A Quality Improvement Initiative to Decrease Behavioral Health Unit Readmission Rates

The United States Centers for Disease Control and Prevention recently published the 2021 results of the National Youth Risk Behavior Survey, concluding youths are experiencing a behavioral health crisis.¹ It is estimated that 1 in 6 children have a behavioral health condition. Occurring at the same time as the behavioral health crisis is a shortage of behavioral health specialists. Parents and caregivers often struggle to find access to therapists and/or psychiatrists to help manage their child’s behavioral health needs.² 

More than a quarter of all acute care hospital days among children and adolescents are due to a behavioral health diagnosis.³ Adding to this growing crisis is a deficiency of child and adolescent behavioral health inpatient beds. One factor contributing to the lack of available beds is readmission to pediatric BHUs.^4–6 Available data reveal that 1 in 4 youth are readmitted to a behavioral health hospital within 1 year of discharge with most readmissions occurring within 3 months.⁷ Studies report 30-day readmission rates between 7.8% to 13% in pediatric BHUs.^8,9 Higher readmission rates to pediatric BHUs contribute to the deficiency of behavioral health beds.^4,5 

Evidence suggests that characteristics such as public insurance, severity of illness, prior behavioral health hospitalizations, LOS, diagnoses, poor family functioning, and aftercare can affect readmission rates.^4,8–14 Additionally, studies have reported that discharge instructions crucial to follow-up care and parental care at home are frequently misunderstood by parents.^10,15 Using this information along with assessing reasons for readmission for our patients, we developed interventions to standardize processes and procedures with the goal of reducing the 7-day and 30-day readmission rates on our pediatric BHU by 20% by May of 2023.

This quality improvement (QI) project included patients admitted to our closed BHU from July 2020 through December 2023. The 24-bed BHU is housed within our 181-bed freestanding children’s hospital in the Midwest. Admission criteria for the BHU included suicidal and homicidal patients as well as those with active mania or psychosis. Exclusionary criteria to the unit included patients with nonverbal autism, patients unable to perform activities of daily living due to intellectual disabilities, and patients who had a documented history of violence against medical personnel. Demographic data around sex and insurance were collected. ^8,10 

Unit staff included 4 board-certified child and adolescent psychiatrists, 1 nursing clinical team lead, 4 nurses, 5 behavioral health therapists (master’s level counselors or social workers), 1 activity therapy team lead (licensed recreation therapist), 2 activity therapists (music or recreation therapist), 4 mental health technicians (associate’s degree in behavioral health–related field), 2 patient care assistants, 2 behavioral health specialists (bachelor’s degree in behavioral health–related field) to support discharge planning, and 4 administrators (1 psychiatrist, 2 nurses, and 1 licensed professional clinical counselor with supervision). In July 2022, a weekend-only child and adolescent psychiatrist was added to reduce the workload of the full-time child and adolescent psychiatrist covering weekends and weekdays.

This work was done in the context of our hospital’s intermediate QI course. Our key driver diagram (Figure 1) maps our relationships among interventions, drivers, and overall aims. Reasons for readmission were evaluated between May and June of 2021 to determine possible drivers and interventions (Figure 2). These were collected in real time through interviews with the attending psychiatrist, therapist, and medical record review and captured in a data collection form for later analysis. Baseline data were collected from July 2020 to July 2021. We used Standards for Quality Improvement Reporting Excellence 2.0 guidelines for reporting this QI project.

PDSA cycles were employed for each intervention, with multiple related cycles organized into ramps to accelerate learning. Most cycles were adapted. When no further learning was anticipated, cycles were adopted. No cycles were abandoned.

The first PDSA, completed in October of 2021, involved creating a readmission track. All patients who were readmitted would be considered for the readmission track. Patients with active psychosis or mania would be exempt from this track due to their inability to participate. The readmission track involved alternative individualized programming that considered the child and family’s needs. Programming was led by the behavioral health therapist, who actively collaborated with the patient’s care team throughout the admission. Individual programming could include more one-to-one sessions with the therapist, activity/music therapist, or journaling that would provide treatment in areas identified as triggers to the admission. Dialectical behavioral therapy (DBT) skills, such as distress tolerance, interpersonal skills/communication, and emotional regulation, which have been shown to decrease self-harm, suicide attempts, and depressive symptoms, were focused on.^16,17 Each day, a child would be able to earn 1 to 2 groups based on their progress in their individualized programming. This encouraged participation among patients. Individualized programming needs were reassessed daily, and modifications were made to the treatment plan as clinically indicated.

