Suicide and Life-Threatening Behavior 44 (6) December 2014 © 2014 The American Association of Suicidology DOI: 10.1111/sltb.12097
“Signs of Suicide” Shows Promise as a Middle School Suicide Prevention Program ELIZABETH A. SCHILLING, PHD, MARTHA LAWLESS, BA, LAUREL BUCHANAN, MA, ROBERT H. ASELTINE JR., PHD
Although the Signs of Suicide (SOS) suicide prevention program has been implemented at both the middle and high school levels, its efficacy has been demonstrated previously only among high school students. The current study evaluated SOS implemented in high military impact middle schools. Compared to controls, SOS participants demonstrated improved knowledge about suicide and suicide prevention, and participants with pretest ideation reported fewer suicidal behaviors at posttest than controls with pretest ideation. These results provide preliminary evidence for SOS’s efficacy as a suicide prevention program for middle school students. Recent data from 2010 indicate that suicide becomes a measureable public health problem once children reach middle school. Suicide is the third leading cause of death in this age group; 267 children aged 10 to 14—approximately 1.29 per 100,000—completed suicide in 2010 (McIntosh & Drapeau, 2012). Middle school students are in a “sensitive period” of rapid biological, social, and psychological development which may heighten the risk of anxiety, depression, and related psychological difficulties (Reardon, LeenFeldner, & Hayward, 2009). Early adolescence marks the rise in risk of suicide; as children age into late adolescence and early adulthood, the suicide rate increases dramatically (McIntosh & Drapeau, 2012). Thus, middle school may offer a prime ELIZABETH A. SCHILLING, MARTHA LAWLESS, LAUREL BUCHANAN, and ROBERT H. ASELTINE JR., Center for Public Health and Health Policy, The University of Connecticut Health Center, East Hartford, CT, USA. Address correspondence to Elizabeth A. Schilling, Center for Public Health and Health Policy, The University of Connecticut Health Center, 99 Ash. St., MC7160, East Hartford, CT 06108; E-mail: [email protected]
venue in which to sensitize children and their families to the warning signs of suicide and to educate them about effective ways to intervene. Middle schools that have a high percentage of students with parents in the military, “high military impact” schools, may be in particular need of efficacious suicide prevention efforts because many risk factors for adolescent suicide are consequences of, or inherent in the experience of, parental deployment. Sequelae of parental deployment include emotional intensity and conflict in the family (Chandra, Lara-Cinisomo, et al., 2010; Esposito-Smythers et al., 2010; Huebner, Mancini, Wilcox, Grass, & Grass, 2007; Johnson et al., 2007), depression and anxiety in the nondeployed parent (Esposito-Smythers et al., 2010), depression and anxiety in the adolescent (Chandra, Lara-Cinisomo, et al., 2011, 2011), weaker connection to school (Richardson et al., 2011), and maltreatment (Campbell, Brown, & Okwara, 2011), all of which are risk factors for adolescent suicide (Bridge, Goldstein, & Brent, 2006; Wells & Heilbron, 2012). Adolescents with parents in the military who are deployed, or preparing to deploy, are at higher risk of emotional
654 and behavioral difficulties (see EspositoSmythers et al., 2010, for a review). Parental deployment inherently involves “ambiguous loss” (Huebner et al., 2007), and adolescent suicide is often precipitated by events that involve interpersonal conflict or loss (Bridge et al., 2006). Despite the potential for effective suicide prevention at the middle school level, most adolescent suicide prevention research has taken place at the high school level. None of the studies in a recent review of 15 school-based universal suicide intervention program evaluations included participants younger than the eighth grade (Robinson et al., 2013), and none focused on schools with high percentages of students experiencing parental military deployment. Of the eight suicide prevention programs included in another recent review of high quality programs, only two programs were implemented in middle schools and neither included a follow-up assessment (Cusimano & Sameem, 2011). These evaluations generally demonstrated positive results. Of the 15 evaluations reviewed by Robinson et al. (2013), 6 studies reported reductions of suicide-related outcomes, 9 studies reported improvements in knowledge of suicide, and 7 of the 11 studies that measured attitudes toward suicide reported significant improvements. Cusimano and Sameem (2011) reviewed only “well performed, controlled” (p. 48) studies, and concluded that suicide prevention research has generally demonstrated the efficacy of these programs for improving knowledge, attitudes, and helpseeking. The Signs of Suicide (SOS) program (Aseltine & DeMartino, 2004; Aseltine, James, Schilling, & Glanovsky, 2007) was included in both of the above reviews. As one of the most well-researched suicide prevention programs at the high school level, SOS is currently listed in the National Registry of Evidence-based Programs and Practices (NREPP; Substance Abuse & Mental Health Services Administration, 2011). Three separate randomized controlled studies have documented statistically significant
decreases in suicide attempts (Aseltine & DeMartino, 2004; Aseltine et al., 2007; Schilling, Aseltine, & James, 2013). In addition, compared to control students, SOS participants demonstrated increases in knowledge about depression and suicide and increases in favorable attitudes toward obtaining help for depression and suicidal thoughts. Differences in self-reported levels of help-seeking were generally not detected in any of these evaluations, however. The philosophy behind the SOS program is that suicide is usually a consequence of depression and that adolescents are most likely to turn to their peers for help when they are facing crises and considering suicide. The SOS program aims to sensitize youth to the symptoms of depression and signs of suicide, to educate them that depression is a very treatable condition, to emphasize the importance of obtaining help for themselves and their friends, and to provide youth with information about whom they should approach for help. In addition, the depression screening is intended to raise students’ awareness of their own risk related to depression. Finally, the program trains “trusted adults,” such as teachers and parents, to provide a supportive environment and to encourage their “approachability” by youth. Thus, the program aims to reduce suicidal thoughts and behavior by increasing students’ recognition of depression in themselves and their friends so that they can obtain help and by encouraging supportive home and school environments so that adolescents feel able to approach parents and school personnel for help. The middle and high school SOS programs are similar. Although they follow the same format, the content differs so that it is appropriate to the students’ developmental level. The middle school version features a 17-minute DVD that includes (1) three age-appropriate vignettes which are less intense than the high school version; (2) a group discussion by middle school students about depression, suicide, bullying, self-injury, and getting help; and (3) a
