# Cancer Registry Abstraction Services

> Registry abstraction for CoC cancer programs, ACS-verified trauma centers and NCDR/STS cardiac programs. Every field drafted to the manual and page-cited.

Canonical page: https://medrecords.ai/solutions/clinical-registry-abstraction/

---
[Home](https://medrecords.ai/) ›[Solutions](https://medrecords.ai/solutions/) ›[Case Types](https://medrecords.ai/solutions/case-types/) ›Clinical Registry Abstraction
CANCER REGISTRY ABSTRACTION SERVICES

## The abstract, drafted to the manual — oncology, trauma, cardiac.

Cancer registry abstraction services draft the registry abstract to the manual that governs it: STORE and NAACCR v25 with AJCC 8th staging for the tumor registry, NTDS and AIS 2015 for trauma, NCDR and STS dictionaries for cardiac. One engine, three rulebooks, every field cited to the page it came from.

[Test a file ](https://medrecords.ai/test-a-file/?src=clinical-registry-abstraction)
 [Book a demo](https://medrecords.ai/demo/?src=clinical-registry-abstraction)
Click any field → the source page it cites
Abstract draft · Case #IME-4812
 for registrar review
Adams, Timothy
 · registry source file
342 pp / 2 packets received
 logged
11 duplicate pages removed
 free
p.140 — wrong patient
 quarantined
Pages 342
 Fields 38
 Cited 100%

### The manual is published. So is the grader.

We do not build for a case type until we can name the standard that defines a correct output and the person who grades the file against it. Registry abstraction is the cleanest example on the list: the field definitions are in print, and the audit is on a calendar.

The rulebook and the referee
The rulebook
- STORE, the NCDB's Standards for Oncology Registry Entry, plus the SEER Program Coding and Staging Manual
- NAACCR Data Standards v25 for the field set and the record layout
- AJCC Cancer Staging Manual, 8th edition, for the staging elements
- Solid Tumor Rules and the Heme & Lymphoid Coding Manual for multiple primaries and histology
- CoC Optimal Resources Standards 6.1–6.4 for what the registry itself has to deliver
- Alongside them: NTDS with AIS 2015 for trauma, and the NCDR and STS data dictionaries for cardiac and stroke

The referee
- The CoC surveyor, on a site visit every three years, with live case review
- The NCDB Call for Data, whose hard edit rejects run against 250+ NAACCR edits
- RQRS and CP3R concordance reporting on quality measures
- State central registry re-abstraction QC studies
- NAACCR Gold and Silver certification, graded on completeness and accuracy

Outsourced abstraction runs $25–45 per analytic cancer case, $20–35 per trauma record, and $18–40 per cardiac case

 Flat 10¢/page here, duplicates free
Abstract draft · Case #IME-4812
 cited 100%
Production342 pp / 2 packets · 11 deduplicatedlogged
Primary site · histologyCited to the pathology reportcited
Staging elementsDrafted from the documented findingsregistrar reviews
First course of treatmentDates cited to op notes and visitscited
38 template fields · cited 100% · nothing submitted for you
What the file actually looks like

### One case is 250 pages of somebody else's formatting.

A single abstractable case runs 100 to 400 pages, around 250 in the middle, and it almost never arrives as one clean production. Pathology comes from the lab system, operative notes from the surgical record, imaging and oncology flowsheets from somewhere else again, and the outside-facility packet arrives as a scan of a fax.

That is what makes the work slow rather than hard: degraded scans, handwritten margin notes on the flowsheet, and the same discharge summary appearing in three packets. In the demo case, 11 duplicate pages came out before the read started and page 140 turned out to belong to a different patient.

~2.5M
hospital abstracts a year, against an 18M+ survivor follow-up cohort
~40M
records in the NCDB from 1,500+ programs, capturing 72% of new US diagnoses
What we'd build with you

### Four steps, starting with one real case.

Registry abstraction is a build, not a download. We start from a case you have already abstracted, so you can grade our read against a known answer before anything is configured.

01

#### Send one real file

One case, as it sits in your queue, in whatever formats it arrived in. Bring one you have already completed if you want a control.

02

#### We return the read

The fields drafted to the manual that governs the case, each one cited to the page it was abstracted from, plus the duplicates and any wrong-patient pages called out separately.

