Community Trial to Decrease Ambulance Diversion Hours: The San Diego County Patient Destination Trial

Transcription

1 EMERGENCY MEDICAL SERVICES/ORIGINAL RESEARCH Community Trial to Decrease Ambulance Diversion Hours: The San Diego County Patient Destination Trial Gary M. Vilke, MD Edward M. Castillo, PhD, MPH Marcelyn A. Metz, RN Leslie Upledger Ray, MA, MPPA Patricia A. Murrin, RN, MPH Roneet Lev, MD Theodore C. Chan, MD From the County of San Diego, Division of Emergency Medical Services, San Diego, CA (Vilke, Castillo, Metz, Upledger Ray, Murrin); the Department of Emergency Medicine, University of California San Diego Medical Center, San Diego, CA (Vilke, Chan); and Scripps- Mercy Hospital, San Diego, CA (Lev). See editorial, p Study objective: Emergency department (ED) ambulance diversion is a major issue in many communities. When patients do not reach requested facilities, challenges in care are compounded by lack of available medical records and delays in transferring admitted patients back to the originally requested facility. We seek to evaluate a community intervention to reduce ambulance diversion. Methods: This was a community intervention in a county of 2.8 million individuals. Ambulance diversion guidelines were revised for all ambulance agencies and EDs. Participation by EDs was voluntary, and main outcome measures, which included ambulance transports, ambulance diversions, and bypass hours, were compared for the pretrial, trial, and posttrial periods. Results: A total of 235,766 patients were transported to an ED by advanced life support ambulance during the 2-year study period. There was a significant decrease in the number of patients who did not reach the requested facility because of ambulance diversion for the trial period (n=322) and posttrial period (n=449) compared with the pretrial period (n=1,320; 998 diverted patients per month [95% confidence interval (CI) 1,162 to 833 patients] and 871 diverted patients per month [95% CI 963 to 780 patients], respectively). There was also a significant decrease in average monthly hours on diversion for the trial period (n=1,079) and posttrial period (n=1,774) compared with the pretrial period (n=4,007; 2,928 hours on bypass [95% CI 3,936 to 1,919 hours on bypass] and 2,232 hours on bypass [95% CI 3,620 to 2,235 hours on bypass], respectively). Conclusion: A voluntary community-wide approach to reducing hospital ED diversion and getting more ambulance patients to requested facilities was effective. [Ann Emerg Med. 2004;44: ] /$30.00 Copyright Ó 2004 by the American College of Emergency Physicians. doi: / j.annemergmed OCTOBER :4 ANNALS OF EMERGENCY MEDICINE 295

2 Editor s Capsule Summary What is already known on this topic Excessive ambulance diversion may adversely affect emergency medical services (EMS) system efficiency and impair the care of patients transported to alternate emergency departments (EDs). What question this study addressed Investigators monitored hospitals ambulance diversion hours before and after stricter countywide ambulance diversion rules were implemented. The rules limited diversion episodes to 1 hour in duration and mandated acceptance of at least 1 patient before going off diversion. What this study adds to our knowledge During the 2-year study period, 235,766 EMS patients were transported. Ambulance diversion hours decreased 56%, fewer patients failed to be transported to their hospital of choice, and fewer patients required subsequent interfacility transfers. How this might change clinical practice Simple rules can significantly decrease ambulance diversion and benefit EMS, patients, and hospitals alike. INTRODUCTION Historically, ambulance diversion was created to offer temporary relief to emergency departments (EDs) that were full or had critically ill patients whose safety and care could be jeopardized by the arrival of a similar patient. This process was initially designed to be used rarely and for short periods to allow an ED to quickly recover from its temporary condition. The use of diversion has increased across the country as a mechanism to attempt to treat the growing problem of ED crowding. 1-3 The overuse of diversion has grown to epidemic proportions. ED crowding has been widely reviewed and written about in the lay press and the medical literature. 4-9 In fact, previous work has shown that ambulance diversion can lead to delays for patients to obtain definitive medical care and can result in adverse outcomes, including death In San Diego County, a data review for quality improvement purposes revealed that there has been a marked increase in diversion hours during the past 5 years, to the point that hospitals were on diversion status 1 of every 4 hours cumulatively. Some communities have addressed this issue with political mandates by nonmedical personnel banning the use of diversion. 