A Long-term Analysis of Physician Triage Screening in the Emergency Department

Transcription

1 ORIGINAL RESEARCH CONTRIBUTION A Long-term Analysis of Physician Triage Screening in the Emergency Department Jonathan G. Rogg, MD, Benjamin A. White, MD, Paul D. Biddinger, PhD, Yuchiao Chang, MD, and David F. M. Brown, MD Abstract Objectives: The problem of emergency department (ED) crowding is well recognized; however, little data exist on the sustainability of potential solutions, including physician triage and screening. The authors hypothesized that a physician triage screening program (Supplemented Triage and Rapid Treatment [START]) sustainably improves standard ED performance metrics. Methods: This retrospective, observational, before-and-after study compared performance measures over 4 years in a tertiary care urban academic medical center with approximately 90,000 annual ED visits. Patients seen between December 2006 and November 2010 were included. Outcome measures included length of stay (LOS) for ED patients, percentage of patients who left without completing assessment (LWCA), percentage of patients treated and dispositioned by START without using monitored beds, and door-to-room time. Descriptive statistics were used. Results: Median LOS for START patients was 56 minutes/patient lower when comparing 2010 to 2007 (p < ) and for non-start patients 22 minutes/patient lower (p < ). The percentage of patients who LWCA decreased from 4.8% to 2.9% (p < ) during the same time period. In START s first halfyear, 18% of patients were discharged without using monitored beds. This increased to 29% by year 3. In addition, median door-to-room time decreased from 18.4 to 9.9 minutes during the same 3-year interval. Conclusions: Physician screening appears to provide sustainable improvements in ED performance metrics including ED LOS, percentage of patients who LWCA, door-to-room time, and percentage of patients treated without using a monitored bed, despite increasing ED volume. Physician screening delivers additional incremental benefits for several years after implementation and can effectively increase ED capacity by allowing emergency physicians to more efficiently use monitored beds. ACADEMIC EMERGENCY MEDICINE 2013; 20: by the Society for Academic Emergency Medicine Emergency department (ED) visits continue to rise in the United States, with the million visits in 2007 representing a 23% increase over the From the Harvard Affiliated Emergency Medicine Residency (JGR), Boston, MA; the Department of Emergency Medicine (BAW, PDB, DFMB), General Medicine Division, and the Department of Medicine (YC), Massachusetts General Hospital, Boston, MA; the Division of Emergency Medicine (BAW, PDB, DFMB) and the Department of Medicine (YC), Harvard Medical School, Boston, MA; and the Department of Health Policy and Management, Harvard School of Public Health (PDB), Boston, MA. Received July 6, 2012; revision received September 27, 2012; accepted October 16, Presented at the Society for Academic Emergency Medicine annual meeting, Chicago, IL, May The authors have no relevant financial information or potential conflicts of interest to disclose. Supervising Editor: Jesse Pines, MD. Address for correspondence and reprints: David F.M. Brown, MD; dbrown2@partners.org. prior 10 years. 1 In addition, given continued capacity limitations, ED crowding remains a significant problem that has detrimental effects, including increased wait times, decreased staff and patient satisfaction, and most concerning, poor patient outcomes. 2 7 While it is now widely accepted that ED crowding is a result of hospital crowding, and specifically the practice of boarding inpatients in the ED, recent evidence indicates that new intake models may represent part of the overall crowding solution. 8 In the conceptual framework for ED process flow (the input-throughput-output model) of Asplin et al., 9 these measures tend to focus on throughput, an area of focus more readily within the sphere of influence of emergency physicians. 9,10 As an example, physician triage models have been shown to reduce waiting times in the ED We have previously reported our experience with a physician screening program known as START (Supplemented Triage and Rapid Treatment) comparing a 3-month time period before and after implementation of the program and, separately, the financial ramifications of such a program. 18,19 However, most of the published research of physician screening throughput interventions is limited by small ISSN by the Society for Academic Emergency Medicine 374 PII ISSN doi: /acem.12113

