Among 340 youths approached, 254 were eligible (62%), 86 were ineligible (48 parents not present to consent/participate, 10 not English or Spanish-speakers, 28 met other exclusion criteria), 210 of the 254 eligible youths (83%) completed baseline assessments, 29 were excluded after baseline (27 pilots, 2 determined ineligible after baseline, and 181 were enrolled in the RCT (eFIGURE 1
). Patients’ mean age was 14.7±2.0, 69% were female, and 67% ethnic/racial minorities (). The ED visit was due to a suicide attempt in 53% of youths, with the remainder due to suicidal ideation. Past-year suicide attempts were reported by 66% of youths, with 27% reporting multiple (≥ 2) past-year attempts. Mental health/functioning problems were common at baseline: 78% reported severe depression (CES-D ≥ 24); 53% screened positive for post-traumatic stress disorder;24
17% reported probable substance abuse;25
and 70%, 58%, and 73% of youths scored in the clinical range on the Child Behavior Checklist (CBCL) internalizing, externalizing, and total problems, respectively.20
FISP and CONTROL groups were similar at baseline, with the exception of marginally higher total problem scores in the FISP condition.
Baseline Background, Service Use, and Clinical Characteristics*
Most youths were hospitalized after ED evaluation/treatment (70%), with no significant between group differences. Based on retrospective assessments completed at follow-up, 40% (N=56/139) of youths met DISC-IV criteria for depressive disorders in the year before the emergency department visit (major depression, N=53; dysthymic disorder, N=3), with no significant group differences. Youths not meeting criteria for depressive disorders still had high levels of depressive symptoms (71%, or N=59/83, with a CES-D score≥ 24) and externalizing (54%, or N=42/78, in the clinical range) and internalizing (65%, or N=51/78, in the clinical range) problems.
Linkage to Outpatient Community Mental Health Treatment
presents the intervention effect on linkage to outpatient treatment, with and without multiple imputation for missing values. FISP patients were significantly more likely than controls to be linked to outpatient treatment (92% vs 76%; OR=6.2; 95% CI=1.8–21.3, p=.004). FISP patients also had significantly higher rates of psychotherapy (76% vs 49%; OR=4.0, 95% CI=1.9–8.5, p=.001), combined psychotherapy and medication vs. monotherapy (psychotherapy or medication alone) or no treatment (58% vs 37%; OR=3.3, 95% CI=1.5–7.0, p=.003), and significantly more outpatient treatment visits (FISP mean=5.3±7.0, median=3.0, range=0–36; CONTROL mean=3.1±5.5, median=.5).
Intervention Effects on Linkage to Outpatient Mental Health Treatment After ED/Hospital Discharge (N=160)*
Inpatient hospitalization was also associated with increased linkage (91% vs 67%, X2=14.69, df=1, p=.001). However, the intervention effect remained significant when hospitalization was included in the model (X2=8.37, df=1, p<.004), significant when the sample was restricted to hospitalized patients (97% FISP vs 86% CONTROL, X2=4.18, df=1, p<.05, n=114); within the smaller sample of non-hospitalized youths (n=45), the between-group difference was larger but marginal (82% FISP, 57% CONTROL).
Suicidality & Exploratory Outcomes
At follow-up, nine youths had attempted suicide (6%), four who received the FISP intervention (6%) and five who received enhanced usual emergency care (6%). There was one completed suicide. Suicidal ideation was observed among 18 youths (8 in FISP, 13%; 10 in the control group, 13%) on the DISC-IV. There were no statistically significant intervention effects on suicidality or other clinical/functioning outcomes (eTABLES 1–2
Results from random effects models revealed statistically significant improvements from baseline to follow-up: CES-D total score (t=−8.5, df=130, p<.0001), severe CES-D (OR=.24, 95% CI=.14–.41, p<.0001); CBCL total problems in clinical range (OR=.52, 95% CI=.30–.90, p=.02); parent CES-D (t=−2.15, df=96, p=.04); and CBQ (t=−10.12, df=128, p<.0001).
Did Outpatient Treatment Linkage Affect Clinical Outcomes?
Given the significant intervention effect on linkage to outpatient treatment and non-significant effects on clinical outcomes, we conducted exploratory instrumental variables (IV) analyses examining whether linkage was associated with improved clinical outcomes. These IV analyses estimate the effect of linkage while adjusting for selection effects which can lead to non-significant treatment-outcome relationships using traditional analyses.26–28
IV analysis relies on identifying an instrument that predicts the probability of treatment, but has no independent effect on outcomes. We used randomized intervention status as the instrument, linkage to any outpatient mental health treatment, and examined three youth outcomes: suicidal behavior (HASS Suicidal Behavior), severe depression (CES-D ≥ 24), and overall psychopathology (CBCL Total Problems, Clinical Range). For HASS score, we fit two step treatment-effects model using treatreg command in STATA version 11.1. For two binary outcomes, we fit bivariate probit regression model with biprobit command to jointly model clinical outcome and treatment linkage, explicitly taking into account the correlation. In all models, the effect of intervention on linkage was significant (p <.05), but no statistically significant benefits of treatment linkage on clinical/functioning outcomes (HASS, CES-D, Total Problems) emerged and treatment linkage was associated with more severe CES-D depression (eTABLE 3