tiate program impact from broader co-occurring trends. In the case of the comprehensive, multilevel educational programs, insufficient attention has been paid to documenting which program components are responsible for the reported outcomes.

An additional limitation of currently available data on the impact of universal education programs is their short-term focus. It is not clear if ongoing interventions might serve as “booster shots” to enhance and reinforce a program's impact. In addition, follow-up evaluations of these programs have been rare, and thus little is currently known about their impact on reducing suicidal behavior among the targeted group. Longitudinal controlled studies that look at youth several years after participating in educational programs are needed to address the question of long-term behavioral change. This will require addressing the fact that neither high schools nor colleges currently have a reliable system for reporting suicidal behaviors among students, thus hampering collection of reliable data to determine an educational program's impact. Also, students graduate and leave the school environment, making follow-up difficult.

Long-term controlled studies of gatekeeper training programs are likewise needed to determine the frequency or the effectiveness of participants' direct interventions during the years following the training. Because little is known about particular approaches that make referral efforts safe and effective, further evaluation is needed of the impact of such programs on referral processes, adequate treatment, and, in turn, the reduction of suicide risk factors and suicidal behavior among youth.

Some concerns have been voiced by high school personnel and parents that overt discussion of suicide in the school curriculum may increase suicidal thoughts and behavior, and adequate attention has generally not been given by evaluators to documenting adverse effects. One study found statistically significant increases in hopelessness and maladaptive coping resources among some male students after exposure to a suicide awareness curriculum (Overholser, Hemstreet, Spirito, & Vyse, 1989). Studies by Shaffer and colleagues (1990, 1991) and Vieland and colleagues (1991) found that students who had pre viously made a suicide attempt were less likely to recommend suicide awareness programs in the schools, and were more likely to feel that talking about suicide in the classroom would increase suicidal behavior among some students. While the small number of students reporting past suicidal behavior limit generalization of these findings, they point to the need for evaluations of school-based suicide education programs to include better assessment of potential harmful effects and identification of adolescents who may be vulnerable to adverse effects. Educational programs should also include a plan for clinical assessment and referral for students who are identified to be at risk for suicidal behavior. It is essential that school personnel be made aware of referral sources in the community and for the school to have in place a plan of action for identified students that includes a debriefing component for peers and faculty who are involved in making referrals.

In the case of college-based programs, concerns about effects on the institution's legal liability, reputation, and student enrollment sometimes encourages campus officials to avoid or minimize the problem of student suicide, which appears to have limited the development of educational programs directed to this population. In addition, providing suicide education to college students poses unique issues. In contrast to high school students, who follow a tightly prescribed core curriculum that typically includes at least a minimal amount of health education, college students are not generally required to take any courses in which education about depression and suicide may be appropriately incorporated. Other than a limited number of mandatory orientation sessions, few opportunities exist to reach large numbers of college students with information about mental health issues and services. Involvement of parents in educational programs on such issues is also extremely limited in most college settings.

Finally, it should be noted that most suicide prevention programs directed to young adults are designed specifically for college students, who represent less than half of all persons aged 18–24 in the United States. Although few research studies have examined suicide risk among young adults not in college, this population may

have particular risk factors, including more involvement with substance use, as well as less access to mental health resources.

One effort that may have applicability to youth in noncollege settings is the U.S. Air Force suicide prevention program, which has focused on removing the stigma of seeking help for mental health and psychosocial problems, enhancing understanding of mental health, and changing policies and social norms within the service. Introduced in 1996–97, the Air Force program has been described as highly effective in reducing suicide and other adverse outcomes, including family violence and homicide, among its five million members. A recent evaluation that compared 5-year cohorts before and after program implementation reported a 33% relative risk reduction for suicide and reductions ranging between 18% and 54% for other outcomes (Knox, Litts, Talcott, Feig, & Caine, 2003). The impact of the program on young servicemen in particular has not been reported, and thus the program is not listed as a youth suicide prevention program in the current review.

Screening for depression in adults has been demonstrated to increase the likelihood of depressed adults seeking mental health treatment (Greenfield et al., 1997, 2000). Universal screening programs as a youth suicide prevention strategy (listed in Table 22.3) are designed to identify young people at risk for suicidal behavior and refer them to treatment. Some programs focus specifically on identifying symptoms of psychopathology known to be related to adolescent and young adult suicidal behavior, while others assess specifically for signs of suicidality.

The primary assumption underlying screening programs is that because anxiety, depression, substance abuse, and suicidal preoccupation among youth often go unnoticed and untreated, a systematic, universally applied effort is needed to improve identification of at-risk individuals. Although not always explicitly stated, screening programs also rest on the assumptions that iden tification of youth with psychiatric disorders will substantially increase the number receiving treatment, the treatment will be sufficiently effective, and effective treatment will decrease suicides.

Reynolds (1991) described one of the first high school–based screening programs for youth at risk for suicide. The program involved a two-stage method, in which a general population of students was first screened using the Suicide Ideation Questionnaire (Reynolds, 1988). Students with scores above a defined cutoff value were subsequently evaluated clinically with the Suicide Behavior Interview, and those identified as being at risk were referred for treatment.

The program devoted particular attention to determining an appropriate cutoff score for identifying at-risk youth, comparing two different scores with regard to sensitivity (the ability to identify correctly those who have the problem, with few false negatives) and specificity (the ability to identify correctly those who do not have the problem, with few false positives). Reynolds found that increasing the cutoff score led to missing a disproportionate number of at-risk youth. The impact on suicidal behavior and the adherence to treatment recommendations were not reported.

Perhaps the most widely used high school screening program, the Columbia TeenScreen Program (CTSP), likewise employs a multistage procedure. In one variant of the CTSP, students complete a brief, self-report questionnaire, i.e., the Columbia Suicide Screen. Those who screen positive on this measure are given a computerized instrument, the Voice DISC 2.3, a version of the Diagnostic Interview Schedule for Children, which has been found to accurately identify a comprehensive range of psychiatric disorders in children and adolescents (Shaffer, Fisher, Lucas, Dulcan, & Schwab-Stone, 2000; Shaffer et al., 2004). This stage of the screen is regarded as particularly important for avoiding overidentification of students at risk. In the final stage, youth who have been identified through Voice DISC 2.3 as meeting specific diagnostic cri