Fetal monitoring strips in birth injury cases

Electronic fetal monitoring (EFM) records 2 channels on 1 time axis. The top channel is the fetal heart rate in beats per minute. The bottom channel is uterine activity: the timing of contractions and, depending on the sensor, their strength. The printed or displayed result is called the strip, the tracing or, in some hospitals, the CTG (cardiotocograph). In a lawsuit it gets called all 3, sometimes in the same deposition.

Each channel can come from 1 of 2 sources, and the source changes what the tracing can show.

The source matters to a reviewer for a plain reason. ACOG's 2025 guideline notes that when staff have ongoing difficulty telling the maternal from the fetal heart rate, "placing a fetal scalp electrode may provide more accurate monitoring." If the chart has a stretch where staff wrote that the tracing was hard to read, the expert will want to know which sensor was in use, when it changed and whether maternal pulse was recorded at the same time. Those facts sit in different places: the monitor stores the sensor mode with the tracing, the nurse writes the placement in the flowsheet, and the physician may mention it in a progress note.

What a printed strip page carries

A printed or exported page from a perinatal system carries more than the 2 lines. Learn to read the frame before anyone argues about the line.

Paper versus the archive

In older units the strip came out of the bedside monitor on folded paper and was boxed after delivery. Most hospitals now run a perinatal surveillance system that captures the tracing electronically, shows it at a central station and archives it. The printed pages in your production are often a print job someone ran after a request arrived. The 2009 bulletin was direct about storage: "Computer storage of the FHR tracing that does not permit overwriting or revisions is reasonable, as is microfilm recording." That sentence is why the native archive, not the print job, is the record to ask for (chapter 7).

The tracing is often not inside the main EHR at all. The EHR may hold only what the nurse charted about it: baseline, variability, decelerations, category, and notes. That split is the source of most strip problems in litigation, and it is why a release-of-information office can honestly send "the complete chart" without a single page of tracing.

Expert reports in birth injury cases lean on ACOG guidance, and the guidance changed in October 2025. ACOG Clinical Practice Guideline No. 10 replaced Practice Bulletin 106 (July 2009), Practice Bulletin 116 (November 2010) and a 2022 practice advisory on oxygen for Category II or III tracings. Most cases filed in 2026 involve deliveries before that date. The right question for every guideline cited in a report is: which version was in force on the day of the labor?

Standard of care is a legal question answered by expert testimony, and guidelines are only part of it. Still, the change between versions decides what a reviewer should look for in the chart, so it belongs in the case file from day 1.

2 practical effects follow. First, for a labor after October 2025, the review interval comes from the hospital's own policy, so the unit's EFM policy in force on the delivery date becomes a core document. Ask for it with the records, not after the depositions. Second, an intervention charted in 2019 may look different against the 2025 text. Oxygen is the clearest example: the same flowsheet entry reads differently under the 2 versions. Your expert decides what it means. Your job is to give them the dated version of every guideline next to the entries it touches.

"An FHR tracing may move back and forth between the categories depending on the clinical situation and management strategies used."

The 2008 NICHD workshop set the 3-tier vocabulary that labor nurses and obstetricians chart in the United States, and the 2025 ACOG guideline keeps it while noting its limits, especially inside Category II. Our HIE guide covers the categories and their link to timing an injury. This chapter does something narrower: for each category, what a reviewer should expect to find written in the chart near it. Whether care was right is the expert's call. Whether the chart says anything at all is yours.

The 2025 guideline lists features that make a Category II tracing higher risk: absent variability, prolonged and otherwise unexplained minimal variability, an unexplained change from a normal baseline to tachycardia, and recurrent late, recurrent variable, or more than 1 prolonged deceleration. It names moderate variability and accelerations as protective. A reviewer does not decide which features were present. A reviewer does check whether the chart describes them in words, because an expert who reads the strip as high-risk Category II will look for the nurse's description of the same minutes.

