Characterization of emergency department abandonment using a real-time location system

a b s t r a c t

Background: Patients who present to emergency departments (EDs) for evaluation but are noted to have Left without being seen (LWBS) are potentially at great risk. Governmental agencies, such as the Centers for Medicare and Medicaid, as well as hospitals and health organizations, are examining the factors which drive LWBS, includ- ing accurately quantifying patient tolerance to wait times and targeting interventions to improve patient toler- ance to waiting.

Objective: Compare traditional methods of estimating time to LWBS with an objective method using a real-time location Tracking system (RTLS); examine temporal factors associated with greater LWBS rates.

Methods: This is a retrospective cohort study of all ED visits to a large, suburban, quaternary care hospital in one calendar year. LWBS was calculated as patient registration to nurse recognition and documentation of patient abandonment (traditional method) vs registration to last onsite RTLS timestamp (study method). Descriptives of patterns of patient abandonment rates and patient demographic data were also included.

Results: Our study shows that traditional methods of measuring LWBS times significantly overestimate actual pa- tient tolerance to Waiting times (median 70, mean 92 min). Patients triaged to resource intensive categories (Emergency Severity Index 2, 3) wait longer than patients triaged to less resource intensive categories (ESI 4, 5).

Conclusion: Compared to traditional methods, RTLS is an efficient and accurate way to measure LWBS rates and helps set the stage for assessing the efficacy of interventions to reduce LWBS and reduce the gap between those seeking evaluation at emergency departments and those ultimately receiving it.

Introduction

Background

Left without being seen (LWBS), or patient abandonment is a major problem in emergency departments (EDs), with potential for negative impact on patient safety and patient-centered healthcare. The Centers for Medicare and Medicaid Services utilize LWBS rates as one of its de- terminants of payment in its Hospital Outpatient Quality Reporting Pro- gram [1], highlighting the gap between patients presenting for evaluation and care and those receiving it as an area with significant

? Presented at: ACEP Scientific Assembly Research Forum 2017.

* Corresponding author at: Department of Emergency Medicine, Mayo Clinic, 200 First St. SW, Rochester, MN 55906, United States of America.

E-mail address: [email protected] (H.A. Heaton).

potential for improvement. Previous studies have identified length of wait, updates on expectED wait time, triaged priority level, time of day, and month of year [2-6] as important factors which increase likeli- hood of LWBS and impact time to abandonment. The most commonly identified reason for leaving was the length of waiting time, and many studies have focused on identifying and implementing interventions to cut down on wait times in an attempt to reduce LWBS rates, with some success [6-11]. However, accurately quantifying exact ED aban- donment times has historically been challenging, with up to 31% of cal- culated LWBS waiting times deemed to be inaccurate [12]. Methods of capturing time to ED abandonment vary and are often not clearly de- scribed. Typically, the measurement process starts at the time of patient registration and ends at a point when ED staff recognize and document their departure, which is often only when the patient is called to a room and noted to have already left (a “latest possible” approach) [3,10,13]. Previous studies have measured overestimation of patient

https://doi.org/10.1016/j.ajem.2019.06.025

760 J.M. Geers et al. / American Journal of Emergency Medicine 38 (2020) 759–762

abandonment times at approximately 25 min between when the pa- tient left and when they were noted to have left by recording in the chart [14], or simply stating that the “latest possible” approach inher- ently overestimates time waited [13].

Importance

Better understanding of patient tolerance to wait times allows EDs to develop realistic targets for wait times [3], identify potential interven- tions that reduce waiting times, or alternately establish interventions to increase patients’ tolerance to waiting [2,11,15]. Several previous studies showed that patient tolerance to waiting is around 70 min be- fore leaving [13,14], though acknowledged that this time was a “latest possible” time due to reliance on staff recognition that the patient had left [13]. By implementing a more objective patient tracking system, in- stitutions can more accurately quantify patients’ tolerance to waiting, allowing for potential interventions prior to abandonment. Additionally, this information can be compared to national averages to compare an ED’s LWBS to national benchmarks and provide information for goal wait-times across Emergency Severity Index levels.

Goals of this investigation

This study’s objective was to compare discrepancies of ED abandon- ment times as collected through a Real-Time Location System (RTLS), thought to best represent actual patient abandonment time, with times collected through the traditional timestamp method (which relies on staff recognition of patient absence). In doing so we can better quan- tify actual patient tolerance to wait-times in our ED, and describe the gap between the “latest possible” method and actual patient abandon- ment times.

Methods

Study design

This study is a retrospective cohort analysis. Data used for this report were collected as part of existing processes and practices, and this work was determined as exempt of review by our Institutional Review Board (IRB).

Study setting

This study is based on data for an entire year, from January through December 2016, within our ED at a large academic hospital in a Midwest suburban community, which is a quaternary care and Level 1 trauma center. The ED manages over 77,000 patient visits annually. In the study period of 2016, 82% of patients were categorized as adults (N17 years). From the ED, 35% of adult patients and 16% of pediatric pa- tients were admitted.

