Using Quality Improvement Methodology to Increase Communication of Discharge Criteria on Rounds Free

This quality improvement initiative took place between July 2019 and June 2021 (baseline data collection July to August 2019 and intervention November 2019 to June 2021) at a 296-bed, tertiary care, pediatric academic medical center. It focused on 2 PHM teaching services, with attending physicians supervising residents and students. Our PHM has 4000 discharges annually. PHM rounds typically occur as daily PFCR with the interprofessional team at the bedside with the child and caregiver and includes the attending, senior resident, 2 to 3 interns, 2 to 3 students, the bedside nurse, case manager, and other clinical and support service teams as applicable to the patient, such as pharmacists, dieticians, therapists, or interpreters. At baseline, an interprofessional committees is in place that focuses on improving local discharge practices. An interprofessional QRIP is also in place to observe and enhance the quality of rounds. The QRIP uses a checklist to teach learners elements to include in PFCR. Our electronic health record (EHR) is Epic (Verona, WI).

Our improvement team consisted of 2 providers, 1 nurse, 2 medical students, 1 case management specialist, 1 EHR specialist, 1 data analyst, and 1 project manager. We conducted a current state analysis using direct observation to understand the baseline frequency and approach to discharge discussions during rounds. A driver diagram was developed to identify drivers that contributed to achieving our aim and change ideas to test (Fig 1).

The Model for Improvement was used with plan–do–study–act cycles conducted to implement and test 5 main interventions:

We identified variable perspectives regarding the perceived importance of including discharge criteria in PFCR and estimated frequency of inclusion at baseline. To address this, we developed targeted education for student orientations, resident housestaff conferences, and PHM faculty meetings regarding the rationale for inclusion of discharge criteria discussions in PFCR, relevant literature to support its inclusion, and baseline rates to identify opportunities for improvement.

Before this initiative, our institution used a checklist to highlight key elements to include in PFCR. This checklist is used to educate learners regarding best practices for PFCR and share data with teams regarding their performance. To support the inclusion of discharge criteria during PFCR, it was added as a checklist item. We found that teams were more likely to include discharge in PFCR for patients near discharge. Teams reported a lack of comfort with how to include this for patients earlier in their course. To account for this, scripting was added to the checklist outlining how to incorporate discharge criteria for patients at various stages of their hospital course to promote inclusion regardless of timing.

Rounds observations and key stakeholder interviews identified that EHR notes often guide presenters on what content to include in their PFCR presentations. We therefore next sought to add discharge criteria as a standard element in EHR progress notes to prompt its inclusion in PFCR. A pop-up flowsheet discharge SmartForm was developed (Fig 2) to support documentation of discharge criteria, and hyperlinks were embedded into note templates to increase SmartForm access. The SmartForm included selectable options for common criteria and free text options to add additional criteria as needed. Once this content was entered on the SmartForm, it would automatically populate into daily progress notes. These SmartForm responses were recorded within flowsheet rows for efficient data collection.

We performed quarterly audits of measures broken down by teams and shared reports at resident housestaff and PHM faculty meetings as a reminder of the project aims to promote sustainability.

An additional barrier we identified was trainee discomfort determining discharge criteria for patients. We therefore added buttons to the SmartForm for common diagnoses that automatically selected predetermined criteria for specific diagnoses including asthma, bronchiolitis, neonatal fever, and hyperbilirubinemia (Fig 2). Discharge criteria for these diagnoses had been developed for local clinical practice guidelines before this initiative but not widely referenced by teams. The SmartForm was modified so that those diagnosis-specific criteria would populate when the respective diagnosis-specific button was selected. The criteria could then be modified as needed based on patients’ individualized goals.

Measures are outlined in Table 1. Our outcome measure was the percentage of all PHM PFCR observed through our QRIP that included discussion of the patient’s discharge criteria. At baseline, rounds are observed weekly through our QRIP program to assess for inclusion of key elements. For this initiative, an element was added noting whether discharge criteria were included. Rounds were initially observed by 2 observers simultaneously and scored independently, noting if discharge criteria were discussed to ensure consistency between observers. After consistency was established, 1 observer collected the remainder of observations. Teams were informed when observers were present and instructed to proceed as they typically would. No prompting was provided by the observer during PFCR. All observed PFCR were included in the data regardless of family presence.

