Foundations

Usability & context of use

What this system is trying to produce, for whom, and how anyone would know it worked. Every other page describes attributes; this page describes the outcome those attributes are supposed to contribute to.

Stable · v1.0 Framework · ISO 9241-11:2018 Framework · ISO 9241-210 IEC 62366-1
Framework, not conformance

ISO 9241-11 is a definitions and concepts standard. It contains no requirements and its scope states that it does not describe processes or methods — you cannot conform to it and this system does not claim to. It is used here for one reason: it is where the vocabulary in your IEC 62366-1 usability engineering file is defined, so structuring the design system the same way lets the two documents reference each other instead of talking past each other.

Dashed pills throughout this system mark a framework or vocabulary reference. Solid pills mark a standard a product can actually be assessed against.

Usability is an outcome, not a property

ISO 9241-11 defines usability as the extent to which a system can be used by specified users to achieve specified goals with effectiveness, efficiency and satisfaction in a specified context of use. The consequence is easy to state and easy to forget:

Usability is a property of a use, not of a component. A 44 px button is not usable. A clinician acquiring a lead-off-free ECG on a 15-inch cart display while gloved, and correctly identifying an anterior STEMI within four minutes, is a usable outcome — to which that button contributed.

This matters for how you read every other page in this system. A component page can only supply attributes believed to contribute to usability in an intended context. Whether they actually do is settled by observing use, not by reading documentation. The same is true of the Clinical safety notes sections: each one is a design rationale, not a hazard-analysis output, and it only holds inside the context of use described below.

Human-centred quality

Usability is one of four outcomes of use. Together they are called human-centred quality, and this system organises itself around all four rather than treating safety as the only one that matters in a medical device.

OutcomeQuestion it answersWhere it lives here
UsabilityDid specified users achieve their goals, at what cost, and how did it feel? This page, plus Outcomes of use on each component
AccessibilityDoes it work for the widest range of user needs and capabilities? Accessibility on every page
User experienceWhat did users perceive and feel, before, during and after? Satisfaction below; Confidence disclosure
Avoidance of harm from useWhat could go wrong, and what stops it? Clinical safety notes on every page

Specified

Users, goals and context are written down, not assumed. Guidance that does not say who it is for is guidance nobody can check.

Whole

Effectiveness, efficiency and satisfaction are all reported. A system optimised for one while silently degrading another has not improved.

Observed

Claims about usability are settled by watching real users in a real context, never by the confidence of the documentation.

Specified users

Users are not only the people who operate the software. ISO includes people who use its output without producing it, and people who support or maintain it. Omitting those two groups is the most common way a clinical design system ends up unusable for half the people it affects.

This page works one example through

Users, goals and context are properties of a device, not of a design system — so this page demonstrates the method on a single application, AcuteLine, rather than attempting a table that covers all of them. The other reference applications carry their own profiles on Reference applications, and they differ sharply: a screening operator who cannot grade the image, or a patient dosing at home with nobody to ask, is not a variant of the users below. Do not inherit this profile for your own device. Repeat the method.

GroupRelationshipCharacteristics that influence usability
ED registrar / consultantOperates — primary Frequently interrupted; gloved; standing; variable experience with the tool; shift fatigue; reads at 50–70 cm on a workstation or 1 m on a cart
ED nurseOperates — acquisition Acquires the ECG, manages electrodes; may never open the review screen; time-poor at triage
Cardiology on-callOperates — escalation recipient Arrives without prior context; may be off-site on a small screen; needs the whole picture in one view
Cath lab teamUses the output Never touches the software; receives a decision and a time. Their usability depends on what the notification contains
Clinical systems administratorSupports Configures alarm thresholds, escalation chains, filters. Errors here affect every clinician silently
Biomedical engineerMaintains Selects and calibrates displays, sets the deployment scale factor. Works from Scaling & displays
Quality / audit reviewerUses the output Reads the audit trail months later with no memory of the shift. Their usability depends on the record
The widest range of capabilities

Within every group above, assume the full range: colour vision deficiency, presbyopia, tremor, hearing loss, non-native language, first shift with the tool, and twelfth hour of a night shift. Accessibility in ISO's sense is broader than WCAG conformance — WCAG is the testable floor, not the definition.

Specified goals

A goal is an intended outcome and is independent of the means. A task is one particular means of achieving it. Most design documentation — including most of this system before this page existed — describes tasks and never states the goal, which makes effectiveness undefinable.

AcuteLine · goal decomposition
Goal — a patient presenting with chest pain receives definitive treatment appropriate to their diagnosis, within the guideline window.
Subgoal — an acute ischaemic pattern present on the ECG is detected.
Task: acquire a 12-lead ECG; software analyses it.
Subgoal — a qualified clinician reviews the evidence and reaches their own conclusion.
Task: open ECG review; assess the trace; agree, disagree or record "cannot assess".
Subgoal — the conclusion reaches the people who act on it.
Task: activate the local pathway; notify the cath lab.
Subgoal — the decision is attributable and reconstructable afterwards.
Task: acknowledge the alarm; record the assessment.
Subgoal — a patient without ACS is not sent down the pathway.
Task: override; request serial ECG.
Goals conflict, and the conflict is the design problem

The organisation's goal is to reduce door-to-balloon time. The clinician's goal includes not activating the cath lab unnecessarily at 3 a.m. Optimising the interface purely for the first produces over-triage; purely for the second produces missed infarcts. Almost every contested decision in this system — default sort order, alarm priority thresholds, whether disagreement is one tap — is a position on that trade-off, and should be argued as such rather than as a usability preference.