In November 2021, follow-up calls were offered to all patients admitted to the BHU regardless of readmission status. Calls were made within 24 to 48 hours of discharge. During these calls, the behavioral health specialist asked the following 4 questions:

1. How is your child doing?

2. Do you have any questions/concerns about your child’s safety plan?

3. Have you had any problems obtaining medications your child was prescribed?

4. Do you have any new concerns about getting to your child’s follow-up appointment(s)?

Based on the needs identified during the follow-up phone calls, the behavioral health specialist could link the guardian back to the treatment team to further discuss the safety plan/medications or could link the guardian to additional community resources to reduce barriers to accessing aftercare appointments.

Interpersonal conflict was frequently cited by patients and families as a contributing factor to readmission. To address this, conflict management interventions were standardized within the course of care for all patients admitted to the BHU in February of 2022. Groups targeting aspects of conflict management skills, including interpersonal communication skills, healthy boundaries, empathy, and anger management, were included. Groups pertaining to these topics were held for 1-hour increments 2 to 3 times daily. These skills were reviewed individually with patients and integrated into patient safety plans. Conflict management skills were also reviewed with guardians during the patient’s safety plan meeting.

The readmission checklist (Supplemental Figure 1) was developed to ensure that the family and patient were ready for discharge. This checklist reminded users of readmission track programming, referrals to wrap around services, parent education around conflict management, and discharge follow-up phone calls. This checklist was applied to all readmissions within 7 and 30 days but was only tracked for those patients who were readmitted within 30 days. This process was standardized as our fourth PDSA in November of 2022. The goal of the readmission checklist was to improve the reliability of our process by standardizing, simplifying, and ensuring staff followed the readmission interventions previously listed.

Two main outcome measures were captured: (1) 7-day and (2) 30-day readmission rates of patients readmitted per 100 hospitalizations. Because the BHU is a closed unit with specific indications for admission, these were considered unit-cause readmissions (in contrast with all-cause or condition-specific readmissions). Days were calculated from the index hospitalization discharge date. Data were collected monthly. The numerator included those patients who were readmitted. The denominator included all patients discharged for the month from the BHU. Rates were normalized per 100 hospitalizations. We monitored the BHU LOS as a balancing measure to ensure that the initiation of our readmission track did not prolong the amount of time needed to complete treatment. Adherence to a readmission checklist was tracked and served as a process measure for all patients admitted within 30 days. The numerator included all those patients that had the readmission checklist used. The denominator included all patients readmitted to the BHU in 30 days.

We used annotated attribute U-charts to examine changes in outcomes and associations with interventions. Standard criteria were used to identify special cause variation. Centerlines were established with 20 data points and shifted based upon special cause variation associated with practice changes.

Because this work was considered part of normal operations and internal QI, it was deemed exempt from full review by our hospital’s institutional review board. No external funding was received for this work.

There were 390 patients readmitted over the 42-month period. During this time, the demographic data for the entire BHU followed a similar trend to readmission data in becoming less female and more public paying (Table 1). These demographic data included 3-month periods at the beginning and end of the project time periods.

The primary outcome measure was 7-day and 30-day readmission rates. The rates of patients readmitted within 7 days decreased from a baseline mean of 5.54 to 2.83 readmissions per 100 hospitalizations (Figure 3). This equates to a decrease in readmissions of 49%. The rates of patients readmitted within 30 days decreased from a baseline mean of 11.52 to 7.38 readmissions per 100 hospitalizations (Figure 4), which equates to a rate of decrease of 36%. Interventions are annotated on each of these U-charts at their respective time points in the project. Special cause variation was seen on both charts (Figure 3, 4) after the incorporation of conflict management into groups. We observed that each of these improvements in readmission outcomes was sustained over 22 months, despite a medical surge in March 2023 and a change in leadership over the summer of 2023. Upper control limits also tightened over the same period.

Our primary balancing measure was LOS (Figure 5). The average LOS was noted to have a decrease from 5.58 to 5.09 days.

Successful implementation was sustained through the creation of standardized workflows and readmission checklists. Over the project period, the readmission checklist was used for 81 out of 83 patients readmitted within 30 days, or 97.5%. The 2 incidents in which we did not use the measure at all were secondary to going live in the electronic health record (EHR) and a discrepancy in our EHR readmission report.

