The eggshell plaintiff rule and pre-existing conditions: how prior records decide the case
For personal injury attorneys, paralegals, adjusters and IME physicians working a claim with a medical history behind it. You walk away with the 3 doctrines kept apart, the imaging studies that reframe degeneration, a worked timeline and 3 templates for a before/after record review.
The eggshell plaintiff rule says a defendant takes the injured person as they find them: if a hidden weakness makes the harm worse than a healthy person would suffer, the defendant is liable for the full harm the incident caused. It does not make the defendant pay for the condition itself. Aggravation law limits damages to the worsening, and apportionment splits a single harm between causes when the records give a reasonable basis.
Almost every adult injury claimant arrives with a medical history: an old back strain, a knee that clicks, a neck MRI from 3 years ago, a chiropractor they saw for a season. The law on pre-existing conditions fits on a page. The fight is over the prior records, and whoever reads them first and most carefully usually frames the damages case.
Pre-existing conditions in 8 numbers
3 doctrines that get blurred together
Lawyers, adjusters and doctors use "eggshell plaintiff" loosely, as if it covered every claimant with a history. It does not. There are 3 separate ideas at work, and each 1 answers a different question.
The eggshell plaintiff rule (thin skull rule)
The rule answers the question of extent: can a defendant argue that a normal, healthy person would have been hurt less? No. The Restatement (Second) of Torts puts it this way:
"The negligent actor is subject to liability for harm to another although a physical condition of the other which is neither known nor should be known to the actor makes the injury greater than that which the actor as a reasonable man should have foreseen as a probable result of his conduct."
The Restatement (Third) of Torts: Liability for Physical and Emotional Harm, section 31, carries the same rule forward and extends it to mental conditions and to harm of a different type as well as a greater amount.
Aggravation of a pre-existing condition
Aggravation answers a different question: what exactly is the defendant paying for? The answer is the worsening, not the condition. A claimant with a lumbar disc that was already degenerated and already painful recovers for the added pain, added treatment and added limitation the incident caused. They do not recover for the back they already had. This is where the prior records carry the whole load, because "worse" only means something against a documented "before".
Apportionment of damages
Apportionment answers the question of division: when 2 causes combine, can the harm be split? Section 433A of the Restatement (Second) says damages are apportioned among 2 or more causes when there are distinct harms, or when there is a "reasonable basis for determining the contribution of each cause to a single harm". A broken wrist in the crash and a separate knee problem from a later fall are distinct harms. A single disc that was degenerated and then herniated is a single harm, and it can be split only if the evidence gives a reasonable basis. When it cannot be split, many courts hold the defendant liable for the whole (chapter 2 covers who carries that burden).
| Question | Eggshell plaintiff rule | Aggravation | Apportionment |
|---|---|---|---|
| What it decides | Whether fragility reduces damages (it does not) | That damages run from the baseline, not from zero | Whether 1 harm can be divided between causes |
| Typical claimant | Hidden weakness, often no prior symptoms | Known condition, often with prior treatment | Prior condition plus the incident, or 2 incidents |
| Main authority | Restatement (Second) 461; Restatement (Third) 31; CACI 3928 | CACI 3927 and state equivalents | Restatement (Second) 433A; state case law |
| Who it helps | Mostly the plaintiff | Both: it caps damages and protects the added harm | Whoever has the better baseline evidence |
| Records that decide it | Post-incident treatment and the causation opinion | Prior records for the same body part, compared with post-incident records | Dated, quantified before and after measures: doses, visit counts, restrictions, imaging |
What jurors are told, and who has to prove the split
Pattern jury instructions are the clearest statement of how a state applies these doctrines, because they are the words the jury actually hears. California splits the ideas into 2 instructions, and reading them side by side shows how the doctrines work together.
"[Name of plaintiff] is not entitled to damages for any physical or emotional condition that [he/she/nonbinary pronoun] had before [name of defendant]'s conduct occurred. However, if [name of plaintiff] had a physical or emotional condition that was made worse by [name of defendant]'s wrongful conduct, you must award damages that will reasonably and fairly compensate [him/her/nonbinary pronoun] for the effect on that condition."
"You must decide the full amount of money that will reasonably and fairly compensate [name of plaintiff] for all damages caused by the wrongful conduct of [name of defendant], even if [name of plaintiff] was more susceptible to injury than a normally healthy person would have been, and even if a normally healthy person would not have suffered similar injury."
Read 3927 as a fence and 3928 as a floor. The fence keeps the old condition out of the award. The floor stops the defense from arguing the claimant should have been sturdier. A jury that hears both is being asked to do 1 thing: measure the change. It can only do that if someone shows it a clear before and a clear after.
