Background: Health literacy (HL) strongly influences patient outcomes (1), yet many hospitals rely on rapid screening tools such as the Rapid Estimate of Adult Literacy in Medicine–Short Form (REALM-SF), which assesses medical word recognition but not comprehension (2). The Joint Commission’s 2022 Health Equity Standards require hospitals to identify and address social needs, including literacy, to promote equitable care delivery (3). As with other mandatory screenings, institutions may defer to quick, single domain tools to meet documentation requirements. However, patients who can accurately pronounce medical terms may still lack the comprehension needed for safe communication and adherence. This study evaluated whether REALM-SF word-recognition scores predict comprehension among hospitalized adults.
Methods: This prospective cross-sectional study quantified the association between medical word recognition and comprehension using a Poisson regression framework. A total of 466 hospitalized adults completed the REALM-SF, reading seven medical terms aloud and providing definitions for each. Comprehension scores (0–7) were modeled as a count outcome, with word-recognition score as the primary predictor. Covariates included age, race, gender, and ethnicity. Since the Pearson chi-square statistic (χ² = 0.78) indicated under-dispersion, a quasi-Poisson model was used. Model fit was evaluated with McFadden’s pseudo-R².
Results: Participants had a mean age of 66.8 years; 52.8% were female. The cohort was 80.9% White, 10.3% Black, 7.5% Hispanic, and 8.8% other race/ethnicity. The quasi-Poisson coefficient for word recognition was β = 0.260 (incidence-rate ratio [IRR] = 1.297, 95% CI 1.206–1.394, p < 0.001), indicating each one-point increase in word recognition was associated with a 29.7% higher expected comprehension score. McFadden’s pseudo-R² = 0.035 demonstrated minimal variance explained. Among demographic covariates, only female gender predicted higher comprehension (β = 0.098, IRR = 1.103, p = 0.038). Although 81% of participants read all seven words correctly, only 4.5% accurately defined all terms.
Conclusions: Although REALM-SF word-recognition scores were significantly associated with comprehension (IRR = 1.297, p < 0.001), the relationship was weak, with reading ability explaining only 3.5% of the variance in understanding. Despite 81% of patients reading all seven medical words correctly, only 4.5% could define all terms, demonstrating that word recognition substantially overestimates true comprehension. These findings indicate that REALM-SF failed to identify most patients with comprehension deficits. Continued reliance on such tools may leave hospitals noncompliant with Health Equity Standards and contribute to preventable communication failures. Incorporating comprehension-focused strategies, such as teach-back and plain-language discharge materials, is essential for accurately identifying at-risk patients and improving safety and equity in hospital care.
.jpg)
.jpg)