Our rate of patient readmissions at 7 and 30 days decreased by 49% and 36%, respectively, surpassing our SMART aim of 20%. Interventions had similar effects on 7-day and 30-day readmission rates, which was likely due to these 2 groups experiencing similar care. Decreasing the readmission rate allows for quicker access to inpatient care with less disruption to their personal and academic life. In addition, this is a cost-saving measure for the medical system in which readmissions have been proven to be very expensive.^18,19 Additionally, by lowering the readmission rate for both 7 and 30 days, we were able to slightly decrease our LOS. This was felt to be secondary to the addition of a weekend psychiatrist position. Sustainment was continued due to the standardization of our process through the readmission checklist. Within the context of a pediatric behavioral health system struggling to keep up with the need for behavioral health admissions, this represents a significant improvement in process and outcome.

Very little research exists on the clinical outcomes of therapeutic programming of child and adolescent BHUs.^20–25 BHUs often focus on group therapy, which has mixed reviews on how beneficial it is vs individual therapy.^23,25,26 The readmission track attempted to focus on creating more individual programming and therapeutic sessions through cognitive behavioral therapy (CBT) treatment modalities such as journaling to ensure that patients were focusing on learning DBT and CBT skills. Through journaling, patients would learn techniques like problem-solving, emotional regulation, and distress tolerance, which have been shown to help with self-harm, suicidal behaviors, and depression.^{16,17,22,27–29} 

The importance of conflict management and communication among family and peers cannot be understated because this has been repeatedly linked to suicide attempts and completed suicides in multiple studies.^30–33 Successful communication through mood check-ins, anger management tools, and exploration of empathy was addressed by adding conflict management groups and education to family meetings. It has been shown that high emotional dysregulation and lack of empathy can create conflict in the home.³⁴ This ongoing family conflict can spill over to academic dysfunction and negative moods, which can lead to worsening mental health symptoms and readmissions.³⁵ The addition of conflict management and communication addressed these problems and allowed patients and families to have some basic strategies to lean on during times of distress.

Follow-up phone calls within 48 hours of discharge are theorized to improve successful transfer of care between inpatient to outpatient. In adults, it has been shown that phone calls can decrease rehospitalization and decrease suicidal behaviors.^36,37 A recent study found that 30% of youths were not making their aftercare appointments despite having appointments prearranged by the hospital.³⁸ When a loved one is in a BHU, often the entire family is experiencing a heightened level of stress. Families may not realize discharge instructions are given or have a lack of understanding of what is being said.¹⁰ The follow-up phone calls provide an opportunity to review the discharge instructions after stress levels have improved.

The severity of illness is often linked to the risk of readmission.^4,13 Illnesses such as psychosis, depression, or anxiety can lead to suicidal thoughts. In these cases, medication can be imperative for successful treatment. The inability to obtain medications has been linked to readmissions.^7,10 Phone calls helped ensure families obtained their medications and were not experiencing side effects. At least once, it was found that a patient’s medications were not obtained due to cost. In this situation, medication changes were offered to prevent decompensation. Also, there were multiple instances of guardians reporting medications not available at the pharmacy upon arrival. The discharge follow-up phone call served as a reminder to return to the pharmacy to obtain all prescribed medications.

The readmission checklist was developed to standardize and increase reliability in discharge planning. Studies have shown that postdischarge care, including case management and access to discharge medications, can affect readmission rates.¹³ Although every child admitted to the BHU received aftercare appointments, the readmission checklist attempted to increase the amount of aftercare a child could receive. This was accomplished through referrals to wrap around programs or day treatment programs, which have been shown to decrease readmission rates.³⁸ 

This work has limitations. First, data collection was completed manually. All attempts were made to ensure data accuracy, but there remains a possibility of human error. Second, this was a project at a single center, which may limit external validity. Additionally, our process was highly dependent on having adequately trained staff and interventions. We attempted to improve reliability with a standardized checklist that likely led to sustained improvement in our system. Additionally, we did not measure the process of follow-up phone calls, which would have provided more robust evidence of this intervention’s efficacy. These interventions may need to be adjusted and further investigated individually to have similar effects in more complex health systems.

This work demonstrates that through rigorous QI interventions, changes can be made by institutions to affect behavioral health readmission rates. Implementation of consistent protocols and development of tools that address known factors for readmission were associated with reduced 7-day and 30-day readmission rates while decreasing the average LOS.

Thank you to John Duby, MD, and Shehzad Saeed, MD, for their review of the manuscript. Thank you to Tara Lee, who helped with data analytics, and Beth Williams for assistance in developing some of our graphs. Thank you also to Dayton Children’s Hospital, which supported an internal quality improvement course, where this work began.

BHU

behavioral health unit

CBT

cognitive behavioral therapy

LOS

length of stay

PDSA

Plan-Do-Study-Act

QI

quality improvement