402 The Effect of Patient Demographics on the Odds of Restraint Use for Agitation in the Emergency Department

Abstract

Agitated patient encounters in the Emergency Department (ED) are on the rise, with a recent estimate of 1.7 million events per year and a 2.6% prevalence. To protect staff and prevent self-harm, physical restraints are sometimes used. However, these are associated with safety risks and the potential for stigmatization of a vulnerable population. We aim to determine factors that are associated with odds of being restrained in the ED. We conducted a retrospective cohort analysis of all patients (≥ 18 yo) placed in restraints during an ED visit to three hospitals within a large tertiary health system from Jan 2013-Aug 2018. We undertook descriptive analysis of the data and created a generalized linear mixed model with a binary logistic identity link to model restraint use and determine odds ratios for various clinically significant demographic factors. These include sex, race, ethnicity, insurance status, alcohol use, illicit drug use, and homelessness. Our model accounted for patients nested across the three EDs and also accounted for multiple patient visits. In 726,417 total ED visits, 7,090 (1%) had associated restraint orders. Restrained patients had an average age of 45, with 64% male, 54% Caucasian and 29% African American. 17% had private insurance, 36% endorsed illicit substances, 51.4% endorsed alcohol use and 2.3% were homeless. Using logistic regression, we found that African Americans had statistically significant odds of being restrained compared to Caucasians with adjusted odds ratio (AOR) of 1.14 (1.08,1.21). Females (AOR 0.75 [0.71, 0.79] had lower odds of being restrained compared to their male counterparts while patients with Medicaid (AOR 1.57 [1.46, 1.68]) and Medicare (AOR 1.70 [1.57, 1.85]) had increased odds compared to the privately insured. Furthermore, illicit substance use (AOR 1.55 [1.46, 1.64]), alcohol use (AOR 1.13 [1.07, 1.20] and, homelessness (AOR 1.35 [1.14, 1.16], portended increased odds of restraint use. To our knowledge, this is the first study of its kind to show statistically significant effects of patient demographics on odds of restraint use in the ED. The increased odds based on race, insurance status, and substance use highlight the potential effects of implicit bias on the decision to physically restrain patients and underscores the importance of tools that facilitate a more objective assessment of these patients. Using such tools could assist providers in avoiding further marginalization of an already vulnerable population.