Our process measure was the percentage of all patients discharged from the PHM service with discharge criteria documented in the SmartForm before discharge, as this was a key intervention implemented. This was obtained from a flowsheet report that indicated whether criteria had been entered on the SmartForm.

Our balancing measure was PHM rounds’ length to assess if inclusion of discharge criteria inadvertently prolonged rounds length. This was determined by the teams tracking the total length of rounds in minutes and dividing it by the number of patients rounded on. Additional balancing measures were average LOS and 7- and 30-day all-cause readmission rates for all PHM patients.

We used plan-do-study-act cycles to make iterative changes to test interventions.

We used statistical process control charts (P and X bar/S charts) created using QI Charts Macro for Microsoft Excel (Denver, CO) to assess for changes in measures. Standard tests¹⁸  to distinguish special cause variation from common cause variation were applied, including shifts when 8 or more consecutive points above or below the center line or points outside upper and lower control limits. Control limits were included corresponding to ± 3 σ limits from the mean. The study team met weekly to review data, address input received, and determine next steps.

Our institutional review board determined these activities nonhuman subjects research. SmartForms is a trademark of Epic Systems Corporation.

Data were obtained from 68 baseline PFCR observations from June to August 2019 and 632 intervention period observations from November 2019 to June 2021. EHR records for 5568 patients were reviewed during the intervention period to assess for documentation of discharge criteria in the EHR SmartForm, LOS, and readmission rates.

To establish a baseline frequency of discharge criteria discussions during PFCR, rounds were observed for 68 PHM patients, with discharge criteria discussions occurring a mean of 32% of the time. After interventions of sharing rationale for including discharge criteria during PFCR and modification of the PFCR checklist to include discharge criteria, this increased to a mean of 77%, indicating special cause variation. Rates further increased to a mean of 90% after data audit with feedback sharing and SmartForm adaptation to include standardized criteria for common diagnoses. This has been sustained for 10 months (Fig 3).

At baseline, there was no centralized location to document discharge criteria. After development of the SmartForm, criteria were documented in the SmartForm for 17% of patients. After modifying the note template to include a hyperlink to the SmartForm, this increased to 58%. With the subsequent addition of standardized criteria for common diagnoses to the SmartForm, this improved to 70% (Fig 4). Special cause variation was also seen intermittently when data points were outside the control limits. January to February 2020 and October 2020 correlated with recent intervention implementations with subsequent higher rates that were not sustained. June 2020 may be correlated with an increase at the end of the academic year before a new class starting the following month.

Rounds length remained stable at ∼11 minutes per patient throughout the intervention. Overall LOS and 7- and Supplemental Figs 5 and 6).

Using quality improvement methodology, we successfully achieved our aim of increasing the percentage of observed PHM PFCR that included discharge criteria discussions to >75% without increasing rounds length.

Effective communication is paramount to a safe transition home. It has been well documented that breakdown in communication can be detrimental to the discharge process.^2,4–6  Initiatives such as this, which increase discharge criteria communication with the interprofessional team and family, are an important step in enhancing the discharge process.

There is evidence to support the value of communicating discharge criteria and the positive impact it can have on the quality and timeliness of discharge.^9–11  Less is known, however, on how to feasibly communicate this information throughout an interprofessional team and with patients and families. Our study was unique in the ability to successfully increase both verbal and electronic communication of discharge criteria and sustain that communication over multiple months.

We intentionally focused on communicating discharge criteria rather than solely expected discharge dates.^19,20  Concerns have been cited that communicating dates, especially early in the hospitalization, may lead to sharing inaccurate information subject to frequent changes. Particularly in pediatrics, predicting course trajectories is challenging.^21,22  We instead sought to focus on sharing discharge criteria which teams and families could mutually be aware of that would remain more constant throughout the hospitalization.

We also sought to focus on communication with both the broad interprofessional team and patients and families, recognizing the key roles all members have in an effective discharge. Although initiatives such as whiteboards²¹  can communicate information with individuals present in the room, this may not be readily viewable by some members of the team. This may be increasingly true in the postpandemic era of personal protective equipment preservation when room entries by some members of the team may be reduced to conserve resources.²³  On the other hand, initiatives that focus on staff communication may exclude families from these vital discussions.²⁴  With the Centers for Medicare and Medicaid Services mandate requiring family engagement in discharge planning and family discharge preparedness playing a significant role in ensuring timely and effective discharges, it is invaluable to include patients and families in these discussions, as well. PFCR is an excellent platform to discuss discharge criteria, given the involvement of both the team and families.

