A Systems Approach to Front-End Redesign With Rapid Triage Implementation.

The most common site for hospital sentinel events due to care delays, secondary to waiting and/or inefficient processes, occurs in the emergency department (ED). Decreasing patient length of stay in an ED is a key initiative for many hospitals in order to maximize both quality and efficiency. The purpose of this practice improvement project was to (1) standardize front-end processes across a 6-hospital health system, (2) move non-sorting-related clinical questions out of triage, and (3) improve door-to-triage and door-to-provider times. The project occurred within a 6-hospital East Coast health system. This was a continuous quality improvement initiative utilizing the Donabedian theoretical model, plus the DMAIC method, for process improvement. A system-wide performance work team was formed including ED leaders and staff; site-specific implementation teams were also formed. Rapid triage implementation was effective in producing statistically significant improvement in door-to-triage, door-to-provider, and ED length of stay for discharged patients at 3 of the 6 sites. Further performance improvement projects in this area are needed to better understand the generalizability of this process in other EDs. Furthermore, from a leadership perspective, additional investigation is needed into the cost savings as well as shared labor opportunities that may exist when policies and processes are standardized across a system's service line.

CONTRIBUTION TO EMERGENCY NURSING PRACTICE

The purpose of this practice improvement project was to (1) standardize front-end processes across a six-hospital health system, (2) move non-sorting-related clinical questions out of triage, and (3) improve door-to-triage and door-to-provider times.

The primary outcome of this practice improvement project was statistically significant improvement in door-to-triage, door-to-provider, and ED length of stay for discharged patients.

Key implications for emergency nursing practice based on this project include the value of standardized policies/processes across a health system as well as the impact of a streamlined triage process to improve door-to-provider times.

INTRODUCTION

Problem Description

The emergency department (ED) is the most common site for hospital sentinel events due to care delays secondary to waiting and/or inefficient processes (Murrell, Offerman, & Kaufman, 2011). The Joint Commission maintains a “Quick Safety” issue reinforcing the negative impact delays have on patient care and outcomes (The Joint Commission, 2015). Some EDs have maintained antiquated processes, with bottlenecks resulting in these care delays. One such outdated process is the inclusion of mandatory screening questions and other data collection queries as part of the triage process (Foley & Durant, 2011). This process occurs at some sites even when there are available patient treatment spaces in the department as well as a qualified medical provider ready to see the patient.

An East Coast six-hospital faith-based health system embarked on a 4-year journey to rightsize the organization for optimum quality care at the most efficient cost. The health system comprises 1,928 certified hospital beds, 790 nursing home beds, and approximately 18,400 employees. Its six EDs care for more than 228,000 annual visits; each ED's annual volume ranges from 21,644 to 85,208 visits. Although left without being seen rates were below 2% at every site, lengths of stay (LOSs) were an identified area of performance improvement across the entire system, as was the arrival-to-provider time metric at certain sites.

A system-wide service line assessment identified wide variations in both facility triage policies and practices. The health system set out to standardize and improve the arrival-to-provider process.

Available Knowledge

All six EDs utilize the Emergency Severity Index (ESI) as its triage stratification tool (Gilboy, Tanabe, Travers, & Rosenau, 2020). Three months of preintervention throughput metrics and volumes for each site are detailed in Table 1.

Table 1.

Preintervention metrics

Aug–Oct 2018	ED1	ED2	ED3	ED4	ED5	ED6	Combined
Discharge visit volume	14,491	5,372	5,704	3,727	3,681	2,885	35,860
Admission visit volume	6,117	2,133	2,286	1,529	2,884	4,269	19,218
Total visit volume	21,302	7,864	8,431	5,411	6,800	7,283	57,091
Actual annual volume (Aug 18–Oct 19)	85,208	31,456	33,724	21,644	27,200	29,132	28,364
Mean door-to-triage (min)	3.4	1.7	11.7	8.8	4.5	4.9	5.3
Mean door-to-provider (min)	22.8	24.6	32	43	21.8	32.9	27.5
Mean ED discharge LOS (min)	239.9	234.6	280.6	210.2	212.5	319.8	246.1
Mean ED admit LOS (min)	599.3	440.1	582.4	589.5	467.5	627.4	565.3
Mean LOS all patients (min)	352.2	293.4	363	321.3	326.3	499.9	358.5

Note. Time intervals rounded to the 10th decimal point. ED = emergency department; LOS = length of stay.

Four sites conducted a comprehensive data collection process in a triage booth for all patients arriving by means other than ambulance; ED1 already utilized a rapid triage process. ED6 utilized a two-step triage process prior to bed placement.

Rationale

There are many approaches to improving the door-to-provider metric including, but not limited to, the provider-in-triage method (Pierce & Gomley, 2016), immediate bedding (Howard, 2011), and split flow (Bish, McCormick, & Otegbeye, 2016), among others (Martin, 2012; Murrell et al., 2011). In this case, a standardized process was needed that could work in EDs of varying footprint and patient volume.

