PATIENT CROWDING in the emergency department (ED) has become a daily challenge to providinghigh-quality care in EDs worldwide and is a critical problem affecting more than 114 million patients annually in the United States alone(Institute of Medicine, 2006; Pitts, Niska, Xu, & Burt, 2008; Richardson, 2006). According to the Crowding Resources Task Force of 2002, an ED is considered crowded whenthe identified need for emergency services is greater than the available resources in the ED (Asplin et al., 2003). American College of Emergency Physicians (2002) amended this definition to be quantifiableby defining ED crowding as any time when "inadequate resources to meet patient care demands lead to a reduction in thequality of care." In 2006, the Institute of Medicine (IOM) released a series of reports stating that emergency medical services in the United States are overwhelmed and underfunded. In response to the Institute of Medicine reports, The Joint Commission has deemed thatevery hospital must have a plan to address the growing problem of crowding in the ED (Fee, Weber, Maak, & Bacchetti, 2007). However, fewhospitals have been able to develop an effective strategy to combat the daily occurrence of ED crowding. Many factors such as increasedpatient volume, nursing staff shortages, decreased inpatient beds, increased acuity of patients entering the ED, and increased number ofpatients boarded in the ED have been shown to consistently contribute to ED crowding (Magid, Asplin, & Wears, 2004; Weiss, Ernst, &Nick, 2006).

In an attempt to understand the causes and implications of ED crowding, research on the topic continues. The number of articles focusing on ED crowding increases dramatically every year although the topic of ED crowding was first described more than 2 decades ago. Much of the research currently being generated regarding ED crowding is at an organizational level. Urgent Matters, a division of the Robert Wood Johnson Foundation, has become the leading organization in ED research and has focused on finding best practices for hospitals to address ED crowding. Urgent Matters has also focused its resources on improving patient flow. However, less attention has been paid on the effect that ED crowding has on patient outcomes. During periods of crowding, emergency nurses report perceived decreases in the quality of care provided to patients (Pines, Garson, et al., 2007).

Measures of quality care are outcomes that describe the impact on health resulting from care provided by the health care system (Centers for Medicare & Medicaid Services, 2007). The purpose of this review was to summarize the findings of published reports that investigated quality-related outcomes and crowding.

METHODS

A search in PubMed for articles that included the terms "ED" or "emergency department" with "crowding" was performed. As there is no consensus for a measurable definition of ED crowding, no definitive threshold was used to determine whether a study truly measured a crowded ED. Editorial and review articles not written in the English language were excluded. The remaining articles, were reviewed. Only articles that used a quantitative research design were included. The articles that focused on causes of ED crowding were excluded.

RESULTS

The search revealed a total of 276 research articles related to ED crowding. Twenty-three articles focused on patient outcomes related to ED crowding (see Figure 1). Several studies were of large populations from multiple hospitals. Quality-related outcomes identified in published reports from the ED setting include delays in antibiotic and analgesia medication, cardiac intervention, decreased patient satisfaction, and increased mortality.

Figure 1. Flow diagram of reviewed, excluded and included articles.

The 23 research articles that examined associations between patient outcomes and ED crowding were grouped in three categories: delay in treatment [(e.g., percutaneous intervention for myocardial infarction (MI) or the administration of antibiotics, analgesia, or thrombolytics)], decreased satisfaction, and increased mortality.

DISCUSSION

All 23 data-based articles had a physician as the first author. Only four articles had a nurse in the author listing. Although there are several publications related to quality of nursing care in the ED, no report links nurse-sensitive indicators of care quality to ED crowding. Herein is a systematic review of the 23 articles that link outcomes of care to ED crowding so as to inform future investigations into the effects of ED crowding on patient outcomes.

DELAY IN INTERVENTIONS

Eleven studies that examined the effects of ED crowding reported delays in antibiotic administration, analgesia use, and cardiac intervention (Table 1). Delays in these treatments have effects on patients that range from inconvenient to severe health impairment. Delays in analgesics lead to undue suffering in patients with severe pain (Barrett & Schriger, 2008), back pain (Pines, Shofer, Isserman, Abbuhl, & Mills, 2010), hip fractures (Hwang, Richardson, Sonuyi, & Morrison, 2006), and other painful extremity injuries (Abbuhl & Reed, 2003). Delays in the administration of aspirin and thrombolytic medications have an impact upon the severity of myocardial damage (Reikvam & Aursnes, 1999) and cerebrovascular accident patients (Roden-Jullig, Britton, Malmkvist, & Leijd, 2003). Delays in antibiotic administration have been associated with increased length of stay in patients diagnosed with community-acquired pneumonia (Blot et al., 2007) and mortality in meningococcal disease (Bugden, Coles, & Mills, 2004).

