EMDOC (Emergency Department Overcrowding) Internet-Based Safety Net Research

The Journal of Emergency Medicine, Vol. 35, No. 1, pp. 101–107, 2008 Copyright © 2008 Elsevier Inc. Printed in the USA. All rights reserved 0736-4679/08 $–see front matter

doi:10.1016/j.jemermed.2007.03.022

Administration of Emergency Medicine EMDOC (EMERGENCY DEPARTMENT OVERCROWDING) INTERNET-BASED SAFETY NET RESEARCH Robert Steele,

MD, FACEP

and Attilla Kiss,

MD

Department of Emergency Medicine, Synergy Medical Education Alliance, Saginaw, Michigan Reprint Address: Robert Steele, MD, Loma Linda University Medical Center, 11234 Anderson Street, Loma Linda, CA 92354

e Abstract—Emergency Department (ED) overcrowding is a national crisis with few prospective data to document its occurrence. The objective of this study was to prospectively collect data on variables involved in Emergency Department overcrowding (EMDOC) using an Internet-based data entry model. A prospective observational Internet-based study involving 18 hospitals over a 13-month period was designed. Investigators input data into the EmDOC Internet site at 10:00 p.m. on 7 random days each month. The study found that the primary reason for ED overcrowding was lack of inpatient beds. Important means were: patient-to-nurse ratio ⴝ 2.85, diversion was 7.4 h/24 h, and hospital census was 83%. From ED waiting room to an ED bed took a mean time of 209 min. The mean number of makeshift beds was 3.1. There was no single variable that was noted to define or predict overcrowding. Documentation of factors involved in ED overcrowding found that overcrowding was not just an ED problem, but a problem that occurs due to overcrowding in the entire institution. © 2008 Elsevier Inc.

this country (1). This has received attention not only in publications specific to emergency medicine but in the mainstream media as well (2,3). The United States is not alone in this crisis. It has been well documented in Australia, Great Britain, and Taiwan (4). Emergency Department overcrowding has been blamed on the closing of hospitals, decreasing hospital stays, and the outpatient management of complex medical problems (5,6). This trend of decreased hospital stays, the aging of America, and the increase in the working uninsured continues to place stress on an already stressed system (5–7). There is a relative paucity of documented studies that quantify ED overcrowding in a prospective manner (6). This is partly because it is hard to define overcrowding, but it is easy to recognize (8). In addition, it is hard to find someone to input prospective data from EDs regarding overcrowding at the peak hours in the ED. We used the Internet to bridge that gap. In this study we attempted to quantify ED overcrowding utilizing known ED overcrowding variables. We then tried to simplify the input of those variables using Internet technology.

e Keywords—Emergency Department; overcrowding; Internet

INTRODUCTION METHODS

It has been well publicized that there is a crisis of overcrowding in Emergency Departments (EDs) across

Study Design and Setting

Presented as an abstract at the American College of Emergency Physicians annual Scientific Assembly, October 2002, Seattle, WA. Currently, Dr. Steele works for the Loma Linda University Medical Center, Loma Linda, California.

The objective of this study was to prospectively collect data on variables involved in ED overcrowding using an Internet-based data entry model. Inclusion criteria were that an institution had to have ⬎ 15,000 ED visits per

RECEIVED: 22 August 2005; ACCEPTED: 2 October 2006 101

102

year and accept ambulance traffic. This was done to exclude satellite facilities and urgent care facilities that provide urgent but not emergent care. Express Care data located within the confines of the ED were not included. All hospitals in Michigan were notified of the study by letter and invited to participate. Eighteen EDs in Michigan were willing to participate; these 18 hospitals accounted for approximately 20% of all ED visits in the state. There were five academic and 13 community institutions participating in this study. This study was approved by the Institutional Review Board of the sponsoring institution. This approval was accepted by the other participating institutions.