The words you will see in the flowsheet

Baseline

The mean heart rate over a 10-minute window, rounded to 5 bpm, present for at least 2 minutes of that window.

Variability

Fluctuation around the baseline: absent (undetectable), minimal (5 bpm or less), moderate (6 to 25), marked (over 25).

Acceleration

An abrupt rise; at 32 weeks or later, 15 bpm or more for 15 seconds or more but under 2 minutes.

Late deceleration

A gradual drop whose lowest point comes after the peak of the contraction.

Variable deceleration

An abrupt drop of 15 bpm or more, lasting 15 seconds to under 2 minutes, not tied to contraction timing in a fixed way.

Prolonged deceleration

A drop of 15 bpm or more lasting 2 minutes or more but under 10.

Sinusoidal

A smooth, wave-like pattern of 3 to 5 cycles a minute lasting 20 minutes or more.

Tachysystole

More than 5 contractions in 10 minutes, averaged over 30 minutes.

Category is a snapshot. ACOG's 2025 text says categorization "evaluates the fetus at that point in time." So when the flowsheet says "Category II" at 01:00, 01:30 and 02:00, you have 3 snapshots of a continuous record. The rest of this guide is about the minutes between them.

Tracing information is spread across 6 to 10 places in a typical labor chart. Missing 1 of them is how a timeline ends up with a 40-minute hole that was never a hole. Search each location below, in this order, and log every entry with its page and its clock.

Late entries and copy-forward text

Labor and delivery charting is often finished after the fact. A note filed at 05:10 that describes events at 02:30 is a late entry, and most EHRs store both the event time the nurse chose and the time the note was saved. Log both. When the 2 are far apart, the expert and counsel will want to know, and the place to confirm the save time is the EHR's audit trail, which is a separate request (see EHR audit trails in medical malpractice).

Ambient AI scribes now draft some physician notes in labor units. A scribe-drafted note can carry the time the draft was generated, not the time of the exam it describes, and it may echo phrasing from earlier notes. Treat those times like any other late entry: as the time of writing until something else proves the time of the event.

Hypothetical. 2 versions of 1 note, both dated to the same event time, saved 3 hours apart. Neither version proves what happened at 02:20. Both belong in the timeline with their save times.

The fetal monitor, the EHR, the anesthesia system, the pump and the lab analyzer each keep their own clock. They are supposed to agree, and often they are a few minutes apart. In a case where the argument is about a 10-minute delay, a 6-minute clock offset is the whole case. Our HIE guide explains how to anchor the clocks at the time of birth. Strip work needs more anchors than that, spread across the labor, because offsets can drift and because the monitor may have been restarted or swapped.

How to find the offset

1. List every clock. Monitor or archive server, EHR, anesthesia record, infusion pumps, lab analyzer, any paper form timed from a wall clock.
2. Find events recorded by 2 systems. Good anchors: an epidural bolus (anesthesia record and a tracing annotation), an FSE placement (tracing mode change and flowsheet entry), the move to the OR (end of tracing and OR log), birth.
3. Compute the difference for each anchor. Write the times as recorded and the difference in minutes. Do not adjust anything yet.
4. Check that the offset is stable. 3 anchors that agree within a minute support 1 offset for the labor. Anchors that disagree point to drift, a restart or a wrong entry, and each needs a note.
5. Show both times in the timeline. Keep the time as recorded in its own column and add a converted time only where the offset is supported. Never overwrite the original.

Hypothetical. The same labor appears in chapter 6 and in the sample timeline.

Annotations versus note times

Annotations are useful because they are stamped by the monitor at the moment someone entered them. They are limited because they are short, often abbreviated, and entered by whoever was at the central station. A note saying "MD notified 02:22" and an annotation reading "MD aware" at 02:31 monitor time (02:25 corrected) are 2 statements about 1 event, 3 minutes apart after correction. Put both in the timeline with their sources. Do not choose between them.

