Devices · diagnostics · 4 minutes reading

Clinical trials digest

Device investigations and diagnostic performance studies

A medical device trial is not a drug trial with a serial number. UDI, lot accountability, ISO 14155, MDR, and FDA IDE shape the file. An IVD performance study adds specimen chain of custody, reader studies, and accuracy methods.

MDRIVDRUDIDiagnostics
← All articles

4 min

Reading time

EU+US

Doctors and sites

1 record

Drug, device, IVD

Why device studies break drug EDC

Drug EDC assumes a molecule, a dose, and a visit. A device investigation assumes a product with a UDI, a lot, sometimes a software version, and accountability from shipment to explant or return. ISO 14155 and the MDR describe the investigation. FDA IDE describes the US path. Serious adverse device effects have EU vigilance and US MDR reporting clocks that are not SUSAR clocks. Sites that run both kinds of study should not pretend they are the same form with a different title. ValidTrials treats device accountability as a first-class part of the participant record. We configure it from the protocol and the device list, not from a spreadsheet the CRA keeps privately.

Why device studies break drug EDC

Full article for doctors, investigators, and readers who follow clinical trials.

UDI, lots, and who touched the product

Every unit that reaches a participant should be traceable: which lot, which serial, which site, which person, which procedure date. Recalls and deficiencies make that trace a safety tool, not an inventory hobby. The record should survive a staff change at the clinic. Inspectors will ask for it. ValidTrials records each kit with its UDI, expiry, status, and the participant it was assigned to. That is the difference between a device investigation and a drug trial that borrowed a serial-number field.

IVD performance is a different study again

In vitro diagnostic performance studies under the IVDR and ISO 20916 add specimens, chain of custody, kit lots, and often a reader study. A sample has a life: drawn, labelled, shipped, stored, tested, result reported, leftover handled. If that chain breaks, the number on the report is unattached. Reader studies add cases, readers, and sometimes a reference standard. Doctors in laboratory medicine, pathology, and radiology live in this world. Accuracy methods — CLSI EP12, EP24, EP05, EP15, Bland–Altman, Deming, Passing–Bablok, ROC — belong with the locked snapshot, not in a statistician’s laptop after an export. ValidTrials holds the specimen and the method with the study. It still does not write that a test is clinically validated. A person signs that conclusion.

IVD performance is a different study again

Full article for doctors, investigators, and readers who follow clinical trials.

AI-enabled devices and locked versions

When the product includes an algorithm, the version that ran for a participant is part of the evidence. A model that silently updates is not a locked intervention. Reader studies, subgroup performance, and a predetermined change control plan are how authorities now expect that story to be told. ValidTrials records the algorithm version per participant. It does not use generative tools. The newsletter will say this as often as vendors say the opposite.

One clinic, three product types

Hospitals run drug trials, device investigations, and diagnostic studies with the same investigators and the same nurses. They should not need three electronic systems to do it. The authority packages differ. The site work is visits, consent, safety, and accountability. That is the gap ValidTrials is built to close. If you run a device or IVD study and your EDC still looks like a drug CRF with a comment field for the serial number, write to mail@validtrials.com with the protocol.

One clinic, three product types

Full article for doctors, investigators, and readers who follow clinical trials.

Read more from the digest.

Subscribe, or bring us the protocol.

The digest is for doctors and trial readers. The study conversation starts with us.