GUDID Data-Quality Checklist: What Distributors Should Verify Before a Contract

A hospital value-analysis team does not usually verify one device identifier at a time. It inherits a contract — hundreds or thousands of SKUs across dozens of labelers — and has to answer a portfolio-level question: is the device data we will load into our ERP, EHR, and inventory systems complete and reliable enough to trust? The FDA's Global Unique Device Identification Database (GUDID) is the source of truth for that data, but GUDID is only as good as what each labeler submits, and labelers submit unevenly.

This article is a data-quality checklist for distributors, GPOs, and hospital value-analysis teams operating at contract scale. It is distinct from our single-device UDI lookup workflow, which covers how to verify one identifier before a purchase. Here the question is broader: across a whole contract, where are the GUDID gaps that will break downstream systems, frustrate recall matching, and create patient-record noise?

How GUDID Data Is Structured (and Where It Fragments)

Every device that requires a UDI has a Device Identifier (DI) record in GUDID, submitted by the labeler (usually the manufacturer, but also repackagers, relabelers, and specification developers). Each DI record carries the fixed identification elements — labeler name and DUNS, brand name, version or model number, catalog number, product code(s), device description, and lifecycle status — and the UDI production identifier flags (lot, serial, expiration date, manufacture date). Required data elements are defined under 21 CFR 830.310, but several fields that are operationally critical are optional, and that is where quality fragments.

Labelers submit via the GUDID web interface, HL7 SPL files, or a third-party submitter, and the data becomes public through the AccessGUDID portal. FDA can remove fraudulent data, but day-to-day completeness and accuracy are the labeler's responsibility.

What the Data Actually Looks Like

To size the data-quality problem, we analyzed the complete FDA GUDID public device record extract. The headline: the database is enormous, but it is patchy in predictable places.

Across 5,083,948 device records covering 5,061,910 unique device identifiers (DIs) from 11,779 distinct labelers:

Two findings dominate. First, more than one in five GUDID device records has no device description — and device description is the field that lets clinicians and systems distinguish between similar devices. Second, labeler activity is heavily concentrated: the top 20 labelers account for 27.7% of all device records, with Cardinal Health alone holding 370,347 records (about 7.3%) and Medline Industries 251,266 (about 4.9%). Concentration cuts both ways — a handful of large labelers are easy to validate in bulk, but a long tail of small labelers drives most of the data-quality risk.

Data source: FDA AccessGUDID public device record database; analysis by MedDeviceGuide of the public GUDID extract dated 2026-06-10, covering 5,083,948 device records and 11,779 labelers.

The GUDID Data-Quality Checklist for a Contract Portfolio

Run these eight checks across every device the contract introduces. The first four are field-completeness checks; the last four are integrity checks that catch the subtle failures.

1. Device description is populated and specific

Device description is optional in GUDID, but it is the single most important field for identifying a device in a clinical record and distinguishing it from look-alikes. With 20.3% of records missing it entirely, this is the first thing to flag. A description that reads only a generic category name (or is blank) should be sent back to the labeler for completion before the contract loads the item.

2. Labeler identity is consistent and DUNS-linked

Every DI record should resolve to a single labeler DUNS number and a consistent company name. Check for labelers that appear under multiple name variants or that changed their company name without updating historical DIs — a common cause of "phantom" duplicate records in a portfolio. The data shows 5,061,910 unique DIs against 11,779 labelers (an average of about 430 DIs per labeler), so even small naming inconsistencies can produce large deduplication problems downstream.

3. Version/model and catalog numbers are present and matchable

These are essentially complete in GUDID (brand name and version/model were effectively 100% populated in our analysis), so a missing value here is a red flag rather than the norm. When a version or model number is absent, the labeler's catalog number should be populated in both the catalog and version/model fields per the recognized best practice. One structural caveat that catches value-analysis teams: catalog number is not a required GUDID data element, even though providers rely on it to match GUDID records against their own item master, ERP, and charge master. Contracts that depend on catalog-number matching should confirm catalog numbers are populated for every DI before relying on automated reconciliation.

4. Product-code linkage is clean

Each DI should map to a valid FDA product code. About 0.6% of records carry no product code, and 11.4% map to more than one. Multi-code records are legitimate for some products, but they complicate recall matching and adverse-event rollups: when a recall targets a product code, a multi-code DI may or may not be captured. Flag multi-code records for manual review during contract onboarding.

