How to write a clear case summary or referral

A useful case is not necessarily rare. It may show an easily missed common condition, a misleading initial frame, a test used well, a treatment sequence that mattered, or a moment when the team recognized that the expected course was not occurring. State that learning purpose to yourself before writing. If the only appeal is surprise at the final diagnosis, the narrative may become a guessing game rather than a reusable reasoning resource.

Keep the claim proportional to one case. A case can illustrate a possibility or a decision process; it cannot establish prevalence, comparative effectiveness, or a universal rule. Avoid “always,” “never,” and unsupported superlatives. If the lesson depends on a guideline or study, link the authoritative source rather than making the case carry evidence it cannot provide.

Give the reader enough to orient: the main concern, clinically relevant background, time course, and setting. Use an age band unless exact age is necessary and safely publishable. Then tell the story in the order information became available. Chronology lets readers reason alongside the team and prevents hindsight from making the answer look obvious.

Prefer precise observation to decorative detail. “Progressive exertional breathlessness over six weeks” carries more signal than “feeling unwell for a while.” Explain what prompted escalation or referral. Do not paste the full chart; omit routine information that neither shaped the differential nor affected management.

A useful one-liner is not the first sentence of a chart copied into shorter form. It is a concise synthesis of the context that matters, the time course, and the clinical syndrome or unresolved problem. Prefer an age band for a public without identifying details case. Include only background that changes interpretation, and translate raw details into careful clinical abstractions without claiming more certainty than the evidence supports.

Copy this structure: “[Relevant context] with [time course] [syndrome or unresolved problem], characterized by [two to four discriminating findings]; the current question is [decision or uncertainty].” Then test every phrase: would removing it change the listener’s interpretation, and was it actually known at that point in the chronology? If not, remove it or move it to the longer summary.

Terminology varies across clinical-education sources. Some authors use “problem representation” for the clinician’s internal mental model and “summary statement” for its spoken or written expression; others use “problem representation” for the one-liner itself. The label matters less than making the statement concise, diagnostically relevant, revisable, and honest about uncertainty. This template is an educational aid, not a validated clinical instrument or a substitute for local presentation standards.

• Relevant context: age band or population context, setting, and only the baseline or risk factors that change interpretation.
• Time course: acute, subacute, or chronic; stable, intermittent, progressive, recurrent, or resolving where supported.
• Syndrome: the main pattern expressed with precise observations or semantic qualifiers rather than a premature final diagnosis.
• Discriminators: two to four positive or negative findings that genuinely change the working possibilities.
• Current question: the diagnostic, management, escalation, or consultation decision that remains unresolved.
• Update rule: rewrite the statement when the clinical problem, evidence, or purpose of the presentation changes.

Name the leading possibilities at meaningful turning points and explain which finding raised or lowered them. Include relevant negatives when they truly influenced reasoning, not as an exhaustive review of systems. Separate what was observed from what was inferred. A laboratory value, imaging pattern, or treatment response is evidence; the interpretation should remain explicit and open to correction.

Tests are most educational when the reader knows the question each one was meant to answer. Report results with units and reference context where needed, but avoid false precision that increases identification risk or distracts from the pattern. If an early result was misleading, explain why it was reconsidered rather than treating error as a plot twist.

• What was the working explanation before this result?
• Which dangerous alternative had to be addressed first?
• What new fact caused the plan to change?
• Which uncertainty remained after the test or intervention?

Copy the headings below into a draft, then remove any field that is unnecessary for the learning question.

A fictional one-sentence frame might read: “Adult with six weeks of progressive exertional breathlessness, a remote thrombotic event, clear lung fields, and new right-heart pressure signs; the unresolved question is whether pulmonary vascular obstruction explains the trajectory.” It states the timeline and decision problem without revealing a final answer or adding decorative detail.

• Learning question: What decision, pitfall, or reasoning pattern should the reader understand?
• Problem representation: Age band, relevant context, main concern, time course, and current severity.
• Timeline: Onset → meaningful changes → prior tests or treatment → reason for escalation.
• Key findings: Important positives and negatives, with units or reference context when needed.
• Differential and turning points: What was considered, and what evidence changed the ranking?
• Resolution: How the final explanation was supported—or what uncertainty remained.
• Management and outcome: What happened, over what follow-up period, and what cannot be inferred from the response.
• Learning points and sources: Two or three bounded lessons plus authoritative references.
• Privacy and review: De-identification decision, consent/governance status, author, reviewer, and review date.