The "bound to worsen" discount
The defense has a counterweight to the eggshell rule. In Steinhauser v. Hertz Corp., 421 F.2d 1169 (2d Cir. 1970), a teenager with no physical injuries developed schizophrenia after a collision; the experts agreed she had a predisposition and that the accident precipitated the illness. The Second Circuit held the defendant liable, and said that if the defendant proves the pre-existing condition would have worsened anyway, the damages should be discounted for what would have happened without the accident. That is an evidence question. It turns on the trajectory in the prior records: stable, improving or progressing.
Who bears the burden when the harm cannot be split
States differ here, and the difference changes how you build the review. Maine's highest court put the burden on the defendant in Lovely v. Allstate Insurance Co., 658 A.2d 1091 (Me. 1995), a case about an elbow injured in March 1985 and hurt again in an April 1985 car accident. As quoted by the federal court in Bouchard:
"single injury rule places any hardship resulting from the difficulty of apportionment on the proven wrongdoer and not on the innocent plaintiff."
The same opinion noted that even the cases placing the burden on the plaintiff to produce apportionment evidence accept that the defendant is liable for all of an indivisible injury. So the practical split is this: in some states, the plaintiff must bring evidence that lets the jury divide the harm, and in others the defendant must. Either way, the side without a documented baseline loses the argument by default.
What apportionment evidence looks like in practice
Bouchard is worth reading in full for how a judge actually used records. The claimant had serious problems before a 2003 collision, including opioid dependence treated with methadone. The treating physician's records showed 10 milligrams 4 times a day just before the accident and a dose nearly 5 times that level afterward. The court treated that numeric change as the best evidence of the accident's effect, found the government had met its burden to separate out 1/5 of the post-accident methadone treatment as pre-existing, and found it had "largely failed" to apportion the rest of the claimant's combined injuries. The judgment was $1,100,000.
The lesson: apportionment is built on quantities from dated records (doses, visit counts, pain scores, restrictions). Adjectives like "chronic" do not split anything.
The 4 layers of a pre-existing condition claim
It helps to picture every body part in the claim as a stack of layers. Each layer has its own evidence, and each side is trying to move harm from 1 layer to another. The figure below is an illustration of a hypothetical lumbar claim, not data from any case.
The widths are invented for illustration. In a real file, each layer's size is argued from dated records and expert opinion.
- Baseline condition
- What the body part looked like and how it worked before the incident: symptoms, findings, treatment, restrictions and activity. The defendant owes nothing for this layer.
- Acute injury
- New harm the incident caused on its own: a fracture, a sprain, a new tear. Pre-existing conditions do not affect this layer at all.
- Aggravation
- The worsening of the baseline condition. Lawyers also call it "activation" or "lighting up" when a quiet condition turns symptomatic. The defendant pays for this layer.
- Acceleration
- The incident brought on sooner what the condition would have produced later. This is where the Steinhauser discount lives.
- Asymptomatic but present
- A condition visible on imaging that caused no complaints, no treatment and no limits. The strongest baseline a plaintiff can have, and one the records must prove, not assume.
A 2021 therapy discharge saying "goals met, pain 1/10, full duty" shrinks the baseline layer. A 2023 note saying "chronic low back pain" swells it. A 2025 fall down stairs creates a disputed layer on its own.
Degeneration on imaging in people with no pain
The most common defense exhibit in a spine or joint claim is a post-incident MRI that says "degenerative" and a prior study that says the same. The most useful response is not rhetoric. It is the published prevalence of those same findings in people who have no symptoms at all.
The largest summary is a systematic review in the American Journal of Neuroradiology by Brinjikji and colleagues (2015). It pooled 33 studies covering 3,110 people without low back pain, almost all imaged with MRI, and modeled how often each lumbar finding appears at each age.
By 40, a majority of people with no back pain have disk degeneration on imaging. The finding alone cannot tell you whether someone hurt before.
The same table covers 7 more findings. Bulges climb with age like degeneration; protrusions barely move; facet degeneration and spondylolisthesis start rare and become common late in life.
| Lumbar finding (no back pain) | Age 20 | Age 40 | Age 60 | Age 80 |
|---|---|---|---|---|
| Disk degeneration | 37% | 68% | 88% | 96% |
| Disk signal loss | 17% | 54% | 86% | 97% |
| Disk height loss | 24% | 45% | 67% | 84% |
| Disk bulge | 30% | 50% | 69% | 84% |
| Disk protrusion | 29% | 33% | 38% | 43% |
| Annular fissure | 19% | 22% | 25% | 29% |
| Facet degeneration | 4% | 18% | 50% | 83% |
| Spondylolisthesis | 3% | 8% | 23% | 50% |
Other joints show the same pattern.
Of the people in the Framingham knee sample who had a meniscal tear on MRI, 61% reported no pain, aching or stiffness in that knee during the previous month. Englund et al., NEJM 2008, 991 subjects.
