Improving Efficiency on a Pediatric Hospital Medicine Service With Schedule-Based Family-Centered Rounds Free

This was a single-center pilot quality improvement project on the PHM service at a 361-bed free-standing children’s hospital in northern California. The PHM service has ∼1400 discharges annually, cares for children with general and complex medical conditions and serves as the primary service for the neurology, endocrinology, hematology, and genetics/metabolic services. There are 2 attending physicians (1 university-based and 1 private practice) who are assigned on the basis of the patient’s primary care provider. The team is composed of 1 senior resident, 1 to 2 interns, 1 to 2 medical students, 1 case manager, and, at times, a PHM fellow or chief resident. Interns rotate in 4-week blocks. During the first 7 months of the intervention, senior residents rotated in 2-week blocks, which changed to 4-week blocks in July 2019. Attending physicians, fellows, and chief residents are on service for 7-day blocks. Patient census varies seasonally with an average of 10 patients per day. Rounds started at 8:50 am, and senior residents were required to attend a multidisciplinary meeting at 11:20 am on weekdays. A second PHM team with a similar team composition was started during the middle of the project period (June 2019) and thus was not involved in this intervention. Before the intervention, the rounding order was decided by the senior resident as rounds progressed, and a case manager was responsible for calling nurses during weekdays when the team arrived at the patient’s room, but there was no advanced notice before that call.

A multidisciplinary workgroup of stakeholders met monthly for 9 months before the intervention period to develop the process and metrics. This group included representatives from the university-based PHM division, the private practice group, pediatrics residents and residency program leaders, case management, nursing, interpreter services, clinical informatics, patient and family experience, patient education, and our family advisory council.

The Model for Improvement was used as a framework for the project. Before implementation, we surveyed attending physicians and residents to understand overall satisfaction with rounds and perceptions of efficiency and teaching. Both groups identified patient census and complexity, family and team expectations about the length of rounds, and inconsistent workflow with rotating team members as drivers of inefficiency. We also surveyed nurses to understand barriers to attending rounds, which included lack of timely notification, concurrent rounds on other patients, and other urgent patient care responsibilities. Attending physicians, residents, and nurses were all in support of pursuing a scheduled rounds process to address these concerns.

Our interventions were designed to address the following key drivers: a shared mental model for team member roles in rounds coordination and daily rounding plan, enhanced communication between team members about coordinating rounds, resident ownership of the schedule, data transparency, and family awareness of the scheduled rounds process (Fig 1). Details of the interventions are described below.

We adapted an EHR-linked scheduling tool created by our clinical informatics department and another subspecialty service at our institution (Fig 2).¹⁴  The tool used a simple user interface created by our internal web developer that was accessible through the EHR (Epic Systems, Verona, WI). It combined automatically populated data from the patient’s record (patient name, room number, preferred language, bedside nurse name, and direct contact number) with custom fields to identify the attending group (university-based or private-practice) and early discharges. Early discharges were seen by the attending before the start of rounds, were not included in that day’s schedule, and did not affect the start time of rounds. Although the early discharge process was established before this intervention,⁸  the scheduling tool allowed for them to be denoted visibly in the EHR. A prearranged schedule with default order based on the attending group, unit, and interpreter need was automatically generated. Default durations for individual rounding encounters were assigned on the basis of interpreter need (10 minutes for English-speaking patients, 12 minutes for non-English-speaking patients; increased to 11 and 13 minutes, respectively, in July 2019). These durations could be altered as needed and were based on the average durations of rounds encounters collected during the preintervention phase. The tool also allowed users to rearrange the patient order and add blocks for transit, teaching, anticipated admissions, or other needs. Based on resident feedback, the scheduling tool was adapted twice during the intervention period. In December 2018, we changed the order of the columns to improve usability and added additional buffer slots. In July 2019, we improved the clarity of the instructions, increased default rounding durations, and added a teaching slot.

Before starting service, senior residents were emailed instructions and an instructional video on the use of the scheduling tool (Supplemental Fig 4) and expectations regarding rounds workflow (Supplemental Fig 5). A document detailing expectations for the coordination of rounds for attending physicians was presented at division meetings, distributed via e-mail, and posted in the attending workroom. Strategies to support ending rounds by 11:20 am during times of high census were reinforced during weekly meetings, which included identifying early discharges, adjusting default rounding durations, and seeing patients with a smaller group of providers outside of the rounding schedule, when applicable. Education on the process for nurses was presented at staff meetings and reviewed regularly in nursing huddles.

