- Email: [email protected]
Self-reported abuse history and adolescent problem behaviors. I. antisocial and suicidal behaviors
JOURNAL OF ADOLESCENT HEALTH 1999;24:163–172
ORIGINAL ARTICLE
Self-Reported Abuse History and Adolescent Problem Behaviors. I. Antisocial and Suicidal Behaviors LILLIAN SOUTHWICK BENSLEY, Ph.D., JULIET VAN EENWYK, Ph.D., SUSAN J. SPIEKER, Ph.D., AND JUDY SCHODER, M.N.
Purpose: To examine the associations of self-reported abuse and sexual molestation with self-reported antisocial behavior and suicidal ideation/behavior in a general population of adolescents. Methods: We used a stratified cluster sampling procedure with replacement to sample 4790 students in Washington State public schools in Grades 8, 10, and 12. Students were asked in a self-administered questionnaire whether they had ever been abused by an adult and whether they had ever been sexually molested. They were also asked about antisocial behavior and suicidal ideation and behavior in the past year. We conducted polytomous logistic regressions, controlling for gender and grade, using Software for the Statistical Analysis of Correlated Data (SUDAAN). Results: Reported abuse history was associated with antisocial behavior and with suicidal ideation and behavior. The associations were stronger for abuse and molestation than for nonsexual abuse or molestation alone, and stronger at higher levels of severity (e.g., suicide attempts vs. suicidal thoughts). For example, adjusted odds ratios and 95% confidence intervals (in parentheses) for abuse and molestation were 4.4 (3.1– 6.2) for suicidal thoughts, 6.8 (4.4 –10.4) for suicide plan, 12.0 (7.9 –18.4) for noninjurious suicide attempt, and 47.1 (23.3–95.3) for injurious suicide attempt. For abuse alone, these figures were 2.3 (1.7–3.2), 3.1 (2.1– 4.6), 5.1 (3.3–7.8), and 11.8 (4.4 –31.9), respectively. Conclusions: Efforts to reduce antisocial behavior and suicidal ideation/behavior in adolescence, particularly early or severe manifestations of the behaviors, should consider the possible role of a history of maltreatment, From the Washington State Department of Health, Olympia, and the University of Washington, Seattle. Address reprint requests to: Lillian Bensley, Ph.D., Washington State Department of Health Office of Epidemiology, 1102 SE Quince St., Olympia, WA 98504-7812. Manuscript accepted May 12, 1998.
especially the possibility of sexual abuse. © Society for Adolescent Medicine, 1999 KEY WORDS: Adolescence Child abuse Physical abuse Sexual abuse Antisocial behavior Suicide Survey Gender differences
Adolescent depression, anxiety, and delinquent behavior tend to co-occur (1) and a history of being maltreated has been reported as a risk factor for each. However, the degree of risk associated with different types of maltreatment and the extent to which findings from studies of specialized groups generalize to adolescents in the general population are not known. Large-scale surveys, which rely on self-report of both abuse history and long-term sequelae, are important to describe the prevalence of abuse and its sequelae in populations such as school students so that the extent of the need for prevention and intervention services can be documented. A 17-year longitudinal community study using self-reports of abuse history linked abuse to a variety of negative outcomes in adolescence and adulthood, including depressive symptomatology, psychiatric disorder, suicidal ideation, and suicide attempts (2), and smaller studies of clinical samples have found similar associations. However, these studies have included relatively small numbers of maltreated individuals, restricting their ability to examine differ-
© Society for Adolescent Medicine, 1999 Published by Elsevier Science Inc., 655 Avenue of the Americas, New York, NY 10010
1054-139X/99/$–see front matter PII S1054-139X(98)00111-6
164
SOUTHWICK BENSLEY ET AL.
ent types of maltreatment. Brent et al. (3,4) conducted a case-control study of adolescents who completed suicide. These authors identified mood disorders, substance abuse, and conduct disorder as risk factors and found that past suicidal ideation with a plan was as strongly associated with completed suicide as a past attempt (3). The association between a history of maltreatment and suicide in the Brent et al. sample did not achieve statistical significance; however, this was likely owing to the small sample sizes (8% of the 67 suicide completers reported physical abuse compared to 0% of the community controls) (4). Research on delinquency has also shown relatively consistent associations with a history of maltreatment. Ireland and Widom (5), Widom and Ashley (6), and Widom and Kuhns (7) used a prospective cohort design in which validated cases of child sexual abuse, physical abuse, and neglect were matched by age, sex, race, and socioeconomic status with controls for whom there were no records of child abuse or neglect. Childhood sexual abuse, physical abuse, and/or neglect were predictors of later adolescent delinquent and adult criminal behavior; arrest for prostitution; adult arrest for alcohol- and/or drug-related offenses; and, for males, violent sex crimes. However, this study included only cases of childhood abuse and neglect that came to the attention of the authorities and official criminal histories, so that only relatively severe forms of both maltreatment and antisocial behavior were included. A few large-scale student surveys have included questions on physical and sexual abuse. For example, Nelson et al. (8) added questions to Oregon’s 1993 Youth Risk Behavior Survey. Students who reported that they had been sexually abused within the past year were more likely to have a low self-image, to carry weapons, and to have been in physical fights, have seriously considered suicide in the past year, and have used alcohol or drugs in the past month compared with students who reported never having been sexually abused. This study confirms the findings of others that sexual abuse and adolescent high-risk behaviors are highly correlated. However, there was no way to determine if the abuse in the past year occurred in a peer dating situation or had occurred with someone older. In addition, the sole focus on sexual abuse may be misleading. Maltreatment rarely exists in pure forms, as more than one type of abuse is usually associated with a given incident (9). By reporting only on sexual abuse (and further restricting that to abuse in the past year), Nelson et al. could not conclude which outcomes
JOURNAL OF ADOLESCENT HEALTH Vol. 24, No. 3
were associated with sexual abuse and which were associated with other forms of abuse that may have coexisted but were not included in their analyses. As there is increasing evidence that different types of maltreatment (physical abuse and sexual abuse) are associated with different outcomes (10), it is important to ask respondents in a population survey to differentiate types of abuse. Some studies have used student surveys to examine associations among aggressive and violent behavior and other health risk behaviors (11,12). These studies have not examined family characteristics, such as child abuse, associated with these patterns. Other longitudinal studies found that family characteristics such as low support, cohesion, and adaptability predicted future suicidal ideation and behavior in adolescents (13,14), but did not measure maltreatment apart from other family factors. We examined the health risk behaviors associated with a history of physical abuse, sexual abuse, and combined physical abuse and sexual abuse in a general population sample of adolescents in Washington public schools. The sample size was large (4790 individuals), so that we were able to assess the degree of risk associated with different types of maltreatment. In this report, we present detailed methodological information about the study and the associations of abuse history with suicidal ideation/ behavior and antisocial behavior. In an accompanying report (15), we present evidence of validity of the abuse history self-reports and the associations of abuse history with alcohol and drug use and early initiation of substance use. [For related evidence about the validity of self-reports of abuse, see (16,17)]. The specific questions addressed by this study were: 1. How prevalent are self-reported histories of physical abuse; sexual abuse; combined physical abuse and sexual abuse; antisocial behavior; suicidal thoughts and behavior; and combined antisocial behavior and suicidal thoughts and behavior in adolescents from the State of Washington? 2. What are the associations between self-reported history of physical abuse; sexual abuse; and combined physical abuse and sexual abuse on the one hand and antisocial behavior; suicidal thoughts and behavior; and combined antisocial behavior and suicidal thoughts and behavior on the other hand in adolescence? 3. Are there age or gender differences in these associations that might be helpful in targeting interventions?
March 1999
Methods Sampling We sampled students in Washington State public schools in Grades 8, 10, and 12 using a stratified cluster sampling procedure with replacement. (Sixth graders were also surveyed but were not asked about abuse history and so are not included in this report.) Schools were the primary sampling unit. Stratification was by geographic regions (East, Southwest, Puget Sound, and Northwest) and school size (large or small). To boost the representation of racial/ ethnic minorities in the state sample, a sampling with probability proportionate to size (PPS) procedure was used in which schools with higher minority enrollments received higher probabilities of selection. Replacement schools were selected using the above procedures and with the additional criterion that they were similar in level of urbanicity (metropolitan core, small city, suburban, or rural). In other words, when a given school refused to participate, another school from that region of the same size, urbanicity, and minority enrollment was recruited to take its place. Within the participating schools, all students at the target grades were asked to participate. A weighting factor was applied to each student record to adjust for the varying probabilities of selection in an effort to maximize the representativeness of the results to the population of public school students in Washington State. Students in 44 middle and high schools participated, and the school response rate was 22%. Comparisons between participating and nonparticipating schools did not reveal differences for a number of parameters including percentage of children participating in the free or reduced lunch program; graduation rates; percentage of high school dropouts and students whose status was unknown; percentage of eighth graders with computers at home; percentage of eighth graders whose mothers did not have a high school diploma; and 11th-grade test scores. Of the students present on the day of the survey, the student response rate was 96%. Of the completed surveys (including sixth graders), approximately 8% were discarded because of dishonesty or inconsistent responses. (More detail is provided below in the section on quality assurance.) These procedures led to a final sample of 4790 completed questionnaires. Questionnaire and Data Collection The questionnaires addressed a broad range of student health risk behaviors including physical fitness,
ABUSE AND ADOLESCENT PROBLEM BEHAVIORS
165
unintentional and intentional injury behaviors, alcohol, tobacco and other drug use, human immunodeficiency virus/acquired immunodeficiency syndrome education, and risk and protective factors for adolescent health risk behaviors. The questionnaire was administered in the classroom during a regular class period. Completing the questionnaire took approximately 50 min. Instructions stated that the questionnaire was voluntary and that responses were anonymous and confidential. Students recorded their responses directly on the questionnaire, which was in scannable format. All students were provided with an 800 number to call for referral if they wished to talk with someone about issues raised by the survey. Parental consent procedures included mailings to the parents providing information about the content and reasons for the survey. Parents were given the opportunity to review the entire survey at the school and to notify the school if they did not want their children to participate. The Washington State Department of Social and Health Services/Department of Health Human Research Review Board approved the research procedures. Coding Abuse history. Students were asked two questions about their abuse histories: “Have you ever been abused or mistreated by an adult?” and “Has anyone ever touched you in a sexual place, or made you touch them, when you did not want them to?” Both questions were coded “yes”/“no.” To create a summary measure of abuse history, we coded students who answered “no” to both questions as not abused, “yes” to the first and “no” to the second as abused, “yes” to the second and “no” to the first as molested, and “yes” to both as abused and molested. Evidence of the predictive validity of this measure is provided in the accompanying report (15). Antisocial behavior. The measure of antisocial behavior was constructed from five questions measuring how many times in the past year the respondent had carried a handgun, sold illegal drugs, stolen or tried to steal a motor vehicle such as a car or motorcycle, been arrested, or taken a handgun to school. The measure of antisocial behavior indicated whether the respondent reported “none”, “one”, or “two or more” different behaviors. This definition emphasizes the degree of diversity of antisocial behavior over the number of times a particular behavior was committed.