How to use these numbers honestly
These studies cut both ways, and a careful reviewer uses them that way. Brinjikji's authors say their findings apply mainly to degenerative changes seen incidentally in people without back pain, and they do not claim degenerative findings are always harmless when they match the patient's symptoms and exam. So the numbers support a narrow point: a degenerative finding, standing alone, does not prove the claimant was symptomatic before. They do not prove the incident caused anything either.
What carries weight is change and correlation. A new extrusion at the level that matches new leg pain, in a dermatome no prior note mentions, is a different fact from a bulge that looks the same on both studies. That is why chapter 9 asks you to compare imaging by level, side and finding, and to get the images themselves along with the report.
What the defense looks for in prior records
A defense record review of a pre-existing condition claim follows a predictable order. Plaintiff teams should run it on their own client first. The ranking reflects how directly each item answers the aggravation question.
Prior complaints for the same body part
Primary care, urgent care and ED notes that mention neck, back or knee pain, even in passing during a visit for something else. The side, level and words matter: "low back pain radiating to left leg" in 2022 is a very different fact from "LBP" on a problem list.
Prior imaging of the same region
X-rays, MRIs and CTs, with the impression and the date. The defense wants the same finding on both sides of the incident, and the phrase "unchanged from prior" in the post-incident report.
Chiropractic and physical therapy
Visit counts, intake pain scores, functional goals and discharge status. A chiropractic intake questionnaire often holds the frankest history in the file.
Pain management and prescriptions
Injections, opioid agreements, muscle relaxants and refill histories. A pharmacy printout can show ongoing symptoms that no visit note mentions.
Gaps in treatment
A gap after the incident invites "the claimant got better, then something else happened". A short gap before the incident invites "the claimant was still treating".
Other events and prior claims
Earlier accidents, workers' comp claims, disability applications, sports injuries and falls, before or after the incident. They show up in histories, in claims data and in intake forms.
Reading the same pages from both sides
What the defense reads for
Continuity: the same problem, before and after
- The same complaint in similar words before and after
- Imaging described as "chronic", "degenerative" or "unchanged"
- Treatment after the incident that matches the prior pattern
- A post-incident gap followed by a new event
- Intake forms where the claimant denied prior problems that the records show
What the plaintiff reads for
Change: a new problem, or a clear step up
- A long quiet period with no complaints for the region
- Discharge notes: goals met, full duty, pain resolved
- New symptom types: radicular pain, numbness, weakness
- New findings at the matching level and side
- Escalation: injections, surgical referral, new specialists
Why the claimant's own words carry weight
The history in a treating record is evidence. Federal Rule of Evidence 803(4) lets in, as a hearsay exception, a statement made for medical diagnosis or treatment that describes "medical history; past or present symptoms or sensations; their inception; or their general cause" (FRE 803(4)). Many states have a similar rule. A 2022 intake form where the claimant wrote "back pain on and off for 10 years" can come in against them. So can the post-incident ED note where they said "never had back problems before". Find both before the deposition.
Watch the copy-forward trap. A "history of low back pain" line repeated in every note for 6 years may trace back to 1 visit in 2018. Problem lists and past medical history sections get carried forward by EHR templates, and ambient AI scribes that draft notes from recorded conversation can now put an offhand remark into the history of present illness. Find the original entry and date it before you let anyone call the condition "chronic".
How the plaintiff proves a baseline
A baseline is not a feeling that the claimant "was fine before". It is a set of documented facts about symptoms, findings, treatment, work and activity during a defined period before the incident. The best baselines combine medical records with records that were never written for litigation.
| Evidence | What it proves | Where to find it | Weakness to expect |
|---|---|---|---|
| Quiet medical period | No complaints, visits or prescriptions for the region | Primary care notes, especially annual physicals with a normal musculoskeletal exam | Absence of complaint is not proof of absence of pain |
| Discharge notes | Prior treatment ended with goals met | PT and chiropractic discharge summaries | Many patients simply stop attending; no discharge note exists |
| Prior imaging comparison | Which findings are new at the matching level and side | The prior study's images and report, compared with the new study | Different machines, protocols and readers |
| Work history | Full duty, no restrictions, no missed time | Payroll, attendance records, job descriptions, employer testimony | Jobs change; the defense will look for light-duty periods |
| Activity | What the claimant did before: sports, gym, yard work, childcare | Gym check-ins, league rosters, photos, witness statements | Social media from the post-incident period can cut the other way |
| Pharmacy history | No refills for the region's medications | Pharmacy printouts, health plan claims | Cash purchases and samples leave no trail |
"Asymptomatic but present" is a claim you have to prove
The strongest plaintiff baseline is a condition visible on imaging that caused nothing. It is also the easiest to lose: 1 overlooked urgent care note ends it. Prove it in 3 parts:
- Define the window. State the look-back period you searched, often several years for spine and joint claims, and list every provider in it. There is no universal rule for how far back; courts decide discovery scope case by case, and the right answer depends on the body part and the claimant's history.
