When Language Becomes a Safety Risk: English Proficiency and Hospital Restraint Practices

Highlights

• In a large retrospective cohort of 132,767 hospitalizations, limited English proficiency (LEP) was associated with higher adjusted odds of physical restraint or antipsychotic use.
• The disparity persisted despite similar unadjusted event rates between LEP and non-LEP groups, suggesting confounding by illness severity and hospitalization characteristics.
• Among patients with delirium, the association was stronger, underscoring the importance of communication in a syndrome that depends heavily on reorientation and nonpharmacologic management.
• The findings raise concern that language barriers may contribute to potentially avoidable escalation to restraints or antipsychotics, both of which carry recognized harms.

Study Background

Physical restraints and antipsychotic medications are generally considered last-resort interventions in hospitalized patients who are at immediate risk of harming themselves or others. In modern inpatient care, first-line management for agitation, confusion, and unsafe behavior emphasizes identifying reversible causes, providing environmental support, and using verbal de-escalation and reorientation. These approaches are especially important in delirium, where communication, reassurance, and repeated orientation can reduce distress and prevent escalation.

Patients with limited English proficiency may be disproportionately vulnerable in this setting. When clinicians and patients cannot communicate easily, subtle symptoms may be missed, distress may be misinterpreted as aggression, and opportunities for de-escalation may be lost. Interpreter access can mitigate these risks, but in real-world hospital workflows it may not always be rapid, consistent, or sufficiently integrated into urgent bedside decision-making. This study addressed an important and underexplored question: whether English proficiency is associated with restraint and antipsychotic use during hospitalization.

Study Design

This was a retrospective cohort study conducted at a single academic medical center. The investigators examined adults discharged between 1/2019 and 6/2023. Limited English proficiency was defined using the electronic medical record designation of non-English primary language. The primary outcome was a composite of either physical restraint use or antipsychotic use during hospitalization. Secondary analyses evaluated each component separately, and a subgroup analysis focused on patients diagnosed with delirium.

Physical restraints were identified through orders, and antipsychotic exposure was identified through pharmacy charges. To reduce confounding, the investigators used multivariable generalized estimating equations controlling for demographic, hospitalization, and clinical characteristics. This approach is appropriate for repeated hospitalizations and allows estimation of adjusted odds while accounting for correlated observations.

Key Findings

The cohort included 132,767 hospitalizations, with a mean age of 62.9 years; 48.3% were female, and 10.8% were classified as LEP. The overall incidence of either physical restraints or antipsychotic use was 11.2% (14,802 hospitalizations).

At the unadjusted level, outcome rates were similar between LEP and non-LEP groups. For the composite outcome, the rates were 10.1% versus 11.3%, respectively. Physical restraint use occurred in 7.9% of LEP hospitalizations and 8.9% of non-LEP hospitalizations. Antipsychotic use was identical at 4.9% in both groups. These crude comparisons could suggest no disparity, but they likely reflect differences in patient mix and illness severity across language groups.

After adjustment, LEP was associated with higher odds of the composite outcome. The adjusted odds ratio (aOR) was 1.18 (95% CI, 1.08-1.28) for either physical restraint or antipsychotic use. When analyzed separately, LEP remained associated with physical restraint use (aOR 1.30; 95% CI 1.17-1.43) and with antipsychotic use (aOR 1.12; 95% CI 1.01-1.25).

The most clinically informative result came from the delirium subgroup. Among patients with diagnosed delirium, the association between LEP and the composite outcome was stronger: aOR 1.42 (95% CI 1.22-1.65). This is biologically and clinically plausible. Delirium management depends heavily on frequent verbal reorientation, assessment of orientation and attention, calming communication, and family engagement when possible. Language discordance can directly interfere with each of these elements.

Although the effect sizes were modest, they are meaningful because the exposure and outcomes are common and the harms are nontrivial. Physical restraints can cause injury, pressure damage, agitation, loss of dignity, and worsening delirium. Antipsychotics may be associated with sedation, extrapyramidal effects, QT prolongation, falls, and other adverse outcomes, particularly in older adults and medically ill patients. Even a small relative increase may translate into a substantial absolute burden across large inpatient populations.

Interpretation and Clinical Context

This study suggests that language barriers may function as a patient safety issue, not merely a communication inconvenience. The finding that adjusted disparities emerged despite similar unadjusted rates is noteworthy. It indicates that LEP patients may be systematically different in ways that obscure the underlying association in crude analyses, and that the true relationship may only become visible after accounting for confounders such as diagnosis burden, severity of illness, and hospitalization characteristics.

