Characteristics of Restraint Use on Inpatient Medical/Surgical and Psychiatric Units at a Tertiary Care County Hospital

Background

Racism contributes to worse patient care and health outcomes in the USA. Disparities in restraint application have been documented in the emergency department context, yet inpatient medical/surgical and psychiatric settings are less studied.

Objective

The primary objective of this analysis is to determine factors associated with restraint use within the inpatient medical/surgical and psychiatric populations at a tertiary, academic medical center. We predict that Black, American Indian, Native American, or Alaska Native, or Hispanic/Latinx racialized identity, age, presence of medical/cognitive disorders, substance use disorders, and psychiatric illnesses will be associated with higher odds of restraint use (H1), of locked restraint use (H2), and of seclusion (H3), and with more time spent in restraints (H4).

Design

Logistic regression models were used to predict odds of restraint, odds of locked restraint, and odds of seclusion, while linear regression models were used to predict time spent in restraints.

Participants

We included admitted inpatients greater than 17 years, discharged between January 1, 2018, and December 31, 2022, at an urban, Level I academic trauma center who stayed at least one night in a medical/surgical or psychiatry unit.

Main Measures

Four outcome variables included restraint, locked restraint, seclusion, and time spent in restraints. Other key variables included race/ethnicity, age, substance use, medical/cognitive disorders, psychiatric illness, sex, interpreter use, insurance status, being recently unhoused, and having a history of violence.

Key Results

For medical/surgery patients, higher odds of restraint use and of locked restraints were found for American Indian, Native American, or Alaska Native patients and Latinx/Hispanic patients are restrained for nearly 30 h more than white patients. For psychiatry patients, Black patients had higher odds of seclusion, but this significant association disappeared when covariates were added.

Conclusions

This is the first paper to examine differences in time spent in restraints and the type of restraint used, suggesting that variability exists in how racialized inequities appear across dimensions of restraint.