Coding the injury: extracting a Workers Comp First Report of Injury

A Workers Compensation claim lives or dies on coding. The same injury described three different ways — “hurt his back,” “lumbar strain lifting,” “L4-L5 sprain” — has to resolve to one body part, one nature, one cause, or the reserve, the OSHA log and the loss triangle all go wrong. Here is an actual First-Report-of-Injury packet — a lifting injury at Cedar Valley Manufacturing — run into one coded claim record with the medical roll-up computed exactly.

The packet

• broker_claim_email.pdf — the producer’s notice: insured, policy, employee, what happened.
• first_report_of_injury.pdf — the FROI: employee, employer, injury detail, treatment, return-to-work.
• medical_bill.pdf — the treating provider’s statement of charges, line by line.

Step 1 — Classification

Step 2 — Extraction, with citations

Step 3 — NCCI coding

The narrative — “sharp lower-back pain while manually lifting a 70-lb carton onto a conveyor” — is coded against the injury glossaries. This is the difference between a claim that reserves and reports correctly and one that doesn’t.

Field	As written	Coded (NCCI-aligned)
body_part	“lower back”	Back - Lumbar
nature_of_injury	“sharp pain / pulled”	Strain / Sprain
injury_cause	“lifting a 70-lb carton”	Overexertion / Lifting
line_of_business	“workers comp”	Workers Compensation - Indemnity

Body part × nature × cause is the reserving key for lost-time claims, the input to the OSHA 300 log, and the axis every WC loss triangle is built on. Coding it at intake — not weeks later in an audit — is where the value is.

Step 4 — The medical bill

Statement of charges — Columbus Orthopedic medical bill p.1

Service date	Procedure	Billed
2026-06-18	Initial evaluation & exam	$1,850
2026-06-20	MRI lumbar spine w/o contrast	$2,400
2026-06-24	Physical therapy (6 visits)	$1,320
2026-07-01	Office follow-up	$260
Total billed		$5,830

Step 5 — Deterministic roll-ups

The claim record

Why it holds up

• Coded at intake — body part, nature and cause resolve to NCCI-aligned values the day the claim opens, so the reserve, the OSHA log and the loss triangle are right from the first touch.
• OSHA & lost-time flags — recordability and lost-time vs medical-only are captured as fields, not inferred later, and a return-to-work date drives the indemnity exposure.
• Exact medical roll-up — total billed is summed over the extracted line items by rule, so it always ties to the bill — no model arithmetic, no truncation on a long statement.
• Cited & versioned — every value traces to a document and page, and the coding rules are pinned to a published version for reproducibility.