5. Packaging hierarchy and unit-of-use integrity

GUDID models a packaging hierarchy (unit of use, base package, higher-level packages), and each configuration gets its own DI. Verify that the contract's purchase units resolve to the correct DI level — a common error is loading a base-package DI where a case DI belongs, which breaks quantity math in inventory systems. This is one of the most frequent submission errors FDA and industry best-practice guides call out.

6. Sterile and single-use flags are verified, not assumed

Sterilization status and single-use flags drive reprocessing decisions and patient-safety controls — but they are inconsistently populated. In our analysis, single-use flags were set on the records that carried them (62% single-use), but the sterile-status attribute was not reliably populated across the extract. The practical rule for value-analysis teams: do not rely on GUDID's sterile flag for a safety-critical decision. Confirm sterilization state directly with the labeler and on the label.

7. MRI safety information is present where relevant

MRI safety status is a separate GUDID field that is frequently left blank, even though it is a recognized common submission error. For any contract that includes devices that may be present during MRI (implants, markers, accessories), confirm the MRI-safety field is populated or obtain the information from the labeler's MRI-conditional labeling.

8. Lifecycle status is current

GUDID records carry a publish and discontinuation lifecycle. Contracts sometimes inherit DIs that the labeler has marked discontinued or superseded. Confirm that every DI in the contract is in an active/published state and that no listed device has been quietly replaced by a new DI with a different sterilization state, model number, or configuration.

Common GUDID Submission Errors That Drive These Gaps

Most portfolio-level data-quality problems trace back to a handful of labeler submission errors:

• Incorrect DI formatting against the issuing agency (GS1, HIBCC, ICCBBA) specification
• Missing or vague device descriptions, especially on older or private-label records
• Incorrect packaging hierarchy — unit-of-use, base package, and higher-level DIs assigned incorrectly
• Inconsistent descriptions across a product line, so sibling devices are hard to distinguish
• Failure to update discontinued devices, leaving stale DIs in circulation
• Missing MRI-safety and sterilization fields that downstream teams assume are reliable

Running the Check at Portfolio Scale

For a single device, the AccessGUDID web lookup is sufficient. At contract scale, value-analysis teams typically work from a GUDID data extract (available through NLM and openFDA) and automate the eight checks above as a pre-contract gate: compute description completeness and product-code cleanliness per labeler, flag multi-code DIs and missing critical fields, and require labeler remediation before the items go live in the ERP. The data suggests where to focus: prioritize the long tail of smaller labelers (where completeness gaps cluster) and audit any contract where more than 10–15% of records lack a device description.

When gaps are found, the fix is not internal — only the labeler can correct a DI record. Route each flagged record back to the labeler for update, and, for data that appears incorrect rather than simply missing, submit a ticket to the FDA UDI help desk so the record can be reviewed. Treating data quality as a contract precondition (items do not go live until their GUDID records pass the checks) is the single most effective lever for moving the long tail toward completeness.

Bottom Line

GUDID is the authoritative source for device identification data, but at portfolio scale its quality is uneven in predictable ways — one in five records lacks a device description, the sterile flag cannot be trusted for safety-critical decisions, and a long tail of smaller labelers drives most of the risk. A disciplined distributor or value-analysis team treats GUDID data as something to be checked, not assumed: verify description completeness, labeler identity, product-code linkage, packaging hierarchy, and lifecycle status across the contract before a single item loads. That front-loaded check is what separates a clean launch from months of recall-matching failures and patient-record noise.

Sources

• FDA / NLM, AccessGUDID public device record database — https://accessgudid.nlm.nih.gov/
• FDA, "Global Unique Device Identification Database (GUDID)" guidance — required data elements under 21 CFR 830.310
• AHRMM, "Global UDI Database (GUDID) Best Practices Guide" — critical data elements frequently missing or inaccurate, including device description, version/model number, and MRI safety
• RegDesk, "GUDID Compliance: How to Submit and Maintain UDI Data" — common GUDID submission errors and labeler responsibilities
• The FDA Group, "Guidance Breakdown: The FDA Updates its GUDID Final Guidance" — DI record structure and account roles
• eCFR, 21 CFR Part 830 — Unique Device Identification — https://www.ecfr.gov/current/title-21/chapter-I/subchapter-H/part-830