How to Take a Focused History in Four Steps

Your senior says, "just take a focused history," and you either ask everything or freeze. A focused history is built around the presenting complaint, where you choose which questions to ask and, just as much, which to leave out.

Here's the method in four steps, then two fictional AI-patient runs: Hamish Thorne, 29, chest tightness and can't breathe; and a 6-week-old boy vomiting after every feed, with his father giving the history. We'll stay complaint-driven, ask what splits the differentials, keep the pertinent negatives and safety questions, and drop the rest on purpose.

How to Take a Focused History (and What to Leave Out)

What is a focused history?

A focused history is a complaint-driven interview that targets the chief complaint, a concise history of present illness, and review-of-systems points tied to that problem. If you're asking how to take a focused history, the answer is restraint, every question has a diagnostic reason, you drop unrelated sections on purpose, and length tracks complexity, not the clock.

Compared with a full history for a new, nonemergency patient, a focused history centers on the problem at hand. AMBOSS splits it in AMBOSS's medical history entry: problem-focused, "only includes CC and brief HPI; usually taken in emergency setting." and expanded problem-focused, "includes CC, brief HPI, and pertinent ROS; usually when patient is already under the ongoing care of a provider".

For the full admission view, see a complete admission clerking.

When do you take a focused history instead of a full one?

Take a focused history when time or context demands it, like the emergency department, an urgent review, a follow up for someone already under care, or a timed station. Do a full history for new, nonemergency patients, for example an admission, or when you are asked for the long version.

That matches AMBOSS: problem focused histories usually in the emergency setting, expanded problem focused when the patient is already under ongoing care, and a full history for new, nonemergency patients. In a timed station or a busy shift, the skill is question triage, and even a focused history keeps safety questions, covered in step 4.

Flow diagram of four focused history steps, from problem naming to differentials, split questions, and safety checks.

Step 1: read the complaint, then name the problem in one line

Before you walk in, read the triage note so you know the chief complaint, then take a minute to think about what could be wrong. That is straight from Columbia's pediatrics clerkship objectives, which says this gives a better chance to ask about pertinent positives and negatives.

Open with an open question and let them talk. Then say back the problem in one line that you will use as your anchor for every question after.

Case A starts with, "It's happened again, my chest goes tight and I can't get my breath, and I honestly think I'm dying." He volunteers that attacks come on suddenly, out of the blue, often with no trigger. Fourth visit to the emergency department in three months, heart tracing, heart blood test and chest X-ray normal every time.

Your one line: recurrent sudden attacks of chest tightness and breathlessness, fourth emergency visit in three months with normal heart tests each time.

Step 2: characterize the complaint with SOCRATES or OLDCARTS

If you're working out how to take a focused history, start by fully characterizing the main symptom, every time. The mnemonic is a checklist for the complaint itself, not for the whole patient, and it's the bit you never skip.

Use whatever you learned. SOCRATES is the common UK pain mnemonic. In the US, OLDCARTS is everywhere, Onset, Location, Duration, Character, Alleviating/aggravating, Radiation, Timing, Severity, see the OLDCARTS breakdown. AMBOSS teaches its own, LIQOR AAA for pain, COLD REARS SIT for the HPI.

In a focused history, this is where strong discriminators hide, especially timing and triggers. In Case A, each attack peaks within about ten minutes and is completely gone within half an hour. The chest tightness isn't brought on by exercise, it happens sitting still, at his desk, or in bed.

Step 3: pick your differentials, then the questions that split them

Pick 3 to 5 differentials from the one-line problem, then for each pick the single question that would move it up or down. Anything that does not change your list can wait.

Building that ranked list lives here, with cases, generating a ranked differential.

Case A, palpitations in discrete attacks, carried this set, acute coronary syndrome, supraventricular tachycardia, pulmonary embolism, hyperthyroidism, pheochromocytoma, hypoglycemia, alcohol withdrawal, generalized anxiety disorder.

Does exertion bring it on, does rest settle it?
Any clot risk factors?
Heat intolerance, weight loss, hand tremor?
During attacks, what is the blood pressure, any headaches?
Caffeine, alcohol, or drugs near onsets?

He was well between attacks, could run for a train, carry shopping upstairs, and sleep flat. Normal exercise tolerance between episodes "effectively excludes angina, heart failure and significant pulmonary disease."

