A performance evaluation is not a performance study

This blog has already covered the three tests an IVDR performance evaluation has to run — scientific validity, analytical performance, and clinical performance, the evidence package Article 56 asks a manufacturer to assemble before a CE mark issues. That post treats the performance evaluation as a body of evidence a manufacturer gathers. It doesn't cover the separate question of where some of that evidence has to come from: a performance study run on human subjects, which IVDR regulates as its own activity, under its own chapter, with its own approval requirements that have nothing to do with the file the evaluation eventually produces.

One is the evidence; the other is how some of it gets made

Article 2's definitions keep the two terms apart on purpose. A performance evaluation is the assessment and analysis of data to establish or verify a device's scientific validity, analytical performance, and clinical performance — the output Article 56 requires, built from existing literature, prior clinical experience, and testing already available wherever that's sufficient. A performance study is a study undertaken to establish or confirm that data, involving human subjects, whose results feed into that same evaluation. The relationship runs one direction: a performance study, when one is needed, produces data the performance evaluation consumes. Not every evaluation requires a study at all, and IVDR doesn't ask a manufacturer to run one just to have run one.

Most performance studies don't need approval — some need it before they start

IVDR's default assumption is that a study using samples already collected, or samples collected for another purpose and left over, doesn't require the same regulatory gate as one that reaches into a subject specifically for the study. Article 58 narrows the authorization requirement to the cases where that assumption breaks down: studies involving surgically invasive sample-taking done solely for the study's purposes, interventional clinical performance studies, and studies imposing additional invasive or burdensome procedures on subjects, alongside a second category Article 58 covers by name — performance studies of companion diagnostics, regardless of how the sample is taken. A study that falls into neither category can often proceed on lighter footing; a study that does needs sign-off before a single sample is drawn for it.

Forty-five days, not a formality

Where Article 58 applies, Article 66 sets the review clock: the Member State has to notify the sponsor of its decision on the application within 45 days of the application's validation date, extendable by a further 20 days where the Member State needs to consult experts. That's a real review period a study timeline has to build around, not a nominal filing step slotted in alongside the rest of the submission — a first-patient-in date set without accounting for it is a date the regulation doesn't actually support.

A third path exists for studies on devices already CE marked

Article 70 covers a narrower case: a performance study run on a device that already carries the CE mark, where the study goes beyond the device's intended purpose or otherwise involves additional invasive or burdensome procedures — the post-market performance follow-up work a manufacturer runs after launch rather than before it. That path doesn't require the same authorization Article 58 does; it requires notifying the competent authority at least 30 days before the study starts. A post-market study file that treats this notification as optional paperwork, or skips it because the device is already on the market, has missed the one obligation Article 70 actually imposes on work IVDR would otherwise let a manufacturer run unannounced.

Where this meets the file

A performance-evaluation plan has to answer a gating question before it answers anything else: does closing the evidence gap require a performance study at all, and if it does, does that study fall under Article 58's authorization requirement, Article 70's notification requirement, or neither. The PMPF plan this blog has already covered is exactly where an Article 70 study is most likely to surface, tucked inside a post-market commitment rather than flagged as its own regulatory gate. A performance-evaluation worksheet built around this three-way split is previewed in the launch catalog. If your program routes performance studies differently, the shelf takes that correction directly.