JMIR Publications

ABSTRACT

Background:

In critical care units (CCUs), clinically important patient information is often fragmented across screens and text-heavy notes. Visual notes may reduce the cognitive work of reviewing narrative nursing notes, but direct comparisons with conventional written notes are limited.

Objective:

This pilot study aimed to compare critical care nurses' perceived workload after reviewing patient information in a conventional written note format versus an investigator-developed visual note format using the National Aeronautics and Space Administration Task Load Index (NASA-TLX). We hypothesized that the visual format would be associated with lower total workload and lower subscale scores than the written format.

Methods:

We conducted a single-site, within-subject, nonrandomized fixed-order study at an urban medical center in the northeastern United States between November 2025 and February 2026. Eligible participants were registered critical care nurses working in 2 critical care units. Each participant reviewed the same deidentified intensive care unit nursing subjective, objective, assessment, and plan (SOAP) notes for a patient, obtained from the PhysioNet MIMIC critical care text corpus, in 2 sequential formats: first as a series of conventional written notes (WN) and then as an investigator-developed visual notes (VN) display. The VN used a large language model-supported structuring process to convert the narrative SOAP content into a structured chart. Subjective workload was measured after each condition using the raw NASA-TLX. Paired comparisons were performed using Wilcoxon signed-rank tests.

Results:

A total of 41 nurses completed paired observations. Mean total NASA-TLX score decreased from 54.95 in the written condition to 30.68 in the visual condition, corresponding to a mean reduction of 24.27 points (44.2%; P<.001). The visual format was associated with lower workload on all 6 NASA-TLX subscales: mental demand (mean difference 5.44; P<.001), physical demand (mean difference 2.41; P=.020), temporal demand (mean difference 4.76; P<.001), performance-related workload (mean difference 1.39; P=.041), effort (mean difference 4.63; P<.001), and frustration (mean difference 5.63; P<.001). The largest relative reductions were observed for frustration (55.7%), temporal demand (49.6%), physical demand (48.8%), and mental demand (44.7%). Internal consistency of the NASA-TLX subscales was good in both conditions (Cronbach's α = 0.879 for written and 0.830 for visual).

Conclusions:

Review of the VN was associated with a lower perceived workload than conventional written SOAP notes (WN). This pilot study used a single patient's deidentified CCU notes and a fixed-order design in which all the nurses reviewed the WN before the VN. Therefore, findings should be interpreted as preliminary pilot data. A lower workload with VN may have reflected the order effect due to familiarity with the patient case, given that the nurses reviewed the WN first. Further studies should use multiple patient cases and randomized or counterbalanced designs to better evaluate the usability and clinical relevance of VN displays for nursing priorities.