The intent was to significantly reduce the collection of information during triage that is not relevant to the act of sorting. Wolf et al. (2018) described cautionary trends when implementing a rapid triage process. They described hospitals triaging patients without any physiological data, subsequently raising concerns about the impact on patient outcomes. For the purposes of this project, the team remained conservative in its approach to what was collected during triage; full vital signs collection, among other queries, continued on all patients during triage.

Specific Aims

The specific project aims were to improve door-to-triage and door-to-provider times. This was to be achieved through moving non-sorting-related clinical questions out of triage. This project also focused on consolidating site-specific front-end policies into one system-wide policy.

METHODS

Context

This project was completed using the “structure-process-outcome” framework described by Donabedian (1988). This model has been frequently utilized by researchers as well as those in public policy to map out the mechanics of a particular situational process. The continuous quality improvement initiative utilized the DMAIC method for process improvement. DMAIC stands for Define, Measure, Analyze, Improve, and Control (Moran, Burson, & Conrad, 2017). A system-wide team of ED nursing leaders, ED physicians, clinical educators, clinical informaticists, and other key stakeholders was formed to (1) consolidate the six triage policies into one system-wide policy with intended processes, and (2) redesign the electronic health record (EHR) components while keeping rapid triage in mind.

Site-specific performance work teams were subsequently formed to design the individual site-specific process mechanics of the redesigned approach. Participation and frequency of meetings varied by site to the degree of change that was needed in each ED. Because ED1 already utilized a rapid triage approach, the impact only consisted of educating the nurses about the forthcoming EHR changes, which were more intuitive to the process already in place. Conversely, this was a major undertaking for ED3, ED4, and ED5. These teams were much larger and involved more staff-level participation, including physicians and advanced practice providers.

Intervention(s)

As part of the intervention, the EHR was parsed into the three distinct groups described in Table 2:

Table 2.

Updated EHR documentation configuration

Primary triage	Screenings	Primary assessment
Arrival InfoaArrival DocaChief compliantaTriage assessVitalsaPain assessmentaOB/GYN statusHeight/weightaAllergiesaSuicide riskRespiratory screenTravel screenaED triage notesESI/destinationaLWBS or to L&D	Sepsis screenStroke screenAbuse indicatorsHIV testingTetanus statusSBIRT screeningAdvanced directives	Primary assessmentPain assessmentOB/GYNBreastfeedingHistoryMedications taken PTAOutside medsFall riskImplantsExternal medicalCare everywhereTriage interventionsOrdersOrder sets

Note. ED = emergency department; EHR = electronic health record; ESI = Emergency Severity Index; L&D = labor and delivery; LWBS = left without being seen; PTA = prior to arrival.

aDenotes a policy/process required field in triage.

“Primary triage” is completed upon patient arrival.

“Screenings” is completed by a registered nurse (RN) at the earliest opportunity after the primary triage. If another patient is waiting for triage, this part is deferred for completion at a later time, likely after bed placement.

“Primary assessment” is completed by an RN at the earliest opportunity after the patient is placed in his or her treatment space; this is most often completed by the primary RN.

Utilization of routine metric sharing with staff and the intentional desire of using the Hawthorne effect as a positive change catalyst were also woven into the leadership implementation strategies.

Measures

The measures chosen for studying the process outcomes included the door-to-triage, door-to-provider, and ED discharge LOS time intervals. ED admission LOS and overall LOS data were also collected. However, it is hypothesized that LOS metrics were not directly impacted by the front-end process redesign due to the numerous extraneous variables.

Raw data were sent by health system information technology personnel to a third-party vendor via a HIPAA-compliant analytics platform. Data were obtained from the platform in both aggregate and raw formats; none of the data for the throughput metrics had unique patient identifiers.

Visits were excluded if they met any of the following criteria:

Negative ED LOS;

Cases where facility name, discharge disposition, time of ED disposition, or time of arrival are null; and

Invalid time exclusions: Any record where the time of arrival is after the triage time or physician contact time.

Operational Definitions

Comprehensive data collection is defined as collecting information that is not relevant to the act of sorting. Examples of this include, but are not limited to, lethality screening, tuberculosis screening, obtaining a medication history not related to the chief complaint, fall risk, immunization status, and domestic violence victim screening among others (Foley & Durant, 2011). Because of triage processes varying on the basis of setting, location, and situation, this project focused on triage operations in U.S. EDs during normal operations; that is, operations that do not include disaster situations, mass casualty situations, and pandemic situations.

Front-end process refers to an institution's specific process implemented to receive, register, triage, and initiate provider evaluation. Although the terminology refers to a process as opposed to a time interval, the process start and ending points are synonymous with the door-to-provider interval.

Assurances of Data Completeness

Analytics platform information is received from the EHR via a data transfer; this represents the complete data for a specific time frame and as such no sampling techniques were utilized. Analytics platform data were refreshed daily and viewable in both daily and monthly formats.

Ethical Considerations

This performance improvement project did not meet the definition of research under 45 C.F.R. 46.102(d). Access to relevant patient data occurred through a HIPAA-compliant analytics platform.

