Vista previa del material en texto
1 6 8 A N N A L S O F E M E R G E N C Y M E D I C I N E 3 9 : 2 F E B R U A R Y 2 0 0 2 C O N C E P T S Daniel W. Spaite, MD*‡ Fran Bartholomeaux, RN, MS‡ John Guisto, MD*‡ Elizabeth Lindberg, MD*‡ Becky Hull, RN, MS‡ Alicia Eyherabide, RN, MS‡ Sally Lanyon, RN‡ Elizabeth A. Criss, RN, MEd* Terence D. Valenzuela, MD, MPH*‡ Carol Conroy, PhD* From the Arizona Emergency Medicine Research Center, the Department of Emergency Medicine, College of Medicine, University of Arizona,* and the University Medical Center,‡ Tucson, AZ. Received for publication November 8, 1999. Revisions received November 3, 2000; May 24, 2001; and September 10, 2001. Accepted for publication October 11, 2001. Address for reprints: Daniel W. Spaite, MD, Arizona Emergency Medicine Research Center, Arizona Health Sciences Center, Box 245057, Tucson, AZ 85724-5057; 520-626-7957, fax 520-626-2480; E-mail dan@aemrc.arizona.edu. Copyright © 2002 by the American College of Emergency Physicians. 0196-0644/2002/$35.00 + 0 47/1/121215 doi:10.1067/mem.2002.121215 See related article, p. 159. Academic emergency departments are traditionally associated with inefficiency and long waits. The academic medical model presents unique barriers to system changes. Several non– university-based EDs have undertaken process redesign, with significant decreases in patient waiting time intervals. This is the presentation of a rapid process redesign in a university- based ED to reduce waiting time intervals. We present the application of a process-improvement team approach to evalu- ate and redesign patient flow. As a result of this effort, the median waiting room time interval (triage to patient room) decreased from 31 minutes in January 1998 to 4 minutes in July 1998. ED throughput times also decreased, from 4 hours, 21 minutes in January 1998 to 2 hours, 55 minutes in July 1998. Urgent care waiting room time intervals decreased from 52 minutes to 7 minutes and throughput times from 2 hours, 9 minutes to 1 hour, 10 minutes. Patient satisfaction evaluations by an independent institute demonstrated dramatic improve- ment and establishment of a new benchmark for academic EDs. Process redesign is possible in a busy, complex, tertiary-care ED, with decreases in waiting time intervals and improvement in patient satisfaction. Major sustained support from top-level hospital administrators and physician leadership are funda- mental prerequisites. With these in place, a process improve- ment team approach for evaluating and redesigning the patient care system can be successful. [Spaite DW, Bartholomeaux F, Guisto J, Lindberg E, Hull B, Eyherabide A, Lanyon S, Criss EA, Valenzuela TD, Conroy C. Rapid process redesign in a university-based emergency department: decreasing waiting time intervals and improving patient satisfaction. Ann Emerg Med. February 2002;39:168- 177.] Rapid Process Redesign in a University-Based Emergency Department: Decreasing Waiting Time Intervals and Improving Patient Satisfaction E M E R G E N C Y D E P A R T M E N T R A P I D P R O C E S S R E D E S I G N Spaite et al F E B R U A R Y 2 0 0 2 3 9 : 2 A N N A L S O F E M E R G E N C Y M E D I C I N E 1 6 9 I N T R O D U C T I O N The structure of an academic medical center often pre- sents unique barriers to system redesign, even when the design is aimed at improving patient services. In recent years, several non–university-based emergency depart- ments have undertaken process redesign efforts that have led to significant improvements in patient satisfaction, specifically in patient waiting times.1 We present such an effort initiated between hospital administration and aca- demic emergency physicians at a university-based medi- cal center. At University Medical Center (UMC) in the early 1990s, the ED volume was approximately 32,000 annu- ally. As a result of a projected increase in patient volume, a new ED was opened in 1994. Subsequent years yielded significant increases in volume, reaching 48,000 in 1996. By the end of 1997, the ED had ground to a halt, with extremely long waits, large numbers of disgruntled patients leaving without being seen by a physician, and overall inefficiency related to providing patient care ser- vices. The ED was a source of many complaints, and in late 1997, the hospital’s chief executive officer discussed the possibility of a major improvement initiative with the leaders of the academic division of emergency medicine. The goals of this joint effort were to improve overall effi- ciency and to satisfy patients. UMC is a traditional university health sciences center in a medium-sized metropolitan area (population 750,000). The hospital has approximately 350 acute-care beds and provides full tertiary-care services (Level I trauma, pedi- atric critical care, neonatal intensive care, high-risk obstetrics, and organ and bone marrow transplantation). The ED has 4 fully equipped trauma bays, 12 acute medi- cal rooms