Notice what is not in the timeline: no category assigned by the reviewer, no statement that a deceleration was late or variable, no comment on whether 02:40 was soon enough. Those are clinical judgments. A reviewer who writes them into the timeline gives opposing counsel a reason to attack the whole document.

Notice also what the timeline makes easy. An obstetric expert can open tracing segment 19 and the progress note side by side. A nursing expert can see the 2 versions of the 02:20 note. Counsel can see exactly which tracing minutes were never produced and draft the follow-up request from the gap row (chapter 7).

The common versions of the problem: a stack of printed pages that covers the last 40 minutes, a scan of a scan with an unreadable time axis, a production with no tracing at all, or a letter saying the tracing is not part of the legal medical record. Our HIE guide has a decision path for what to ask next. This chapter is the procedure: what the archive holds, how to ask for it, how to document gaps and what the law says when data is lost.

What the perinatal archive holds

Perinatal surveillance systems store the tracing as data, not as pictures. A complete export usually includes both channels at full resolution, the sensor mode for each period, annotations with their times, alarms or alerts where the system logs them, and system events such as disconnections. Many systems can also report who viewed, printed or annotated a tracing and when. The release-of-information office often cannot produce any of this because it does not run the system. Nursing informatics or clinical engineering does.

For a patient's own request, HIPAA gives a right to a copy in the electronic form and format requested if it is readily producible (45 CFR 164.524(c)(2)(ii)), within 30 days with 1 extension of up to 30 more. Whether the tracing falls inside the hospital's designated record set, and how discovery works in litigation, depends on the hospital's policies, state law and your court's rules. Ask anyway, by name, and get the answer in writing.

Documenting the gaps

A complaint that "the strip is incomplete" goes nowhere. A table of specific missing minutes, each tied to a chart entry showing monitoring was running, forces a specific written answer. Build it like this:

Some gaps have plain explanations: the transfer to the OR, a bathroom break, a monitor swap, a period of intermittent auscultation under unit policy. Log the explanation the hospital gives and its source. The gap table is a record of questions and answers, not an accusation.

Spoliation, in plain terms

When evidence that should have been preserved is lost, courts have remedies, and they vary. In federal court, Rule 37(e) governs lost electronically stored information. It limits the harshest remedies to cases where a party acted with intent:

"only upon finding that the party acted with the intent to deprive another party of the information's use in the litigation may" the court presume the information was unfavorable.

Short of intent, a federal court that finds prejudice may order measures "no greater than necessary to cure the prejudice." Most birth injury cases are in state court, where spoliation doctrine, adverse inference instructions and any separate tort claim differ from state to state. Some states have specific rules. Virginia's birth injury compensation act deems fetal monitoring strips, "whether printed or in electronic format," part of the infant's medical records for that program. Counsel decides what to do with a gap. The reviewer's job is to make sure the record of the gap is exact: what was asked, when, by whom, and what came back.

Interpreting a fetal heart rate tracing is a clinical opinion. In litigation it comes from a qualified expert: an obstetrician, a maternal-fetal medicine specialist, or a labor and delivery nurse expert for nursing standard of care, depending on the issue and the court's expert rules. AWHONN's 2024 position statement says fetal heart monitoring "requires advanced assessment and clinical judgment skills" and should not be delegated to people without the licensure, education and skills validation for it. That is a statement about clinical care, but it is a fair guide for litigation work too.

The split has a practical reason beyond credentials. Tracing interpretation varies between readers, and the presentation of the tracing affects it. In a 2016 study, 14 physicians read the same 19 tracing segments at 1 or 3 cm/min; at 1 cm/min they more often called variability absent rather than minimal and minimal rather than moderate, and agreement between readers was higher at 3 cm/min. If a reviewer sends an expert a rescaled printout without saying so, the reviewer has changed the evidence.