TABLE 1. Studies that examined the effects of ED crowding

TABLE 1. Studies that examined the effects of ED crowding (

TABLE 1. Studies that examined the effects of ED crowding (

TABLE 1. Studies that examined the effects of ED crowding (

TABLE 1. Studies that examined the effects of ED crowding (

TABLE 1. Studies that examined the effects of ED crowding (

Antibiotic Delay

Three studies looked at the effect that ED crowding had on antibiotic administration (Fee et al., 2007; Pines, Hollander, Localio, & Metlay, 2006; Pines, Localio, et al., 2007). All three studies included adult patients. Two studies were limited to one hospital, whereas one study (Pines et al., 2006) included patients from 24 hospitals. Sample size varied from 405 to 694 patients. All studies used a descriptive, retrospective design.

Each research group measured crowding, using different methods. Pines et al. (2006) used ED length of stay and the left-without-being-seen rate to measure crowding. Fee et al. (2007) studied ED volume instead of crowding because of the multiple definitions of crowding and the lack of agreed-upon standard of ED crowding. Pines and colleagues (2007) determined crowding levels by recording two input measures of (1) waiting room number at triage and (2) number of newly registered patients in the 6-hour period before triage, two throughput measures of (1) arithmetic mean ED LOS for patients who were discharged from the ED in 6 hours before triage and (2) arithmetic mean ED LOS for admitted patient who were transferred to inpatient beds in the 6 hours before triage, and three output measures of (1) the number of patients discharged from the ED in 6 hours before triage, (2) the number of admitted patients who were transferred to inpatient beds in 6 hours before triage, and (3) the boarding burden (total patient-care hours for those patients who had to be admitted to the hospital as determined by a computerized bed request). The researchers also used the total patient care measures (a sum of all hours for all patients currently treated in ED at triage) as a global measure of ED crowding.

All researchers reported that crowding was associated with delays. Pines et al. (2006) found that length of stay was inversely associated with the probability of receiving antibiotics within 24 hours. Fee et al. (2007) found that antibiotics administration within 4 hours was less likely with a higher census. Pines, Localio, et al. (2007) found that increasing levels of ED crowding were associated with delayed or no antibiotic administration. When both the waiting room and recent length of stay were at the lowest quartiles, the predicted probability of delayed antibiotics within 4 hours was 31%. Alternatively, when both waiting room and recent length of stay were at the highest quartile, the predicted probability was 72%.

Pines, Localio, et al. (2007) explain that these crowding measures, chosen and adapted from a list of potential measures (Solberg, Asplin, Weinick, & Magid, 2003) and assigned at triage, were chosen for their simplicity, ease of measure, and face validity. However, there is no evidence that construct and content validity have been established in these measures. Fee et al. (2007) reported their results, using odds ratios. Although this demonstrates the increase risk of delay with the presence of each additional patient, it does not provide information as to how long a delay is extended with the presence of more patients.

Findings from all three studies indicate that the degree of harm due to delays in antibiotic administration depends on the patients' presenting diagnosis. Because antibiotics are used to treat many types of infections, a delay in their administration for an ear infection carries a less severe implication than a potentially life-threatening diagnosis of bacterial meningitis. Thus, although crowding is associated with delayed antibiotic administration, the implications for delay can vary by condition, and patient condition has not been separately examined for its influence on the occurrence or amount of delay.

Delay in Cardiac Intervention

Three articles were located that specifically examined the effect of ED crowding on delays of administering thrombolytics or delays of percutaneous coronary intervention (Kulstad & Kelley, 2009; Pines et al., 2006; Schull, Vermeulen, Slaughter, Morrison, & Daly, 2004). All three articles were a retrospective, descriptive design. Sample size of adult subjects varied from 17 at a single hospital site to 3,452 from 25 hospitals in Ontario, Canada. Crowding was measured by ED length of stay and the left-without-being-seen rate (Pines et al., 2006), EDWIN score (Kulstad & Kelley, 2009), and ambulance diversion status (Schull et al., 2004). The results of the studies varied. Pines et al. (2006) found that no crowding measures demonstrated an association with time to percutaneous intervention (PCI) for patients with acute MI. However, Kulstad and Kelley (2009) found a significant difference between high crowding and low crowding times for the time to PCI and Schull et al. (2004) found an increase in delay during high crowding times. It may be that EDs with specialized cardiac services or personnel are less likely to experience delays in cardiac interventions, regardless of crowding.

