Reducing a voidable admissions through the Emergency Department at Massachusetts General Hospital

Abstract

Despite efforts to address capacity constraints with a massive expansion less than five years ago, the Emergency Department (ED) at Massachusetts General Hospital (MGH) is again displaying consistent and serious symptoms of overreacting, including rising patient wait times and routine activation of capacity-related emergency management protocols. As MGH grapples with these challenges, it is imperative to understand precisely what is driving the congestion. In this thesis, will show there has been significant volume growth and ii) study whether these visits resulted in inpatient admissions that could have utilized alternative care pathways while preserving patient safety and quality of care. After collaborating with hospital staff to analyze ED patient volume in 2015, we conclude that avoidable admission candidates who transferred to MGH from other facilities occupied nearly 6 percent of the hospital's General Medicine capacity. Furthermore, the utilization growth associated with these patients was equivalent to 1.3 percent of all General Medicine beds. meaning transfers alone can account for the overcrowding symptoms mentioned above. In a second analysis. applying unsupervised and supervised learning methods to short-stay inpatients reveals that even generalized order data can reliably predict conditions associated with avoidable admissions. Building on this insight, we then develop a scoring method to identify avoidable admission candidates without requiring manual case review by a physician.