Start

Contributing

Changes to a system that renders clinical information are changes to a risk control, whether or not anyone calls them that. This page says what a proposal has to contain, who has to agree, and which changes are not accepted at all.

Stable · v1.0 ISO 14971 IEC 62366-1

The standard a change is held to

Every rule in this system exists because something can go wrong without it. So a proposal is not assessed on whether it is nicer, more modern or more consistent with a fashion. It is assessed on one question:

The question

What can a clinician now do, misread, or fail to notice, that they could not before? A proposal that cannot answer this is not ready. An answer of "nothing" is acceptable and common — it just has to be stated rather than assumed.

The corollary matters as much. A rule that forces clinicians into a workaround is a worse control than no rule, because it trains people to route around the interface. If you are here because a rule blocked real work, that is one of the most valuable things you can report.

What a proposal contains

SectionWhat it has to say
The use situation Who, doing what, under what conditions. In the terms of Usability & context of use: specified users, goals, context. Not a persona — a situation.
The failure today What goes wrong now, and how you know. An observation beats an opinion; a near-miss beats a preference.
The change The smallest version of it. Large proposals are almost always several small ones that should be argued separately.
New failure modes What the change makes possible that was not possible before, including for users of assistive technology and in degraded network or data states.
Standards touched Whether it affects alarm priority (IEC 60601-1-8), contrast or interaction (WCAG 2.2), or a stated risk control on a component page.
Do's and don'ts At least one pair, authored with the change. If you cannot write the "don't", the rule is not yet clear enough to enforce.

Three classes of change

ClassExamplesReview
Editorial Wording, examples, a clearer explanation of an existing rule, a new do/don't pair that restates something already true. One reviewer. Merged when the prose is right.
Behavioural A new component, a new state, a changed keyboard model, a new open token, a relaxed or tightened rule. Two reviewers, one of whom must have clinical or human-factors background. Requires the full proposal above.
Governed Anything touching --alarm-*, --input, --ring, --destructive, alarm priority mapping, the filter escape rule, or a stated risk control. Two reviewers plus a written risk assessment recorded in tokens.json or the component's safety notes. Never merged the same day it is opened.
Governed changes

A governed change is a change to a controlled characteristic. It carries a rationale that a manufacturer's risk file can cite, and the CI report from the change is retained as evidence. "It looked better" is not a rationale; neither is "the brand guidelines say so". See Tokens & governance.

Reporting an unsafe behaviour

This is the highest-priority class of report and it skips the normal proposal format. Include:

  1. The component and page, and the version you are on.
  2. What a clinician would conclude from what is on screen, and what is actually true.
  3. Whether it is reachable in normal use, or only in a degraded state.
  4. A screenshot or reproduction with fabricated data only — never real patient data, never a real MRN, never a screenshot of a live system.

If you are a manufacturer and the behaviour is reachable in a product you have placed on the market, your own vigilance and post-market surveillance obligations apply independently of anything reported here, and on their own timelines. Reporting it here is not a substitute for them.

Changes that are not accepted

These are not permanent in principle — they are permanent until someone brings evidence, not a preference. Two of them are narrower than they were because someone did.

Do's and don'ts

Do

"On a night shift, a nurse acknowledging from the worklist twice acted on the wrong row because the list re-sorted between look and click. Observed twice in two weeks."

A use situation, a failure, and evidence. This is a proposal a reviewer can act on within an hour.

Don't

"The worklist feels cluttered. Suggest we simplify the rows."

No user, no situation, no failure. There is nothing here to agree or disagree with, and nothing to verify afterwards.

Do

Propose the smallest change that addresses the failure, and say what it makes newly possible.

"This lets a clinician dismiss the banner without reading the provenance line" is the sentence that gets a proposal reviewed properly.

Don't

Bundle a component rewrite, a token change and a documentation restructure into one proposal.

A governed change buried in a large diff gets the review a small editorial change deserves, which is how controls quietly disappear.

Do

Report a rule that forced you into a workaround, with the workaround described.

A rule people route around is producing use errors somewhere else. The workaround is the diagnosis.

Don't

Fork the component locally, remove the rule, and carry the patch forward silently.

Your product now differs from its documentation in a way no reviewer of either will notice.

Do

Use fabricated data in every example, screenshot and reproduction.

Every example in this system is fictional for the same reason. There is no version of a real MRN in an issue tracker that is acceptable.

Don't

Attach a screenshot from a live clinical system, "anonymised" by cropping.

Cropping is not anonymisation. Timestamps, bed numbers and value combinations re-identify.

Before you open it

NotJustAnyMed.Tech Design System · Contributing · v1.0 · draft for review
Reference applications named in this system are fictional; all patient data and reference ranges shown are fabricated and illustrative.