Short-Term Suicide Awareness Curriculum
Abstract
ABSTRACT: The present study attempted to determine the effectiveness of a suicide awareness curriculum and to assess two methods of presenting the information. It was hypothesized that students receiving instruction via lectures and handouts would score significantly higher than students who only received handouts. Dependent variables included the Suicide Intervention Response Inventory (SIRI), the Knowledge of Suicide Test (KOST), the Suicide Prevention Questionnaire (SPQ), and specific suicide-related vignettes. There was a significant main effect for treatment. Univariate analyses demonstrated significant treatment effects for all dependent measures. Two orthogonal contrasts demonstrated significant differences on the KOST and the vignettes, suggesting that the lectures were instrumental in the increased performance on those tasks.
Young persons (aged 15–24) who die this year will probably experience a violent death. Suicides are second only to accidents and more prevalent than homicides (National Center for Health Statistics. 1986). Suicides are increasing more rapidly in this group than in any other group. Rates of suicide in the young have risen over 300% in the past 20 years (Shaffer & Fisher, 1981). Young suicides represented fewer than 5% of all suicides only 20 years ago; however, reported suicides by 15- to 24-year-olds now constitute 17.2% of reported suicides. Approximately 13.7 young people commit suicide each day, accounting for 5,026 per year (Hopkins, 1983; National Center for Health Statistics. 1986).
Statistics on completed suicides have escalated in recent years; however, the problem is further magnified when the young people who plan or attempt suicide are considered. An excellent initial analysis of the number of high school students who experience suicidal ideation, attempt suicide, and plan suicide (Smith & Crawford, 1986) has highlighted the inconsistent estimates of the ratio of attempters to those who complete the act. Smith and Crawford correctly indicate that one group of studies estimating the numbers of attempters may understate the problem, because these studies have based their data only on young people who sought help. The wide range of estimates varies from 200:1 to 50:1 (Hawton, 1982; Jacobziner, 1965; Peck, 1982; Rosenkrantz, 1978). The studies using surveys that directly questioned large numbers of individuals, rather than taking information only from those who sought help, present a different picture. The rates of attempters were 13% of all students in a northern California sample (Ross, 1985), 10% in a New York sample (Klagsbrun, 1981), 15% of college students in a Detroit sample (Mishara, Baker, & Mishara, 1976), and 10.5% in the Midwest sample (adjusted for sex = 8.5%) of Smith and Crawford (1986). The disagreement appears to be a product of two dimensions of this research area: (1) the differential reporting of attempted and completed suicides and (2) the absence of universally recognized definitions of such terms as a suicide “attempt.”
Some efforts have been made to determine the level of suicidal ideation among students. Approximately 2% of all high school students in one study admitted to either thinking of committing suicide or being concerned with suicidal thoughts (Sartore, 1976). Within a college population (Mishara et al., 1976), it was reported that 65% of the students were able to describe the means they would use if they were to attempt suicide. Peck (1982) estimated that the percentage of students experiencing suicidal ideation, gestures, or attempts is as high as 10%. The more complete and accurate data from the Smith and Crawford (1986) study indicate that approximately 25.2% of all students surveyed were distressed to the point of making a plan or an attempt. Another study (Kinkle, Bailey, & Josef, in press), which assessed high school students, indicated that 36.6% had very serious thoughts about suicide; this figure is similar to that for the Smith and Crawford “planners.” It seems apparent that suicidal thoughts, by whatever means assessed, are highly prevalent among high school students and young adults. These data should direct researchers to differentially assess personality and vulnerability stressor variables for both groups.
Working with individuals at risk is essential. However, efforts to prevent the suicidal deterioration process at an earlier stage are necessary. Efforts to untangle the constellation of suicide-predicting variables have been previously summarized (Ross, 1985). A great deal is known about risk factors and warning signs. This knowledge is not as effective as it might be unless it leads to prevention.
