Utilization Patterns at a Specialized Children's Comprehensive Psychiatric Emergency Program

Psychiatric emergency visits for children and adolescents have increased dramatically over the past 20 years (1–5), rising 20.9% from 2006 to 2011 alone (5). Despite this escalation in demand, emergency departments (EDs) lack the structural and clinical resources to adequately meet these patients’ needs. Children are commonly seen in medical EDs that are not able to keep youths safe, both because the physical facilities are inadequate and because child psychiatric specialists are not available around the clock. In other hospitals, youths are managed in psychiatric emergency programs designed for adult patients. Like the medical EDs, these programs generally lack staff with pediatric expertise and can be frightening to children.

This staffing deficit represents an important obstacle to the appropriate evaluation and treatment of youths in psychiatric crisis. Bridge and colleagues (6) reported that almost half of youths who presented to EDs for self-injury did not receive any mental health assessment. Self-injury, with its close association to suicidal behavior, is exactly the type of presentation that should warrant a specialist evaluation. Yet across the country, EDs lack access to these specialists. Baraff and others (7) reported that EDs at only 25% of academic medical centers and 8% of community hospitals had child psychiatrists available to evaluate youths.

Pediatric psychiatric visits place an extra burden on EDs because they contribute to longer wait times and lengths of stays and high rates of boarding in the ED (8,9). Much of this burden is driven by the dearth of child psychiatric specialists in most EDs, in combination with lack of access to immediate outpatient mental health services. There are significant consequences for patients as well. Many youths do not receive appropriate mental health evaluation and are discharged without adequate treatment planning (6). Another 30% to 40% are held for inpatient admission, although many might have been managed in outpatient treatment had immediate access to care been available (10–13).

Studies attempting to characterize psychiatric admissions of children and adolescents have often focused on demographic characteristics, in part because of the availability of these characteristics in the large administrative data sets commonly used to study ED decision making. Older age (12,14–16), white ethnicity (12,14), and residing in an urban area (14) have been associated with psychiatric admission, whereas prior outpatient treatment has been inconsistently associated with disposition (15,17–19). Whether these factors truly influence need for hospitalization or merely reflect differential access to care is unclear, a reflection of the limitations of retrospective studies of claims data.

Mood disorders, such as depression (12,20,21), and psychotic illnesses (12,22,23) are consistently associated with hospitalization. However, in some studies (21,24), externalizing behavior disorders and aggression are associated with admission, particularly among younger children (25), but in other studies, they are associated with discharge (20,26). Substance use disorders, developmental disabilities, and intellectual disorders have also been inconsistently associated with hospitalization (16,17,26–30). Suicidality (including suicidal ideation and behavior) and high-risk behaviors, such as nonsuicidal self-injury and aggression, especially in the context of multiple comorbid diagnoses, are usually associated with admission (14,21,30,31). Hospitalizations for suicidality and nonsuicidal self-injury have increased in recent years, making it particularly important to understand how EDs evaluate children with these behaviors. The increase in admissions for suicidality and nonsuicidal self-injury may be related to increased screening, economic factors affecting the family’s or child’s stress level and access to care, or overall escalation in the prevalence and severity of pediatric mental illness, of which suicidality is a marker (5,14).

Although hospitalizations for suicidality and self-injury are increasing, the likelihood that an individual patient presenting to the ED with suicidal behavior is admitted for psychiatric treatment still depends largely on resource factors. These include type of insurance (12,14,26), availability of psychiatric inpatient beds in the area, and geographical location (14,26). The odds of hospitalization for suicidal behavior depend significantly on whether a child presents to a children’s hospital or to an ED staffed by pediatricians or psychiatrically trained professionals rather than to a general medical ED without pediatrics or psychiatry available (14,26).^.It is concerning that psychiatric admission is dependent on staffing rather than on clinical need, given the dearth of child psychiatric specialists in most EDs. To make matters worse, three-quarters of EDs lack pediatricians or pediatric emergency medicine physicians (32). This is particularly worrisome given the significant increases in the number of youths presenting with suicidal and self-injurious behavior (5) and the increased risk of completed suicide after a suicide attempt (33–35).

In 2011, Bellevue Hospital Center opened the Children’s Comprehensive Psychiatric Emergency Program (CCPEP) to provide a safe and appropriately staffed program for youths in psychiatric crisis. The CCPEP is an adaption of a model of emergency psychiatry care for adults that includes extended observation beds for brief stabilization, specialized staff to provide comprehensive psychiatric and psychosocial evaluation and treatment planning, and interim crisis clinic services to provide acute outpatient treatment and a bridge to longer-term services. This model has been quite successful in increasing the quality of care provided to adults in psychiatric crisis (36).