No calf swelling, flights, surgery, or hemoptysis. No family clots. No weight loss, heat intolerance, shaky hands between attacks, diarrhea, or neck swelling. Blood pressure was normal at every visit, including during an attack, with no headaches, and the case file calls a normal pressure during an episode powerful evidence against pheochromocytoma. On a specific question he reported about six strong coffees a day plus energy drinks.

Diagram of five differentials for a 29 year old with chest tightness, each with one key question and the patient answer.

Step 4: the pertinent negatives and safety questions you never drop

Focus trims plenty, but not this set: the pertinent negatives that rule out the dangerous picks on your list, the red flag safety questions, and the patient's ideas, concerns, expectations.

Case A, an AI patient, reports mood dropped over the last month, sleeping badly, lost interest. The file is clear that suicidality must be screened for explicitly. He had no thoughts of harming himself and no thoughts of suicide, but only because someone asked, directly.

Pertinent negatives are chosen, not guessed. They track your differential and document the safety work you did. If you need a quick framework for that pruning, see choosing negatives on purpose.

Then ask what they think and fear. In Case A, he's convinced there is a heart problem the tests keep missing, he's worried he'll drop dead during an attack, and he expects a proper explanation and, in his words, "a scan of my heart." Try asking what they're worried about.

That last answer tells you what your explanation must address, and what you record as pertinent negatives and safety checks.

What you can leave out, and how to say so

In a focused history, cut what won't split your differential. That means symptoms from unrelated systems, long social or family detours, and past history that doesn't touch today's complaint.

Say it up front: "I took a focused history for the chest tightness; I haven't covered the full systems review yet." That marks a decision, not a gap. You can circle back if asked. See tone in presenting a patient on rounds.

Full review of systems without red flags tied to your differential.
Detailed social history beyond smoking, alcohol, or exposures that change risk for your list.
Past medical or surgical history with no bearing on the complaint.

Worked example on two cases

Case A: Hamish Thorne, 29, fictional AI patient.

Sudden attacks peaking at ten minutes and resolving within thirty, well between with full exertion. Differential: acute coronary syndrome, supraventricular tachycardia, pulmonary embolism, hyperthyroidism, pheochromocytoma.

Timing, peak in ten and gone in thirty. Full exertion between, yes. Pins and needles around the mouth and in fingers with hand curling, yes on a specific probe. Avoidance, stopped the gym, the Tube, and being alone.

Negatives kept: normal blood pressure during an attack, no headaches, no clot risks, no thyroid symptoms, mood screened, no suicidal thoughts. Answer: panic disorder, a positive diagnosis on pattern recognition, not one of exclusion.

Case B: a 6-week-old boy, history from his father, fictional AI patient.

Opening line from his father: "He's been sick after every feed for about two weeks and it's getting worse, it goes right across the room now." Differential: reflux, viral gastroenteritis, urinary tract infection, sepsis, intussusception.

Ever green, milk only and never green, on a specific question. Hungry straight after, ravenously. How many wet diapers, about three a day not six or seven. Family stomach surgery as a baby, his father, not volunteered until asked.

Negatives kept: no fever, cough, rash, diarrhea or unwell contacts, reflux treatment changed nothing. Answer: pyloric stenosis.

In both fictional Diagnosica AI-patient cases, written and signed off by a doctor, the story pointed somewhere comfortable, his heart then reflux, and the key facts surfaced only on a specific probe. Avoid them with spotting red herrings.

Try a full case Run a full Diagnosica case where facts surface only when you ask the patient. Start a case

Drill it in two minutes a day

Drill how to take a focused history with tiny reps. The quick case takes two minutes. The patient says one line, five clues follow one by one, you guess after each, then the answer and a one-line teaching point. It keeps a daily streak.

Talk by voice or type to an AI patient on screen as a video avatar, order investigations, commit to a diagnosis and management plan, and get a scorecard with teaching points after every case. Free tier included, no card needed.

More than 130 cases, each written and signed off by a doctor. Pediatric cases are told by a parent. No physical examination practice. "It doesn't give advice about real patients, and it's not a diagnostic system. Treat every output as educational."

See diagnosis games worth playing.

Before each case, write your one-line problem and top three, then count how many questions you needed.

Educational use only. Not medical advice. AI-generated; verify clinically against primary sources.

Clinical review pending.