The nation's schools have large numbers of nondisturbed individuals, all of whom may be considered as potentially able to prevent suicides if they are sufficiently aware of what to do. The majority of suicidal young adults attempt some communication about their impending death (Ross, 1980, 1985). Education of individuals within the school system regarding warning signs and prevention techniques is logical, and may be imperative.
Some steps have been taken to educate those who could be termed “gatekeepers” within the school system. Health care professionals (Neimeyer & MacInnes, 1981), teachers (McKenry, Tishler, & Christman, 1980), counselors (McBrien, 1983; Morgan, 1981; Wellman, 1984), and school nurses (Hart & Keidel, 1979) have all been the focus of increased educational efforts. However, there appears to be some reluctance to educate those who may need it the most—the students. Research shows that suicidal adolescents and young adults are more likely to discuss their problems with a peer than to talk to a parent, teacher, or counselor (Greuling & DeBlassie, 1980; Ross, 1980, 1985). Therefore, high school and college students should receive suicide prevention education. Some studies indicate that college students have higher suicide rates than the 15- to 24-year-old group as a whole (Hendin, 1982; Mishara, 1982; Ross, 1969). However, there is some question whether suicide among the noncollege group may be just as prevalent (Schwartz, 1980; Schwartz & Reifler, 1980).
Student curricula on suicide have been heralded as the most promising means of primary prevention (Berg, 1972; Jacobziner, 1965, Klagsbrun, 1981; O'Roark, 1982; Ross, 1980, 1985; Sartore, 1976). Following this lead, the Florida State Legislature passed a bill mandating the teaching of “positive mental health” (interpreted to include suicide awareness) to all Florida high school students as a small part (a few lectures during 1 week) of their semester-long Life Management Skills course (State of Florida. 1985). The teachers are required to teach suicide awareness, and have been provided instructional materials that were neither scientifically derived nor academically assessed (State of Florida. 1985). Rigorous assessment and evaluation of suicide awareness curricula is imperative.
The main objective of the present study was to determine whether or not knowledge instrumental in suicide awareness and potential intervention could be imparted to college students in a short time span through the use of an instructional package specifically designed for this purpose. The secondary objective was to evaluate two methods of presenting the suicide awareness material. It was hypothesized that participants who received both lectures and handouts for individual study (the “lecture-plus-individual-study” condition) and those who received the handouts only (the “individual-study” condition) would score significantly higher than those in a control group. It was further postulated that those in the lecture-plus-individual study condition would score significantly higher than those in the individual-study condition on all dependent measures.
Method
Participants
Participants were 73 college undergraduates enrolled in a course at Florida State University. Subjects were screened for prior crisis intervention training and potential suicide risk (two individuals were excluded). Consent to participate was granted by 100% of the class (ages 18–22).
Dependent Measures
Suicide Intervention Response Scale. The Suicide Intervention Response Scale (SIRI; Neimeyer & MacInnes, 1981) was developed to assess paraprofessional competence following suicide intervention training. It includes 25 items, each of which offers an initial “client” remark, followed by two possible “helper” replies. According to crisis intervention theory, one response is facultative and the other detrimental. Item content was derived from a variety of sources, including case reports and anecdotal accounts of professional therapists. An effort was made to devise items incorporating common problems encountered in replies to suicidal callers on hotlines. The score on the SIRI is derived by the number of correct responses chosen by the respondent. Original reliability for the SIRI (Kuder—Richardson 20) was .845 (n = 164); test-retest reliability was .864 over a 3-month interval (Neimeyer & MacInnes, 1981). The pretest reliability (internal consistency) for the present study (Kuder-Richardson 20) was .675 with a posttest reliability of .756.
Knowledge of Suicide Test and Suicide Prevention Questionnaire. The Knowledge of Suicide Test (KOST) and Suicide Prevention Questionnaire (SPQ) were developed after a review of existing instruments (deemed not entirely appropriate) to measure suicide awareness knowledge (McIntosh et al., 1985).
