Screening for Suicide Risk Among Youths With a Psychotic Disorder in a Pediatric Emergency Department

In response to the Joint Commission’s recommendation to incorporate screening as routine care to increase the early detection of suicide risk, the Johns Hopkins pediatric ED introduced suicide-risk screening with the ASQ. Beginning in 2013, a multidisciplinary committee of key stakeholders was formed to implement suicide-risk screening as routine care in the pediatric ED. In March 2013, ASQ screening was implemented among all patients ages 8 years and older who were treated in the ED for a psychiatric or behavioral chief complaint. The Johns Hopkins Hospital pediatric leadership decided to implement universal screening in the pediatric ED and all age-relevant inpatient units as of January 1, 2017. Thereafter, all medical patients ages 10 years and older and all patients ages 8 years and older who were treated for a psychiatric or behavioral chief complaint were screened by nursing staff with the ASQ as routine care during triage. Faculty and staff from the Johns Hopkins Bloomberg School of Public Health provided training to nursing staff on ASQ implementation and have continually studied the effectiveness of ASQ screening in detecting suicide risk among youths as a standard quality improvement effort.

This was a retrospective cohort study of a consecutive case series of patients in the Johns Hopkins Hospital pediatric ED from March 2013 through December 2018. Patients under age 8 and over 18 years were excluded from the analysis. This report focuses on patients diagnosed as having a psychotic disorder, including schizophrenia, schizoaffective disorder, major depressive disorder with psychotic features, or bipolar disorder with psychotic features, as recorded in the electronic health record (EHR). Only the initial visit during the study period for each patient was used for the analyses. A medical record review was approved by the Johns Hopkins School of Medicine Institutional Review Board.

The ASQ is a four-item suicide-risk screening instrument that can be rapidly administered to patients and scored by nursing staff and does not require extensive training to use (6). The ASQ includes four questions with binary (yes/no) response options: In the past few weeks, have you wished you were dead? In the past few weeks, have you felt that you or your family would be better off if you were dead? In the past week, have you been having thoughts about killing yourself? Have you ever tried to kill yourself? A positive response to any item is considered a positive screen. Horowitz and colleagues (6) reported a sensitivity of 96.9% (95% confidence interval [CI]=91.3–99.4), specificity of 87.6% (95% CI=84.0–90.5), and negative predictive values of 99.7% (95% CI=98.2–99.9) for medical and surgical patients and 96.9% (95% CI=89.3–99.6) for psychiatric patients.

Treatment as usual at the time of study initiation assessed suicide risk on the basis of the presence or absence of a chief complaint related to suicide, including self-disclosed suicidal ideation or attempts and involuntary visits due to suicidal ideation or attempts. Because of inconsistent documentation within the spectrum of suicidal ideation and behavior, this determination is coded as a binary variable denoting the presence versus absence of a chief complaint of suicidal ideation, attempts, or related behavior.

Patient demographic characteristics were drawn from the EHR. Because of low frequencies of some racial-ethnic categories, race-ethnicity was indicated as black, white, or other. Gender was coded as a binary variable (male or female) because a nonbinary option was not yet included in the EHR. Patient disposition was likewise drawn from the EHR and coded as discharged or admitted/transferred (transfer versus admission was primarily determined by the availability of inpatient hospital beds rather than by clinical need).

Youths who screened positive versus negative on the ASQ were compared on demographic characteristics, chief complaint, and admissions disposition by using bivariate analyses (t tests for age, chi-square for all other variables). Exploratory analyses were conducted to better understand the group that screened positive on the ASQ but did not present with a suicide-related chief complaint. Statistical significance was tested on the basis of two-tailed α=0.05 for all analyses.

A total of 87 patients were diagnosed as having a psychotic disorder among our cohort of 15,007 patients (0.6%) (8). The sample was predominantly non-Latino black, slightly more than half were female, and about 20% of patients had a suicide-related chief complaint (Table 1). Nearly half of the sample (N=42, 48%) screened positive on the ASQ, with females more likely than males to screen positive (odds ratio [OR]=27.1, 95% CI=3.4–216.2), but no differences were found by race or age. Nearly all positive screens included the endorsement of at least one of the items assessing current suicidal ideation or behaviors; only three (7%) positive screens were due to positive responses to the item inquiring about lifetime suicide attempts alone. Youths with a suicide-related chief complaint were more likely to screen positive (OR=4.0, 95% CI=1.6–10.1). Further, the ASQ missed only one youth with a suicide-related chief complaint. However, treatment as usual on the basis of chief complaints missed many youths; 37% (N=26) of patients without a suicide-related chief complaint nonetheless screened positive on the ASQ. Conversely, among those who screened positive, only 38% (N=16) had come into the ED with a chief complaint related to suicidal ideation or behavior. The ASQ therefore increased detection of suicide risk almost threefold (2.63 times) relative to the chief complaint alone. Notably, many of those who screened positive on the ASQ were discharged (50%, N=21) rather than admitted or transferred, although the discharge rate was slightly lower (43%, N=17) when excluding those who screened positive solely on the basis of the lifetime suicide attempts item. We compared respondents who screened positive on the ASQ on the basis of whether or not they had a chief complaint of suicidal thoughts or behaviors. Among those who screened positive, patients who presented with suicidal ideation or behavior did not vary with patients who did not present with those indications in terms of age (mean±SD=15.4± 2.3 versus 14.7± 1.5, t=1.0, df=1, N=42, p=0.320), gender (χ²=1.8, df=1, N=42, p=0.177), or race (χ²=0.5, df=2, N=42, p=0.775).