student interview with a school-based counselor to model getting help. Middle school vignettes involve (1) cyber-bullying, (2) a student exhibiting anger as a symptom of depression, and (3) a girl talking to her sister about being gossiped about by her friends. Following the discussion, the students are screened for depression and suicidal thoughts or behavior and then complete a follow-up card stating whether they would like to speak to an adult. Teachers are provided with detailed instructions regarding presentation and implementation of the program (Jacobs, 2013), and the kit includes instructions for the presentation of a parent training session. In the current study we aimed to replicate results from previous SOS evaluations implemented in high schools (Aseltine & DeMartino, 2004; Aseltine et al., 2007; Schilling et al., 2013) among a middle school, high military impact, population. SOS has features that are compatible with implementation in high military impact schools in a number of ways recommended by Esposito-Smythers et al. (2010): (1) because it is a universal program, it does not single out children at high risk for suicide, thus precluding stigma; (2) it is structured and directive, which may be more palatable to military families; and (3) because SOS is relatively inexpensive and very accessible, it is sustainable. In addition, SOS’s emphasis on peer help-seeking and intervention is particularly relevant to middle school student suicide prevention (Herring, 1990). Three categories of hypotheses were tested. The first involved suicidal behaviors and cognition. Participation in SOS was expected to be associated with less suicidal ideation or planning and/or fewer attempts. The second category involved the assessment of information acquired from participation in the program. Participation in SOS was expected to be associated with increased knowledge about depression and suicide, as well as with participants’ endorsement of better attitudes toward getting help for themselves or their
655 friends. The third category included help-seeking behaviors. Participation in the SOS program was expected to be associated with increased levels of help-seeking behaviors.
The University of Connecticut Health Center’s and the Department of Defense’s institutional review boards approved all procedures for this study. Participants Participants were fifth through eighth graders in middle schools with a high proportion of students who had parents in the military (high military impact schools). Participants in the SOS evaluation included students in one grade at each school. The demographic characteristics of consented students are presented in Table 1. Students were more likely to be White (40.8%) than Hispanic (20%), multiethnic (18.4%), or African American (11.7%). Students were also somewhat more likely to be female (52.6%) than male (47.4%). A minority of students were English language learners (ELL; 18.7%), and 36% were eligible to receive free or reduced lunch. A large majority (85%) had at least one parent or caregiver serving in the military, and a quarter of students (25.8%) reported having a deployed parent. Procedures Twenty middle schools identified by the Department of Defense as high-impact were approached for participation in the study. The SOS program was scheduled to be presented independent of the evaluation at all of these schools. Eleven schools agreed to participate. Of these 11 schools, one dropped out before the study began, and two did not adhere to the study protocol and were dropped from the evaluation. Thus, eight schools provided valid data for
Demographic Data for Intervention and Control Groups Control n
Race or ethnicity 44 50.6 113 White (nonHispanic) Black 9 10.3 36 (nonHispanic) Hispanic 13 14.9 64 Asian 3 3.4 18 Other 2 2.3 12 Multiethnic 16 18.4 55 Gender Male 40 46.0 143 Female 47 54.0 156 ELL* Yes 9 10.3 63 No 78 89.7 236 Grade in school* 6 0 0 42 7 67 77.0 248 8 20 23.0 9 Free lunch Yes 22 31.9 84 No 47 68.1 138 Average course grades A 49 56.3 158 B 29 33.3 105 C 8 9.2 26 D 1 1.1 6 F 0 0 0 Parent or caregiver in military Yes 74 85.1 251 No 13 14.9 46 Parent or caregiver deployed* Yes 6 7.0 93 No 80 93.0 205
37.9 157 40.8
21.5 6.0 4.0 18.5
77 20.0 21 5.5 14 3.6 71 18.4
47.8 183 47.4 52.2 203 52.6 21.1 72 18.7 78.9 314 81.3 14.0 42 10.9 82.9 315 81.6 3.0 29 7.5 37.8 106 36.4 62.2 185 63.6 53.6 207 54.2 35.6 134 35.1 8.8 34 8.9 2.0 7 1.8 0 0 0 84.5 325 84.6 15.5 59 15.4 31.2 99 25.8 68.8 285 74.2
ELL, English language learners *p < .05 for difference between intervention and control groups.
analyses. The evaluation included students who were scheduled to receive the program (thus, students were not matched on grade). Control schools included seventh (79%)
and eighth (21%) graders; intervention schools included sixth (14%), seventh (84%), and eighth (3%) graders. All students scheduled to receive the SOS program were potentially eligible to participate in the evaluation. School personnel who presented the program were trained to collect data for the evaluation with a webinar and individualized instruction; research personnel were available remotely to answer any questions but were not physically present in the schools at any time. Consent forms approved by the University of Connecticut Health Center and the Department of Defense were provided to the participating schools. Only students who were given consent by a parent and also assented to participation (“consented students”) were included in the evaluation. The forms were sent home with the students or included in direct mailings by schools to parents and guardians. Once consent forms were returned to school counselors, the consent of parents or guardians was verified. As an incentive to complete the consent forms, all completed forms (with either “yes” or “no” to participation marked) were entered into a random drawing. Two prizes were awarded at each school, a $50 Amazon gift card and an iPod Touch valued at approximately $190. Among the eight schools that participated in the evaluation, 51% (682/1,326) of eligible students returned consent forms; of these, 212 (31%) indicated “no” to participation and 470 (69%) indicated “yes” (i.e., were consented). Schools were randomly assigned to intervention and control groups. However, the randomization was compromised by two events that altered treatment assignments. The group assignment of one school changed from control to intervention due to unalterable scheduling issues. In addition, another school, originally assigned to the control group, presented the program prematurely; this school was thus reassigned to the intervention group. Thus, the final assignment included two schools in the control group and six in the intervention group.