2 ACADEMIC EMERGENCY MEDICINE April 2013, Vol. 20, No sample size and brief study periods; as a result the magnitude and sustainability of the gains of such intake programs remain unknown. 3,20 Changes to ED systems and process flow require significant time, effort, and expense. As such, ED leaders would benefit from knowing whether the positive effect of an intervention was likely to be sustainable over time. If a physician screening system such as START results in sustained improvement in ED throughput, there are likely to be multiple direct and indirect benefits, including decreased waiting times, reductions in length of stay (LOS), improved patient satisfaction and safety, and increased ED revenue. However, at present, there is no literature reporting the sustainability of these interventions, and ED leaders thus cannot truly evaluate physician screening. In this study, we examined the hypothesis that a physician screening program (START) sustainably improves standard ED performance metrics, including patient LOS, and percentage of patients who leave without completing assessment (LWCA). Based on our previous research, we anticipated that after 3 years, our initial LOS reduction of 29 minutes would persist and patients who LWCA would remain decreased by 1.7% compared to the period prior to physician screening. 18 We also measured the number of patients treated and dispositioned by START without using monitored beds and the median patient door-to-room time and trended these metrics over time to further define the effects of START on ED throughput. METHODS Study Design This retrospective, observational, before-and-after study compared performance measures over 4 years (December 2006 through November 2010). This period included 3 years of physician screening and 1 year prior to the intervention and was chosen to provide adequate sample size and sufficient time to demonstrate sustainability of the effects of physician screening. The ED underwent physical expansion in summer 2011, so this 4-year study period represents the largest consecutive interval after initiation of START without expansion of the ED. The ED staff and all participants were unaware of the data collection or analysis, but could not be blinded to the intervention. The study had institutional review board approval, qualifying as exempt from full review as no protected health information was collected and only aggregate data were used. Study Setting and Population The study was performed in a large, urban, academic tertiary care ED with an annual visit volume of approximately 90,000. The hospital is a Level I trauma center for adult and pediatric patients as well as a regional burn center. The admission rate is approximately 26%, and approximately 31% arrive by ambulance. Study Protocol Patients triaged directly to the fast track (i.e., low acuity) section of the ED, to the acute area (those with unstable vital signs or high acuity), or to the hospital s separate acute psychiatric service (APS) were excluded from the primary analysis. All other patients presenting to the ED during the study period, December 2006 through November 2010, were included in the analysis. A subset of data for secondary outcome analysis, patients dispositoned directly from START without using monitored beds, and patient door-to-room time was collected for 3 years between April 2008 and March Our ED information system (EDIS) could not collect these data prior to April START began at our institution in December Data were obtained from a computerized EDIS. This is a home-grown ED tracking system and database. Demographic data were collected for all ED patients eligible for START. In addition, ED daily volume and boarding times were recorded. ED LOS was defined as the time interval between patient registration and patient disposition (discharge, admission, or observation). A monitored bed is defined as a bed that has cardiac and pulse oximetry monitoring capabilities. Our definition of a boarding patient was any patient in the ED for more than 2 hours after an admission bed request had been made. Details of the START process can be found in our previous study. 18 Briefly, the START intervention was designed by a multidisciplinary group of physicians, nurses, administrators, and support staff over a 12-month period, building on models and initiatives developed in other institutions. START is a multifaceted system, designed to facilitate and accelerate 1) identification of subtle presentations of life threats, 2) initiation of appropriate diagnostic testing, 3) administration of important therapies such as analgesia and antibiotics, and 4) identification of a subset of patients whose disposition (either admission or discharge) may be determined after a brief clinical evaluation. The START system was first deployed in December 2007 and operates 12 hours per day during the hours of highest ED arrival volume, from 11:00 a.m. and 11:00 p.m., 7 days per week. During the hours when START is not operational, standard triage and patient flow systems are used. Attending physicians are designated to staff this area in two shifts, with a change of shift at 3:30 p.m., and were present in both the preintervention and the postintervention periods. There were no other significant and identifiable operations changes affecting patient flow between the two study periods. Although the system has multiple operational components, START has two major clinical functions, greeting and screening. Greeting. Patients who do not arrive by ambulance are first met by a greeter nurse. The greeter performs a brief evaluation (under 2 minutes) and determines whether the patient should receive immediate evaluation or intervention in the acute care area, needs an electrocardiogram within 10 minutes, requires isolation for infection control or the potential for violence, or is the victim of sexual abuse. If the patient meets any of the above criteria, the greeter nurse immediately sends the patient to the appropriate destination in the ED. If the patient meets none of the above criteria, the greeter sends the patient on to a triage station for standard triage, including vital signs.