Specified context of use

Context of use is the combination of users, goals and tasks, resources, and environment. It is the part of this system that was always present in substance and never named.

The table below is one application's context, worked through in full. Every row would read differently for a device that sits elsewhere on the axes in Reference applications — a home-use device has no shared workstation, no hierarchy between registrar and consultant, and an expendable resource (the user's own cognitive capacity) that AcuteLine never has to consider.

ComponentAcuteLine
UsersThe seven groups above.
Goals & tasksAs decomposed above. Tasks are frequently interrupted and rarely completed in one sitting.
Reusable resources ECG acquisition hardware; workstation, cart and gantry displays; the patient record and any prior ECG; local chest-pain protocol; colleagues, including the on-call cardiologist; assistive technology.
Expendable resources Clinician time and attention — the binding constraint. Also electrodes, network connectivity, and cath lab capacity.
Technical environment Displays from 10″ carts to 4 K reading monitors; variable pixel density; touch and pointer input; intermittent connectivity; shared workstations. See Scaling & displays.
Physical environment Bright resuscitation bay to darkened reading room; noise; gloves; vehicle motion; viewing distance 50 cm to 4 m; screens read off-axis.
Social, cultural, organisational Handover at the keyboard; twelve people per workstation; hierarchy between registrar and consultant; multiple languages; local protocol variation; the knowledge that actions are attributable and auditable.
Prescribed versus actual

Where the way a task is actually performed diverges from the way it is prescribed, that divergence is information about the context, not a discipline problem. The clearest example in this system: slow sign-in on a shared workstation is prescribed as individual authentication and actually performed as one shared account nobody logs out of. The design response is a faster badge-first sign-in and read-only locking — not a policy reminder. See Sign in / sign out.

The three components

Effectiveness — accuracy and completeness

The extent to which actual outcomes match intended outcomes, and the extent to which all intended outcomes are achieved. The two are independent: a queue can be perfectly accurate about every case it shows and radically incomplete because a filter hid eight of them.

This is why no default filters and an empty state that proves liveness are completeness controls, not conveniences.

Efficiency — resources used for the results achieved

Resources are time, human effort, financial cost and materials. In this context the first two dominate, and human effort is the one most often left unmeasured.

Underload is also a cost

Expenditure of human effort can involve excessive demand or underload, and either can cause negative consequences. A model tuned so conservatively that a clinician rarely disagrees produces vigilance decrement: the reviewer stops genuinely reviewing. Efficiency measures that only ever reward "fewer actions" will miss this entirely.

Satisfaction — physical, cognitive and emotional responses

The most neglected component in clinical software, and the one where this system already does real work without having named it.

Use error

The term this system uses — imported into ISO 9241-11 from IEC 62366-1 — is use error: a user action, or lack of action, that leads to a different result than the manufacturer intended or the user expected. Never "user error" or "human error".

The distinction is not politeness. Use error is defined to include the user being unable to complete a task at all, and it typically arises from a mismatch between the user, the interface, the task and the environment — which places it squarely inside the designer's responsibility. Two consequences for how pages here are written:

Measuring it

At least one measure for each of effectiveness, efficiency and satisfaction is normally required, and each can be measured objectively or as perceived by the user. The two frequently disagree, and the disagreement is itself a finding — a clinician who believes they reviewed a trace carefully and did not is a usability problem that no objective measure alone will surface.

ComponentObjective measuresPerceived measures
Effectiveness Proportion of true findings correctly actioned · wrong-patient actions per 1 000 sessions · cases in scope not reviewed · agreement with adjudicated truth Clinician's confidence that they reviewed every relevant case · belief that the queue is complete
Efficiency Time to acknowledge by priority · time from acquisition to decision · alarms per clinician-hour · screens traversed to reach evidence Perceived effort of a review · perceived interruption burden · perceived time spent versus actual
Satisfaction Voluntary reuse when an alternative exists · override rate and its stability · acknowledge-without-open rate as a disengagement proxy Calibrated trust in the model · perceived safety of acting on an output · comfort over a full shift · willingness to recommend to a colleague

No single measure represents overall usability, and different measures of the same component can support opposite conclusions. Acknowledge-without-open rate appears above under satisfaction as a disengagement proxy and also functions as an efficiency measure; that overlap is expected and is a reason to report several rather than to pick a winner.

Learnability and maintainability

Both are usability considered for a different goal, not separate qualities.

Do's and don'ts

Do
Specified users
ED registrar · ED nurse · cardiology on-call · cath lab team · systems administrator · biomedical engineer · audit reviewer

Named groups with characteristics, including the people who use the output and the people who maintain the system.

Don't
Specified users
Clinicians

One word covering a registrar on their first shift and a consultant with presbyopia. Nothing can be evaluated against it.

Do
Effectiveness
96%
correct priority actioned
Efficiency
41s
median time to acknowledge
Satisfaction
0.12
acknowledge-without-open

All three components reported together, so a gain in one that was bought from another is visible.

Don't
Efficiency
41s
median time to acknowledge

Time alone. Optimising this rewards acknowledging without reading — the disengagement the system exists to prevent.

Clinical safety notes

How this page is used

This page is the context statement that every other page's guidance depends on. Trace it from the context-of-use section of your IEC 62366-1 usability engineering file.

NotJustAnyMed.Tech Design System · Usability & context of use · v1.0 · draft for review
Structured on the concepts of ISO 9241-11:2018. Terms are paraphrased for this context; consult the standard itself for authoritative definitions.