Effect of a Motivational Interviewing–Based Intervention on Initiation of Mental Health Treatment and Mental Health After an Emergency Department Visit Among Suicidal Adolescents

This randomized clinical trial examines whether a motivational interviewing–based intervention increases linkage of adolescents to outpatient mental health services and reduces depression symptoms and suicidal ideation in adolescents seeking emergency care for non–mental health–related concerns.


Introduction
Suicide is the second leading cause of death in adolescents aged 12 to 17 years in the United States, and the rate of adolescent suicide increased 87% between 2007 and 2016. 1 The annual prevalence of adolescent suicide attempts, the most robust risk factor of youth suicide, 2 is 8.6%, and the rate of attempts that require emergency medical care has increased. 3 With suicide prevention a national priority, 4,5 The Joint Commission now recommends that hospitals screen all medical patients for suicide risk. 6 The emergency department (ED) is a promising venue for screening, brief intervention, and referral to treatment because it is the portal into mental health services for most suicidal patients. 7 Many patients at risk for suicide go unrecognized in the ED, 8 thereby precluding referrals to evidence-based outpatient services. 9 To our knowledge, only 2 pilot studies 10,11 have evaluated the efficacy of interventions for adolescents seeking emergency care for non-mental health-related concerns who screen positive for suicide risk. In 1 study, 10 adolescents who received the Teen Options for Change intervention had a significant reduction in depression symptoms compared with an enhanced usual care (EUC) group during the 2 months after their ED visit, but no significant group differences on suicidal ideation and mental health service use were found. A previous pilot randomized clinical trial 12 assessed the effectiveness of a brief, ED-based mental health service engagement intervention to increase linkage to outpatient mental health services in teens presenting with medical chief concerns and no recent history of mental health issues who screened positive for suicide risk factors using the Columbia Suicide Severity Rating Scale (C-SSRS). In that pilot study, 11 24 participants were randomly assigned to the intervention (n = 11; short motivational interview, barrier reduction, outpatient appointment established, and reminders before scheduled appointment) or standard referral (n = 13; telephone number for a mental health care practitioner). Results suggested that adolescents receiving the intervention were more likely than those in the standard referral group to attend a mental health care appointment during the follow-up period.
In this 2-site randomized clinical trial, we tested the efficacy of the Suicidal Teens Accessing Treatment After an ED Visit (STAT-ED), a brief mental health treatment engagement intervention, for adolescents seeking ED treatment for nonpsychiatric concerns but identified via systematic screening as being at risk for suicide. STAT-ED is rooted in motivational interviewing (MI), a nonconfrontational style of communication to help patients and their families resolve ambivalence about engaging in a health-related behavior. 13 A meta-analysis by Cushing et al 14 found that MI had a small but significant association with improvement in a wide variety of adolescent health behaviors, but suicidal ideation was not one of the conditions studied.
Building on theoretical models of treatment engagement, 15,16 previous empirical findings in treatment engagement studies of adolescent suicide attempters, [17][18][19] and our own pilot work, 11 STAT-ED targeted family engagement, problem solving, assistance with referral, and limited case We hypothesized that adolescents receiving STAT-ED would have a higher rate of initiating mental health treatment and attend more mental health treatment sessions in the 2 months after the ED visit compared with adolescents receiving EUC. We also hypothesized that STAT-ED would be superior to EUC in reducing suicidal ideation and depression symptoms at 2 and 6 months. Finally, in exploratory analyses, we examined intervention effects on 6-month mental health treatment initiation and attendance, suicide attempts during the follow-up period, and potential demographic moderators of intervention effects on 2-month treatment outcomes.

Trial Design
This 2-site, single-blind randomized clinical trial used a parallel design to compare STAT-ED and EUC with a 1:1 allocation ratio (Figure 1). Parents provided written informed consent for youth to participate in the study, and youth provided their assent in writing. All data were deidentified. All study procedures were approved by the institutional review boards of the Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio (prime), and the Nationwide Children's Hospital, Columbus, Ohio (reliance). Adolescent and parent participants were compensated for their participation ($25 each at baseline and $20 for each completed follow-up assessment). This study followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline. 20 The trial protocol can be found in Supplement 1. Study recruitment occurred between April 2013 and July 2015.
Intention-to-treat analyses were performed from September 2018 to October 2019.

Participants
Participants were a convenience sample recruited at times when both a research coordinator and a study-trained and nonstudy ED psychiatric social worker were on duty (7 days per week).
Adolescents presenting to the ED for nonpsychiatric concerns at Cincinnati Children's Hospital Medical Center or Nationwide Children's Hospital were recruited. Patients were eligible if they met  Participants were excluded if they presented to the ED with a chief concern of suicidal behavior or a primary or secondary psychiatric concern or altered mental status attributable to illness or medication, lacked telephone access, were unable to understand the study process, or were unable to speak or read English adequately to participate in study procedures.