The Bellevue CCPEP is an adaptation of this model that is specialized to serve children ages zero to 18. The program is staffed 24/7 by child psychiatrists and child psychiatric nurses; social workers, psychologists, and caseworkers provide daily coverage. The CCPEP has six brief-stabilization beds, where patients can be held for up to 72 hours, and an interim crisis clinic that provides immediate outpatient follow-up irrespective of insurance or catchment area. Each patient and family presenting to the CCPEP receives a comprehensive evaluation by a child and adolescent psychiatrist. Patients can be admitted to the six-bed brief-stabilization unit within CCPEP or to an inpatient psychiatric unit elsewhere or discharged with connection to outpatient care, often provided in the crisis clinic. The program serves children from throughout the New York City metropolitan area. A small proportion of the youths served in CCPEP are patients of the hospital’s outpatient programs, but most are not, referred instead from school, other clinics, private providers, or family members or brought to the program by emergency services or police.

Given that much of the extant literature on disposition after an ED visit is based on administrative databases with minimal clinical content, we sought to better understand the clinical characteristics of youths accessing ED care for psychiatric complaints and the drivers of hospitalization among these patients through a retrospective analysis of clinical and sociodemographic factors of youths presenting to this specialized ED program. This study provided data on the disposition of child psychiatric emergencies among youths who were evaluated in an appropriately staffed emergency program, where each patient is evaluated by a child and adolescent psychiatrist and a range of appropriate and immediate dispositions is available. These data allowed us to explore dispositional outcomes of child psychiatric emergencies independent of the common confounds of inadequate child psychiatric staffing, poor access to follow-up care, and pressure to discharge children in psychiatric crisis from medical EDs.

The objective of this study was to identify demographic and clinical characteristics that were associated with the disposition of children experiencing a psychiatric emergency into one of three categories: evaluate and release, brief stabilization in the CCPEP, or psychiatric admission.

Methods

We conducted a retrospective chart review of patients seen in the CCPEP during its first year of operation (January 4, 2011, to January 3, 2012). Data collection took place between February 1, 2013, and May 15, 2015. The Institutional Review Boards of New York University Medical Center and Bellevue Hospital Center approved this study. There were 1,180 visits by 890 discrete patients ages two to 18. The analytic sample consisted of data from the first visit for the 885 patients for whom disposition status was known. Disposition status was identified by using ED logs and electronic health records.

Charts were reviewed by a research assistant blinded to study hypotheses. Data were reviewed for accuracy by the study team coordinator, who reviewed and double coded every tenth chart. All clinical variables were also independently reviewed by the coordinator. Any data discrepancy was followed up with direct chart review of the variable by the research assistant and study team coordinator to ensure that the final database accurately reflected the information contained in the chart. Variables reviewed included demographic factors, urgency of visit, reason for visit, recent mental health service utilization, current and past clinical characteristics, and DSM-IV diagnosis. Data about child exposure to trauma and adversity were recorded; these will be published separately. The outcome measure was disposition following the ED visit (evaluate and release, brief stabilization in the CCPEP, or psychiatric admission).

Categorical measures were summarized by using counts and percentages, whereas continuous measures were described by using means and standard deviations. Bivariate analyses comparing child and clinical characteristics by disposition category were assessed by using the chi-square test and the Wilcoxon rank sum test. Post hoc pairwise comparisons were performed if the overall p value was <.01. Measures of theoretical interest (suicidal ideation, self-injurious behaviors, and externalized aggression as the reason for visit) as well as measures with a bivariate association of p<.10 were further assessed by using multinomial logistic regression analyses. Main effects models were fitted, and exposures that were statistically significant at p<.01 were retained in the final model. The results are summarized by using adjusted odds ratios (AORs) and their 95% confidence intervals.

Results

Disposition Following ED Evaluation

Following ED evaluation, visits by 59% of children were categorized as evaluate and release, 13% as brief stabilization, and 28% as psychiatric admission.

Demographic Factors

Half of patients were female (49%), with a mean±SD age of 12.9±3.4years (Table 1). Patients were racially and ethnically diverse (39% black and 33% Hispanic) and predominantly poor (56%) and covered by public insurance (74%). Bivariate associations between child demographic characteristics and disposition category are shown in Table 1. No demographic characteristic except age was significantly associated with disposition; children who were admitted were significantly younger than those who were held for brief stabilization or evaluated and released, and children who were held for brief stabilization were significantly older than children who were evaluated and released.