- Show the silence is real. Prove the claimant saw doctors during the window for other reasons. A quiet period in which the claimant had 4 physicals, each with a normal back exam, is far stronger than a quiet period with no medical visits at all.
- Tie the imaging to the silence. If a prior study exists, show its findings were incidental: ordered for something else, or followed by no treatment. Then show what is new on the post-incident study at the level that matches the new symptoms.
When the claimant was symptomatic before
Prior treatment does not end a claim. It changes the job. Now the plaintiff must show a step change: new symptom types, a new level of treatment (from chiropractic to injections to a surgical referral), new restrictions, or a clear drop in function. The Bouchard dose comparison is the model. Pick the measures that were recorded before and after in the same units, and put them next to each other.
Worked example: 5 years of prior care, a gap, then a crash
The file holds 1,140 pages from 9 providers. Page cites below use a provider prefix and page number. Each entry shows what the record says and how each side will use it.
- 2019-06-04Chiropractic intake: low back pain after lifting at home
Pain 6/10, no leg symptoms. 8 visits through August. Defense: prior treatment, same region. Plaintiff: axial pain only, no radicular symptoms.
CHIRO p. 1 to 22 - 2019-07-15Lumbar X-ray
Mild L4-5 disc space narrowing. Defense: degeneration documented 5 years early. Plaintiff: common finding for age (chapter 4).
IMG p. 3 - 2020-11-02Primary care: back flare after yard work
Cyclobenzaprine 10 mg, 20 tablets. Defense: recurring episodes. Plaintiff: 1 visit in 16 months, resolved without follow-up.
PCP p. 31 - 2021-02-10Lumbar MRI
L4-5 desiccation, mild broad-based bulge, no stenosis, no nerve root contact. Defense: structural disease before the crash. Plaintiff: no nerve involvement at baseline.
IMG p. 7 to 8 - 2021-04-22Physical therapy discharge
6 visits. "Goals met, pain 1/10, returned to full duty." Plaintiff: the anchor of the baseline. Defense: pain was still present at discharge.
PT1 p. 18 - 2021-04 to 2024-0537 months with no back complaints
5 primary care visits, including 2 annual physicals with a normal back exam. Last cyclobenzaprine fill March 2021. Plaintiff: a documented quiet period with medical contact. Defense: absence of complaint is not absence of pain.
PCP p. 40 to 66, RX p. 2 - 2024-05-17Rear-end collision, ED visit
Neck and low back pain, new left leg pain below the knee. ED history: "denies prior back problems." Defense: the denial contradicts 2019 to 2021 records. Plaintiff: new radicular pain no prior note mentions.
ED p. 2 to 9 - 2024-06-20Lumbar MRI
L4-5 left paracentral extrusion contacting the descending left L5 root. Desiccation "unchanged from 02/2021". Defense: "unchanged". Plaintiff: the extrusion and root contact are new, at the level and side of the new symptoms.
IMG p. 12 to 13 - 2024-07 to 2024-10Physical therapy and 2 epidural injections
18 PT visits; injections in August and October. Plaintiff: treatment level above anything before. Defense: conservative care similar in kind to 2019 to 2021.
PT2 p. 1 to 64, PM p. 5 to 19 - 2024-11 to 2025-019-week gap in visits
No visits; gabapentin refills continue monthly. Defense: improvement, or a new cause. Plaintiff: authorization delay, and the refills show symptoms continued.
RX p. 4, PM p. 20 - 2025-02-12Spine surgeon: microdiscectomy recommended
Restrictions: no lifting over 20 lb. Plaintiff: first-ever surgical recommendation and first restriction. Defense: surgery for a degenerative level.
ORTHO p. 44 to 47
Both sides cite the same 11 entries. The case turns on 4 of them: the 2021 MRI, the 2021 discharge, the 37-month quiet period and the 2024 MRI.
What this timeline tells each side
For the plaintiff, this is an aggravation case with a strong baseline. An "asymptomatic before" theory would be false. The honest theory is better anyway: a treated, resolved lumbar strain with degeneration and no nerve involvement, followed by a 37-month documented quiet period, followed by a new extrusion with root contact and new radicular pain. Damages start from the 2021 discharge, which is a strong place to start.
For the defense, the ED denial is the most useful page in the file for credibility, and the 9-week gap is worth a question at deposition. But the imaging is the weak point for a pure "pre-existing" theory: the prior MRI says no nerve root contact, and the new one says contact at the matching level and side. An evaluation that ignores that comparison will be wrong.
For the IME physician, the report has to address each of the 4 anchor entries by page. An opinion that calls the extrusion "degenerative" without addressing the 2021 study's "no nerve root contact" invites the obvious cross-examination.
You found a prior record for the same body part: what next
The moment a prior record surfaces is where most reviews go wrong, in both directions. Plaintiff teams panic and minimize it. Defense teams declare victory and stop reading. Run it through these questions instead.