Interventions
If randomized to the EUC intervention, the adolescent received a mental health evaluation and referral that followed standard-of-care guidelines for emergency behavioral health assessments in the participating EDs. The EUC social workers did not receive any new training or feedback. If it was safe for the adolescent to be discharged home and clinically indicated, the EUC social worker facilitated a referral to a mental health care practitioner during the visit or the next day if after hours.
If randomized to the STAT-ED intervention, the adolescent and parent received brief MI to target mental health care-seeking behavior, barrier reduction discussion, and referral (eMethods 1 in The case management protocol included a telephone call within 2 days of discharge, a call the day before the scheduled appointment, and 1 or 2 follow-up telephone calls after ED discharge. The STAT-ED social worker talked with the parent, obtained a release of information, spoke to the mental service practitioner who the parent had chosen (if the parent gave written permission to do so), and was available to assist the family if problems arose. The rationale for contacting families is to reinforce the general principles of suicide risk reduction and review the actionable plan for initiating and sustaining mental health treatment recommended during the ED visit. If the initial plan for engaging in treatment proved difficult to follow, the practitioner would work with the family to develop a new one to enhance initiation and adherence with mental health treatment.
Both EUC and STAT-ED interactions in the ED were audio recorded. We randomly selected 10% of both intervention and control audiotapes, blocking information on site, study practitioner, and phase of the study (early vs late). An MI expert, who did not provide periodic feedback to study practitioners and was blinded to randomization status, rated the audiotapes using Motivational Interviewing Treatment Integrity, version 3.1. 23 The STAT-ED practitioners had a mean (SD) global proficiency score of 3.56 (1.15), whereas EUC practitioners had a mean (SD) score of 2.38 (0.70) (t 15 = 2.51; P = .02). A global score of 3.5 is considered the beginning proficiency level of the 5-point scale. 23 There was a significant site effect, indicating that the Nationwide Children's Hospital had higher global proficiency ratings across study conditions.
All adolescents in the study reported suicidal ideation or behaviors, and each received a risk assessment in the ED. After a risk assessment, patient disposition was determined in consultation with a psychiatrist, ED attending physician, or nurse practitioner. Although the plan was to admit to inpatient psychiatric care any adolescents thought to be at too high of a risk to send home, no study patients had conditions severe enough to be admitted during the study.

Outcome Measures
The prespecified primary mental health engagement outcomes included treatment initiation, defined as attendance of at least 1 mental health visit in the 2 months after discharge, and the number of mental health treatment sessions attended in the 2 months after the ED visit. Both primary mental health treatment outcome measures were assessed by an independent evaluator and verified with the mental health practitioner. Independent evaluators were master's degree-or PhD-level assessors outside the ED and not part of the screening, triage, or intervention process.
Mental health treatment initiation was also assessed by the parent via telephone calls performed by interviewers (L.S.) blind to treatment allocation. The advantage of parent report is that it can cover mental health treatment use across a range of practitioners, whereas for agency report we were restricted to practitioners for whom we obtained release of information forms signed by the parents for agencies where the child was most likely to receive services. The Service Assessment for Children and Adolescents (SACA) 24  Three exploratory outcomes that were not included in the statistical analysis plan of the original trial protocol (Supplement 1) were examined: (1) 6-month mental health treatment initiation and attendance, (2) suicide attempts during the 6-month follow-up period, and (3) time to mental health treatment initiation (eMethods 2 in Supplement 2). The exploratory outcome of suicide attempt was defined as self-injury with at least some stated or inferred intent to die and was assessed using the C-SSRS. 12 The C-SSRS has demonstrated good convergent and divergent validity in both adult and adolescent suicide attempters. 12

Screening Measure
The ASQ 21

Sample Size
The study was designed to have power of 80% (2-sided α = .05). By using a χ 2 test of proportion with continuity correction, power estimates for detecting group differences in treatment initiation rates

Method for Randomization
Participants were randomized into the EUC or STAT-ED group via computer-generated blocks of 10.
A data analyst separate from the study statistician generated a random sequence of numbers printed and placed in numbered, sealed, radio-opaque envelopes. On consent, the research social worker broke the seal of the next envelope and determined the adolescent's group allocation, thereby ensuring allocation concealment. Research staff performing follow-up telephone surveys were blind to participant group allocation.