Our ED utilizes a real-time location system (RTLS) which is com- prised of Radio-frequency Identification (RFID) hardware (Quake Global, Inc.), including 194 in-ceiling passive RFID readers with 734 an- tennas broadcasting over 212 locations. The locations span the ED and the adjoining radiology department, covering 54,450 total square feet. ED employees, including providers, nurses, health unit coordinators, ra- diology technicians, phlebotomy technicians, electrocardiogram (ECG) technicians, child life specialists, and finance/registration staff have pas- sive RFID tags in their identification badges. ED patients are given an RFID-embedded wristband with a unique identifier at the time of ar- rival. Subsequently, reader antennas pick up signals from these wrist- bands to generate a real-time location for each patient. RFID data are recorded in real time, and stored for one year on an internal server. An internally developed departmental tracking board (YES Board (R), Ambi- ent Clinical Analytics) is used throughout the department to display pa- tient information, as well as staff and patient RTLS locations. The RTLS

was in operation and reliable for three months prior to this data collection.

Methods and measurements

Data on all patients that abandoned the ED after registration during the period of study were collected. The timestamp of the electronic medical record (EMR) using the traditional method was compared against RTLS timestamp of abandonment for each individual patient. Traditional, or timestamp, abandonment time was computed as the du- ration between the patient registration and the earliest time when the nurse noticed and documented the patient being absent. This approach mimics the current standard of practice in almost all EDs. RTLS aban- donment was computed as the duration between patient registration and the timestamp of the patient physically leaving the ED waiting area (last known timestamp recorded) from RTLS system.

Discrepancy 1/4 Abandonment_Timestamp-Abandonment_RTLS

Analysis

The goal of this study was to quantify the discrepancy between a tra- ditional “timestamp” method of capturing abandonment, which is the current standard of practice, and the use of RTLS. The analysis per- formed included descriptive statistics and comparisons. Abandonment behavior is characterized by determining variation by ESI levels, num- ber of patients in the waiting area when patients abandoned, and vari- ation by time of day, day of week and month of year. The discrepancy is computed as mean and standard deviation.

Results

Characteristics of study subjects

A total of 1181 patients abandoned the ED in the 2016 calendar year. Of these, only 1.1% (n = 13) were pediatric patients (b18 years). Adults (18-64 years) comprised a total of 83% of the LWBS population, further broken down in our analysis as youth (18-24) at 13%, young adults (25-34) at 22.8%, then adults (35-64) at 47.2%. The elderly population (>=65) comprised the remaining 16%. By gender, males represented 43.9% and females 56.1% of patients who left without being seen. The breakdown is seen in Table 1.

Results

Actual patient abandonment times obtained through RTLS (Table 2) had a mean of 98 min (median 68, SD 194). Patients triaged as more re- source intensive ESI categories (ESI 2-3) tended to tolerate longer wait times, while those triaged to ESI 4-5 were quicker to leave without being seen. On average there were 13 (median 13, SD 7) patients in the waiting area when patients abandoned, as shown in Table 3; this did not vary significantly by ESI.

Table 1

Demographics.

	Number of patients	Percentage
Age
Pediatrics (age b 18)	13	1.1%
Youth (age 18-24)	153	13.0%
Young adult (age 25-34)	269	22.8%
Adult (age 35-64)	557	47.2%
Senior (age >= 65)	189	16.0%
Gender
Male	519	43.9%
Female	662	56.1%

J.M. Geers et al. / American Journal of Emergency Medicine 38 (2020) 759–762 761

Table 2

Time waited by patients before abandonment (in minutes).

	Mean	Median	Standard deviation
ESI 2	182	62	634
ESI 3	97	70	92
ESI 4	84	65	78
ESI 5	56	33	60
Total	98	68	194

Of all patients leaving prior to an evaluation, 23% left between 5 and 7 p.m., and nearly 50% left between 4 and 9 p.m. (Fig. 1). Tuesday had the highest rate of abandonment, and weekends had the least (Fig. 2). Summer months had the highest number of patients leaving without being seen with August the highest followed by July and September; January had the lowest rate (Fig. 3).

Table 4 defines the discrepancy between “timestamp” methods of documenting patient abandonment time and RTLS-determined times. Notably, traditional methods overestimated the time after abandon- ment for every patient in the entire cohort, by a mean of N1.5 h (mean 92 min, median 70, SD 111).

