Lack of Provider Impact on Patient Length of Stay in the Emergency Department
Daniel J. Fahey MD MBA, Heather A. Lindstrom PhD, Paul R. May, MA, Robert F. McCormack MD MBA
Abstract
Introduction: Decreasing ED length of stay (LOS) is a common goal. ED LOS is affected by provider, nursing, and institutional factors. We studied the effect of provider role in time to patient disposition when an existing ED provider group was completely replaced in a busy suburban ED. There were no significant changes to non-provider staffing and general ED operations during the transition.
Objective: Our objective was to determine if providers have a significant role in ED LOS. We hypothesized the new group with a higher proportion of residency trained, board certified physicians and increased physician and mid-level hours of coverage would decrease time to patient disposition.
Methods: We conducted a retrospective review of administrative data on time to disposition for patients who presented to the study ED from September 1, 2011 through August 31, 2012 (original ED provider group) and September 1, 2012 through August 31, 2013 (new ED provider group). The portion of ED LOS most directly under control of providers can be divided into two periods: ED bed assignment to first provider sign-up and first provider sign-up to provider order for patient disposition to admission or discharge. Patient data was separated by inpatient admissions or outpatient discharges. Data was analyzed for one full year in each group, and for the first 6 months of the study period (original group) and the final 6 months of the study period (new group) to remove the impact of factors related to the original group’s contract being terminated and the new group’s period of acclimation to the ED. T-tests were used to compare mean times to provider sign-up and patient disposition.
Results: Time to disposition data was available for 33,038 patient encounters for the original provider group and 32,693 for the new provider group. For inpatient admissions, totaltime to disposition increased by 1.06 min under the new group for the 1 year data, and increased by 7.03 min under the new group for the 6 month data. For outpatient discharges, total time to disposition increased by 8.81 min under the new group for the 1 year data, and increased by 16.94 min under the new group for the 6 month data. Mean total ED LOS was 466 min for admissions and 258 min for discharges. The new provider group was younger, had more emergency medicine residency trained physicians all of whom were board certified, and had increased hours of attending and midlevel coverage. Minor changes in ancillary services also occurred.
Conclusion: Time to patient disposition and provider sign-up varied by ED provider group. While statistically significant, the difference in times is too small to be clinically significant (<5% total time to disposition). Decreasing ED LOS may be best achieved through means other than changing provider groups.
INTRODUCTION:
The goal of an emergency department (ED) is to ensure prompt, quality care for patients.1 For this reason, patients expect immediate assessment, rapid initial management, and either a timely discharge home or an efficient transfer to the appropriate hospital inpatient unit.Any delay in this emergency department processcan result in a prolonged ED length of stay for patients. Prolonged lengths of stayin the ED are not only an inconvenience to the patient, but they are also a safety risk potentially leading to poorer outcomes.2,3
Emergency department efficiency and decreased length of stay (LOS) has become a top priority for ED physicians and hospital administration in order to improve patient care and reduce costs. ED trends showan increasing number of patient visits along with a decreasing number of EDs.4,5With the use of electronic medical records, new ways of recording, monitoring and improvingED efficiency are being implementedby hospitals. In addition, payers such as the Centers for Medicare and Medicaid Services (CMS) have begun looking into ways to link pay to ED efficiency in a pay for performance model.6 CMS now reportsED timing performance metricson its Hospital Compare Website. With an increased focus on ED efficiency, improving LOS has become a common goal for ED’s.7
Emergency department length of stay is a broad term encompassing a series of events which begin at the moment a patient arrives at the ED.Length of stay is affected by many factors in the ED, some of which include triage efficiency, nurse and provider staffing, physician practice patterns, disposition decision making, laboratorytesting and imaging time, attainment of admission orders, and inpatient bed placement.8
One of the longest intervals duringthe ED LOS is time to patient disposition. There are limited studies which look at physician impact on time to patient disposition, and the research that does exist shows that physicians have minimal impact on this component of length of stay.9-12Despite this evidence, physicians are often viewed by administrators as a key contributor to ED LOS.
In early 2012, administrators of a suburban, community hospital decided to terminate the contract with an ED physician group due to concerns about physician productivity and patient satisfaction.A contract was negotiated with a new ED physician group to take over care in the ED. The existing physician group had been providing care in that particular ED for 13 years. A unique opportunity arose to assess the impact of physicians on patients’ EDLOS in an otherwise unchanging clinical environment.
It was hypothesized that the new physician group, which had a higher proportion of residency trained/ board certified physicians andincreased physician and mid-level hours of coverage, would have decreased times to patient disposition.The primary objective in this study was to determine if providers have a significant impacton time to patient disposition, a major component of ED LOS.