The prior record changes the theory of the case. It rarely ends it.
Requesting what the prior record points to
A prior note is a map to other records. The referral in a 2020 primary care note points to an orthopedist. The "per outside MRI" line points to an imaging center. The pharmacy on the medication list points to a refill history. Request those before you rely on the baseline. Under HIPAA, a covered provider must act on a patient's own access request within 30 days, with 1 extension of up to 30 days, and may charge only a reasonable, cost-based fee (45 CFR 164.524). Requests by authorization or subpoena follow different rules; see how to get medical records for a lawsuit.
1. Prior-records request checklist
Use it at intake and again after the first production arrives. Fill in the brackets; strike lines that do not apply and say why.
PRIOR-RECORDS REQUEST CHECKLIST Matter: [CASE NAME] Claimant DOB: [DOB] Incident date: [DATE] Claimed body parts: [e.g. lumbar spine L4-5, left knee] Look-back window: [START DATE] to [INCIDENT DATE] Reason for window: [WHY] A. Providers the claimant named [ ] Primary care: [NAME] ............ requested [DATE] received [DATE] [ ] Urgent care / ED: [NAME] ........ requested [DATE] received [DATE] [ ] Chiropractic: [NAME] ............ requested [DATE] received [DATE] [ ] Physical therapy: [NAME] ........ requested [DATE] received [DATE] [ ] Orthopedics / spine: [NAME] ..... requested [DATE] received [DATE] [ ] Pain management: [NAME] ......... requested [DATE] received [DATE] B. Providers the records point to (fill in as you read) [ ] Referral in [PROVIDER] note of [DATE], p. [X]: [SPECIALIST] [ ] "Outside imaging" mentioned in [PROVIDER] note, p. [X]: [CENTER] [ ] Pharmacy on medication list, p. [X]: [PHARMACY] C. Records that are not visit notes [ ] Imaging: the images (DICOM) as well as the reports, for [STUDIES] [ ] Pharmacy refill history for [REGION] medications [ ] Health plan claims history for the window [ ] Intake questionnaires and pain diagrams (chiro, PT, pain mgmt) [ ] Prior workers' comp or disability claim files [ ] Employer attendance and duty-status records [ ] Prior accident claim files for [DATES] D. Close-out [ ] Every provider in A and B requested or ruled out, with reason [ ] "No records" certifications on file for providers with none [ ] Window and search method written into the review memo
Building the before/after record review
The deliverable that decides these cases is a comparison, not a summary. A narrative of each visit in order is a medical chronology, and you need that first. The before/after review sits on top of it: the same measures, for the same body part, on both sides of the incident, each with a page cite.
- Build the full chronology first. Every provider, every date, in order, each entry cited to its page. You cannot compare what you have not put in order.
- Tag each entry by region, side and level. "Lumbar, L4-5, left" beats "back". Mixed-region notes get more than 1 tag.
- Pick measures recorded on both sides. Pain scores, radicular symptoms, exam findings (straight leg raise, reflexes, strength), range of motion, imaging findings by level, medications and doses, visit frequency, work status, restrictions.
- Fill the comparison table. Last known pre-incident value, first post-incident value, current value. Leave a cell empty rather than guess, and mark it "not recorded".
- Mark what is new. New symptom types, new findings at the matching level, a new level of treatment, a first restriction. These are the aggravation facts.
- List what hurts your side. A fair review names the ED denial, the gap, the matching complaints. The other side will find them anyway.
- Cite every cell. An expert, a mediator or a jury will not accept a baseline built from memory.
2. Baseline vs post-injury comparison table
1 table per body part. Paste into a spreadsheet or a word processor table. Every cell gets a page cite or "not recorded".
