ARCHIVED: Chapter 2: Coping with Becoming a Teen: When You Have Been Maltreated as a Child

To better understand the prevalence, risk, and underlying mechanisms involved in substance use behaviours of youth involved with child welfare, the current report seeks to address the following:

• Objective 1: Using parallel substance use items drawn from the 2005 Ontario Student Drug Use Survey (OSDUS) and the MAP, what are the overall rates of substance use in the MAP sample, and how do rates of substance use among MAP youth compare with OSDUS youth who do not report some level of lifetime involvement with Ontario child welfare?
• Objective 2: Among youth in the MAP sample, what is the relationship between specific types of childhood maltreatment experienced and types of substance use? Is there an association among multiple forms of maltreatment and particular substances, or particular types of maltreatment and multiple substance types used?
• Objective 3: How do rates of substance use differ for male and female youth in the MAP versus the OSDUS dataset? Is substance use associated with age among male and female youths in the MAP sample?
• Objective 4: Given the significant link between childhood maltreatment and symptoms of PTSD, and the negative long-term consequences associated with PTSD symptoms, is there a significant gender- moderated mediation of the relationship between childhood maltreatment and substance use by PTSD symptomatology considering sub-clinical rather than disorder levels?

Method

The 2005 cycle of the OSDUS asked the question about lifetime involvement with Ontario child welfare, as a result we were able to compare findings on the Ontario Maltreatment and Adolescent Pathways (MAP) youth with findings on the 2005 OSDUS youth who were not welfare involved. This comparison is based on the MAP Year-1 assessment point (collected March 2004 to December 2007) that utilized the OSDUS questionnaire package.

MAP Year-1 Sample

The MAP longitudinal study collects data from youth (N = 388; 52% females) initially and every six months to a two-year follow-up point. The MAP study has a 70% recruitment rate, with 388 youth tested out of 554 eligible youth contacted. The study's retention rate was 83.5% at the one year testing point. The MAP study was approved by the University of Western Ontario (UWO) Research Ethics Board and the ethics committees at participating child protective agencies, and is reviewed annually by the UWO Research Ethics Board. Since risk behaviours typically start in early adolescence, mid-adolescence was the age range selected to examine ongoing risk behaviour rates. MAP teens were, on average, 15 years at intake (M = 15.67 years; SD = 1.08), and one-third indicated bi- or multi-racial background. Most teens were living away from their biological families (37% foster home, 27% group home, 14% independent living/other). Caseworkers were also surveyed for a subsample of the MAP participants; for this subsample, most youth experienced more than one form of substantiated maltreatment (sexual, physical, emotional abuse and/or neglect). MAP participants were more likely to have experienced sexual abuse if they were female than male. One year after the MAP testing, the OSDUS questionnaire package was administered to MAP participants, providing the basis for the comparisons of the prevalence of substance use between child welfare youth and youth from the general population. This report is based on 177 MAP youth (61% female), ranging in age from 15 to 19 years old (M = 16.80, SD = 0.99).

Measures of Maltreatment

For these analyses, two different approaches were used to capture maltreatment histories among MAP youth:

• caseworker assessments of substantiated maltreatment or risk for a subsample (n = 50); and
• youth self-reports of experiencing lifetime maltreatment, using the Childhood Trauma Questionnaire (Bernstein & Fink, 1998; see Appendix A), and the Childhood Experiences of Violence Questionnaire (CEVQ; Walsh et al., 2000; (Walsh, MacMillan, Trocmé, Jamieson, & Boyle, 2008; see Appendix B).

Caseworker Assessments

A short questionnaire was given to caseworkers when they referred youth to the MAP study. Workers were asked to assess each youth's risk, experience, and severity of maltreatment, as well as the youth's risk of substance use, intimate partner violence perpetration and/or victimization, risky sexual practices, and mental health problems. Caseworkers were also asked to assess youth global functioning based on diagnostic criteria outlined in the DSM-IV (American Psychiatric Association, 1994). Although caseworkers provided information on multiple domains, the analysis that follows includes only data regarding maltreatment status. This provides convergent validity to youth self- reports of maltreatment experience.

Youth Self-reports

As noted above, experiences of childhood maltreatment were assessed via a brief version of the Childhood Trauma Questionnaire (CTQ; Bernstein & Fink, 1998). This CTQ short form (Bernstein et al., 2003) assesses maltreatment via a standard stem (e.g., While you were growing up...). Participants rate 29 items on a 5-point scale (1 = never true, to 5 = very often true) across five subscales:

• emotional neglect;
• physical neglect;
• sexual abuse;
• physical abuse; and
• emotional abuse.

The CTQ does not capture witnessing domestic violence . Please see Appendix A for a sample of the items.

