Pediatric emergency department (PED) overcrowding and prolonged boarding times (admission order to PED departure) decrease quality of care. Timely transfer of patients from the PED to inpatient units is a key driver that relieves overcrowding. In 2015, PED boarding time at our hospital was 10% longer than the national benchmark. We described a resident-led quality-improvement initiative to decrease PED mean boarding times by 10% (from 173 to 156 minutes) within 6 months among general pediatric admissions.
We applied Plan-Do-Study-Act (PDSA) methodology. PDSA 1 (October 2016) interventions were bundled to include streamlined mobile communications, biweekly educational presentations, and reminder signs. PDSA 2 (August 2017) provided alternative workflows for senior residents. Outcomes were mean PED boarding times for general pediatrics admissions. The proportion of PICU transfers within 12 hours of admission served as a balancing measure. Statistical process control charts were used to analyze boarding times and PICU transfer rates.
Leading up to PDSA 1, monthly mean boarding times decreased from 173 to 145 minutes and were sustained throughout the study period and up to 1 year after study completion. The X-bar chart demonstrated a shift with 57 consecutive months of mean boarding times below the preintervention mean. There were no changes in PICU transfer rates within 12 hours of admission.
Resident-led quality improvement efforts, including education and streamlined workflow, significantly improved PED boarding time without causing harm to patients.
Pediatric emergency department (PED) overcrowding decreases quality of care and PED throughput. Decreased quality of care due to PED overcrowding can include delayed administration of analgesics to children presenting with sickle cell pain crises or long bone fractures.1,2 PED overcrowding has also been associated with a decrease in quality of care for patients with asthma exacerbation, specifically, increased time to corticosteroid, β-agonist, and asthma score administration as well as increased numbers of patients not receiving one or more of these interventions.3 In addition to effects on quality of care, PED overcrowding has been independently associated with delays in time to triage, delays in time to being seen by a physician, and increased rates of leaving without being seen.4
One of the key drivers that relieves overcrowding in the PED is the timely transfer of patients from the PED to inpatient units (ie, reduced boarding time).5 Boarding time, defined as time from admission order placement to emergency department (ED) departure, is a national quality metric.6 The national benchmark for the median boarding time in academic PEDs in 2016, according to the Academy of Administrators in Academic Emergency Medicine (C. Harbertson, BS, and J. Scheulen, PA, MBA, unpublished data, 2016), was 114 minutes. Compared to this, the median boarding time in 2016 at our urban, academic, tertiary-care children’s hospital was 27 minutes longer (141 minutes).
Researchers addressing the problem of ED overcrowding have largely focused on the downstream impact of inpatient-bed availability on ED boarding times and demonstrated an association between increased hospital occupancy and ED length of stay (LOS).7–11 Resident physicians are integral to a teaching hospital’s admission workflow; however, no studies to date have focused on how the habits of a resident workforce can impact the admissions process for the entire hospital. This quality improvement (QI) study was driven by pediatric residents and targeted resident practices. We aimed to decrease the mean PED boarding time by 10% (from 173 to 156 minutes) within 6 months among patients admitted to general pediatric units by targeting pediatric resident admission workflow.12
This study was conducted at an urban, academic, tertiary-care hospital in Baltimore, Maryland. During the study period, the institution’s PED received ∼34 000 visits per year consisting of Maryland residents, specialty referrals (primarily from DE, PA, VA, DC, and WV), and international patients. Among these visits, 5000 to 6000 hospital admissions occurred annually, with 29% to 40% of these admissions going to the general pediatric inpatient service.
Patients admitted to the general pediatrics service from the PED were the target study population. Excluded were pediatric subspecialty (given that fellows were more actively involved in the admission process for such patients), surgery, psychiatry, and PICU admissions because these departments had different admission protocols.
The admission workflow consists of a cascading sequence of notifications, starting with the PED attending physician notifying the PED charge nurse of an admission and ultimately ending with the floor resident paging acceptance of the patient. Once the PED attending physician makes the decision to admit a patient, the PED charge nurse contacts the shift coordinator for a bed and inpatient team assignment. The PED resident caring for the patient in the PED then sends a page to the senior resident of the assigned inpatient team notifying them of the admission. Once the acceptance page is received by the PED, floor charge nurses, and shift coordinator, the PED bedside nurse is notified of the acceptance. The PED bedside nurse then calls the inpatient unit nurse who will be caring for the patient to sign out. After nurse-to-nurse sign out, the patient is then transported from the PED to the inpatient unit.