BASELINE VS POST-INJURY COMPARISON Body part: [REGION, LEVEL, SIDE] Incident: [DATE] Reviewer: [NAME] Look-back window: [START] to [INCIDENT DATE] Measure | Last pre-incident | First post-incident | Current | New? -------------------------|-----------------------|-----------------------|-----------------------|----- Pain score (0-10) | [X] ([SRC] p. [X]) | [X] ([SRC] p. [X]) | [X] ([SRC] p. [X]) | [Y/N] Symptom type | [axial / radicular] | [ ] | [ ] | [ ] Radiation / dermatome | [ ] | [ ] | [ ] | [ ] Neuro exam (SLR, reflex) | [ ] | [ ] | [ ] | [ ] Range of motion | [ ] | [ ] | [ ] | [ ] Imaging finding at level | [study, date, finding]| [ ] | [ ] | [ ] Medications and doses | [ ] | [ ] | [ ] | [ ] Visits per month | [ ] | [ ] | [ ] | [ ] Treatment level | [none/PT/chiro/inj] | [ ] | [ ] | [ ] Work status | [full / light / off] | [ ] | [ ] | [ ] Restrictions | [ ] | [ ] | [ ] | [ ] Activities stopped | [ ] | [ ] | [ ] | [ ] Gaps in care (dates, length, explanation in record with page): [ ] Other events before or after the incident (with page): [ ] Entries that hurt our position (with page): [ ]
The table, filled in for the worked example
| Measure | Last pre-incident | First post-incident | Current | New? |
|---|---|---|---|---|
| Pain score | 1/10 (PT1 p. 18, 2021) | 8/10 (ED p. 4) | 6/10 (ORTHO p. 44) | Higher |
| Symptom type | Axial low back only (CHIRO p. 1) | Back plus left leg below knee (ED p. 3) | Same (ORTHO p. 44) | Yes |
| Imaging at L4-5 | Bulge, no root contact (IMG p. 7) | Extrusion contacting left L5 root (IMG p. 12) | Not re-imaged | Yes |
| Medications | None since March 2021 (RX p. 2) | Gabapentin monthly (RX p. 4) | Same | Yes |
| Treatment level | PT, discharged (PT1 p. 18) | PT plus 2 injections (PM p. 5) | Surgery recommended (ORTHO p. 46) | Yes |
| Work status | Full duty (PT1 p. 18; PCP p. 66) | Off work (ED p. 9) | 20 lb restriction (ORTHO p. 47) | Yes |
Hypothetical values from chapter 7. Notice what the table does not say: it does not say the crash caused the extrusion. It shows what changed, cited, and leaves causation to the physicians and the jury.
Before you rely on the review
0 of 12 checked
Common mistakes
- Trusting the intake history. Patients forget or minimize old complaints. A post-incident history is a claim to check, not a substitute for the prior records.
- Comparing a report with a report. Different radiologists describe the same disc differently. Get the images compared by 1 reader where it counts.
- Writing conclusions into the table. "Aggravated by MVA" is an opinion. The table holds facts and cites; the expert holds opinions.
IME physicians and treating doctors on aggravation
The jury decides causation and damages, but it decides them on medical opinions, and those opinions are only as good as the records the physician saw. In federal court, a party's physical or mental condition can be examined when it is "in controversy", on motion for good cause (FRCP 35(a)); states have their own versions. For an overview of the exam itself, see what is an IME.
What an aggravation opinion needs to address
- The records reviewed, listed by provider and date range. An opinion that never mentions the 2021 MRI gets impeached with it.
- The pre-incident baseline, with cites. Symptoms, findings, treatment and function during the look-back window.
- What changed. Each new symptom, finding and treatment level, by page.
- Alternative explanations. Natural progression, later events, gaps in care, and why each is or is not the better explanation.
- Apportionment, if asked, with its basis. A percentage without a reason is the easiest part of any report to attack. Tie any split to quantified before and after measures.
- The standard used. Many jurisdictions require opinions to a reasonable degree of medical probability. Use the words your jurisdiction requires.
The IME failure that shows up most in cross-examination is asymmetry: quoting every prior complaint and skipping the discharge note. A report that addresses the strongest record for the other side survives deposition. For how evaluators get judged on their record work, see the IME physician field manual.
3. Questions for a treating physician or IME on aggravation
For a records-review letter, a treating physician conference or a deposition outline. Attach the comparison table from template 2 so the physician answers against cited pages.
AGGRAVATION QUESTIONS: [CLAIMANT], [BODY PART], incident [DATE]
Records provided: [PROVIDERS AND DATE RANGES]; comparison table attached.
Baseline
1. Before [DATE], what was the condition of [BODY PART]? Cite the records.
2. Was the condition symptomatic, treated or limiting work in the
[X] months before the incident? Cite the last pre-incident entry.
3. Were the prior imaging findings at [LEVEL] incidental, or did they
correlate with symptoms at the time? Basis?
Change
4. Which symptoms, findings or treatments after [DATE] had not appeared
before? Cite each.
5. Compare [PRIOR STUDY, DATE] with [NEW STUDY, DATE] at [LEVEL, SIDE].
Which findings are new, which unchanged? Did you review the images?
6. Is the mechanism of the incident consistent with the new findings?
Alternatives
7. Would the condition likely have progressed to its current state
without the incident? If so, when, and on what basis?
8. Do any events after [DATE] (gaps, falls, other injuries) explain
the current state better? Cite them.
Apportionment (only if your jurisdiction and the question call for it)
9. Can the current condition be divided between the pre-existing
condition and the incident on a reasonable basis? What measures
support the division?
10. To what degree of medical certainty or probability do you hold
each opinion above?
AI medical record review for prior-history cases: where it helps and where it fails
Pre-existing condition work is mostly finding and sorting: pulling every entry for 1 body part out of 1,000 to 5,000 pages from a dozen providers, putting them in date order and lining them up against the incident. That is the part of the job AI does well. Deciding what the entries mean for causation and damages is the part it should never be trusted with.
