Ambulance Diversion as a Proxy for Emergency Department Crowding: The Effect on Pediatric Mortality in a Metropolitan Area Rohit P. Shenoi, MD, Long Ma, MS, Jennifer Jones, MS, Mary Frost ,RN, BSN, Munseok Seo, Dr PH, and Charles E. Begley, PhD
Abstract Objectives: The objective was to determine the prevalence of emergency department (ED) ambulance diversion among Houston pediatric hospitals and its association with mortality of pediatric patients. Methods: Hospital diversion and patient data between August 2002 and December 2004 were used to examine the impact of diversion on mortality of children under age 18 years. Patients were assumed to be exposed to ED crowding if diversion and admission or ED arrival times overlapped. Univariate and logistic regression were performed to determine if diversion was associated with mortality while controlling for age, illness severity, injury, and transfer status. Results: Mean hospital diversion hours as a percentage of operating hours were 10.58 (standard deviation [SD] ± 9). Overall, of 63,780 admissions, there were 4,095 (6.4%) children admitted during diversion. Fewer severely ill patients were admitted during diversion than nondiversion times (odds ratio [OR] = 0.72; 95% confidence interval [CI] = 0.66 to 0.78). The presence of diversion was protective for mortality (OR = 0.51; 95% CI = 0.34 to 0.77) in bivariate analysis. Mortality was associated with presence of major or extreme illness (OR = 60.7; 95% CI = 45.2 to 81.5), injury (OR=1.7; 95% CI = 1.4 to 2.1), and transfer status (OR = 6.3; 95% CI = 5.4 to 7.3). Using conditional logistic regression, major or extreme illness (OR = 50.7; 95% CI = 37.7 to 68.3), injury (OR 3.7; 95% CI = 2.9 to 4.7), and transfer (OR = 2.7; 95% CI = 2.2, 3.2) were associated with mortality, but diversion did not show any association with mortality. After combining ED and inpatient deaths, no association between diversion and mortality was observed. Conclusions: Hospital diversion due to ED crowding is common in pediatrics. The authors found no evidence of an association between diversion and ED and inpatient pediatric mortality. ACADEMIC EMERGENCY MEDICINE 2009; 16:116–123 ª 2008 by the Society for Academic Emergency Medicine Keywords: ambulance diversion, emergency department crowding, pediatric mortality
mergency departments (EDs) are a vital component in our health care safety net and are available at all times for those requiring care. ED crowding has been a serious issue for the past two decades and is rapidly becoming an increasingly significant problem affecting pediatric patients.1 Pediatric patients account for a quarter of all ED patients.2 Hospitals resort to diversion of emergency medical services (EMS) due to From the Section of Emergency Medicine, Department of Pediatrics, Baylor College of Medicine (RPS, LM, JJ), Houston, TX; the Emergency Center, Texas Children’s Hospital (MF), Houston, TX; and the School of Public Health, University of Texas Health Science Center (MS, CEB), Houston, TX. Received June 30, 2008; revision received September 30, 2008; accepted October 9, 2008. Presented at the Research Forum, American College of Emergency Physicians, Seattle, WA, October 2007. Address for correspondence and reprints: Rohit Shenoi, MD; e-mail: [email protected]
ISSN 1069-6563 PII ISSN 1069-6563583
various factors that cause ED crowding.3 Crowding leads to delays in treatment.4,5 Hospitals on diversion may decline or delay requests for transfer of patients from other hospitals, and these patients may need to be transported to other, more distant hospitals for care. The problem can be important for children treated at community hospitals who need transfer to pediatric hospitals for a higher level of care. The increasing frequency of ED diversion has led to concerns about its impact. Studies have examined the impact of EMS diversion on mortality,6–10 and on the volume of patients treated at overloaded hospitals.11,12 A recent study showed an increase in diversion in Houston from 1999 to 2002 and a possible association between diversion and mortality in adults.6 However, none of these possible impacts have been studied in children. To examine whether these circumstances exist for pediatric patients, we have tested the hypothesis that there is an association between diversion and mortality in Houston hospitals.
ª 2008 by the Society for Academic Emergency Medicine doi: 10.1111/j.1553-2712.2008.00317.x
ACAD EMERG MED • February 2009, Vol. 16, No. 2
The objectives of this study were twofold: 1) to investigate the magnitude of ambulance diversion at hospitals that admit children in Houston and 2) by using ambulance diversion as a proxy for ED crowding, to determine if there was a statistical association between ambulance diversion and pediatric mortality after controlling for age, illness severity, transfer, and injury status.