Put the reason for referral and the question you want the receiving service to answer near the top. Then include the smallest clinically sufficient set of history, findings, tests, treatments, medicines, risks, and access needs that lets the service triage and prepare. AHRQ specifically advises including the reason, pertinent medical history, medicines, and test results and sending the information directly rather than relying on the patient to relay it.

Copy the structure below into the authorized clinical record or the receiving service’s required form. Replace every prompt, remove irrelevant fields, attach the records the destination requests, and review the final document before sending. Do not paste identifiable patient information into SameCase, a public template, or a consumer AI system.

This is a general educational drafting aid, not a universal referral standard. For example, the active NHS England Clinical Referral Information Standard applies to UK primary-to-secondary outpatient referrals and organizes a wider required dataset that includes medicines, previous history, allergies, current symptoms, and the patient’s concerns and expectations. Other systems define different mandatory fields, consent and disclosure rules, language-access duties, payment authorization, urgency pathways, and follow-up responsibilities. The destination’s current instructions and local policy take precedence.

• To: [named service, specialty, or clinician] · From: [referrer, role, organization, secure contact] · Date: [date sent].
• Patient and routing details: [identifiers and administrative fields required by the authorized receiving system—do not copy these into a public tool].
• Reason for referral: [one sentence stating the current problem and why referral is being made now].
• Explicit request: [diagnostic clarification, procedure assessment, management advice, shared care, transfer of care, or another defined task].
• Urgency and rationale: [local category plus the clinical facts supporting it; use the local urgent or emergency route when indicated rather than assuming a routine letter will be read immediately].
• Concise problem representation: [relevant context] with [time course] [syndrome or unresolved problem], characterized by [key discriminating findings].
• Relevant history and current status: [pertinent conditions, operations, exposures, examination findings, function, and important positives or negatives].
• Workup: [completed tests with dates and key results] · [pending tests and who will review them] · [reports or images attached or transferred].
• Treatment so far: [interventions, dates, response, adverse effects, and why treatment stopped or changed where relevant].
• Medicines and safety information: [current relevant medicines with dose/use, allergies with reactions, and other material clinical warnings].
• Working assessment: [supported possibilities and material uncertainty; distinguish findings from interpretation and do not imply a confirmed diagnosis when one is not established].
• Coordination: [what the referring team will continue], [what the receiving team is asked to do], [how and when feedback should return], and [best secure contact for clarification].
• Patient communication and access: [what the patient understands about the referral, communication language or accessibility needs, and copy-sharing or preference details required locally].

Sending a letter is only one step. Confirm that it reached the intended service, record the triage or booking outcome, track whether the consultation occurred, and make sure the specialist’s report returns to the clinician responsible for ongoing care. AHRQ recommends a defined tracking process, especially for high-risk or urgent referrals, plus documentation of completion and results. Assign ownership for tests that are pending when the referral is sent; neither team should have to infer who will act on a result.

Explain the reason and practical next step to the patient in language they understand. UK GMC guidance, for example, tells doctors to explain why the referral is being made, what should happen next, who is responsible for overall care, when the patient can expect to be seen, whom to contact, and what relevant information will be disclosed. Where the health system or local policy supports it, provide or help the patient obtain a copy of clinician correspondence. Do not make the patient the only courier of clinical information, but do give them enough context to recognize the purpose, destination, preparation instructions, expected timing, and contact route if the referral does not progress.

Research on referral-letter improvement is useful but limited. A 2018 review found that templates can improve documentation and awareness of relevant factors, while also reporting heterogeneous interventions and methodological limitations. It does not prove that this particular template improves clinical outcomes. Treat the fields as a completeness prompt that supports—not replaces—clinical judgment, destination-specific criteria, peer feedback, and a reliable referral process.

• Delivery: Was the referral accepted by the intended destination through an approved channel?
• Triage: Was it accepted, redirected, returned for more information, or declined—and who will act next?
• Pending work: Who owns each result or task that may complete before the specialist visit?
• Completion: Did the consultation occur, and did the report return to the responsible clinician?
• Patient communication: Does the patient know the outcome, next action, timing, and contact route for a delay or change?
• Learning review: If a service repeatedly requests missing fields, update the local template or referral agreement rather than adding unfiltered chart text.

Choose the document before choosing the detail. A formal medical case report is a scholarly publication for readers beyond the care team. A referral letter requests a defined contribution from another service and must support safe triage and coordination.