What a large language model does well here
- Finding every mention of a region. Clinical NLP catches "LBP", "lumbago", "lumbar strain" and "back pain" in the history section of a visit for a sore throat, which is where prior complaints hide.
- Ordering thousands of pages by date. An AI medical chronology puts 2019 chiropractic, 2021 imaging and 2024 ED records in 1 sequence with page citations, which is the precondition for any before/after table.
- Making gaps visible. Once every visit is dated, a 37-month quiet period or a 9-week post-incident gap is obvious.
- Extracting prior history. Prior accidents, surgeries and conditions from intake forms and histories, listed with cites.
Where it fails
- OCR on faxed and scanned records. Optical character recognition can misread "L4-5" as "L4-S" or a date by a year. In this work a date error moves an entry across the incident line.
- Handwritten notes. Chiropractic travel cards and handwritten intake forms are often the best baseline evidence and the hardest pages to read.
- Copy-forward text. A model reads "history of low back pain" in 40 notes and may report 40 complaints. A person has to trace it to the first entry.
- Side and level. Left and right, C5-6 and C6-7: small tokens, big differences. Every region tag needs checking.
- Hallucination. Ask a general chatbot "was this condition pre-existing?" and generative AI can write a confident answer that no page supports. Apportionment and causation are expert and jury questions, not model outputs.
Why page-level citations are the whole point
A baseline is only worth what its cites are worth. The defense will test every pre-injury line against the old chart, so each one needs grounded, page-level citations the adjuster can open. A human-in-the-loop review then decides which prior complaints were the same condition and which were not. Courts have made the stakes plain: in Mata v. Avianca, Inc. (S.D.N.Y. 2023), lawyers were sanctioned under Rule 11 for filing a brief with case citations a chatbot had fabricated. The same discipline applies to an AI demand letter or an IME report: open the page before the sentence goes out.
Choosing legal AI tools for this work
The market for AI document review, technology assisted review and agentic AI for law firms is crowded. For prior-history work, check these before any records leave your office:
| Requirement | Why it counts in pre-existing condition cases |
|---|---|
| A citation on every line | Every baseline entry must open to its source page, or it cannot go in a demand or an expert letter |
| HIPAA compliant AI with a signed BAA | Prior records are protected health information; no business associate agreement, no upload |
| SOC 2 | Independent evidence of security controls over years of a claimant's history |
| No training on your data | A claimant's prior records should not improve someone else's model |
| Handles scans, faxes and handwriting | The best baseline evidence is often the worst-quality paper |
| An audit trail of AI use | A record of who uploaded, viewed and edited the output, for defensibility |
| Leaves decisions to people | A tool that scores causation or apportionment is answering a question it cannot answer |
For more on the privacy side, see HIPAA compliant AI medical record review; for accuracy in litigation, see is AI accurate enough for court.
Where Medrecords AI fits in a pre-existing condition review
Medrecords AI is medical record review software. You upload the records you have; it drafts a cited chronology across every provider and date, so prior-care entries for the same body part land in order, each linked to its source page. It does not retrieve records, decide causation, apportion damages or score a case. Flags are signals for a person to check, not verdicts.
- Cited medical chronology. Every entry, pre-incident and post-incident, in 1 dated sequence with a link to its page. This is step 1 of chapter 9, drafted.
- Patient history and recovery factors. Prior conditions, prior accidents and prior surgeries extracted from histories and intake forms, each cited. Extraction only.
- Missing records identification. Visits, providers and date ranges that the records imply but the production lacks, such as a referral with no specialist records, each flag cited to the evidence behind it. Treatment gaps show up in the chronology for you to judge.
- Causation and apportionment graph. The chain from incident to today with prior conditions and competing causes laid out and cited, as a map for your expert. The opinion stays with the expert.
- Questions answered from the record and handwritten record extraction. Ask when a medication was last filled; read the chiropractic travel card.
You review, you revise, you sign. The comparison table, the demand and the opinion are yours.
See a prior-history file put in order, every entry cited.
Book a demo on a claim with years of prior care, then run your first case free on us. Every line comes back cited to its source page. You review, you revise, you sign.
Scheduling only. No records move from a public page.
Frequently asked questions
- What is the eggshell skull rule in simple terms?
- A defendant who injures someone cannot reduce damages by arguing that a healthier person would have been hurt less. If a hidden weakness made the harm worse, the defendant answers for the harm that actually followed. Restatement (Second) of Torts section 461 and California's CACI 3928 state the rule. It does not make the defendant pay for the condition itself.
- What is the difference between the eggshell plaintiff rule and aggravation of a pre-existing condition?
- The eggshell rule is about extent: fragility does not reduce damages. Aggravation is about scope: the defendant pays for the worsening, not for the condition that existed before. California splits them into CACI 3928 and CACI 3927. Most contested files are aggravation cases, decided by comparing prior and post-incident records.