‘‘open’’ hospital and appropriate medical facility within 15 minutes’ transport from the incident location. If there are no ‘‘open’’ facilities within this time frame, the ambulance will be directed to the most appropriate facility, regardless of diversion status. At the time of this study, diversion requests managed by the HFD were honored for up to 8 hours from the time of notification. HFD seldom transports patients to hospitals outside their jurisdiction. There are special policies for the transport of pediatric patients.14 Patients with moderate or serious illness are transported to hospitals with PICU facilities. In non–trauma-related situations with cardiopulmonary resuscitation (CPR) in progress or with an uncontrollable airway, patients are transported to the closest facility regardless of diversion status with two exceptions: first, if the closest hospital is on diversion due to internal disaster, and second, if a second hospital that is open is nearly as close as the one on diversion. Seriously and critically injured trauma victims are transported directly to the Level 1 centers. Patients with trauma-related CPR in progress or an uncontrollable airway are transported to the nearest appropriate medical facility. Although there is a policy for diversion due to trauma saturation, pediatric trauma patients are exempted from this policy. Decisions on accepting interhospital transfer requests must be made by the destination hospital within 30 minutes. Destination hospitals can decline transfers due to divert status. Transferring hospitals are responsible for locating a suitable hospital that will accept patients who need a higher level of care.
METHODS Study Design This was a cross-sectional study conducted to review hospital patient data and ambulance diversion data. The study was approved and granted exempt status by the institutional review board of Baylor College of Medicine. Study Setting and Population We studied Houston area hospitals that treated pediatric patients in the ED or as inpatients. The city of Houston is a 600 square mile area with a population of 2 million people. There are approximately 540,000 children under the age of 18 years.13 The Houston Fire Department ⁄ Emergency Medical Services (HFD ⁄ EMS) is the sole EMS provider for the city of Houston. HFD along with other EMS agencies provide emergency services to some hospitals outside of the city limits. The hospitals that admitted children consisted of one children’s hospital with a dedicated ED and pediatric intensive care unit (PICU), two Level 1 trauma hospitals with a separate pediatric ED and PICU, and eight general hospitals with a common ED for children and adults. Patients who require admission for subspecialty and critical care are transferred to one of the three tertiary care hospitals. These hospitals are situated within a half-mile radius of each other. About 5%–9% of pediatric ED patients in these hospitals are transported by EMS. There were 12 other general hospitals without inpatient pediatric facilities that treat children in the ED. Houston Fire Department follows standard protocols with online medical direction.14 The severity of illness in pediatric patients is based on patient’s vital signs in the field. Under the local diversion policy, hospitals can request ED diversion of EMS for the following reasons: ED saturation (ED is heavily saturated and noncritical patients will have to wait excessively long periods before receiving treatment), critical medical saturation (the facility does not have the capacity or capability to accept additional critical medical patients), internal disaster (an environmental or physical plant situation that disrupts the staff’s ability to provide care), and trauma saturation (the facility does not have the capacity or capability to accept additional critical trauma patients).15 The decision to request diversion is made by each hospital independently, and the criterion on which this is based varies with each hospital. Diversion status is communicated by the respective hospitals to a central regional website that alerts EMS providers of hospital divert status. Diversion requests are honored, provided that the ambulance estimates that it can reach an
Selection of Subjects and Participating Hospitals Two groups of patients were studied: inpatients and patients who died in the ED. Using the Texas Health Care Information Collection (THCIC) hospital inpatient database, we selected children up to age 18 years who were admitted to all hospitals in the Houston-SugarlandBaytown metropolitan statistical area during August 15, 2002, through December 31, 2004. We then selected only those hospitals that accepted patients transported by the HFD. This consists of hospitals in the city of Houston and a few neighboring areas. This study group consisted of all children who needed emergency or urgent admission to these hospitals, including transfers from other hospitals within or outside the metropolitan area. We excluded newborn admissions in maternity hospitals, pediatric admissions to specialty hospitals (psychiatry and oncology), and elective admissions. The second group of patients consisted of children up to age 18 years who died in Houston-area hospital EDs served by the HFD for the same study period. These patients are not admitted to hospital. Their death records were obtained from the Harris County Child Fatality Review Team. Study Protocol We obtained data from three sources: 1) EMSystems (Milwaukee, WI), which records the hours each hospital in the region is on ambulance diversion; 2) THCIC database, which is a record of deidentified inpatient
Shenoi et al.