Historically, interns were trained by their senior residents to admit patients in the following sequence: receive handoff from the PED physician, conduct a history and physical examination in the PED, place admission orders, and then send the acceptance page. Per hospital policy, nurses in the general pediatric inpatient unit would not receive patients from the PED without the acceptance page from the residents.
We analyzed the factors associated with PED boarding times. A key driver diagram and process map of admissions from the PED were developed to identify a bottleneck in the workflow (Figs 1 and 2). Using the key driver diagram, we identified resident communication, nursing communication, inpatient-bed availability, resident availability, nursing availability, availability of clinical support staff, and PED-workflow efficiency as key drivers. The process map of PED admissions revealed that the inpatient resident team performing a history and physical examination in the PED before admission was a bottleneck in the workflow. To further inform our key driver identification and process-map derivation, we conducted a survey of PED nurses to better delineate barriers to timely admission. When delays were noted, delayed resident acceptance for PED admission was the most cited barrier to timely admission (Fig 3). We theorized that these delays were due to either floor residents forgetting to send acceptance pages or floor residents conducting in-person history and physical examinations of patients in the PED before sending an acceptance page, both of which are easily modifiable aspects of resident workflow.
This was a QI study using Plan-Do-Study-Act (PDSA) cycles from October 2016 to May 2018. The 11-person, multidisciplinary QI team was led by 7 pediatric resident physicians, 1 PED nurse, 2 PED attending physicians, and 1 general pediatrics attending physician. For PDSA 1, we focused on reminding admitting resident teams to send their acceptance page and recommended using a mobile paging application so the acceptance page could be sent on the spot, which would leave fewer opportunities for the resident to become distracted and forget to send the page. The bundled intervention consisted of visual reminders (displayed on all resident computer workstations throughout the children’s center) to send acceptance pages in a timely manner and in-person, biweekly educational meetings during floor residents’ morning conference. Educational meetings reinforced the message that patients with an uncomplicated medical diagnosis (eg, cellulitis) did not require a secondary evaluation in the PED by floor residents before being accepted to the floor. During PDSA 1, we received feedback from the residents that some were hesitant to accept patients to their service without at least seeing the patient first regardless of their medical diagnosis. Therefore, for PDSA 2, we specifically targeted senior residents with an alternative workflow that would provide residents with the opportunity to see admitted patients in the PED as well as expedite the admissions process at the same time. This alternative workflow is indicated as steps 1 to 5 in Fig 2. Briefly, we proposed that senior residents do a quick assessment of the patient in the PED and send their acceptance page before obtaining their full history and physical. The rationale for targeting senior residents was twofold: (1) senior residents are most often responsible for paging acceptance, and (2) we theorized that changing institutional culture would need to start with changing the behavior of senior residents, who play a large role in educating interns on efficient hospital workflow.
Study of the Intervention
Statistical process control (SPC) analysis consisted of X-bar and s charts to assess monthly mean PED boarding times (QI Macros, version 2019.03; KnowWare International Inc, Denver, CO).
Primary outcomes were monthly mean PED boarding times extracted from the electronic health record (EHR) (Epic; Epic Systems Corporation, Verona, WI).6 Boarding time was defined as time from admission order placement to PED departure. For a balancing measure, we assessed the proportion of patients transferred to the PICU within 12 hours of admission from the PED to ensure that our intervention was not increasing the rate of PICU transfers after admission.
The Institute for Healthcare Improvement rules were applied to identify special cause variation in our control charts over time.
This study was reviewed by the hospital’s institutional review board (IRB00132352) and was determined to not constitute human subjects research under Department of Health and Human Services or Food and Drug Administration regulations.