166
SOUTHWICK BENSLEY ET AL.
Suicidal ideation and behavior. The measures of suicidal ideation/behavior were identical to those used on the Youth Risk Behavior Survey (18,19). Four questions were: “During the past 12 months, have you ever seriously thought about attempting suicide?” “During the past 12 months, did you make a specific plan about how you would attempt suicide?” “During the past 12 months, how many times did you actually make a suicide attempt?” and “If you attempted suicide during the past 12 months, did that attempt result in an injury, poisoning, or overdose that had to be treated by a doctor or a nurse?” Students were coded as the most serious category (thoughts, plan, noninjurious attempt, or injurious attempt) that they reported.
Data Analysis All analyses were conducted using Software for the Statistical Analysis of Correlated Data (SUDAAN) (20). The primary sampling unit was school. Observations were weighted by the school characteristics (region, size, and minority concentration) that were used in sampling. Major analyses. Major outcome measures were antisocial behavior, which had three levels of severity (“none”, “one”, or “two or more” different antisocial behaviors) and suicidal ideation/behavior, which had five levels of severity (“none”, “thoughts”, “plans”, “noninjurious attempts”, and “injurious attempts”). We compared each other level of severity to the group reporting none of that type of behavior (e.g., for each of the analyses of suicidal ideation/behavior, the referent group was those students who reported no suicidal thoughts). We also compared respondents who reported a combination of antisocial and suicidal behavior (at least one antisocial behavior and at least one suicidal ideation/behavior) to those who reported neither. Preliminary analyses using polytomous logistic regression models (21) tested whether significant Grade 3 Abuse history or Gender 3 Abuse history interactions existed across the levels of severity of each major outcome measure. (The interaction test of suicidal ideation/behavior did not include one level of severity that had low-prevalence, injurious suicide attempts, owing to empty cells.) An additional dichotomous logistic regression analysis tested the same possible interactions on the combined antisocial/suicidal outcome measure. We did not identify any interactions that were statistically significant (all
JOURNAL OF ADOLESCENT HEALTH Vol. 24, No. 3
interaction p’s . .05), and so we used a main-effects model for the outcomes in this report. For each of the major outcome measures (the three-level measure of antisocial behavior and the five-level measure of suicidal ideation/behavior) we conducted a polytomous logistic regression in which the predictor variables were the four-level abuse history variable (“none”, “abused”, “molested”, or “abused and molested”), gender (male or female), and grade (8, 10, or 12). From these logistic regressions, we calculated odds ratios (ORs) and 95% confidence intervals (CIs) for each predictor variable, adjusting for levels of the other variables. We also conducted an identical dichotomous logistic regression analysis on the combined antisocial/suicidal outcome measure. Quality assurance. In a section on drug use, students were asked whether they had ever used derbisol (a fictitious drug) and whether they had answered the questions honestly, to identify students who were providing false responses. Students who admitted answering dishonestly or somewhat honestly, claimed use of a fictitious drug, and/or responded inconsistently to three or more pairs of related items (e.g., claimed 30-day use of a substance on one item and no use in lifetime on another item) were omitted from all analyses (8%).
Results Sample Characteristics The sample included 4790 students in Washington State public schools, Grades 8, 10, and 12. Sample characteristics are presented in Table 1. The sample was approximately evenly divided between males and females and was predominately white. Differing sample sizes by grade levels reflected the fact that there were fewer total students in Grade 12 compared to the other grades. This was owing in part to differences in cohort sizes and in part to dropouts between Grades 10 and 12. The total number does not include 320 students who had missing data on gender or abuse history and so were omitted from all analyses.
Abuse History The results of the abuse history questions and weighted percentages are presented in Table 1. Overall, 73.9% reported no maltreatment history, 11.4% reported being abused, 5.9% reported being mo-
March 1999
ABUSE AND ADOLESCENT PROBLEM BEHAVIORS
Table 1. Sample Characteristics Characteristic Gender Female Male Grade 8 10 12 Race/ethnicity White, not Hispanic Black, not Hispanic Hispanic Asian/Pacific Islander Native American Other/unknown Abuse history Abuse 1 sexual molestation Molestation Abuse None Antisocial behaviors in past year None One Two or more Didn’t answer question Suicidal ideation/behaviors in past year None Suicidal thoughts Suicide plan Noninjurious suicide attempt Injurious suicide attempt Didn’t answer question
n
Weighted %
2518 2272
52.1 47.9
1960 1661 1169
40.8 34.7 24.6
3617 88 479 269 123 214
75.0 2.4 9.1 5.9 3.2 4.3
441 279 533 3537
8.8 5.9 11.4 73.9
3852 567 328 43
80.3 11.7 7.0 0.9
3891 364 185 245 63 42
81.1 7.4 4.0 5.3 1.4 0.8
167
1. Of the sample, 11.7% reported one antisocial behavior and 7.0% reported two or more different antisocial behaviors in the past year. As noted earlier, for suicidal ideation/behavior, respondents were categorized into the most serious category of behavior they reported. The most seriously injurious suicidal ideation/behaviors reported by students were as follows: suicidal thoughts 7.4%, suicide plan 4.0%, noninjurious suicide attempt 5.3%, and injurious suicide attempt 1.4%. However, as might be expected, most people with a suicide attempt or a suicide plan also reported suicidal thoughts. If this overlap is included, then the total numbers and weighted percentages of students reporting suicidal thoughts are 852 (18.0%), suicide plan 420 (9.2%), and suicide attempt 308 (6.7%). A total of 264 individuals (5.7%) reported both suicidal ideation or behavior and antisocial behavior. Abuse History and Antisocial Behavior in the Past Year
lested, and 8.8% reported being abused and molested. Abuse histories differed significantly for boys and girls (x2 5 29.91; p , .0001). Of the girls, 65.6% reported no maltreatment history, 10.7% reported being abused, 9.2% reported being molested, and 14.6% reported being abused and molested. Of the boys, 82.9% reported no maltreatment history, 12.19% reported being abused, 2.4% reported being molested, and 2.6% reported being abused and molested.
Girls were less than one-third as likely as boys to report one type of antisocial behavior and one fifth as likely to report two or more different antisocial behaviors in the past year. We did not identify significant associations between grade and antisocial behavior. Most important, abuse history was associated with antisocial behavior, and the associations were stronger for combined abuse and molestation than for nonsexual abuse or molestation alone, and stronger for more severe antisocial behavior (two or more different behaviors) than for less severe manifestations (one type of behavior). For example, those students who reported combined abuse and molestation had more than a threefold risk of reporting one antisocial behavior and more than a sevenfold risk of reporting two or more antisocial behaviors compared to students who reported no abuse. Those students who reported abuse alone had a doubled risk of reporting one antisocial behavior and a fourfold risk of reporting two or more antisocial behaviors compared to students who reported no abuse. Table 2 shows numbers and weighted percentages of students reporting antisocial behavior by grade, gender, and abuse history, and Table 3 shows adjusted ORs and 95% CIs.
Antisocial Behavior and Suicidal Ideation/Behavior
Abuse History and Suicidal Behavior in the Past Year
Levels of the target behaviors and weighted percentages across the entire sample are presented in Table
Girls were 1.5 times more likely than boys to report suicidal thoughts and twice as likely to report non-
The overall n was 4790. Percentages may not sum to 100% because of small amounts of missing data. An additional 320 students for whom gender or abuse history were coded as missing were not included in any of the analyses. Antisocial behaviors included carried handgun, sold illegal drugs, stole car, was arrested, or took handgun to school.
168
SOUTHWICK BENSLEY ET AL.