Senior residents were responsible for publishing the schedule by 7:50 am (1 hour before the start of rounds). Unit secretaries would then text a photo of the schedule to the nursing staff via our Health Insurance Portability and Accountability Act-compliant secure text messaging system (Voalte Mobile App, Hillrom, Sarasota, FL).

Each morning, attending physicians would check in with senior residents regarding early discharges, acute patient care issues, and the rounding plan. Either the case manager (on weekdays) or the attending (on weekends) notified nurses just before rounds by phone or secure text message. This in-the-moment notification supplemented advanced notification of the scheduled time as nursing understood that there was a 30-minute window to give some allowance for earlier arrival times or delays. As our institution has a large Spanish-speaking population, there were different workflows for Spanish and non-Spanish interpreters. During weekdays, Spanish interpreters would reference the schedule and join rounds at the scheduled times. Non-Spanish interpreters were scheduled individually.

Weekly check-in meetings were held with nursing, case management, clinical informatics, and the medical team to discuss issues with the workflow and scheduling tool in real time. Data on metrics for the previous week were reviewed at this meeting. To promote data transparency, a change was made 6 months after implementation to also post these metrics in the resident workroom.

After scheduled rounds had been implemented for 6 months and data revealed that rounds encounters started consistently within +/− 30 minutes of the scheduled rounding time, nursing staff notified families of a 1-hour timeframe during which rounds should occur (eg, scheduled time of 9:30 am corresponded to a timeframe of 9–10 am). Nursing also educated families at the time of admission about the scheduled rounds process and the overall purpose and duration of rounds (Supplemental Fig 6). This intervention was disseminated through nursing staff meetings, daily huddles, weekly check-in meetings, and nursing badge cards.

We targeted 3 primary outcome measures. First, we aimed to start 90% of rounds encounters within 30 minutes of scheduled rounding time (defined as +/− 30 minutes). If a patient’s scheduled rounding time was 9:30 am, the target was met if the encounter started between 9:00 and 10:00 am. We chose this metric because it could be consistently met even with some deviations from the schedule. Second, we aimed to have 90% nursing presence during rounds, as this has been shown to enhance team communication and family involvement.¹  Third, we aimed to have 80% of daily rounds completed by 11:20 am to allow for residents to attend a daily required multidisciplinary meeting and a noon teaching conference. We also analyzed this metric separately for days with census <12 and ≥12. We tracked the creation of the schedule as a process measure. We reviewed formal rotation evaluations before and during the intervention to assess ratings of teaching as a balancing measure.

Data were collected daily on the time that rounds encounters started for each patient, nursing presence on rounds, overall rounds start and end times, family member presence on rounds, interpreter use, daily patient census, schedule creation, and scheduled rounding times. Preintervention baseline data collection was limited to 2 time periods, 4 weeks over March to April 2018, and 4 weeks over October to November 2018 just before the initiation of the intervention, because these were time periods when we had sufficient staffing resources for an observer (case manager or resident on quality improvement elective) to collect data during weekdays. For 1 year after the start of the intervention (November 2018–November 2019), data were collected on rounds daily by a case manager on the weekdays and by attending physicians on the weekends. The case manager collected these data using an electronic tablet to fill out a Research Electronic Data Capture (REDCap) form hosted at Stanford University.¹⁵  Attending physicians collected the same data on a paper form, which was then entered into the same REDCap database. This quality improvement work was deemed to be nonhuman subjects research by our institution’s institutional review board.

Data from REDCap were first analyzed by using Stata version 15 (College Station, TX; StataCorp LLC) to calculate weekly performance on metrics and then analyzed by using run charts. Overall comparisons between the preintervention and intervention periods were made by using a t test for continuous variables and χ² test for categorical variables.

Baseline data captured 360 rounds encounters over 40 days during March to April 2018 and October to November 2018, and the postintervention data captured 2846 encounters over 350 days from November 2018 to November 2019. The average daily census was 10 patients (standard deviation [SD] 2.8), and 14% of days had at least 1 early discharge (range 1–3). Approximately 11% of encounters used an interpreter. A family member was present on rounds for 75% of encounters during the preintervention period and 79% of encounters during the intervention period (P = .07). A schedule was created for 349 of 350 days (99.7%). The 1 day it was not created was due to a technical issue with the scheduling tool which was fixed later that day.