We saw an increased rate of discharge criteria communication during PFCR associated with the interventions of education on rationale for sharing this information, as well as the addition of discharge criteria communication and related scripting to our rounds checklist. It is possible this pairing of the “why” behind the initiative along with the “when” and “how” contributed to the initial increase in our outcome measure.

We also saw a subsequent increase in verbal communication of discharge criteria during PFCR after an increase in electronic documentation of criteria. Although hardwired EHR-based interventions can be impactful tools in quality improvement work to affect change, it can be challenging to develop those types of interventions to target verbal communication goals such as our primary outcome. We therefore sought to leverage EHR-based interventions to change practices closely tied to verbal communication. We incorporated clinical support tools in the form of discharge SmartForms to promote documentation of criteria, recognizing the impact that documentation has on the subsequent verbal communication of that content during PFCR.

We found a final higher rate of verbal communication of criteria on rounds than electronic communication in the EHR SmartForm. This may be related to differentiating starting points in these metrics because verbal communication occurred 32% of the time at baseline, whereas electronic SmartForm communication was a new process implemented as part of this initiative. It is also possible the improvement in verbal communication was associated with the cumulative interventions targeting this aim, including adjustments to the rounds checklist, which could have improved verbal communication but may not have impacted electronic communication in the EHR.

We did not see any special cause variation in LOS or readmission rates. One of the goals of improved communication regarding discharge criteria is to support proactive discharge planning, which may support a more efficient discharge process and limit postdischarge issues. On the other hand, there was concern that increased concrete documentation of discharge criteria might prolong LOS because of stricter interpretation of specific parameters. We therefore sought to assess for associations between this initiative and changes in LOS and readmissions with no special cause variation seen. This may be related to the wide range of factors impacting those metrics and the diverse patients included in this initiative. There would be value in further exploring specific elements of those metrics, including LOS broken down for specific diagnoses and the time from medical discharge readiness to the time of discharge. We noted a barrier to assessing the latter metric locally because there was no standardized way of documenting discharge criteria to objectively identify the time of medical readiness. An additional benefit of this project is increased documentation of discharge criteria to allow for more objective tracking of the time of medical readiness in future states.

There were various limitations to this study. It is a single site initiative and may not be generalizable to other institutions. Our metrics primarily focused on communication of criteria, and next steps include consideration of additional patient-centered metrics, including time from medical discharge readiness until discharge and number of postdischarge safety events. Although we have seen a subjective improvement in safety events with improved communication about discharge criteria, this improvement is challenging to accurately capture because of inconsistent documentation of those events. Our institution is currently implementing a new process to track safety events, and we hope to be able to further explore the impact of communication on safety events as next steps. Our primary outcome measure of inclusion of discharge criteria discussions on rounds was obtained through rounds observations from our QRIP and may not reflect teams’ baseline behavior when not being observed. The outcome measure was also reliant on observer availability and therefore had intermittent months of missing data because of observer unavailability and a relatively brief baseline timeline for this metric. Lower observations were feasible in March 2020 because of limited observer access early in the coronavirus disease 2019 pandemic and higher observations were feasible in July 2020 because of increased observer availability that month. In addition, we primarily focused on the inclusion of discharge discussion on rounds, but this does not account for the accuracy of the criteria discussed, which could impact the utility of those discussions. We also did not assess the team or family’s knowledge of criteria, which is an opportunity for next steps. Our metrics were obtained from a baseline rounds observation program and EHR flowsheet data pull developed locally. These resources may not be accessible at other programs limiting the generalizability of this work. Information was also not collected on race, ethnicity, or language preference so metrics could not be stratified by those domains.

Improvement methodology was associated with an increase in the frequency of discharge criteria discussions during PFCR without negatively impacting rounds length. Next steps include ongoing expansion of this initiative to subspeciality teams and ICUs across our institution and assessing the impact of discharge criteria discussions during PFCR on additional patient-centered metrics, including team and family knowledge of discharge criteria, LOS by diagnoses, and postdischarge safety events.

We thank Christopher Spahr, MD, Robyn Woolever, and Elizabeth Berkowitz for their contribution and oversight on this project.