JOURNAL OF ADOLESCENT HEALTH Vol. 24, No. 3
Table 2. Numbers and Weighted Percentages Reporting Antisocial Behavior* in Past Year, by Gender, Grade, and Abuse History
Abuse history Abuse 1 molestation Molestation Abuse None Gender Female Male Grade 8 10 12 Total
None
One Antisocial Behavior
Two or More Antisocial Behaviors
n (Weighted %)
n (Weighted %)
n (Weighted %)
313 (69.7%) 213 (78.4%) 366 (67.9%) 2960 (83.7%)
72 (17.5%) 39 (13.0%) 81 (15.2%) 375 (10.4%)
53 (12.5%) 24 (7.5%) 81 (15.4%) 170 (5.1%)
2227 (88.1%) 1625 (71.9%)
186 (7.6%) 381 (16.1%)
89 (3.5%) 239 (10.8%)
1594 (81.2%) 1332 (79.7%) 926 (79.8%) 3852 (80.3%)
223 (11.8%) 193 (11.6%) 151 (13.1%) 567 (11.7%)
125 (6.0%) 120 (8.5%) 83 (6.6%) 328 (7.0%)
* Antisocial behaviors include carried handgun, sold illegal drugs, stole car, was arrested, or took handgun to school. † Weighted percentages do not add to exactly 100% because of small amounts of missing data. The total n is 4790.
injurious suicide attempts. Risk of having a suicide plan or an injurious suicide attempt was not significantly related to gender. Risk of a noninjurious suicide attempt was highest in 8th grade, followed by 10th and then 12th grade. Most important, abuse history was associated with suicidal ideation/behavior, and the associations were stronger for combined abuse and molestation than for nonsexual abuse or molestation alone, and stronger for more severe Table 3. Adjusted Odds Ratios* and 95% Confidence Intervals for Risk of Reporting Antisocial Behavior† in Part Year, by Gender, Grade, and Abuse History
Abuse history Abuse 1 molestation Molestation Abuse None Gender Female Male Grade 8 10 12
One Antisocial Behavior
Two or More Antisocial Behaviors
Adjusted Odds Ratio (95% Cl)
Adjusted Odds Ratio (95% Cl)
3.7 (2.5–5.6) 2.2 (1.6 –3.2) 2.0 (1.5–2.6) 1.0 (referent)
7.2 (4.1–12.9) 3.2 (1.8 –5.8) 4.3 (3.0 – 6.2) 1.0 (referent)
0.3 (0.2– 0.4) 1.0 (referent)
0.2 (0.1– 0.2) 1.0 (referent)
0.9 (0.7–1.3) 0.8 (0.7–1.0) 1.0 (referent)
1.0 (0.6 –1.6) 1.4 (0.9 –2.1) 1.0 (referent)
* Odds ratios for each predictor variable are adjusted for levels of the other predictor variables. † Antisocial behaviors include carried handgun, sold illegal drugs, stole car, was arrested, or took handgun to school.
suicidal behavior. For example, those students who reported combined abuse and molestation had more than a fourfold risk of reporting suicidal thoughts and more than a sixfold risk of reporting a suicide plan, a 12-fold risk of reporting a noninjurious suicide attempt, and a 47-fold risk of reporting an injurious suicide attempt compared to students who reported no abuse. Those students who reported abuse alone had a doubled risk of reporting suicidal thoughts, a threefold risk of reporting a suicide plan, a fivefold risk of reporting a noninjurious suicide attempt, and a more than 11-fold risk of reporting an injurious suicide attempt compared to students who reported no abuse. Table 4 shows numbers and weighted percentages of students reporting suicidal ideation/behavior by grade, gender, and abuse history; Table 5 shows adjusted ORs and 95% CIs.
Combination of Antisocial Behavior and Suicidal Ideation/Behavior in Past Year Girls were less likely than boys to report both antisocial and suicidal behavior, adjusted OR (95% CI) 5 0.4 (0.3– 0.6). Younger adolescents were more likely than older adolescents to report this combination. Adjusted ORs (95% CIs) were as follows: 8th grade, 2.3 (1.5–3.5); 10th grade, 1.7 (1.3–2.3); 12th grade, 1.0 (referent). Most important, abuse history was significantly associated with the combination of antisocial and suicidal behavior, and the increase in estimated risk was greater for combined abuse and molestation
March 1999
ABUSE AND ADOLESCENT PROBLEM BEHAVIORS
169
Table 4. Numbers and Weighted Percentages Reporting Most Serious* Suicidal Ideation/Behavior in Past Year, by Gender, Grade, and Abuse
Abuse history Abuse 1 molestation Molestation Abuse None Gender Female Male Grade 8 10 12 Total
None
Suicidal Thoughts
Suicide Plan
Noninjurious Suicide Attempt
Injurious Suicide Attempt
n (Weighted %)
n (Weighted %)
n (Weighted %)
n (weighted %)
n (Weighted %)
223 (46.7%) 199 (71.6%) 367 (69.6%) 3102 (87.8%)
64 (15.7%) 39 (9.5%) 63 (10.5%) 204 (5.7%)
39 (9.5%) 20 (8.3%) 38 (6.5%) 88 (2.6%)
83 (19.7%) 20 (6.7%) 50 (10.3%) 92 (2.7%)
31 (8.1%) 6 (3.1%) 11 (2.5%) 15 (0.3%)
1932 (76.3%) 1959 (86.3%)
237 (9.1%) 127 (5.5%)
104 (4.4%) 81 (3.6%)
184 (7.6%) 61 (2.7%)
49 (2.1%) 14 (0.6%)
1561 (80.0%) 1348 (80.2%) 982 (84.2%) 3891 (81.1%)
137 (6.4%) 134 (8.1%) 93 (8.0%) 364 (7.4%)
80 (3.8%) 61 (4.1%) 44 (4.2%) 185 (4.0%)
132 (7.3%) 76 (4.9%) 37 (2.5%) 245 (5.3%)
30 (1.5%) 26 (1.6%) 7 (0.9%) 63 (1.4%)
* Respondents were categorized into the most serious category they reported so that, for example, the percentage reporting suicidal thoughts does not include those respondents who also reported a suicide attempt. † Weighted percentages do not add to exactly 100% because of small amounts of missing data. The total n is 4790.
than for nonsexual abuse or molestation alone. ORs ratios (95% CIs) were 24.6 (15.5–39.1) for abuse and molestation, 5.2 (3.0 –9.0) for molestation, 5.6 (3.6 – 8.8) for abuse, and 1.0 (referent) for no reported abuse, respectively.
Discussion Both mild and severe antisocial behavior and suicidal ideation and behavior were associated with abuse history. The associations were especially strong for the more severe forms of the behaviors
(such as injurious suicide attempts) and for the combination of antisocial and suicidal behaviors. Although the reason for this pattern is not known, it may be related to the potentially damaging role of maltreatment in many aspects of psychological development (22). A variety of factors undoubtedly influence problem behaviors in adolescence, and some of these (such as peer pressure to perform behaviors disapproved by adults, disappointments in relationships, or academic difficulties) may be relatively transient or even normative influences. Less serious forms of behavioral problems may be
Table 5. Adjusted Odds Ratios* and 95% Confidence Intervals for Risk of Reporting Suicidal Ideation and Behavior in Past Year, by Gender, Grade, and Abuse History
Abuse history Abuse 1 molestation Molestation Abuse None Gender Female Male Grade 8 10 12
Suicidal Thoughts
Suicide Plan
Noninjurious Suicide Attempt
Injurious Suicide Attempt
Adjusted Odds Ratio (95% Cl)
Adjusted Odds Ratio (95% Cl)
Adjusted Odds Ratio (95% Cl)
Adjusted Odds Ratio (95% Cl)
4.4 (3.1– 6.2) 1.9 (1.2–2.8) 2.3 (1.7–3.2) 1.0 (referent)
6.8 (4.4 –10.4) 3.9 (2.2– 6.7) 3.1 (2.1– 4.6) 1.0 (referent)
12.0 (7.9 –18.4) 2.7 (1.5– 4.8) 5.1 (3.3–7.8) 1.0 (referent)
47.1 (23.3–95.3) 11.6 (3.2– 42.3) 11.8 (4.4 –31.9) 1.0 (referent)
1.5 (1.1–2.1) 1.0 (referent)
1.0 (0.7–1.4) 1.0 (referent)
2.2 (1.4 –3.3) 1.0 (referent)
1.9 (0.7–5.2) 1.0 (referent)
0.9 (0.6 –1.3) 1.1 (0.8 –1.4) 1.0 (referent)
1.1 (0.7–1.6) 1.0 (0.8 –1.4) 1.0 (referent)
3.7 (2.7–5.2) 2.1 (1.3–3.4) 1.0 (referent)
2.3 (1.0 –5.7) 2.0 (0.7–5.3) 1.0 (referent)
* Odds ratios for each predictor variable are adjusted for levels of the other predictor variables.