3 376 Rogg et al. A LONG-TERM ANALYSIS OF PHYSICIAN TRIAGE SCREENING Screening. After triage, all patients except those triaged directly to the fast track area (nonemergent) or APS area receive a screening examination by a physician in one of four dedicated screening rooms. These walled rooms have limited monitoring and examination supplies and doors that can close to facilitate patient privacy during physical examination. After examination and evaluation, the physician has two tasks: 1) ordering appropriate laboratory and radiographic studies, as well as important early therapies such as antipyretics, antiemetics, or antibiotics, and 2) identification of patients whose disposition may be accelerated without further need for work-up in the clinical areas of the ED. Every patient triaged to START is seen by a physician and evaluated, and no standing orders are used. After being screened, patients wait in a dedicated internal waiting room area termed postscreening, staffed by nurses who initiate the therapies and interventions ordered by the physician. This area consists of a large room lined with chairs, converted from previous space, and has a nursing station and a work station with three computer terminals. In addition, the adjacent hallway space is used for patients requiring stretchers during their evaluations. Patients can be discharged or admitted directly from postscreening without further evaluation in a clinical unit of the ED and without requiring monitored beds. Patients not dispositioned from this area are cared for in ED clinical units as beds become available. Generally patients who require private rooms may have their work-ups initiated in START but are not eligible for disposition from START. Examples of patients who need private rooms for reasons beyond cardiorespiratory monitoring include victims of sexual assault, patients in poor behavioral control, patients with significant physical care needs (e.g., frequent incontinence, vomiting, complete immobility, and no support), patients who require medical procedures (e.g., suturing, splinting), and patients who require isolation for immunosuppression or infection control reasons. While imaging and laboratory testing may be ordered from START, these patients are given, or are put in the queue, for a private ED bed. Many, but not all patients are disrobed. All patients can be fully undressed for examination but it is dependent on the complaint and history of illness. Patients in START, like all other patients in the ED, are reassessed and can be retriaged based on their needs. Sometimes patients seen in START are down-triaged to fast track if the screening physician feels they can be appropriately treated there. Patients who the screening physician feels are unstable or who become unstable in START are retriaged to the acute area of the ED. During the study period there were relatively small changes in staffing. For implementation of START, 12 hours per day of nursing time was added for the greeting function. In 2009, an additional 12 hours per day of nursing time was added, and in 2010 enough nurses were hired to increase nursing staff by one nurse 24 hours per day. An attending physician is always on duty in START with a midlevel provider (nurse practitioner or physician assistant [NP/PA]). The NP/PA coordinates testing, consultation, and disposition of patients seen by an attending physician. During the study period midlevel providers did not see new patients. Postgraduate year (PGY)-3 and PGY-4 residents work in START when there are extra residents; however, START shifts often do not have residents. From its inception, attending physicians were told START had the four priorities mentioned. Initially there was a lot of heterogeneity in the performance of the group. Over the study period, there have been many multidisciplinary groups convened to analyze and reanalyze the process, and several best practices have been identified. Those best practices have been shared in faculty meetings, via , and in targeted one-on-one feedback sessions. We continue working on improving START. There are no other formal guidelines, however, for the specifics of the screening intervention. Measures The primary outcome measure was ED LOS for all START patients. Secondary outcomes included percentage of patients who LWCA, door-to-room time, and the percentage of START patients directly dispositioned from START without using monitored beds. An additional analysis describes LOS reductions for all ED patients and for non-start patients and can be found in Data Supplement S1 (available as supporting information in the online version of this paper). Data Analysis Median and interquartile ranges (IQRs) were used to summarize the distribution for continuous variables while count and percentage were used for categorical variables. Wilcoxon rank sum tests were used to compare ED LOS between the periods before and after intervention implementation, while chi-square tests were used to compare the percentage of patients who LWCA. Because some patients had multiple ED visits before and after START implementation, we also presented