FormalPara Clinician’s Capsule

What is known about the topic?

Validated pediatric-specific tools to guide care of patients with mental health emergencies are underutilized.

What did this study ask?

Can a care bundle with pediatric-specific mental health tools and a referral from triage option be reliably implemented?

What did this study find?

Care bundle elements were implemented with high fidelity, and eligible low-risk patients were assessed in a partnered clinic outside the emergency department.

Why does this study matter to clinicians?

It presents an innovative model for pediatric mental health emergencies and the use of quality improvement methods to support implementation.

Introduction

Children often access the emergency department (ED) for mental health concerns [1, 2]. During the visit, risks are identified, concerns assessed, and disposition decided. Despite the best efforts of healthcare providers, care is often inefficient, inaccessible, and not patient-centered [3]. While some EDs use pediatric-specific mental health tools to guide care, this practice is not widespread [1, 4,5,6]. Local feedback from patients and parents in Alberta suggested the need to refine mental health care to improve care efficiency, timeliness, and access to mental health providers in the ED and following discharge [7]. We designed a prospective implementation study of a mental health care bundle to improve suicide risk screening at triage, to standardized needs assessment, to provide timely follow-up visits, and to offer an option for low-risk patients to choose a scheduled appointment over remaining in the ED. For the bundle to improve patient care, ED flow, and outcomes, all bundle elements needed to be reliably implemented. Therefore, embedded in the study, was a quality improvement (QI) initiative to test and fully implement each bundle element.

The project aims were that within 6 months of bundle introduction, 80% of eligible patients would receive suicide risk screening, 80% would receive a needs assessment score, and 100% of eligible discharged patients would be offered a follow-up appointment within 96 h.

Methods

Context

The initiative took place from February to October 2021 in the Stollery Children’s Hospital, a tertiary care pediatric ED in Edmonton, Alberta, Canada with ~ 50,000 annual pediatric visits including ~ 2300 annual visits for mental health. Prior to the initiative, mental health visits began with an assessment by a triage nurse who would classify urgency into one of five acuity levels using the Canadian Triage and Acuity Scale (CTAS) score; however, score determination lacked standardized, validated questions to identify suicide risk [8]. Following triage, children were assessed by an ED mental health nurse who would provide a summary to the ED physician, who would decide on the need for a psychiatrist consultation and determine disposition. During this process, a standardized tool was not used to guide assessments and narrative summaries did not consistently assist physician decision-making. Most discharge instructions required families to organize their child’s follow-up care.

Intervention

The care bundle (Fig. 1) was developed with patient and parent partners. It includes three elements that have a strong evidence base and reflect recommendations from the American Academy of Pediatrics (AAP) and American College of Emergency Physicians [1]. Elements were introduced sequentially in 2-week intervals to allow the project team to give each change appropriate attention and give staff time to adapt. Impact on patient outcomes will be evaluated once patient-reported follow-up data have been collected (study registration # NCT04292379) [9].

Fig. 1
figure 1

Bundle overview

Full size image

Suicide risk assessment at triage

Nurses utilized the Ask Suicide-Screening Questions at triage with children ≥ 8 years presenting with one of the following chief complaints [10, 11]: anxiety, depression/suicidal, deliberate self-harm, hallucinations/delusions, homicidal behavior, ingestion, pediatric disruptive behavior, situational crisis, or violent behavior. The tool is highly sensitive for detecting immediate suicide risk in this age group [12,13,14]. If a child answered ‘yes’ to at least one of four questions about suicidality, a fifth question was asked about acuity.

Children were admitted to the ED if they answered ‘yes’ to any question or were unable/refused to answer. Families of patients who answered ‘no’ to the first four questions (negative screen), and did not have other medical or safety concerns, were met by an ED mental health nurse to discuss whether they would prefer to schedule an urgent appointment in a mental health clinic in lieu of waiting to be seen in the ED. This approach ensured that low-acuity patients would have an expedited path to assessment and avoid a lengthy ED wait.

Focused mental health assessments

We introduced the HEADS-ED tool for use by ED mental health nurses who assessed patients prior to ED physicians. The tool facilitated focused assessments with patients and guided nurse/physician decision-making for psychiatrist consultation and discharge planning [2]. The tool has excellent inter-rater agreement [16] and was originally developed to aid physician assessment [15, 16].

The tool has seven domains for assessment: Home, Education, Activities/Peers, Drug/Alcohol, Suicidality, Emotions and Behavior, Discharge Resources. Nurses assigned a score to each domain to indicate an impairment/action level: 0 (no/minimal functional impairment; no action needed), 1 (moderate functional impairment; needs action but not immediate), or 2 (severe functional impairment; needs immediate action) [17]. Total scores ≥ 8 (reflecting moderate-to-severe impairments across several domains) or a suicidality domain score of 2 (at-risk for immediate harm) were used to identify children potentially in need of psychiatry consultation. ED physicians could consult psychiatry regardless of score if deemed clinically important. Children with lower scores were deemed suitable for ED discharge with an urgent appointment offered to those without a mental health provider.