- Can you sue for aggravation of a pre-existing condition?
- Generally yes. Most states allow damages for the worsening of a pre-existing condition, measured from the pre-incident baseline. The defendant is usually not responsible for the condition itself. Rules on proving the split differ by state, so check your jurisdiction.
- Who has to prove how much harm was pre-existing?
- It depends on the state. In Maine, Lovely v. Allstate (1995) puts the burden of apportionment on the defendant when the injury cannot be divided. Other states require the plaintiff to produce evidence that allows a split. Either way, the side without a documented baseline is at a disadvantage.
- Does degeneration on an MRI defeat a personal injury claim?
- Not by itself. A 2015 systematic review of 3,110 people without back pain found disk degeneration in 37% of 20-year-olds and 96% of 80-year-olds. The question is whether symptoms, findings at the matching level, treatment or function changed after the incident.
- How far back should prior medical records go?
- There is no single rule. Spine and joint claims often look back several years, and courts decide discovery scope case by case. Pick a window that covers every provider likely to hold records for the same body part, follow every referral in the prior notes, and write the window down.
- Can AI find pre-existing conditions in medical records?
- AI can find and cite mentions of prior complaints, imaging, treatment and accidents across thousands of pages, including history lines buried in unrelated visits. It cannot decide whether a condition was pre-existing in the legal sense or how to apportion harm. A person checks each cite and makes those calls.
- Can ChatGPT summarize prior medical records for a personal injury case?
- A general chatbot can summarize text, but it may invent details, it may not cite pages, and uploading records to a consumer tool without a business associate agreement raises HIPAA problems. Use a tool built for medical records with a BAA and page-level citations, and verify every line before it goes into a demand or report.
- Is it HIPAA compliant to upload a claimant's medical records to AI software?
- It can be, if the vendor signs a business associate agreement, secures the data (SOC 2 is common evidence) and does not train models on your files. Consumer chat tools usually do not meet those terms. Check the agreement before you upload.
- Will AI software decide causation or apportionment?
- It should not, and Medrecords AI does not. Software can put the prior and post-incident records side by side with citations and flag gaps as signals. Physicians give causation opinions, and the judge or jury decides.
Sources and method
Rules, instructions, cases and study figures were checked against primary sources or published court opinions in September 2026. Quoted instruction text and the Restatement section 461 text are verbatim; the Lovely quotation is as quoted in the Bouchard opinion. Restatement sections 31 and 433A are paraphrased. Everything in chapters 3, 7 and 9 labeled hypothetical is invented for illustration. Product facts come from this site's product pages. Nothing here is legal or medical advice; rules vary by state.
- Restatement (Second) of Torts sections 433A and 461; Restatement (Third) of Torts: Liability for Physical and Emotional Harm, section 31 (American Law Institute).
- CACI No. 3927, Aggravation of Preexisting Condition or Disability, and CACI No. 3928, Unusually Susceptible Plaintiff, Judicial Council of California Civil Jury Instructions (2026 edition).
- Lovely v. Allstate Insurance Co., 658 A.2d 1091 (Me. 1995), as quoted and applied in Bouchard v. United States, No. 05-187-B-W (D. Me. Aug. 13, 2007): methadone dose comparison, 1/5 apportioned, $1,100,000 judgment.
- Steinhauser v. Hertz Corp., 421 F.2d 1169 (2d Cir. 1970): predisposition precipitated by an accident; discount if the condition would have developed anyway.
- Federal Rule of Evidence 803(4) and 803(6); Federal Rule of Civil Procedure 35(a).
- 45 CFR 164.524, HIPAA right of access: 30 days, 1 extension of up to 30, cost-based fee.
- Brinjikji W, Luetmer PH, Comstock B, et al., "Systematic literature review of imaging features of spinal degeneration in asymptomatic populations", AJNR 36(4):811 to 816, 2015. 33 studies, 3,110 people; Table 2 age-specific estimates.
- Englund M, Guermazi A, Gale D, et al., "Incidental meniscal findings on knee MRI in middle-aged and elderly persons", NEJM 359(11):1108 to 1115, 2008. 991 subjects.
- Sher JS, Uribe JW, Posada A, et al., "Abnormal findings on magnetic resonance images of asymptomatic shoulders", JBJS Am 77(1):10 to 15, 1995. 96 subjects.
- Nakashima H, Yukawa Y, Suda K, et al., "Abnormal findings on magnetic resonance images of the cervical spines in 1211 asymptomatic subjects", Spine 40(6):392 to 398, 2015.
- Mata v. Avianca, Inc., 678 F. Supp. 3d 443 (S.D.N.Y. 2023): Rule 11 sanctions for filing fabricated AI-generated case citations.
Related guides: workers' comp apportionment and medical records, altered medical records, the AI-native personal injury law firm, personal injury plaintiff record review and personal injury defense record review.