What a Medical Summary Is Actually For

What is a medical summary actually for?

Most personal injury firms treat the medical summary as a clerical step: collect the records, put the visits in order, hand it to the attorney. That framing explains why so many summaries run long and still leave the demand weak. A chronology tells the reader what happened. The adjuster, and later defence counsel, wants to know something narrower: whether the treatment is connected to this accident, whether it was reasonable, and whether anything in the file gives them room to discount it.

If the summary does not answer those questions, the adjuster answers them on their own, from the same records, and with the opposite incentive.

Why do adjusters read medical records with suspicion?

Claims professionals are trained to look for patterns that do not match the injury, and that instinct has data behind it. The Insurance Research Council studied more than 30,000 auto injury claims closed in 2012 where the most serious injury was a back or neck sprain. The share of personal injury protection claimants receiving at least one MRI ranged from around half in New York and New Jersey to under 10% in Kansas, Massachusetts and Utah, for the same category of injury (IRC, Interstate Differences in Medical Utilization in Auto Injury Claims). The study is over a decade old, but the lesson for plaintiff firms has not changed: the industry measures treatment against what is typical, and anything outside that pattern gets questioned.

So the summary is not a neutral record. It is the first argument about reasonableness, and the firm either makes that argument deliberately or leaves it to the other side.

What six questions should a medical summary answer?

The useful summaries are organised around the questions an adjuster will ask, not around the provider list. Each one gets its own section, with page references back to the source records:

1. Chronology: When did treatment start relative to the accident, and is the timeline continuous?
2. Providers: Who treated the client, in what sequence, and was each referral documented?
3. Treatment gaps: Where are the breaks in care, and what explains each one (work, childcare, insurance, a scheduling issue the records actually show)?
4. Prior injuries: What does the history show about the same body parts before the accident, and how does the current presentation differ?
5. Causation notes: Where does a treating provider connect the complaints to this accident in their own words?
6. Billing totals: What was billed, by provider, and does it reconcile with the treatment described?

A gap the firm explains in the summary is a fact. A gap the adjuster finds on their own becomes a reason to cut the offer.

What does this look like in a real file?

A pattern I saw again and again in PI files: a rear-end collision, neck and lower back complaints, several months of chiropractic care and physical therapy, then an MRI. The summary that came back was accurate and long, with every visit in date order. Three things were missing. The primary care records from before the accident had never been requested, so nobody read them until the adjuster did and found an earlier visit for neck stiffness. A break of several weeks in therapy sat unexplained, even though the therapist’s own notes recorded why the visits had stopped. And the billing ledgers had never been reconciled with the treatment notes, so the totals in the demand did not match the provider bills. Each of those became a reason to discount the claim.

Rebuilt around the six questions, the same records told a different story. The prior complaint got its own section, showing a single earlier visit with no imaging or referral, set against the new symptoms documented from the first post-accident visit. The gap was explained from the provider’s own notes, with page references. Billing was reconciled provider by provider. Nothing new was added to the file; the summary simply answered the adjuster’s questions instead of leaving them open.

In my experience the bottleneck is rarely summarising speed. It sits upstream: records requested without a prior-history request, files sent for summary before the record set is complete, and nobody owning the reconciliation of bills to treatment.

Where does AI fit in medical summaries, and where does it not?

Medical summarisation is one of the most common uses PI firms describe for generative AI, and expectations are rising. In 8am’s 2025 survey of more than 325 legal professionals, 37% of personal injury respondents said they personally use generative AI at work, and 19% of PI firms expect AI to replace outsourced work, against 12% across the wider legal market (8am, 2025 Legal Industry Report: Personal Injury Insights).

Speed is not the issue. A tool can extract dates and providers from a thousand pages quickly. The harder part is judgement: deciding which prior complaint matters, which gap needs an explanation, and which provider note carries the causation argument. Responsibility does not move with the work either. ABA Formal Opinion 512 says that uncritical reliance on AI output without an appropriate degree of independent verification can breach the duty of competence, and that the lawyer remains fully responsible for the work (ABA Formal Opinion 512). The same logic already applied to outsourced summaries under Formal Opinion 08-451, which requires lawyers to make reasonable efforts to supervise outside providers (ABA Formal Opinion 08-451). For how this plays out across the US, UK, Canada and Australia, see Who Is Responsible When AI Makes a Mistake in Legal Work?

In practice, whether the summary comes from a paralegal, an LPO team or a model, someone at the firm has to check it against the six questions before it reaches the demand.

What should a PI firm change this month?

• Rewrite your summary template so the headings are the six questions, not the provider names.
• Require a page reference for every finding, so the attorney can verify it in seconds.
• Add a one-page reviewer checklist that the attorney signs before the demand goes out, whoever produced the summary.
• Track which sections adjusters push back on, and tighten those first.

The firms that get better offers are usually not the ones with longer summaries. They are the ones whose summaries leave the adjuster fewer questions to ask.

Want to see the format? Email sagar@whobringsthebusiness.com with the subject line SAMPLE and I will send a redacted medical summary built around these six questions, with the reviewer checklist.

Sources

• Insurance Research Council: Interstate Differences in Medical Utilization in Auto Injury Claims (2013)
• 8am: 2025 Legal Industry Report, Personal Injury Insights
• ABA Formal Opinion 512: Generative Artificial Intelligence Tools (2024)
• ABA Formal Opinion 08-451: Outsourcing Legal and Nonlegal Support Services (summary)