with cardiac monitors, 7 intermediate beds, and 9 urgent care (UC) beds. The UC portion of the ED is an integrated fast-track unit that sees approximately 35% of ED volume. R A P I D R E D E S I G N P R O J E C T D E V E L O P M E N T The theoretic construct centered in the belief that the cul- tural, operational, and process changes required to improve the service provided to ED patients could be achieved by a combination of (1) a major commitment by senior hospital administrators, (2) emphasis on emer- gency physician leadership, and (3) proper application of organizational resources by using a multidisciplinary process-improvement team approach. The improvement project focusing on these areas for change began in January 1998. The first change was to name a senior faculty member from the division of emer- gency medicine as administrative director of the ED. The administrative costs for this full-time position were borne by the hospital in a contract with the college of medicine. The director reports to the hospital’s vice president for patient care services. The director works collaboratively with a registered nurse who serves as patient care manager. The director has responsibility and authority for adminis- tration, hiring, and firing of nonphysician personnel, and operational (wage and nonwage) and capital budgets. The decision to have a physician faculty member in the role of a full-time hospital administrative director was unprecedented within the medical center (a nonprofit corporation separate from the medical school). However, it was clear that the many previous attempts to fix the ED had ultimately failed in a morass of faultfinding. Previous evaluations by the hospital had concluded that the prob- lem with the ED was inadequate physician staffing, varia- tions in physician practice, and lack of physician commit- ment to addressing the hospital’s issues. On the other hand, evaluations by the physicians had concluded that the hospital did not provide adequate nonphysician staffing, efficient processes, rapid turnaround time for diagnostic studies, or quick inpatient bed availability for timely admissions. Because of this longstanding gridlock, the leaders of the hospital and the emergency medicine faculty group agreed that the only way to succeed would be for the his- torical tug of war to end. By appointing and funding a physician director, the hospital had someone who was theirs and who also had clout with the physicians. At the same time, the physicians had a colleague who had re- sponsibility and authority for process, personnel, and budgets in the ED. The project was initiated with the support of UMC’s Strategic Initiatives Department. Under the leadership of the director and patient care manager, multidisciplinary process-improvement teams were formed. Each process improvement team included at least 1 senior hospital administrator (vice president), as well as appropriate physician, nursing, and ancillary personnel. Team mem- bership was no less than 8 and no greater than 12. Each team wasguided and supported by a trained facilitator from the strategic initiatives department. Teams met on a weekly basis using the following process: initial discus- sion, data collection, data analysis, process evaluation, recommendation formation, and presentation of recom- E M E R G E N C Y D E P A R T M E N T R A P I D P R O C E S S R E D E S I G N Spaite et al 1 7 0 A N N A L S O F E M E R G E N C Y M E D I C I N E 3 9 : 2 F E B R U A R Y 2 0 0 2 throughput times, a decrease in patients who left without being seen, and improved patient satisfaction. P R O J E C T I M P L E M E N T A T I O N A N D O U T C O M E S Recommendations from the process improvement teams were approved by the senior hospital administration be- tween February and April 1998. The project recommen- dations were implemented from late April through June 1998 (Table 1). The process improvement team evaluating internal procedures and staffing found numerous challenges. Rooms in the ED/UC areas were frequently full, backing patients up into the waiting room for long periods of time. A 1996 consultant report concluded that the new ED was already at capacity at 46,000 annual visits. The only way to decompress the department was to use the current beds more efficiently and to add beds in the hallway. The first problem identified by the process improve- ment team was staffing patterns. All nursing zones had at least 1 monitored bed, often resulting in less ill patients competing unsuccessfully for nursing resources. In addi- tion, nurse staffing had lagged far behind volume in- creases. The team recommendation was to change the nursing zones from 5 or 6 beds per nurse to no more than 4 beds per nurse. This change increased nursing availabil- ity and redistributed nursing resources so that each zone contained similar patients. By redistributing nursing re- sources, less ill patients were not held hostage to sicker