Improving Shared Situation Awareness for High-risk Therapies in Hospitalized Children

Our institution is an academic quaternary care freestanding children’s hospital with >28 000 admissions and 31 000 surgical encounters annually. The 48-bed unit included in this study primarily admits neurology, neurosurgery, trauma surgery, and hospital medicine patients. The interdisciplinary improvement team included a physician from hospital medicine, a pediatric resident physician, unit nurse leadership, the lead pharmacist for medication safety, 2 nurse leaders with extensive experience in the use of HRTs, a respiratory therapist, a hospital administrator, and the parent of a technology-dependent child with medical complexity who experiences frequent hospitalizations. This study was systems improvement, which did not necessitate institutional review board review or exemption.

Our team used the Model for Improvement to develop a current state process map (Fig 1A) and construct a modified failure modes and effects analysis.^14,15  In addition to the aforementioned safety event reviews and the team’s expert consensus, we reviewed all safety reports filed in the institutional event reporting system for this unit over a 6-month period to determine the medications and medical devices most implicated in adverse events. Team members observed huddles on both the study unit and other units with extensive HRT experience (eg, bone marrow transplantation) to determine how the “H” flag impacted staffing, patient care, and SA. Furthermore, several qualitative survey questions were asked of nurse safety leaders to obtain perspectives on the current state of HRT SA.

Using these learnings, the team developed an ideal state process map and HRT algorithm (Fig 1B) that made several enhancements to the original process. We identified key stakeholders on the patient’s care team who may identify an HRT, including the patient and family, clinicians, pharmacist, all bedside staff, and charge nurses. We expanded the definition of an HRT to include medications and devices that were unfamiliar (eg, chemotherapy on a nononcology unit) or familiar with an inherent high risk (eg, external ventricular drains). Importantly, removal of a medication or device that places a patient at high risk was also added (eg, removal of an antiepileptic medication). The new algorithm also designated that the charge nurse was responsible for entering the “H” flag and descriptor into the EHR. A specific HRT huddle was added to the process, during which the HRT identifier and the charge nurse would discuss concerns with relevant stakeholders (eg, respiratory therapist, pharmacist) and determine the need for additional resources. The team then discussed the patient at preexisting unit safety huddles until the HRT was removed, at which time the “H” flag was resolved to mitigate alert fatigue.

Beyond unit-based huddles, our institution-wide SA system involved safety huddles 3 times daily. Participants included charge nurses from each unit, a physician safety officer, a nurse safety leader, and patient flow coordinators.¹¹  The current state process involved discussing all HRT patients with “H” flags by listing their room numbers aloud. However, no additional dialogue was facilitated, frequently leading to no actionable discussion and an ineffective use of huddle time. Therefore, specific criteria were added as to when patients receiving HRTs should be discussed at this huddle, including (1) additional staffing needs, (2) resources required from another unit more familiar with the therapy or facilitating transfer of the patient to another unit as indicated, and (3) elevation in unit microsystem stress due to the quantity and quality of all SA concerns (eg, HRTs, risk for clinical deterioration), which may necessitate mobilization of resources to that unit to provide safe care. Our team then developed a key driver diagram (Fig 2) to facilitate unit- and institution-wide SA for patients receiving HRTs and ensure optimal workflows and patient safety.¹⁴ 

The original HRT process on this unit involved a list of 5 HRTs that contained only the name of the HRT, without any information on the risks of the therapy, mitigation plans, or escalation plans. The improvement team developed a comprehensive table of HRTs for use by bedside staff (Fig 3, Supplemental Fig 6). Meetings with unit nurse leaders and discussion with the primary medical services leadership led to a consensus table of 13 designated HRTs, with the option for any other therapy meeting algorithm criteria to be included as indicated. The table included the name and a graphic of the therapy, the primary risks, the unit risk prevention and mitigation plans, and the medical service that should be contacted with questions or concerns.

After finalizing the new HRT algorithm and reference table, unit staff and the medical services were educated by using a variety of modalities, including both live and virtual staff meetings, the development of an interactive educational video, and newsletter communication. To facilitate sustaining this knowledge, the new algorithm and reference table were added to monthly unit staff education modules and the orientation education for new staff and posted in laminated form in unit huddle rooms, break rooms, and leadership offices.