Discussion

LWBS is a major problem as it has the potential to lead to adverse pa- tient outcomes and significant morbidity. Capturing this data through traditional, “timestamp” methods with reliance on ED staff overesti- mates times of abandonment by an average of over 1.5 h. This study found that actual patient tolerance for waiting times is overestimated by traditional measurement methods, namely methods which rely on staff to note patient absence from the waiting area and document this time in patient records, and is better quantified by a real-time location identification system. An objective system such as the RTLS is consis- tent, reliable, and reproducible, and can be used to capture actual wait times more accurately than traditional methods. Our data show that pa- tient tolerance for waiting time was consistent across ESI levels 2-4 at median 62-70 min, but the lowest acuity patients in ESI level 5 aban- doned after a median of only 33 min. We used median wait times to make our analysis robust against outliers, especially the ones observed in the smaller ESI 2 population which skewed mean wait times in this group.

Consistent with previous study findings were our average total LWBS rates of 1-3%, [3,13,16] as well as previous studies’ findings of a discrepancy between actual abandonment times and traditional means of capturing this data. One study showed a discrepancy of roughly 25 min between the earliest recorded time a patient was noted to have left and the official end time of their medical chart [14]; this study shows an even larger gap, with a mean discrepancy of 92 (median 70) minutes between staff-recorded and RTLS departure time. This should be interpreted in the light of findings that LWBS rates vary significantly by the demographics of populations they serve – for example, hospitals serving patients who are younger, more often uninsured, or low income tend to have higher LWBS rates independent of wait times [17]. Our ED population as measured by RTLS waited a me- dian of 68 min before leaving, which may represent a significantly higher tolerance to waiting than other institutions whose “latest

Table 3

Number of patients in waiting room at time of abandonment.

	Mean	Median	Standard deviation
ESI 2	14	14	7
ESI 3	13	13	7
ESI 4	13	13	7
ESI 5	10	10	8
Total	13	13	7

Fig. 1. Abandonment of ED by hour of the day.

possible” approach showed mean LWBS times of 70 min, [14] suggest- ing they actually left earlier than this. Further studies in a variety of dif- ferent hospital sizes and settings will help in contributing to the robustness of these statistics and establishing and understanding trends in LWBS rates.

Further explored by this study was the hypothesis that patient aban- donment rates are related to times at which the ED was historically busiest and indeed showed clinically and statistically significant rises in rates of abandonment in the afternoon and evening hours. This study’s data also suggests higher abandonment rates in summer months and lower rates on weekends.

Findings would suggest that establishing an ideal wait time may de- crease LWBS rates. Previous studies suggest goal wait times of 45 min for ESI 3 and 60 min for ESI 4/5, to achieve a goal of LWBS rates of b2.0% [3]. Other studies have aimed interventions directed at decreasing LWBS through methods such as providing information on wait times and nursing care in the waiting area, such as providing analgesic medi- cations, ice packs, and bandages, which did seem to increase tolerance to wait times [2]. One prospective survey showed patient interest in having a “time tracker” board in the ED to provide wait time informa- tion, with the strongest preference from patients who were triaged as ESI 4 or 5, [18] in keeping with previous findings that higher-acuity pa- tients (ESI 2-3) found wait time to be less important than lower-acuity patients (ESI 4-5) in their decision to stay vs leave without being seen.

Strengths of the study

As far as the authors are aware, this is the first study which objec- tively quantifies patient wait times, and compares this objective mea- sure with traditional measures of time to patient abandonment of the

Fig. 2. Abandonment of the ED by day of the week.

762 J.M. Geers et al. / American Journal of Emergency Medicine 38 (2020) 759–762

Table 4

Discrepancy of traditional means in capturing abandonment (in minutes).

	Mean	Median	Standard deviation
ESI 2	124	70	313
ESI 3	89	72	70
ESI 4	93	69	80
ESI 5	55	63	40
Total	92	70	111

Declaration of Competing Interest

None.

Fig. 3. Abandonment of the ED by month of the year.

ED. The RTLS provides reliable, reproducible LWBS data, and does not rely on staff attention or intervention to quantify patient abandonment. The study ran over a full year and included all visits to the ED, resulting in a robust sample size and excellent power. Additionally, due to its de- sign, this study could easily be repeated and lends itself to further re- search on actionable interventions to reducing wait times and improving patient tolerance for waiting.

Weaknesses of the study

The RTLS, or similar system which tracks patient location and can recognize real-time patient abandonment, is not a universal or even standard feature in most emergency departments across the nation. The cost of installing such a system will likely be prohibitive for a num- ber of organizations. Additionally, this is a single-center study with a pa- tient population which has unique characteristics; this demographic and the study’s results may therefore not be directly applicable to other hospitals and should be interpreted with care.

Conclusion

ED abandonment is an issue which can have detrimental effects on patient care, safety, and patient experience. While this is of significant importance, lack of objective data has inhibited studies and design of in- terventions. This study leveraged RTLS to characterize accurate aban- donment times and discrepancy with traditional recording methods. With a better understanding of actual wait times and patient tolerance to wait times, further studies can be conducted aimed at redesigning op- erations, with the goal of improving patient wait time tolerance and overall decreasing ED abandonment rates.

Grant support

None.

References

AJEM

Characterization of emergency department abandonment using a real-time location system

a b s t r a c t

Introduction