170
SOUTHWICK BENSLEY ET AL.
influenced relatively more by transient or normative influences, and more serious forms influenced relatively more by extreme influences such as abuse, at least in the realm of behaviors with a self-destructive or antisocial component such as those included in this report. Girls were more likely than boys to report abuse histories that included sexual molestation, whereas the two genders were similar in their reports of physical abuse alone. Girls were also more likely to report suicidal thoughts and noninjurious suicide attempts, and boys were more likely to report antisocial behavior. It should be noted, however, that the antisocial behavior items in this study reflected the more aggressive and overt antisocial behaviors typically committed by boys (23) and did not include the more covert types of antisocial behaviors typically committed by girls, such as shoplifting and relational aggression (harming or threatening to harm another’s relationships— e.g., getting even by excluding someone from a group) (24). The fact that we found higher rates of one and two or more antisocial behaviors for males may be a consequence of our measure of antisocial behavior, and not a basic difference in rates of antisocial behaviors between adolescent boys and girls, per se. Boys do have higher rates of overt aggression and violent offending throughout the life span (25), but it has been argued that by the adolescent years, the male predominance in antisocial behavior is no longer statistically significant if both overt and covert antisocial behaviors are considered (26). This means that the girls reporting the antisocial behaviors measured in this study may be more deviant, relative to same sex norms, than the boys. Future research should address whether a history of abuse also increases risk for covert antisocial behavior. The combination of abuse and sexual molestation was consistently more strongly associated with both antisocial and self-destructive behaviors in adolescence than nonsexual abuse or molestation alone. There is an increase in suicidal behavior among youth who show both antisocial and depressive features (27). When a child is depressed and aggressive, aggressive impulses tend to be directed against the self. This study has several limitations. First, whereas the response rate for the students was high (96% of students whose schools agreed to participate in the survey and who were present on that day completed questionnaires), only 22% of the schools asked to participate agreed to administer the survey. Therefore, it is possible that the results reported here are
JOURNAL OF ADOLESCENT HEALTH Vol. 24, No. 3
not generalizable to other groups of adolescents. However, we did not detect differences between the schools which participated and those which refused to participate on a number of parameters. While we did not have the data to detect subtle differences, the fact that there are not large differences indicates that the results may be representative of other adolescents in Washington public schools. Second, it should also be noted that if students who are absent or drop out from school are more likely than others to have been abused, then the prevalence rates in the current study may be underestimates of the actual numbers abused. Students who are in schools other than public schools (such as private schools or alternative schools outside the public school system) may also have different rates of abuse compared to students in public schools. Prevalence rates of other factors measured in this report may also have been influenced by dropouts, absentees, and the inability to assess students outside of the public school system, as the survey only reached students who were present in public schools. However, only about 7% of Washington students are in private schools. Third, the use of self-reports introduces the possibility that students were intentionally or unintentionally providing false responses. To minimize the possibility of including questionnaires with false responses, we eliminated questionnaires for dishonesty, internal inconsistencies, and reporting use of a fictitious drug. The anonymity of the self-reports might be expected to mitigate these problems, especially for boys. Widom and Morris (7) reported that only 16% of men known to have been sexually abused in childhood disclosed this in a face-to-face interview, compared to 60% of women. In addition, the prevalence rates obtained in this study for alcohol and cigarette use were similar to results from national studies using self-administered questionnaires in a school setting (19). We provide validity evidence for self-reports of abuse in the accompanying report (15). Fourth, because of the school setting, we were not able to ask important questions about the abuse, such as the relationship to the perpetrator, age of first occurrence, number of occurrences, and details about the nature of the abuse. Therefore, we were not able to address the question of chronicity of abuse, which we found to be important in a related study (10). Also, it is possible that some of the students who were coded as molested were reporting experiences with peers instead of abuse by someone older. Finally, the cross-sectional nature of the study does not
March 1999
allow firm statements about causality. This is discussed more completely in the accompanying report (15). The behavior most strongly associated with abuse history was injurious suicide attempts. Maltreatment may be a predominant risk factor for serious suicidal behaviors. By comparison, behaviors which involve pleasurable activities with peers, such as substance use, or activities with an illegal but goal-oriented focus, such as some delinquent activities (e.g., theft) may have relatively more alternative risk factors. For example, while relatively small numbers of adolescents may be influenced by peers to attempt suicide, relatively large numbers may be influenced by peers in deciding whether to drink in settings such as parties. Further research is needed to identify factors which differentiate among maltreated adolescents who display different needs and behaviors. Adolescents who display antisocial behavior in particular are likely to be dealt with in the juvenile justice system and to have their own history of being abused overlooked (28). Legal and disciplinary problems are, in turn, a risk factor for suicide in the next year (4). Both effective programs aimed at intervening with youth who are at risk for suicide (4,29) and antisocial behavior and programs aimed at preventing the occurrence of maltreatment are needed. The evidence to date suggests that effective programs should include parents, and in particular target the discord between parents and children (4,30 –32). This research was presented in part at the annual meeting of the Society of Behavioral Medicine, New Orleans, March, 1998.
ABUSE AND ADOLESCENT PROBLEM BEHAVIORS
8.
9.
10.
11.
12.
13. 14.
15.
16.
17.
18.
19.
20.
21.
References 1. Hinden BR, Compas BE, Howell DC, Achenbach TM. Covariation of the anxious-depressed syndrome during adolescence: Separating fact from artifact. J CCP 1997;65:6 –14. 2. Silverman AB, Reinherz HZ, Giaconia RM. The long-term sequelae of child and adolescent abuse: A longitudinal community study. Child Abuse Negl 1996;20:709 –23. 3. Brent DA, Perper JA, Moritz G, et al. Psychiatric risk factors for adolescent suicide: A case-control study. J Am Acad Child Adolesc Psychiatry 1993;32:521–9. 4. Brent DA, Perper JA, Moritz G, et al. Stressful life events, psychopathology, and adolescent suicide: A case-control study. Suicide Life-Threat Behav 1993;23:179 – 87. 5. Ireland T, Widom CS. Childhood victimization and risk for alcohol and drug arrests. Int J Addict 1994;29:235–74. 6. Widom CS, Ashley AM. Criminal consequences of childhood sexual victimization. Child Abuse Negl 1994;18:303–18. 7. Widom CS, Kuhns JB. Childhood victimization and subsequent risk for promiscuity, prostitution, and teenage preg-
22.
23.
24. 25.
26. 27.
28.
171
nancy: A prospective study. Am J Public Health 1996;86:1607– 12. Nelson DD, Higginson GK, Grant-Worley JA. Using the Youth Risk Behavior Survey to estimate prevalence of sexual abuse among Oregon high school students. J Sch Health 1994;64: 413– 6. Briere J, Runtz M. Differential adult symptomatology associated with three types of child abuse histories. Child Abuse Negl 1990; 14:357– 64. Spieker SJ, Bensley L, McMahon RJ, et al. Sexual abuse as a factor in child maltreatment by adolescent mothers of preschool aged children. Dev Psychopathol 1996;8:497–509. Orpinas PK, Basen-Engquist K, Grunbaum JA, et al. The co-morbidity of violence-related behaviors with health-risk behaviors in a population of high school students. J Adolesc Health 1995;16:216 –25. Valois RF, McKeown RE, Garrison CZ, et al. Correlates of aggressive and violent behaviors among public high school adolescents. J Adolesc Health 1995;16:26 –34. Lewinsohn PM, Rohde P, Seeley JR. Psychosocial risk factors for future suicide attempts. JCCP 1994; 62:297–305. Garrison CZ, Addy CL, Jackson KL, et al. A longitudinal study of suicidal ideation in young adolescents. J Am Acad Child Adolesc Psychiatry 1991;30:597– 603. Bensley LS, Spieker SJ, Van Eenwyk J, Schoder J. Self-reported abuse history and adolescent problem behaviors. II. Alcohol and drug use. J Adolesc Health 1999;24:173–180. Widom CS, Shepard RL. Accuracy of adult recollections of childhood victimization: Part 1. Childhood physical abuse. Psychol Assess 1996;8:412–21. Widom CS, Morris S. Accuracy of adult recollections of childhood victimization: Part 2. Childhood sexual abuse. Psychol Assess 1997;9:34 – 46. Brener ND, Collins JL, Kann L, et al. Reliability of the Youth Risk Behavior Survey questionnaire. Am J Epidemiol 1995;141: 575– 80. Center for Disease Control and Prevention. Reliability of the Youth Risk Behavior Surveillance—United States, 1995. MMWR 1996;45:1– 84. Shah BV, Barnwell BG, Bieler GS. SUDAAN user’s manual, Release 7.5. Research Triangle Park, NC: Research Triangle Institute, 1997. Ananth CV, Kleinbaum DG. Regression models for ordinal responses: A review of methods and applications. Int J Epidemiol 1997;26:1323–33. Cicchetti D, Carlson V, eds. Child Maltreatment: Theory and Research on the Causes and Consequences of Child Abuse and Neglect. Cambridge, England: Cambridge University Press, 1993. Dishion TJ, French DC, Patterson GR. The development and ecology of antisocial behavior. In: Cicchetti D, Cohen DJ, eds. Development of Psychopathology, Vol. 2. New York: Wiley, 1995:421–71. Crick NR, Grotpeter JK. Relational aggression, gender, and social-psychological adjustment. Child Dev 1995;66:710 –22. Moffitt TE. “Adolescence-limited” and “life-course-persistent” antisocial behavior: A developmental taxonomy. Psychol Rev 1993;100:674 –701. Zoccolillo M. Gender and the development of conduct disorder. Dev Psychopathol 1993;5:65–78. Shaffi N, Carrigan S, Whittinghill JR, Derrick A. Psychological autopsy of completed suicide of children and adolescents. Am J Psychiatry 1985;142:1061– 4. Howing PT, Wodarski JS, Kurtz PD, Gaudin JM Jr. Maltreatment and the School-Age Child: Developmental Outcomes and System Issues. Binghamton, NY: Haworth Press, 1993.
172
SOUTHWICK BENSLEY ET AL.
29. Eggert LL, Thompson EA, Herting JR. A measure of adolescent potential for suicide (MAPS): Development and preliminary findings. Suicide Life-Threat Behav 1994;24:359 – 81. 30. Hengeller S, Rodick J, Borduin C, et al. Multisystemic treatment of juvenile offenders: Effects on adolescent behavior and family interaction. Dev Psychol 1986;22:132– 41.
JOURNAL OF ADOLESCENT HEALTH Vol. 24, No. 3
31. Kazdin A, Bass D, Siegel T, Thomas C. Cognitive-behavior therapy and relationship therapy in the treatment of children referred for antisocial behavior. JCCP 1989;57:522–35. 32. Patterson G, Chamberlain, P, Reid J. A comparative evaluation of a parent-training program. Behav Ther 1982;13: 638 –50.