Rounds encounters started within 30 minutes of scheduled rounding time for a median of 83% of encounters in the first 1.5 months, then increased to 87% of encounters for the next 6 months of the intervention with weekly variability (Fig 3A). The centerline increased to 95% for 1.5 months after improved resident education, decreased to 83% for 2 months, and then increased back to 96% for the last 1.5 months.

Nursing presence remained high throughout the intervention. The centerline increased from a baseline median of 79% to 90%, where it was sustained for 5 months, decreased to 86% for 2 months, and then increased to 94% for the last 5 months (Fig 3B). We reached our target for nursing presence through notification of scheduled rounding times every morning and real-time notification just before rounds.

The percentage of rounds ending by 11:20 am increased from a median of 0% to 57%, where it was sustained through the first 6 months of the intervention (Fig 3C). The centerline then increased to 86% for the subsequent 3 months, during the period of resident data transparency and improved resident education, decreased to 43% for a month, then increased to 86% for the last month. Performance on this metric was met on 195 of 220 days (75%) with a census of <12 and 25 of 90 days (28%) with a census of ≥12 (P < .001).

The average duration of rounds was shorter during the intervention period when compared with the preintervention period (140 [SD 39] vs 167 [SD 40] minutes, P = .003). For days with census >12, the average rounds duration was also shorter during the intervention period (intervention: 167 minutes [SD 31], preintervention: 187 minutes [SD 26], P = .01). There was no significant difference in the average duration for days with a census <12 (intervention: 131 minutes [SD 37], preintervention: 139 minutes [SD 42], P = .2).

Weekly meetings revealed the following challenges: resident unfamiliarity with the scheduling tool and process, ending rounds on time when the census was high, and planning for extended discussions on complex patients. Residents generally found the scheduling tool to be easy to use and helpful for creating a rounding order ahead of time. In evaluating our balancing measure, resident rotation evaluations revealed that perceptions of the quality of teaching before and during the intervention remained high.

Our PHM service started rounds encounters within 30 minutes of scheduled rounding time 96% of the time and showed a significant increase in bedside nursing presence within 1 year after implementation of scheduled-based rounds. The multifaceted nature of our intervention, which focused on a shared mental model of a daily rounding plan and team member roles for rounds coordination, enhanced team member communication about coordinating rounds, and resident ownership of an EHR-linked scheduling tool, was likely a major factor in its success.

Nursing presence on rounds has been shown to improve team communication and support parent participation,³  and previous literature has linked the process of scheduling rounds encounters with improvements in nursing presence and efficiency.^11,14,17  One institution implemented an appointment-based rounds process and improved bedside nurse attendance from 30% to 40% to 72% along with reported improvement in patient/family satisfaction.¹¹  A subspecialty service at our institution with patients on a single geographic unit implemented scheduled rounds and improved nursing presence from 69% to 87%.¹⁴  Another institution incorporated a rounding schedule on their complex care service as 1 element of a comprehensive rounds intervention across all PHM teams, which improved efficiency as measured by rounds end times, attending note completion, and nursing presence.¹⁷  Standard processes for rounds workflow have been associated with improvements in efficiency,^18,19  and having a schedule is 1 method to provide a shared mental model of the rounding plan. The combination of having a scheduled rounding time easily accessible through an EHR-linked schedule and notification just before rounds sent by a consistent team member likely contributed to our >90% nursing presence, higher than revealed by other similar studies.^11,14,17 

Our improvement in the metric of ending rounds by 11:20 am was associated with a decrease in the average daily census, and our performance on this metric was much lower with a census ≥12 patients. Despite reminders of scheduling strategies that could be used when the census was high and the ease of planning them using the EHR-linked scheduling tool, it is possible that unfamiliarity with these scheduling strategies contributed to the performance on this metric, given that residents and attending physicians change frequently. Another contributing factor may have been the complexity of our patient population and the team and/or family’s desire to extend the rounds discussion beyond the scheduled duration. Additionally, with our defined rounds start and end times and default rounding durations, there was a limit to the number of encounters that could be fit into this timeframe without using the previously discussed scheduling strategies. Despite the lower performance on this metric during days with a high census, the decrease in average rounds duration on days with a census >12 suggests that the scheduled rounds process may have contributed to overall rounding efficiency even if the team did not end by the target time. Our improvement in this metric also coincided with the introduction of a second PHM team, which allowed for a decreased census on the team with scheduled rounds. During times of high census, a combination of scheduling strategies and staffing interventions, such as having an additional attending to assist with triage, discharges, and admissions during rounds, or adding additional provider teams may be optimal to improve efficiency.^17,20 

