Hooven and colleagues (2010) studied the long-term effectiveness of Care, Assess, Respond, Empower (CARE), using a quasi-experimental design with three intervention/treatment groups and a comparison group. The study sample was drawn from 20 high schools in urban and suburban areas of the Pacific Northwest. A total of 2,000 youths were screened with the suicide risk screen (SRS) instrument. Students who screened in as at risk of suicide were retained for the study. Students who were not deemed currently at risk were excluded from the study. The initial screening identified 615 youths at risk for suicide. Overall, the sample was 61 percent female and 67 percent white, with a mean age of 15.9 years.
On completion of the SRS instrument, participants were randomly assigned to receive one of four conditions: Parents CARE (P–CARE), Counselors CARE (C–CARE), a combination of the youth and parent interventions (P&C–CARE), or the minimal-intervention comparison condition (MI). Youths assigned to any of the treatment conditions completed a 1½- to 2-hour assessment interview, followed by a brief counseling protocol and the facilitation of social connections with parents and school personnel, which lasted about another 1½ to 2 hours. At 2½ months from baseline assessment and initial counseling, a booster session consisting of a follow-up interview, a brief counseling protocol, and a repeat of the social connection was completed. The random assignment of teens and their parents to the intervention and comparison conditions produced the following groups: 155 subjects receiving P–CARE, 153 subjects receiving C–CARE, 164 subjects receiving P&C–CARE, and 143 receiving the MI comparison condition.
Parents of teens assigned to the P–CARE intervention completed two home visits with CARE staff, lasting about 2 hours each. During these visits parents learned suicide prevention “first aid” and how to support their teen through difficult times with skills adapted from the CARE curriculum given to youths. A parent booster session, conducted by phone, was administered 2½ months after the baseline assessment. During this phone call the teen’s suicide-risk status was reviewed and specific strategies that were previously taught and aimed at reconnecting the youth to social support were reinforced.
The teens assigned to the C–CARE condition received the aforementioned assessment and brief counseling session as the other interventions. However, in this condition parents did not receive additional training or home visitations to further support the programmatic elements of CARE. Teens receiving the P&C–CARE intervention had the same assessment, counseling sessions, and booster sessions as the other interventions. Their parents also received the home visits and suicide prevention “first aid” that parents in the P–CARE intervention had been given. Separating out the intervention groups in this manner allowed the researchers to determine the effectiveness of each component of CARE as well as the effects of combining them.
The MI comparison group was included as a necessary part of the research design and ethical concerns when dealing with youths at risk for suicide. Teens assigned to MI participated in a brief (15- to 30-minute) assessment interview, using the 22-item Screen for Youth Suicide Risk, followed by a risk feedback and mobilization of support resources at home and at school. This was designed to simulate the usual intervention provided by school personnel. If at any point during the study a youth was considered to be at imminent risk of suicide, regardless of group assignment, all available help and counseling were to be brought in to take care of the teen, who would be removed from the study. No such incidents occurred during the course of the study.
To assess the long-term effects of CARE, there were two follow-up study contacts made during young adulthood by phone. These phone calls were 1-hour assessment interviews used to check up on study subjects. The time from baseline—ranging from 2.5 to 8.0 years for all follow-up interviews—was incorporated into the analysis. The retention rate was 86 percent (530 study participants) of the original 615 participants who were included in the long-term young adult follow-up study. Fourteen percent had left the initial study before the start of the long-term follow-up.
All participants completed the High School Questionnaire (HSQ) that measures suicidal behaviors (thoughts, threats, attempts), depression, and drug involvement. Suicide-risk behaviors were measured in two ways: direct behavior and distress factors. Direct behavior consisted of suicidal thoughts, notes, threats, and attempts. Protective factors acquired at baseline include self-efficacy/coping and family support. A respondent was deemed to be at an elevated suicide risk if he or she reported a previous suicide attempt or had high scores on the depression scale (4 and above on a scale of 6). Substance use and abuse were measured as the frequency of use within the last 30 days of alcohol and illicit drugs (cocaine, opiates, inhalants, etc.).
Latent class growth models were used to determine patterns of change from postintervention to long-term follow-up. This method of longitudinal analysis is well suited for detecting differences among groups over time. Such an analytic plan allows for estimates of growth trajectories (behavior change over time) while accounting for the differences in the length of time from baseline to follow-up. Missing data was considered missing at random, and proper adjustments were made in the model to account for this.