The program was presented in schools in the intervention group from November 2009 through March 2010; during the same period, students in the control group completed the pretest questionnaires but did not participate in the program. The program was presented mainly by school psychologists and counselors; in one school, a health teacher also presented the program. These school personnel were trained in the program presentation with a detailed manual and DVD included with their SOS program kit, as well as an online interactive module that qualified for continuing education credits for school counselors, social workers, guidance counselors, and school psychologists. Pretest questionnaires for this evaluation were collected in class at all schools before the SOS program implementation during the 2009–2010 school year (419/470 = 94% of consented students). Participants and control students completed posttest questionnaires 3 months following completion of the pretest questionnaires. Questionnaires used for the evaluation were marked with an ID, and each questionnaire was placed in a sealed envelope with the student’s name written on the envelope only; the research protocol instructed students to discard the envelope after completing the questionnaire. Schools did not have the key and could not identify respondents. Responses on the questionnaires were confidential and were not used to identify students with suicidal ideation or intent. Ninety-two percent of students who completed the pretest completed the posttest (386/419). Students were informed prior to survey administration that their responses were confidential and would not be shared with the school. Measures The questionnaire included items relevant to three specific categories of outcome: (1) self-reported suicidal ideation and suicide attempts, (2) knowledge and attitudes about depression and suicide, and (3) help-seeking. The primary endpoint was a
657 combination of three single-item measures of self-reported suicide ideation, planning, and attempts which were adapted from the Centers for Disease Control and Prevention’s (CDC) Youth Risk Behavior Survey (YRBS) to assess behavior in the past 3 months instead of the past 12 months (Kann et al., 2000): “During the past 3 months, did you actually attempt suicide?” (yes or no); “During the past 3 months, did you ever seriously consider attempting suicide?” (yes or no); and “During the past 3 months, did you make a plan about how you would attempt suicide?” (yes or no). Lifetime attempts were also assessed with the question: “Have you ever attempted suicide?” The measures of knowledge and attitudes about depression and suicide were adapted from instruments previously used to evaluate school-based suicide prevention programs (Shaffer, Garland, Vieland, & Underwood, 1991; Spirito, Overholser, Ashworth, & Morgan, 1988). Some items were modified by changing item wording, recall periods, or response categories. Knowledge of depression and suicide was measured with seven true or false items that reflect the central themes of the SOS program (e.g., “People who talk about suicide don’t really kill themselves”; “Depression is an illness that doctors can treat”). Scores on this variable reflected the number of correct answers. The test–retest reliability of the knowledge measure in the control group was low (r = .33 for the total score and test–retest kappas of individual items ranging from .25 to .60). The measure of attitudes was a 10-item summary scale that assessed attitudes toward intervening with friends who exhibit symptoms of depression and/or suicidal intent and toward getting help for themselves (e.g., “If someone really wants to kill him/herself, there is not much I can do about it”; “If a friend told me he/she is thinking about committing suicide, I would keep it to myself”). Responses to these questions ranged from strongly disagree to strongly agree on a 5-point scale, with higher values indicating more favorable attitudes toward intervening (Cronbach a = .73).
Eight questions were used to assess help-seeking behavior. Students were asked whether in the past three months, “you received treatment from a psychiatrist, psychologist, or social worker because you were feeling depressed or suicidal” (yes or no); whether “you talked to a [parent or guardian, brother or sister, teacher or guidance counselor, other adult, friend, crisis or telephone hotline worker] because you were feeling depressed or suicidal” (yes or no for each type of person); and whether “you talked to an adult about a friend you thought was feeling depressed or suicidal” (yes or no).
Comparability of Treatment and Control Groups Preliminary analyses were conducted to assess the comparability of the intervention and control groups in terms of race or ethnicity, gender, ESL status, grade in school, eligibility for free lunches, grades,
and parental military service and deployment. Significant differences between intervention and control groups are designated by an asterisk in Table 1. Gender, race or ethnicity, free lunch status, and grades did not differ between intervention and control groups. However, chi-square tests revealed statistically significant differences in the composition of intervention and control groups by ELL status, X2 (1, N = 386) = 5.1, p < .03, grade in school, X2 (1, N = 386) = 48.3, p < .001, and parental deployment, X2 (1, N = 384) = 20.5, p < .001, with the intervention group having a higher percentage of students in ELL and with deployed parents. Students’ knowledge and attitudes about depression and suicide and the prevalence of suicidal thoughts and behaviors are presented in Table 2. Most measures of suicidal thoughts and behaviors were comparable at pretest among the students in the intervention and control groups. Exceptions involved suicidal ideation and planning, which differed at pretest in the intervention and control groups, with the intervention group reporting significantly lower rates of ideation, Χ2 (1, N = 381) = 6.1, p < .05, and
Percentage of Students Reporting Suicidal Thoughts and Behaviors, Knowledge, and Attitudes at Pretest for Students with Valid Posttest Data Total
Suicidal ideation during past 3 months* Suicide plan during past 3 months* Suicide attempt during past 3 months Suicidal behavior during past 3 months Lifetime suicide attempt
10.5 11.6 0.0 12.9 7.1
3.7 4.7 1.7 7.8 5.9
5.2 6.3 1.3 9.0 6.1
381 381 382 380 374
Scale Knowledge of depression/suicideb Attitudes toward help-seeking for depression and suicide
4.55 (1.31) 4.05 (.57)
4.58 (1.20) 4.01 (.59)
4.57 (1.22) 4.02 (.58)
a Valid number of cases for the item or scale. The number of cases for the full sample is N = 386, with n = 87 in the control group and n = 299 in the intervention group. b Missing values are ignored in the count unless all items are missing in which case result (count) is missing. * p < .05 for difference between intervention and control groups.