Statistical Analysis
Our primary analyses compared STAT-ED and EUC using intention-to-treat analyses, which requires that outcomes are analyzed by randomization without regard to participant adherence to study protocol. 31,32 Groups were compared on baseline characteristics and 2-month and 6-month outcomes by using the χ 2 or Fisher exact test for categorical variables and independent group t tests and Mann-Whitney statistic, as appropriate. Logistic regression models were used for the binary outcomes associated with initiation in mental health services. Multinomial logistic regression was used to compare groups on treatment attendance categories (0, 1, or Ն2 appointments at 2 months; 0, 1-11, or Ն12 appointments at 6 months). Unadjusted and adjusted odds ratios (ORs) and 95% CIs were calculated with logistic regression; adjusted analyses controlled for site, age at enrollment, sex, and race/ethnicity. Repeated-measures analyses of variance were used to compare groups and group × time interactions on the primary clinical outcomes (suicidal ideation and depression symptoms). The exploratory analysis of time to mental health treatment initiation was computed using the Kaplan-Meier method and compared with the log-rank test. Our approach to assessing moderators of outcome followed the approach outlined by Kraemer and colleagues. 33 Table 2). The overall rate and number of mental health care appointments for youth in the STAT-ED group also were significantly higher at 6 months than for youth in the EUC  Table 3). Among nonwhite or Hispanic youth, no group differences in mental health treatment initiation were observed.
However, white, non-Hispanic youth in the STAT-ED group had higher rates of mental health treatment initiation at 2 months compared with youth in the EUC group. There were no reported adverse events associated with STAT-ED or EUC.

Discussion
Contrary to expectations, no statistically significant group differences were observed on mental health treatment initiation or attendance at 2 months. Similarly, the intervention groups did not differ on trajectories of depression symptoms and suicidal ideation over time. In exploratory analyses, STAT-ED emerged as more efficacious than EUC for linking patients to mental health treatment at 6 months. Youth in the STAT-ED intervention also attended more mental health treatment sessions by 6 months. In exploratory moderator analyses, white, non-Hispanic participants had higher rates of 2-month mental health treatment initiation in the STAT-ED condition relative to EUC. However, there was no intervention effect among nonwhite or Hispanic participants.  The log-rank test indicated a significant difference favoring the STAT-ED intervention compared with EUC (χ 2 1 = 4.65; P = .03). As described in the text, 37 participants were lost to follow-up; therefore, it was not possible to assess their engagement with mental health services as assessed by agency report. These participants were included in the analysis but censored immediately after baseline.
Overall, the sample reflected the predominantly disadvantaged inner-city characteristics of the population served in the catchment areas of the 2 sites. Anecdotally, several parents reported that their child was unable to initiate counseling within 2 months because of logistic and scheduling barriers, especially long wait lists for initial appointments. Visual inspection of the survival distributions reveals clear between-group separation in mental health treatment initiation beginning approximately 90 days after ED discharge and persisting throughout the follow-up period. Although no overall group differences emerged at 2 months, tests of moderation revealed significant STAT-ED intervention effects only among white, non-Hispanic youth. Future research would benefit from consideration of perspectives from African American and Hispanic families to guide adaptations of the intervention and maximize mental health treatment engagement in a culturally effective way.
The STAT-ED intervention used psychiatric social workers trained to deliver the intervention under routine practice conditions, which is likely to increase sustainability over time. This approach differs from prior ED-based studies 10,17,18,35

Limitations
This study has several important limitations. First, it was conducted in 2 large, academic children's hospitals, and findings may not generalize across all racial/ethnic groups, geographic locations, and hospitals that lack strong academic and research infrastructure. Second, by design, there was no overlap of study and nonstudy psychiatric social workers; thus, the likelihood of contamination of the intervention into EUC was small, but the possibility of contamination effects cannot be completely excluded. Third, the study assessed the use, but not the quality, of mental health services. Fourth, there was limited MI proficiency in the STAT-ED condition. Additional studies are needed to confirm whether the lack of an intervention effect on primary outcomes was associated with limited MI proficiency. Fifth, we did not systematically record the interval between the ED encounter and initial scheduled mental health care appointment. If the STAT-ED social workers attempted to facilitate quicker referrals relative to EUC social workers, then differences in MI delivery may not explain our intervention differences at 6 months.

Conclusions
This randomized clinical trial shows that a brief MI intervention did not have a significant benefit on mental health treatment initiation or attendance at 2 months and mental health outcomes at 2 and 6 months. However, in exploratory analyses, STAT-ED outperformed EUC at 6 months in linking youth screening positive for suicide risk to initial and ongoing mental health treatment. Future research should explore barriers to mental health treatment for adolescents at risk for suicidal behavior.