top of page

Your Test Lab Is Part of Your Quality System. FDA Keeps Saying So.

Writer: Lexim AI
Lexim AI
Nov 6
7 min read

Most device quality teams audit their own factories and their critical component suppliers. Fewer apply the same scrutiny to the contract laboratory that generated the bench, biocompatibility, or cybersecurity testing sitting inside a 510(k). FDA has now spent nearly three years telling manufacturers that this is a gap, and the warnings have become more specific.

In February 2024, FDA's device center sent a letter to industry describing an increasing number of third-party test labs generating data that is fabricated, duplicated from other device submissions, or otherwise unreliable. In June 2026, FDA published an updated notice on data integrity and linked three general correspondence letters to firms that concluded their testing facility had copied the results of another study or created falsified or otherwise invalid data. Meanwhile, FDA has been removing labs from its voluntary accreditation program. And in October, the center put the ASCA program on its fiscal 2027 list of guidances to finalize.


What FDA says has happened

The June 25, 2026 notice says FDA has seen an increase in unreliable testing data generated by third-party facilities on behalf of device manufacturers and sponsors, and that this has left the agency unable to reach a substantial equivalence determination or otherwise authorize marketing for devices whose submissions rely on that data. The three letters it links state FDA's conclusion that the testing facility copied another study's results or created falsified or invalid data, and that the quality and integrity of the study data the firm generated cannot be ensured.

This follows a pattern FDA has described before. The 2024 letter warned that unreliable lab data calls into question the integrity of the entire submission, not just the affected test, and noted that suspect submissions included data from numerous facilities based in China and India. Reporting on the agency's statements has since added that concerns about lab data have led CDRH to refuse to accept or reject device submissions, that FDA has issued warning letters to some labs for good laboratory practice violations, and that reviewer time has been diverted to evaluating possible data integrity problems. The public notices I reviewed do not say how many submissions or labs are affected.


How ASCA is supposed to work

The Accreditation Scheme for Conformity Assessment is a voluntary program authorized under section 514(d) of the FD&C Act. It works in two steps. A testing laboratory is first assessed by an ASCA-recognized accreditation body against ISO/IEC 17025:2017, the general competence standard for testing laboratories, plus ASCA program specifications for each eligible standard and test method. The lab then applies to FDA for ASCA Accreditation, which exists only within the program and is separate from any other accreditation the lab holds.

The design goal is to reduce documentation, and that is the technically important part. A lab accredited for a test method performs the testing, gives the manufacturer the complete results, and also produces an ASCA Summary Test Report: a standardized report of roughly two pages with a declaration of conformity. The manufacturer includes the Summary Test Report as received in its submission, and FDA reviews the conformity assessment from that summary instead of from the full report, provided the standard and method were within the lab's ASCA scope at the time of testing. The eligible standards include biocompatibility methods, such as the ISO 10993-5 cytotoxicity assays, and basic safety and essential performance testing under the IEC 60601 series.

That design shifts where trust sits. FDA reads less underlying data because it trusts the accreditation. A lab that falsifies or copies results therefore has more leverage inside the program than outside it, and the manufacturer, which holds the complete results, is the party best placed to check them.


ASCA is a screen, not a guarantee

The Accreditation Scheme for Conformity Assessment is a voluntary program under which FDA-recognized accreditation bodies accredit test labs to specific consensus standards, and FDA has encouraged manufacturers to use ASCA-accredited labs. But FDA's own ASCA page now says that, during audits of accredited labs, it found some did not meet the requirements of section 514(d) of the FD&C Act and the related guidance, and in some cases it identified concerns about the integrity of the data those labs submitted. Labs in that situation appear in the ASCA database with an "FDA Initiated Withdrawal" status, and ASCA Summary Test Reports from them are no longer accepted.

In practice, that means accreditation status can change after you award the work, and a lab's status on the day you contract is not its status on the day you submit.


Why this is a quality system problem

FDA's position is consistent: the manufacturer is responsible for scrutinizing data generated by a third party before putting it in a submission. That is a purchasing-controls obligation as much as a regulatory-affairs one. Under the QMSR, which incorporates ISO 13485, supplier evaluation and control records are inside the scope of an FDA inspection, and a testing lab that produces evidence for your design and regulatory files is a supplier whose output you rely on. If a lab's data is later found to be copied or falsified, the question will not only be what happened at the lab. It will be what you did to verify it, and whether your records show that.