planning, Χ2 (1, N = 381) = 5.3, p < .05. Knowledge and attitudes did not differ at pretest in the intervention and control groups. Numbers and percentages of helpseeking behaviors are presented in Table 3. Two behaviors differed at pretest: (1) whether the participant had talked to a friend in the last 3 months about being depressed or suicidal, Χ2 = 8.1, df = 1, p < .01, and (2) whether he or she had talked to an adult about a friend in the last 3 months, Χ2 = 4.4, df = 1, p < .05. For both of these behaviors, the control group reported higher rates. Outcome Evaluation: Indications of Program Efficacy Regression and logistic regression analyses were performed in SAS version 9.3 (Cary, NC: SAS Institute Inc.). Intervention effects were estimated in models that accounted for the clustered sampling design in which students were nested within schools. The effects of SOS participation on suicidal behavior, help-seeking, knowledge, and attitudes (S2) were estimated with the following model:
S2 ¼ B0 þ B1 S1 þ B2 G1 þ B315 Controls315
where S2 is the outcome at follow-up; S1 is the relevant pretest score(s) defined below; G1 is a dummy variable for intervention status; Controls3-15 refers to a series of demographic and control characteristics which include dummy variables for sex (female vs. male), race or ethnicity (Black, Asian, Hispanic, Multiracial, and Other race) with White race as the omitted reference category, parental military status at pretest (yes vs. no), parental deployment status at pretest (yes vs. no), average grades, grade in school (6 and 7 vs. 8 as reference category), and posttest report of professional treatment for depression or suicidality in the previous 3 months. In addition, a dummy variable for ELL status was added to the model because it significantly distinguished treatment from control groups at pretest. The predictors in Equation (1) varied slightly for the three categories of outcomes: (1) suicidal behavior at posttest (suicide attempt, planning, and ideation), (2) knowledge and attitudes, and (3) help-seeking
Help-Seeking Behaviors at Pretest in the Previous 3 Months for Students with Valid Posttest Data Help-Seeking Behaviors Treatment from a psychiatrist, psychologist, social worker Talked to adult about a friend* People talked tob: Parents and guardians Brother or sister Teacher or guidance counselor Other adult Friend* Crisis or telephone hotline worker a
19.5 11.5 12.6 9.2 35.6 1.1
23.7 7.5 8.5 6.5 20.7 0.3
22.8 8.4 9.5 7.1 24.1 0.5
382 381 380 380 381 380
Valid number of cases for the item or scale. The number of cases for the full sample is N = 386, with n = 87 in the control group and n = 299 in the intervention group. b Item stem: “Have you talked to any of the following people because you were feeling depressed or suicidal?” * p < .05 for difference between intervention and control groups.
660 behavior. For models predicting suicidal outcomes, S1 included two variables: (1) a dummy variable indicating pretest suicidal ideation, because it is considered the first stage in the process toward planned suicide (Perez, 2005); and (2) a dummy variable indicating pretest report of lifetime suicide attempt. Because a history of one or more suicide attempts has been identified as the most robust predictor of future suicidal behavior (Joiner et al., 2005), it was considered to be the best indicator for high suicide risk in our data. For models predicting knowledge and attitudes, S1 was the pretest score of knowledge or attitudes, respectively, and the posttest report of professional treatment for depression was excluded. For models predicting help-seeking, S1 included three variables: pretest help-seeking, posttest suicidal ideation, and posttest report of a lifetime attempt. The posttest report of suicidal ideation was used to coincide with the same time period for which the help-seeking behavior was assessed (i.e., 3 months prior to the posttest). A fourth model predicting suicidal behavior was designed to test whether the SOS program influenced the theoretical progression of planned suicide that begins with ideation and ends in a suicide attempt (Perez, 2005); thus, it was constructed to estimate the effect of the SOS program on posttest suicidal behaviors for middle school students who reported suicidal ideation at pretest. Suicidal behavior was defined as a dichotomous variable indicating self-report at posttest of (1) suicide attempt, (2) suicidal ideation, and/or (3) suicide planning. In addition to the variables in Equation (1), this model included the interaction of pretest suicidal ideation and intervention status. The effects of the SOS program on students’ suicidal behavior, and on knowledge of and attitudes toward depression and suicide, are shown in Table 4. For attitudes and knowledge, coefficients from regression analyses in Proc MIXED are presented. For suicidal behavior, coefficients from logistic regression analyses in Proc GLIMMIX are presented.