14 Others have been told by local emergency medical services (EMS) agencies that diversion will no longer be an option. Locally, the EMS community thought that these were extreme options, but if our current practice of ambulance diversion did not decrease, then political mandate was a real possibility. This trial was designed to develop a forum to align the EMS community participants with a common goal to improve the number of patients reaching their requested facilities primarily, and with a secondary goal of decreasing overall ambulance diversion hours. We hypothesized that more ambulance patients requesting specific EDs could reach their desired destination and that overall bypass hours could be decreased with a voluntary community-wide effort to improve out-of-hospital patient care. MATERIALS AND METHODS Selection of Participants Blinded, aggregated, out-of-hospital records for all patients, including pediatric patients, who were transported to an ED by advanced life support ambulance after accessing paramedics via 911 from October 1, 2001, through September 30, 2003, were included. Data for the pretrial period (October 1, 2001, through September 30, 2002) had been collected using the County Quality Assurance Network computer network during routine system surveillance. The data for the trial and posttrial periods were prospectively collected after the initiation of the study. The County Quality Assurance Network is an online real-time computer network that connects all receiving hospitals, base hospitals, and trauma centers and many ambulance stations. Trauma patients transported to a trauma center resuscitation suite were not included in this study. The study was performed in a county with approximately 2.8 million people residing in urban, suburban, rural, and remote areas. Approximately 250,000 calls are received annually by nine 911 dispatch agencies, resulting in 150,000 emergency patients. Eighteen advanced life support ground transport agencies and a single aeromedical rotor-wing agency operate in the county, employing about 900 paramedics to provide EMS to these areas. The location has 21 EDs (2 of which are military) with 1 Level I trauma center and 5 Level II trauma centers, one of which is the pediatric trauma center. Medical direction is provided by written protocols, with available online medical direction by a mobile intensive care nurse or base hospital physician when necessary. In San Diego County, out-of-hospital providers and the mobile intensive care nurses enter data from calls run throughout the county into the County Quality Assurance Network s electronic out-of-hospital patient record. Aside 296 ANNALS OF EMERGENCY MEDICINE 44:4 OCTOBER 2004

3 from collecting out-of-hospital patient data, the County Quality Assurance Network also provides real-time updates on the resource status of all receiving hospitals and trauma centers. The time an individual ED went on diversion status and the duration of that status were collected and evaluated. Additionally, the ED to which a patient had requested transport, whether the requested ED was on diversion at the time, and the facility to which the patient was ultimately transported were also collected. There were no changes in the data collection process during the prospective and retrospective phases of the study. The existing county policy is to transport the patient to the closest, most appropriate emergency facility capable of meeting the needs of the patient. Patient requests are to be honored, unless triage status of the patient or extreme distances, defined as greater than 20-minute transports beyond local hospitals, precludes safely honoring the request. If a facility is on ambulance diversion, that facility does not receive ambulance traffic. If multiple hospitals in the same region are on ambulance diversion, then operations run as if none of the facilities are on diversion, and each gets its own requests and a sharing of the remaining patients who do not have requests. All agencies follow this policy. The study process was initiated by establishing a medical oversight committee, through the San Diego County Medical Society, that consisted of representatives from each receiving hospital ED (including ED nurse managers and ED directors), paramedic agencies, the San Diego County Division of Emergency Medical Services, and the local health care association, which represents the chief administrative officers and directors from the community hospitals. The community issues about patient care hindered by diversion were discussed, and consensus was reached on the need to make an intervention. The rules for diversion were established by the committee and agreed on by all participants. First, each episode of diversion status would last only 1 hour. After an hour, the ED would have to remove itself from diversion status. Second, diversion status could not be reestablished until the facility had taken at least 1 ambulance patient. Third, while on diversion, a hospital would still take patient requests to that facility, unless there was a significant patient safety issue. A consensus of when ED diversion was allowed was created by the San Diego County Medical Oversight Committee in 2000 and was redistributed to all county EDs. This consensus was only a list of recommendations because there were no mandates or penalties. The appropriate indications for diversion are shown in Figure 1. Participation was voluntary, but all hospitals agreed to participate. Meetings of the oversight committee were held weekly for the first month of the trial and then biweekly for months 2 and 3. Ongoing review