results from the cluster analysis, which used the generalized estimating equations (GEEs) techniques with compound symmetry covariance structures to account for the repeated-measures data. For ED LOS, we also conducted a multivariable regression analysis to compare the third year after implementation to the year before START was initiated, controlling for patient age, sex, hospital billing level, and the ED volume at the time of patient visit. The improvement between START and non-start was compared by testing the interaction between period and patient group. A p 0.05 was considered as statistically significant. With the large sample size available for analysis, the study has 80% power to detect a difference as small as 0.4% in LWCA. Therefore, we also noted whether the difference reached clinical importance. All data analyses were conducted using SAS version 9.3 (SAS Institute, Cary, NC). RESULTS Table 1 provides characteristics for those patients eligible for START from the year before implementation (December 2006 through November 2007) and for those who participated in START after the implementation (December 2007). During the study period, both ED

4 ACADEMIC EMERGENCY MEDICINE April 2013, Vol. 20, No Table 1 START Demographics and ED Characteristics Variable December 2006 November 2007 December 2007 November 2008 December 2008 November 2009 December 2009 November 2010 START volume 39,142 42,723 48,756 50,249 Age (yr), median (IQR) 43 (24 61) 43 (24 61) 41 (22 60) 43 (23 61) Male, population (%) 19,245 (49.2) 20,952 (49) 24,811 (50.9) 25,283 (50.3) Hospital billing level, n (%) 1 65 (0.2) 115 (0.3) 228 (0.5) 252 (0.5) 2 1,975 (5) 2,578 (6) 3,447 (7.1) 3,236 (6.4) 3 7,714 (19.7) 9,226 (21.6) 12,449 (25.5) 13,148 (26.2) 4 10,677 (27.3) 11,114 (26) 12,252 (25.1) 11,974 (23.8) 5 16,528 (42.2) 18,004 (42.1) 18,412 (37.8) 19,949 (39.7) Critical care 452 (1.2) 469 (1.1) 385 (0.8) 888 (1.8) Critical care not available 1,731 (4.4) 1,217 (2.8) 1,583 (3.2) 802 (1.6) Arrival by ambulance, n (%) 11,409 (29.1) 12,028 (28.2) 13,377 (27.4) 14,703 (29.3) Disposition, n (%) Discharged 22,774 (58.2) 26,032 (60.9) 31,054 (63.7) 30,995 (61.7) Admitted to hospital 11,210 (28.6) 11,881 (27.8) 12,725 (26.1) 12,758 (25.4) ED observation unit 3,231 (8.3) 3,427 (8) 3,604 (7.4) 4,986 (9.9) Other 1,917 (4.9) 1,383 (3.2) 1,373 (2.8) 1,510 (3) ED characteristics Total ED volume, n 81,310 85,205 91,212 91,176 Boarders per day, 44 (39 49) 44 (37 50) 43 (36 50) 43 (35 48) median (IQR) Boarding hours per ( ) ( ) ( ) ( ) day, median (IQR) Boarding hours per patient, median (IQR) 2.31 ( ) 2.21 ( ) 2.05 ( ) 2.08 ( ) Beginning of START was December IQR = interquartile range; START = Supplemented Triage and Rapid Treatment (physician triage screening program). volume and the volume of patients eligible for START increased. Annual ED volume rose to 91,176 in the last year of the study, a 12% increase from the year prior to START. The number of patients eligible for START increased 28% from 39,142 patients December 2006 to December 2007 to 50,249 patients December 2009 to December Patient demographics were stable during the 4-year period, with median age around 43 years and the percentage of males at 49%. Disposition for patients seen in START is provided in Table 1. For all ED patients, median boarding hours per patient decreased (2.31 hours in December to 2.08 hours in December ) although the median number of ED boarders was similar (43 to 44 patients per day) in all years of the study. During the 4-year period, the median LOS for START ED patients decreased by 56 minutes/patient (p < from both Wilcoxon rank sum test and taking into account clustering) when comparing the final year to the year before START was initiated (Table 2). In the multivariable repeated-measures regression model controlling for patient age, sex, hospital billing level, and the ED volume at the time of patient visit, START implementation remained a significant predictor for reducing ED LOS (p < ). Discharged patients had the largest decrease in median LOS, showing a reduction of 60 minutes in December compared with December There was a reduction in START ED patients who LWCA from 4.8% to 2.9% (p < from both chi-square test and taking into account clustering) during the same time period. Median LOS for all ED patients decreased from 292 minutes 1 year prior to START to 262 minutes in the third year after START (p < ; see Data Supplement S1). For patients not eligible for START during that period, median LOS decreased by 22 minutes (232 minutes to 210, p < ; see Data Supplement S1). LOS decreased by 34 additional minutes in START patients (56 minutes vs. 22 minutes, p < ). In START s first half-year, 18% of patients were discharged without using monitored beds. This percentage increased to 29% by year 3 (Figure 1A). In addition, median door-to-room time decreased from 18.4 minutes to 9.9 minutes during the same 3-year interval (Figure 1B). DISCUSSION Physician screening in many smaller studies has shown reductions in LOS for patients, although all previous studies have examined physician screening over a short period of time and cannot be used to assess the sustainability of this intervention. 12,13,15,16,18 However, at least one study showed only mild effect in LOS for most patients and negligible effect for admitted patients. 21 In our study, the first year of physician screening showed only a moderate reduction in overall LOS (362 to 342 minutes). However over the entire study period median LOS for ED patients decreased to 306 minutes, a 56-minute reduction compared to before START was implemented. This corresponds to a reduction of 100

