STAFF BOARD

CHAIRMAN
Maruxa Hernández Corredoira
EDITOR IN CHIEF
Manuela Velázquez Prieto
Tomás Casasín Edo
MARKETING MANAGER
EDITORIAL BOARD
María B. Badía Tahull
Lluís Campins Bernadas
Juan Carlos Juárez Giménez
Carles Quiñones Ribas

Fixing the “No Milk” Problem in Medication Labels with a Randomized Eye-Tracking Study of Negation and Food-Interaction Pictograms

Elfriede Thomas Bernhard1, Dikshit Ishita1
1 Medical University of Graz
Corresponding Email: thomaselbernhard@yahoo.com

Background: Patients with limited literacy remain vulnerable to medication errors, especially for instructions that involve negation and food/drug interactions. Pictograms are widely used, but guidance on which visual grammar works best is limited. Methods: We ran a three-arm, randomized experiment with low-literacy adults (N=360) assigned between subjects to one of three icon grammars: A) ISO-style black/white with red slash; B) photo-icon composites with overlaid symbols; C) culturally localized line-art with explicit red “X”/green tick. Each participant judged 12 common label instructions enriched for historically difficult items. The primary outcome was immediate, item-level correct comprehension (binary). Secondary outcomes were response time, confidence, and eye-tracking metrics (time-to-first-fixation, dwell time, scanpath entropy). Analyses used mixed-effects logistic/linear models with participant and item random intercepts (intention-to-treat). Results: Grammar C outperformed A and B on the primary endpoint (overall marginal risk difference vs. A: +10 percentage points, 95% CI +7 to +13; vs. B: +7 pp, by subtraction). Grammar B exceeded A by +3 pp (95% CI 0 to +6). The largest gains for Grammar C occurred on high-risk items: “Not by mouth” (A 62%, B 68%, C 78%) and “Do not take with dairy” (A 55%, B 60%, C 72%); timing/handling items also improved (e.g., “Every 8 hours”: A 69%, B 73%, C 81%). Eye-tracking corroborated these differences: median time-to-first-fixation on critical regions was 1.42 s (A), 1.31 s (B), and 1.08 s (C), indicating more efficient visual search under Grammar C. Results were consistent across literacy strata and languages; no subgroup showed harm. A 48-hour retention subset (n=210) maintained the C > B > A pattern. Conclusions: How pictograms are drawn matters. Culturally localized line-art with explicit negation cues produced faster and more accurate comprehension, especially for negation and interaction instructions. We provide a practice checklist—use explicit negation near the action/object; pair action+object+context; prefer simplified line-art—that can be embedded in label-printing systems without added counseling time. Future pragmatic trials should link these comprehension gains to adherence, error reduction, and cost outcomes.

Keywords: Pictograms; Medication Labeling; Health Literacy; Eye-Tracking; Human Factors; Visual Design