The KOST consists of 30 multiple-choice questions. The SPQ consists of 10 questions to be answered on an 11-point Likert scale; subjects respond to each question with a number from −5 (“strongly disagree”) to +5 (“strongly agree”), and the test is scored by summing all answers algebraically. Pretest internal consistency for the KOST was .530; post-test KOST reliability was .721; pretest SPQ was .741; and the posttest SPQ coefficient was .819.
Suicide-Related Vignettes. A series of five suicide-related vignettes labeled “Application of Course Materials” was presented. These included (1) what to ask a friend to determine whether he or she is suicidal; (2) what to say and/or do if a friend says he or she is thinking of suicide and makes you promise not to tell anyone; (3) what to say and/or do if you suspect a friend may be suicidal; (4) what to say and/or do if you notice a friend is depressed, has been losing weight, and has begun failing in school; and (5) what to say and/or do if a friend says he or she is considering suicide but currently has no plan or method. Each person received a single score for each vignette based upon predetermined criteria. Reliability was determined by having an expert score 15% of randomly determined vignettes previously scored by trained undergraduates. The percentage of agreement was established by dividing the number of answers in agreement by the number of answers in disagreement plus the number of answers in agreement (mean reliability = .89; range, .83–.94). Content validity was ascertained by consultation with two PhD clinical psychologists with expertise in the area of suicide (93% agreement).
Subjects' Assignment to Groups
Participants were randomly assigned to one of three treatment conditions, which were controlled for gender: an individual-study group (n = 16 females and 7 males); a control group; and an individual-study-plus-lecture group (n = 17 females and 8 males). Students were informed that the class would be divided into thirds, with each division receiving information about issues concerning adolescents. Students were motivated by making 25% of their semester grade contingent on this portion of the course.
Students in all conditions were pretested using the SIRI, KOST, and SPQ. Following the 2-week experimental period, all subjects were post-tested using the pretest instruments and the five suicide-related vignettes, which were paired with five non-suicide-related vignettes about adolescent issues (alcohol use, drug abuse, conflict with parents, religion, and sexual orientation). The additional vignettes were used in an attempt to overcome an exclusively suicide-oriented mode of thinking and to produce a more authentic representation of the participants' performance. The vignettes were administered two per day, for 5 consecutive class days (1.5 weeks) after the last experimental lecture. The SIRI, KOST, and SPQ were administered together in posttesting the students after the completion of the vignette series (4 weeks after the pretest).
Experimental Design
Experimental sessions were administered according to the schedule depicted in Table 1.
| Day | Condition 1 Individual study only | Condition 2: Control | Condition 3: Individual study plus lectures | |
|---|---|---|---|---|
| Week 1 | F | Pretest | Pretest | Pretest |
| Week 2 | M | Lecture (nonsucidal) | ||
| W | Suicide lecture | |||
| F | Lecture (nonsuicidal) | |||
| Week 3 | M | Suicide lecture | ||
| W | Lecture (nonsuicidal) | |||
| F | Suicide lecture | |||
| Week 4 | M | One suicide vignette plus one nonsuicide vignette for all | ||
| W | One suicide vignette plus one nonsuicide vignette for all | |||
| F | One suicide vignette plus one nonsuicide vignette for all | |||
| Week 5 | M | One suicide vignette plus one nonsuicide vignette for all | ||
| W | One suicide vignette plus one nonsuicide vignette for all | |||
| F | Posttest | Posttest | Posttest | |
Experimental Conditions
Condition 1: Individual Study. Participants in the first condition received a packet of information on the pretest date, which they were to study for a test to be given 4 weeks from the date they were given the material. They were instructed not to come to class during the 2-week period immediately following the pretest until they were given the posttest.
Condition 2: Control Group. Participants in the second condition were excluded from receiving the suicide intervention handouts on the pretest date. They were told which days they were required to attend class, and they received non-suicide-related lectures (socialization problems) by the same instructor who taught the suicide-related lectures, followed by the posttest (including a test over the lecture material they had received.
