We performed a retrospective cohort study that included all non-ICU, non-surgical patients ≥50 years-old hospitalized at a quaternary academic hospital from 06/01/2023 to 05/31/2024. Demographic/clinical data were collected from the electronic medical record, including results of delirium screening (Nursing Delirium Screen Scale) systematically administered by nursing staff. Primary outcomes included prevalent (on admission) and incident (hospital-acquired) delirium. Secondary outcomes included safety attendant and restraint use, and in-hospital mortality. Data were analyzed using non-parametric tests and multivariable regression models (Stata v18.0).

4249 subjects were included, with 948 (22%) subjects identifying as NELP. Compared with those with English preference, those with NELP were older and more likely to be women. In unadjusted analyses, subjects with NELP had higher prevalent delirium (16.2% vs 8.8%), incident delirium (12.2% vs 9.4%), restraint use (4.7% vs 3.1%), safety attendant use (16.8% vs 10.9%), and in-hospital mortality (4.5% vs 1.2%), all with p-values <0.05. Multivariable logistic regression models adjusting for age, sex, and Charlson Comorbidity Index (CCI) revealed NELP was a statistically significant predictor of prevalent delirium (OR 1.32, 95% CI 1.05–1.65) and mortality (OR 3.24, 95% CI 2.05–5.11), but not incident delirium, restraint use, or safety attendant use. The association between NELP and mortality remained after also adjusting for any delirium (OR 3.14, 95% CI 1.97–4.99).

NELP was a significant predictor of prevalent delirium and mortality in hospitalized patients. This study highlights possible language-related disparities in healthcare outcomes and could serve to guide interventions to improve inpatient delirium care.