The coefficient for the effect of the SOS program on posttest suicidal attempts did not converge because of the very small number of attempts. Intervention effects on posttest suicidal ideation and planning were not significant (see models 1 and 2 in Table 4). However, in model 3, the SOS program was associated with significantly less risk of suicidal behavior (ideation, planning, and/or attempts) among students reporting pretest ideation in the intervention group compared to the control group, controlling for pretest levels of lifetime suicide attempt (p < .05). The contrast effect of the SOS program on suicidal behavior was 3.28 [(5.79 + .84 - 4.12)– 5.79 = 3.28], which, when converted to an odds ratio (OR), indicates that students with pretest ideation in the intervention group at posttest were approximately 96% less likely to report suicidal behavior in the previous 3 months than students with pretest ideation in the control group (OR = e3.28 = .038, t(341) = 2.03, p < 05). To illustrate this effect, a comparison of the proportions of students in the intervention and control groups who reported suicidal thoughts and/or behavior at posttest, and who had reported these same behaviors at pretest, is shown in Figure 1. The effects of participation in the SOS program on attitudes toward, and knowledge of, depression and suicide are presented in models 4 and 5, respectively, in Table 4. Participation in the SOS program resulted in greater knowledge of depression and suicide. The effect of the SOS program on attitudes was not significant (b = .13, SE = 0.09, p > .1). The effect of the SOS program on knowledge was statistically significant with an effect size of 2/5 of a standard deviation (e.g., .53/ 1.31 = .40, p < .05). This is a small to medium effect size (Cohen, 1977). The effects of participation in the SOS program on help-seeking are presented in Table 5 for models which converged (help-seeking from parent, brother or sister, or friend). Models predicting help-seeking for a friend, or from professionals, teachers,
Effects of Participation in SOS on Suicidal Behavior, Knowledge, and Attitudes at Posttest
Effect Intercept Pretest scorea Pretest ideation Pretest lifetime attempt SOS program Female Black Asian Other Hispanic Multiethnic Parent serve Parent deployed Grades ELL Grade 6 Grade 7 Treated by professionala Pretest ideation*SOSa n
Any Suicidal Behavior
2.08 – 3.71* 1.23
3.05 – 0.76 0.78
2.30 – 2.54* 2.29*
2.72 – 0.76 0.71
0.28 – 5.79* 2.53*
2.36 – 1.41 0.73
1.55* 0.56* 0.37* 0.05
0.27 0.04 0.12 0.11
2.37* 0.37* 0.02 0.01
0.55 0.05 0.27 0.25
0.53 0.29 1.52 1.01 1.42 0.25 0.65 0.67 0.28 0.36 1.21 1.93 2.08* 0.66
1.01 0.61 1.42 1.21 1.31 0.88 0.77 0.97 0.67 0.38 0.89 1.63 0.95 1.20
0.35 0.39 0.19 1.83 2.55* 1.57 1.40 0.27 0.16 0.38 1.17 0.10 0.94 0.46
0.86 0.55 1.12 1.29 1.09 0.83 0.83 0.89 0.65 0.32 0.81 1.85 1.32 0.93
0.84 0.18 0.33 1.36 1.81 0.84 0.83 1.62 1.02 0.43 1.57 2.09 0.94 1.17
0.73 0.49 0.89 1.25 1.03 0.74 0.71 1.04 0.66 0.31 0.84 1.40 0.87 0.83
0.13 0.09 0.04 0.06 0.11 0.02 0.11 0.14 0.05 0.01 0.07 0.11 0.03 –
0.09 0.05 0.08 0.12 0.13 0.07 0.07 0.07 0.06 0.04 0.07 0.14 0.11 –
0.56* 0.33* 0.04 0.39 0.29 0.21 0.03 0.31 0.05 0.21* 0.25 0.36 0.38 –
0.25 0.12 0.20 0.27 0.31 0.16 0.16 0.17 0.14 0.08 0.16 0.36 0.26 –
ELL, English language learner aDashes indicate that the variable was not included in model. *p < .05.
or a crisis center or hotline did not converge. Pretest help-seeking strongly predicted posttest help-seeking for all types of help-seeking. Suicidal ideation significantly predicted more help-seeking from a friend (b = 1.67, SE = 0.66, p < .05), but students who reported a lifetime attempt at posttest were less likely to seek help from a friend in the previous 3 months (b = 1.41, SE = 0.70, p < .05), controlling for pretest help-seeking and suicidal ideation during the same time period. On the whole, SOS participation did not significantly affect help-seeking from a parent, brother or sister, or friend.
This is the first evaluation of the SOS program among middle school students, with results suggesting promise for its efficacy with younger adolescents. In addition, we evaluated the SOS program in schools with a relatively high proportion of military dependents, many of whom are in some respects more emotionally vulnerable than the general public school population (Chandra, Martin, Hawkins, & Richardson, 2010; Chandra, Lara-Cinisomo, et al., 2010; Esposito-Smythers et al., 2010). Results
20 N=0 N=5
Post-test report of suicidal behavior Figure 1. Percentage of students who reported suicidal behavior at posttest among those who reported same behavior at pretest.
Effects of Participation in SOS on Help-Seeking at Posttest
Effect Intercept Pretest score Posttest ideation Posttest lifetime attempt SOS program Female Black Asian Other Hispanic Multiethnic Parent serve Parent deployed Grades ELL Grade 6 Grade 7
Brother or Sister
2.30 2.93* 0.60 1.01 0.31 0.27 1.17 2.03 0.81 0.26 0.28 0.13 0.20 0.25 0.19 0.20 0.10
1.60 0.38 0.67 0.69 0.68 0.36 0.75 1.16 0.92 0.49 0.51 0.52 0.46 0.26 0.50 1.04 0.70
1.51 2.31* 1.67* 1.41* 0.29 0.53 0.83 0.67 0.24 0.52 0.15 0.12 0.68 0.37 0.37 1.02 1.17*
1.40 0.33 0.66 0.70 0.41 0.33 0.62 0.75 0.82 0.42 0.45 0.47 0.37 0.22 0.43 0.77 0.57
2.16 3.79* 1.63 1.47 1.44 0.31 2.14 1.20 0.59 0.17 0.77 0.34 1.14 0.63 0.29 1.28 0.50
2.18 0.67 1.17 0.79 1.12 0.52 1.37 1.36 1.43 0.72 0.65 0.74 0.73 0.34 0.65 1.37 0.88
ELL, English language learners. *p < .05 (two-tailed).