Many factors likely contributed to our intervention’s success. Establishing buy-in from multiple stakeholders was crucial to promoting acceptance of workflow changes. The EHR-linked scheduling tool’s ease of use fostered quick adoption of the process and integration into the senior resident’s daily workflow. Another institution that reported success with a scheduled rounding process required a half-time coordinator to manage the schedule daily,¹¹  which may not be feasible in many settings. Institutions interested in this process could work with their informatics teams to build a similar tool by utilizing a variety of technical tools from spreadsheet macros to a web application such as ours. We also promoted senior resident ownership of the process because they were the most knowledgeable of the patient care and team needs relevant to planning the schedule. Designing our metrics to allow the team a window of time to round on a patient allowed for flexibility when unexpected situations arose, such as longer conversations or unforeseen patient care needs, and allowed for greater acceptance. Having a system to contact nurses in real time through phone calls or secure text messages was crucial to coordination. Lastly, weekly meetings allowed for timely review of metrics and discussion of real-time barriers, facilitators, and potential solutions. We noted that the metrics revealed occasional downward shifts after periods of higher performance. We believe this was likely due to a variety of factors, including unfamiliarity with the process due to changing team members, unexpected interruptions and/or delays, and addressing urgent care needs. Consequently, these weekly meetings were instrumental in providing ongoing education about the process and strategies to address possible barriers to rounding efficiency in real time.

This study has several limitations. First, this intervention was implemented at a quaternary children’s hospital with many complex patients often requiring extended discussions. Therefore, the relationship between ending rounds on time and patient census may vary by patient complexity. Second, we were unable to track the fidelity with which nurses provided families with information about the scheduled rounds process and/or the daily rounding timeframe. However, weekly conversations with nursing leadership reassured us that this information was reviewed regularly with the nursing staff. Third, we were unable to collect preintervention data for >8 weeks because of resource limitations. Fourth, this work was conducted with ongoing clinical informatics support and case manager support during rounds, which may not be generalizable to other settings. Lastly, we did not collect data on quality or safety of care, the patient/family experience, or health care worker burnout, and thus cannot evaluate the impact of a scheduled rounds process on these outcomes.

A multifaceted scheduled rounding intervention focused on shared expectations, team member communication, and resident ownership of an EHR-linked scheduling tool improved the efficiency of our rounding workflow and bedside nursing presence during family-centered rounds. This process continues to be part of our standard workflow >3 years after implementation and has also been started on our second PHM team and another subspecialty service. Ongoing work since the end of the study period has focused on improving family notification of rounding times and the implementation of a short educational video to be shown to families at admission to orient them to the rounds process. Future steps include a qualitative evaluation of family perspectives, automating notification of rounding times, and spreading this intervention to other teams at our institution.

This project was supported by the Lucile Packard Children’s Hospital Stanford (LPCHS) Value Improvement Program. We acknowledge the valuable contributions of the Stanford Division of Pediatric Hospital Medicine, Palo Alto Medical Foundation Pediatric Hospitalist Group, Stanford Pediatric Residency Program, Stanford Pediatric Cardiology Scheduled Rounds Workgroup, and the following departments from Stanford Children’s Health/LPCHS: Case Management, Nursing, Unit Secretaries, Interpreter Services, Information Services, Performance Improvement, Clinical Care Liaisons, Office of Patient and Family Experience, Patient Education, and Family-Centered Care. We also thank the following individuals: Shirley Cheung, MSN, for her guidance during the planning process and assistance with baseline data collection, Katherine Muse, MD, for assisting with baseline data collection and resident education efforts, Ben Schwartz and Matthew Randolph for assisting with data entry and analysis, Amit Borcar and Joshua Faulkenberry for their work in the programming of the EHR-linked scheduling tool, and Lauren Destino, MD, for her helpful comments in the preparation of this manuscript. Finally, we thank the patients and families of Lucile Packard Children’s Hospital Stanford.