5 378 Rogg et al. A LONG-TERM ANALYSIS OF PHYSICIAN TRIAGE SCREENING Table 2 LOS for START Patients and Patients Who LWCA From 1 Year Before START and 3 Years After Implementation Variable December 2006 November 2007 December 2007 November 2008 December 2008 November 2009 December 2009 November 2010 Difference* p-value ED length of stay overall 362 ( ) 342 ( ) 310 ( ) 306 ( ) 56 < Discharged patients 317 ( ) 293 ( ) 263 ( ) 257 ( ) 60 < Admitted patients 461 ( ) 456 ( ) 431 ( ) 425 ( ) 36 < Other disposition 209 ( ) 152 (82 267) 139 (74 247) 139 (71 251) 70 < LWCA 4.80% 3.10% 2.70% 2.90% 1.90% < Beginning of START was in December IQR = interquartile range; LOS = length of stay; START = physician triage screening program. *Difference is comparing year prior to START to the most recent year. A Percent B Minutes 35% 30% 25% 20% 15% Apr'08-Sep'08 Oct'08-Mar'09 Apr'09-Sep'09 Oct'09-Mar'10 Apr'10-Sep'10 Oct'10-Mar'11 Date Apr'08-Sep'08 Oct'08-Mar'09 Apr'09-Sep'09 Oct'09-Mar'10 Apr'10-Sep'10 Oct'10-Mar'11 Date Figure 1. (A) START dispositions without using monitored beds. (B) ED arrival to room time. START = Supplemented Triage and Rapid Treatment. patient waiting hours per day compared to the time period prior to the implementation of physician screening. This reduction occurred despite a 12% increase in total ED volume, and a 28% increase in volume of patients seen in START. The ED saw decreased LOS during the interval studied; however, patients in START had the most significant reduction in LOS (56 minutes for START patients vs. 22 minutes for non-start patients). It is unclear whether START was responsible for the reduction in LOS for non-start patients or whether other ED process improvements contributed to this decrease in LOS. Regardless, the LOS improvement in START patients was more than double that of non- START patients, with a net difference of 34 minutes. The percentage of patients who LWCA decreased after the implementation of physician screening. This is consistent with previous studies of physician triage. 15,16,21 This may be related to the fact that patients were being placed in rooms more quickly, as decreases in waiting time have been shown to decrease the percentage of patients who LWCA. 22 Reducing the number of patients who LWCA is an important factor of quality for both physicians and patients, resulting in higher patient satisfaction as well as adherence to federally mandated emergency care. 23 Prior to initiation of START, few non fast track patients were seen and discharged by physicians in our ED without using monitored beds. The ability of the screening physician to see patients as they arrive and order appropriate testing allows patients to have their work-ups started even when there is significant ED boarding. Soon after the initiation of START, 18% of patients were dispositioned without using monitored beds. After 3 years of START, 29% of patients seen at START were dispositioned without using monitored beds. We hypothesize that better use of often scarce resources like monitored beds is helpful to reduce crowding. Studies have shown that fast track work silos can decrease resource use. 24,25 It is likely that identifying patients who do not require monitored beds, or even stretchers in some cases (i.e., keeping vertical patients vertical ), might have a similar effect, although we did not directly examine resource use in this study. Still, more efficient use of monitored beds effectively increases ED capacity to take care of sicker patients and the increased ED volume that has become a national phenomenon. We did see an increase in the percentage of patients using the START system over the course of this study period. In the year prior to START, 48% of ED volume would have been eligible for START. In the

6 ACADEMIC EMERGENCY MEDICINE April 2013, Vol. 20, No third year of START, 55% of ED volume went through START. We can only speculate that other factors such as crowding in other areas, or increased staff comfort with the START system, led to this rise. Additionally, median door-to-room time decreased from 18.4 minutes during the half year beginning April 2008 to 9.9 minutes for October 2010 through March This is a 46% reduction in median waiting time for patients and is consistent with other metrics of this study, which show additional improvements over time. Other studies have shown physician triage to reduce time spent waiting to see a physician Shorter door-to-room time or door-to-doctor time should have a positive effect on patient safety. In addition, patients tends to be more satisfied when they are quickly seen by a provider, and while we did not directly measure door-to-doctor time, in our center the two are closely correlated. 