ones and could be more efficiently treated. In addition, a nurse was added to staff 5 nonmonitored hallway beds for 16 hours a day. The changes in nurse staffing yielded an increase of 11.2 full-time equivalents (23%). The team identified the processing of orders and tele- phone communication as another constraint in patient flow. A single UA was responsible for both tasks. Analysis revealed that answering the telephone often consumed mendations to senior hospital administration. Specific teams were established to evaluate staffing and patient distribution, registration and triage time, interaction between the ED and laboratory and diagnostic radiology departments, and inpatient bed availability (Table 1). Data gathered for analysis included triage time, regis- tration time, time of arrival in room, and time of disposition from the department. Triage, room arrival, and disposition times were handwritten on the chart and subsequently entered into the hospital’s information system by a unit assistant (UA). Registration time was entered electroni- cally at the time of registration. Additional times specific to the laboratory and diagnostic radiology departments were gathered from logs maintained by each specific de- partment. Patient satisfaction was evaluated by using 6 validated questions (Figure 1).2,3 Patients or proxies were con- tacted by telephone in the month before change imple- mentation (April 1998) and in the month after implemen- tation (July 1998). The end points of effectiveness to be measured for the redesign project were a decrease in patient waiting room time, a decrease in overall ED/UC Table 1. Process improvement changes. Staffing/internal process: Decreased nursing ratios from 6:1 or 5:1 to 4:1 Rearranged nursing zones to create 1 zone with no monitored patients Added a new intermediate level (nonmonitored) zone by adding 5 hallway beds Doubled UA staffing and separated responsibilities into telephone-communication and paperwork-order processing Increased emergency medicine resident staffing by 25% (July 1, 1998) Triage-registration Status board monitors in triage Brief triage and place patient immediately in room In-room registration Doubled registration staff Two-way radios for communication between triage nurse, charge nurse, and registration staff Diagnostic radiology Electronic order entry Tripled dedicated ED radiology staffing Located new radiograph printer in ED Radiographs hung immediately, irrespective of whether old films available Laboratory Electronic order entry Bar-code labeling done in ED Bright visual cue for laboratory staff ED laboratory samples take priority over all other laboratory samples Bed availability ED-based nursing admit team Figure 1. Patient satisfaction survey questions. 1. How would you rate your waiting time? 2. How would you rate the courtesy of the ED staff? 3. How would you rate the overall care that you received? 4. Did you have to wait long to see a physician? 5. Did you have to wait long for your radiograph or laboratory results? 6. Would you recommend this ED to your friends or family? E M E R G E N C Y D E P A R T M E N T R A P I D P R O C E S S R E D E S I G N Spaite et al F E B R U A R Y 2 0 0 2 3 9 : 2 A N N A L S O F E M E R G E N C Y M E D I C I N E 1 7 1 this position. The single UA was responsible for answer- ing incoming calls, initiating pages for nursing and physi- cian staff, and transferring calls to patient rooms. These activities precluded efficiently carrying out the other responsibilities (eg, laboratory and radiographic order processing and initiating inpatient beds). In response, a new 24-hour-a-day UA position was created, separating order processing from the communications process. The evaluation of physician staffing found that ED vol- ume had increased by nearly 50% over the previous 3 years. Although faculty coverage had increased by 54%, emergency medicine resident coverage had increased by only 20%. During this time, both the emergency medi- cine residents and the residency director noted the educa- tional experience was compromised as a result of the dramatic increase in the number of patients seen per resi- dent. The residents also noted a decreased ability to inter- act with attending physicians because of increased patient volume. This prompted a request to the college of medi- cine to increase the emergency medicine residency from 8 to 10 residents per year, allowing an increase of 25% in ED resident staffing. Although the justification for the residency expansion was made purely on educational grounds, the parallel needs of education and service were both met by increasing the numbers. Before the changes, the ED used a traditional triage and registration process. When patients presented, a triage nurse performed a relatively complete nursing assess- ment. Unless an emergency condition was present, the patients would wait in line to be registered. Then the patients would remain in the