Browsing by Subject "Instrumental variable"

Now showing 1 - 1 of 1

The impact of crowding in emergency departments on patient health and health outcomes : an instrumental variable analysis
(2017-01-09) Kao, Emily Chih-Wen; Olmstead, Todd; Heinrich, Carolyn J.; Angel, Jacqueline L; Hayward, Mark D; Milling, Jr., Truman J; Richardson, Samuel S; Stolp, Chandler W
It is commonly presumed that emergency department (ED) crowding results in the delivery of inferior quality health care and deleterious effects for acute patients. Yet a growing body of literature assessing outcomes for patients treated under crowded conditions has yielded inconsistent results. This dissertation identifies whether and to what extent crowding in hospital-based EDs has a causal impact on health outcomes. The conflicting evidence stems from both conceptual and empirical weaknesses in the crowding literature. Conceptually, no universal definition of crowding exists. Although most frequently defined as the incapacity to meet demand for ED services, there is a lack of consistency in how capacity is related to crowding (e.g., number of arrivals, occupants, proportion of unserved patients, etc.). The wide variation in conceptualization contributes to the use of myriad proxies to estimate crowding’s effect and to a rudimentary understanding of its impact. Additionally, studies measure an assortment of outcomes with discord among researchers over the most appropriate to assess causality. Policy and program recommendations have been advanced to mitigate crowding based on the perception that it is a problem, but evidence-based research reveals incongruency in the findings. Furthermore, empirical studies have not identified any more than an associative crowding-outcomes relationship, challenging inferences made about crowding’s true effects. This is primarily due to insufficient controls for endogenous variables, contributing to biased estimators. The obscurity in whether crowding affects patient health fuels debate about the underlying issue, consequences thereof, and efficacious and cost-effective strategies to address it. This dissertation aims to: (1) offer a new, standardizable tool to define and more accurately measure the degree of crowding; and (2) assess the relationship between ED crowding and health outcomes using a new construct of crowding – “patient arrival density,” to instrument for ED occupancy rate. This empirical study employs an instrumental variable estimation approach to evaluate the effect of crowding on probability of mortality, severe sepsis, diabetic ketoacidosis, and subsequent ED visits. Findings show crowding has no causal effect on any of these outcomes, suggesting resources targeted at reducing crowding in EDs may be better directed at identifying the true mechanisms that can lead to sentinel events. By taking a novel approach, this dissertation advances existing research and contributes to our knowledge of the crowding phenomenon, which subsequently can help better direct resources and future research aimed at improving the delivery of care in the acute healthcare system.