Insurance audit dataset de-identification is the removal of the 18 HIPAA Safe Harbor IDs (45 CFR §164.514(b)) from a sampled pull. anonym.plus runs on your device. The reviewed fields stay clear, but the rows no longer name anyone.
When this applies
A pull samples many records to test billing accuracy. Sending that to a cloud tool is a disclosure risk. Local work clears the IDs and keeps the sample.
How anonym.plus handles it
- Point anonym.plus at the export on your server.
- It scans the ID columns and the free-text fields.
- Steady aliases keep links across sampled rows.
- Keep the sampled fields and the error flags.
- Replace each ID with a steady alias, or redact it.
- Save the clean pull on your device.
What you need to provide
- The pull as CSV, JSON, or a bundle.
- A column map for known ID fields.
- An operator: Replace, Redact, or Mask.
PHI entity types detected
| Category | anonym.plus entity type | Example |
|---|---|---|
| Names | PERSON | patient_name → [PATIENT_n] |
| Member ID | US_HEALTH_PLAN_BENEFICIARY | mbr_id → [MEMBER_ID_n] |
| Account | ACCOUNT_NUMBER | acct field → [ACCOUNT_n] |
| Dates | DATE_TIME | service_date → shifted [DATE] |
| Record IDs | MEDICAL_RECORD_NUMBER | mrn field → [MRN_n] |
| Free text | PERSON / LOCATION | inline names → aliases |
Compliance achieved
- Strips all 18 ID classes for HIPAA Safe Harbor (45 CFR §164.514(b)).
- Catches member and account numbers as identifiers.
- Keeps the sampled fields and error flags for review.
- Fully offline — no BAA for the tool.
Anonymize audit datasets offline — see plans & start free →
Limitations & cautions
These pulls mix tidy columns with messy free text. Column rules handle the first well. Free-text fields need the same review as any note. Test a sample before a full run, and check that date-shifting keeps the gaps your review needs.
Frequently asked questions
Can rows stay linkable after the swap?
Yes. A steady alias map swaps each ID the same way, so sampled rows for one person still join while no real identity is left.
Do the sampled fields survive the swap?
Yes. The reviewed fields and error flags stay. Only the IDs change, so the pull is still useful for review.
Why work locally rather than in the cloud?
Sending raw rows to a cloud tool is itself a disclosure. Local work skips that risk.