from this study largely corroborate findings from previous SOS evaluations conducted among high school students (Aseltine, 2003; Aseltine & DeMartino, 2004; Aseltine et al., 2007). The SOS program had an impact on students’ self-reported suicidal behavior 3 months following exposure to the program, controlling for pretest reports of suicidal ideation and lifetime suicide attempt. Students who reported pretest suicidal ideation in the intervention group reported fewer suicidal behaviors (ideation, planning, and/or attempts) at posttest. Suicidal ideation is considered to be the initial stage in a progression that leads through planning to attempt and completion (Bridge et al., 2006; Perez, 2005). In addition, suicidal ideation varies in terms of seriousness (Carlton & Deane, 2000). Thus, suicidal ideation at pretest can remit, continue at the same level, or progress to more serious ideation, planning, or attempt. Remission of suicidal ideation would clearly constitute the most favorable outcome. Thus, our results support the interpretation that SOS may, for some students with suicidal ideation, interrupt the progression from suicidal ideation to more active and serious stages of contemplation, planning, and attempt. Although preliminary, this is an encouraging result. However, the impact of the SOS program on the prevalence of suicidal behaviors among middle school students with pretest suicidal ideation is a finding unique to this study, and as such, needs replication. Participation in the SOS program was associated with positive changes in knowledge of suicide and depression. The incorporation of a pretest assessment into the design of the current study enabled tests of change in knowledge and attitudes over the follow-up period to be performed by controlling for pretest scores in regressions on the posttest score (Kessler & Greenberg, 1981). The greater increase in knowledge of depression and suicide among SOS participants compared to controls was comparable to, although slightly
663 lower than, levels previously observed in high school samples. Although the magnitude of the effect of the SOS program on changes in favorable help-seeking attitudes was similar in magnitude to the effect in previous evaluations, it was not statistically significant. Consistent with results from evaluations of SOS in high school samples, no significant effects of the SOS program on changes in help-seeking behaviors were found in the middle school sample. This negative result is not a function of low statistical power because the direction of the effect of participation in the SOS program was opposite to the hypothesized direction; that is, for all three types of help-seeking successfully modeled, the direction of the nonsignificant effects was negative (in the direction of less help-seeking). As with the other results in this evaluation, replication in another, larger middle school sample is necessary. However, as of yet, the efficacy of the SOS program in reducing suicidal behaviors has not been associated with increased helpseeking as hypothesized. Although help-seeking will continue to be an important behavior to assess in future SOS evaluations, other potential mechanisms also need to be assessed. Without prompting actual help-seeking behavior in participants prone to suicidal ideation, attitude change toward help-seeking might alter cognitive structures, such as schemas or assumptions, by lowering perceived barriers to help-seeking (Cigularov, Chen, Thurber, & Stallones, 2008). For example, exposure to the SOS curriculum might reduce rigid expectations of self-reliance and the vignettes modeling communication with a counselor might increase participants’ confidence in their ability to approach adults about their problems if they become desperate. In addition, “trusted adult” training might result in more supportive and approachable school and home environments which might soften a depressed student’s sense of alienation from his or her “ecological microsystem” of
664 home and school (Bronfenbrenner, 1979). Finally, simply by acknowledging and addressing the phenomena of teen depression and suicide, SOS may weaken feelings of “differentness” and loneliness that depressed youths often feel (Groholt, Ekeberg, Wichstrom, & Haldorsen, 2005). Such cognitive changes in vulnerable adolescents might engender a greater sense of integration into their social systems that, in turn, might lessen feelings of isolation which play an important role in the suicidal desire to die (Joiner, Hollar, & Van Orden, 2006; Ribeiro & Joiner, 2009). Future SOS evaluations should expand the assessment of mediating mechanisms in an effort to identify the “active ingredients” of the program. In addition to the positive effects of the SOS program, the statistical models include effects that, although tangential to the evaluation, are consistent with findings reported previously in the literature. For example, the only type of help-seeking associated with suicidal ideation during the follow-up period was elicited from a friend, a consistent finding in other research (Cigularov et al., 2008; Michelmore & Hindley, 2012). These tangential findings provide support for the internal validity of the study and the integrity of the data. This support is particularly important because of the difficulties involved in conducting the evaluation. Participating schools were already scheduled to implement the program as part of their curriculum before being recruited for the evaluation; thus, researchers directed the evaluation off-site and did not control program implementation. The limited influence of research staff on program implementation and evaluation resulted in a number of limitations that were the consequences of evaluating a program implemented in real-world conditions, a common characteristic of effectiveness, as opposed to efficacy, research (Flay et al., 2005). Other limitations relate to the sample. First, because two schools originally assigned to the control group essentially self-selected into the intervention group,
our sample was closer to a convenience sample of high military impact schools than to the designed random sample. Second, high military impact middle schools may differ from civilian middle schools in terms of their suitability for the SOS implementation. Third, if differences between the consented and targeted samples existed, our results may have been affected by selection bias. All of these limitations have the potential to compromise the generalizability of the results. Fidelity to the intervention protocol is also another area of limitation. Because researchers did not directly collect the data in this study, the fidelity of program training and procedures was not monitored or assessed. Thus, we do not have evidence that all aspects of the SOS program were implemented as directed in the SOS Procedure Manual (Jacobs, 2013), and the variability of the implementation of SOS in the current study was likely closer to that of a widely disseminated program instead of a controlled efficacy study. In addition, a related limitation is that independent evaluators did not collect the study data; the school personnel who administered the program also administered the pretest and posttest surveys. Thus, it is possible that bias could have been introduced into the data. However, the comparability of these findings to previous studies of the SOS program conducted under better controlled conditions suggests that the positive impact of the program can be achieved in real-world conditions. Limitations involving our data are also notable. First, our sample did not include enough self-reported suicide attempts to evaluate whether participation in SOS at the middle school level was associated with fewer attempts. Future studies will need sufficiently large sample sizes to provide the statistical power necessary to evaluate the impact of SOS on specific suicidal behaviors, especially attempts, in middle school adolescents because the prevalence of suicide attempts is very low in this population (McIntosh & Drapeau, 2012). Second, the test–retest reliability of the knowledge measure in the con-
