Operative report extraction turns a surgeon's note into named fields: the team and their roles, the diagnosis before and after surgery, implants with lot numbers, specimens, blood loss, drains and procedure times. Capture each value exactly as written, keep its unit and its source section, and send lot numbers, conflicting times and late addenda to a person.
This post assumes you already know what an operative report is and which sections it usually has. That overview lives in the guide to operative report data extraction by section. Here the focus is narrower: the fields that look simple on the page and go wrong once they become data.
Operative report header fields: surgeon, assistants and anesthesia
The operative report header names the people in the room and the role each one played: the primary surgeon, any assistants, the anesthesia provider and the anesthesia type. Capture each person as a separate entry with a role label, because a name with no role cannot tell an auditor who performed the procedure and who helped.
A header line often packs several people into one string, such as "Surgeon: Dr. A. Lindqvist; Assistant: Dr. P. Oyelaran, R. Basu PA-C". Split that string into one row per person. Keep the role exactly as printed ("Assistant", "First assistant", "Co-surgeon") instead of mapping it to your own list on the way in. Mapping can happen later, once a person has seen the original wording.
Credentials belong in their own field. "PA-C", "CRNA" and "MD" tell a reader what kind of clinician held the role, and they are easy to lose when a name is copied into a single text box.
Anesthesia needs two fields, not one. The first is the anesthesia provider, a person. The second is the anesthesia type as written, such as "General endotracheal" or "MAC with local". Anesthesia is one of the standard operative note sections listed in the overview of operative report sections.
The signature belongs with the header fields too. The hospital surgical services rule, 42 CFR 482.51 on the eCFR reader (last amended September 30, 2019), says the operative report must be "signed by the surgeon". Record the signer's name and the signature date and time, so the signer can be compared with the surgeon named in the header.
Preoperative and postoperative diagnosis fields in an operative report
An operative report carries two diagnosis fields because the surgeon's understanding can change once the patient is open. The preoperative diagnosis records what was expected before the first incision. The postoperative diagnosis records what the surgeon concluded afterwards. Extraction keeps both, word for word, even when they are identical.
Both fields usually sit under their own headings, such as "Preoperative Diagnosis" and "Postoperative Diagnosis", or the shorter "Pre-op Dx" and "Post-op Dx".
The two values often differ in small ways. A preoperative diagnosis of "Right inguinal hernia" may become "Right indirect inguinal hernia" afterwards. Store both strings exactly as written. Do not merge them into one field, and do not keep only the second because it looks more complete. The difference between the two is information, and only a clinician or coder should decide what it means.
"Same" is a common postoperative entry. When the report says "Postoperative diagnosis: Same", record the word "Same" in the field and add a flag that it refers to the preoperative value. Copying the preoperative text into the postoperative field looks tidy, but it changes what the surgeon wrote.
Assigning codes is a separate job. Where printed ICD-10 or CPT codes show up beside clinical text, and how they relate to the words, is covered in the guide to where ICD-10 and CPT codes appear in clinical documents.
How should implants and lot numbers be captured from an operative report?
Implants in an operative report should be captured one device per row, with the device name as written, the manufacturer, the catalog or reference number, the lot number, the size and the body site. Lot numbers are the field most likely to be misread, so every one should be checked against its source by a person.
The HL7 C-CDA Procedure Implants Section describes the content as "any materials placed during the procedure including stents, tubes, and drains." On paper, implants can show up in several places:
- The implants line in the operative report itself, often brief: "Implants: 15 x 10 cm polypropylene mesh".
- The procedure narrative, where the surgeon describes placing the device.
- A separate implant log, where staff record each device used.
- Manufacturer stickers peeled from the packaging and stuck on a log page, carrying the catalog and lot numbers in small print.
The sticker is usually the best source for the lot number, because it was printed by the manufacturer rather than typed. It is also the hardest to read: small fonts, curved labels, glare from scanning and barcodes beside the text. Capture the value as printed, keep the page reference, and send it to review.
Watch for characters that look alike. Zero and the letter O, one and the letter I, and 5 and S are frequent swaps in lot and catalog numbers. A lot number with a swapped character still looks valid, so no format check will catch it. Only a comparison against the sticker will.
When the operative report and the implant log disagree, keep both values with their sources. Choosing one quietly hides a discrepancy that someone may need to see.
Specimens, estimated blood loss and drains as operative report fields
Specimens, estimated blood loss and drains are the count-and-amount fields of an operative report. Each needs a number and a unit kept together, plus the original wording. A specimen list becomes one row per specimen, blood loss becomes a value with its unit, and each drain becomes its own entry with type and site.
The HL7 C-CDA Procedure Specimens Taken Section covers "the tissues, objects, or samples taken from the patient during the procedure including biopsies, aspiration fluid, or other samples sent for pathological analysis." For specimens, capture each one separately: the label as written ("A. Hernia sac"), the destination if stated ("to pathology") and the count if given. Keep specimen letters or numbers, because the pathology report will use the same labels. The pathology result itself is a separate document, covered in the post on pathology report extraction.