Birth injury files are large. A long induction can produce hundreds of flowsheet rows, dozens of notes, pump logs, anesthesia records and a few hundred tracing pages. Legal AI tools are good at some of this work and should not touch the rest.

What AI should not do here

An AI system should not read the tracing image and assign a category, call a deceleration late or variable, or say whether a response was timely. Those are clinical opinions. The 2025 ACOG guideline notes there are no clinical trials showing that computer-based tracing interpretation improves newborn outcomes. A litigation tool that offered strip interpretation would be offering an unsigned clinical opinion. Medrecords AI does not interpret strips, assign NICHD categories or read the tracing line. It works on the text of the record: entries, times, orders, annotations printed on the pages, and what is missing.

Failure modes to check for

• Hallucinated factsA large language model (LLM) can write a fluent sentence that no page supports. Require a page-level citation on every line and click a sample of them.
• OCR errors on timesFaxed and scanned flowsheets turn 01:38 into 01:58. Check every time that drives a conflict flag against the page image.
• Missed handwritten notesPaper delivery records and resuscitation forms are often handwritten. Confirm the tool flagged low-confidence pages and that someone read them.
• Copy-forward text dated wrongText carried forward from an earlier note can carry an old time. Compare the note's save time to the event time.
• Monitor and EHR times mergedA chronology that silently mixes clocks hides the offset. Keep the source clock in a column.
• Generative AI summaries without sourcesA paragraph summary of labor is useful only if each sentence links back to the record.

0 of 6 checked

Courts have sanctioned lawyers under Rule 11 for filing fabricated citations produced by generative AI; Mata v. Avianca (S.D.N.Y. 2023) is the case most people know. The same discipline applies to medical facts: every date and time in a filing should trace to a page someone opened. Retrieval-augmented generation with grounded citations makes that checking fast. It does not remove the need for it. Human-in-the-loop review is the method, not a disclaimer.

Vendor checklist for birth injury work

1. HIPAA compliant AI with a signed business associate agreement (BAA). No BAA, no upload.
2. SOC 2 report available on request, and a clear answer on whether your data trains any model (it should not).
3. A citation on every line of the chronology, linked to the source page.
4. Source clock kept with each time, so monitor and EHR times are never merged silently.
5. Low-confidence OCR pages flagged, including handwritten notes.
6. No clinical interpretation of the tracing, in the product or the marketing.
7. A log of who ran what on your file inside the tool, for your own chain of custody.

Copy these into your own forms. Adapt the request to your jurisdiction and your rules of discovery; counsel decides what gets sent.

EFM RECORDS REQUEST CHECKLIST
Patient: [NAME]   DOB: [DATE]   Admission: [DATE] to [DATE]
Facility: [HOSPITAL]   Delivery date and time: [DATE, TIME]

1. Name of the perinatal surveillance or EFM archive system in use on the
   delivery date, its version, and the department that administers it.
2. The complete fetal heart rate and uterine activity tracing from the first
   minute of monitoring (including triage and any antepartum testing in this
   admission) to delivery, in native electronic format or a complete export
   that keeps the time axis, both channels and system time stamps.
3. A viewer or instructions to open the native file, if needed.
4. All annotations and event marks stored with the tracing, with times.
5. Sensor mode history: external ultrasound or fetal scalp electrode, toco or
   intrauterine pressure catheter, with the time of each change.
6. Maternal heart rate or pulse oximetry channel, if recorded.
7. Alarms, alerts and system event logs (disconnections, downtime, restarts)
   for the admission.
8. Access and print history for the tracing: who viewed, printed, annotated
   or exported it, and when.
9. A written statement of any period with no tracing data and the reason.
10. The unit's EFM and documentation policies in force on the delivery date,
    including tracing review frequency and notification procedures.
11. Monitor and archive server time synchronization records, if kept.
12. Retention policy for tracings in force on the delivery date.

