Quick Summary
A paper in npj Digital Medicine makes a clean distinction most EHR debates blur: system usability primarily reduces extraneous cognitive load, while data usability increases germane cognitive load. In the paper’s words, EHRs were “designed to enhance clinical decision-making,” yet “the way data are organized and displayed can create significant cognitive demands for physicians.”
The paper, directly
The research is titled When Better Data Meets Better Design: How EHR Data Usability and System Usability Shape Physicians’ Cognitive Load. The authors study how EHR data usability (data quality, data completeness, data-driven use) and system usability jointly shape physicians’ cognitive load, using survey responses from 564 physicians across 32 specialties.
In the abstract, they summarize the core finding with unusual clarity: “Findings show that stronger data usability increases germane cognitive load,” while “higher system usability reduces extraneous cognitive load by aligning interface design with clinical workflow and minimizing navigation-related effort.”
Two kinds of cognitive load, and why it matters
Most EHR conversations treat cognitive load as one thing to reduce. The paper splits it into two different categories, which is the point.
- Extraneous cognitive load is wasted mental effort created by interface friction: hunting, clicking, switching screens, re-entering information, recovering context.
- Germane cognitive load is productive mental effort: the reasoning work that supports interpretation, synthesis, and diagnosis.
Quotable line: Better screens save effort. Better data improves thinking.
The authors are explicit about the “two levers for improving cognitive performance”: “enhancing system usability lowers unnecessary cognitive effort and documentation-related errors, while improving data usability supports reasoning-intensive diagnostic work.”
Information overload is not just volume
The paper also puts a sharp definition on overload. They argue cognitive overload is not a database problem. It is a workflow and interface problem: “Cognitive overload does not stem from the underlying database architecture, but rather from interface complexity, navigation burden, and misalignment between the system workflows and clinical reasoning processes.”
Information overload partially mediates the effects. In the abstract: better data usability helps physicians “filter irrelevant data and stay focused on diagnostically relevant cues.”
Quotable line: The enemy is not data. The enemy is unfiltered data.
The most uncomfortable line in the introduction
The paper cites prior work showing how bad usability can get at scale: a national study reported a mean EHR usability score of 45.9 on the System Usability Scale, “equivalent to a failing grade (‘F’).” It also notes that “the difficulty of using EHR systems during clinical work has even been likened to texting while driving.”
What builders should take from this
This paper is a warning against one-dimensional fixes.
- If you improve UI polish but ignore data usability, you reduce pain but you do not strengthen reasoning.
- If you improve data quality but leave interface friction untouched, the system still burns attention.
- If you do both, you get balance: less wasted effort, more meaningful interpretation.
The authors describe this as a “rebalancing of cognitive load rather than a reduction in its overall magnitude,” where higher-quality, context-relevant data increases germane load while usability improvements reduce extraneous load.
Quotable line: Reduce the effort that does not matter, so the effort that matters can happen.
The Aether angle: longitudinal care needs longitudinal usability
A longitudinal view of health is only as useful as the cognitive experience it creates. A timeline that forces doctors to hunt, scroll, and reconcile duplicates is not a timeline, it is a stress test.
The paper points to practical levers that match how we think about the problem at Aether: reduce navigation steps, collocate key diagnostic information in a single view, suppress duplicate entries, and add provenance indicators to signal reliability.
Read the paper here: https://www.nature.com/articles/s41746-025-02243-4