RESULTS

The postintervention metrics are detailed in Table 3. The t-test correlation coefficient was used to establish the presence or absence of statistical significance with this project. The t-test calculations at a 95% confidence interval are further detailed in the “Discussion” section. Key outcomes are graphed in Figure 1.

Table 3.

Postintervention metrics

Aug–Oct 2019	ED1	ED2	ED3	ED4	ED5	ED6	Combined
Discharge visit volume	13,776	5,142	5,751	3,351	3,539	2,846	34,405
Admission visit volume	6,137	2,075	2,436	1,756	2,827	4,782	20,013
Total visit volume	20,770	7,556	8,565	5,314	6,568	7,788	56,561
Annualized volume (Aug 19–Jul 20)a	83,080	30,224	34,260	21,256	26,272	31,152	226,244
Mean door-to-triage (min)	3.3	2.2	8.3	7.7	3.3	3.3	4.4
Mean door-to-provider (min)	20.4	14.4	17.5	48.7	19.1	35	23.7
Mean ED discharge LOS (min)	250.9	241.3	246	213.7	202.6	283.5	242.8
Mean ED admit LOS (min)	713.5	418.2	595.6	645.7	384.5	514.7	567.3
Mean LOS all patients (min)	395.9	294	347.4	361.6	282.1	426.1	361.9

Note. Time intervals rounded to the tenth decimal. ED = emergency department; LOS = length of stay.

aAnnualized volume is based on visits through February 29, 2020.

Figure 1.

Project key outcomes. ED = emergency department; LOS = length of stay.

The preintervention exclusionary cases included 164 cases with negative LOS or null components, as previously identified. In addition to this, 156 cases were excluded because of invalid times. Thus, 320 of 57,411 cases were excluded, or 0.56%.

The postintervention exclusionary cases included 181 cases with negative LOS or null components, as previously identified. In addition to this, 207 cases were excluded because of invalid times. Thus, 388 of 56,949 cases were excluded, or 0.68%.

DISCUSSION

Summary

There were no significant improvements at ED1; in fact, overall ED LOS significantly increased in all groups. ED2 demonstrated significantly longer door-to-triage times; however, the door-to-provider metric was significantly reduced (p < 0.0001). Statistically significant reduced door-to-triage times were seen at ED3 (p < 0.0001), ED4 (p = 0.0175), ED5 (p < 0.0001), and ED6 (p < 0.0001); ED3 and ED5 also had a significantly reduced door-to-provider times (each p < 0.0001). A statistically significant reduction in LOS of discharged patients was observed at ED3 (p < 0.0001), ED5 (p = 0.00102), and ED6 (p < 0.0001).

Analyzing all visits as an overall system, the project resulted in statistically significant improvements in the door-to-triage (p < 0.0001), door-to-provider (p < 0.0001), and ED discharge LOS (p < 0.0185) time metrics.

Interpretation

The lack of performance improvement at ED1 was secondary to the ED already having a rapid triage process in place. Furthermore, there was significant construction in the triage area during the postintervention measurement period. The construction may have led to inefficiencies that increased processing times throughout patient ED visits.

The results at ED2 interestingly show overall significant improvements in the door-to-provider metric but not the door-to-triage metric. It was identified in the preintervention period that patients queuing for triage may have not had properly back-timed arrival times. This was corrected through education, but the result of more accurate arrival times has lengthened LOS. Remarkably, the door-to-provider times were still shorter. This may be due to a Hawthorne effect and focus on improving those times; it may also be simply due to the fact that the time interval is longer and thus variations in 1 or 2 min did not impact overall calculations.

Limitations

This project represents a performance improvement initiative at six different sites within one organization; although it was an effective approach as a system, further investigation at other organizations is needed to determine its generalizability.

Furthermore, although these changes were an intense focus for the ED, performance improvement initiatives do not occur in a vacuum. There were many other concurrent hospital improvement initiatives at each facility aimed at reducing throughput times throughout the organization. There is confidence that this intervention reduced door-to-triage and door-to-provider times; however, there are other potential extraneous factors that could have influenced discharge LOS, and it is highly probable that admission LOS and overall LOS had significant influence from outside factors, such as inpatient bed availability and staffing, among others.

CONCLUSION

The primary outcome of this practice improvement project was statistically significant improvement in door-to-triage, door-to-provider, and ED LOS for discharged patients at three of the six sites; the improvement was also statistically significant for the overall health system.

Key implications for the emergency nursing practice based on this project include the value of standardized policies/processes across a health system, as well as the impact of a streamlined triage process to improve door-to-provider times. Multidisciplinary staff-level involvement was a key strategy to designing a staff-owned process with buy-in. It is important for the emergency advanced practice provider to recognize the value of rapid and accurate sorting. Getting the patient to the qualified medical provider in this expeditious process accelerates patient care and throughput.

Further performance improvement projects in this area are needed to better understand the generalizability of this process in other EDs. Furthermore, from a leadership perspective, additional investigation is needed into the cost savings as well as shared labor opportunities that may exist when policies and processes are standardized across a system's service line.