What we have learned from a decade of ED crowding research.

Abstract

In this issue of Academic Emergency Medicine, two articles explore the link between emergency department (ED) crowding, quality of care, and outcomes. One article by Stang et al.1 is a systematic review and identifies existing measures of ED crowding linked to quality of care across the six Institute of Medicine (IOM) quality domains. From 1980 to 2012, the authors identified 32 ED crowding articles with measured links to quality of care, mostly commonly the number of waiting room patients, ED occupancy, and the number of patients awaiting inpatient beds. A second article by Wu et al.2 describes the relationship between ED crowding and outcomes in critically ill trauma patients with hemorrhagic shock, linking ED crowding with delays in resuscitation and surgery and higher rates of traumatic coagulopathy in the intensive care unit (ICU). As illustrated in the review by Stang et al., the past decade has seen great advances in how crowding and long waits for care affect ED care and outcomes. The majority of this work was published in this decade on the heels of the 2006 IOM report, “Hospital-Based Emergency Care: At the Breaking Point,” which described the causes and consequences of ED crowding in the United States.3 Examining the literature as a whole, several observations can be made. The first is that measuring ED crowding has clearly been a challenge for researchers. This is clear from the large variety of metrics that have been used to measure ED crowding, 15 different ways, in fact. To explain why this has happened, we will begin by describing some scientific issues in measuring ED crowding. Conceptually, the term “crowded” refers to the state of an ED where there are insufficient resources—whether personnel, space, or other resources—to adequately meet patient demands. This supply–demand mismatch leads to queuing or waiting, where patient care is delayed, sometimes for long periods of time. Translating this concept to standard epidemiologic methods, where we link an exposure—such as smoking or whether someone took a particular drug—to an outcome, is problematic. This is because ED crowding is not really a single “exposure,” but rather a dynamic situation that can change not only hour to hour but minute to minute. A patient may arrive at a crowded ED at 4 pm but his or her care may span the next 9 hours; when he or she departs at 1 am, the ED may be much less crowded. In addition, a single ED crowding measure often does not capture why the ED is crowded. Crowding may be related to a rapid inflow of new patients, prolonged processing times for current patients (e.g., high volumes of pending tests or multiple critically ill patients occupying staff), or outflow problems such as ED boarding. It may also be a combination of factors, and this combination can change over time. Capturing why the ED is crowded is important in part because often the cause of crowding may also be the reason why patient quality or outcomes suffer. For example, the medical errors affecting admitted patients in the ED may be related to, and more prevalent during, periods of high ED boarding. By contrast, safety issues emanating from the arrival of several critically ill patients within a short period of time may be entirely different in nature. In the latter situation, the next critically ill but as yet undifferentiated patient (e.g., occult sepsis or smoldering diabetic ketoacidosis) may have to wait for an extended time before important interventions. Stang et al. found that three measures of ED crowding showed consistent links to outcomes, including the number of waiting patients and inpatients and ED occupancy.1 These measures capture the relative “crowdedness” in an ED at a single point in time; however, they do not measure crowding across a patients' ED visit or the reason(s) why the ED is crowded. Yet despite the imperfection of crowding metrics, more than 30 publications have linked crowding (often measured differently) to a wide variety of quality of care and outcomes, underscoring the robustness of the crowding–quality relationship. The second challenge with ED crowding research has been finding measurable ways to capture ED quality and outcomes that have plausible relationships to ED crowding. Many studies have used “process metrics” as outcomes, such as whether the effective or timely care was delivered. In Stang et al.'s review, only a handful of studies have tied higher ED crowding to less effective care, specifically guideline-concordant care for asthma and the actual use of ED pain medication in patients reporting severe pain or ineffective care in the form of higher rates of medical errors.4, 5 By comparison, the great majority of ED crowding studies test whether crowding delays important care and for the most part demonstrate that longer lines (i.e., ED crowding) lead to longer waits (i.e., time to care). This is seen in non–critically ill patients, such as discharged patients with asthma where delays of an hour or more are seen in time to treatment during crowded times.6 It is also seen in the critically ill, where systemwide ED crowding leads to delays in thrombolysis for acute myocardial infarction (AMI), although observed delays in AMI care were short—only a few minutes on average.7 It is also important to mention that some studies have found no association between ED crowding and effective or timely care. For example, one study found that time to care in patients who are explicitly prioritized as candidates for tissue plasminogen activator in acute stroke was no different during crowded versus not crowded times.8 In general, patients who “jump the line” because their presenting condition is explicitly prioritized (e.g., stroke, AMI, trauma) tend to be less affected by ED crowding. Another challenge in crowding research has been to link ED crowding to actual outcome metrics, such as patient experience, mortality, and complications. Some of these outcomes have clear conceptual links to crowding, such as patient experience: when the ED is crowded, patients wait longer and are less happy with their care.9 By comparison, mortality and complication metrics have less of a conceptual link to crowding. It could be argued that some waiting room deaths are directly caused by ED crowding when patients decompensate from care delays. Or alternatively, crowding may cause providers to deliver hurried care, missing critical steps in diagnosis or treatment, which may later cause complications, misdiagnosis, and ultimately mortality. However, it is also possible that most deaths occurring days or weeks after an ED visit may be unrelated to ED crowding. In this case, observed relationships may be due to unmeasured confounding, reflecting that perhaps patients treated during crowded times are slightly different—and higher risk—than those treated when the ED is less congested, even after adjusting for measurable factors. The study by Wu et al. uses both process metrics (timeliness and effectiveness) and outcomes to explore how high-risk, explicitly prioritized patients—those with hemorrhagic shock related to trauma—are affected by ED crowding. They found that both care timeliness, measured as time to initiate blood transfusions and surgery, and care effectiveness, measured as actual amounts of blood products and crystalloid received, as well as whether urinary output was measured, were lower when the ED was crowded. They also linked ED crowding with higher rates of traumatic ICU coagulopathy, but not ICU length of stay or 30-day mortality.2 This adds to the literature on the topic because of the tight conceptual link between delays in care and outcomes in this population and the fact that large delays were seen—more than 2 hours for time to surgery during episodes of ED crowding—which is both statistically significant and clinically important. In addition it demonstrates that crowding not only affects patients with minor illnesses in important ways, it also affects those who are explicitly prioritized. It also further illustrates the challenge of linking crowding to downstream mortality even when a conceptual link exists and linkages to intermediate process and outcome metrics are present. However, the lack of association may be due to the study being underpowered to find differences in mortality: although not statistically significant, the mortality rate was actually nearly 5% higher comparing the highest crowding level to the lowest. Despite the challenges inherent in studying the effects of crowding, researchers across the world have created a robust literature linking ED crowding to quality and to outcomes in the decade since the publication of the IOM's “Emergency Care: At the Breaking Point” report.3 We know now that crowding has variable effects on different types of patients. Some ED patients may have poorer experiences during episodes of crowding, others may experience important delays, and others may receive less effective care leading to greater complications and mortality. There is now proof for what until recently was merely a suspicion: ED crowding is a worldwide public health problem and clearly an important patient safety issue. We hope that crowding researchers will continue this important work with an eye toward finding solutions to improve ED flow and mitigating the effects of ED crowding on observed quality of care and outcomes.