Case Reports: When and How to Write One

If I deleted the word "rare" from your case report draft, what would be left?

For most of the case reports I've handled, the honest answer is very little. You'd have a patient, a diagnosis, a treatment that worked about as well as the textbook says it should, and a literature review restating what's already known. Rarity feels like a reason to publish. To an editor it usually isn't. A rare thing that behaves exactly as expected teaches the reader nothing they can use on Monday morning.

I'm not trying to talk you out of writing case reports. I've published a few myself, and I've accepted many more as an editor. A good case report can be the first signal of a drug reaction, a new disease or a diagnostic trap. Some of the most cited clinical observations of the last century started as single cases. But the ones that get accepted and get read share a few features. The rejected ones also share a few, and you can see them coming from the first paragraph.

What makes a case worth publishing

When a case report lands on my desk, I ask one question before I read past the abstract: What will a clinician or researcher do differently after reading this? If I can't answer that, the manuscript goes back.

Cases that pass that test usually fall into one of these groups:

• An unexpected association. Two conditions, or a drug and an outcome, turning up together in a way that hasn't been described and could plausibly be causal.

• An unreported or rarely reported adverse effect, especially of a newly approved drug, device or procedure. These are some of the most useful case reports we publish, because post-marketing signals often start this way.

• A diagnostic pitfall. A common disease presenting in a way that led you, or someone before you, down the wrong path. These are underrated. They're also the cases readers remember.

• A new mechanism or management approach, described cautiously, with enough detail that someone else could try it or test it.

• A genuinely new condition or variant. This is the rarest category, and it's the one most authors think they're in.

Notice what's not on the list: "the first case at our institution," "a rare case of X successfully treated with standard therapy," and "a reminder to consider X." A reminder is not a finding.

The ten-minute novelty check

Before you write a word, search properly. Not two keywords in a general search engine. Search PubMed with the condition, the presenting feature and the intervention, and include synonyms and older terminology. Do the same in a second database relevant to your specialty. A quick PubMed search on case reports will show you how crowded some of these spaces are.

If you find twelve previous reports of the same thing, you don't necessarily have to stop. Your case might add something the twelve didn't: longer follow-up, a biopsy result, a response to a rechallenge. But your manuscript now needs to say exactly what that addition is, in one sentence, near the top.

A story about reframing

A few years ago, Priya, an intensive care physician who does research alongside a heavy clinical load, sent me a draft about a young patient with a rare autoimmune encephalitis. It was carefully written, and it was going nowhere. The literature already had dozens of similar cases.

When we talked it through, the interesting part turned out to be buried in her timeline. The patient had been admitted twice and treated for a primary psychiatric illness for nearly three weeks before anyone ordered the right test. Priya had written that as background. It was actually the paper.

She rewrote it around the diagnostic delay: what the early signs were, why they were missed, and which specific findings in retrospect should have triggered the workup. Same patient, same data. It was accepted at a solid general medicine journal, and she still gets emails from residents about it. The case wasn't rare in any useful sense. The lesson was.

Consent: do it early, do it in writing, show the patient the manuscript

This is where I see the most avoidable problems, and where the stakes are highest. A case report is about one identifiable human being. Even with the name removed, a combination of age, sex, rare diagnosis, hospital and dates can make a patient recognizable to their neighbors, their employer or themselves.

The ICMJE recommendations are the baseline most medical journals follow. They say that identifying information shouldn't be published unless it's essential for scientific purposes and the patient (or parent or guardian) has given written informed consent for publication. They also say the patient should be shown the manuscript to be published. Most journals I've worked with now expect written consent for every case report, identifiable or not, and many require their own consent form.

A few practical points that trip people up:

• Consent to treatment is not consent to publication. The general research or data-use clause in your hospital's admission paperwork almost never covers it.

• Get consent before you write, not after acceptance. If the patient says no, you want to know before you've spent forty hours on a draft.

• Use the target journal's form if it has one. If you don't know the journal yet, use a well-regarded standard form, and be ready to go back to the patient if the journal insists on its own.

• Deceased patients still need consent, usually from next of kin or a legal representative. Check your local law as well as the journal's policy.

• Patients who can't be reached are a real problem. Some journals allow publication without consent in narrow circumstances, often with ethics committee approval and heavier anonymization. Many won't. Don't assume.

• Anonymization isn't a workaround. Black bars across the eyes in a clinical photograph haven't counted as anonymization for a long time.

On ethics approval: practices vary a lot between countries and institutions. Many institutions don't classify a single retrospective case report as human subjects research, but some do, and some require a formal waiver. Email your ethics office and keep the reply. Journals increasingly ask for it.

The visiting researcher and the missing signature

Tomás was a visiting researcher from abroad who spent a year in our department. He brought a beautiful case from his home hospital: a rare drug interaction with a clear temporal relationship and positive rechallenge, exactly the kind of thing I'd want to publish. He'd discussed it with the patient during treatment, the patient had agreed enthusiastically, and the conversation was noted in the chart.

The journal wanted a signed form. By then the patient had moved and wasn't responding to letters. It took Tomás's former colleagues nearly five months to track the patient down and get a signature. The paper survived. Plenty don't.

The lesson I took from it, and now pass on to everyone: the moment you think "this might be publishable," get the signature. A two-minute conversation at the bedside saves months later.