Where this lands in ISO 13485 and 21 CFR Part 58

The QMSR incorporates ISO 13485:2016, and three parts of that standard bear directly on a contract lab. Clause 4.1.5 requires an organization that outsources a process affecting product conformity to monitor it and keep control over it. Clause 7.4.1 calls for documented criteria to evaluate, select, monitor, and re-evaluate suppliers, proportionate to the risk of the device. Clause 7.4.3 requires verification that purchased product meets requirements; for a test lab, the purchased item is data, and verification means checking that it is genuine and complete.

Good laboratory practice adds a second layer where it applies. 21 CFR Part 58 governs nonclinical laboratory studies of safety, which includes much biocompatibility testing, and requires elements such as a study director, a quality assurance unit, and retained raw data. It does not cover every bench performance test, so for much of the testing in a typical 510(k) there is no GLP framework to rely on, and the manufacturer's own verification carries more of the weight.


What independent verification can look like

  • Check status twice. Look up the lab and its scope of accreditation in the ASCA database before awarding work and again before you submit, including for any withdrawal status.

  • Go beyond accreditation. Qualify the lab on its own merits: audit it, confirm applicable GLP compliance where it applies, and look at its subcontracting.

  • Contract for the evidence. Require access to raw data, instrument logs, and audit trails, a right to audit, record retention, and prompt notice of any FDA correspondence the lab receives.

  • Review raw data, not only summary reports. Look for the warning signs FDA describes: results duplicated across unrelated tests or customers, implausibly uniform values, and records that do not match instrument or personnel logs.

  • Cross-check critical results. Compare lab output against your own development data, and consider repeating a sample of the highest-risk tests in-house or at a second lab.

  • Document the verification. Put what you checked, who checked it, and what you found in the supplier file, so you can show it.

  • Look back. For submissions already cleared or pending, review which test labs produced the data and whether any have since lost accreditation status.

  • Plan for retesting. A lab withdrawn mid-project can mean repeating studies. Knowing the cost and schedule in advance is part of choosing the lab.


What raw-data review looks for

FDA's description of the problem, data that is fabricated, duplicated from other submissions, or otherwise unreliable, suggests where to look. These are practical examples of reviewer checks, not an FDA list:

  • Duplication. Identical numeric series, chromatograms, spectra, or images across supposedly independent samples, tests, or customers, and identical standard deviations or rounded values that repeat too neatly.

  • Provenance. Test article IDs, lot numbers, and sample counts that do not match what you shipped, and names, dates, or product references from other projects left in file metadata or report templates.

  • Timing. Timestamps outside the lab's operating hours, overlapping runs on a single instrument, or test durations shorter than the method requires.

  • Traceability. Missing calibration or maintenance records for the instruments used, no instrument audit trail, or raw files that exist only as screenshots or PDFs.

  • Fit. Results that are implausibly clean compared with your own development data or with the method's known variability.

The common thread is the ALCOA principles of data integrity: data should be attributable, legible, contemporaneous, original, and accurate. Raw files with intact metadata and instrument audit trails let you test each of those. A two-page summary report cannot.


Keeping this in proportion

The notices describe a trend and specific enforcement, not a verdict on contract testing generally, and most labs are doing legitimate work. The point is asymmetry. Independent verification costs some audit time and a few raw-data requests. A refused submission, or a file whose integrity FDA questions as a whole, costs months and the credibility of everything else in it.

Quality leaders who treat test labs as suppliers, with the same qualification, monitoring, and records as any other, will not need to hope that the lab is honest.


Sources: FDA, Notifications on Data Integrity — Medical Devices, notice of June 25, 2026; Lachman Consultants, Data Integrity in Device Submissions, June 30, 2026; FDA, ASCA-Accredited Testing Laboratories and ASCA-Recognized Accreditation Bodies pages; FDA Letter to Industry on third-party test lab data, February 2024, as reported by RAPS Regulatory Focus and Knobbe Medical; Akin Gump, FDA Debuts a New Communications and Compliance Tool for Device Data Integrity Concerns; FDA Law Blog, CDRH's FY 2027 Proposed Guidances, October 2026; FDA, ASCA program pages and ASCA program guidance (docket FDA-2020-D-0981); ISO 13485:2016, clauses 4.1.5, 7.4.1 and 7.4.3; 21 CFR Part 58.

Comments


bottom of page