Condition 3: Individual Study plus Lectures. Participants in the third condition received a packet of information on the pretest date, which they were to study for a test to be given 4 weeks from the pretest date, and they were informed of the schedule of attendance required. These participants received lectures given by the course professor, which covered the individual-study handouts and were supplemented by the asking and answering of questions. Lectures provided examples of young adult suicide intervention from clinical practice.
Results
The results indicated a statistically significant main effect for treatment after multiple dependent measures were controlled for, by means of a multivariate analysis of covariance (MANCOVA), F (6, 130) = 19.20, p < .001. The correlation coefficient for the first canonical variate equaled .716, which indicated a high degree of commonality among the three variables; Wilk's lambda equaled .2811, indicating the percentage of unexplained variance among the dependent variables. The control and experimental pretest scores were not significantly different.
As expected, there was a moderate correlation between each pre-and posttest measure (standardized beta coefficient = .51065 for the KOST, .4464 for the SIRI, and .4620 for the SPQ).
Univariate analyses of variance (ANOVAs) indicate significant treatment effects for all three dependent variables. For the KOST, F (3, 67) = 8.89, p < .001; for the SIRI, F (3, 67) = 8.78; p < .001; for the SPQ, F (3, 67) = 6.49, p < .001.
Following the ANOVAs, two orthogonal contrasts were performed for each dependent variable. In the first contrast, the means for the individual-study condition were tested against the means for the individual-study-plus-lectures condition. The results of the first contrast were significant only for the KOST: The individual-study mean was 26.30 and the individual-study-plus-lectures mean was 28.04, t (46) = 2.42, p < .02. For the SIRI, the individual-study mean was 23.61 and the individual-study-plus-lectures mean was 24.56, t (46) = 1.14, p > .05; for the SPQ, the individual-study mean was 80.78 and the individual-study-plus-lecture mean was 83.28, t (46) = .650, n.s. There were no significant differences between the individual-study and individual-study-plus-lecture conditions on the SIRI or the SPQ.
The control means were tested against the average of the individual-study and individual-study-plus-lecture means for the second contrast. The results of the second contrast were significant for all three of the dependent measures. On the KOST, the control mean was 20.76 and the average of the other two groups' means was 27.17, t (71) = 12.52, p < .001. For the SIRI, the control mean was 21.60 and the average of the other two groups' means was 24.09, t (71) = 3.84, p < .001. On the SPQ, the control mean was 65.84 and the average of the other two groups' means was 82.03, t (71) = 7.31, p < .001.
A one-way ANOVA was employed to assess treatment effects on the final dependent measure—the suicide-related vignettes. The main effect for treatment was significant, F (2, 70) = 25.848, p < .0001. Orthogonal contrasts demonstrated significant differences between each of the conditions. A comparison of means indicated that the control mean was significantly different from the average of the individual-study and individual-study-plus-lecture means, t (2, 70) = 5.072, p < .001. In addition, the individual-study and the individual-study-plus-lectures means were significantly different, t (2, 70) = 4.971, p < .001.
Discussion
The Present Results
This study may appear merely to substantiate that students given instruction gained more knowledge than students deprived of instruction. And, at the most basic level, that perception is accurate. However, it is not accurate to assume that this demonstration is insignificant. In developing suicide prevention curricula, one must begin at the beginning; that is, one must demonstrate that a method of imparting knowledge has done what it set out to do. Researchers can assume nothing. Now that it has been demonstrated that the present two educational packages had at least some desired effects, we can begin to refine the dependent variables further and develop more sensitive measures. This study does not necessarily demonstrate what students may do to intervene in a possible suicide; however, the study has defined and attempted to assess what young adults think about suicide intervention and how they think they would respond (suicide-related vignettes).