Urgent appointments

We introduced an urgent appointment within 96 h that was booked prior to the family leaving the ED. For low-risk children whose families chose an appointment over an ED visit, the appointment occurred at a mental health clinic located within the children’s hospital. For children discharged from the ED with an appointment, it occurred at the hospital-based clinic or one of two community-based clinics. Partnerships with community-based clinics were created to provide appointments and additional funding was obtained to support the hospital-based clinic.

The appointment was modeled after the Choice and Partnership Approach to care, a shared decision-making framework that focuses on prioritizing the needs and preferences of children and their caregivers [18]. Per the framework, mental health providers generated a case formulation with the child/caregiver and concluded the appointment with a joint decision on what resources/services best matched needs and goals [19].

Implementing and supporting practices

We created clinical workflows for each bundle element. Healthcare providers received element-specific training by a nurse educator and/or QI nurse. The QI nurse also used change management strategies, based on the Prosci ADKAR® model [20], including providing just-in-time, one-on-one support, and group communications (emails, posters, huddles) (Fig. 2). We developed aims (Table 1) and key driver diagrams for each bundle element [21]. We conducted a value stream mapping exercise to verify the bundle’s potential to increase efficiency, despite addition of new tools, by supporting more focused assessments and standardizing flow. Feedback was collected from healthcare providers to inform necessary modifications.

Fig. 2
figure 2

Implementation Activities and PDSA Cycles

Full size image
Table 1 Overview of the aims and measures used in the evaluation
Full size table

Plan-do-study-act cycles

We tested bundle implementation using iterative plan-do-study-act cycles. Initial tests of change were small (e.g. testing a change with one provider, one patient, or during one ED shift) to learn quickly if a component could be used as planned or required adjustments. Tests of change became larger in scale as learning accrued from previous cycles [22].

Measures and data collection

Outcome measures are identified in Table 1 alongside each aim. For measures related to Ask Suicide-Screening Questions and HEADS-ED, an aim of 80% of all patients with an eligible chief complaint was selected to allow for cases in which the use of the tools would not be possible (e.g. refusal, intoxication, developmental age < 8 years). Clinical data were collected from the patient’s electronic medical records.

Analysis

We used weekly run charts to determine if change was resulting in improvement. Run charts were developed according to established standards [22], and chosen over statistical process control P-charts because we preferred a simple tool that uses a median as the center line (an advantage for mitigating the effect of outliers), could detect a specific special cause (shift), and did not require special software so it could be maintained by ED staff. Since none of the bundle elements were used prior to implementation, we did not analyze pre-implementation data. The baseline median represented the initial use of the tools upon introduction and was established using the first ten data points. Non-random signals of change were identified by a shift (≥ 6 consecutive points above/below the median) [22]. Other outcome data were analyzed using descriptive statistics.

Ethical considerations

The plan-do-act cycles undertaken for this initiative were deemed QI per the University of Alberta’s Guidelines for Differentiating among Research, Program Evaluation and Quality Improvement [24]. REB approval was obtained for the implementation study.

Results

Suicide risk assessments at triage

Tests of change with individual providers began February 1, 2021 with full use of the Ask Suicide-Screening Questions beginning April 2021. A shift occurred in September 2021 with tool use occurring with 93% of eligible patients (Fig. 3). From April to October 2021, 80.3% of 1251 total mental health presentations were screened with tool. Among patients who screened negative, 20 (61.0%) of 36 families were offered an urgent appointment, with 95.0% of appointments scheduled within 96 h. Two families (9.0%) offered appointments declined and remained in the ED for an assessment. The opportunity to offer an urgent appointment was missed with 14 families (39.0%). The median time to an appointment was 11.3 h. (IQR 13.4–23.3 h.). All families attended their appointment, and no patients were redirected to the ED with safety concerns.

Fig. 3
figure 3

The percentage of patients who were screened with the Ask Suicide-Screening Questions at triage

Full size image

Focused mental health assessments

Small tests of change began in February 2021 with full HEADS-ED use beginning April 2021. A shift was noted in August 2021 with scores generated for 87% of patients (Fig. 4). From April to October 2021, 80.8% of 795 patients had scores generated on the HEADS-ED with 39.9% of 642 having a HEADS-ED score ≥ 8 (n = 152) or suicidality score of 2 (n = 104). Of these, 52.3% of the 256 received a psychiatric consultation and 47.7% received a comprehensive assessment from the nurse. Of the 386 patients with a HEADS-ED score ≤ 7 and suicidality score of 0 or 1, 21.0% received a psychiatric consultation.