During the intervention period (October 2016 to May 2018), 9082 patients were admitted from the PED to the children’s hospital. Of those, 2649 patients were admitted to the general pediatrics service from the PED. During the baseline period (August 2014 to September 2016), there was an average of 122 general pediatrics admissions per month (median 116; range 70–174). The intervention period had a monthly average of 134 general pediatrics admissions per month (median 141; range 77–186).
Downward movement in mean boarding times began one year before our intervention. Using SPC analysis of monthly mean boarding times, we identified a shift in mean boarding times from 173 to 145 minutes beginning in August 2016, which was 2 months before the start of PDSA 1 (Fig 4A). SDs of the mean boarding time also decreased around this time (Fig 4B). After implementation of PDSA 1, monthly mean boarding times were maintained at 145 minutes. After PDSA 2, in which we focused our educational sessions on senior residents with an alternative admission workflow, we found that there was no significant change in boarding times. Of note, the mean boarding time of 145 minutes and reduction in SD that began in August 2016 was maintained up to one year after completion of our study.
The balancing measure was the proportion of PICU transfers during the first 12 hours of admission, analyzed by using SPC (p-chart) (Fig 5). The average number of general pediatrics admissions per month for the entire study (August 2014 to May 2018) was 127 (median 134; range 52–186), and the average number of 12-hour PICU transfers was 1.6 (range 0–6). The average proportion of 12-hour PICU transfers was 1%. There was no shift in the proportion of patients transferred to the PICU within the first 12 hours of admission, and the process remained in control throughout the duration of our study.
Using process mapping, we identified the resident admission workflow as a potential cause of prolonged PED boarding times due to delays in transferring admitted patients from the PED to the inpatient floor. We targeted a change in resident admission workflow with a bundled intervention, which included paging using a mobile phone application, biweekly educational presentations, and reminder signs. SPC charts showed a significant reduction in mean boarding times leading up to our intervention, and this reduction was maintained during our interventions without causing increased transfers to the PICU. We demonstrated sustainability and cultural change through sustained PED boarding times 43 months after the onset of our interventions. Future directions include building the acceptance paging process into Epic so that PED residents and admitting residents can communicate through a single system. Currently, a separate paging system must be used to notify the admitting team. Bed allocation and team assignment is also done manually by the shift coordinator. Automation of this process through Epic could streamline the admissions process from the time the admission order is placed to notifying the resident team of an admission. Eliminating the acceptance paging process altogether or defaulting the admission to accepted unless the senior resident sends a page to stop the admission are other possibilities under consideration.
Our results demonstrate that resident-led interventions directed at resident workflow efficiency contributed to reduced PED boarding time and thus helped to improve an important PED quality metric at this academic medical center. Several researchers have examined the effect of trainees in their staffing models on ED crowding and resource use. These researchers have shown that teaching hospitals at which patients are seen by a trainee and supervised by an attending physician have longer ED LOS when compared with hospitals at which patients are seen only by attending physicians.13–16 Although the literature reports that EDs staffed by trainees have increased LOS, our results suggest that throughput metrics can be improved through interventions targeting resident efficiency. Future studies might include evaluation of other ED or hospital throughput metrics (eg, total ED LOS, inpatient LOS, time to hospital, or time to ED discharge) after applying similar interventions.
Beginning in December 2015, we observed that mean boarding times were decreasing before the start of our intervention. There were 3 changes to resident team workflow during the 2015–2016 academic year that may account for the decreased boarding times. First, the 2 general pediatrics inpatient teams began to alternate admissions, which may have distributed the workload more evenly among the teams. In previous years, admitting teams were assigned on the basis of the geographic location of the patient. Thus, if several beds were available in one area of the hospital, then all new admissions would have gone to the team covering that area, whereas the other team would receive none of the admissions. Second, the general pediatrics teams stopped doing admissions for the pediatric neurology service during daytime hours. Third, general pediatrics teams began to only cofollow the general pediatric surgery patients as opposed to all surgical patients in the children’s center (orthopedics, urology, otolaryngology, and neurosurgery). Coupled with lower patient volume in the spring and summer months, these changes may have had the overall effect of decreasing resident workload and therefore allowing residents to be more available for evaluating and accepting admissions from the PED. However, if decreased boarding times could be attributed to improved workload alone, then one would expect to see an increase in mean boarding times during the fall and winter months when patient volume increased. Instead, we continued to see decreased boarding times into the winter months after implementation of our intervention in October 2016.