What to tell the patient

Patients usually say yes, especially when they understand why. Something like this works well:

"Your case taught us something that could help doctors recognize this problem sooner in other patients. We'd like to write it up for a medical journal. We won't use your name, but someone who knows you well might recognize the details. Once it's published, it may be freely available online and we can't take it back. You can read the draft before we submit it, and saying no won't affect your care in any way."

Notice the bit about being recognizable and about permanence. Leaving those out isn't kindness. It's an incomplete consent.

Following the CARE guidelines without writing like a checklist

The CARE guidelines (short for CAse REport) were developed through an international consensus process and published in 2013. They're listed on the EQUATOR Network alongside CONSORT, PRISMA and the other major reporting guidelines, and many journals now ask you to upload a completed CARE checklist with your submission. If you want the background papers, search PubMed for the CARE guideline publications.

The checklist covers the title, keywords, abstract, introduction, patient information, clinical findings, a timeline, diagnostic assessment, therapeutic interventions, follow-up and outcomes, discussion, the patient's perspective and informed consent. You don't need to memorize it. You need to open it while you write, not after.

Most advice treats CARE as a compliance exercise. I think that's backwards. The checklist is really a list of the questions a skeptical reader will ask, and filling it in honestly tells you where your case is thin.

Title: say what the case teaches

Before: A rare case of hepatotoxicity: case report and review of the literature

After: Delayed severe hepatotoxicity after starting [a named drug class] in an older adult: a case report

CARE asks for "case report" in the title and for the area of focus. The second version tells a busy reader whether to click. Drop "and review of the literature" unless you've actually done a systematic review. Usually you haven't, and editors notice.

The timeline is the part most authors skip

CARE specifically asks for a timeline, and it's the item I most often see missing or crammed into prose. A simple table does the job and makes reviewers noticeably happier:

Use relative time (day 0, week 8) rather than calendar dates. It's clearer, and it removes one more identifier.

Discussion: claim less than you want to

This is where good cases get rejected. A single patient can't establish causation, efficacy or incidence, and reviewers will hold you to that.

Before: This case demonstrates that early plasma exchange is an effective treatment for this condition and should be considered first-line.

After: Our patient improved after early plasma exchange, although spontaneous recovery and concurrent steroid treatment may have contributed. This observation supports evaluating plasma exchange in larger case series or a registry.

The second version is more likely to be accepted, and more likely to be cited honestly. If you're writing about an adverse event, say how you assessed causality, for example with a standard causality scale, and mention the alternative explanations you ruled out.

Don't skip the patient perspective

CARE includes a section for the patient's own view of their care, when they're willing to share it. Authors leave it out more than any other item. A short paragraph in the patient's words about what the diagnostic delay felt like, or why they chose one treatment over another, often turns out to be the most memorable part of the paper. Ask during the consent conversation. Some patients will happily write three sentences.

A pre-submission checklist I actually use

1. Can I state in one sentence what this case adds, and is that sentence in the abstract?

2. Did my literature search use at least two databases and include older terminology?

3. Is the signed consent for publication on file, on a form the journal accepts? Has the patient seen the manuscript?

4. Have I removed or generalized dates, locations, occupations and other indirect identifiers that aren't scientifically necessary?

5. Do I have a timeline table, with relative time?

6. Have I completed the CARE checklist with page numbers, and are any "not applicable" items genuinely not applicable?

7. Does the discussion acknowledge alternative explanations and avoid efficacy or causation claims?

8. Does the author list reflect who actually did the work? Use ICMJE authorship criteria and a CRediT contribution statement. The resident who managed the patient overnight isn't automatically an author, and the department chief who never read the draft isn't either.

Choosing where to send it

Several large general journals have cut back sharply on case reports, and dedicated case report journals have filled the gap. Some are excellent. Some exist mainly to collect publication fees from clinicians who need a line on their CV, and case reports are a particular target for predatory publishers because the authors are often first-timers.

Before you submit, run the journal through Think. Check. Submit., check whether an open-access title is indexed in the Directory of Open Access Journals, and see whether it's a member of COPE. If a journal promises peer review in a few days and acceptance in a week, believe them, and then go somewhere else. A case report published in a predatory journal is effectively lost: hard to find, sometimes not indexed, and occasionally a mark against you on a promotion committee.

Look at what the journal has actually published recently, too. If every case in the last three issues is a diagnostic pitfall, your new-mechanism paper may be a poor fit, however good it is.

Where the time really goes

Writing the case usually goes fast. The slow part is everything around it: the literature search, checking whether the twelve prior reports really match your case, keeping the CARE items straight across drafts, and getting co-authors to comment on the same version. That's where tools can genuinely help. I've watched residents use PACR, an AI-native research workspace, to run the novelty search, pull the prior reports into one place with short summaries, and then draft and share the manuscript with co-authors without copying notes between five different apps. If you want to try that workflow on your own case, you can start by exploring the existing literature on your condition and see how crowded the field already is.

The tool won't decide whether your case is worth publishing, though. That's still your judgment, and it should be.

So go back to your draft. Cross out "rare." Write one sentence about what a reader will do differently because of your patient. If that sentence comes easily, get the consent form signed this week and open the CARE checklist. If it doesn't, you may have a good teaching case for your next grand rounds instead of a paper, and that's a perfectly respectable outcome too.

When you're ready to pull it all together, set up a free PACR workspace and keep the search, the drafts and the checklist in one place.