waiting room until all infor- mation had been entered into a computer and a patient- care record was printed at the ED or UC nursing station. The paperwork would then be placed in a bin to await room assignment. When a room became available, a nurse or patient care technician (PCT) would go to the waiting room and bring the patient to his or her room. This process was completely redesigned. Now, the patient arrives in the waiting room and goes to the triage nurse, where a brief triage is performed and a decision is made regarding whether the patient should go to UC, a nonmonitored ED bed, or a monitored ED bed. The chart is handwritten, available immediately, can be initiated from any location in the department and requires no wait- ing for data entry or subsequent printing. The room status boards in the ED and UC are visible to the triage nurse through a television monitor, allowing the triage nurse to assign patients to open rooms. All nursing and registra- tion staff have 2-way radios that allow for immediate and continuous communication. The final process is as fol- lows: the triage nurse does a briefassessment, documents on a chart, identifies a room, radios both the charge nurse and registration staff of the patient’s destination, and then the patient is placed in a room. The nurse in that zone greets the patient and does a complete nursing assessment. Often, the physician will also be in the room and may obtain a simultaneous his- tory. Soon after room assignment, the admitting clerk brings a mobile computer dumb terminal into the room and plugs into the hospital’s central information system through a standard data port. Registration is performed at the bedside, and patient labels are generated at the nursing station. The financial aspects of the visit are completed on discharge, completely unlinked to the registration process. Analysis of registration staffing revealed that to do in-room registration, staffing of admitting clerks would have to be nearly doubled. Evaluation of diagnostic radiology identified 2 major problems. First, the process of ordering and obtaining radiographs was cumbersome. Second, the volume of radiographs had markedly increased during recent years, whereas radiology staffing had not. The existing process of obtaining a radiograph included the physician writing (often illegibly) on a radiograph order form. The UA would then call the radiology techni- cian (or page them during the hours that no dedicated radiology staff were scheduled), often interrupting a radiograph being done on another patient. When the technician was ready to do the radiograph, he or she would come to the nursing station, retrieve the radiograph order, go back to the radiology area, and electronically enter the information. Then the technician would trans- port the patient from his or her ED room to the radiology suite and perform the radiograph. The patient was re- turned to his or her room after film processing. Whereas the radiology suite and the digital information processor were located in the ED, the film printing actually occurred in central radiology, more than 100 yards away. A film librarian was responsible for retrieving and labeling the films, getting any old radiographs, and delivering them to the ED. Often the film librarian would accumulate multi- ple patients’ films before bringing them to the ED. Time sampling revealed that mean time from radiograph order until the films were available to the emergency physician for review was 77 minutes. In the revised process, the UA transcribes the physician radiograph order electronically. The order is electronically transmitted and printed in the ED radiology suite without E M E R G E N C Y D E P A R T M E N T R A P I D P R O C E S S R E D E S I G N Spaite et al 1 7 2 A N N A L S O F E M E R G E N C Y M E D I C I N E 3 9 : 2 F E B R U A R Y 2 0 0 2 were too busy to accept the patient. The throughput inter- val for UC patients decreased by 47% (Figure 5). The improvements in time intervals occurred despite the fact that there were significant increases in ED/UC volume every month when compared with volumes in the previous year. The overall volume in 1998 was up 17% over 1997. Table 2 places the process changes, on the interruption of the technician’s current activities. A film printer was placed in the ED radiology suite to rectify the problem of distant film printing. Delivery of old films from the film library was uncoupled from hanging of the films. Radiology staffing was increased from 1 technician 16 hours per day to 2 technicians 24 hours per day and a radiology assistant 16 hours per day. The process yielded order-to-hang-time intervals of 9 to 15 minutes in the months after redesign (an 80% to 88% reduction). The existing laboratory process included the physi- cian’s written order being sent with a specimen to the lab- oratory, where it would sit in a collection bin until a labo- ratory technologist was available