The shift in mean boarding times began in August 2016 as the first of 8 points below the center line. Our QI project was inspired by ongoing informal efforts by PED attending physicians to expedite the admissions process by verbally reminding admitting residents to send acceptance pages or to send their acceptance page before seeing stable patients. These efforts leading up to our intervention could explain the shift in mean boarding times before the start of PDSA 1. Implementation of Epic on the inpatient floors in July 2016 could have also contributed to improvement in boarding times through facilitating information sharing by enabling the admitting team on the inpatient floor to access the medical charts of patients in the PED. The interventions implemented by our QI project formalized existing efforts and took advantage of a newly unified EHR system to maintain improved boarding times, which persisted up to 1 year after the completion of the study.
There was one outlier of monthly mean boarding times in Fig 4A that warranted exploration. This outlier of elevated PED boarding times occurred during February 2018. On review of medical records for this month, there were several patients with extremely prolonged boarding times in the PED due to lack of bed availability when our hospital was at full capacity. Additionally, this month coincided with the 2018 influenza epidemic peak during the influenza season, with the greatest number of influenza cases seen over the past 4 seasons.17 Our study highlights that resident leadership in QI initiatives at academic medical centers can result in a significant and sustainable culture shift among residents and thus improved hospital quality metrics. By identifying bottlenecks in resident workflow, resident leaders were able to address inefficient behaviors in pediatric residents that had been propagated by tradition from year to year. We conducted a systematic review of the literature examining the effectiveness of improving patient care through resident involvement in QI initiatives and found only 28 published articles in which residents were actively engaged in all components of the QI project (of which only 16 articles mentioned improvement in process outcomes).18 Our study was unique in that we addressed resident workflow culture and its direct relation to patient throughput. Active resident leadership in this QI initiative was critical to its success by identifying key drivers in resident workflow and facilitating cultural acceptance of our intervention.
Our study had several limitations. First, the acceptance paging process was a system unique to our institution, and as such, it may limit the generalizability of this work to other hospital systems. We did not have a system in place to collect acceptance page times and therefore used PED boarding times as a surrogate measure of improvements to resident workflow. We were unable to assess other factors contributing to prolonged PED boarding times, including bed availability as well as nursing workflow. Lastly, our data were collected through the electronic medical record and subject to potential misclassification of data. Occasionally, PED providers misclassified the admitting service when placing the admission order (eg, EHR lists a patient under general pediatrics when in fact they were admitted to pediatric hematology). These misclassified patients were also occasionally included in this analysis. An audit of 3 months chosen at random before and after our intervention revealed that up to 4% of patients may have been misclassified before the intervention and up to 10% of patients may have been misclassified after the intervention. The impact on the data of these patients being included in our study would be a false increase in the mean boarding time per month because of the fact that the admitting process for specialty services, in general, requires several more steps in comparison with a general pediatrics service. Thus, we may have seen greater improvements had they been excluded from the study. In addition, admitting residents on specialty teams were still exposed to the intervention at various points, although they were not formally targeted (eg, visual reminders were placed on resident workstations in the PED).
Resident-led QI efforts directed at resident workflow efficiency were critical in maintaining reduced PED boarding times, which is an important PED quality metric. Other academic centers may improve hospital and PED metrics by targeting interventions to resident workflow. Resident behaviors impact teaching hospital throughput, which is relevant to medical centers addressing overcrowding while aiming to improve the patient experience of care.
Drs Kouo and Kleinman conceptualized and designed the study, drafted the initial manuscript, designed the data collection instruments, collected data, and reviewed and revised the manuscript; Dr Fujii-Rios drafted the initial manuscript, collected data, and reviewed and revised the manuscript; Dr Falco collected data and reviewed and revised the manuscript; Drs Badaki-Makun and Kim conducted the initial data analyses and critically reviewed the manuscript for important intellectual content; Dr Canares supervised study design, coordinated and supervised data collection, conducted the initial data analyses, and critically reviewed the manuscript for important intellectual content; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.