Data Collection The information was to be entered at 10:00 p.m. on 7 random nights per month. Data were entered into an Internet-based data acquisition form on the website www. schi.org/cgi/emdoc_days.txt. The form was developed based on previous overcrowding surveys (Figure 1) (8–10). Drop-down boxes were used to facilitate rapid input of standardized information. Information was gathered from multiple sources including triage nurses, charge nurses, bed coordinators, tracking systems when available, and working ED physicians. It was the duty of the site investigator to input collected data and the Internet form ensured that all data were standardized.

Outcome Variables The variables used for overcrowding were divided into patient volume, patient waiting times, and hospital staffing. Patient volume variables were: total number of registered patients in the ED, number of admitted patients in the ED, total number of patients in the hospital, time on diversion over the last 24 h, number of patients in the waiting room, and the number of patients in makeshift beds. Other variables based on patient wait times were: mean wait times in the ED waiting room at 10 p.m., ED disposition times, and radiology and laboratory times (Figure 1). Variables based on staffing were nurse staffing, and physician staffing at 10 p.m. Acuity was assessed based on the number of resuscitations and the number of intensive care unit (ICU) patients in the ED because these two groups utilize the greatest amount of ED resources in a short period of time. Hospital census was obtained from the working bed coordinator at 10 p.m. The nurse in triage calculated ED waiting room census and ED waiting room wait times at 10 p.m. Makeshift beds were defined as any bed created that was not part of the original patient bed design. This

R. Steele and A. Kiss

meant that any patient bed in the hallway, a doublebunked patient bay, or a bed located in isolated storage closets was considered a makeshift bed. Diversion times were calculated over a 24-h period. The charge nurse determined the number of nurses on staff in the ED. Only registered nurses were counted; Emergency Medical Technicians and registered-nurse-extenders were not factored into the equation. For acuity, we tried to obtain recent resource-utilizing events that may slow down an ED. We used codes (pediatric, trauma, and medical) and numbers of ICU patients to ascertain acuity in the ED at that moment of data collection. If there was not a tracking system in place, radiology and laboratory times were estimated by the working ED physician. The goal of this study was to provide a glimpse of the ED at a time when administrators are not in the hospital. Internal Continuous Quality Improvement projects from several institutions showed 10 p.m. to be a particularly busy time in the ED. Thus, 10 p.m. was picked because it is typically a very busy time and it is not a time when administrators are in the ED. The Internet was used for data collection because it created an instant standardized database. The site investigator would input the data. Drop-down boxes were used to help standardize quantifiable data. For example, diversion time could only be recorded in minutes and there was a range of 0 –30 min, 30 – 60 min, 60 –120 min, 120 –180 min, etc. The information obtained was protected at the server level to ensure that tampering with the information could not occur. Randomly assigned numbers were given to the institutions to assure anonymity in this process. An initial month was used as a pilot month to examine any technical flaws, which were corrected before implementation of the study. Data from the pilot month were not included in this study. There was an EmDOC (Emergency Department OverCrowding) logo that accessed a link to the study questions to be completed by the site investigators at each participating institution. When the site investigators filled out the study questions, the information was automatically converted into an Excel (Microsoft Inc., Redmond, WA) file format. There was no third-party data entry. Follow-up by e-mail on a daily to weekly basis was used to let participants know that the information was being transmitted correctly and to address any problems. The data were password-protected and were accessible only to the principal investigator at any time. This also meant the study could be monitored at all times from anywhere. Left without Being Seen (LWBS) data were obtained from all of the participating hospitals as part of an ongoing ED Quality Assurance process. We incorporated these data into the EmDOC database retrospectively. All other data were obtained prospectively.

FemVein Access Study

103

Figure 1

Statistical Analysis Mean values were calculated for patient wait times, patient volumes, and staffing ratios. Confidence intervals

were included with the mean value. Pearson’s correlation was calculated to compare inpatient wait times with ambulance diversion times, as well as ambulance diversion times with LWBS volumes. A Pearson’s correlation

104

R. Steele and A. Kiss

Figure 2

was also calculated for physician-estimated laboratory time vs. computer-tracked laboratory time. Diversion was plotted for seasonal variation.