administrative hospital data;16 and 3) Harris County Child Fatality Review Team for ED death reports. Emergency department diversion requests are tracked by the HFD indicating the exact time each hospital is on diversion for any reason. We collected data to ascertain the reason why each hospital resorted to diversion. However, since there were multiple reasons for single diversion episodes, we were unable to analyze the effects of specific reasons for diversion. Texas Health Care Information Collection discharge data for the same period as the diversion data were used to examine patient characteristics and outcomes. All licensed hospitals in Texas are required to submit discharge data to the THCIC. The data contain information on patient demographics, hour and date of admission, admission source, length of stay, discharge status (alive or dead), diagnosis (including primary and secondary International Classification of Diseases, 9th Revision, Clinical Modification [ICD-9CM] and diagnosis-related group [DRG] codes), charges, payment source, and procedure codes. The data also contain the severity of illness measure based on the all patientrefined DRG (APR-DRG)17 scale with four levels: minor, moderate, major, or extreme. The APR-DRG severity of illness scale categorizes severity on the basis of the hospital ICD-9CM discharge diagnosis and procedure codes. The scale is developed for a patient according to the presence of secondary diagnoses and procedures associated with physiologic decompensation or organ system loss of function. A patient’s severity of illness with a specific condition is major if there is significant organ system loss of function. Patients with multiple disease or injury conditions along with significant organ system loss of function are considered to have extreme severity of illness. We decided a priori to dichotomize patients into two groups based on their APR-DRG: mild ⁄ moderate and major ⁄ extreme, because moderate to severely ill patients are transported directly to pediatric tertiary care centers by the HFD. Hospitalized patients transferred from other hospitals were identified in the discharge database. Transferred patients were included for the destination hospital. For this study, only hospital-to-hospital transfers were counted because such transfers typically are made to facilitate a higher level of care for acutely injured and sick patients. We did not include transfers from hospitals to hospice care. The inpatient database does not specify if the transfers occur directly between hospital wards. Data processing and quality checks performed for THCIC data are described elsewhere.16 The accuracy of THCIC data was assessed by comparing the number of admissions with the admission record statistics of one hospital during the study period. THCIC admissions numbered 18,768 for the period from October 10, 2003, through December 31, 2004. During the same period, there were 19,192 admissions reported by the same hospital. This corresponds to a difference of 2.2% between the two data sets. The reason for this difference is that data on some patients needed to be modified by the hospital after it had been sent to THCIC. Noninclusion of this modified data in the final THCIC data led to slightly lower numbers in the THCIC database.
EFFECT OF AMBULANCE DIVERSION ON PEDIATRIC MORTALITY
Data of patients who died in the ED were collected on a standardized abstraction form and entered into an Access database (Microsoft Corp., Redmond, WA). Information about the patients’ age, gender, ethnicity, hospital, date and time of arrival and death in the hospital ED, presence of injury, and diagnosis was obtained. Information about transfer status was unavailable. We used the following definitions: 1) Ambulance diversion included instances when a hospital was on ambulance diversion for any reason. 2) Mortality represented inpatient or ED mortality. It was measured according to the day of admission and represented the number of patients admitted on that day who died before discharge or in the ED. Deaths during EMS transfer were not included. 3) Injury cases included those patients with ICD-9CM codes 800 to 959.99, except 905–909 (late effects of injuries), 930–939 (effects of foreign bodies entering through orifice), 958 (traumatic complications), and 820–820.99 (hip fractures in elders). These are widely accepted codes for trauma cases.18 4) Pediatric tertiary care hospital included hospitals with a PICU. Other hospitals were grouped as general hospitals. Data Analysis Inpatients. Emergency department diversion and THCIC hospital discharge data were combined into a single hospital-specific database covering the study period. Patients were divided into two groups based on the presence of diversion during the time of admission. The exact time of ED arrival is not available, and admission time is available in only 1-hour intervals. For the purposes of analysis, we assumed that patients were exposed to diversion if the hospital’s time of diversion overlapped19 the patient’s hour of admission by at least 30 minutes. This was chosen to have a robust definition for exposure to diversion. Descriptive statistics were performed to compare demographics, admissions during diversion, illness severity, presence of injury, and mortality by type of admitting hospital. Age, gender, race, illness severity, injury, transfer, and diversion status were coded as categorical variables. Pearson chi-square and odds ratios (ORs) with 95% confidence intervals (CIs) were derived for 2 · 2 tables relating those who died or were discharged alive to diversion. A statistical significance level of