ORIGINAL ARTICLE
Self-Reported Abuse History and Adolescent Problem Behaviors. I. Antisocial and Suicidal Behaviors LILLIAN SOUTHWICK BENSLEY, Ph.D., JULIET VAN EENWYK, Ph.D., SUSAN J. SPIEKER, Ph.D., AND JUDY SCHODER, M.N.
Purpose: To examine the associations of self-reported abuse and sexual molestation with self-reported antisocial behavior and suicidal ideation/behavior in a general population of adolescents. Methods: We used a stratified cluster sampling procedure with replacement to sample 4790 students in Washington State public schools in Grades 8, 10, and 12. Students were asked in a self-administered questionnaire whether they had ever been abused by an adult and whether they had ever been sexually molested. They were also asked about antisocial behavior and suicidal ideation and behavior in the past year. We conducted polytomous logistic regressions, controlling for gender and grade, using Software for the Statistical Analysis of Correlated Data (SUDAAN). Results: Reported abuse history was associated with antisocial behavior and with suicidal ideation and behavior. The associations were stronger for abuse and molestation than for nonsexual abuse or molestation alone, and stronger at higher levels of severity (e.g., suicide attempts vs. suicidal thoughts). For example, adjusted odds ratios and 95% confidence intervals (in parentheses) for abuse and molestation were 4.4 (3.1– 6.2) for suicidal thoughts, 6.8 (4.4 –10.4) for suicide plan, 12.0 (7.9 –18.4) for noninjurious suicide attempt, and 47.1 (23.3–95.3) for injurious suicide attempt. For abuse alone, these figures were 2.3 (1.7–3.2), 3.1 (2.1– 4.6), 5.1 (3.3–7.8), and 11.8 (4.4 –31.9), respectively. Conclusions: Efforts to reduce antisocial behavior and suicidal ideation/behavior in adolescence, particularly early or severe manifestations of the behaviors, should consider the possible role of a history of maltreatment, From the Washington State Department of Health, Olympia, and the University of Washington, Seattle. Address reprint requests to: Lillian Bensley, Ph.D., Washington State Department of Health Office of Epidemiology, 1102 SE Quince St., Olympia, WA 98504-7812. Manuscript accepted May 12, 1998.
especially the possibility of sexual abuse. © Society for Adolescent Medicine, 1999 KEY WORDS: Adolescence Child abuse Physical abuse Sexual abuse Antisocial behavior Suicide Survey Gender differences
Adolescent depression, anxiety, and delinquent behavior tend to co-occur (1) and a history of being maltreated has been reported as a risk factor for each. However, the degree of risk associated with different types of maltreatment and the extent to which findings from studies of specialized groups generalize to adolescents in the general population are not known. Large-scale surveys, which rely on self-report of both abuse history and long-term sequelae, are important to describe the prevalence of abuse and its sequelae in populations such as school students so that the extent of the need for prevention and intervention services can be documented. A 17-year longitudinal community study using self-reports of abuse history linked abuse to a variety of negative outcomes in adolescence and adulthood, including depressive symptomatology, psychiatric disorder, suicidal ideation, and suicide attempts (2), and smaller studies of clinical samples have found similar associations. However, these studies have included relatively small numbers of maltreated individuals, restricting their ability to examine differ-
© Society for Adolescent Medicine, 1999 Published by Elsevier Science Inc., 655 Avenue of the Americas, New York, NY 10010
1054-139X/99/$–see front matter PII S1054-139X(98)00111-6
164
SOUTHWICK BENSLEY ET AL.
ent types of maltreatment. Brent et al. (3,4) conducted a case-control study of adolescents who completed suicide. These authors identified mood disorders, substance abuse, and conduct disorder as risk factors and found that past suicidal ideation with a plan was as strongly associated with completed suicide as a past attempt (3). The association between a history of maltreatment and suicide in the Brent et al. sample did not achieve statistical significance; however, this was likely owing to the small sample sizes (8% of the 67 suicide completers reported physical abuse compared to 0% of the community controls) (4). Research on delinquency has also shown relatively consistent associations with a history of maltreatment. Ireland and Widom (5), Widom and Ashley (6), and Widom and Kuhns (7) used a prospective cohort design in which validated cases of child sexual abuse, physical abuse, and neglect were matched by age, sex, race, and socioeconomic status with controls for whom there were no records of child abuse or neglect. Childhood sexual abuse, physical abuse, and/or neglect were predictors of later adolescent delinquent and adult criminal behavior; arrest for prostitution; adult arrest for alcohol- and/or drug-related offenses; and, for males, violent sex crimes. However, this study included only cases of childhood abuse and neglect that came to the attention of the authorities and official criminal histories, so that only relatively severe forms of both maltreatment and antisocial behavior were included. A few large-scale student surveys have included questions on physical and sexual abuse. For example, Nelson et al. (8) added questions to Oregon’s 1993 Youth Risk Behavior Survey. Students who reported that they had been sexually abused within the past year were more likely to have a low self-image, to carry weapons, and to have been in physical fights, have seriously considered suicide in the past year, and have used alcohol or drugs in the past month compared with students who reported never having been sexually abused. This study confirms the findings of others that sexual abuse and adolescent high-risk behaviors are highly correlated. However, there was no way to determine if the abuse in the past year occurred in a peer dating situation or had occurred with someone older. In addition, the sole focus on sexual abuse may be misleading. Maltreatment rarely exists in pure forms, as more than one type of abuse is usually associated with a given incident (9). By reporting only on sexual abuse (and further restricting that to abuse in the past year), Nelson et al. could not conclude which outcomes
JOURNAL OF ADOLESCENT HEALTH Vol. 24, No. 3
were associated with sexual abuse and which were associated with other forms of abuse that may have coexisted but were not included in their analyses. As there is increasing evidence that different types of maltreatment (physical abuse and sexual abuse) are associated with different outcomes (10), it is important to ask respondents in a population survey to differentiate types of abuse. Some studies have used student surveys to examine associations among aggressive and violent behavior and other health risk behaviors (11,12). These studies have not examined family characteristics, such as child abuse, associated with these patterns. Other longitudinal studies found that family characteristics such as low support, cohesion, and adaptability predicted future suicidal ideation and behavior in adolescents (13,14), but did not measure maltreatment apart from other family factors. We examined the health risk behaviors associated with a history of physical abuse, sexual abuse, and combined physical abuse and sexual abuse in a general population sample of adolescents in Washington public schools. The sample size was large (4790 individuals), so that we were able to assess the degree of risk associated with different types of maltreatment. In this report, we present detailed methodological information about the study and the associations of abuse history with suicidal ideation/ behavior and antisocial behavior. In an accompanying report (15), we present evidence of validity of the abuse history self-reports and the associations of abuse history with alcohol and drug use and early initiation of substance use. [For related evidence about the validity of self-reports of abuse, see (16,17)]. The specific questions addressed by this study were: 1. How prevalent are self-reported histories of physical abuse; sexual abuse; combined physical abuse and sexual abuse; antisocial behavior; suicidal thoughts and behavior; and combined antisocial behavior and suicidal thoughts and behavior in adolescents from the State of Washington? 2. What are the associations between self-reported history of physical abuse; sexual abuse; and combined physical abuse and sexual abuse on the one hand and antisocial behavior; suicidal thoughts and behavior; and combined antisocial behavior and suicidal thoughts and behavior on the other hand in adolescence? 3. Are there age or gender differences in these associations that might be helpful in targeting interventions?
March 1999
Methods Sampling We sampled students in Washington State public schools in Grades 8, 10, and 12 using a stratified cluster sampling procedure with replacement. (Sixth graders were also surveyed but were not asked about abuse history and so are not included in this report.) Schools were the primary sampling unit. Stratification was by geographic regions (East, Southwest, Puget Sound, and Northwest) and school size (large or small). To boost the representation of racial/ ethnic minorities in the state sample, a sampling with probability proportionate to size (PPS) procedure was used in which schools with higher minority enrollments received higher probabilities of selection. Replacement schools were selected using the above procedures and with the additional criterion that they were similar in level of urbanicity (metropolitan core, small city, suburban, or rural). In other words, when a given school refused to participate, another school from that region of the same size, urbanicity, and minority enrollment was recruited to take its place. Within the participating schools, all students at the target grades were asked to participate. A weighting factor was applied to each student record to adjust for the varying probabilities of selection in an effort to maximize the representativeness of the results to the population of public school students in Washington State. Students in 44 middle and high schools participated, and the school response rate was 22%. Comparisons between participating and nonparticipating schools did not reveal differences for a number of parameters including percentage of children participating in the free or reduced lunch program; graduation rates; percentage of high school dropouts and students whose status was unknown; percentage of eighth graders with computers at home; percentage of eighth graders whose mothers did not have a high school diploma; and 11th-grade test scores. Of the students present on the day of the survey, the student response rate was 96%. Of the completed surveys (including sixth graders), approximately 8% were discarded because of dishonesty or inconsistent responses. (More detail is provided below in the section on quality assurance.) These procedures led to a final sample of 4790 completed questionnaires. Questionnaire and Data Collection The questionnaires addressed a broad range of student health risk behaviors including physical fitness,
ABUSE AND ADOLESCENT PROBLEM BEHAVIORS
165
unintentional and intentional injury behaviors, alcohol, tobacco and other drug use, human immunodeficiency virus/acquired immunodeficiency syndrome education, and risk and protective factors for adolescent health risk behaviors. The questionnaire was administered in the classroom during a regular class period. Completing the questionnaire took approximately 50 min. Instructions stated that the questionnaire was voluntary and that responses were anonymous and confidential. Students recorded their responses directly on the questionnaire, which was in scannable format. All students were provided with an 800 number to call for referral if they wished to talk with someone about issues raised by the survey. Parental consent procedures included mailings to the parents providing information about the content and reasons for the survey. Parents were given the opportunity to review the entire survey at the school and to notify the school if they did not want their children to participate. The Washington State Department of Social and Health Services/Department of Health Human Research Review Board approved the research procedures. Coding Abuse history. Students were asked two questions about their abuse histories: “Have you ever been abused or mistreated by an adult?” and “Has anyone ever touched you in a sexual place, or made you touch them, when you did not want them to?” Both questions were coded “yes”/“no.” To create a summary measure of abuse history, we coded students who answered “no” to both questions as not abused, “yes” to the first and “no” to the second as abused, “yes” to the second and “no” to the first as molested, and “yes” to both as abused and molested. Evidence of the predictive validity of this measure is provided in the accompanying report (15). Antisocial behavior. The measure of antisocial behavior was constructed from five questions measuring how many times in the past year the respondent had carried a handgun, sold illegal drugs, stolen or tried to steal a motor vehicle such as a car or motorcycle, been arrested, or taken a handgun to school. The measure of antisocial behavior indicated whether the respondent reported “none”, “one”, or “two or more” different behaviors. This definition emphasizes the degree of diversity of antisocial behavior over the number of times a particular behavior was committed.