METHODS:
Study Design
A retrospective review of administrative electronic medical records (aEMR)of the time to disposition for patients who presented for treatment at the study ED was conducted. The study period was from September 1, 2011 through August 31, 2013. During this time, two different emergency medicine groups were contracted to provide services in the ED: September 1, 2011- Aug 31, 2012 (original EM physician group) and Sep. 1, 2012- August 31, 2013 (new EM physician group).
Study Setting and Population
The study was conducted at a suburban, community ED with an affiliated residency training program and an annual census of approximately 40,000 visits. The ED is divided into two treatment areas: a lower-acuity “fast track”area and a higher-acuity main area. Patients are assigned to areas of the ED based on triage acuity and time of day. Emergency medicine providers include full-time attending physicians, resident physicians, nurse practitioners, and physician assistants. During the period of the study, there were nosignificant changes in ancillary services (e.g. imaging, laboratory), nurse staffing, electronic medical record system usage, in-patient beds, or availability of consulting services thatwould have impactedtime to disposition.
Study Protocol
Patient visit data was abstracted from the administrative electronic medical record system (aEMR), Magellan. Magellan Scorecard is a copyrighted tool of Oracle Business Intelligence designed to allow administrators to monitor healthcare delivery and performance metrics. Eligible visits were those that occurred during the study period (September 1, 2011 through August 31, 2013), occurred in the main ED, and resulted in the patient either being admitted to an in-patient floor or discharged to home or an outpatient facility. Patient visits were excluded from the analysis if the visit was to Fast Track or time data was incomplete.
Emergency DepartmentLOSdata was abstracted for all eligible patient visits. The total ED LOS can be broken into multiple time intervals.Two specific intervals were analyzed: “ED bed assigned to first provider signup” and “first provider signup to provider order for patient disposition. These twoelements of ED LOS were selected for analysis because theyare the intervalsmost directly influenced by the ED physician. Total ED LOS was also calculated. Total ED LOS was defined as the time interval between a patient’s arrival at the emergency department and a patient’s actual discharge from the ED/admission to an inpatient bed.
Aggregate data was also collected for attending physicians in the old and new emergency medicine physician groups. This data was collected from the hospital’s administrative physician database as well as the twoED physician groups’ scheduling records. For each physician group, the meanattending physician age, percent of male physicians, percent of EM residency trained/ EM board certified physicians, and the number of staffing hours was obtained. Attending physicians were considered to be members of the emergency medicine physician group if they were on the group’s schedule at any time during the study period whether full- or part-time, and even if they did not work for the physician group for the entirety of the study period.
The study was reviewed and approved by the University’s Institutional Review Board (IRB).
Outcome Measures
The primary outcome measure was time to patient disposition which was defined as the time from first provider sign-up to provider order for patient disposition to either discharge or admission.
The secondary outcome measure was total ED LOS defined as the time from patient sign in to patient departure from the ED.
Data Analysis
Data was analyzed for one full year in each group, and for the first 6 months of the study period (original group) and the final 6 months of the study period (new group) to remove the impact of factors related to the original group’s contract being terminated and the new group’s period of acclimation to the ED. A statistical analysis was performed using a Welch’s T-test on Microsoft Excel® to compare mean times to provider sign-up and to patient disposition under each ED physician group. A correction for multiple comparisons was performed with the T-test.
RESULTS:
During the study period there was a total of 81,317 visits to the ED with 68,913patient visits to the main ED (Figure 1). Time to disposition data was available for 33,038 patient encounters for the original provider group and 32,693 patient encounters for the new provider group.
The primary outcome was time to patient disposition. Mean time to patient disposition differed significantly between the original and new physician groups for both the full year and 6 month data (Table 1). When a full year of data was analyzed for mean time to patient disposition for each physician group, the new group was4.5min slower than the original group for inpatient/observation admissions and10.3 min slower than the original group for outpatient discharges. A similar pattern was observed for the six month data. Thenew group’s time to disposition was 7.1 minslower for inpatient/observation admissions and 13.5 min slower for time to outpatient discharge than the oririginal group.
The time from emergency department bed assignment to first provider sign-up for each patient visit was also analyzed. Time to first provider sign-updiffered significantly between the original and new physician groups with the exception of the 6 month data for inpatient/observation admissions which showed no significant difference between the physician groups (Table 1). The new physician group had faster mean times to first provider sign-up than the original group for all of the data examined except for inpatient/observation admissions in the 6 month data analysis.