SameCase's templates are educational drafting aids. They do not create a clinical record, transfer responsibility, establish research or ethics status, confer permission to publish, or satisfy a journal merely because every prompt is filled in.

• Formal clinical case report: a public or scholarly account organized for transparent reporting, literature context, publication consent, authorship, disclosures, journal review, and durable publication.
• Referral letter: action-oriented clinician-to-clinician communication stating the reason, urgency evidence, relevant history and results, explicit request, and coordination plan through an authorized care pathway.
• Learn from real cases shared by patients and doctors. Compare the details and discuss useful ideas with the doctor caring for you.
• Learn from real cases shared by patients and doctors. Compare the details and discuss useful ideas with the doctor caring for you.
• Case-study research: a research methodology with defined questions, evidence sources, analysis, and governance; it is more than a renamed clinical case report.

Start with one bounded clinical message and a plausible target journal. Confirm that the episode is suitable to report and that the required publication-consent, privacy, institutional, authorship, and disclosure steps can be completed. Reconstruct what was known at each point from the source record, then draft the chronology, diagnostic assessment, interventions, follow-up, outcomes, limitations, literature context, and patient perspective where appropriate. Finally, check the manuscript against the target journal's current instructions, the official CARE checklist, and any relevant specialty extension before submission.

CARE is a reporting guideline, not a universal manuscript format. Its 13 top-level areas run from title, keywords and abstract through patient information, clinical findings, timeline, diagnostic assessment, interventions, outcomes, discussion, patient perspective and informed consent. The copyable outline below paraphrases those areas and adds workflow controls; it is not the official CARE checklist and should not replace the journal's version.

Copy these prompts into an approved workspace, then compare the completed draft with the official CARE checklist and the target journal's current author instructions. Replace every bracketed prompt, mark unavailable information as unavailable rather than inventing it, and remove notes that are not needed for the report's bounded message. This outline is CARE-shaped, not a reproduction of the official checklist or a claim of formal CARE compliance.

• Venue and scope: [target journal or conference], [article type], [word/figure/reference limits], [required CARE or specialty extension], and [submission deadline or version date checked].
• Title and keywords: name the main phenomenon and identify the work as a case report using the journal's preferred terminology; add specific, searchable keywords without claiming novelty that has not been established.
• Abstract: [why the case matters], [main presentation and findings], [diagnosis or unresolved status], [intervention if relevant], [outcome and follow-up], and [one bounded lesson].
• Introduction: [brief literature context], [what gap, variation, harm signal, diagnostic issue, or educational point the case may illustrate], and [why one case is worth reporting without implying frequency].
• Patient information and clinical findings: [minimum relevant demographic and contextual information], [presenting concerns], [medical, family, psychosocial and treatment history that changes interpretation], and [relevant examination findings], all within the approved privacy and consent boundary.
• Timeline: [dated or relative sequence of presentation, tests, working explanations, interventions, changes, adverse events, follow-up and outcome]; show what was known at each stage and identify estimates or missing records.
• Diagnostic assessment: [methods and results with units/reference context], [differential diagnoses], [diagnostic criteria or confirmation], [challenges and uncertainty], and [facts that weakened competing explanations].
• Interventions: [type, administration or procedure details needed to interpret the case], [changes and reasons], [co-interventions], [monitoring], and [important harms]; do not turn a publication account into instructions for another patient.
• Follow-up and outcomes: [duration], [clinician- and patient-reported outcomes where available], [objective follow-up], [adherence or tolerability where relevant], [adverse or unanticipated events], and [later uncertainty or loss to follow-up].
• Discussion: [comparison with the most relevant literature], [strengths and limitations of the report and management], [alternative explanations], [why temporal sequence is or is not persuasive], and [two or three proportionate takeaways].
• Patient perspective: [the patient's account or reason it is not included], using only the content and wording permitted by the patient, journal, and local process.
• Consent, privacy, and governance record: [publication-consent form and version], [who obtained and stores it], [what manuscript/images the patient or authorized proxy reviewed], [institutional or ethics determination], [removing identifying details review], and [journal-specific statement]. Keep this record in the approved system, not in SameCase or a public AI tool.
• Publication record: [authors and accountable contributions], [funding and relationships], [acknowledgments], [AI assistance if the journal requires disclosure], [references checked], [figures/files inspected for identifiers], and [official checklist or extension submitted if required].