Eggert and colleagues (2002) used a three-group, repeated-measures, randomized prevention trial to study the effectiveness of CARE on 341 potential dropouts from seven high schools. The design included random assignment by school to one of two experimental conditions or to the “usual care” control condition. Adolescents were assigned to intervention groups through block randomization procedures. The starting condition for each school was randomly determined, and then study conditions were rotated according to the following sequence; usual care control condition, then C–CARE, then Coping and Support Training (CAST), and finally a no-intervention or “pause” condition. Study conditions were not repeated at any high school, nor were they nested within schools. Randomizing intervention conditions by school kept experimental and control conditions from occurring in the same school simultaneously, minimizing the possibility for contamination. The no-intervention condition allowed potential carryover effects within schools to dissipate over time.
Participants, potential dropouts, and youths at risk of suicide were selected using a two-part process. First, the pool of potential subjects was screened to identify potential dropouts by drawing on indicators that have been known to predict dropouts (low academic performance, poor attendance, and any earlier attempt or history of dropping out). Potential dropouts were then invited to join the study, at which point they were given the suicide risk screen instrument to identify those at risk for suicide. Risk level was based on seven items, which included suicide risk behaviors (thoughts, threats, prior attempts), depression, and drug involvement. Youths determined to be at risk of suicide were retained for the study; all others were excluded.
The final sample included students from 9th to 12th grade, ranging from 14 to 19 years old. The racial composition was 39.9 percent white, 12.9 percent biracial or mixed ethnicity, 12.9 percent Asian/Pacific Islander, 12.3 percent African American, 7.0 percent Hispanic, 2.1 percent American Indian, 3.8 percent other, and 9.1 percent unknown. More than half the sample (56.6 percent) had one or more school moves in middle or high school, and just under half (48.3 percent) had experienced a divorce. The death of a parent was experienced by 12 percent of the sample.
Youth at high risk for suicide were then assigned to one of two experimental groups: C–CARE (n=117) or CAST (n=103). Teens assigned to C–CARE immediately started receiving the 4-weeklong intervention. Youths placed into the CAST group started receiving the 6-weeklong intervention after the C–CARE intervention was finished. Those assigned only to C–CARE did not receive the CAST intervention. The control group (n=121) received the individualized “usual care” protocol at the same time the intervention groups were receiving C–CARE.
The C–CARE intervention is a one-to-one, 2-hour assessment interview followed by a 1½-hour to 2-hour counseling session and social “connections” intervention with parents and school personnel. C–CARE is delivered by trained research staff—typically advanced practice nurses or social workers. The motivational counseling session concentrates on delivering empathy and support, sharing personal information, and reinforcing positive coping skills and help-seeking behaviors.
The CAST intervention is a combination of the above C–CARE intervention, followed by a small group prevention program consisting of 12 individual sessions each 1 hour long. These sessions consist of small groups that work on skills training and social support. The sessions were delivered over 6 weeks to groups of six or seven adolescents by trained master’s level high school teachers, counselors, or nurses with considerable school-based experience. Each session incorporated key objectives, skills, and an implementation plan detailing activities to be carried out by the CAST Leader. The subject and content of these sessions targeted mood management, drug-use control, and school performance. Each session consisted in part of helping youth apply newly acquired skills and gain support from family and other trusted adults.
The usual care control condition simulated the procedures typically used by school staff to deal with youth at suicide risk. Beck’s Suicide Ideation and Intent Scales were used to assess youths at risk of suicide. After the brief (30 minute) assessment, staff followed a short social-connections procedure and notified parents and select school personnel.
All participants filled out the HSQ at baseline assessment and at three follow-up assessments that coincided with the delivery of the intervention. The first follow-up assessment occurred at 4 weeks from baseline, when the C–CARE or “usual care” conditions had been implemented. The second follow-up occurred at 10 weeks from baseline or with the conclusion of the CAST intervention. The last assessment was completed 9 months from baseline.
The HSQ was used to measure suicide-risk behaviors, related risks, and protective factors. Suicide risks were seen as a group of self-destructive thoughts and behaviors including prior suicide attempts, written or verbalized threats of suicide, and suicidal thoughts. Related risks included measures of depression, anxiety, illicit drug use and control problems, and adverse drug consequences denoting addiction or dependency. Repeated-measures analysis of variance (ANOVA) was used to assess trends in the data over time. This trend analyses can detect changes in behavior, measured as trajectories, and as such can determine whether an intervention altered the behavior in question.