RESULTS The results stated are over a 9-month period of time when all institutions were participating. The mean number of patients in the waiting room at 10 p.m. was 27.0 (95% CI 26.9 –27.1). The mean number of patients waiting for an inpatient bed was 4.0 (95% CI 3.8 – 4.2). If they waited for longer than 4 h, then 85% (95% CI 83– 87) of those patients were waiting for an ICU bed. The median number of empty beds in the ED at 10 p.m. was 0. The mean number of patients in makeshift beds

was 3.1 (95% CI 2.4 –3.8). The mean time to get from the ED waiting room to an ED bed was 209 min (95% CI 199 –221). On average, 1 (95% CI 0 –2) patient was held in the ED longer than 4 h awaiting bed placement. The median time from initial triage to discharge was 7 h (95% CI 6.9 –7.1) for patients in the ED at 10 p.m. The mean diversion time per day was 7.4 h (95% CI 7.2–7.6). The mean diversion event lasted 2.2 h (95% CI 2.1–2.3). The two main reasons given for ambulance diversion were lack of inpatient beds (95% CI 40.6 %) and ED patient volume exceeding ED resources (95% CI 31.3%). We tried to examine why patients were waiting in the ED. The types of in-house beds patients were waiting for are shown in Figure 2. The role of the consultant was evaluated to see if that made a difference in ED wait times. Patients who did not have a consultant see them in the ED waited ⬎ 3 h for an inpatient bed only 17.0% of the time. Patients who had a consultant see them in the ED waited ⬎ 3 h for an inpatient bed 59.0% of the time (Figure 3). This time does not include the ED workup. This wait started at the time of ED physician disposition. Inpatient bed wait times correlated strongly with ambulance diversion times. A Pearson correlation of 0.93 (95% CI 0.90 – 0.95) was found. As inpatient bed wait times increased, so did ambulance diversion times. As ambulance diversion times increased, so did the incidence of LWBS with a Pearson correlation of 0.88 (95% CI 0.83– 0.92). LWBS rates also correlated to inpatient bed wait times with a Pearson correlation of 0.90 (95% CI 0.87– 0.92). Of the patients who left without being seen, 65% did so when the inpatient bed wait times were

Figure 3

FemVein Access Study

105

What was the wait for an Inpatient bed when patients LWBS?

80

65.2

60 % LWBS

40 20

10.9

8.7

6.5

8.7

0 <

r 1h

2 1-

s hr

3 2-

s hr

4 3-

s hr

>4

s hr

Inpatient Bed Wait Times Figure 4

⬎ 4 h (Figure 4). Overall, the rate of LWBS was 1.2 (1.1–1.3) per institution per night. The seasonal variation of diversion times peaked in February and June (Figure 5). The mean nurse-per-patient ratio was 2.85 (95% CI 2.74 –2.96). The mean hospital census was 83% (95% CI 81.8 – 84.2) at 10 p.m. at night. The mean time for laboratory results (complete blood count, cardiac enzymes, and electrolytes) was just under 1 h. In eight of the institutions, a computer tracked the laboratory times, whereas the other 10 institutions required the ED physician to estimate laboratory times. Mean radiology times for a chest X-ray study and computed tomography scan of the head as estimated by the ED physician were found to be 1 h.

