Multiple-choice question (MCQ) Applied knowledge and choosing the best option among close distractors In review, justification of the best answer and why distractors fail Choosing by recall without checking stem details
Short-answer question (SAQ) Written reasoning on a defined scenario with specified response requirements Ranked differential, discriminating test and management in clear sentences Listing facts without linking them to the diagnosis
Short case or clinical station Spoken reasoning under time, with examiner prompts Problem summary, targeted questions and a volunteered safety net Stopping after the diagnosis and omitting escalation
Case-based discussion (CBD) — discussing your own real cases with an assessor Reasoning across assessment, investigation and management in a real case Context-specific plan including resource and referral limits Generic plan that ignores local access

For MCQs, identify exactly what the lead-in asks before selecting an answer. During review, explain why the nearest distractor fails in this patient, not why it’s generally incorrect.

For SAQs, connect each answer to the scenario and obey the requested response count. A long list of possible tests can conceal the fact that you haven’t chosen the test that answers the clinical question.

In short cases, speak selectively. Give the examiner your synthesis and decisive findings rather than narrating every thought. If you are preparing for the Australian Medical Council (AMC) clinical examination, the AMC clinical exam guide is a format-specific starting point.

Context-sensitive answers

The Australian College of Rural and Remote Medicine (ACRRM) uses a CBD as a Fellowship assessment. Its 2025 Case Based Discussion public report describes it as an assessment of clinical reasoning: assessing patients, formulating differential diagnoses, ordering relevant investigations and applying management plans. Candidates submit 12 cases from general practice, a small rural emergency department or both; six are selected and discussed across three one-hour online sessions, and a pass needs the expected standard in five of the six. In 2025, 135 of 149 candidates passed (90.6%).

For a rural CBD, an imaging request isn’t a complete plan if imaging isn’t locally available. Explain interim care, senior advice, telehealth consultation where appropriate, transfer arrangements and the circumstances requiring immediate escalation rather than waiting for remote advice.

Aboriginal and Torres Strait Islander health contexts require culturally safe communication and attention to the individual’s circumstances. Ask about preferences, family involvement, access and relevant local services rather than treating identity as a substitute for history.

Transport limitations and referral pathways can change whether outpatient follow-up is safe.

Assessment rule: Make the decision the task requires, then show the patient-specific reason for it. Don’t substitute a generic management list for a workable local plan.

How to Practise Clinical Reasoning Step by Step

Practise the decision process before checking the model answer. Reading a convincing explanation creates familiarity, but it doesn’t reveal whether you could have generated the explanation yourself.

Use the checklist as a repeatable exercise, then record your actual decisions beside it.

A routine you can run this week

  1. Choose a case and time the history. Use a written scenario or a de-identified clinical encounter, with timing matched to your assessment. Progress means reaching the decisive questions without rushing the patient’s opening account.

  2. Write the problem representation before revealing the answer. Leave out details that don’t alter your interpretation. Look for a summary that another clinician can understand without needing the whole transcript.

  3. Generate a short ranked differential. Record supporting and weakening features for each option. You’re improving when every retained diagnosis has a reason to be there.

  4. Name what would most change your view. This might be a historical detail, examination finding or investigation result. A useful answer identifies a genuine discriminator rather than more evidence for an already favoured diagnosis.

  5. Choose the next investigation and its consequences. Write what you’ll do if it’s positive, negative or inconclusive. Progress means you can explain the action attached to the result before ordering the test.

  6. Write management and safety-net instructions separately. Include ownership of follow-up and what the patient should do if specified symptoms occur. Check whether a patient could act on your wording without interpreting medical jargon.

  7. Review against the scoring domains. Mark where you hesitated, skipped a comparison or lost the link between findings and action. Finish with a correction you can test in your next case.

The clinical scenario questions guide can help you select prompts for this routine. Choose cases that require decisions, not just a diagnosis label.

Turn review into a change in behaviour

Explain your answer aloud to a peer and ask them to challenge its weakest link. They should ask what finding you haven’t explained, not merely whether they agree with your diagnosis.

Keep a brief error log with the error type, the observable behaviour and your next countermeasure. For example, record anchoring when you retained the referral diagnosis despite contradictory findings, then practise rewriting the summary without that label.

For coursework connected to these cases, BromeAI’s study tools for students are one option to use alongside primary-source reading. Keep identifiable patient information out of any external tool, and verify clinical claims against the original documents.

A manageable starting cadence is several cases across the week followed by a dedicated review session. Judge progress by earlier synthesis, fewer unexplained differentials and safety nets that no longer need prompting.

Build Your Reasoning Routine Before Your Next Exam

Choose your upcoming assessment format first, then work backwards. If you’re preparing for MCQs, practise distinguishing close options under the conditions in the stem. If you’re facing SAQs or short cases, practise producing concise written or spoken decisions without relying on recognition.

For case-based discussion, select cases where context affected care. Be ready to explain an alternative you rejected, a limitation you worked around, and the point at which you sought help.

Choose material you can check

Before committing to a resource, check its relationship to your actual assessment.

The medical exam preparation guide can help organise that format-first decision. Keep the preparation plan narrow enough that you can review your errors rather than complete more questions.

This week, choose a case in your upcoming format, complete the practice routine and log the specific error you’ll target next. That gives your next study session a purpose beyond getting through more material.

CliniRoo is building preparation material for Australian medical school, specialty selection and college assessments. Nothing is on sale yet. If you want to be told when there is something to use, CliniRoo takes an email address and does nothing else with it.