Diagnostic Factors

One-third of children were diagnosed as having a depressive disorder (Table 2); children with this diagnosis or a diagnosis of psychotic disorder (3.5%) were more likely than children without these respective disorders to be admitted or held for brief stabilization than to be evaluated and released (p<.001 for both comparisons). Nearly 40% of children were diagnosed as having one or more disruptive behavior disorders, although these were not associated with disposition. Three hundred and sixty-two (41%) children had multiple diagnoses, and having more diagnoses was associated with brief stabilization and admission (p<.001).

Suicidality, Self-Injurious Behavior, and Externalized Aggression

Visits by most (55%) children were classified as emergent (immediate danger to self or others), 31% as urgent (significant exacerbation of symptoms over the past five days), and 14% as neither emergent nor urgent (for example, the child needed a medication refill) (Table 3). Over one-third (36%) of patients reported current suicidal ideation, 14% had attempted suicide, and 21% had current self-injurious behavior. These patients were more likely than patients without these behaviors to be admitted or held for brief stabilization than to be evaluated and released (p<.01 for all comparisons). Patients with both current suicidal ideation and prior suicidal behavior were particularly more likely to be hospitalized (admission or brief stabilization, p<.01 for all comparisons). Patients whose visit was due to externalized aggression were less likely to be admitted or held for brief stabilization than to be evaluated and released (p<.01). Children who presented both with suicidal and self-injurious behavior and with externalized aggression were more likely to be admitted than children whose visit was due to externalized aggression only (p<.05). There was a high degree of comorbidity between aggressive behaviors and suicidality, with 30% of youths presenting for externalized aggression also reporting current suicidal ideation.

Prior Service Utilization Factors

Previous psychiatric hospitalization and recent use of outpatient mental health services were common (Table 3), and both were associated with greater likelihood of admission or brief stabilization (p<.01 for all comparisons).

Multinomial Logistic Regression Results

The multinomial logistic regression model showed that multiple clinical characteristics and diagnoses were risk factors for being held for brief stabilization or admitted compared with being evaluated and released: diagnosis of psychotic disorder and depressive disorder or mood disorder NOS; current suicidal ideation; recent outpatient treatment; and previous psychiatric hospitalization (Table 4). Patients diagnosed as having a psychotic disorder had significantly higher odds of being held for brief stabilization (AOR=8.8) or admitted (AOR=13.8) than of being evaluated and released. Patients with depressive disorder or mood disorder NOS had twofold increased odds of brief stabilization and threefold increased odds of being admitted versus being evaluated and released. Children with current suicidal ideation had 2.2-fold increased odds of brief stabilization or admission versus evaluation and release compared with children without suicidal ideation. Children whose visit was related to aggression (suicidal or self-injurious behavior, externalized aggression, or both) had fivefold increased odds of being held for brief stabilization and 2.4-fold increased odds of being admitted versus being evaluated and released, compared with children whose visits were unrelated to these behaviors. Patients who recently received outpatient mental health services had increased odds of being held for brief stabilization (AOR=1.7) or admitted (AOR=1.8) versus being evaluated and released compared with children who had not received outpatient mental health services recently. Children with a previous psychiatric hospitalization had significantly higher odds of brief stabilization (AOR=2.4) or admission (AOR=2.05) versus evaluation and release compared with children without a previous psychiatric hospitalization.

Two factors reduced the likelihood of admission versus brief stabilization or evaluation and release (Table 4). Children diagnosed as having adjustment disorder versus other diagnoses had significantly lower odds of admission compared with brief stabilization (AOR=.33) or evaluation and release (AOR=.25). Age was also protective: for each one-year increase in age, the odds of admission versus brief stabilization decreased by a factor of .83, and the odds of admission versus evaluation and release decreased by a factor of .88.

Discussion

The detailed clinical and demographic data collected in this study contribute to the existing literature on child psychiatric emergencies in several ways. First, they reflect care provided in a setting where every child, regardless of insurance status and socioeconomic status, receives an immediate, thorough psychiatric and psychosocial evaluation by child mental health specialists. These data provide a detailed clinical and diagnostic picture missing from much of the existing literature. Second, these evaluations occurred in a setting in which appropriate and immediate dispositions are available, including brief stabilization, inpatient admission, and discharge with urgent outpatient follow-up. These two factors reduce the challenges inherent in interpreting much of the data on disposition of child psychiatric emergencies in the existing literature: confounds caused by lack of appropriate child psychiatric specialist evaluation of patients and pressures to discharge children managed in medical EDs.