of the data was carefully monitored by the committee, and significant deviations from the established rules (ie, diversion status more than the predetermined 1 hour) were brought to the attention of the ED nurse manager and ED director of the involved facility. Letters and mailings went out to each of the hospitals and advanced life support agencies, with information and education points on the process to be given to the ED staff, as well as the out-of-hospital workforce. Additionally, reporting of ED crowding issues, safety issues from requested patients being brought to the ED, and offload delays were reported at these meetings Figure 1. The San Diego County Medical Society EMS Medical Oversight Committee Emergency Department Ambulance Diversion Guideline; San Diego County, CA, October Purpose To apply a uniform guideline among all San Diego EDs when an ED declares itself to be on saturation, requesting that emergency ambulance patients be diverted to a different facility. Authorization for ED Diversion 1. Only the combination of the on-site emergency physician and on-site charge nurse, at a minimum, may declare that their ED is saturated and authorize diversion. 2. After no more than 3 hours of diversion caused by ED saturation, an appropriate hospital administrator or manager should be notified of the prolonged saturation status and be asked to attempt to remedy the problem. Length of ED Diversion Each ED must reassess its diversion status every hour to ensure compliance with the agreed-on causes for ED diversion. Acceptable Causes for ED Diversion The ED does not have bed or staffing capacity to treat a new acute-care patient regardless of whether the patient presents by ambulance, walkin, or in-house. Example: All monitored beds in the ED are in use with patients requiring continuous telemetry, and therefore no new monitored beds are available for a new patient requiring this level of care. Example: All nurses are involved with critical procedures or medical care, and therefore no nurses are available to care for a new patient requiring intensive medical attention. Example: Cardiac or psychiatric patient requiring continuous monitoring or observation. Unacceptable Causes for ED Diversion Caused by ED Saturation Full waiting room No ICU beds Computed tomographic scanner down Staffing shortage Long waiting room time OCTOBER :4 ANNALS OF EMERGENCY MEDICINE 297

4 throughout the study period by the various departments and transporting agencies. The oversight committee set the trial period to last 3 months (October 1, 2002, to December 31, 2002). This period allowed time to initiate the process, as well as to review the data to see whether the intervention was causing any meaningful changes. Before initiation, the research plan called for reassessment at 3 months time. A decision about whether to extend the trial period if more data were needed, stop the study and return to the previous operational status, or stop the study and continue the process as the new mode of operation was to be made at that time. The trial could be stopped at any time if patient safety or patient care was believed to be compromised by the study. After the 3-month trial period, the oversight committee thought that the trial should be stopped and the process be continued as the standard operating procedure. Data for the subsequent 9 months (posttrial period) were monitored to assess the success of maintaining the effects on the system of the initial intervention. Meetings of the oversight committee continued monthly throughout the posttrial period, with data reported to the hospitals weekly. No additional education or interventions were continued during this phase. Primary outcome measures included total number of patients transported by ambulance, total number of patients who requested a particular ED but were taken to another ED because of diversion status, and total number of hours on diversion. Secondary outcome measures included the total number of ED visits, hospital admissions, and ED interfacility transfers out. These secondary data points were unavailable from the county Quality Assurance Network system and had to be obtained from each facility directly. In many cases, these data were not routinely collected and required hand search. Therefore, these secondary data were collected only for the 2 months preceding the trial (pretrial) and the first 2 months of the trial (trial). This study was approved by the University of California San Diego, Human Research Protections Program. Primary Data Analyses Descriptive statistics are presented for demographic characteristics, advanced life support transports, patients who were diverted from their requested facility, and hours on bypass for the pretrial, trial, and posttrial periods. Advanced life support transports, patients who were diverted from their requested facility, and hours on bypass for the trial and posttrial period were compared with that of the pretrial period using point and interval estimates of effect size, specifically the difference in means and the 95% confidence interval (CI) on the difference. Additionally, relative risks (RR) and corresponding 95% CIs are presented for diverted patient comparisons. Data were analyzed with