History and Physical Extraction, Section by Section

History and physical extraction turns each section of an H&P into fields without changing what the note says. Lists such as past history become one entry per item, narrative such as the HPI stays as text, denied symptoms are stored as negatives, templated exam text is marked as template, and the plan is copied as written, never interpreted.

The sections an H&P contains, and the timing rules that make it an audit target, are set out in the guide to history and physical document structure. This post starts where that one stops: what to do with each section once you have found it.

Chief complaint and HPI in history and physical extraction

The chief complaint and the history of present illness (HPI) open an H&P and describe why the patient is being seen. The chief complaint is short and becomes one text field. The HPI is a narrative paragraph and should stay as narrative text, with only a few clearly stated facts lifted into separate fields beside it.

The chief complaint is usually a phrase, sometimes in the patient's own words: "Chief complaint: 'my knee gave out'". Keep the quotation marks if the note has them. They tell a reader that the words are the patient's, not the clinician's.

The HPI is different. It is a story told in sentences, with times, causes and context mixed together. Splitting it into fields risks pulling a phrase away from the words that qualify it. "Pain started three days ago after a fall, worse with stairs, no prior injury to this knee" holds an onset, a cause, an aggravating factor and a negative. Taken apart, the negative can drift free and the onset can lose its anchor.

A safer pattern keeps the full HPI as one text field and adds only facts that the note states plainly:

• Onset as written, such as "three days ago", kept as text rather than turned into a date unless the note gives one.
• Source of history, when stated, such as "per patient" or "per daughter".
• The HPI author, if the section was written by someone other than the note's signer.

The HL7 Consolidated CDA History and Physical template (version 5.0.0, active as of 2026-06-13) requires a chief complaint and reason for visit, and lists the History of Present Illness section as recommended, per the HL7 C-CDA History and Physical template. So a scanned H&P without an HPI heading is not unusual. The story may sit under "Subjective" or run straight on from the chief complaint.

Past medical, surgical, family and social history fields in an H&P

The history sections of an H&P cover past medical history, past surgical history, family history and social history. Each one becomes its own list, with one structured entry per item. An entry keeps the item as written, any date or age given, and for family history the relative it applies to.

These sections are usually written as short lists, which makes them the easiest part of an H&P to extract. The work is in keeping the lists apart and keeping each item intact.

Past medical history gives one entry per condition, with the wording as written and any date or year stated: "Hypertension, dx 2019". Do not clean "HTN" into "Hypertension" at capture time. Keep the original and expand it in a separate, reviewed step.

Past surgical history gives one entry per procedure, with the procedure as written, the year or age if stated and the side if stated: "R knee arthroscopy 2015". Laterality is easy to drop and hard to recover, so keep it as its own field.

Family history needs a relative field on every entry. "Mother: breast cancer at 52; father: MI at 60" is two entries, each with a relative, a condition as written and an age as written. "Noncontributory" is a statement and should be stored as written, not as an empty list.

Social history mixes several topics: tobacco, alcohol, other substances, occupation and living situation. Store each topic separately with the note's exact wording. "Former smoker, quit 2010, 10 pack-years" stays together as one tobacco entry.

Medications and allergies usually sit between these sections in an H&P. Their entry rules belong to a separate post, which sets out the fields every medication list entry needs.

How should review of systems negatives be extracted from an H&P?

Review of systems negatives should be extracted as findings with a negative status attached, never as bare symptom names. "Denies chest pain" becomes an entry for chest pain marked "denied", with the original phrase kept beside it. Dropping the word "denies" turns a negative into a positive, the most damaging error an H&P extraction can make.

The review of systems (ROS) is a system-by-system list of symptoms the clinician asked about. The HL7 C-CDA Review of Systems Section describes it as holding current symptoms "as well as a potentially large number of pertinent negatives, for example, symptoms that the patient denied experiencing."

Each ROS entry needs four parts:

1. The body system, such as "Cardiovascular" or "Respiratory".
2. The symptom as written, such as "chest pain".
3. The status: reported, denied, or unclear.
4. The source phrase, such as "Denies chest pain, palpitations".

One sentence can hold several negatives. "Denies fever, chills, night sweats" is three denied entries, and the word "denies" covers all three. Then watch for the turn: "Denies fever or chills but reports fatigue" switches status at "but".