166
SOUTHWICK BENSLEY ET AL.
Suicidal ideation and behavior. The measures of suicidal ideation/behavior were identical to those used on the Youth Risk Behavior Survey (18,19). Four questions were: “During the past 12 months, have you ever seriously thought about attempting suicide?” “During the past 12 months, did you make a specific plan about how you would attempt suicide?” “During the past 12 months, how many times did you actually make a suicide attempt?” and “If you attempted suicide during the past 12 months, did that attempt result in an injury, poisoning, or overdose that had to be treated by a doctor or a nurse?” Students were coded as the most serious category (thoughts, plan, noninjurious attempt, or injurious attempt) that they reported.
Data Analysis All analyses were conducted using Software for the Statistical Analysis of Correlated Data (SUDAAN) (20). The primary sampling unit was school. Observations were weighted by the school characteristics (region, size, and minority concentration) that were used in sampling. Major analyses. Major outcome measures were antisocial behavior, which had three levels of severity (“none”, “one”, or “two or more” different antisocial behaviors) and suicidal ideation/behavior, which had five levels of severity (“none”, “thoughts”, “plans”, “noninjurious attempts”, and “injurious attempts”). We compared each other level of severity to the group reporting none of that type of behavior (e.g., for each of the analyses of suicidal ideation/behavior, the referent group was those students who reported no suicidal thoughts). We also compared respondents who reported a combination of antisocial and suicidal behavior (at least one antisocial behavior and at least one suicidal ideation/behavior) to those who reported neither. Preliminary analyses using polytomous logistic regression models (21) tested whether significant Grade 3 Abuse history or Gender 3 Abuse history interactions existed across the levels of severity of each major outcome measure. (The interaction test of suicidal ideation/behavior did not include one level of severity that had low-prevalence, injurious suicide attempts, owing to empty cells.) An additional dichotomous logistic regression analysis tested the same possible interactions on the combined antisocial/suicidal outcome measure. We did not identify any interactions that were statistically significant (all
JOURNAL OF ADOLESCENT HEALTH Vol. 24, No. 3
interaction p’s . .05), and so we used a main-effects model for the outcomes in this report. For each of the major outcome measures (the three-level measure of antisocial behavior and the five-level measure of suicidal ideation/behavior) we conducted a polytomous logistic regression in which the predictor variables were the four-level abuse history variable (“none”, “abused”, “molested”, or “abused and molested”), gender (male or female), and grade (8, 10, or 12). From these logistic regressions, we calculated odds ratios (ORs) and 95% confidence intervals (CIs) for each predictor variable, adjusting for levels of the other variables. We also conducted an identical dichotomous logistic regression analysis on the combined antisocial/suicidal outcome measure. Quality assurance. In a section on drug use, students were asked whether they had ever used derbisol (a fictitious drug) and whether they had answered the questions honestly, to identify students who were providing false responses. Students who admitted answering dishonestly or somewhat honestly, claimed use of a fictitious drug, and/or responded inconsistently to three or more pairs of related items (e.g., claimed 30-day use of a substance on one item and no use in lifetime on another item) were omitted from all analyses (8%).
Results Sample Characteristics The sample included 4790 students in Washington State public schools, Grades 8, 10, and 12. Sample characteristics are presented in Table 1. The sample was approximately evenly divided between males and females and was predominately white. Differing sample sizes by grade levels reflected the fact that there were fewer total students in Grade 12 compared to the other grades. This was owing in part to differences in cohort sizes and in part to dropouts between Grades 10 and 12. The total number does not include 320 students who had missing data on gender or abuse history and so were omitted from all analyses.
Abuse History The results of the abuse history questions and weighted percentages are presented in Table 1. Overall, 73.9% reported no maltreatment history, 11.4% reported being abused, 5.9% reported being mo-
March 1999
ABUSE AND ADOLESCENT PROBLEM BEHAVIORS
Table 1. Sample Characteristics Characteristic Gender Female Male Grade 8 10 12 Race/ethnicity White, not Hispanic Black, not Hispanic Hispanic Asian/Pacific Islander Native American Other/unknown Abuse history Abuse 1 sexual molestation Molestation Abuse None Antisocial behaviors in past year None One Two or more Didn’t answer question Suicidal ideation/behaviors in past year None Suicidal thoughts Suicide plan Noninjurious suicide attempt Injurious suicide attempt Didn’t answer question
n
Weighted %
2518 2272
52.1 47.9
1960 1661 1169
40.8 34.7 24.6
3617 88 479 269 123 214
75.0 2.4 9.1 5.9 3.2 4.3
441 279 533 3537
8.8 5.9 11.4 73.9
3852 567 328 43
80.3 11.7 7.0 0.9
3891 364 185 245 63 42
81.1 7.4 4.0 5.3 1.4 0.8
167
1. Of the sample, 11.7% reported one antisocial behavior and 7.0% reported two or more different antisocial behaviors in the past year. As noted earlier, for suicidal ideation/behavior, respondents were categorized into the most serious category of behavior they reported. The most seriously injurious suicidal ideation/behaviors reported by students were as follows: suicidal thoughts 7.4%, suicide plan 4.0%, noninjurious suicide attempt 5.3%, and injurious suicide attempt 1.4%. However, as might be expected, most people with a suicide attempt or a suicide plan also reported suicidal thoughts. If this overlap is included, then the total numbers and weighted percentages of students reporting suicidal thoughts are 852 (18.0%), suicide plan 420 (9.2%), and suicide attempt 308 (6.7%). A total of 264 individuals (5.7%) reported both suicidal ideation or behavior and antisocial behavior. Abuse History and Antisocial Behavior in the Past Year
lested, and 8.8% reported being abused and molested. Abuse histories differed significantly for boys and girls (x2 5 29.91; p , .0001). Of the girls, 65.6% reported no maltreatment history, 10.7% reported being abused, 9.2% reported being molested, and 14.6% reported being abused and molested. Of the boys, 82.9% reported no maltreatment history, 12.19% reported being abused, 2.4% reported being molested, and 2.6% reported being abused and molested.
Girls were less than one-third as likely as boys to report one type of antisocial behavior and one fifth as likely to report two or more different antisocial behaviors in the past year. We did not identify significant associations between grade and antisocial behavior. Most important, abuse history was associated with antisocial behavior, and the associations were stronger for combined abuse and molestation than for nonsexual abuse or molestation alone, and stronger for more severe antisocial behavior (two or more different behaviors) than for less severe manifestations (one type of behavior). For example, those students who reported combined abuse and molestation had more than a threefold risk of reporting one antisocial behavior and more than a sevenfold risk of reporting two or more antisocial behaviors compared to students who reported no abuse. Those students who reported abuse alone had a doubled risk of reporting one antisocial behavior and a fourfold risk of reporting two or more antisocial behaviors compared to students who reported no abuse. Table 2 shows numbers and weighted percentages of students reporting antisocial behavior by grade, gender, and abuse history, and Table 3 shows adjusted ORs and 95% CIs.
Antisocial Behavior and Suicidal Ideation/Behavior
Abuse History and Suicidal Behavior in the Past Year
Levels of the target behaviors and weighted percentages across the entire sample are presented in Table
Girls were 1.5 times more likely than boys to report suicidal thoughts and twice as likely to report non-
The overall n was 4790. Percentages may not sum to 100% because of small amounts of missing data. An additional 320 students for whom gender or abuse history were coded as missing were not included in any of the analyses. Antisocial behaviors included carried handgun, sold illegal drugs, stole car, was arrested, or took handgun to school.
168
SOUTHWICK BENSLEY ET AL.