STRIP-TO-CHART TIMELINE WORKSHEET
Case: [MATTER]   Reviewer: [NAME]   Offset used: monitor [+/- N] min vs EHR
Anchors: [EVENT 1, cites] [EVENT 2, cites] [EVENT 3, cites]

| Time as recorded | Source clock | Corrected EHR time | Source (doc, page or segment) | Entry, quoted or close paraphrase | Saved time if later | Flag |
|------------------|--------------|--------------------|-------------------------------|------------------------------------|---------------------|------|
| [HH:MM]          | [Monitor/EHR/Anes/Pump/Lab] | [HH:MM] | [e.g. Flowsheet p. 214]   | [ENTRY]                            | [HH:MM]             | [conflict / gap / late entry / none] |

Flag rules:
- conflict: 2 sources describe 1 event with times that differ after correction
- gap: chart describes monitoring in a period with no tracing pages
- late entry: saved time more than [N] minutes after the event time
No category assignments or clinical judgments in this worksheet.

FETAL MONITORING TRACING: GAP TABLE
Patient: [NAME]   Facility: [HOSPITAL]   Production reviewed: [BATES RANGE]

| # | Missing period (EHR time) | Minutes | Evidence monitoring was running (cite) | Date asked | Response (date, source) |
|---|---------------------------|---------|----------------------------------------|------------|-------------------------|
| 1 | [HH:MM to HH:MM]          | [N]     | [e.g. Flowsheet 03:05 "FHR 90s", p. 216] | [DATE]   | [RESPONSE]              |

For each row, please state: whether tracing data exists for the period;
if so, produce it in native format; if not, why not (disconnection,
transfer, downtime, intermittent auscultation, loss), with supporting logs.

Medrecords AI is medical chronology software for legal and claims teams. You upload the records you have, and it reads about 100 pages a minute; it drafts a cited AI medical chronology in which every entry links to its source page. For a birth injury file, that means every tracing-related entry the chart contains, the flowsheet assessments, nursing notes, oxytocin rate changes, provider notes, orders, OR and anesthesia times, organized in time order with a citation on every line.

• It flags entries about the same event that carry different times, so a person can check the clocks.
• It flags periods the chart describes that have no matching pages in the production, through missing records identification, each flag cited to the entries that imply the record should exist.
• It shows near-duplicate pages that differ side by side through record alteration detection, which surfaces addenda and late versions. A flag is a signal, not a verdict.
• It routes typed, handwritten and table pages to the right OCR engine and flags low-confidence pages.

What it does not do: it does not retrieve records from the hospital, read the tracing image, assign NICHD categories, judge the standard of care or give legal advice. It does not parse a hospital's EHR audit trail export. You review, you revise, you sign.

To see the output format, open the fetal monitoring timeline sample, a hand-built illustration on a fictional labor. For the wider field of tools, see our list of AI tools for birth injury law firms and the birth injury record review page.

What is a Category 2 fetal heart tracing?

A Category II (category 2) tracing is any fetal heart rate tracing that does not meet the definition of Category I or Category III under the 2008 NICHD system. It is the largest group and covers a wide range of patterns, which is why ACOG's 2025 guideline describes high-risk features within it. Deciding whether a specific tracing was Category II is an expert's clinical judgment.

What is a Category 3 fetal heart tracing?

A Category III (category 3) tracing has absent baseline variability with recurrent late decelerations, recurrent variable decelerations or bradycardia, or a sinusoidal pattern. ACOG's 2025 guideline recommends expedited delivery when a Category III tracing does not respond to initial intrauterine resuscitation, when indicated. Whether a tracing met that definition is a question for a qualified expert.

What is an EFM strip?

An EFM strip is the printed or displayed record from electronic fetal monitoring: the fetal heart rate on top and uterine activity below, on 1 time axis. In US hospitals it is usually printed at 3 cm per minute and stored electronically in a perinatal archive.

What if the fetal monitoring strips are missing?