to electronically enter it into the system. This would lead to the generation of a bar-code label that was attached to the specimen. The specimen was then placed in the appropriate area for analysis. In any given collection bin, there might be as many as 20 to 30 specimens awaiting testing. Although, theoretically, ED specimens were “stat,” there was no visual cue to identify the specimens. Data analysis found that only 88% of specimens were analyzed and reported to the ED within the optimal targets set for the various stat tests. These targets are time ranges mutually agreed on by the ED and laboratory administration. With the redesigned process, the UA performs elec- tronic order entry, including bar-code labeling. The labels are a unique color, which easily identifies ED specimens throughout the entire transport and testing process. When a specimen arrives in the collection bin, it is now treated as the number one priority and is completed ahead of all other laboratory samples. After the imple- mentation, the number of specimens available within the target range increased to 94%. The multifaceted process changes resulted in dramatic improvements in waiting room and throughput intervals. Figures 2 and 3 show the waiting room intervals (time interval from triage to patient room) for ED and UC patients from January through December 1998. These data reveal significant reductions in the medians of 90% in the ED and 91% in UC. Figure 4 shows the effect on ED throughput intervals (interval from triage to patient dis- position). The throughput time for admitted patients was found to decrease by 27% between January and December, whereas nonadmitted patient throughput times declined by 31%. The addition of an admitting registered nurse- patient-care technicians team 12 hours a day to transport patients upstairs, enter their data into the unit computer, and perform routine inpatient admitting duties allowed admissions to occur even when inpatient units said they Figure 2. ED waiting room interval. 1:04 0:57 0:50 0:43 0:36 0:28 0:21 0:14 0:07 0:00 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 1998 Mean time– 83% decrease Median time– 90% decrease Figure 3. UC waiting room interval. 1:12 0:57 0:50 0:43 0:36 0:28 0:21 0:14 0:07 0:00 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 1998 Mean time– 84% decrease Median time– 91% decrease 1:04 E M E R G E N C Y D E P A R T M E N T R A P I D P R O C E S S R E D E S I G N Spaite et al F E B R U A R Y 2 0 0 2 3 9 : 2 A N N A L S O F E M E R G E N C Y M E D I C I N E 1 7 3 basis of our subjective hypothesis, in rank order of effect on patient waiting room time and overall ED/UC through- put times. In addition, it gives the approximate costs and delineates the full-time equivalent associated with per- sonnel additions, when applicable. Three prospective methods for tracking patient satis- faction were used: (1) an evaluation of the effect of the changes on the number of patients who left without being seen; (2) a telephone survey immediately before and after the process improvement changes using validated survey questions; and (3) a comparison of the Picker Institute Survey for University Health System Consortium (UHC) benchmarks.2,3 Figure 6 documents the decline noted in the patients who leave without being seen. Data were obtained from the triage log for patients leaving before seeing the triage nurse and from ED records of patients not responding when called for room assignments. Over the implementa- tion period, a 92% decline was noted in this patient popu- lation. Figure 1 lists the 6 questions from the telephone sur- vey. Major improvements were noted in every category, comparing July with April. Figures 7 and 8 show a sum- mary of data for “poor” and “excellent” responses on a 5- point scale for questions 1 to 3. The “definitely yes” and “definitely no” answers on a 3-point scale, including “def- initely no,” “probably yes,” and “definitely yes,” are shown for questions 4 to 6. The third method of identifying changes in patient sat- isfaction used results from a comprehensive survey that has been mailed annually toED patients for years. This survey includes 54 questions that evaluate patient satisfac- tion with a host of care-and service-related processes.2,3 The results are compiled into an overall rating of patient satisfaction. More than 100 UHC hospitals use the Picker Institute Survey, allowing comparisons and benchmark- ing for hospitals in the consortium. Figure 9 shows the results of the overall rating before (1997), during (second quarter of 1998), and after (fourth quarter of 1998) imple- mentation. The UHC benchmark (not average) is shown. In 1997, we scored substantially below the benchmark. During the quarter of implementation, substantial improvement was noted, bringing results up to bench- mark. After full implementation, a huge improvement occurred, with the result well above the UHC benchmark. It is notable that most of this improvement was caused by an increase in the most satisfied (excellent) group. I N T E R P R E T A T I O N A N D I M P L I C A T I O N S During the data collection and analysis phase of the re- design process, it became clear that many processes had been developed on the basis of the needs of hospital per- sonnel, departments, and operational realities. When these processes were analyzed with a systematic step-wise process improvement approach, their flaws were uncov- ered. For example, including noncritical patients in zones with critical patients delays care for the noncritical patient. From a systems perspective this created a self- Figure 4. ED throughput interval. 