Consistency without conformity of thought

Under structural load on conformity assessment, another pressure runs through the system: comparable evidence is not always approached in a comparable way, within a team, across teams, and across Notified Bodies. Manufacturers feel that as unpredictability. Assessors feel it as the cost of improvising structure under load. Patients feel it indirectly, as delay, rework, and uneven clarity about what “enough evidence” means.

Talking about consistency is delicate. Professional judgment is not a bug. Two competent reviewers can weigh residual uncertainty differently and still act within the law. The problem is not that humans think. The problem is when process, expectations, and evidence pathways vary so much that the framework itself becomes personal, so that “how we work here” is remembered rather than designed.

Consistency worth wanting is consistency of framework: shared structure, inspectable reasoning, and retained human authority, not automated sameness of conclusions.

What manufacturers name first

When MedTech Europe asked medical-device manufacturers what would help most with MDR transition, the leading theme was not a vague wish for “easier rules.” It was alignment. Among respondents who answered that question, 28 of 96 named aligned and clear requirements within a Notified Body and among Notified Bodies, the single most frequent request, closely followed by predictability of timelines (27) and structured dialogue (19) (MedTech Europe 2024). Capacity and leverage of existing evidence also appear, but the top of the list is about knowing what the system expects and that those expectations do not shift with the room.

Related friction shows up in clinical evaluation. Half of medical-device respondents in the same survey reported that at least one certificate was significantly delayed or closed negatively because clinical (or performance) evaluation was challenged, and the top obstacle named was lack of clarity about clinical evidence expectations (MedTech Europe 2024). Clarity is not the enemy of rigor. Unclear, shifting, or person-dependent expectations produce rework without necessarily producing better safety arguments.

The sector already treats consistency as a system goal

It would be a mistake to cast consistency as a vendor slogan imposed on reluctant institutions. Notified Bodies and authorities have spent years naming the same problem from the inside: fragmentation and uneven application undermine predictability, efficiency, and trust in the framework.

A July 2025 joint perspective from Team-NB and NBCG-Med executives describes the EU medical-device regulatory framework as facing fragmentation, inefficiencies, and inconsistent application of rules, and argues for stronger coordination and centralised support so that assessment remains scientifically rigorous while becoming more coherent across the system (Team-NB / NBCG-Med governance paper). Team-NB’s best-practice guidance for technical documentation was built by reviewing many member-specific guides and aiming at a more unified expectation for how Annex II and III material should be structured and presented (Team-NB TD submission BPG). Harmonised submission expectations reduce avoidable variance before assessment even begins.

Policy has moved in the same direction. Commission Implementing Regulation (EU) 2026/977 now sets legally binding maximum timelines and uniform procedural requirements for Notified Body conformity assessment, alongside quotation rules and performance monitoring. The legal project is not “one opinion for every device.” It is more uniform process quality so that manufacturers and assessors operate under clearer, more comparable rules of the road.

Two kinds of variance

Not all difference is the same. Distinguishing them is the only way to talk about consistency without insulting professional judgment.

• Judgment under uncertainty: residual risk, clinical context, state of the art, and proportionate evidence can legitimately support different professional conclusions. Suppressing that difference would not be rigor; it would be theater.
• Process and framework variance: which checklist is used, how findings are phrased, how evidence is linked, how similar prior cases are remembered, whether uncertainty is recorded or smoothed over. This kind of variance is designable. When left to habit, it becomes load: rework, surprise deficiencies, and uneven manufacturer experience for comparable products.

Under system pressure, framework variance gets worse. People under load fall back on personal shortcuts. Institutions that lack shared structure export inconsistency even when every individual is acting in good faith.

The wrong target: conformity of thought

A crude response to inconsistency is to automate conclusions, to treat models, templates, or rigid scripts as substitutes for qualified assessment. That confuses uniformity of output with integrity of process. Conformity assessment exists precisely because some decisions require accountable human judgment under incomplete information. Making every review “the same” by hiding uncertainty or outsourcing disposition would not make the system more trustworthy. It would make it more brittle and less defensible.

Another crude response is to deny the problem: to insist that any call for consistency is an attack on professional autonomy. That confuses autonomy with opacity. A professional who can show the pathway from requirement to evidence to disposition is not less free; they are more able to defend their judgment under scrutiny: by peers, competent authorities, and their own quality system.

The design goal is not conformity of thought. It is a shared lattice in which thought can be compared, challenged, and owned.

What “designed consistency” means

If the target is framework integrity rather than identical opinions, several design obligations follow. None of them require automating certification.

• Shared structure for the case: requirements, evidence, findings, and dispositions arranged so that two reviewers of the same dossier can navigate the same map, even when they disagree on a call.
• Inspectable reasoning: conclusions linked to sources (document, clause, page, history) so that difference of judgment is visible rather than buried in untraceable prose.
• Visible uncertainty: gaps and ambiguity recorded as such, not converted into false confidence for the sake of looking decisive.
• Institutional memory without fossilisation: prior approaches and rationales available as context, not as silent personal folklore that only long-tenured reviewers can apply.
• Human authority retained: software may organise, retrieve, compare, and draft; qualified professionals evaluate evidence and remain responsible for the assessment record.

Consistency should be designed into the assessment environment so that judgment can remain human without remaining accidental.

Load multiplies variance

Under light load, informal excellence can mask weak infrastructure. Experienced reviewers carry structure in their heads; teams correct outliers through hallway conversation; manufacturers learn “how this NB likes it” over years of relationship. Under the load described in series A and B, those informal buffers fail. New staff ramp slowly. Parallel teams diverge. Comparable dossiers receive incomparable journeys.

The remedy is not only more people and not only stricter sameness of opinion. It is better environments for expert work: structured cases, source-linked assessment, designed frameworks, and disposition that stays with the qualified professional. If judgment is the authority, the framework must be good enough that authority can be exercised consistently under pressure.