Summary lines need care too. "ROS: All other systems reviewed and negative" is not a list of negatives for every system. Store it as one statement, as written, rather than generating "denied" entries for symptoms that never appear on the page.

How software detects negation and context across all kinds of clinical text is explained in the guide to negation and context in unstructured clinical text.

Templated normal physical exam text in an H&P

Templated normal exam text is default wording that an electronic record inserts into the physical exam section, such as "Heart: regular rate and rhythm, no murmurs". It needs its own handling because the same words appear in many notes, and extraction cannot tell from the text alone whether a finding was examined or simply left at its default.

The HL7 C-CDA Physical Exam Section defines the content as "direct observations made by a clinician", and says it "does not include laboratory or imaging findings." That definition describes what the section should hold. A template only shows what text was inserted.

You cannot judge from a page whether a clinician examined a system. What you can do is record what the page shows and mark the patterns that suggest default text:

• Identical wording across notes for the same patient, or across patients from the same author.
• Every system "within normal limits", including systems unrelated to the visit.
• Exam text that contradicts the HPI, such as "no tenderness" for the knee the patient came in about.

Mark such entries with a "template suspected" flag rather than removing them. Removing them would change the note. Leaving them unmarked lets default text pass for an observed finding. A contradiction between the exam and the HPI should go to a person, with both passages shown.

Copied-forward text raises similar problems across a series of notes. The guide on whether copy-forward text in clinical notes misleads extraction covers how to spot it.

Is the H&P assessment and plan worth extracting as fields?

The H&P assessment and plan is worth extracting as a list of items as written, each with the problem or action named in the note. It is not worth extracting as interpreted data. Extraction can capture "Admit to medicine" or "MRI right knee" as plan lines, but it should not decide what a plan line means.

The assessment names the clinician's conclusions. The plan names what will happen next. Many H&Ps combine them as numbered problems, each followed by its plan lines:

1. Right knee pain after fall: MRI right knee, crutches, orthopedics consult.
2. Hypertension: continue home medications.

That layout extracts well. Each numbered problem becomes an entry with its text as written, and each plan line becomes a child entry linked to it. Keep the number, because a later note may refer to "problem 2".

Some things should stay out of the fields. Do not infer a diagnosis from a plan line, such as reading "rule out fracture" as a fracture. "Rule out", "possible" and "consider" are words of uncertainty, and the field should keep them. Do not turn "continue home medications" into a medication list. It points to a list held elsewhere.

Imaging ordered in the plan produces its own report later. Pulling exam names, dates and impressions from that report, without interpreting them, is covered in radiology report extraction.

H&P update notes that point to an older exam

An H&P update note is a short entry that refers back to an H&P written earlier, records any change in the patient's condition, and is dated at admission or before a procedure. Extraction should link the update to the original H&P and keep both dates, so neither one is mistaken for the other.

The hospital medical records rule, 42 CFR 482.24 on the eCFR reader (last amended May 1, 2020), calls for "an updated examination of the patient, including any changes in the patient's condition" when the history and physical was completed within 30 days before admission or registration. The hospital surgical services rule, 42 CFR 482.51 on the eCFR reader (last amended September 30, 2019), carries the same requirement before surgery.

An update note can be a single line: "H&P of 2026-02-17 reviewed, patient examined, no changes." For extraction, capture:

• The update's own date, time and author.
• The referenced H&P date, as written in the update.
• The change statement, as written: "no changes" or the changes listed.
• A link to the original H&P record, when you can find it in the same chart.

When the referenced H&P is not in the chart you hold, keep the referenced date anyway and flag the missing document. Do not copy the original's findings into the update record. The update says what it says, and the original stays a separate document with its own date.

Emergency visits that lead to an admission often produce an H&P and an ED record for the same day. How to keep those apart is covered in emergency department record extraction.

A before-and-after of a fictional H&P

This before-and-after shows sections of one fictional H&P beside the fields extracted from them. The patient, author, dates and wording are all examples.

Three rows go to a person: the knee exam that contradicts the HPI, the summary ROS line, and any field with a low confidence score.

The H&P is not named among the document types in Schema Care's supported documents section, so this table describes a method rather than a product feature. Whatever does the reading, the same rule holds: a reviewer should see each flagged value on the original page before it is used.