26 To our knowledge, no other study has shown that physician screening provides sustainable gains. During our 3-year intervention period, START provided significant initial gains to our ED metrics. Equally as important, physician screening continued to show incremental gains over the study period, despite increasing ED volume. LIMITATIONS First, this was a single site study and our findings may not be generalizable to all other ED settings. Our ED triages patients with minor complaints to a fast track area, and those with primary psychiatric complaints to a separate area, which may be different from other EDs. However, ED crowding is ubiquitous, particularly in large urban EDs, and our findings should at least be informative for those institutions. Second, it is possible other ED changes during the study period could have affected the results, and with this study design we cannot prove cause and effect, only an association. There were, however, no other significant changes to our ED system internally, or to our patient population, during the time period studied. Our hospital does work on constantly improving quality. There were no significant changes in ED or hospital resources. For example, there were no increases in ED or hospital beds, and no increase in radiology or laboratory resources. Incremental improvements in our EDIS continued during the study period, and initiatives focused on important diagnoses such as ST-segment myocardial infarction, sepsis, pneumonia, and cerebral vascular accidents continued as well. Externally, ambulance diversion was abolished during the study period, which may explain the increase in patient volume. However, that change would be expected to increase crowding with a concomitant deleterious effect on the ED performance metrics measured; as such, if anything this strengthens our findings. There were limited increases in nursing staffing as mentioned in study protocols. No physician staffing was added. Third, it was not possible to blind physicians and staff to the implementation of START. While a component of the Hawthorne effect is possible, the length of the study period makes this unlikely. Additionally, as a number of patients were not placed in monitored beds, it is possible that a triage bias might cause the ED to consider these patients less sick and perform less testing. Fourth, we have only studied a limited number of metrics, including LOS, percentage of patients who LWCA, and door-to-room time. These are not actual measures of ED patient care quality, and while crowding has been noted to be inversely correlated with quality we did not directly assess quality of our patient care. However, we believe that our metrics are valid surrogates that relate to ED quality. Finally, we did not have a way to precisely measure door-to-doctor time and instead chose as our proxy door-to-room time, based on when a patient actually enters a screening room. While this may not be the exact time a physician sees the patient, the same criterion is used consistently and trends in the data would contain the same amount of systematic error. CONCLUSIONS This physician screening system demonstrates an association with sustainable improvements in ED performance metrics including ED length of stay, percentage of patients who left without completing assessment, doorto-room time, and the percentage of patients treated without using monitored beds, despite increasing ED patient volume. This study suggests that physician screening delivers additional incremental benefits for several years after implementation and that physician screening can effectively increase ED capacity by allowing emergency physicians to more efficiently use monitored beds. References 1. Niska R, Bhuiya F, Xu J. National Hospital Ambulatory Medical Care Survey: 2007 emergency department summary. Natl Health Stat Rep. 2010; 26: Derlet RW, Richards JR. Overcrowding in the nation s emergency departments: complex causes and disturbing effects. Ann Emerg Med. 2000; 35: Wiler JL, Gentle C, Halfpenny JM, et al. Optimizing emergency department front-end operations. Ann Emerg Med. 2010; 55: Liu S, Hobgood C, Brice JH. Impact of critical bed status on emergency department patient flow and overcrowding. Acad Emerg Med. 2003; 10: Pines JM, Hollander JE. Emergency department crowding is associated with poor care for patients with severe pain. Ann Emerg Med. 2008; 51: Sprivulis PC, Da Silva JA, Jacobs IG, Frazer ARL, Jelinek GA. The association between hospital overcrowding and mortality among patients admitted via Western Australian emergency departments. Med J Austral. 2006; 184: Richardson DB. Increase in patient mortality at 10 days associated with emergency department overcrowding. Med J Austral. 2006; 184: Eitel DR, Rudkin SE, Malvehy MA, Killeen JP, Pines JM. Improving service quality by understanding emergency department flow: a White Paper and position statement prepared for the American Academy of Emergency Medicine. J Emerg Med. 2010; 38:70 9.