The vast majority (86%) of the patients presented in crisis (emergent or urgent severity of visit). Yet almost two-thirds (59%) of the patients could be discharged after evaluation, given the access to immediate outpatient follow-up. Another 13% of patients could be stabilized in the brief stabilization unit in under 72 hours. This suggests that brief stabilization with active treatment is an effective alternative to inpatient admission for a subset of patients and should be used more broadly for both quality improvement and cost containment.

Unlike earlier reports, in this study none of the demographic variables, with the exception of age, predicted disposition. Our finding that younger patients were more likely to be admitted differs from earlier reports. There are at least two possible explanations for this finding. One, access to child mental health specialists increases the identification of serious mental disorders among younger children, resulting in higher rates of admission. Two, the immediate access to both brief stabilization and inpatient admission for younger children increases the utilization of these services. This is an important area for further study, given the dramatic increase (151%) in inpatient admissions for suicide, suicidal ideation, and self-injury among children ages ten to 14 between 2006 and 2011 (5).

Increased odds of admission for youths presenting with depressive and psychotic disorders support the most consistent findings in the literature on the impact of diagnosis on disposition. Disruptive behavior disorders did not increase the odds of admission, and adjustment disorders clearly were associated with a decreased likelihood of admission.

Prior studies have reported significantly lower rates (24%−56%) (6,22,31,37) of hospitalizaton for suicidal ideation or behavior than the rates reported here. In this study, 178 of 317 (56%) of patients with current suicidal ideation and 69 out of 85 (81%) patients with current suicidal ideation and prior suicidal behavior were hospitalized (admission or brief stabilization; data not shown). For youths presenting after an episode of self-injurious behavior, 95 out of 186 (51%) were hospitalized (admission or brief stabilization; data not shown), also significantly higher than previously reported (29%−33%) (14,31).

Self-injurious behavior is a known risk factor for suicide, and the months immediately after a suicide attempt are known to carry a markedly elevated risk of reattempt. Yet in many EDs, youths with self-injurious or suicidal behavior are discharged home without psychiatric evaluation. Professional guidelines indicate that youths presenting to the ED with suicidal ideation or deliberate self-harm should undergo an extended mental health evaluation before discharge from the ED (38). In this study, the presence in the ED of child psychiatrists to perform such evaluation likely resulted in greater recognition of suicide risk and higher rates of hospitalization, compared with hospitals without such staffing. Although suicidal and self-injurious behaviors are typically associated with depression and other internalizing disorders, suicidal behavior often occurs in the context of externalizing disorders, such as conduct disorder, and suicidal or self-injurious behavior often co-occurs with aggression toward others. In this study, 30% of youths presenting for aggression toward others also reported current suicidal ideation, and over half of these youths required hospitalization (admission or brief stabilization; data not shown).

A significant proportion of patients utilizing the program had prior mental health treatment, and these patients were more likely to be held for brief stabilization or admitted. Therefore, dedicated psychiatric emergency programs for youths represent an important safety net for children and adolescents living with serious or chronic mental disorders.

This study had several limitations. The design was retrospective; diagnosis and documentation of clinical factors was clinician dependent; and service availability factors, such as bed availability or insurance barriers, were not available. These data are cross-sectional, so causality cannot be determined. This urban, largely low-income patient population may not generalize to all EDs, although studies with national databases also show that a significant proportion of patients accessing EDs for psychiatric care are nonwhite, reside in urban areas, and receive public insurance.

Conclusions

Across the country, EDs serve increasing numbers of children and adolescents in psychiatric crisis and have a crucial role to play in both supporting youths with chronic, serious mental illness and appropriately addressing the needs of those experiencing the first presentation of mental illness. Suicide is now the second leading cause of death among adolescents, and suicidal ideation and behavior affects increasing numbers of younger adolescents (5). In this study of an emergency setting staffed with child psychiatric specialists, clinical factors, especially suicidality, rather than demographic factors, predicted psychiatric admission. Admission rates for youths with suicidality were significantly higher in this study than previously reported, suggesting that the presence of child psychiatric specialists allowed greater ascertainment of suicide risk (and thus hospitalization to mitigate that risk), including among youths for whom suicidal ideation or behavior was not the presenting complaint.

Access to EDs that provide appropriate psychiatric evaluation and the appropriate range of services to manage youths in psychiatric crisis is a necessary improvement to our child and adolescent mental health service system. Use of telepsychiatry consultation and nonphysician specialized child mental health clinicians may allow community hospitals without access to 24/7 on-site child psychiatry to provide such evaluation and ED-based treatment.