Delay in Pain Management

Five studies examined the relationship between ED crowding and pain management (Hwang et al., 2006, 2008; Mills, Shofer, Chen, Hollander, & Pines, 2009; Pines & Hollander, 2008; Pines et al., 2010).

Sample size for these studies ranged from 158 to 13,758 subjects. The populations' presenting complaints included hip fracture and severe back and abdominal pain. All the studies were retrospective and descriptive in design. These studies all found that episodes of ED crowding are associated with poor pain management.

Hwang et al. (2008) set the outcomes for their study as (a) documentation of pain assessment, (b) medications ordered, and (c) times of activities. Although no difference was noted in the documentation of pain assessment or pain follow-up by physicians during high census compared to low census or during periods of high or low boarders, there were direct correlations between ED census and increased time to pain assessment (r = .22, p < .0001), as well as time to administration of pain medications (r = .25, p < .0001). Pines and Hollander (2008) found that nontreatment was associated with various measures of crowding (OR = 1.02)

DECREASED SATISFACTION

Five research articles were located that examined the association between ED crowding and decreased patient satisfaction (McMullan & Veser, 2004; Pines, Garson, et al., 2007; Pines et al., 2008; Sun et al., 2000; Vieth & Rhodes, 2006). The sample sizes for these studies ranged from 629 to 2,899. All studies save one (McMullan & Veser, 2004), which looked at left without being seen rates, were the result of surveys taken of patients, doctors, and/or nurses. All studies were descriptive in design. The results of these studies consistently showed an inverse relationship between patient/visitor satisfaction and ED crowding. The measure of crowding varied between all the studies.

The study by Pines et al. (2008) differed from the other studies in that they examined the association between factors related to ED crowding and patient satisfaction with the ED and the overall hospital. However, only one statement on the survey related to the general ED visit by addressing the degree to which a patient would recommend the ED to others. The elements of ED crowding that were measured included the number of waiting room patients, ED occupancy as a percentage of treatment spaces, and the number of admitted patients.

INCREASED MORTALITY

Mortality is a common patient outcome measure and is commonly used as an indicator of quality of care. The current leaders in ED crowding research use death as a common measure of adverse patient outcomes (Robert Wood Johnson Foundation, 2004). But there are limitations to using mortality as an outcome measure. At times, mortality may not be a relevant measure of quality. Generally, mortality has been considered a poor outcome. However, when a terminally ill cancer patient enters the ED with the wish to die a peaceful, pain-free death, the outcome (e.g., death) may reflect poorly on the mortality rate of the ED, while simultaneously illustrating the best possible care that could have been provided to the patient as well as a desired outcome. Death is a dichotomous variable and does not represent the full spectrum of potential outcomes that an ED patient could experience. Historically, there is a lack of clarity about definitive measures of quality (Department of Health, 2008; Klein, Malone, Bennis, & Berkowitz, 1961) However, data about mortality and crowding in the ED have implications for both practice and research.

Eight studies examined the association between ED crowding and mortality (Chalfin, Trzeciak, Likourezos, Baumann, & Dellinger, 2007; Diercks et al., 2007; Fatovich, 2005; Gilligan et al., 2008; Miro et al., 1999; Richardson, 2006; Shenoi et al., 2009; Sprivulis, Da Silva, Jacobs, Frazer, & Jelinek, 2006). Although ED crowding was measured differently in each study, the majority of these studies found that correlations exist between ED crowding and increased mortality.

Richardson (2006) performed a retrospective stratified cohort study that compared the mortality rate of all patients who entered an Australian ED during shifts classified as "overcrowded" and an equivalent number during "not overcrowded" shifts. During the 736 shifts from both categories, 144 deaths occurred in the overcrowded cohort and 101 deaths occurred in the not overcrowded cohort over the 48-week period (0.42% and 0.31%, respectively; p = .025) with a relative risk for 10-day mortality of 1.34 (95% CI = 1.04-1.72). Although the author reports that patients present during overcrowded shifts received inferior care in terms of standard performance (starting appropriate treatment within triage, left without being seen (LWBS) rate, boarded in the ED), only descriptive data were provided and no statistical analysis within the article verified this conclusion. In addition, Richardson (2006) used 75% occupancy as the definition of ED crowding. The author did explain that the definition of overcrowding chosen was "necessarily weak to obtain similar cohorts." Because of this decision, the possibility that a more acutely ill population of patients contributed to both overcrowded conditions and increased mortality rates may be an explanation for the significant results. A more comprehensive definition of ED crowding may have provided stronger correlations between crowding and mortality rates. Another limitation of this study is the use of logistic regression to analyze the results; the dichotomization of crowding levels does not allow for the analysis of how mortality rates would differ between a crowded ED and an overcrowded ED.

