The causes of ED overcrowding are complex and multifactorial 151617. The primary determinants of ED overcrowding are not related to patient throughput inside the ED but actually originate outside the ED 18. Of these, the two most important determinants of ED overcrowding pertaining to the critically ill are an increasing volume of high-acuity patients presenting to the ED and insufficient inpatient capacity.

Between 1992 and 1999, ED visits rose 14% from 89.8 million to 102.8 million visits annually, with the steepest gains in volume over the last 2 years of that time period 19. While this increase in patient volume took place in the 1990s, 1128 EDs closed their doors 20. As a result, more visits are being concentrated in fewer EDs. An increasing proportion of these patients are high-acuity patients who require critical care. Lambe and coworkers 7 reported a 59% increase in critically ill patients presenting to California EDs from 1990 to 1999. The increasing severity of illness among ED patients has been attributed to age shifts in the population and a higher prevalence of patients with severe chronic medical conditions. Considering the projected future growth in critically ill populations as a whole 21, this trend toward increasing severity of illness among ED patients will probably continue.

Although escalating patient acuity places a large strain on ED resources, the most important cause of ED overcrowding is insufficient inpatient capacity for ED patients who require hospital admission 82223. A lack of inpatient beds is also the most important contributor to ambulance diversion 24. The number of inpatient hospital beds in the USA has declined sharply over the past 2 decades. Between 1981 and 1999 the total number of inpatient beds decreased by 39% 23. This cutback has largely been the result of managed care initiatives and hospital cost-containment strategies. Eliminating inpatient beds maximizes the hospital census and ensures a 'full house' at all times, which is favorable from a financial standpoint 9. However, when hospitals are perpetually functioning at greater than 90% of their inpatient capacity, they are ill equipped to handle surges in the number of admissions 25. The current US nursing shortage exacerbates the lack of inpatient capacity by further decreasing the number of staffed beds available to offload an overcrowded ED.

Inadequate inpatient capacity for a patient population with increasing severity of illness forces the ED to serve as a holding area for admitted patients. The term 'boarding' refers to patients who are admitted to the hospital but who remain in the ED, sometimes for more than 24 hours, because of the lack of available beds 81826. Critically ill patients are no exception because ICU patients may also board in the ED for extraordinarily long periods until an ICU bed becomes available 4. A recent report from the American Hospital Association 14 indicated that the average waiting time for an inpatient acute or critical care bed in US EDs is more than 3 hours, but the average waiting time nearly doubles (5.8 hours) in hospitals that consistently have ED overcrowding 14.

EDs are designed for rapid triage, stabilization, and initial treatment. When boarding in the ED causes a 'gridlock', the ED becomes the site for ongoing (i.e. longitudinal) care in the acute phase of hospitalization. In this scenario EDs must provide ongoing care for critically ill patients, effectively serving as expandable extensions of the ICU or 'de facto ICUs' 27. However, EDs are not designed, equipped, or staffed to provide continuing care for the critically ill patient. Although most EDs have specialized areas in which they care for patients who present with trauma and AMI, most do not have the ability to perform invasive hemodynamic monitoring, including arterial and pulmonary artery catheterization. They are designed for rapid stabilization, including resuscitation from cardiac arrest (i.e. 'code' situations), but not for extended care.

Conventional ED nurse : patient ratios do not typically allow for the focused attention that a patient could receive in a critical care unit because most ED nurses are simultaneously responsible for numerous patients with varying severities of illness. When a nurse is assigned to an ICU patient boarding in the ED, one of two scenarios can be expected to occur; either the ideal 1 : 1 or 1 : 2 critical care nurse : patient ratio will be compromised, or the rest of the ED nursing staff will be required to absorb a greater proportion of ED patients.

Boarding in the ED is not only reported to be a barrier to specialized inpatient care, but it also has been identified as a potential high-risk environment for medical errors 8. Critically ill patients boarding in the ED are physically separated from the watchful eye of the intensivists who are ultimately responsible for their care. All of these factors could potentially lead to delays in recognizing deterioration in a patient's condition and in initiating critical interventions, and may detract from optimal patient care.

ED overcrowding has been reported to compromise patient safety, and the critically ill are an especially vulnerable population and are at-risk for serious adverse events. Although the impact of ED overcrowding on patient outcome has not yet been investigated in rigorous prospective observational studies, survey studies in the literature have linked ED overcrowding to clinically significant delays in diagnosis and treatment, as well as to poor patient outcomes 1015. One report 10 linked ED overcrowding to delays in identification and treatment of time-sensitive conditions, such as acute coronary syndrome, stroke, surgical emergencies, and septic shock. ED overcrowding has also been shown to cause ambulance diversion and significant delays in ambulance transport for patients with acute cardiac emergencies 2829, regardless of the severity of illness 28. A recent study reported that ED overcrowding was associated with delays in door-to-needle time for AMI 30. According to the Joint Commission on Accreditation of Healthcare Organizations 31, one half of all 'sentinel event' cases of poor outcomes that were attributable to delays in therapy originated in the ED, with ED overcrowding playing a role in almost one-third of these cases.