Because the new algorithm designated the charge nurse as the stakeholder responsible for “H” flag and descriptor entry in the EHR (Supplemental Fig 7), this group of nurse leaders was targeted for specific education. Discussion of the new HRT algorithm was integrated into the agenda for regularly scheduled charge nurse meetings. A reminder sign to enter the “H” flag and descriptor was placed at the charge nurse workstation, along with a statistical process control chart to provide weekly feedback on their success in improving this measure.

To facilitate unit shared SA and patient and family engagement, we developed bedside signs for patients receiving HRTs (Fig 4). The sign contained information on the name and potential risks of the therapy, unit risk mitigation plans, clinical signs and symptoms that may be indicative of complications, and escalation plans for questions and concerns. The improvement team worked with the family relations department to develop signage at approximately a seventh-grade literacy level. Signs were developed for each of the 13 designated HRTs (Fig 4), and a blank template was created for other HRTs to foster collaboration between the bedside staff, nurse leaders, the medical team, and the patient and family to create a visual representation of their discussion of the HRT. As part of the HRT huddle in the new algorithm, a sign was placed inside the room on the preexisting whiteboard containing information about the patient’s hospital stay.

For each unit safety huddle, charge nurses prepared a sheet containing information on census, patient flow, SA concerns, and any other relevant operational issues. This sheet served multiple functions, including as a (1) reference for the current charge nurse and unit leadership, (2) source of information escalated to institutional safety huddles, and (3) handoff between charge nurses. We collaborated with nurse leadership to modify this sheet to include the HRT patients flagged as “H” in the EHR, including the appropriate descriptor. This encouraged consistency between the EHR and huddle documentation, and charge nurses used this sheet to verify that EHR information was correct before and during unit safety huddles. We also created a new section of the sheet to address the failure mode that the flag was being used for non‒HRT‒related concerns, such as behavioral escalation and limited code status. This section was referred to as “heightened awareness” and allowed for unit-wide SA regarding these patients without using the HRT flag inappropriately.

Retrospective data for the baseline period were collected via an automated report of the SA concern flow sheet generated from the EHR, which provided quantitative data regarding use of the “H” flag as well as qualitative data on the accompanying descriptor or lack thereof. During the active study period, the improvement team leader received automated daily reports on use of the “H” flag on the study unit. All patients labeled “H” were included in the analysis to allow for the inclusion of other HRTs not listed on the reference table. Survey data to inform the process maps and modified failure modes and effects analysis were collected as described above. During the active study period, our team collected and reviewed qualitative data from key stakeholders via semistructured interviews to assess shared SA and inform the development and adaptation of interventions. Patients and families, bedside nurses, and first-call clinicians for patients labeled “H” in the EHR were asked 3 questions about the HRT: (1) What is the HRT? (2) What are the risks and mitigation plans for the HRT? and (3) To which team will you escalate questions or concerns? After implementation of the new algorithm, the study team leader performed manual chart review of admitted patients over a 2-week period to ensure that patients receiving an HRT designated on the reference table were not inadvertently excluded from “H” EHR designation and thus excluded from the new process and the denominator of our outcome metric.

As a proxy for unit- and institution-wide shared SA, the primary outcome measure was the percentage of “H” flags in the EHR that contained the correct descriptor for the HRT. This was tracked weekly via a prospective time series analysis by using a statistical process control p-chart.¹⁶  Established rules for Shewhart control charts were used to determine if observed changes were due to special cause variation.¹⁷  Shared SA was initially intended as a quantitative outcome measure in addition to a qualitative measure to inform interventions; however, the coronavirus disease 2019 pandemic halted in-person data collection.

While plan-do-study-act cycles were implemented to enhance shared SA, the percentage of patients receiving an HRT and labeled “H” in the EHR with a correct descriptor for the therapy increased from a mean of 11% to 96%, with special cause variation noted on a statistical process control p-chart (Fig 5). A manual review of all patients admitted to the study unit over a 2-week period confirmed that no patients receiving an HRT were inadvertently excluded from “H” EHR designation.

