The Reserve Accuracy Problem Speed-Only Medical Record Review Can't Fix

A fast chronology and an accurate one are always different. For TPAs, the difference shows up on the reserve line long before it ever shows up on an invoice.

Most TPA vendor selection for medical record review runs on two questions: how fast, and how much. Both matter. Neither tells you whether the file you get back holds up once a claim moves past intake.

That gap is where reserves quietly drift. A chronology can be fast, clean, and still miss the one detail that would have changed how a claim was reserved from day one. In reviewing medical record files for TPAs across multiple jurisdictions, we’ve seen this pattern play out repeatedly: a reserve gets set on a clean chronology, then the claimant’s actual medical complexity surfaces during UR or IME, forcing a correction that should never have been necessary.

Key Statistic

According to NCCI’s Workers Compensation Statistical Plan data (cited as of 2026), motor-vehicle-related crashes are the costliest lost-time workers’ comp claims by cause of injury, averaging $91,433 per claim for accident years 2022-2023, well above the $47,316 average across all claim types. Burns ($64,973) and falls or slips ($54,499) follow. On files that size, a missed comorbidity or an unflagged treatment gap is not a rounding error. It is a reserve set on incomplete information.

Why Fast Chronology Isn't Accurate Chronology

A fast chronology tells you what happened. An accurate one tells you what it means for the reserve. The two look identical on a turnaround report and behave nothing alike once a claim is challenged.

Bill review and medical record review vendor comparisons tend to default to a spreadsheet: turnaround time, price per page, and page volume capacity. All measurable, all easy to rank. None of them answers the question that actually determines claim outcome: did the reviewer catch what the record was hiding?

Industry commentary on vendor selection has made this point about bill review for years, and it applies just as directly to record review and chronology work. A quantitative comparison tells you which vendor is cheapest and fastest. It says nothing about which vendor’s output you can rely on when a claim gets contested.

Where Does Speed-Only Review Create Reserve Risk?

Speed-only review fails in three predictable places: comorbidities buried in unrelated specialties, treatment gaps smoothed into continuous care, and causation contradictions resolved into a single clean narrative. Each one is the normal condition of a complex claim file, and each one moves the reserve when caught and hides it when missed.

Every body part named in the initial injury report survives into the current file

A pre-existing condition documented in a cardiology note eight years before the claim doesn’t announce itself. It has to be found, cross-referenced, and tied back to the current injury.

Take a shoulder injury claim where the file includes a decade-old endocrinology note mentioning early-stage diabetic neuropathy. That note has nothing to do with the shoulder on its face. But it matters directly to healing time, treatment duration, and whether ongoing complaints are fully attributable to the work injury. A tool optimized to summarize quickly flattens that note into a single line in a chronology instead of flagging it as a fact that should move the reserve. A reviewer trained on causation reads it differently, because they know what question it’s actually answering.

Treatment Gaps That Change the Story

A claimant who treats steadily for six weeks, disappears for four months, then resumes care right before a UR request or IME is telling you something.

A chronology built to read smoothly presents this as continuous care rather than surfacing the gap as a discontinuity worth reserving around. The reserve set on a “continuous treatment” read and the reserve set on a “treatment gap, resumed under scrutiny” read are not the same number, and they shouldn’t be.

Causation Language That Gets Smoothed Over

Medical records are rarely clean on causation. Notes hedge, specialists disagree, and imaging reports sometimes contradict the narrative in the claim file.

An MRI report reading “degenerative changes consistent with age” sitting next to a demand letter describing a traumatic herniation is a genuine conflict that deserves scrutiny, not a quiet footnote. A review built to produce a coherent output has an incentive to resolve that tension into a single clean narrative. A review built to hold it in tension, the way an adjuster needs to, flags the contradiction instead of erasing it.

None of these are edge cases. They are the normal condition of most complex claim files, and they are exactly what a review focused on speed alone is least equipped to catch.

Why Does Multi-State Volume Compound the Problem?

For TPAs operating across jurisdictions, this problem compounds because the same clinical finding carries different reserve consequences depending on which state’s rules apply.