In the context of increasing child and adolescent suicide rates and in the absence of systematic suicide-risk screening protocols, this study demonstrates both the feasibility and validity of implementing universal suicide-risk screening as routine care in a pediatric ED. Limited empirical evidence has supported the capacity of suicide-risk screening tools to improve the identification of suicide risk in naturalistic settings, particularly among high-risk groups, such as youths in the early stages of psychosis. The main findings of this study are consistent with the implications of the Joint Commission’s sentinel event alert: nearly half of the youths with a psychotic disorder in this study screened positive for being at risk for suicide, and a majority of these individuals had suicide risk that was undetected through the treatment-as-usual approach of relying on the patient’s chief complaint. Furthermore, 50% (N=21) of the children and adolescents in this study who screened positive on the ASQ (including 44% [N=17] of those reporting acute rather than lifetime suicidal thoughts or behavior) were discharged from the ED. In addition, this high-risk group had a significantly higher rate of positive screens when compared with a larger universal screening sample, which had a positive screening rate of 8%, demonstrating the need for screening tools with particularly vulnerable groups (8). ASQ screening is imperative for the identification of youths at risk for suicide, but it needs to be paired with feasible follow-up intervention services.

It is notable that a majority of female patients with psychotic disorders screened positive (62%, N=32) compared with male patients (29%, N=10). This difference may relate to gender differences in symptom profiles in early psychosis, with some evidence for greater affective symptoms among females and greater negative symptoms among males (9). The incidence of death by suicide is higher in males than females in general and among those with schizophreniapi and bipolar disorder, but findings on the prevalence of suicidal ideation and behavior in early psychosis have been inconsistent across studies (10). The finding of higher rates of positive screening on the ASQ among females is timely given a recent report showing the narrowing of the historically large gap in youth suicide rates between males and females (11).

The main potential limitation of this study was that psychotic disorders were not directly coded in the EHR and therefore were identified through chart review; as such, they may have been underidentified in these data. Complicating this issue was the low prevalence of psychotic disorders, particularly in pediatric samples. Despite an initial sample of more than 15,000 youths, we were able to identify only 87 patients with confirmed psychosis diagnoses. Similarly, despite having longitudinal data available for the broader sample, return visits and deaths by suicide were infrequent among this subsample, inhibiting sufficient power for longitudinal analyses. Another limitation was the lack of data on the specific diagnoses, which were not recorded in a sufficiently consistent manner across individuals and over the 5 years of this study to confidently separate the sample into subtypes of psychotic disorder. Suicidal behavior was similarly recorded dichotomously and without consistent differentiation between ideation and attempts. Despite these potential limitations, this study supports the value of standardized ED screening to identify youths at risk for suicide.

Now that we have doubled detection of our patients at risk for suicide by using the ASQ suicide-risk screening tool, we will turn our attention to enhancing follow-up protocols and procedures. On a small scale, we have been testing the value of sending patients brief and supportive text messages after discharge. Unfortunately, patients with psychosis were excluded because of concerns about these messages contributing to paranoia. To better understand the barriers that suicidal individuals face in being connected to mental health care after an ED visit, we are conducting periodic follow-up assessments of a subset of our patients up to 1 year after their discharge from the ED. The longitudinal follow-up includes a battery of assessments on factors that can either facilitate or inhibit an individual’s mental health outcomes. We intend to use the data gathered from the longitudinal follow-up to inform practices that will improve appropriate referrals and long-term mental health outcomes. Clinical and quality improvement staff have identified the use of family navigators as a next step, based on our current analyses, for improving outcomes among those identified as being at risk for suicide. With the appropriate training and supervision, family navigators could assist in conducting follow-up contacts, care coordination, and brief interventions, such as safety planning and lethal means counseling. These patient care enhancements could make a measurable difference in quality of life and health outcomes. Our ongoing efforts will also aim to study whether the ASQ leads to better connection to appropriate community-based mental health care.

The authors thank the patients and nursing staff of the pediatric emergency department at Johns Hopkins Medical Center.