Ask in writing which archive system captured the tracing, request preservation and a native export, and send the custodian a gap table listing each missing period with the chart entries showing monitoring was running. Remedies for lost evidence vary by court and state; counsel decides the next step.

Are fetal monitoring strips part of the medical record?

ACOG guidance treats the tracing as part of the medical record, and Virginia's birth injury act says so for that program. Whether a hospital produces it in response to a given request depends on its legal health record policy, state law and the discovery rules of the court.

Who can interpret a fetal monitoring strip in a lawsuit?

A qualified clinical expert, usually an obstetrician, a maternal-fetal medicine specialist or a labor and delivery nurse expert, depending on the issue and the court's rules. A record reviewer prepares the tracing inventory, the timeline and the gap table that the expert relies on.

Can AI interpret a fetal monitoring strip?

No AI tool should be used to interpret a tracing for litigation. ACOG's 2025 guideline notes no clinical trials show computer-based interpretation improves newborn outcomes. Medrecords AI does not read the tracing or assign categories; it organizes and cites the chart entries around it.

Is it HIPAA compliant to upload birth injury records to AI?

It can be, if the vendor signs a business associate agreement, protects the data under audited controls such as SOC 2, and does not train models on your records. Check those 3 things before any upload.

Can ChatGPT summarize labor and delivery records for a lawsuit?

A general chatbot can produce a fluent summary, but without page-level citations every fact has to be checked by hand, and consumer tools may not offer a business associate agreement. For litigation, use a tool that cites every line to its source page and keep a person reviewing the output.

How does AI medical record review help in a birth injury case?

It pulls every tracing-related entry, medication time, order and procedure time into 1 cited chronology, flags conflicting times and periods with no matching records, and flags low-confidence scanned or handwritten pages for a person to read.

Sources were opened and checked on 24 October 2026. Quotes are verbatim. The anatomy figure, timeline, crosswalk, note versions and gap rows are hypothetical. Nothing here is legal or medical advice.

• ACOG, Clinical Practice Guideline No. 10, Intrapartum Fetal Heart Rate Monitoring: Interpretation and Management, Obstetrics and Gynecology 146(4), October 2025. Replaces Practice Bulletins 106 and 116 and the January 2022 oxygen advisory. Checked 24 Oct 2026; current.
• ACOG, Practice Bulletin No. 106, Intrapartum fetal heart rate monitoring: nomenclature, interpretation, and general management principles, 2009 (reaffirmed 2017, replaced 2025). Definitions, review intervals and storage language.
• ACOG, Practice Bulletin No. 116, Management of intrapartum fetal heart rate tracings, 2010 (replaced 2025).
• Macones and colleagues, The 2008 NICHD workshop report on electronic fetal monitoring, Obstetrics and Gynecology 112(3), 2008.
• AWHONN, Fetal Heart Monitoring position statement, Journal of Obstetric, Gynecologic and Neonatal Nursing 53(3), revised April 2024. Checked 24 Oct 2026; current per AWHONN's position statement list.
• Peleg and colleagues, The effect of chart speed on fetal monitor interpretation, J Matern Fetal Neonatal Med, 2016.
• ACOG and AAP, Neonatal Encephalopathy and Neurologic Outcome, 2nd ed., 2014: tachysystole definition. Covered in depth in our HIE guide.
• 45 CFR 164.524, HIPAA right of access: (b)(2) and (c)(2)(ii).
• Federal Rule of Civil Procedure 37(e), failure to preserve electronically stored information.
• Va. Code 38.2-5004(F), fetal monitoring strips deemed part of the medical record for the Virginia program.
• Mata v. Avianca, Inc., 678 F. Supp. 3d 443 (S.D.N.Y. 2023): Rule 11 sanctions for fabricated AI-generated citations.

Related guides: HIE birth injury medical records, EHR audit trails in medical malpractice, altered medical records and how legal nurse consultants build a medical chronology.