SPSS software (version 11.0, SPSS, Inc., Chicago, IL). RESULTS During the study period, there were a total of 235,766 patients transported to EDs by advanced life support agencies after 911 had been accessed. There was an average of 9,623 patients per month transported by paramedics during the pretrial period, 9,818 patients per month transported during the 3-month trial period, and 10,093 during the posttrial period. Patient demographic characteristics for the 3 periods are presented in Table 1. Table 2 shows the data collected during the 3 study periods, comparing pretrial data with the trial and posttrial periods: number of paramedic transports, diversion hours, and patients who had requested a specific ED but were not taken there because of diversion status. The monthly average number of ambulance runs increased slightly for the trial period (186 patients per month; 95% CI 283 to 655 patients) and significantly for the posttrial period (435 patients per month; 95% CI 78 to 788 patients) compared with the pretrial period. There was a significant decrease in the number of patients who did not get to the requested facility because of ambulance diversion for the trial period and posttrial period compared with the pretrial period (respectively, 998 diverted patients per month, 95% CI 1,162 to 833 patients, RR 0.24, 95% CI 0.21 to 0.27; and 871 diverted patients per month, 95% CI 963 to 780 patients, RR 0.32, 95% CI 0.29 to 0.36). These results correspond to a decreased risk of a patient being diverted from a requested hospital Table 1. Patient and ambulance run characteristics for the 3 periods; San Diego County, CA, October 2001 through September Characteristic Pretrial Period Trial Period Posttrial Period Total runs/month 9,623 9,818 10,093 Mean age, y Male sex, % Age <18 y, % Trauma complaints, % ANNALS OF EMERGENCY MEDICINE 44:4 OCTOBER 2004

5 because of ED saturation of 76% from the pretrial period to the trial period and 66% from the pretrial period to the posttrial period. There was also a significant decrease in average monthly hours on diversion for the trial period and posttrial period compared with the pretrial period ( 2,928 hours on bypass, 95% CI 3,936 to 1,919 hours; and 2,232 hours on bypass, 95% CI 3,620 to 2,235 hours, respectively). Figure 2 displays the number of diversion hours during the 3 study periods by month. Figure 3 displays the patients who made requests but could not reach the requested ED because of diversion status. Despite an overall increase in paramedic runs from the pretrial to trial and posttrial periods, there was a significant decline in total diversion hours and numbers of patients diverted to other facilities. Table 3 represents the reported interfacility transfer data from 17 of the 21 hospitals. The 2 military hospitals did not report the requested data. Additionally 2 civilian EDs also did not report their data. Two EDs reported all data except for total admissions. The transfer data were broken down to reflect transfers that were initiated because of level of care issues (trauma, psychiatric, or pediatrics) and those that were due to payer request. When available, the number of transfer patients who arrived by ambulance as opposed to those who arrived by other modes of transportation was presented. There was a significant decrease for patients transferred for all reasons ( 175 patients per month; 95% CI 259 to 91 patients) and patients transferred for payer request who were brought in by paramedics ( 100 patients per month; 95% CI 177 to 23 patients). During the study period, none of the EDs reported crowding issues above baseline. Additionally, there were no reported safety issues from requested patients being brought to the ED while they were on diversion. Although initially there were 2 ambulance transporting agencies reporting some increased incidence of offload delays, closer analysis by the agencies determined that this incidence was likely seasonally related because it had happened around the same time of year before the initiation of the trial. However, given that their data were reported at the regular meetings, opportunities for discussion and increased awareness in the need for efficient patient turnover occurred. No specific actions were taken by the oversight committee because of these data. During the pretrial year and the study period, there were no new changes in any of the participating hospitals configurations. There were no additions of ICU, telemetry, medical or surgical, or ED beds. Standard-practice staffing increase or decrease was not addressed by the oversight committee and was left at the discretion of each hospital. LIMITATIONS Limitations to this study include that the trial period was 3 months, which was predetermined by the oversight committee and was in the interest of the community; it was believed that changing the process from a trial to the standard operation at the 3-month review would benefit the community by further emphasizing the importance of following the established rules. The committee was concerned that continuing the process as a trial, given the positive findings, would serve only to potentially decrease enthusiasm for this change in the future. The computerized database also had its limitations. During the trial period, the Quality Assurance Network computer system would infrequently go down for short periods, preventing hospitals from taking themselves off or placing themselves on diversion. The database was not modified for these events because there was an equally likely chance that similar events occurred in the pretrial phase. Additionally, this study did not use formal statistical stopping rules, which may be interpreted as a potential limitation. Rather than limiting the evaluation of these new practices to statistically testable data, we focused on Table 2. Mean monthly paramedic runs, bypass hours, and diverted patients for the 3 periods, San Diego County, CA, October 2001 through September 2003.