As expected, the treatment differentially affected the dependent variables. The results of the MANCOVA and univariate analyses demonstrated the significant main effects of treatment for each dependent variable. The experimental groups scored higher than the control group. Participants in the experimental conditions were better able to choose correct responses regarding suicidal warning signs and intervention strategies, and they were better equipped to choose effective reflective listening responses. In addition, those who received individual study only or lectures plus individual-study material experienced a positive change in reactions regarding their competence in suicidal situations and their reactions to the teaching of suicide intervention. Those in the experimental conditions were better able to respond, in essay format, to questions regarding their intended actions in a variety of potentially suicidal situations. In addition, students in the lectures-plus-individual-study condition outperformed those in the individual-study condition on the KOST and suicide-related vignettes. Inasmuch as the fundamental purpose of the present study was to educate students to recognize and respond correctly to a potentially suicidal situation, the KOST and vignette performance of those in the experimental conditions is encouraging.
The suicide-related lectures in the lecture-plus-individual-study condition were supplemented with examples from the professor's experiences as a clinician. These examples were not transcribed for the individual-study condition. This augmenting of written material, plus information garnered through class questions and answers—characteristic elements of a lecture format—may have been responsible for improved performance. The applied examples that were presented in the class lectures may have made the material more relevant.
The undifferentiated performance of the two experimental groups on the SIRI and SPQ may also have been influenced by the additional information provided in the lecture-plus-individual-study condition. The results suggest, however, that choosing appropriate reflective listening responses and developing more positive reactions toward potential help were not influenced by the content and/or style of the present lecture format. It would appear, then, that hearing the applied clinical examples of suicide intervention did not significantly aid students in choosing reflective listening responses. The absence of mean differences between the two experimental conditions was especially apparent on the SPQ, where variation between group means was determined to be at chance levels. The SIRI and SPQ measures may have produced no added effect of the suicide-related lectures because they may be measuring nondiscrete factors; for example, the SPQ may measure perceived self-confidence in potentially suicidal situations. Instruments designed for future researchers may have to be factor-analyzed.
Future Research
Dependent Variables. The uniqueness of the dependent measures for this study (three of the four measures were specifically designed) makes it necessary to do follow-up experiments that assess reliability and examine additional reliability concerns. In addition to the potential problems with the reliability data, the validity of the KOST, SPQ, and suicide-related vignettes requires further exploration. The present study represents a first step in establishing the construct validity. The fact that nearly 72% of the variance among the SIRI, KOST, and SPQ was accounted for by a similar source implies a high degree of correlation among the dependent measures. There is some evidence that the measures are drawing variance from similar constructs.
Experimental Design. The material presented in the lecture-plus-individual-study condition was, by design, not strictly comparable to that presented in the individual-study condition. Therefore, interpretation of the results must be viewed in light of the additional applied examples.
A study could be done that provided the applied clinical examples in transcribed form for the individual-study condition. If this were done, the only difference between the individual-study and lecture-plus-individual-study conditions would be the time spent in lecture format and class questions and answers. Of course, the present experiment deliberately added relevant material, which appears to have increased the students' competence to deal abstractly with the vignettes. The next logical study is to provide students with simulated suicidal “real-life” vignettes, to assess whether abstract concepts can be applied in a more realistic setting (this study is presently in progress).
Generalizability. The participants in the present research were primarily white females, in their junior year of college, between 21 and 24 years of age (there was only one nonwhite). Therefore, the generalizability of the present findings is limited to persons with characteristics similar to those of the present sample.
In particular, researchers should not assume, because the sample consisted of students, that the results would generalize to other populations currently being evaluated for suicide awareness (e.g., elementary and/or high school students). The current sample demonstrated a high degree of prior knowledge about suicide awareness and intervention (some ceiling effects). The procedure may yield different (perhaps more striking) results when utilized with relatively less knowledgeable, younger participants. The present instruments and procedures hold promise for use with other populations; however, the extent of their utility has yet to be demonstrated by experimental and statistical analyses.
The field of suicide intervention education is at a critical stage in its development. Several important issues are at the forefront of research, including potential stress factors in causation; evaluation of more valid, reliable, and cost-effective assessment instruments; and determination of the critical elements in the curricula. The ultimate question is how best to educate students to intercede in the unnecessary and preventable deaths of others.