First provider sign up times were combined with time to disposition to generate the total time to patient disposition (Table 1). For inpatient admissions, the new group increased total time to disposition by 1.06 min for the 1 year data and 7.03 min for the 6 mos data. Similarly, for outpatient discharges, the new group increased total time to disposition by 8.81 min for the 1 year data and 16.94 min for the 6 month data.
The number of patient visits during the study period for which total ED length of stay could be calculated was 65,731. The overall mean ED LOSduring the study period was 466 minutes for admissions and 258 minutes for discharges. For inpatient admissions, there was less than a 1% difference in total ED LOS between the original group and new group for the 1 year data, and ≈ 2% increase in total LOS under the new group for the 6 month data.For outpatient discharges, the new group increased LOS by ≈ 3% for the 1 year data and increased length of stay by ≈ 7% for the 6 mos data.
Provider demographics and attending hours of coverage differed between the original and new physician groups (Table 2). As a whole, the new provider group was younger, had more emergency medicine residency trained physicians all of whom were board certified, and had increased hours of attending and midlevel coverage. There were small changes in ancillary services which occurred during the study period (Table 2).
DISCUSSION:
Emergency departments are complex environments requiring collaborative efforts of diverse members of the emergency department and hospital care teams if they are to run in an efficient and timely manner. Patient length of stay in the ED is frequently considered a key process indicator for performance improvement and clinical and operational efficiency.13 ED length of stay is affected by many factors in the emergency departmentincluding triage efficiency, nurse and provider staffing, physician practice patterns,disposition decision making, lab and imaging testing time, attainment of admission orders, and inpatient bed placement.8
Studies have shown that factors with a high impact on length of stay times in the ED include hospital occupancy, number of hospital admissions, and the number of ICU admissions.11 Other research has shown that laboratory practices, triage techniques, and physical plant improvements also have an impact on length of stay times. 12,14-16
There has been limited research which specifically addresses physician impact on length of stay in the ED.10-12,17Rathlev et alexamined the effect of physician staffing on ED length of stay. Thenumber of ED clinical attending hours was used as a representation of “physician staffing”. Variations in clinical attending hours or “staffing” did not change patient length of stay in the ED.10Therefore, the study concluded that physicians did not affect length of stay in the ED. Buchelli et al demonstrated that additional physician staffing reduced the length of stay for emergency department outpatient visits, but had no impact on length of stay for patients admitted to the hospital.17
This research capitalized on a unique opportunity to study whether physicians impact length of stay in the ED. A situation occurred in which it was possible to study the impact of a complete turn-over of attending physicians working in a large, suburban ED, while at the same time keeping all other all factors related to length of stay relatively constant.
In this study, ED LOS was broken into multiple time intervals. In the opinion of the authors, the intervals most directly influenced by physicians are the time from emergency department bed assignment to first provider sign-up, and the time from first provider sign-up to provider order for patient disposition. Other intervals, for example the time from provider order for disposition to patient discharge, are much more strongly influenced by factors outside the direct influence of the physician.
Although there is a statistically significant difference in time to patient disposition by EM physician group, thus suggesting physicians do have a role in ED LOS, the difference in times has minimal clinical significance. Mean time to patient disposition increased by 1 – 7 min for inpatient admissions and 8 -16 min for outpatient dischargesunder the new physician group. However, the mean total ED length of stay during the study period was466 minutes to actual inpatient admission and 258 minutes to actual outpatient discharge . A difference in time to patient disposition ranging from1 to 16 min hardly seems meaningful when compared to this lengthy stay in the ED, and the clinical impact will be minimal.
A second interesting finding was that the new provider group which had more EM residency trained/ board certified providers than the original provider group and increased the number of hours of physician coverage did not reduce ED LOS times. In fact, ED LOS timesslightly increased under the new group.In this setting, the number of hours of physician and mid-level coverage per day and physician demographics did not have a clinically significant impact on ED LOS times. This supports prior studies demonstrating that physician staffing does not have an impact on length of stay.
Limitations:
The primary limitation of this study is that data collection was limited to one suburban ED site. Unfortunately, it was not possible to expand this study due to the infrequency of a provider groups changing and the lack of publicly released ED throughput times.
A second limitation of this study was the availability of only 12 months of data for each provider group. Comparing the first six months of study period data from the original provider group to the last six months of study period data for the new provider group was an attempt to remove the impact of the original group’s contract being terminated and the new group’s period of acclimation to the ED. Extending the time frames of data collection may have changedthe study’s findings although this seems unlikely.