trol group was low. This may result, at least partially, from the longer time period between tests than the recommended 2 to 4 weeks, which can “drastically underestimate” the test–retest reliability by interpreting true change as measurement error (Carmines & Zeller, 1979). Although we found increases in knowledge in the treatment group, it is possible that a more reliable measure of knowledge would have provided stronger evidence of knowledge acquisition. In any case, it is reassuring that the results regarding knowledge and attitudes are similar to results from previous SOS evaluations. A third limitation involves the assessment of suicidal thoughts and behaviors with only one item each. Although the single items used to assess suicidal ideation and planning are worded to clearly indicate that the questions relate to the actual acts of planning and thinking about committing suicide and not just the concept of suicide in general, and are comparable to screening items used in more in-depth interviews about suicide (Nock, Holmberg, Photos, & Michel, 2007), this is an important limitation. Future SOS evaluations should use an instrument that provides more comprehensive assessment of suicidal ideation and behavior. A final limitation, which is true of all SOS evaluations to date, is the limited length of the follow-up period. Longer follow-up periods are warranted to assess the likely decay in program effects, as well as any potential sleeper effects. Skills gained in
665 middle school—acquisition of knowledge about depression and suicide, and better attitudes toward intervening with friends and getting help for oneself—may impact suicidal behaviors more strongly later in adolescence when the rates of suicidal behaviors increase. In addition, middle school adolescents who participate in the SOS program may be primed to absorb more from suicide prevention programs presented later in high school. In conclusion, the current study provides preliminary evidence that the SOS program is a promising approach to suicide prevention among middle school students. In addition, our study provides evidence that SOS, implemented by school staff and out of the watchful eye of researchers, may be nearly as effective as more controlled implementations. The current study sacrificed the control favoring internal validity for the realism necessary for external validity, and this type of research is essential to the evidence base for widespread dissemination (Schillinger, 2010). We emphasize that replication of the study in other samples is necessary to evaluate the implications of many of the study’s limitations on our results. However, this study provides initial support for the SOS program’s viability as a suicide prevention program for young adolescents. Given the importance of preventing suicide in young people, expanded studies of SOS’ efficacy among middle school student populations are warranted.
REFERENCES ASELTINE, R. H., JR. (2003). An evaluation of a school-based suicide prevention program. Adolescent & Family Health, 3, 81–88. ASELTINE, R. H., JR., & DEMARTINO, R. (2004). An outcome evaluation of the SOS suicide prevention program. American Journal of Public Health, 94, 446–451. ASELTINE, R. H., JR., JAMES, A., SCHILLING, E. A., & GLANOVSKY, J. (2007). Evaluating the SOS suicide prevention program: A replication and extension. BMC Public Health, 7, 161. BRIDGE, J. A., GOLDSTEIN, T. R., & BRENT, D. A. (2006). Adolescent suicide and
suicidal behavior. Journal of Child Psychology and Psychiatry, 47, 372–394. BRONFENBRENNER, U. (1979). The ecology of human development. Cambridge: Harvard University Press. CAMPBELL, C. L., BROWN, E. J., & OKWARA, L. (2011). Addressing sequelae of trauma and interpersonal violence in military children: A review of the literature and case illustration. Cognitive and Behavioral Practice, 18, 131–143. CARLTON, P. A., & DEANE, F. P. (2000). Impact of attitudes and suicidal ideation on adolescents’ intentions to seek professional
666 psychological help. Journal of Adolescence, 23, 35–45. CARMINES, E. G., & ZELLER, R. A. (1979). Reliability and validity assessment. Beverly Hills, CA: Sage. CHANDRA, A., LARA-CINISOMO, S., JAYCOX, L. H., TANIELIAN, T., BURNS, R. M., RUDER, T., ET AL. (2010). Children on the homefront: The experience of children from military families. Pediatrics, 125, 16–25. CHANDRA, A., LARA-CINISOMO, S., JAYCOX, L. H., TANIELIAN, T., BURNS, R. M., RUDER, T., ET AL. (2011). Views from the homefront: How military youth and spouses are coping with deployment. Santa Monica, CA: Rand Corporation. CHANDRA, A., MARTIN, L. T., HAWKINS, S. A., & RICHARDSON, A. (2010). The impact of parental deployment on child social and emotional functioning: Perspectives of school staff. Journal of Adolescent Health, 46, 218–223. CIGULAROV, K., CHEN, P. Y., THURBER, B. W., & STALLONES, L. (2008). What prevents adolescents from seeking help after a suicide education program? Suicide and Life-Threatening Behavior, 38, 74–86. COHEN, J. (1977). Statistical power analysis for the behavioral sciences (rev. ed.). New York: Academic Press. CUSIMANO, M. D., & SAMEEM, M. (2011). The effectiveness of middle and high school-based suicide prevention programmes for adolescents: A systematic review. Injury Prevention, 17, 43–49. ESPOSITO-SMYTHERS, C., WOLFF, J., LEMMON, K. M., BODZY, M., SWENSON, R. R., & SPIRITO, A. (2010). Military youth and the deployment cycle: Emotional health consequences and recommendations for intervention. Journal of Family Psychology, 25, 497–507. FLAY, B. R., BIGLAN, A., BORUCH, R. F., CASTRO, F. G. L., GOTTFREDSON, D., KELLAM, S., ET AL. (2005). Standards of evidence: Criteria for efficacy, effectiveness and dissemination. Prevention Science, 6, 151–175. GROHOLT, B., EKEBERG, O., WICHSTROM, L., & HALDORSEN, T. (2005). Suicidal and nonsuicidal adolescents: Different factors contribute to self-esteem. Suicide and Life-Threatening Behavior, 35, 525–535. HERRING, R. (1990). Suicide in the middle school: Who said kids will not? Elementary School Guidance & Counseling, 25, 129–137. HUEBNER, A. J., MANCINI, J. A., WILCOX, R. M., GRASS, S. R., & GRASS, G. A. (2007). Parental deployment and youth in military families: Exploring uncertainty and ambiguous loss. Family Relations, 56, 112–122. JACOBS, D. (2013). SOS signs of suicide middle school program: Procedure manual. Cambridge, MA: Screening for Mental Health, Inc.