Sprivulis et al. (2006) examined the effect of crowding on mortality of adult ED patients, using a retrospective design in Australian EDs. The authors used ED occupancy as an indicator of crowding and looked at deaths on Days 2, 7, and 30 after ED admission. The authors developed the Overcrowding Hazard Scale to test the combined effects of hospital and ED occupancy. Logistic regression was used to assess the differences in demographics and clinical characteristics of patients who died by Day 30 and showed that they had longer ED LOS (risk ratio per hour of ED stay, 1.1; 95% CI = 1.1-1.1; p < .001) and longer physician waiting times (risk ratio per hour of ED wait, 1.2; 95% CI = 1.1-1.3; p = .01). Using a Cox's regression, a linear relationship was identified between the Overcrowding Hazard Scale and increased mortality rates on Day 7 (r = .98; 95% CI = 0.79-1.00). However, no report was provided on the validity and reliability of this instrument. Therefore, the true relationship between ED crowding and mortality cannot clearly be established using this metric.

A retrospective cross-sectional study performed by Shenoi et al. (2009) used the occurrence of ambulance diversion as a proxy for the presence of ED crowding and assessed the prevalence of ambulance diversion and its association with mortality among 11 pediatric hospitals and their patients. The authors reviewed 63,780 total admissions and found that 4,095 (6.4%) children were admitted during periods of ambulance diversion. Bivariate analysis showed that the occurrence of ambulance diversion was protective for mortality (OR = 0.51; 95% CI = 0.34-0.77). However, this study demonstrates only a limited effect of crowding because the occurrence of ambulance diversion restricted the degree to which crowding occurred during these times.

Similarly, a retrospective study performed by Fatovich (2005) examining the mortality rate of patients in an Australian ED during times of ambulance diversion showed a statistically significant reduction in patient mortality during periods of ambulance diversion. The results of this study may indicate that ambulance diversion diminishes the influx of patients into the ED. By decreasing the number of patients entering the ED, the staff can meet the needs of the patients already present in the ED during this time. The author did not report the mortality rates of the patients in the ambulances during diversion periods.

The results of these studies are mixed. When ambulance diversion is used as an indicator of ED crowding, the mortality rates of patients already in the ED during these times decrease. However, the studies performed by Richardson (2006) and Sprivulis et al. (2006) demonstrate that when ED crowding is measured using alternative methods there is a correlation between mortality and ED crowding. Because of the mixed results of these studies, future research needs to use valid and reliable measure of crowding.

FUTURE DIRECTIONS

The current body of research regarding quality of care has focused on three categories: patient satisfaction, mortality, and delays. To better understand the implications of ED crowding, additional investigation into quality indicators has great potential to develop processes that deliver high quality of care, regardless of crowded conditions. Every year overcrowding in the ED has become more prevalent (American Academy of Emergency Medicine, 2006) so nurses need to be able to adjust to deliver safe effective care in the presence of ED crowding.

Outcome Measures

Measures of effects of ED crowding on patients are limited. Scant research is available regarding nurse-sensitive indicators suchas avoidance of complications from treatments, communication failures/successes, nosocomial infections, and patient falls in the ED. Forexample, studies of medication errors in the ED are confounded by underreporting by staff (Croskerry & Sinclair, 2001; Francis, Spies,& Kerner, 2008; Horwitz et al., 2008; Hostetler et al., 2007; O'Neill, Shinn, Starr, & Kelley, 2004; Trzeciak & Rivers, 2003). No studies that examine failure-to-rescue, hand-off communication or indicators of nursing processes during periods of crowding inthe ED could be located. To obtain a more precise picture of the true effects of ED crowding on patients, a broader spectrum ofpatients' outcomes must be examined. Community-acquired pneumonia was one of the initial areas identified by TJC as a hospital coremeasure for quality of care. Although a 4-hour antibiotic target has been set for patients with pneumonia, as with other quality benchmarks, the impact of patient volume on this target has not been studied. Demonstrating a relationship between volume and quality of care isimportant in bringing resources to bear on solving the crowding issue, as well as improving the ability to meet quality benchmarks. Onestrategy to build relationships between concepts in delivery of care in the ED is to use the Input-Throughput-Output Model (Asplin et al., 2003). This model has been adopted by others researching the outcomes of care in a crowded ED (Pines, Localio, et al., 2007; Schull et al., 2004; Viethe & Rhodes, 2006). Using a consistent conceptual framework may aid in building the science and art of high-quality care in the ED.