DISCUSSION One of the major differences between this study and other ED overcrowding studies is that we were able to use the Internet for prospective data entry and evaluate the ED at a time when administrators were rarely around to view the ED overcrowding problem. We also obtained data from different facets of patient care (11–14). We obtained data from physicians, charge nurses, triage nurses, and bed coordinators while they were working. The data were obtained at what is traditionally the busiest hour in the ED. The data were automatically logged into an Excel file and the principal investigators could monitor the study 24 h a day. By having an Internetbased site, we were able to free up the site investigator from having to drag around a pen and paper and ask four different people the required information seven times a month. There was no delay in survey response time as would be seen with a written survey. There were no incomplete survey entries. The investigator could call to

various parts of an ED to obtain the data. Once the data were input, the principal investigators could review it immediately. Several variables that are important in ED overcrowding did not seem as important in our study; this is probably related to the design of the study. Our study provided a look at 18 EDs at 10 p.m. Variables such as LWBS are not high at 10 p.m. Wait times to see a physician for a sprained ankle did not apply because more than 50% of the study institutions had a working Express Care area designated for low acuity patients. This variable would most likely be more significant at 2 a.m. when patients become tired and leave. A variable that showed poor correlation was acuity. We tried to use resuscitation to help evaluate ED acuity. In retrospect, resuscitations are just too rare an event to show a direct correlation with an everyday event such as ED overcrowding. Point-of-care testing was in place at 13 of the 18 institutions. The time of 209 min from triage to an ED

Percent of Time on Diversion Over Last 24 Hours 100 80 58

60

62 50

38

40

23 15

20

15

10

5

0 O

ct

ov N

D

ec

n Ja

b Fe

Figure 5

M

ar

A

pr

M

ay

ne Ju

106

bed may have occurred because Express Care patients were removed from the study and patients frequently were in the waiting room while laboratory and radiologic studies were being performed. We did not determine the time from triage to first patient study, which, in retrospect, may have been an important variable. We found nurse-to-patient ratios to be adequate overall. Laboratory and radiology times are frequently pointed to as a causal factor in prolonged patient wait times; we did not find this in our study. ED overcrowding literature is important to continue to document a problem affecting our practice and our patients. Administrators needs to be educated that overcrowding continues to be a problem. Data showing prolonged wait times by patients will negatively affect the reputation and, in the end, the financial strength of the hospital, and this will help to influence those administrators (15,16). It is hoped that through continued research like this we can help to solve the ED overcrowding problem. We found specific areas that may aid in solving the ED overcrowding problem. We found that the primary reason for diversion is the lack of inpatient beds, leading to excessive wait times in the ED. Patients being held in the ED take up space that could be used for patients coming in. This relative decrease in the functional number of ED beds leads to an overall increase in the number patients waiting in the waiting room. The majority of admitted patients are waiting for general medical, telemetry, or ICU beds. Although ICU beds cannot be duplicated outside of that unit, it is possible to place telemetry and general medicine patients into observation units. Observation units have been shown to increase the efficiency of hospital bed turnover and aid in increasing inpatient bed availability (17). Use of consultants can lead to prolonged ED wait times. A solution is to allow the ED physician to make the disposition decision. The consultant can then help to make decisions regarding patient management. This means that ED physicians would have to have the authority to admit stable patients into inpatient beds. Diversion times peak in specific months. These data may help in hospital staffing projections. This study shows diversion times peaking in January, February, and June. Employees may be encouraged not to take planned vacation time during those months in an effort to prevent ED overcrowding from occurring. LWBS rates increase as diversion and inpatient bed wait times increase. To put it another way, LWBS rates increase as ED overcrowding increases. This variable shows how lack of inpatient bed availability negatively influences incoming revenue. This may be a way to convince hospital administrators that lack of inpatient bed access correlates to ED overcrowding and lost rev-