JOHNSON, S. J., SHERMAN, M. D., HOFFMAN, J. S., JAMES, L. C., JOHNSON, P. L., LOCHMAN, J. E., ET AL. (2007). The psychological needs of U. S. military service members and their families: A preliminary report. Washington, DC: American Psychological Association. JOINER, T. E., JR., CONWELL, Y., FITZPATRICK, K. K., WITTE, T. K., SCHMIDT, N. B., BERLIM, M. T., ET AL. (2005). Four studies on how past and current suicidality relate even when “everything but the kitchen sink” is covaried. Journal of Abnormal Psychology, 114, 291–303. JOINER, T. E., JR., HOLLAR, D., & VAN ORDEN, K. (2006). On buckeyes, gators, super bowl Sunday, and the miracle on ice: ‘Pulling together’ is associated with lower suicide rates. Journal of Social and Clinical Psychology, 25, 179– 195. KANN, L., KINCHEN, S. A., WILLIAMS, B. I., ROSS, J. G., LOWRY, R., GRUNBAUM, J. A., ET AL. (2000). Youth risk behavioral surveillance– United States, 1999. MMWR Surveillance Summaries, 49, 1–32. KESSLER, R. C., & GREENBERG, D. F. (1981). Linear panel analysis: Models of quantitative change. New York: Academic Press. MCINTOSH, J. L., & DRAPEAU, C. W., for the American Association of Suicidology. (2012). U.S.A. suicide: 2010 offical final data. Washington, DC: American Association of Suicidology. Retrieved November 28, 2012, from http:// www.suicidology.org. MICHELMORE, L., & HINDLEY, P. (2012). Help-seeking for suicidal thoughts and self-harm in young people: A systematic review. Suicide and Life-Threatening Behavior, 42, 507–524. NOCK, M. K., HOLMBERG, E. B., PHOTOS, V. I., & MICHEL, B. D. (2007). Self-injurious Thoughts and Behaviors Interview: Development, reliability, and validity in an adolescent sample. Psychological Assessment, 19, 309–317. PEREZ, V. W. (2005). The relationship between seriously considering, planning, and attempting suicide in the youth risk behavior survey. Suicide and Life-Threatening Behavior, 35, 35–49. REARDON, L. E., LEEN-FELDNER, E. W., & HAYWARD, C. (2009). A critical review of the empirical literature on the relation between anxiety and puberty [Review]. Clinical Psychology Review, 29, 1–23. RIBEIRO, J. D., & JOINER, T. E. (2009). The interpersonal-psychological theory of suicidal behavior: Current status and future directions. Journal of Clinical Psychology, 65, 1291– 1299. RICHARDSON, A., CHANDRA, A., MARTIN, L. T., SETODJI, C. M., HALLMARK, B. W., CAMPBELL, N. F., ET AL. (2011). Effects of soldiers’
deployment on children’s academic performance and behavioral health. Santa Monica, CA: Rand Corporation. ROBINSON, J., COX, G., MALONE, A., WILLIAMSON, M., BALDWIN, G., FLETCHER, K., ET AL. (2013). A systematic review of schoolbased interventions aimed at preventing, treating, and responding to suicide-related behavior in young people. Crisis, 34, 164–182. SCHILLING, E. A., ASELTINE, R. H., & JAMES, A. (2013). The SOS suicide prevention program for high school students: Further evidence of efficacy and effectiveness. Manuscript submitted for publication. SCHILLINGER, D. (Ed.). (2010). An introduction to effectiveness, dissemination and implementation research. San Francisco: University of California, Clinical Translational Science Institute Community Engagement Program. Retrieved July 2013 from http://ctsi.ucsf.edu/ files/CE/edi_introguide.pdf. SHAFFER, D., GARLAND, A., VIELAND, V., & UNDERWOOD, M. (1991). The impact of cur-
667 riculum-based suicide prevention programs for teenagers. Journal of the American Academy of Child and Adolescent Psychiatry, 30, 588–596. SPIRITO, A., OVERHOLSER, J., ASHWORTH, S., & MORGAN, J. (1988). Evaluation of a suicide awareness curriculum for high school students. Journal of the American Academy of Child and Adolescent Psychiatry, 27, 705–711. Substance Abuse and Mental Health Services Administration. (2011). SAMHSA’s national registry of evidence-based programs and practices: SOS signs of suicide. Retrieved March 8, 2011, from http://www.nrepp.samhsa.gov/ViewIntervention.aspx?id=2053. WELLS, K. C., & HEILBRON, N. (2012). Family-based cognitive-behavioral treatments for suicidal adolescents and their integration with individual treatment. Cognitive and Behavioral Practice, 19, 301–314. Manuscript Received: March 1, 2013 Revision Accepted: February 6, 2014
Copyright of Suicide & Life-Threatening Behavior is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.