For estimated blood loss, keep the number, the unit and any qualifier. "EBL: minimal" is not "EBL: 0 mL". "<10 mL" is not "10 mL". Record "minimal" as text with no number, and keep the less-than sign as part of the value.
For drains, the HL7 Surgical Drains Section notes that drain placement may instead appear as text in the procedure description. So search the narrative when there is no drains line. Capture the drain type, size and site as written.
"None" deserves a field value. "Specimens: None" and "Drains: None" are statements, not blanks. Record them as written, so an empty field can still mean "not found on the page".
Which operative report times should you extract?
Operative report times record when each stage of the procedure happened. The fields usually worth extracting are the time the patient entered the room, the incision time, the closing time and the time the patient left the room. Capture each with its date and its source document, because the same event is often timed in more than one record.
The hospital medical records rule, 42 CFR 482.24 on the eCFR reader (last amended May 1, 2020), says all patient medical record entries must be "dated, timed, and authenticated". That applies to the entry itself. The times of events inside the procedure come from what each record prints.
The four times measure different things:
| Time field | What it marks | Where it often sits |
|---|---|---|
| In-room time | Patient enters the operating room | Nursing intraoperative record, anesthesia record |
| Incision time | Surgery starts | Operative report, anesthesia record |
| Closing time | Wound closure is complete | Operative report, nursing record |
| Out-of-room time | Patient leaves the operating room | Nursing record, anesthesia record |
Many operative reports print only a start and end time, without saying which events they mark. Label those as written ("Start", "End") and do not rename them "incision" and "closing".
When the op note and the anesthesia record disagree
The operative report and the anesthesia record are written by different people, often minutes apart. A dictated note may give a rounded time, while the anesthesia record logs events as they happen. When the two disagree, keep both values in separate fields with their source document and author. Then flag the pair for review instead of averaging them or picking the later one.
Operative report addenda and late dictation that break extraction
An operative report addendum is text added after the original report, and late dictation is a report written or signed well after surgery. Both break extraction that assumes one report, one date and one author. Each version needs its own date, time and signer, and an addendum must never overwrite the text it amends.
42 CFR 482.51(b)(6) says the report must be "written or dictated immediately following surgery". In practice, a scanned chart can still hold a report with a dictation date, a transcription date and a signature date days apart. Extract all three. The procedure date comes from the report body, never from the signature line.
Addenda take a few forms:
- A correction, such as a changed implant size or a corrected laterality.
- An addition, such as a specimen the original report left out.
- A late attestation, where an attending signs a report dictated by a resident.
Store the addendum as its own record linked to the original, with its date, time, author and full text. If an addendum changes a field value, keep the original value too, marked as superseded, with a pointer to the addendum. A reviewer can then see the before and after side by side.
The same pattern of draft, final and amended versions appears in imaging. The post on radiology report extraction covers preliminary reads and addenda there.
A field map for a fictional operative report
This field map follows one fictional report from start to finish. The patient, staff, device and every number are examples, not real data.
| Field | Example value | Report section |
|---|---|---|
| Patient | Marisol Quenby (fictional), record EX-204417 | Header |
| Procedure date | 2026-03-04 (example) | Header |
| Surgeon | Dr. Anders Lindqvist, MD (fictional) | Header |
| Assistant | Rhea Basu, PA-C (fictional), role "First assistant" | Header |
| Anesthesia type | "General endotracheal" | Anesthesia |
| Preoperative diagnosis | "Right inguinal hernia" | Preoperative Diagnosis |
| Postoperative diagnosis | "Right indirect inguinal hernia" | Postoperative Diagnosis |
| Procedure | "Open right inguinal hernia repair with mesh" | Procedure |
| Implant | Polypropylene mesh, 15 x 10 cm, lot "MX40-118Q" (example) | Implants, implant log sticker |
| Specimens | "A. Hernia sac, to pathology" | Specimens |
| Estimated blood loss | "<10 mL" | Estimated Blood Loss |
| Drains | "None" | Drains |
| Incision time | 08:12 (op note); 08:09 (anesthesia record), flagged | Procedure, anesthesia record |
| Closing time | 09:01 | Procedure |
| Signature | Dr. Anders Lindqvist, 2026-03-06 16:40 (example) | Signature |
| Addendum | 2026-03-07: mesh size corrected from "15 x 15 cm" | Addendum |
Three rows in this map should go to a person before the data is used: the lot number, the conflicting incision times and the addendum that changed the mesh size.
The history and physical written before this surgery is extracted with different rules. Its negatives and templated exam text are covered in history and physical extraction.
Operative reports are not named among the document types on Schema Care's supported documents section, so treat this field map as a method you can apply with any tool or by hand. The review idea still carries over: every uncertain value goes to a person who can see it on the original page, the approach set out in confidence-based human review.