The Presence of Nurses in the Research

Little evaluation has been performed that examines processes of care including how nurses respond when EDs are crowded. No matter how many patients are present in the ED, the emergency nurse and the advanced practice emergency nurse are still required to perform assessments, assess vital signs, provide care/medication, and facilitate discharge education and follow-up. Current research does not provide information regarding the effect that ED crowding has on the quality of care provided to ED patients by nurses.

There are limited data about quality-related outcomes from nursing care in the ED. Research needs to be conducted to examine the relationship of ED crowding to important patient outcome measures. To address the issue of ED crowding completely, researchers must examine the consequences, not only the causes, of crowding. By examining relationships between ED crowding and nurse-sensitive patient outcomes, researchers can provide evidence on which to base nursing practice. For example, by examining factors that affect the frequency with which the emergency nurse collects, interprets, and responds to information may be altered during ED crowding. How nurses respond to changes in patient status and ways to streamline communication during periods of crowding are potential foci of future research.

Addressing Additional Patient Populations

Research has been conducted that examines the delays of antibiotics to patients with pneumonia, thrombolytics to patients with stroke and acute myocardial infarction (MI), electrocardiogram and aspirin to MI patients, and analgesics to patients with hip fractures. However, many ED patients are not included in any of these populations. No analysis has been performed on patients with abdominal pain of unknown etiology, migraine sufferers, noncardiac chest pain patients, or many other categories of patients who fall into common presentations or lower acuity.

RELEVANCE FOR NURSES

Nurses can advance safe effective care by building an understanding of how ED crowding affects the practice of the emergency nurse and the outcomes of emergency patients. ED nurses need to continue to study the effects of crowding on ED patients, and evaluate factors and potential barriers to the processes of monitoring and intervention as well as provide evidence-based knowledge for practice and education. Determining barriers and facilitators to monitoring under various crowding states has great potential to contribute to consistent, high-quality nursing care. Low-quality or missed nursing care may cause mortality and contribute to return ED admissions, poor communication between clinicians, and missed opportunities to educate patients for self-management resulting in unnecessary suffering, and contribute to patient/family dissatisfaction with an institution or a provider. For example, monitoring and surveillance of especially vital signs, level of consciousness, and patient response to treatment is a common valued nursing intervening process priority.

A discipline is "characterized by a unique perspective, a distinct way of viewing all phenomena" (Donaldson & Crowley, 1978). However, much of the crowding research is being generated by physicians and policy makers, both of whom lack the unique perspective that nurse researchers provide. Studies conducted within the medical profession focus on how to fix the problem of ED crowding and not on how the problem affects a person. Generating knowledge about ED crowding and nurse-sensitive patient outcomes is important to nursing practice and education.

ED crowding is common and recurrent in emergency nursing practice. Policy makers have been mandated by TJC to address crowding issues, and implementation of such policies has a direct impact on how the emergency is required to practice. Florence Nightingale (1820/1910) wrote that the role of nursing practice is "to put the patient in the best possible condition for nature to act upon him" (Nightingale, 2003). This declaration has guided nursing practice to focus on improving quality of care, while meeting the mandated practice standards of regulation agencies by providing an atmosphere of healing for patients.

Nursing contributions to safe effective care during ED crowding may help to guide interventions to help alleviate or prevent adverse outcomes.

CONCLUSION

The effect of ED crowding on patient satisfaction levels, medication administration delays, and mortality is well documented. Studies demonstrate that quality of care is being impacted during crowding, resulting in delays in treatment and medication administration, reduced patient satisfaction, and even death. Unfortunately, ED crowding is a complex problem with no foreseeable solution. Therefore, ED practitioners and administrators need more information about how to provide quality care despite ED crowding.

Researchers agree that several factors consistently contribute to ED crowding: (a) fewer hospitals nationwide, (b) nursing staff shortages, (c) decreased inpatient beds, (d) increased patient volume, (e) increased acuity of patients entering the ED, and (f) the increased number of patients boarded in the ED (Bernstein, Verghese, Leung, Lunney, & Perez, 2003; Estey, Ness, Saunders, Alibhai, & Bear, 2003; Magid et al., 2004; Raj, Baker, Brierley, & Murray, 2006). Because few of these factors are amenable to interventions (e.g., it is unlikely that new hospitals will be built, inpatient beds added to hospitals, or that acuity will suddenly decrease), it may be that effective strategies for the optimization of patient outcomes focus on the processes that will contribute to maintaining high-quality care even during times of crowding. Nurses have a unique and valuable perspective in identifying safe, effective care and are urged to examine practices that support optimal patient outcomes in ED, regardless of crowding status.

REFERENCES