* Characteristic Pretrial Period Trial Period Posttrial Period Pretrial/Trial Difference (95% CI) Pretrial/Posttrial Difference (95% CI) Number of paramedic runs 9,623 9,818 10, ( 283 to 655) 435 (78 788) Hours of bypass 4,007 1,079 1, ( 1,162 to 833) 871 ( 963 to 780) Patients bypassed 1, ,928 ( 3,936 to 1,919) 2,232 ( 3,620 to 2,235) *Data reported as monthly averages. OCTOBER :4 ANNALS OF EMERGENCY MEDICINE 299

6 the qualitative input from practitioners and paramedic agencies working throughout the county. The use of secondary measurements also had limitations. Transfer data were not available from all of the hospitals in the county. The 2 military hospitals were unable to provide the requested data, although they received a low volume of nonmilitary ambulance traffic. Additionally, 4 hospitals did not offer completed data sets, including 2 hospitals that did not reveal total admissions but supplied the transfer data. It is unlikely that these few missing data points would substantially change the noted decrease in interfacility transfers for payer request reasons that were found in this study. DISCUSSION Ambulance diversion is an issue that has received national attention recently. The news media has reported sensational cases of ambulances with patients circling cities in search of an ED that can take their patient. 15 Recent reports at the federal level of government have documented the growing problem. 9 Many reasons for the increased use of ambulance diversion have been postulated at the individual hospital level. However, changes at any one institution are unlikely to be able to be enough to absorb the increased demands placed on a community when multiple EDs are on diversion. We sought to make changes on a community-wide level. Previous research in this community had been performed to demonstrate that diversion was being overused. 16 The first review involved 2 large hospitals located several blocks apart. The first hospital s administration secured additional resources to stay off diversion for a week to assess internally how census and payer mix were affected. The second hospital made no changes. During the trial week, both hospitals stayed off bypass for all but 1.5 hours, which was down compared with the week before and the week after (47 and 56 hours). Additionally, the number of patients bypassed went down from an average of more than 40 to 2 patients per week. There were no statistical differences in ambulance runs or ED census. An oscillatory phenomenon of ambulance patient arrival was also noted to have been eliminated between these 2 facilities during the study week. This oscillatory phenomenon reflected the fact that when one hospital went on diversion, the other hospital then received a disproportionate flow of ambulance traffic and was itself forced to go on ambulance diversion shortly thereafter. As a result, an oscillatory pattern would establish in which one hospital and then the other would go on and off diversion status. EDs were reporting the use of defensive ambulance diversion to protect them from getting a disproportionate share of ambulances when neighboring facilities were on diversion, which inspired a second trial involving 5 hospitals located proximally in a geographic corner of the north and coastal portion of the county. 17 A similar study method was used with data collected a week before and after a trial week in which all 5 hospitals made a commitment to minimize diversion hours during the trial week, without changes in baseline staffing or expenses. Data were collected prospectively and Figure 2. Countywide total ambulance diversion hours for the 3 periods studied; San Diego County, CA, October 2001 through September ANNALS OF EMERGENCY MEDICINE 44:4 OCTOBER 2004

