PSUR, CER and technical file: keeping the three aligned

A well-kept technical file can trip on a detail. The PSUR reports a sales volume that the clinical evaluation report ignores, and the version of the instructions for use quoted in one is not the version filed in the other. Consistency between the PSUR, the CER and the technical documentation is not a housekeeping concern: it is the first thing a notified body checks once it opens the three documents side by side. Here is what each one asserts, where the contradictions come from, and how to reconcile them.

What each document asserts

The three documents answer different questions, on different cycles.

The technical documentation

Annex II to Regulation (EU) 2017/745 sets the content of the technical documentation: device description, information supplied by the manufacturer, design and manufacturing information, general safety and performance requirements, benefit/risk analysis and risk management, product verification and validation. It is the claimed state of the product at a given moment, under an identified version. Annex III adds the technical documentation on post-market surveillance: the surveillance plan required by Article 84, the PSUR of Article 86 and the post-market surveillance report of Article 85.

The clinical evaluation report

Article 61 and Part A of Annex XIV govern clinical evaluation. The CER documents the results of that evaluation and the clinical evidence it rests on, in support of the conformity assessment. The clinical evaluation and its documentation are updated throughout the life cycle of the device, with data from the post-market clinical follow-up plan (Part B of Annex XIV) and from the post-market surveillance plan. MDCG 2020-8 offers a template for the PMCF evaluation report.

The PSUR

Article 86 requires a PSUR from manufacturers of class IIa, IIb and III devices. It summarises the results and conclusions of the analysis of post-market surveillance data gathered under the Article 84 plan, together with a description of any preventive and corrective actions taken. It sets out the conclusions of the benefit/risk determination, the main findings of the PMCF, the volume of sales, and an estimate of the size and characteristics of the population using the device. MDCG 2022-21 provides a template and a way of presenting the data; MDCG 2025-10 (December 2025) places it back within the expected workings of the surveillance system.

Where inconsistency between PSUR, CER and technical documentation begins

Contradictions almost always sit in the same places.

• The figures. Sales volumes, serious incidents, complaints, exposure estimates: each document draws on an extraction run on a different day. Two extraction dates produce two totals.
• The periods. The PSUR covers one collection window, the PMCF evaluation report another, the CER quotes a third. Without a common calendar, comparing figures means nothing.
• The versions. The CER refers to version 4 of the instructions for use, Annex II holds version 5, and the PSUR describes labelling changed in the meantime.
• The scope. A PSUR written for a group of devices and a CER written per variant do not cover the same list of references. That gap is legitimate, provided it is written down and justified.
• The conclusions. The PSUR flags a signal to monitor; the CER carries over last year's wording and says nothing about it. This is the most expensive gap, because it goes straight to the benefit/risk determination.
• The vocabulary. Intended purpose, indication, target population: three wordings for the same thing are enough to cast doubt on the scope covered.

The calendar comes before the method

Frequencies are set by the texts.

• Class IIb and class III: the PSUR is updated at least annually (Article 86).
• Class IIa: the PSUR is updated when necessary, and at least every two years.
• Class I: no PSUR, but the post-market surveillance report of Article 85.
• Class III and implantable devices: the PSUR is submitted to the notified body through the electronic system referred to in Article 92; the PMCF evaluation report and, where indicated, the summary of safety and clinical performance are updated at least annually (Article 61(11)).
• IVDs: for class C and class D devices, the PSUR falls under Article 81 of Regulation (EU) 2017/746 and is updated at least annually.

One simple rule follows from these frequencies: set a common data cut-off date. The PSUR, the PMCF evaluation report and the CER update are prepared on the same data window, with the same extraction date. Where that is impossible, write the reason for the gap into each affected document: an explained gap can be defended, a silent one becomes a notified body question.

A four-step reconciliation method

1. One source per data point

For every figure, name a single source system and a single extraction date: sales, complaints, incidents, field safety corrective actions. The figure is then carried across the three documents unchanged, with its source and date stated. A number recalculated by hand in one document only is a contradiction waiting to surface.

2. A section mapping table

Map the sections once and for all: the PSUR benefit/risk conclusion points to a given part of the CER, which in turn links to a given Annex II element and a given risk management record. The table is not a regulatory deliverable: it supports the review, the onboarding of a new team member, and the answers given during assessment.

3. A cross-review that runs both ways

Reading the PSUR to work out what must change in the CER is not enough. Read the CER too, looking for what it asserts and the PSUR has not confirmed: a performance claim with no post-market data behind it, a claimed population that sales do not cover, a residual risk deemed acceptable that reported incidents contradict.

4. A record of gaps and decisions

Every gap found leads to a dated decision, with an owner and a target document. Some are fixed in the text; others call for a quality system action, or a corrective action. Keep the list of gaps and rulings: it is worth more than a corrected version with no history, because it shows the review took place.

In practice

Three habits prevent most contradictions. Set a common cut-off date across the three documents. Carry every figure from its single source, without retyping it. Have the three read together, not one after the other, by someone who wrote none of them.

The reconciliation stays a line-by-line comparison across long documents, at every cycle. That is the kind of task a tool can prepare: in EryonOne, the internal consistency axis flags passages that contradict each other from one document to the next and quotes each extract, along with the reference text where the finding is regulatory. The report stays working material rather than an approval: the ruling belongs to the expert who signs. You can review the frameworks covered, or submit your own internal requirements (SOPs, claims matrices, past rulings) so that they are applied alongside the official texts.