The stronger association in delirium deserves particular emphasis. Delirium is common in hospitalized older adults and is often triggered or worsened by unmet needs, pain, sleep disruption, unfamiliar surroundings, and sensory impairment. Because patients with delirium may already have impaired attention and perception, the added burden of language discordance can make bedside assessment even more difficult. If staff cannot reliably understand the patient’s concerns, or if the patient cannot understand instructions, nonpharmacologic management becomes harder to deliver effectively.

From a systems perspective, the findings point to several potential mechanisms. First, inadequate access to professional interpreters during urgent behavioral events may reduce the success of de-escalation. Second, communication gaps may lead to misclassification of agitation or noncompliance as dangerous behavior. Third, time pressure and staffing constraints may make restraint or medication appear like the fastest available option. Finally, implicit bias and differential thresholds for escalation may also play a role, although this study cannot establish that directly.

Strengths

Several features strengthen the credibility of the findings. The sample size was large, the clinical setting was real-world inpatient care, and the investigators evaluated both a composite endpoint and the component outcomes. The use of adjusted models helped address confounding, and the delirium subgroup analysis added important clinical specificity. The study also addresses an area with clear policy relevance: communication equity in high-risk hospital care.

Limitations

As with all retrospective single-center studies, causality cannot be inferred. LEP was defined by primary language in the electronic medical record, which may misclassify some patients with functional English proficiency or incomplete documentation. The study also relied on orders and pharmacy charges rather than direct bedside observation, so it may not capture actual restraint duration or whether antipsychotics were administered as intended. Residual confounding is likely, particularly if LEP patients differed in unmeasured severity markers, cognitive status, social support, or frequency of interpreter use.

Generalizability is limited by the single academic center design. Language diversity, interpreter infrastructure, staffing patterns, and restraint policies may differ across institutions. The study also does not identify which non-English languages were involved, nor does it measure the timeliness, quality, or frequency of professional interpreter services. These are crucial variables for understanding why the disparity occurs and how to fix it.

Expert Commentary

Current best practice in delirium and agitation management emphasizes nonpharmacologic measures whenever possible, including reorientation, mobilization, sleep promotion, hydration, vision and hearing support, pain control, and family engagement. Professional interpreter services are central to many of these interventions. From a quality-improvement perspective, the study supports viewing interpreter access as part of inpatient safety infrastructure rather than an optional service.

The findings also align with broader concerns about inequities in hospital communication. LEP has been associated with differences in informed consent, patient experience, length of stay, and adverse events in prior literature. This study extends that concern into a particularly sensitive domain: coercive interventions used when patients are perceived as unsafe. Because restraints and antipsychotics can themselves worsen outcomes, language-related disparities here may have downstream clinical consequences.

Future studies should examine whether improved interpreter availability, structured delirium bundles, rapid-response communication pathways, or culturally tailored de-escalation protocols reduce restraint and antipsychotic use. It would also be valuable to study specific language groups, unit-level practices, and the role of family interpreters versus professional interpreters. Interventional research is especially needed, since identifying disparity is only the first step toward reducing it.

Conclusion

This single-center retrospective study found that hospitalized patients with limited English proficiency had higher adjusted odds of physical restraint or antipsychotic use, with a stronger association in those with delirium. The results suggest that language barriers may contribute to potentially avoidable escalation in inpatient care. Because both restraints and antipsychotics carry important risks, hospitals should treat language access as a patient safety priority and investigate modifiable drivers of this disparity.

Funding and clinicaltrials.gov

Funding information and clinicaltrials.gov registration were not reported in the source abstract provided. This was a retrospective observational study and does not appear to have a clinicaltrials.gov registration.

References

1. Borczuk R, Wilson L, Anderson TS, Herzig SJ. Disparities in Physical Restraints and Antipsychotic use in Hospitalized Patients with Limited English Proficiency. J Gen Intern Med. 2026-06-08. PMID: 42260183.

2. Inouye SK, Westendorp RGJ, Saczynski JS. Delirium in elderly people. Lancet. 2014;383(9920):911-922.

3. American Geriatrics Society Expert Panel on Postoperative Delirium in Older Adults. American Geriatrics Society abstracted clinical practice guideline for postoperative delirium in older adults. J Am Geriatr Soc. 2015;63(1):142-150.

4. Brummel NE, Girard TD. Preventing delirium in the intensive care unit. Crit Care Clin. 2013;29(1):51-65.

5. The Joint Commission. Advancing Effective Communication, Cultural Competence, and Patient- and Family-Centered Care: A Roadmap for Hospitals. 2010.