In addition to delays in therapy, ED overcrowding may also have an impact on the speed at which critical illness is recognized, through ambulance diversion, triage delays, and delays in bringing patients into treatment rooms. ED overcrowding may also result in extraordinarily long waiting times, causing some patients to leave the ED without being seen by a physician. Patients in the early hours of disease presentation who are initially well appearing and triaged as 'nonemergent' have the potential to leave the hospital without treatment and could become severely ill outside the hospital.

Boarding in the ED can subject critically ill patients to recognition and treatment delays at a pivotal point in the hospital course when time-sensitive interventions are necessary. Because optimal delivery of critical care in the early hours of disease presentation is often time-sensitive (i.e. cardiogenic shock 32, hemodynamic optimization in severe sepsis 3334, and the 'golden hour' of trauma 3536), impediments to prompt critical care recognition and delivery in the ED setting could potentially represent a threat to patient safety.

Adding to the current landscape of ED critical care capacity is the potential threat of terrorism and other disasters. The readiness of US EDs to care for critically ill victims is a key element in preparedness for terrorism or bioterrorism. In a terrorist attack with either conventional or biologic weapons, a large volume of patients would be expected to require critical care services, including mechanical ventilation and hemodynamic support 37. Overcrowded EDs could be poorly prepared to handle mass casualty victims 16.

In the ICU-centric model, the critical care consultant would take over responsibility for any critically ill patient in the hospital at the time of patient identification, regardless of location. This would include critically ill patients boarding in the ED. Transfer of care would occur at the time of consultation, assuming that the patient meets criteria for admission to the critical care service. This would best be accomplished with an 'intensivist model' of staffing ICUs 41, in which a physician trained in multidisciplinary critical care is available around the clock 42. For patients boarding in the ED, the emergency physician would still be in close proximity to respond to sudden major physiologic decompensations but they would not be responsible for ongoing critical care. This would allow the emergency physician to focus on patient throughput for the rest of the ED patients, and may also decrease 'hand-off' errors during patient transition because the intensivists would assume control earlier in the patient's course.

In the ED-centric model, responsibilities for patient care would be site defined. The ED physicians would take full responsibility for all critical care provided in the ED, regardless of how long a patient may be boarding there. Responsibility for patient care would not be transitioned to critical care staff until the patient arrives in the ICU.

One example of this model is the early intervention team (EIT) at Henry Ford Hospital (Detroit, MI, USA). Their ED-based EIT program was created to provide focused care for critically ill patients (specifically severe sepsis patients) in the ED setting. The EIT was founded on the principles that optimal delivery of critical care can be time sensitive, and that aggressive ED based critical care interventions can rapidly improve critical physiology and have a positive impact on outcome. The Department of Emergency Medicine provides all of the EIT infrastructure (including additional personnel, physician training, nursing training, physical plant modifications, and critical care equipment) and maintains the program exclusive from all inpatient critical care services. Essentially, they have built an acute-phase critical care unit in the ED. The EIT physicians send the patient to the inpatient ICU after prospectively defined end-points of resuscitation have been met. ED physicians (including resident physician trainees) staff the program and provide ongoing focused critical care at the bedside, allowing the other emergency physicians to focus on throughput for the noncritical ED patients (Rivers EP, personal communication, 2004).

During the transition of patient care from ED to ICU staff, the use of collaborative evidence-based ED–ICU treatment protocols can help to facilitate uniformity of patient care. The use of protocols to complement clinical decision making for the critically ill has been shown to decrease unnecessary variations in clinical practice 43. Protocol-directed care has already gained acceptance in the management of glycemic control 44, sedation 45, and weaning from mechanical ventilation 464748, but acute phase resuscitation can be guided by protocol as well 3334. Although applicable in all three models, collaborative protocols are more beneficial in this model because there will probably be variability in clinical decision making when patients are co-managed by two different teams of physicians. Collaborative protocols may help to reduce transitional errors and to streamline care.

At our institution (Cooper University Hospital, Camden, NJ, USA), a collaborative ED–ICU sepsis resuscitation protocol (an adaptation of the protocol described by Rivers and coworkers 34) was recently adopted as an institutional 'best practice' model. The ED staff is responsible for patient identification and rapid protocol initiation. Per protocol, the critical care physician is automatically consulted at the time of patient identification for ongoing management. The protocol guides early resuscitative efforts in the ED and carries over to the initial phase of the ICU course as patient care is transitioned to the ICU team. Although transfer of patient care responsibilities (from ED to ICU services) officially occurs at the time of critical care consultation, the ED physicians continue to supervise protocol execution while the patient is boarding in the ED. Before instituting the resuscitation protocol, we held in-service training for our ED nursing staff and we trained all of our junior resident physicians in fundamental critical care support 49. In the future, protocol directed resuscitation might be applicable to shock profiles other than sepsis. Conceptually, a collaborative ED–ICU model can facilitate a seamless transition on the continuum of critical care, as envisioned by the late Dr Peter Safar 50.