03

#### We scope the build

Your field set, your registry's conventions, your review workflow. If the case type is not buildable to your standard, that is the answer we give you.

04

#### You run it

Your registrar reviews, corrects, and enters. We do not retrieve records from providers, key anything into your registry software, or submit to any registry on your behalf.

Audit trail · Case #IME-4812
 exportable
08:12Production received · 342 pp / 2 packetssystem
08:3111 pages deduplicatedsystem
08:44p.140 flagged — wrong patientsystem
08:54Abstract drafted · 38 template fields · cited 100%system
09:20Draft reviewed · citations verifiedregistrar
09:26Draft exported · registry entry made off-platformregistrar
Every access logged · file deleted 30 days after delivery
Why Medrecords AI

### The rules the platform never breaks.

EVERY LINE CITED
CASE #IME-4812 · ADAMS, T.342 pp
2/14 — ER visit, right knee
 p.4
4/18 — arthroscopic surgery
 p.61
p.140 — wrong patient
 quarantined
Medrecords AI
Read every page · cite every line

#### HIPAA, under a signed BAA

Every file is handled under our Business Associate Agreement, from the first byte.

#### Never trains a model

Your records are never used to train any AI model — ours or anyone else's.

#### Every line cited

If we can't cite it, we don't say it. Every sentence links to its source page.

#### Deleted after delivery

Files are deleted 30 days after delivery, with a full audit log of every access.

Powered by the platform

### Four capabilities behind every drafted abstract.

Registry abstraction is one deliverable of the same platform that reads, sorts, and cites the whole record.

[

#### Medical Table Extraction

Pathology grids, oncology flowsheets, and lab panels come out as structured values rather than as a picture of a table.

IN ACTION · scanned flowsheet → field-level values
](https://medrecords.ai/product/medical-table-extraction/)
 [

#### Medical Chronology

Diagnosis date, first course of treatment, and every follow-up contact on one timeline, synced to the source page.

IN ACTION · 342 pp → one dated treatment sequence
](https://medrecords.ai/product/chronology/)
 [

#### Smart Lists

Work a queue instead of a stack: cases grouped by what is still missing, so the registrar's day starts already sorted.

IN ACTION · backlog → a review queue, not a pile
](https://medrecords.ai/product/smart-lists/)
 [

#### Verifiable AI Citations

If we can't cite it, we don't say it. Every drafted field links to the page it was abstracted from, which is what makes an edit failure traceable.

IN ACTION · 38 template fields → cited 100%
](https://medrecords.ai/product/citations/)
FAQ

### Registry abstraction, answered.

No. The platform drafts the abstract and cites every field to the page it came from; a Certified Tumor Registrar or ODS-C reviews it, corrects it, and owns what is entered into the registry. It is the same relationship a contract tumor registrar has with your program, except the first pass arrives already page-cited.

Backlog is volume work: 100 to 400 pages per case, and often hundreds of cases behind. The platform reads the whole production and drafts the fields, so your registrar works down a queue of drafted, cited abstracts instead of reading every chart cold. We never enter or submit anything on your behalf.

Whichever one governs the case. For cancer that is STORE with the SEER Program Coding and Staging Manual, NAACCR Data Standards v25, AJCC Cancer Staging Manual 8th edition, the Solid Tumor Rules and the Heme and Lymphoid Coding Manual. For trauma it is NTDS with AIS 2015. For cardiac and stroke it is the NCDR and STS data dictionaries.

Fields are drafted to the manual and cited to the page, which is what makes an edit failure quick to trace back to the source. We do not run your registry software, submit to the NCDB Call for Data, or clear the edit set for you. Your registrar validates and submits.

No. Compliance is graded by a CoC surveyor at the site visit, against your program's own completeness and timeliness, with live case review. What we deliver is drafted, page-cited abstraction work that your registrar reviews. Whether the program meets Standard 6.1 is the surveyor's determination, not ours.

### Send one case. We'll tell you what we can read.

No obligation. If the case type is buildable we'll scope it; if it isn't, we'll say so.

[Test a file ](https://medrecords.ai/test-a-file/?src=clinical-registry-abstraction)
 [Book a demo](https://medrecords.ai/demo/?src=clinical-registry-abstraction)
[See every case type we process →](https://medrecords.ai/solutions/case-types/)
 [Medicare audit response →](https://medrecords.ai/solutions/medicare-audit-response/)