4:33 4:21 4:04 3:36 2:55 2:38 2:09 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 1998 Median time 3:07 Figure 5. UC throughput interval. 2:24 1:55 1:40 1:26 1:12 0:57 0:43 0:28 0:14 0:00 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 1998 Median time– 47% decrease 2:09 E M E R G E N C Y D E P A R T M E N T R A P I D P R O C E S S R E D E S I G N Spaite et al 1 7 4 A N N A L S O F E M E R G E N C Y M E D I C I N E 3 9 : 2 F E B R U A R Y 2 0 0 2 kind and compassionate far more often. This was true despite the fact that there was no specific intervention or effort made to affect this issue. There was also marked improvement in satisfaction with the care rendered, even though there were no specific efforts to change actual care. Recent trends in defining quality in health care have emphasized patient satisfaction. These trends have also been reflected in the medical literature related to ED patients.4-11 The fact that this effort, which was entirely focused on service and process efficiency, led to improve- ments in patient satisfaction and their perception of the quality of care is an interesting corollary to the current trend. One of the most significant effects of the improved effi- ciency was a marked decrease in the number of patients who leave without being seen (Figure 6). The typical patient who leaves an ED without being seen presents with a relatively minor complaint and often waits a significant defeating situation in which the patients who could be cleared most quickly from the system, thereby freeing up personnel, equipment, and space for additional patients, often stayed the longest. The success of the redesign process was most likely the result of the multifaceted approach. Simply improving 1 or 2 of the process issues (eg, radiology or in-room regis- tration) might have led to minimal overall improvement. However, to affect waiting room and throughput intervals by large amounts required improvements in all aspects of intradepartmental and interdepartmental functioning. Having the support of hospital administration and the emergency physician staff was essential in getting over the historical barriers of committing to rapidly assessing and treating all patients. In the patient satisfaction surveys, an interesting trend was noted. Several questions that did not directly relate to efficiency or timing dramatically improved. An example was that, after the changes, the staff were perceived as Table 2. Theoretic and cost implications of process improvement changes in rank order of importance for decreasing waiting time intervals. Theoretic Basis for Full-Time Rank Intervention Improvement Equivalent One Time/Start-Up Costs Ongoing Costs 1 Top institutional priority combined with Administrative change 1.0 MD Contract with college of medicine for physician-led departmental model full-time director 2 Change nursing ratios from 1 RN/5-6 Staffing change 11.2 RN Hiring-orientation Wages–benefits rooms to 1 RN/4 rooms 3 Rapid triage and in-room registration Process change and 7.0 patient 6 computer dumb terminals and Wages–benefits staffing change registration mobile carts: $2,000; terminal coordinators data ports in each room: $3,000; 3 patient ID label printers: $1,200; hiring-orientation 4 Radiology: changes in ordering, producing, Process change Radiograph film dry laser printer: No new cost compared with old printing, and making films available $40,000 process for physicians 5 Radiology staffing for rapid availability Staffing change 4.2 radiology Hiring-orientation Wages–benefits around the clock technologists 6 Unit clerical function: separating com- Process change and 4.2 UAs Computer workstation: $1,500 Wages–benefits munication responsibilities from nursing staffing change orders-laboratory-radiology processing responsibilities 7 RN/patient-care technician admission Process change and 2.1 RN Hiring-orientation Wages–benefits team staffing change 2.1 patient-care technicians 8 Electronic order entry Process change Training: $3,000 9 Create intermediate-severity nursing zone Process change without monitored rooms 10 Laboratory ordering and process changes Process change FTE savings in laboratory: Bar-code printer: $500 Ongoing FTE savings in laboratory offset by change 8 