Qualitative survey data obtained during the early study period informed development of the new HRT algorithm (Fig 1B), our key drivers, and interventions (Fig 2). Both unit and institutional safety leaders agreed that the definition of an HRT needed to be individualized at the unit level because of the varying nature of patient populations cared for on each unit and thus the varying level of familiarity the staff and families had with different therapies. Gaps in the current process included the frequent absence of a descriptor for the “H” designation in the EHR and thus lack of understanding surrounding what was high risk and why. The quantity and nature of patients receiving HRTs on the unit significantly impacted the nursing workflow, and respondents identified several opportunities to improve the SA system both at the micro- and macrosystem levels. Responders identified multiple issues with the current state of institutional safety huddle discussions, including the inefficiency and ineffectiveness of simply listing the names of HRT patient rooms. Staff also identified the connections between HRTs and patients at risk for clinical deterioration and how the identification of one class of SA concern may lower the threshold to seek out and appropriately identify other types of concerns.

Semistructured interviews to assess shared SA between the patients and families, nursing staff, and the primary medical team revealed several important findings, including that the primary medical team was often not aware of which therapies were considered HRTs on the unit, although there was frequent alignment between the medical teams and nursing staff regarding the risks and mitigation plans. Secondly, patients and families were often unaware of the therapies designated as high risk by the unit and, in some cases, had increased familiarity with an HRT as compared with unit staff because of their use of the therapy at home. In these cases, patients and families could serve as an important resource and partner with unit staff to provide optimal and safe care during the hospitalization.

An innovative HRT algorithm with an emphasis on clear definitions and criteria, role designation, improved documentation, enhanced resource accessibility, and key stakeholder engagement was associated with an 85% improvement in accurate EHR designation, and thus institution-wide SA, for patients receiving HRTs. To our knowledge, we are the first to report on a comprehensive, interdisciplinary approach to risk mitigation for patients receiving HRTs in a hospital system. Multiple novel tools were introduced, including a comprehensive reference table, bedside signs, and an optimized unit safety huddle documentation sheet. Although our early plan-do-study-act cycles were associated with a nearly 55% improvement, an additional ∼40% improvement was associated with the unit safety huddle sheet update (Fig 5). This was a critically important step given that this ensured consistency between verbal, written, and electronic communication and also addressed the failure mode of inappropriate use of the “H” flag in the EHR. Our work builds off previous improvement science that revealed that improved SA was associated with a decrease in preventable patient harm in children’s hospitals.¹¹ 

We measured EHR “H” flag designation and description as an indicator of micro- and macrosystem SA for patients receiving HRTs. The direct measurement of unit key stakeholder SA was initially intended as an outcome measure, but the coronavirus disease 2019 pandemic halted in-person data collection. However, learnings from the prepandemic interviews were critical to guiding our interventions to foster shared SA among stakeholders, including patients and families. The lack of knowledge of unit-specific HRTs by the medical teams further emphasized the importance of the EHR “H” flag descriptor because this would alert clinicians to the therapy considered an HRT on the unit and thus encourage further planning and discussion between the medical team and bedside nurses. Furthermore, our team recognized the critical role of patients and caregivers at the bedside as frequent first-line identifiers of signs and symptoms of complications and therefore developed patient- and family-facing bedside signs for active engagement in risk mitigation. We found that families sometimes had more inherent comfort with a therapy than unit staff because of their management of this therapy in the home setting (eg, tracheostomy, insulin). Therefore, family awareness of the HRT process facilitated a shared mental model between families and staff and fostered enhanced engagement in their child’s care while hospitalized.

Although the association of improved SA with a decrease in preventable harm is known, the ideal outcome metric for this work would be an improvement in patient-level outcomes, including a decrease in adverse events related to HRTs. A variety of methods have historically been used to detect adverse drug events and medical device complications in both adults and children, including chart review, billing codes, safety event reporting systems, and trigger tools.^4,6,18  However, these detection methods are fraught with challenges, including that they are often of low sensitivity and specificity, time intensive, and limited to retrospective study. Although trigger tools have proven to be efficient and effective for adverse event detection, with well-described use for adverse drug events, they have yet to be applied specifically to medical device complications in children.^18–22  Future work will be focused on developing improved methods of detection for use in measurement for improvement efforts.

A comprehensive algorithm to enhance SA for patients receiving HRTs on a high-acuity inpatient unit was associated with a significant improvement in EHR designation and description of these patients, facilitating institution-wide visibility of, and SA for, therapy-related risk status. This new algorithm and framework will facilitate sustainability and ongoing improvement via several avenues, including the addition and removal of HRTs as indicated and the automation of successful interventions. Future work will be focused on developing and improving tools for the direct measurement of SA and patient-level HRT-related outcomes, as well as developing an automated method to alert staff to the presence of HRTs in the EHR.