Clinical dashboard vs patient portal: the same data, designed twice

15 June 2026

Patient portal & EHR Clinical vs Patient Interfaces Data Visualisation & Health Metrics

The same lab result reaches a clinician scanning a dozen patients and a patient opening a portal alone at home. One dataset, two different design problems.

A single blood-test result sits in one row of a laboratory system, but it does two different jobs depending on who opens it. A clinical dashboard supports work: scanning, deciding and acting across many patients in a shift. A patient portal supports understanding: what this means for me, whether it is serious, what to do next. Designing both from the same underlying data is not one problem twice, it is two problems that happen to share a source.

Two audiences, two jobs

A clinician meets a result between many other tasks, one entry among dozens that day. A patient often opens it alone, sometimes before anyone has explained it. The NHS service manual's design principles ask services to consider people's emotional, physical and technical needs and the wider context of use - and a lab result read alone at home, possibly while anxious, is a very different context from a lab result reviewed in a shift-handover list. The data can be identical; the questions asked of it are not.

What the clinician's screen has to do

Clinical interfaces are dense on purpose and frequently interrupted, and that combination has a measurable cost. A 2022 stewardship review in Applied Clinical Informatics reports acceptance rates for interruptive pop-up alerts as low as 4 to 11 percent, and argues that poor alerts erode trust in all of them, not just the bad ones. A 2026 systematic review of 22 studies found alert fatigue measured inconsistently across the field, and only 3 of the 22 studies reported whether the alerts staff overrode were actually appropriate to override. The design implication is not "fewer alerts" as a slogan, but a discipline: every alert has to earn a place that interrupts a clinician's attention.

Labelling and navigation carry as much weight as alerting. In a study of a tablet interface for remotely monitoring vital signs in a hospital care unit, 10 staff members found 37 usability problems in the original design, mostly unclear labels and confusing navigation; after a redesign, mean scores on the System Usability Scale (a standard 0-100 questionnaire) rose from 68.5 to 89. The sample is small, but the direction is consistent with the stewardship literature: clarity in a clinical interface is a safety property, not a polish item.

What the patient's screen has to do

Patient portals have historically reused the clinician's table - the same reference range, the same abbreviations - and handed it to someone with none of a clinician's training. A 2021 JMIR Human Factors study surveyed 203 people, interviewed 13 patients and tested a prototype pairing a graphic with a plain-language takeaway message. Its pilot of 8 participants is too small to prove the design works, but every participant rated it an improvement on the existing table-only portal, and some flagged that even mild phrasing such as "not optimal" could cause needless anxiety - a reminder that word choice is part of the interface, not separate from it.

A worked example: one result, two screens

Take a result outside its reference range (the span a laboratory treats as typical). The clinician's view is a compact, sortable list across many patients, flagged and trended, ordered by what needs attention first. The patient's view uses the plain name of the test, shows where the value sits against the range, explains carefully what "outside the range" does and does not mean on its own, and states who to contact. Neither view is a simplified or a technical version of the other - each is built for a different task, from the same number.

Why the timing changed the problem

Both countries this site tracks have moved toward patients seeing results earlier, sometimes before a clinician has reviewed them, which removes an assumption a lot of older portal design quietly relied on: that a conversation had already happened. In the United States, the federal rule implementing the 21st Century Cures Act requires that patients can access their electronic health information at no cost and restricts practices that delay that access, with nine narrow exceptions. In Australia, most pathology reports in My Health Record became viewable by patients as soon as they were uploaded from October 2025, replacing a 7-day delay (some categories such as anatomical pathology and genetic tests keep a 5-day delay); clinicians see everything immediately regardless. A patient portal built on the old assumption - that a clinician has already called - now has to stand alone as the first, sometimes only, interpreter in the room.

Limits of the evidence Several of the studies cited here are small, single-site or self-reported. They show patterns worth designing for, not settled rules. This article is a design education resource, not clinical guidance.

What this means for design teams

Build two views from one data model, not one view with a toggle. A clinician list and a patient explanation rarely share a layout even when they share a source.
Treat alert design as a budget, not a feature list. Each interruptive alert on a clinical dashboard should be justified against the acceptance-rate evidence above, not added by default.
Write the patient-facing explanation to stand alone. Since a result can now reach a patient before a clinician has reviewed it, the interface itself, not a follow-up phone call, may be the first explanation a patient gets.
Reuse a remote-monitoring dashboard's ranking logic for the clinician view: order by what needs attention, not by patient list order or chart date.