# Characterizing Bias and Care Disparities with Physical Restraint Use in the Emergency Setting Using Natural Language and Cognitive Data

> **NIH NIH R21** · YALE UNIVERSITY · 2023 · $209,375

## Abstract

PROJECT SUMMARY
Agitation is defined as excessive psychomotor activity leading to aggressive and violent behavior in patients.
Those presenting with agitation in the emergency setting represent the most marginalized populations.
Coercive measures like physical restraints are currently used routinely on agitated individuals, but are
associated with physical trauma, respiratory depression, and death. Recent studies have shown
disproportionate use of physical restraint on Black patients, those who are homeless, and those with public or
no insurance. Identification of specific interpersonal and structural factors that affect heuristics and decision-
making of healthcare workers regarding restraint use is needed to mitigate systemic bias and discrimination
against these marginalized patients. However, current research is limited to analyzing structured quantitative
data elements, while narrative text better reflects sociocultural contexts, interpersonal interactions, and
clinician thought processes. Natural language processing is an informatics discipline that can parse free text
within clinical notes into quantifiable variables on a large scale that can be combined with mediation analysis to
uncover factors leading to disparities in restraint use. A complementary tool is cognitive task analysis, which
uses qualitative methods to understand how mediators of bias affect clinical decision-making at the bedside.
Our overall objective is to use the combination of these innovative analytical methods to overcome deficiencies
of standard health service research methods in identifying individual, interpersonal, institutional, and systems
factors leading to bias in physical restraint use. In Aim 1, we will use natural language processing and
mediation analysis on a large database of emergency department clinical narrative notes across our regional
healthcare system to extract and identify candidate variables that lead minority populations to increased risk of
physical restraint. This will allow us to verify and test potential factors within our newly derived conceptual
model of bias during restraint use that predict discriminatory practices. In Aim 2, we will use cognitive task
analyses through qualitative interviews and video-informed focus groups with emergency healthcare workers to
characterize drivers and cues that influence decision-making and heuristics against minority populations. This
will complement the results from Aim 1 by providing explanatory models for how interpersonal and structural
factors that lead to bias are manifested at the bedside.
This proposed work will make a positive contribution to minority and health disparities research in the
emergency setting by identifying specific interpersonal and structural factors mediating bias and discrimination
against minority and socioeconomically disadvantaged individuals with behavioral emergencies. Our study is
exploratory and novel as it combines innovative and multidisciplinary approaches from tw...

## Key facts

- **NIH application ID:** 10633167
- **Project number:** 5R21MD017327-02
- **Recipient organization:** YALE UNIVERSITY
- **Principal Investigator:** Ambrose H Wong
- **Activity code:** R21 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2023
- **Award amount:** $209,375
- **Award type:** 5
- **Project period:** 2022-06-02 → 2025-06-30

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/10633167

## Citation

> US National Institutes of Health, RePORTER application 10633167, Characterizing Bias and Care Disparities with Physical Restraint Use in the Emergency Setting Using Natural Language and Cognitive Data (5R21MD017327-02). Retrieved via AI Analytics 2026-05-22 from https://api.ai-analytics.org/grant/nih/10633167. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*