Select the target venue before finalizing the draft. Journals can define different article structures, word limits, consent forms, ethics statements, image rules, authorship requirements, disclosures, data expectations, and checklist-upload procedures. EQUATOR's CARE record, updated in February 2026, also lists specialty adaptations and extensions; use one only when it fits the case and the target journal accepts or requires it. A generic CARE-shaped outline does not override a radiology, anaesthesia, physiotherapy, acupuncture, surgical, or other field-specific standard.

Publication consent is distinct from consent for clinical care. ICMJE states that identifying information should not be published unless it is essential and written informed consent for publication has been obtained; an identifiable patient should be shown the manuscript, potential online availability should be disclosed, and nonessential identifying details should be omitted. ICMJE also notes that laws vary and journals set their own policies with legal guidance. Do not assume that changing dates, masking eyes, or removing a name makes an unusual case anonymous.

COPE's journal guidance similarly warns that anonymity cannot be guaranteed and asks consent processes to explain publication uses and the limits of withdrawing consent after publication. Its preferred consent-storage approach is not identical to every journal's approach, and it is not a substitute for current law or local policy. Follow the target journal's current form and the responsible institution's process; ask the relevant ethics, privacy, or governance office when the case's status or permission route is uncertain.

A reporting checklist asks whether important information is present or transparently unavailable. It does not independently verify the source record, confirm that the final diagnosis is correct, prove that an intervention caused an outcome, remove confounding or hindsight bias, make a selected case representative, establish incidence or comparative effectiveness, or turn a case report into a controlled study. A complete report can still contain a mistaken inference; an incomplete report can contain a real observation that remains difficult to appraise.

Use CARE as a completeness prompt rather than a quality score or approval badge. Keep observation separate from interpretation, identify missing records and alternative explanations, match causal language to the evidence, and support claims beyond the episode with appropriate external sources. Peer review, journal acceptance, indexing, citation, clinical usefulness, and patient benefit are separate outcomes; completing this outline guarantees none of them.

Explain how the final explanation was supported: formal criteria, confirmatory testing, operative or pathology findings, longitudinal response, or specialist consensus. If the case remained uncertain, say so. “Unsolved” can be educational when the boundaries of current knowledge and the safety plan are clear; do not retrofit certainty into an ambiguous course.

Treatment details should be sufficient to understand the reasoning and outcome without reading like a prescription for strangers. Note important sequencing, monitoring, adverse effects, or why an intervention was withheld. Describe follow-up over a meaningful interval and distinguish improvement that followed treatment from proof that treatment caused it.

End with two or three sentences: what clue was most transferable, what common error the case helps prevent, and what limitation readers should remember. A strong learning point sounds like “consider this pattern when…” or “this result did not exclude…” rather than “this presentation means…”. Invite correction when the evidence is contestable.

Before submission, perform a separate removing identifying details review of prose, title, files, and images. Confirm that no patient, relative, facility, or clinician can be recognized from direct or combined details. SameCase’s scanner is a backstop, not permission to skip that review.

The 2013 CARE statement reports a literature review and interviews, a face-to-face consensus meeting, post-meeting feedback, and pilot testing with 27 participants. Its full text says the University of Michigan Department of Orthopaedic Surgery, the university's Office of the Vice-President for Research, and Global Advances in Health and Medicine funded travel expenses, with no honoraria. The 2017 explanation-and-elaboration paper comes from overlapping CARE authors and PubMed indexes a Cancer Research UK grant, but the abstract does not state the grant's role. The CARE papers and EQUATOR's CARE library record are therefore linked evidence, not three independent validations of this SameCase outline.

ICMJE is a small working group of general medical journals and related organizations; COPE is a publication-ethics membership body. Their pages are primary evidence of editorial recommendations and publisher practice, not independent empirical proof, medical law, or a universal journal contract. Their institutional incentives favor consistent and defensible publication processes; that does not show that the recommendations are wrong, but it limits what can be inferred from agreement between them.

No cited source evaluates this exact combined case-report, referral, and summary page. SameCase has not validated the outline for reporting quality, diagnostic accuracy, causal inference, acceptance, citation, safety, or patient outcomes, and no named clinician, journal editor, publication-ethics specialist, or methods reviewer is assigned as of 30 August 2026. Use the visible sources to inspect the original guidance, then follow the target journal, specialty, institution, and jurisdiction.