IVDIn Vitro Diagnostic device

An In Vitro Diagnostic (IVD) device is a reagent, instrument, calibrator, control material, kit, software or system intended for use in the in vitro examination of specimens — including blood, tissue, urine and other body fluids — derived from the human body, for the purpose of providing information for diagnosis, monitoring, screening, prediction, prognosis, companion diagnostics, or determination of physiological or pathological state. FDA's definition is in 21 CFR 809.3; the EU IVDR definition is in Article 2(2).

The IVD definition is purpose-based, not technology-based. The same mass spectrometer is a general-purpose laboratory instrument when sold for proteomics research and an IVD when sold for clinical newborn screening. Labelling and intended use determine regulatory status.

Three federal bodies regulate IVDs in the US, and a fourth often appears for reimbursement. FDA regulates the IVD itself as a device — premarket submission, post-market surveillance, MDR/vigilance, QSR/QMSR. CMS administers CLIA, which regulates the laboratory that runs the test and assigns each test a complexity category (waived, moderate, high). State laboratory licensing programmes (notably New York's Wadsworth Center and CMS-accredited equivalents) add a third layer for some tests. Reimbursement is set by CMS and private payers and is independent of clearance status.

Most IVDs reach market through 510(k) as Class II — substantial equivalence to a legally marketed predicate IVD. Novel IVDs without predicate go through De Novo; the highest-risk IVDs (HIV diagnostics, blood-screening tests, companion diagnostics for high-impact therapies) go through PMA. A small set of low-risk IVDs (e.g. some controls and calibrators) are Class I and 510(k)-exempt.

CLIA categorisation is assigned by FDA at the time of clearance based on a seven-criteria scoring matrix (knowledge, training, reagent prep, operational steps, calibration, interpretation, troubleshooting). Waived tests can be run in any CLIA-certificated site including physician offices; high-complexity tests require a high-complexity laboratory with qualified personnel under 42 CFR 493 Subpart M.

Laboratory Developed Tests (LDTs) are IVDs designed, manufactured and used within a single laboratory. For decades FDA exercised enforcement discretion over LDTs, regulating them only through CLIA. The May 2024 final rule (89 FR 37286) ended that discretion. LDTs are devices under the FD&C Act and must comply with FDA requirements on a five-stage phase-out schedule running through May 2028.

The phase-out: MDR/correction-and-removal reporting (stage 1, May 2025); registration/listing and labelling (stage 2, May 2026); QSR/QMSR (stage 3, May 2027); premarket submission for high-risk LDTs (stage 4, Nov 2027); premarket submission for moderate-risk LDTs (stage 5, May 2028). Several enforcement-discretion carve-outs remain — including for LDTs first offered before May 2024 (1976-style grandfathering with limits), tests for unmet needs, and certain academic medical centre tests.

Litigation challenging the rule is ongoing; clinical laboratories should not bet the programme on the rule being vacated. The defensible position is to plan for compliance and build a list of which existing LDTs would be in scope at each phase.

EU IVDR (2017/746) replaced the IVDD on 26 May 2022 with a transition through extended dates set out in the 2024 amending regulation (Class D until end of 2027, Class C until end of 2028, Class B and A sterile until end of 2029, conditional on QMS implementation and CE-mark application milestones).

Classification under Annex VIII has seven rules and lands every IVD in Class A (low risk — laboratory tools, reagents), Class B (default — most analytes), Class C (higher individual or moderate public-health impact — cancer markers, infectious agents without high public-health risk, companion diagnostics), or Class D (highest — blood/tissue screening, life-threatening infectious agents). Roughly 80% of IVDs were self-certified Class I under IVDD; under IVDR perhaps 80% require Notified Body involvement. The supply-side bottleneck (around a dozen designated IVDR Notified Bodies) is the practical risk to EU market continuity.

Every IVD submission needs two layers of performance. Analytical performance answers 'does the test measure what it claims to measure?' — accuracy, precision (repeatability, reproducibility), linearity, limit of blank / detection / quantitation, analytical specificity (interference), measuring interval, and matrix comparison. The CLSI EP-series guidelines define the protocols and are the accepted reference for both FDA submissions and IVDR technical documentation.

Clinical performance answers 'does the test detect or predict the clinical condition the labelling claims?' — clinical sensitivity, clinical specificity, positive and negative predictive value, likelihood ratios, and where applicable comparator-arm clinical-utility evidence. For IVDR Class C/D and FDA PMA companion diagnostics this requires a prospective clinical study; for 510(k) it can often be method-comparison against an established reference using banked samples.

Most modern IVDs contain software — instrument firmware, middleware, LIS interfaces, result-interpretation algorithms. Standalone IVD software (e.g. a pathology AI algorithm interpreting digital slides) is regulated as SaMD and follows the IMDRF risk framework plus IEC 62304. Embedded IVD software is regulated as part of the IVD device under the same premarket route, with IEC 62304 lifecycle evidence in the submission.

Cybersecurity premarket documentation per FDA's September 2023 final guidance applies to any networked IVD instrument and any IVD with software: security risk management, architecture views, SBOM, security testing, and labelling. Hospital network deployment dramatically widens the IVD attack surface and reviewers treat this as a Class-relevant risk control.

• Treating an analyte change as a labelling tweak. Adding a new specimen type (e.g. plasma in addition to serum) requires fresh analytical performance and often clinical performance.
• CLIA categorisation overlooked at design. A test designed without thought to operational simplicity often lands High Complexity and loses the physician-office market.
• LDT inventory absent. Laboratories cannot plan phase-out compliance without knowing which tests are in scope at each milestone.
• Reliance on the predicate's clinical data without verifying matrix and population match. FDA increasingly asks for clinical bridging data even for 510(k) IVDs.
• IVDR classification underestimated. Companion diagnostics, infectious-disease tests and self-test devices are routinely under-classified by sponsors used to IVDD self-certification.