JOURNAL OF ADOLESCENT HEALTH Vol. 24, No. 3
Table 2. Numbers and Weighted Percentages Reporting Antisocial Behavior* in Past Year, by Gender, Grade, and Abuse History
Abuse history Abuse 1 molestation Molestation Abuse None Gender Female Male Grade 8 10 12 Total
None
One Antisocial Behavior
Two or More Antisocial Behaviors
n (Weighted %)
n (Weighted %)
n (Weighted %)
313 (69.7%) 213 (78.4%) 366 (67.9%) 2960 (83.7%)
72 (17.5%) 39 (13.0%) 81 (15.2%) 375 (10.4%)
53 (12.5%) 24 (7.5%) 81 (15.4%) 170 (5.1%)
2227 (88.1%) 1625 (71.9%)
186 (7.6%) 381 (16.1%)
89 (3.5%) 239 (10.8%)
1594 (81.2%) 1332 (79.7%) 926 (79.8%) 3852 (80.3%)
223 (11.8%) 193 (11.6%) 151 (13.1%) 567 (11.7%)
125 (6.0%) 120 (8.5%) 83 (6.6%) 328 (7.0%)
* Antisocial behaviors include carried handgun, sold illegal drugs, stole car, was arrested, or took handgun to school. † Weighted percentages do not add to exactly 100% because of small amounts of missing data. The total n is 4790.
injurious suicide attempts. Risk of having a suicide plan or an injurious suicide attempt was not significantly related to gender. Risk of a noninjurious suicide attempt was highest in 8th grade, followed by 10th and then 12th grade. Most important, abuse history was associated with suicidal ideation/behavior, and the associations were stronger for combined abuse and molestation than for nonsexual abuse or molestation alone, and stronger for more severe Table 3. Adjusted Odds Ratios* and 95% Confidence Intervals for Risk of Reporting Antisocial Behavior† in Part Year, by Gender, Grade, and Abuse History
Abuse history Abuse 1 molestation Molestation Abuse None Gender Female Male Grade 8 10 12
One Antisocial Behavior
Two or More Antisocial Behaviors
Adjusted Odds Ratio (95% Cl)
Adjusted Odds Ratio (95% Cl)
3.7 (2.5–5.6) 2.2 (1.6 –3.2) 2.0 (1.5–2.6) 1.0 (referent)
7.2 (4.1–12.9) 3.2 (1.8 –5.8) 4.3 (3.0 – 6.2) 1.0 (referent)
0.3 (0.2– 0.4) 1.0 (referent)
0.2 (0.1– 0.2) 1.0 (referent)
0.9 (0.7–1.3) 0.8 (0.7–1.0) 1.0 (referent)
1.0 (0.6 –1.6) 1.4 (0.9 –2.1) 1.0 (referent)
* Odds ratios for each predictor variable are adjusted for levels of the other predictor variables. † Antisocial behaviors include carried handgun, sold illegal drugs, stole car, was arrested, or took handgun to school.
suicidal behavior. For example, those students who reported combined abuse and molestation had more than a fourfold risk of reporting suicidal thoughts and more than a sixfold risk of reporting a suicide plan, a 12-fold risk of reporting a noninjurious suicide attempt, and a 47-fold risk of reporting an injurious suicide attempt compared to students who reported no abuse. Those students who reported abuse alone had a doubled risk of reporting suicidal thoughts, a threefold risk of reporting a suicide plan, a fivefold risk of reporting a noninjurious suicide attempt, and a more than 11-fold risk of reporting an injurious suicide attempt compared to students who reported no abuse. Table 4 shows numbers and weighted percentages of students reporting suicidal ideation/behavior by grade, gender, and abuse history; Table 5 shows adjusted ORs and 95% CIs.
Combination of Antisocial Behavior and Suicidal Ideation/Behavior in Past Year Girls were less likely than boys to report both antisocial and suicidal behavior, adjusted OR (95% CI) 5 0.4 (0.3– 0.6). Younger adolescents were more likely than older adolescents to report this combination. Adjusted ORs (95% CIs) were as follows: 8th grade, 2.3 (1.5–3.5); 10th grade, 1.7 (1.3–2.3); 12th grade, 1.0 (referent). Most important, abuse history was significantly associated with the combination of antisocial and suicidal behavior, and the increase in estimated risk was greater for combined abuse and molestation
March 1999
ABUSE AND ADOLESCENT PROBLEM BEHAVIORS
169
Table 4. Numbers and Weighted Percentages Reporting Most Serious* Suicidal Ideation/Behavior in Past Year, by Gender, Grade, and Abuse
Abuse history Abuse 1 molestation Molestation Abuse None Gender Female Male Grade 8 10 12 Total
None
Suicidal Thoughts
Suicide Plan
Noninjurious Suicide Attempt
Injurious Suicide Attempt
n (Weighted %)
n (Weighted %)
n (Weighted %)
n (weighted %)
n (Weighted %)
223 (46.7%) 199 (71.6%) 367 (69.6%) 3102 (87.8%)
64 (15.7%) 39 (9.5%) 63 (10.5%) 204 (5.7%)
39 (9.5%) 20 (8.3%) 38 (6.5%) 88 (2.6%)
83 (19.7%) 20 (6.7%) 50 (10.3%) 92 (2.7%)
31 (8.1%) 6 (3.1%) 11 (2.5%) 15 (0.3%)
1932 (76.3%) 1959 (86.3%)
237 (9.1%) 127 (5.5%)
104 (4.4%) 81 (3.6%)
184 (7.6%) 61 (2.7%)
49 (2.1%) 14 (0.6%)
1561 (80.0%) 1348 (80.2%) 982 (84.2%) 3891 (81.1%)
137 (6.4%) 134 (8.1%) 93 (8.0%) 364 (7.4%)
80 (3.8%) 61 (4.1%) 44 (4.2%) 185 (4.0%)
132 (7.3%) 76 (4.9%) 37 (2.5%) 245 (5.3%)
30 (1.5%) 26 (1.6%) 7 (0.9%) 63 (1.4%)
* Respondents were categorized into the most serious category they reported so that, for example, the percentage reporting suicidal thoughts does not include those respondents who also reported a suicide attempt. † Weighted percentages do not add to exactly 100% because of small amounts of missing data. The total n is 4790.
than for nonsexual abuse or molestation alone. ORs ratios (95% CIs) were 24.6 (15.5–39.1) for abuse and molestation, 5.2 (3.0 –9.0) for molestation, 5.6 (3.6 – 8.8) for abuse, and 1.0 (referent) for no reported abuse, respectively.
Discussion Both mild and severe antisocial behavior and suicidal ideation and behavior were associated with abuse history. The associations were especially strong for the more severe forms of the behaviors
(such as injurious suicide attempts) and for the combination of antisocial and suicidal behaviors. Although the reason for this pattern is not known, it may be related to the potentially damaging role of maltreatment in many aspects of psychological development (22). A variety of factors undoubtedly influence problem behaviors in adolescence, and some of these (such as peer pressure to perform behaviors disapproved by adults, disappointments in relationships, or academic difficulties) may be relatively transient or even normative influences. Less serious forms of behavioral problems may be
Table 5. Adjusted Odds Ratios* and 95% Confidence Intervals for Risk of Reporting Suicidal Ideation and Behavior in Past Year, by Gender, Grade, and Abuse History
Abuse history Abuse 1 molestation Molestation Abuse None Gender Female Male Grade 8 10 12
Suicidal Thoughts
Suicide Plan
Noninjurious Suicide Attempt
Injurious Suicide Attempt
Adjusted Odds Ratio (95% Cl)
Adjusted Odds Ratio (95% Cl)
Adjusted Odds Ratio (95% Cl)
Adjusted Odds Ratio (95% Cl)
4.4 (3.1– 6.2) 1.9 (1.2–2.8) 2.3 (1.7–3.2) 1.0 (referent)
6.8 (4.4 –10.4) 3.9 (2.2– 6.7) 3.1 (2.1– 4.6) 1.0 (referent)
12.0 (7.9 –18.4) 2.7 (1.5– 4.8) 5.1 (3.3–7.8) 1.0 (referent)
47.1 (23.3–95.3) 11.6 (3.2– 42.3) 11.8 (4.4 –31.9) 1.0 (referent)
1.5 (1.1–2.1) 1.0 (referent)
1.0 (0.7–1.4) 1.0 (referent)
2.2 (1.4 –3.3) 1.0 (referent)
1.9 (0.7–5.2) 1.0 (referent)
0.9 (0.6 –1.3) 1.1 (0.8 –1.4) 1.0 (referent)
1.1 (0.7–1.6) 1.0 (0.8 –1.4) 1.0 (referent)
3.7 (2.7–5.2) 2.1 (1.3–3.4) 1.0 (referent)
2.3 (1.0 –5.7) 2.0 (0.7–5.3) 1.0 (referent)
* Odds ratios for each predictor variable are adjusted for levels of the other predictor variables.