Fig. 4
figure 4

The percentage of patients during the initiative who had a HEADS-ED score generated after their mental health assessment

Full size image

Urgent appointments were offered to all (100%) 159 patients at ED discharge who did not already have a mental health provider. Of these appointments, 89.3% were booked within 96 h of the ED visit. Reasons for appointments outside of the target timeframe were family choice (n = 3) and holiday between ED visit and next appointment (n = 9); reasons were unclear for five appointments. Of the 159 appointments, 8.2% were not attended (no-show), 6.9% were canceled, and 1.9% were redirected back to the ED during their appointment due to newly emerged safety concerns (suicidal ideation with plan, unable to safety plan, agitation).

Urgent appointments

Tests of change began April 2021. Case formulations were completed for 98.5% of 132 eligible appointments, with 2 not completed due to lack of child/caregiver engagement. At appointment conclusion, 76 families expressed an interest in ongoing treatment, and all were referred.

Discussion

Interpretation

In this initiative, we used a novel approach of embedding QI methods in an implementation study to support the introduction of changes. Aims were met in the implementation of two ED-based tools, the Ask Suicide-Screening Questions and HEADS-ED. We did not achieve our aim to offer urgent appointments to all eligible low-risk patients at triage, but met our aim of offering appointments to all eligible patients at ED discharge, with almost all occurring within the 96-h target.

Previous studies

To our knowledge there are no published ED clinical pathways that combine standardized risk screening at triage with a preference-based choice for care if immediate risks are not identified. The Ask Suicide-Screening Questions has been successfully used in other EDs [25,26,27], but our use to introduce care choices to families is unique. Offering select adult medical patients a clinic visit instead of remaining in the ED has been described as safe and effective [28, 29]. The use of HEADS-ED to guide assessments and decision-making has also been studied in other EDs [16, 30, 31]. We introduced it to also support interprofessional communication between ED physicians, mental health nurses, and psychiatrists. Finally, our provision of urgent appointments allowed families to be directly connected to mental health services. Current practices in discharge of pediatric mental health patients have not been published. A study of academic EDs in the United States showed that 72% reported making appointments for patients to support discharge planning, but it was not reported how often this approach was used [32]. Modeling our urgent appointment after the Choice and Partnership Approach [18] was also unique. To date, appointments using a ‘choice approach’ have only occurred within mental health care settings [33]. Our use of this approach after the ED visit removed the burden on families to arrange care and ensured access to services.

Strengths and limitations

A key strength of this initiative was the novel use of QI to augment an implementation study. Meaningful assessment of implementation outcomes requires that the intervention be successfully adopted. Typically, implementation study interventions are fully designed prior to introduction, whereas QI begins with small tests of change with an intervention modified as learning accrues [34]. While our bundle elements were predetermined, each element was introduced with small tests of change. We believe implementing with plan-do-study-act cycles increased acceptability to staff and ensured sufficient adoption, while analyzing run charts ensured progress toward aims.

Our initiative was limited by logistical challenges. Not all eligible patients were offered a choice for care following triage. One potential reason for this may have been the infrequency of negative Ask Suicide-Screening Questions screens. As only 7% of patients (1–2 per week) screened negative, some triage nurses may have forgotten to use the option. In addition, the COVID-19 pandemic presented staffing challenges with a higher than usual proportion of casual staff present during implementation. As casual staff may be less aware of current initiatives, this may have lowered use of bundle elements. Finally, we cannot comment on bundle impact on patient-reported outcomes as this project was not designed to measure bundle effectiveness.

Clinical implications

During bundle design, reducing long ED waits was a parent partner priority. It was important to our team that urgent appointments were offered to low-risk families from triage as an alternative to remaining in the ED so families could choose what they believed to be in their child’s best interest. While this option was not utilized as frequently as anticipated, there were no adverse events associated with this aspect of the bundle, and we believe that its availability enhances family-centered care and may help alleviate ED crowding.

Research and QI implications

Bundle impact on patient-reported outcomes will be addressed in the ongoing prospective implementation study [9]. There are very few published reports on patient/family experiences with ED mental health visits and outcomes following discharge; this will be an area of immense importance in planning future improvement initiatives.

Using QI methods was an effective strategy to ensure reliable implementation of the bundle. Frequent interactions with staff, follow-up on pathway use, and ongoing education supported early success. Additionally, the use of run charts throughout implementation allowed our team to adjust strategies when initial approaches were not resulting in reaching our implementation aims.

Conclusions

Using a multi-disciplinary approach supported by QI methods, we ensured that a mental health bundle to improve screening, risk assessment, and follow-up was reliably delivered to patients presenting to the ED with mental health concerns. Pediatric-specific mental health assessment tools can be used to assess risk and offer an alternative to full ED evaluation to appropriate low-risk patients.