R. Steele and A. Kiss

enue. Communication with patients has been listed as a way of decreasing this variable (18). It is our unproven assumption that decreasing ED overcrowding will also help to decrease LWBS rates. The data were collected over 9 months. The study was intended to run over a 13-month period, with the first month scheduled as a pilot month. After 9 months (excluding the pilot month) institutions started falling out. The reasons for this were study fatigue and failure to replace graduating residents. The non-academic sites stated that they were unable to maintain the time commitment required. Site investigators who dropped out were asked if they were dissatisfied with the study, and 100% stated no study dissatisfaction. If this study were to be done again, we would try to set up at least two site investigators per site in case one investigator suffered from study fatigue. A national study done in a similar fashion would also increase the volume and value of the data obtained. The investigators would use billing codes and ICU admission rates to define acuity and not resuscitations. We would add patient satisfaction surveys as an outcome variable to be monitored along with overcrowding variables currently in use. LIMITATIONS AND FUTURE QUESTIONS The study objectives were not blinded to the principal investigators or the participating institutions. This could have allowed for investigator bias to be introduced into the study. The use of computers to input the information can limit the gathering of information; it did not allow for written explanations. The input of information at a specific time may not have allowed for thorough randomization. The study fatigue and resident graduation were major limitations to the continuance of the study. It is a major hurdle experienced in all studies requiring multiple investigators at multiple sites. In retrospect, we should have asked for a second site investigator at each site in case the resident moved on or the site investigator was unable to continue to input data. This study excludes smaller EDs, primarily those with ⬍ 15,000 patient visits per year. This was done to avoid the inclusion of urgent but non-emergent centers, although there are many centers that provide emergent care and see ⬍ 15,000 patients per year. The study was designed to measure effects only on State of Michigan EDs. Attempts to apply the results of this study may be subject to regional differences. CONCLUSIONS The documentation of factors involved in ED overcrowding found that overcrowding was not just an ED

FemVein Access Study

problem, but a problem that occurs due to overcrowding in the entire institution. REFERENCES 1. Shute N, Marcus M. Crisis in the ER. US News and World Report Sept 10, 2001:54-61, 64, 66. 2. Sugerman M. Crisis in the ER [Television news series]. San Francisco, CA: CBS affiliate, KPIX; May 20, 2001. 3. Gibbs N. Do you want to die? TIME May 28, 1990.59 – 65. 4. Graff L. Overcrowding in the ED: an international symptom of health care system failure. Am J Emerg Med 1999;17:208 –9. 5. McCabe JB. Emergency department overcrowding: a national crisis. Acad Med 2001;76:672– 4. 6. Gordon J, Billings J, Asplin B, Rhodes K. Safety net research in emergency medicine: proceedings of the AEM Consensus Conference on “The Unraveling Safety Net”. Acad Emerg Med 2001;8:1024 –9. 7. American Hospital Association. Hospital statistics 97/98: emerging trends in hospitals. Washington, DC: American Hospital Association; 1998. 8. Weiss S, Arndahl J, Ernst A, Derlet R, Richards J, Nick T. Development of a site sampling form for evaluation of emergency department overcrowding. Med Sci Monit 2002;8:CR549 –53.

107 9. Derlet RW, Richards JR, Kravitz RL. Frequent overcrowding in U.S. emergency departments. Acad Emerg Med 2001;8:151–5. 10. Richards JR, Navarro ML, Derlet RW. Survey of directors of emergency departments in California on overcrowding. West J Med 2000;172:385– 8. 11. Lambe S, Washington D, Fink A. Waiting times in California’s emergency departments. Ann Emerg Med 2003;41:35– 44. 12. Redemeier 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 1994;23:43–7. 13. Henry MC. Overcrowding in America’s emergency departments: inpatient wards replace emergency care. Acad Emerg Med 2001; 8:188 –9. 14. Viccellio P. Emergency department overcrowding: an action plan. Acad Emerg Med 2001;8:185–7. 15. Lambe S, Washington D, Fink A. Trends in the use and capacity of California’s emergency departments, 1990 –1999. Ann Emerg Med 2002;39:389 –96. 16. Kazzi A. Give emergency medicine true departmental control. West J Med 2000;172:388 –9. 17. Ross M, Wilson AG, McPherson M. The impact of an ED observation bed on inpatient bed availability. AEM 2001;8:576. 18. Arendt K, Sadosty A, Weaver A, et al. The left-without-being seen patients: what would keep them from leaving? Ann Emerg Med 2003;42:317–23.