170
SOUTHWICK BENSLEY ET AL.
influenced relatively more by transient or normative influences, and more serious forms influenced relatively more by extreme influences such as abuse, at least in the realm of behaviors with a self-destructive or antisocial component such as those included in this report. Girls were more likely than boys to report abuse histories that included sexual molestation, whereas the two genders were similar in their reports of physical abuse alone. Girls were also more likely to report suicidal thoughts and noninjurious suicide attempts, and boys were more likely to report antisocial behavior. It should be noted, however, that the antisocial behavior items in this study reflected the more aggressive and overt antisocial behaviors typically committed by boys (23) and did not include the more covert types of antisocial behaviors typically committed by girls, such as shoplifting and relational aggression (harming or threatening to harm another’s relationships— e.g., getting even by excluding someone from a group) (24). The fact that we found higher rates of one and two or more antisocial behaviors for males may be a consequence of our measure of antisocial behavior, and not a basic difference in rates of antisocial behaviors between adolescent boys and girls, per se. Boys do have higher rates of overt aggression and violent offending throughout the life span (25), but it has been argued that by the adolescent years, the male predominance in antisocial behavior is no longer statistically significant if both overt and covert antisocial behaviors are considered (26). This means that the girls reporting the antisocial behaviors measured in this study may be more deviant, relative to same sex norms, than the boys. Future research should address whether a history of abuse also increases risk for covert antisocial behavior. The combination of abuse and sexual molestation was consistently more strongly associated with both antisocial and self-destructive behaviors in adolescence than nonsexual abuse or molestation alone. There is an increase in suicidal behavior among youth who show both antisocial and depressive features (27). When a child is depressed and aggressive, aggressive impulses tend to be directed against the self. This study has several limitations. First, whereas the response rate for the students was high (96% of students whose schools agreed to participate in the survey and who were present on that day completed questionnaires), only 22% of the schools asked to participate agreed to administer the survey. Therefore, it is possible that the results reported here are
JOURNAL OF ADOLESCENT HEALTH Vol. 24, No. 3
not generalizable to other groups of adolescents. However, we did not detect differences between the schools which participated and those which refused to participate on a number of parameters. While we did not have the data to detect subtle differences, the fact that there are not large differences indicates that the results may be representative of other adolescents in Washington public schools. Second, it should also be noted that if students who are absent or drop out from school are more likely than others to have been abused, then the prevalence rates in the current study may be underestimates of the actual numbers abused. Students who are in schools other than public schools (such as private schools or alternative schools outside the public school system) may also have different rates of abuse compared to students in public schools. Prevalence rates of other factors measured in this report may also have been influenced by dropouts, absentees, and the inability to assess students outside of the public school system, as the survey only reached students who were present in public schools. However, only about 7% of Washington students are in private schools. Third, the use of self-reports introduces the possibility that students were intentionally or unintentionally providing false responses. To minimize the possibility of including questionnaires with false responses, we eliminated questionnaires for dishonesty, internal inconsistencies, and reporting use of a fictitious drug. The anonymity of the self-reports might be expected to mitigate these problems, especially for boys. Widom and Morris (7) reported that only 16% of men known to have been sexually abused in childhood disclosed this in a face-to-face interview, compared to 60% of women. In addition, the prevalence rates obtained in this study for alcohol and cigarette use were similar to results from national studies using self-administered questionnaires in a school setting (19). We provide validity evidence for self-reports of abuse in the accompanying report (15). Fourth, because of the school setting, we were not able to ask important questions about the abuse, such as the relationship to the perpetrator, age of first occurrence, number of occurrences, and details about the nature of the abuse. Therefore, we were not able to address the question of chronicity of abuse, which we found to be important in a related study (10). Also, it is possible that some of the students who were coded as molested were reporting experiences with peers instead of abuse by someone older. Finally, the cross-sectional nature of the study does not
March 1999
allow firm statements about causality. This is discussed more completely in the accompanying report (15). The behavior most strongly associated with abuse history was injurious suicide attempts. Maltreatment may be a predominant risk factor for serious suicidal behaviors. By comparison, behaviors which involve pleasurable activities with peers, such as substance use, or activities with an illegal but goal-oriented focus, such as some delinquent activities (e.g., theft) may have relatively more alternative risk factors. For example, while relatively small numbers of adolescents may be influenced by peers to attempt suicide, relatively large numbers may be influenced by peers in deciding whether to drink in settings such as parties. Further research is needed to identify factors which differentiate among maltreated adolescents who display different needs and behaviors. Adolescents who display antisocial behavior in particular are likely to be dealt with in the juvenile justice system and to have their own history of being abused overlooked (28). Legal and disciplinary problems are, in turn, a risk factor for suicide in the next year (4). Both effective programs aimed at intervening with youth who are at risk for suicide (4,29) and antisocial behavior and programs aimed at preventing the occurrence of maltreatment are needed. The evidence to date suggests that effective programs should include parents, and in particular target the discord between parents and children (4,30 –32). This research was presented in part at the annual meeting of the Society of Behavioral Medicine, New Orleans, March, 1998.
ABUSE AND ADOLESCENT PROBLEM BEHAVIORS
8.
9.
10.
11.
12.
13. 14.
15.
16.
17.
18.
19.
20.
21.
References 1. Hinden BR, Compas BE, Howell DC, Achenbach TM. Covariation of the anxious-depressed syndrome during adolescence: Separating fact from artifact. J CCP 1997;65:6 –14. 2. Silverman AB, Reinherz HZ, Giaconia RM. The long-term sequelae of child and adolescent abuse: A longitudinal community study. Child Abuse Negl 1996;20:709 –23. 3. Brent DA, Perper JA, Moritz G, et al. Psychiatric risk factors for adolescent suicide: A case-control study. J Am Acad Child Adolesc Psychiatry 1993;32:521–9. 4. Brent DA, Perper JA, Moritz G, et al. Stressful life events, psychopathology, and adolescent suicide: A case-control study. Suicide Life-Threat Behav 1993;23:179 – 87. 5. Ireland T, Widom CS. Childhood victimization and risk for alcohol and drug arrests. Int J Addict 1994;29:235–74. 6. Widom CS, Ashley AM. Criminal consequences of childhood sexual victimization. Child Abuse Negl 1994;18:303–18. 7. Widom CS, Kuhns JB. Childhood victimization and subsequent risk for promiscuity, prostitution, and teenage preg-
22.
23.
24. 25.
26. 27.
28.
171
nancy: A prospective study. Am J Public Health 1996;86:1607– 12. Nelson DD, Higginson GK, Grant-Worley JA. Using the Youth Risk Behavior Survey to estimate prevalence of sexual abuse among Oregon high school students. J Sch Health 1994;64: 413– 6. Briere J, Runtz M. Differential adult symptomatology associated with three types of child abuse histories. Child Abuse Negl 1990; 14:357– 64. Spieker SJ, Bensley L, McMahon RJ, et al. Sexual abuse as a factor in child maltreatment by adolescent mothers of preschool aged children. Dev Psychopathol 1996;8:497–509. Orpinas PK, Basen-Engquist K, Grunbaum JA, et al. The co-morbidity of violence-related behaviors with health-risk behaviors in a population of high school students. J Adolesc Health 1995;16:216 –25. Valois RF, McKeown RE, Garrison CZ, et al. Correlates of aggressive and violent behaviors among public high school adolescents. J Adolesc Health 1995;16:26 –34. Lewinsohn PM, Rohde P, Seeley JR. Psychosocial risk factors for future suicide attempts. JCCP 1994; 62:297–305. Garrison CZ, Addy CL, Jackson KL, et al. A longitudinal study of suicidal ideation in young adolescents. J Am Acad Child Adolesc Psychiatry 1991;30:597– 603. Bensley LS, Spieker SJ, Van Eenwyk J, Schoder J. Self-reported abuse history and adolescent problem behaviors. II. Alcohol and drug use. J Adolesc Health 1999;24:173–180. Widom CS, Shepard RL. Accuracy of adult recollections of childhood victimization: Part 1. Childhood physical abuse. Psychol Assess 1996;8:412–21. Widom CS, Morris S. Accuracy of adult recollections of childhood victimization: Part 2. Childhood sexual abuse. Psychol Assess 1997;9:34 – 46. Brener ND, Collins JL, Kann L, et al. Reliability of the Youth Risk Behavior Survey questionnaire. Am J Epidemiol 1995;141: 575– 80. Center for Disease Control and Prevention. Reliability of the Youth Risk Behavior Surveillance—United States, 1995. MMWR 1996;45:1– 84. Shah BV, Barnwell BG, Bieler GS. SUDAAN user’s manual, Release 7.5. Research Triangle Park, NC: Research Triangle Institute, 1997. Ananth CV, Kleinbaum DG. Regression models for ordinal responses: A review of methods and applications. Int J Epidemiol 1997;26:1323–33. Cicchetti D, Carlson V, eds. Child Maltreatment: Theory and Research on the Causes and Consequences of Child Abuse and Neglect. Cambridge, England: Cambridge University Press, 1993. Dishion TJ, French DC, Patterson GR. The development and ecology of antisocial behavior. In: Cicchetti D, Cohen DJ, eds. Development of Psychopathology, Vol. 2. New York: Wiley, 1995:421–71. Crick NR, Grotpeter JK. Relational aggression, gender, and social-psychological adjustment. Child Dev 1995;66:710 –22. Moffitt TE. “Adolescence-limited” and “life-course-persistent” antisocial behavior: A developmental taxonomy. Psychol Rev 1993;100:674 –701. Zoccolillo M. Gender and the development of conduct disorder. Dev Psychopathol 1993;5:65–78. Shaffi N, Carrigan S, Whittinghill JR, Derrick A. Psychological autopsy of completed suicide of children and adolescents. Am J Psychiatry 1985;142:1061– 4. Howing PT, Wodarski JS, Kurtz PD, Gaudin JM Jr. Maltreatment and the School-Age Child: Developmental Outcomes and System Issues. Binghamton, NY: Haworth Press, 1993.
172
SOUTHWICK BENSLEY ET AL.
29. Eggert LL, Thompson EA, Herting JR. A measure of adolescent potential for suicide (MAPS): Development and preliminary findings. Suicide Life-Threat Behav 1994;24:359 – 81. 30. Hengeller S, Rodick J, Borduin C, et al. Multisystemic treatment of juvenile offenders: Effects on adolescent behavior and family interaction. Dev Psychol 1986;22:132– 41.
JOURNAL OF ADOLESCENT HEALTH Vol. 24, No. 3
31. Kazdin A, Bass D, Siegel T, Thomas C. Cognitive-behavior therapy and relationship therapy in the treatment of children referred for antisocial behavior. JCCP 1989;57:522–35. 32. Patterson G, Chamberlain, P, Reid J. A comparative evaluation of a parent-training program. Behav Ther 1982;13: 638 –50.