A comorbidity or apportionment issue that clearly reduces a reserve in one state may sit on entirely different legal footing in another. AMA Guides editions alone vary by state: some jurisdictions still apply older editions while others have moved to newer ones, and the impairment percentage attached to the same clinical finding shifts depending on which edition governs. A review vendor working on national claim volume needs reviewers who understand which framework applies to which file, because a single template applied everywhere will miss the variance that actually changes the reserve. A quantitative-only vendor comparison rarely tests for this at all, because turnaround time and cost per page look identical whether the reviewer understands jurisdictional variance or not.

	Quantitative-only comparison	Quality-inclusive comparison
What gets measured	Turnaround time, cost per file, page volume	Turnaround, cost, and reviewer accuracy on flagged issues
What it misses	Whether the output is defensible under challenge	Nothing structural, but requires a harder RFP question set
Where it fails	High-complexity files: apportionment, comorbidities, disputed causation, multi-state claims	Rarely fails on complexity; the review is built for it
Downstream cost	Reserve corrections, re-review, and delayed UR and IME scheduling	Fewer corrections; reserves set closer to the accurate first time
What it rewards	Whichever vendor is cheapest and fastest on paper	Reviewers who catch what changes the outcome

What Does Accuracy-First Medical Record Review Look Like?

Accuracy-first review pairs AI for volume with a certified reviewer for judgment, and flags gaps and contradictions instead of resolving them into a clean story. For TPAs managing intake-to-output pipelines across large claim volumes, the practical version looks like this:

Deduplication and clean intake, so reviewers aren’t working from redundant or fragmented files.
AI-assisted extraction and organization, so volume doesn’t create a backlog.
A certified medical reviewer validating the output while there’s still time to act on it, before a claim escalates, rather than after.
Explicit flagging of gaps, contradictions, and comorbidities, rather than a chronology that resolves them into a clean narrative.
Reviewers who know which state’s rules apply to which file, especially on apportionment and impairment rating.
Consistent output format across jurisdictions, so adjusters aren’t reformatting every file that comes in.

This is not a case against automation. Extraction, sorting, and organizing thousands of pages is exactly where AI earns its place in the workflow. The judgment calls, the parts that determine whether a reserve is right, are where a trained reviewer still has to be the one reading.

summary accuracy across AI-assisted, human-validated review

99.98%

average reduction in turnaround time versus manual-only review

80%

Years of experience

25+

certified for information security

ISO 27001

Frequently Asked Questions

How is medical record review different from bill review?

Bill review checks charges against fee schedules and coding accuracy. Medical record review and chronology work assess the clinical narrative itself: what happened, in what order, and whether the story the records tell is complete. TPAs typically need both, but they are different disciplines with different failure points.

What happens when a reviewed chronology still gets challenged?

A chronology that flagged the gaps, comorbidities, and contradictions up front holds up under challenge in a way a clean, unflagged one does not. The claim can still be contested. What changes is whether the reserve was already set on the full picture, or whether the challenge exposes something the review should have caught in the first place.

Does adding a human review layer slow down turnaround?

Not meaningfully, when the workflow is built around it rather than bolted on afterward. AI-assisted extraction handles the volume; the human reviewer validates and flags, rather than re-reading every page from scratch.

Why does jurisdiction matter in medical record review?

The same clinical finding can produce a different impairment rating and a different reserve depending on which state’s rules and which AMA Guides edition apply. A reviewer who understands jurisdictional variance catches apportionment issues that a single national template misses.

How does AI fit into an accuracy-first review?

AI handles what it does best: deduplication, extraction, and organizing thousands of pages at speed. The clinical judgment, catching a buried comorbidity or an unflagged treatment gap, stays with a certified reviewer. The two together deliver volume without sacrificing the accuracy the reserve depends on.

The Bottom Line

Speed-only medical record review saves time but costs accuracy, and the cost lands on the reserve line first. Accuracy-first review requires a different process, one that treats every complex file as a potential reserve error waiting to be caught rather than a box to be checked. When TPAs ask vendors only about turnaround time and cost per page, they are selecting for volume. On medical record review, the reserve line rewards them for selecting for accuracy.