Emergency Department Crowding. Objective Modelling based on Workload

Abstract: Emergency Departments (ED) have a central role in modern healthcare, providing emergent care regardless of complaint 24/7. However, EDs are often characterized by long waiting times for patients and a stressful working environment for staff. These are features or a resource and demand mismatch, internationally known as ED crowding. Although several reports and news articles have indicated that ED crowding is a problem in Sweden, there is no systematic work to assess it, nor a validated method to measure it. These things are essential to understand both the severity and extent of the problem, and to evaluate initiatives to alleviate crowding. With this thesis, I aim to begin this process by looking at different aspects of crowding assessment.The thesis is built upon four studies, each providing a different perspective on how to measure ED crowding. In the first paper we derived a model that can measure crowding, defined by ED staff, based on data from the digital information system in the ED. The model was derived in 5 EDs in the county of Skåne in Sweden and validated in 2 of these EDs. We propose a model that includes the number of patients in the ED, their waiting times and acuity, and that shows promising ability to measure crowding both in the derivation and validation; The Skåne Emergency Department Assessment of Patient Load (SEAL) model. Paper II investigates the prevalence of crowding and boarding, ie when patients are waiting for an inhospital bed in the ED, on a national level in Sweden. The results suggest that crowding is prevalent on a national level in Sweden, with 37% of the survey EDs reporting high occupancy rate during the 24 hour study period. Based on the data collected in paper II, the third paper in the thesis explores the relation between crowding, assessed by staff, and the ratio of patients to treatment beds, also known as the Occupancy Rate. The analysis in paper III indicates that high occupancy rates may not predict crowding as assessed by staff equally between EDs, and that crowding may be influenced by the organisation of the ED. This highlights the importance of systematic measurements of crowding adjusted for ED-specific features, like number of treatment beds. The last paper in the thesis examines the association between crowding and mortality. The results indicate an association between high levels of crowding, measured by a modified SEAL model, and an increased 7-day mortality, confirming that ED crowding is a real threat to the safety of our patients.This thesis indicates that ED crowding is prevalent on a national level in Sweden and associated with increased mortality for our patients. We would therefore strongly encourage a systematic assessment of crowding as an essential part of the regular quality insurance work, both locally and on a national level. Given the complexity of both crowding and the care in the ED, the assessment should likely include multiple different measures. We suggest using the modified SEAL model since it is a validated measure of crowding with good predictive value and an ability to identify situations where crowding is associated with increased mortality. Further studies should focus on predicting imminent crowding and on methods to reduce its impact on patients and staff.