7 demonstrated a significant decrease in diversion hours and diverted patients during the trial week compared with the week before and the week after. Ambulance patients, ED census, ICU admissions, telemetry admissions, and medical and surgical admissions all remained consistent and stable throughout the 3 weeks of study. These 2 previous trials laid the basis for the current study. In our efforts during the first 3 months of this study, we were successful in decreasing ambulance diversion by 73%. We were also able to significantly decrease the number of patients (75%) who had requested a particular ED and could not reach it because of diversion status. We postulate that in doing so, we improved the timeliness and reduced the resources needed to care for patients in our community s EDs. By reducing diversion, patients were more likely to get to the hospital of choice where their medical records and private physicians would be available, which would theoretically improve ED utilization of existing beds by getting patients admitted sooner because of the availability of records and reduce the need to transfer to other hospitals. In addition, it would reduce the use of ED beds for patients monitored while waiting for interfacility transportation. For example, the system does not benefit from having a patient who was released from the hospital after surgery being sent to a neighboring hospital because the patient s own facility was on ambulance diversion. The examination would be delayed, records would not be easily available, and, if the patient required admission, the patient would need to be transferred back to the originating facility, occupying the ED bed even longer and perpetuating the ED congestion. This scenario was frequent before initiation of the trial. Our data demonstrate that interfacility transfers because of payer request also decreased during the period that had the decrease in diversion hours and numbers of diverted patients. There was a significant decrease in total interfacility transfers for payer request, with essentially no change in the numbers of patients being transferred for level of care indications. In other words, the overall decrease in the use of interfacility transfers comes mainly from the decrease in needing to transfer patients back to their home hospitals. When the data were evaluated more carefully, it was noted that approximately 70 to 80 patients a month who were transported to EDs by paramedics and required subsequent transfer for payer request were transferred to a facility that did not have an ED and thus could not take the patients directly. These data clearly have implications that affect financial issues and resource use. During the posttrial period, the numbers of hours on diversion and patients not getting to their requested facility increased slightly. Beyond seasonal changes (including increased transports during the winter flu seasons), there was in an increase in total ambulance transports during this time. Seasonal changes have been seen in other systems, as well. 18 Additionally, without as aggressive an approach in reporting the data and meeting as frequently, there might be a slight regression. However, Figure 3. Ambulance patients who made ED requests but who could not reach the requested ED because of diversion status for the study periods; San Diego County, CA, October 2001 through September OCTOBER :4 ANNALS OF EMERGENCY MEDICINE 301

8 the new average in the posttrial period is still almost 67% below the pretrial baseline for patients not reaching their requested facilities. The trend throughout the posttrial period remained relatively flat, and we do not anticipate a dramatic return to pretrial levels. Although deviations from the practice standards established occurred intermittently, such as facilities staying on diversion for several hours at a time, the peer-driven approach and reminders from the oversight committee have kept these events to a reasonable minimum. Previous work has been done in an attempt to determine factors that contribute to ED ambulance diversion. Although there have been a few community-wide studies of these factors, most studies have been limited to evaluating single institutions. 9,19,20 Silka et al 21 retrospectively reviewed reasons for ambulance diversion in a large urban setting, noting that diversion increases the total out-ofhospital interval and patient transfer intervals. Schull et al 19 noted that, in their system, admitted patients in the ED, and not nurse or physician staffing, were the most important determinants of ambulance diversion. Previous work that attempted to decrease ambulance diversion in a community was reported by Lagoe et al, 22 who reduced ambulance diversion hours 34% during a year in a community served by 4 hospitals. This was accomplished by sending daily diversion statistics to hospital chief executive officers and ED directors and managers, along with each hospital individually implementing its own measures to reduce hours. Table 3. Reported interfacility transfer data from the 17 hospitals* in aggregate form, y San Diego County, CA, August 2002 through November ED Information Pretrial Period Trial Period Difference (95% CI) ED visits 51,784 49,375 ÿ2,409 (ÿ8,449 to 3,631) Hospital admissions 8,036 7,967 ÿ69 (ÿ545 to 407) Patients transferred (all 1, ÿ175 (ÿ259 to ÿ91) reasons) Patients transferred for ÿ14 (ÿ212 to 184) level of care Patients transferred for ÿ141 (ÿ296 to 15) payer request Patients transferred for payer request brought in by paramedic ÿ100 (ÿ177 to ÿ23) *Two hospitals did not report admission data for both periods. y Data reported as monthly averages. A number of our initiatives to reduce diversion have been attempted in other localities. For example, a number of communities have enacted policies to honor patient requests regardless of diversion status and to limit the total time of diversion for each hospital. 