7 380 Rogg et al. A LONG-TERM ANALYSIS OF PHYSICIAN TRIAGE SCREENING 9. Asplin BR, Magid DJ, Rhodes KV, Solberg LI, Lurie N, Camargo CA Jr. A conceptual model of emergency department crowding. Ann Emerg Med. 2003; 42: Asaro PV, Lewis LM, Boxerman SB. The impact of input and output factors on emergency department throughput. Acad Emerg Med. 2007; 14: Terris J, Leman P, O Connor N, Wood R. Making an IMPACT on emergency department flow: improving patient processing assisted by consultant at triage. Emerg Med J. 2004; 21: Choi YF, Wong TW, Lau CC. Triage rapid initial assessment by doctor (TRIAD) improves waiting time and processing time of the emergency department. Emerg Med J. 2006; 23: Subash F, Dunn F, McNicholl B, Marlow J. Team triage improves emergency department efficiency. Emerg Med J. 2004; 21: Travers JP, Lee FC. Avoiding prolonged waiting time during busy periods in the emergency department: is there a role for the senior emergency physician in triage? Eur J Emerg Med. 2006; 13: Holroyd BR, Bullard MJ, Latoszek K, et al. Impact of a triage liaison physician on emergency department overcrowding and throughput: a randomized controlled trial. Acad Emerg Med. 2007; 14: Partovi SN, Nelson BK, Bryan ED, Walsh MJ. Faculty triage shortens emergency department length of stay. Acad Emerg Med. 2001; 8: Han JH, France DJ, Levin SR, Jones ID, Storrow AB, Aronsky D. The effect of physician triage on emergency department length of stay. J Emerg Med. 2010; 39: White BA, Brown DFM, Sinclair J, et al. Supplemented triage and rapid treatment (START) improves performance measures in the emergency department. J Emerg Med. 2010; 42: Soremekun OA, Biddinger PD, White BA, et al. Operational and financial impact of physician screening in the ED. Am J Emerg Med. 2012; 30: Rowe BH, Guo X, Villa-Roel C, et al. The role of triage liaison physicians on mitigating overcrowding in emergency departments: a systematic review. Acad Emerg Med. 2011; 18: Han JH, France DJ, Levin SR, Jones ID, Storrow AB, Aronsky D. The effect of physician triage on emergency department length of stay. J Emerg Med. 2010; 39: Fernandes CM, Price A, Christenson JM. Does reduced length of stay decrease the number of emergency department patients who leave without seeing a physician? J Emerg Med. 1997; 15: Polevoi SK, Quinn JV, Kramer NR. Factors associated with patients who leave without being seen. Acad Emerg Med. 2005; 12: Hampers LC, Cha S, Gutglass DJ, Binns HJ, Krug SE. Fast track and the pediatric emergency department: resource utilization and patients outcomes. Acad Emerg Med. 1999; 6: Simon HK, Ledbetter DA, Wright J. Societal savings by fast tracking lower acuity patients in an urban pediatric emergency department. Am J Emerg Med. 1997; 15: Bursch B, Beezy J, Shaw R. Emergency department satisfaction: what matters most? Ann Emerg Med. 1993; 22: Supporting Information The following supporting information is available in the online version of this paper: Data S1. Length of stay for all ED patients and patients not included in START.