specimen labels 11 Communication between triage, charge Process change 20 radios: $5,000 Radio maintenance and replacement: nurse, registration, nursing staff $1,000/y E M E R G E N C Y D E P A R T M E N T R A P I D P R O C E S S R E D E S I G N Spaite et al F E B R U A R Y 2 0 0 2 3 9 : 2 A N N A L S O F E M E R G E N C Y M E D I C I N E 1 7 5 amount of time. When the wait becomes too long from the patient’s perspective, they choose to seek medical care elsewhere or not at all. Typically, this group represents people who have become very angry and frustrated with the inattention to their needs. The decrease in waiting intervals was associated with a 92% decrease in patients who leave without being seen. A chronic problem for our institution has been the unavailability of inpatient beds. Our institution has his- torically had an inpatient census of more than 80%. Because of this, the ED often functions as a holding unit when inpatient beds are tight. However, we chose to emphasize the fact that 80% to 85% of the patients that present to our department are discharged. Thus, rather than spending a major institutional effort to deal with inpatient bed availability (an effort that has previously failed many times), we chose to affect the waiting times of patients over whom we had primary control. We were sur- prised to find that the median throughput intervals for admitted patients improved nearly as much (27%) as the intervals for nonadmitted patients (31%). This occurred despite the fact that the physician consultation-admission team was unable to agree on recommendations for change in admitting physician practices. The improvement in throughput time for admitted patients appeared to be related to 3 major changes: (1) the triage-to-patient-room interval was significantly improved for both admitted and nonadmitted patients; (2) decisionmaking related to inpa- tient admission occurred earlier as a byproduct of faster availability of diagnostic results and additional emergency Figure 6. Patients who left without being seen. 300 250 50 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 1998 92% decrease 100 150 200 Jan 98 250 Dec 98 21 Figure7. Telephone survey: “poor” responses. 70 60 50 40 30 20 10 0 Poor wait Poor courtesy Poor overall Poor physician wait Poor results wait Would not recommend April July % Figure 8. Telephone survey: “excellent” responses. 80 70 60 50 40 20 10 0 Wait time April July 30 Courtesy Overall care Physician wait Results wait Definitely recommend % Figure 9. Overall ED rating. 2Q98, Second quarter of 1998; 4Q98, fourth quarter of 1998. 100 90 80 70 60 50 40 30 20 10 0 22.1 24.5 23.9 1997 31.2 24.3 20.2 2Q98 22.2 27.2 36.7 4Q98 23.5 25.6 26.1 UHC benchmark Excellent Very good Good E M E R G E N C Y D E P A R T M E N T R A P I D P R O C E S S R E D E S I G N Spaite et al 1 7 6 A N N A L S O F E M E R G E N C Y M E D I C I N E 3 9 : 2 F E B R U A R Y 2 0 0 2 tionship between the physician director and the patient care manager and their joint direct reporting to a vice president has been particularly effective. The ability to successfully implement these changes elsewhere may be limited by the variations found among EDs. However, we believe that many of the philosophic and operational changes are relevant to most departments. To aid in the potential for implementation elsewhere, we refer the reader to the rank-order list in Table 2. This may give some insight into the relative importance of each of the interventions. The reliance on handwritten times entered by nursing staff for each of the times that were analyzed is a limita- tion. Undoubtedly, there is some compromise of the accu- racy of the times on the basis of variations in individual timepieces in the department, as well as variation in attention to accuracy among different personnel. How- ever, we have no reason to believe that the times were sys- tematically different in the before and after cohorts. The decreases in patients who left without being seen, as well as major improvements in patient satisfaction, make it clear that the improvements could not have been phan- tom changes based on time reporting alone. It is also possible that some, or even all, of the improve- ments noted in the department were caused by the Haw- thorne effect. Hospital administration was more actively involved in daily activities in the ED, often showing up and discussing issues with the staff. More staff was added, which improved staff attitude. Daily and weekly re- minders about target ranges and successes kept the progress in the forefront of the department’s operation. Thus, some of the improvements could be related to the focus of the organization rather than being caused solely by the process changes themselves. In summary, we report the results of a major effort to decrease waiting time intervals and improve efficiency in a university-based ED. As an institutional priority, and with full integration of the emergency medicine faculty into the process, major improvements were noted in patient waiting and throughput intervals. This occurred in the face of significant increases in patient volume. We believe that these improvements are the result of major administrative, philosophic, operational, and bud- getary changes aimed at efficiency and patient satisfaction. R E F E R E N C E S 1. Berwick DM, Espinosa J, Kosnik L. Reducing Waits and Delays and Improving Patient Satisfaction in the Emergency Department. Boston, MA: Institute for Healthcare Improvement; 1999. 