9 A few large metropolitan areas have established oversight task forces to study and track the diversion issue in their communities. 9 However, this article represents the first published report of decreasing diversion hours using multiple initiatives on a community-wide basis in a county of this size. In summary, a community-wide effort to improve getting patients to requested EDs and decreasing ambulance diversion hours can be successful in a large community with an urban, suburban, rural, and remote population distribution. The success of such a process had the additional effect of decreasing the need for ED interfacility transfers for payer request reasons. We thank the San Diego County hospitals, emergency departments, paramedic base hospitals, advanced life support agencies, mobile intensive care nurses, and paramedics. We also thank the San Diego County Medical Society and the Healthcare Association of San Diego and Imperial Counties. Author contributions: GMV, RL, and TCC conceived the study and designed the trial. GMV, LUR, EMC, and PAM supervised the conduct of the trial and data collection. GMV, MAM, and RL undertook recruitment and education of the participating centers and agencies. EMC, TCC, and MAM managed the data, including quality control. EMC and LUR provided statistical advice on study design and analyzed the data. GMV and EMC drafted the manuscript, and all authors contributed substantially to its revision. GMV takes responsibility for the paper as a whole. Received for publication February 2, Revisions received April 1, 2004, and May 10, Accepted for publication May 11, Available online August 25, Presented at the Society for Academic Emergency Medicine annual meeting, Boston, MA, May 2003; and the American College of Emergency Physicians Research Forum, Boston, MA, October The authors report this study did not receive any outside funding or support. Reprints not available from the authors Address for correspondence: Gary M. Vilke, MD, Department of Emergency Medicine, University of California San Diego Medical Center, 200 West Arbor Drive, Mailcode #8676, San Diego, CA 92103; , fax ; gmvilke@ucsd.edu. REFERENCES 1. Asplin BR. Does ambulance diversion matter? Ann Emerg Med. 2003;41: Lagoe RJ, Jastremski MS. Relieving overcrowded emergency departments through ambulance diversion. Hosp Top. 1990;68: ANNALS OF EMERGENCY MEDICINE 44:4 OCTOBER 2004

9 3. Warden CR, Bangs C, Norton R, et al. Temporal trends in ambulance diversion in a mid-sized metropolitan area. Prehosp Emerg Care. 2003;7: Derlet RW, Richards JR. Overcrowding in the nation s emergency departments: complex causes and disturbing effects. Ann Emerg Med. 2000;35: Andrulis DP, Kellerman A, Hintz EA, et al. Emergency departments and crowding in United States teaching hospitals. Ann Emerg Med. 1991;20: Schull MJ, Szalai JP, Schwartz B, et al. Emergency department overcrowding following systemic hospital restructuring: trends at twenty hospitals over ten years. Acad Emerg Med. 2001;8: Reeder TJ, Garrison HG. When the safety net is unsafe: real-time assessment of the overcrowded emergency department. Acad Emerg Med. 2001;8: Schneider S, Zwemer F, Doniger A, et al. Rochester, New York: a decade of emergency department overcrowding. Acad Emerg Med. 2001;8: US General Accounting Office. Hospital Emergency Departments: Crowded Conditions Vary Among Hospitals and Communities: Report to the Ranking Minority Member, Committee on Finance, US Senate. Washington, DC: US General Accounting Office; GAO Redelmeier DA, Blair PJ, Collins WE. No place to unload: a preliminary analysis of the prevalence, risk factors, and consequences of ambulance diversion. Ann Emerg Med. 194;23: Neely KW, Norton RL, Young GP. The effect of hospital resource availability and ambulance diversions on the EMS system. Prehosp Disaster Med. 1994;6: Schull MJ, Morrison LJ, Vermeulen M, et al. Emergency department overcrowding and ambulance transport delays for patients with chest pain. CMAJ. 2003;168: Upfold J. Emergency department overcrowding: ambulance diversion and the legal duty to care. CMAJ. 2002;166: Anderson B. Fresno County bans diversion of ambulances. Sacramento Bee. February 25, 2003;sect A: Andrews W. Deadly ambulance diversions? [February 18, 2002, CBS News] Available at: main shtml. Accessed March 30, Vilke GM, Brown L, Skogland P, et al. Approach to decreasing emergency department ambulance diversion hours. J Emerg Med. 2004;26: Vilke GM, Loh A. A prospective study of minimizing ambulance diversion and its effects on emergency department census and hospital admissions [abstract]. Prehosp Emerg Med. 2003;7: Lagoe RJ, Hunt RC, Nadle PA, et al. Utilization and impact of ambulance diversion at the community level. Prehosp Emerg Care. 2002;6: Schull MJ, Lazier K, Vermeulen M, et al. Emergency department contributors to ambulance diversion: a quantitative analysis. Ann Emerg Med. 2003;41: Cameron P, Scown P, Campbell D. Managing access block. Austr Health Rev. 2002; 25: Silka PA, Geiderman JM, Kim JY. Diversion of ALS ambulances: characteristics, causes, and effects in a large urban system. Prehosp Emerg Care. 2001;5: Lagoe RJ, Kohlbrenner JC, Hall LD, et al. Reducing ambulance diversion: a multihospital approach. Prehosp Emerg Care. 2003;7: OCTOBER :4 ANNALS OF EMERGENCY MEDICINE 303