All terms

    De-Identification of Health Data

    The HIPAA-defined process of removing identifiers from PHI so the resulting data is no longer subject to the Privacy Rule.

    Reviewed by Christian Espinosa, Founder, Blue Goat CyberLast reviewed September 19, 2026

    Definition

    Under HIPAA, health information is de-identified when it cannot reasonably be used to identify an individual. HHS recognizes two methods: the Safe Harbor method (removal of 18 specified identifiers and no actual knowledge that the residual data could re-identify) and the Expert Determination method (a qualified expert applies statistical/scientific methods and documents that re-identification risk is very small). De-identified data is not PHI and not subject to HIPAA.
    What the regulation says
    HHS OCR's 2012 Guidance Regarding Methods for De-identification of Protected Health Information remains the definitive reference. NIST SP 800-188 (2023) and ISO/IEC 20889 cover broader de-identification techniques. Misuse of the de-identification label is a frequent OCR enforcement target.

    What this means in practice

    MedTech teams that want to use clinical data for AI/ML training, analytics, or research typically need de-identified data. Choosing Safe Harbor is procedurally simpler but data-utility-poor; Expert Determination preserves more analytic value but requires documented expert work and ongoing risk monitoring.

    Examples

    • A hospital wanting to share its patient outcomes data with external researchers for a study on treatment efficacy might de-identify the data first, removing all direct identifiers like names and addresses, and ensuring dates are aggregated.
    • A MedTech company developing an AI-powered diagnostic tool trains its algorithms on a de-identified dataset of medical images and associated clinical notes to avoid using Protected Health Information (PHI).
    • A public health agency uses de-identified patient data to analyze disease trends across different demographics without compromising individual patient privacy.
    Common pitfalls
    • •Calling pseudonymized data 'de-identified' - pseudonymization preserves a re-identification key.
    • •Combining de-identified datasets that, together, re-identify (the mosaic effect).
    • •Skipping the Expert Determination documentation - without it, the data isn't legally de-identified.

    Frequently asked questions

    Safe Harbor data is de-identified by HIPAA definition but may still be re-identifiable in adversarial settings. Expert Determination explicitly bounds residual risk. Neither is risk-free.
    Grouped by theme
    Cited by

    Where this term appears across MedTech Terms.

    Latest in MedTech

    Sources

    3 sources

    Sources support the definition or relevant regulatory context, not necessarily every claim on this page. Each is graded against our source-tier hierarchy so you can see what rests on binding law versus commentary.

    Tier 2Regulator guidance and consensus· 2Tier 4Trade press and expert commentary· 1
    Link health: 3 verified· last checked 2026-06-20
    HHS OCR·1NIST·1HSCC·1
    1. 1
      HHS Guidance: De-Identification of PHI
      Tier 2 Verified
      HHS OCRhhs.gov
    2. 2
      NIST SP 800-188 De-Identifying Government Datasets
      Tier 2 Verified
      NISTcsrc.nist.gov
    3. 3
      HSCC - Health Sector Coordinating Council
      Tier 4 Verified
      HSCChealthsectorcouncil.org

    Inline markers like [1] jump to the matching reference above.