2. Cleary PD, Edgman-Levitan S, Roberts M, et al. Patients evaluate their hospital care: a national survey. Health Aff (Millwood). 1991;10:254-267. physician coverage; and (3) an ED-based registered nurse–patient-care technician admission team was imple- mented that eased the workload for the admitting nurses on the inpatient units. The financial issues involved in this project are rele- vant to any department that considers making sweeping changes. Obviously, major financial investment was nec- essary. However, a pleasant surprise has been that the changes have more than paid for themselves. The ongoing costs of this project have exceeded $1 million annually. However, despite this major increase in cost, financial analysis has shown that the department has gone from being approximately $600,000 in the red to more than $300,000 in the black after implementing these process improvements. It is impossible to be sure that there is a direct cause-and-effect relationship between these events. However, this improvement of the bottom line was possi- bly the result of 3 factors. First, the marked decrease in throughput interval had the net effect of adding rooms to the department. This provided an increase in capacity to see new patients. In fact, in the year after implementation, the volume (which had reached a plateau at 49,000 for the 3 previous years) increased by more than 17%. Second, the improved efficiency allowed us to signifi- cantly decrease the number of patients who leave without being seen. These patients represented a recovery of lost revenue. Finally, the payer mix markedly improved. The cause of this may be multifactorial. However, it appears that a major factor in this improvement was related to the fact that improved service and patient satisfaction encouraged patients with a financial choice to drive past other EDs to come to our facility. In fact, among all of the points of entry to our hospital, the ED is now the site with the best payor mix. This is very striking because before the changes, the ED was the point of entry with the worst payor mix. It has become clear to the hospital administration and the emergency physicians that there are 2 foundational prerequisites for success in an effort such as this. First, there must be major commitment and support from the highest levels of the hospital. The absence of this major focus would doom an undertaking such as this to failure. Second, integrating the emergency physicians and ED administration by appointing a physician to have respon- sibility and authority for all clinical operations provided a level of coordination that had never been possible under any other administrative model. This director has full authority and responsibility for process, operations, per- sonnel, and budgetary decisions. The collaborative rela- E M E R G E N C Y D E P A R T M E N T R A P I D P R O C E S S R E D E S I G N Spaite et al F E B R U A R Y 2 0 0 2 3 9 : 2 A N N A L S O F E M E R G E N C Y M E D I C I N E 1 7 7 3. Cleary PD, Edgman-Levitan S, McMullen W, et al. A national survey of hospital patients: the relationship between reported problems with care and patient evaluations. Qual Rev Bull. 1992;18:53-59. 4. Hall MF, Press I. Keys to patient satisfaction in the emergency department: results of a multiple facility study. Hosp Health Service Admin. 1996;41:515-532. 5. Bursch B, Beezy J, Shaw R. Emergency department satisfaction: what matters most? Ann Emerg Med. 1993;22:586-591. 6. Whipple TW, Edick VL. Continuous quality improvement of emergency services. J Health Care Marketing. 1993;13:26-30. 7. Moore CH. Experimental design in health care. Qual Manag Health Care. 1994;2:13-26. 8. Rhee KJ, Allen RA, Bird J. Telephone vs. mail response to an emergency department patient satisfaction survey. Acad Emerg Med. 1998;5:1121-1123. 9. Mayer TA, Cates RJ, Mastorovich MJ, et al. Emergency department patient satisfaction: customer service training improves patient satisfaction and ratings of physician and nurse skill. J Health Care Manag. 1998;43:427-440. 10. Rhee KJ, Bird J. Perceptions and satisfaction with emergency department care. J Emerg Med. 1996;14:679-683. 11. Thompson DA, Yarnold PR, Williams DR, et al. Effects of actual waiting time, perceived waiting time, information delivery, and expressive quality on patient satisfaction in the emer- gency department. Ann Emerg Med. 1996;28:657-665.