Methods

Study design and setting

This study was conducted at Parkland Health & Hospital System, a 900-bed academic public hospital in Dallas, Texas, which houses the 24th busiest ED in the USA.11 We conducted a simple interrupted time-series study in which the primary outcome measures were (1) ED time to admission orders (EDTAO), defined as the time from inpatient bed request to receipt of admission or transitional orders, and (2) EDBT, defined as the time from bed request to departure from the ED (figure 1). We included all patients admitted to the internal medicine service from the ED between 23 May 2007 and 30 June 2008. The study period was divided into five phases (baseline: 23 May 2007 to 30 June 2007; postintervention phase 1: 1 July 2007 to 30 September 2007; phase 2: 1 October 2007 to 31 December 2008; phase 3: 1 January 2008 to 31 March 2008; and phase 4: 1 April to 30 June 2008). The admission process was redesigned over a 9-month period. The redesign process included three steps: (1) detailed assessment of the current state process; (2) development of a future state process, termed RAP; and (3) creation of a change management architecture. The University of Texas Southwestern Medical Center Institutional Review Board approved the research protocol.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Rapid Admission Protocol (RAP) Workflow.

Assessment of the current state

A small team of clinical and administrative leaders from the Departments of Internal Medicine and Emergency Medicine was convened to lead this effort. Members of the team walked through the entire admission process from entry in the ED to arrival in the inpatient unit. The clinical workflow was examined using a variety of parameters, including level of acuity and inpatient destination. Each step in the admission process was documented on a process map, which was subsequently revised through a series of focus groups with key informants. During this examination, the process map was layered with cultural and organisational factors that could not be captured in a linear diagram; many of these factors reinforced inefficiencies in the admission process. The final process map revealed 50 to 75 steps between bed request and time to admission orders. Most of the participants agreed that the process was excessively complex, redundant and, in some capacities, unsafe. Nearly 25 problems with the admission process were identified; these were rank ordered by the team to select areas requiring immediate focus for investigation. The top 10 problems are reproduced in table 1.

Development of a RAP

After considering problems inherent in the existing admission process, the project team developed a set of eight guiding principles to inform the construction and implementation of a new RAP (table 2).

Perhaps similarly to other academic hospitals, the team discovered that the admission process had grown knotty and tangled over the years. It was felt that the process required a sudden and transformational change, implemented at a fixed point in time. Once the decision was made to discard the old process, the team sketched various admission strategies over 2 months using the guiding principles. Once the first draft of RAP was agreed upon, the team sought feedback from staff in multiple clinical departments (internal medicine, emergency medicine, radiology, surgery and pathology), resulting in many minor and at least four major revisions of the process. In the 2 months before going live, negotiations became more intense between departments as we sought approval at the chairperson level; the RAP underwent 52 further revisions during this period. The final admission process was reduced from 50 steps to 10.

The team set 1 July 2007 as the implementation date since a new class of house staff would begin that day. Setting a deadline for implementation also lent a sense of urgency to the process, a component of successful change efforts noted in other industries.13 Three weeks before the implementation, physician and nurse staff received extensive in-service education about RAP through town-hall meetings and small-group sessions.

Among many changes, the new RAP specifically differed from the original process in four important ways. First, physician roles were narrowed and simplified. Previously, a single internal medicine house-staff officer, named the admitting officer on duty, was responsible for evaluating admitted patients to determine if additional studies or consults were needed, assigning the admitting service, communicating details of the admission to the receiving team and temporarily managing care of the patient. This convergence of responsibilities produced bottlenecks and resulted in unnecessary care transitions. In the new process, the role of the admitting officer on duty was narrowed to a handful of clinical activities (figure 1). Second, we instituted transitional orders that could be initiated by ED physicians to help expedite the transfer of clinically stable patients to the admitting unit of the hospital. These transitional orders are completed only after a structured hand-off communication occurs with the admitting team; this communication often results in considerations or recommendations from the receiving service. Transitional orders are valid for a maximum of 4 h and immediately superseded by admission orders from the primary team. To make the transition as safe as possible, we did not allow transitional orders for patients admitted to the ICU or for ward patients who were deemed clinically unstable by the admitting officer on duty or emergency physicians.

Third, the internal medicine team assumed full responsibility for the patient after either the admission or transitional orders were received. In our focus groups, ED nurses expressed frustration that it was sometimes unclear which set of physicians needed to be contacted for patient issues during the hand-off between services. With the new protocol, a “bright line” was created that clearly articulated responsibility for the patient at any stage in the admission process (figure 1). Fourth, before RAP, the desire to fully “package” a patient by completing all diagnostic tests before departure from the ED impeded flow. In the new protocol, routine patients are transferred to the admitting hospital unit as soon as a bed is available, even if the primary admitting team has not yet evaluated the patient. If the patient arrives to the unit, and full admission orders are not available, the primary team is paged and expected to evaluate the patient quickly.

Change management architecture

Heeding John Kotter's principle of developing a “guiding coalition” for any change effort, we developed a committee of clinical and administrative leaders in the internal medicine and emergency medicine departments to oversee implementation.13 This group was very familiar with RAP and met weekly for 2 months after implementation to discuss the performance of the initiative, review patient care problems, assess statistical process control charts and approve amendments to the protocol as needed. Members of this committee served as highly visible champions for RAP.

Statistical analysis

We calculated the EDTAO and EDBT for all patients in the study period using timestamps in the clinical information systems at Parkland Hospital. To assess differences with the baseline period, we calculated a median value for both outcome variables by phase and inpatient destination. Both EDTAO and EDBT were non-normally distributed; thus, we used the two-sample Wilcoxon rank-sum test to compare each of the postintervention phases with the baseline phase. All statistical tests were performed using STATA V.10 (Stata, College Station, Texas, USA). Statistical significance was set at p<0.05.