Two-week test-retest reliability of the CTQ for a MAP youth subsample (n = 52) was moderate [physical abuse (r = 0.64), sexual abuse (r = 0.52), emotional abuse (r = 0.70), emotional neglect (r = 0.63), and physical neglect (r = 0.56)], while internal validity or consistency was high, when considering Cronbach's alpha [physical abuse (α = 0.92), sexual abuse (α = 0.88), emotional abuse (α = 0.85), emotional neglect (α = 0.87), and physical neglect (α = 0.68)]. Youth reports and caseworker ratings of childhood maltreatment were significantly correlated in terms of physical abuse (r = 0.48), sexual abuse (r = 0.58), and physical neglect (r = 0.26). For emotional abuse and emotional neglect, youth reports and caseworker assessments were not significantly correlated.

The Childhood Experiences of Violence Questionnaire (CEVQ) (Walsh, MacMillan, Trocmé, Jamieson, & Boyle, 2008; Wekerle et al., 2006) also assessed self-reported maltreatment. The CEVQ assesses:

• physical abuse;
• sexual abuse;
• emotional abuse;
• witnessing domestic violence;
• peer-to-peer violence; and
• exposure to corporal punishment.

Domestic violence refers to violence involving the actions of parents, step-parents, or guardians, or actions involving a parent, step-parent, or guardian and another adult in the home. The CEVQ does not capture neglect experiences . This self-report measure queries age at time of maltreatment, frequency, outcome, and perpetrator characteristics (see

Appendix B for a sample of the items). The CEVQ demonstrates good test-retest reliability (kappas ranging from 0.61 - 0.91) and validity-as determined by clinician assessment-with estimates falling in a similar range (kappas for physical and sexual abuse were 0.68 and 0.74, respectively).

Two-week test-retest reliability of the CEVQ among the MAP sample ranged from moderate to high [physical abuse (r = 0.88), sexual abuse (r = 0.71), emotional abuse (r = 0.51)], while internal validity or consistency also ranged from moderate to high, when considering Cronbach's alpha [physical abuse (α = 0.82), sexual abuse (α = 0.70), emotional abuse (α = 0.68)]. The CEVQ is used to provide more detailed description of maltreatment (or failure to protect). It can indicate where caregivers are the perpetrators, which would be the chief concern in child welfare cases.

Finally, a modification was made to the MAP CTQ, adding one item (neglect) to complete the categories of maltreatment. The CTQ had three items (sexual abuse, physical abuse, emotional abuse) to assess youth perceptions or self-labeling of maltreatment. Using the same CTQ question stem ("While growing up") and response options, youth responded to the items:

• "I believe that I was sexually abused;"
• "I believe that I was physically abused;"
• "I believe that I was emotionally abused;" and
• "I believe that I was neglected."

Prior research (e.g., Wekerle et al., 2001) indicated that youth perceptions were associated with the level of PTSD symptomatology reported. Thus, it may be important to simultaneously consider multiple measures of maltreatment: the official threshold based on caseworker assessment, the youth's recollected experiences, and the youth's interpretation of his/her experiences. In the analyses and tables that follow, both the CEVQ and CTQ are included to capture the full range of youth self-reported maltreatment history and its relationship to substance use variables.

Trauma Symptomatology

PTSD symptomatology was assessed with the Trauma Symptom Checklist for Children (TSCC) (Briere, 1996), a common measurement tool in maltreatment research. The TSCC is a 54-item self-report measure consisting of six clinical scales (anxiety, depression, anger, PTSD, dissociation, and sexual concerns) and two validity scales (under-response and hyper-response). While the TSCC does not have a set time reference, it does ask whether symptoms occur at a frequency of "almost all of the time," "lots of time," "sometimes," or "never." The measure was normalized on teens and was intended to evaluate children who have experienced traumatic events (see Appendix C for a sample of the items). Reliability is high (internal consistency, α = 0.82 - 0.89) and good convergent, discriminant, and construct validity were established.

The two-week test-retest reliability of the MAP subsample on the TSCC was moderate (r = 0.50) and internal validity was very high (α = 0.97). The TSCC provides a clinical cutoff for each subscale. In keeping with developmental traumatology hypotheses on the importance of sub- clinical symptoms, a total score of any clinical elevation among the six clinical subscales was used. From 4% to 20% of MAP youth showed clinical elevations, with significantly more females than males above the clinical cut-off on the anxiety and sexual concerns subscales.

Specific symptoms may provide some further ideas regarding gender differences; however, these must be considered with caution as they do not capture the entire subscale, syndrome, or diagnosis, and items are not reported for teens that have not experienced maltreatment.

By TSCC item (see Appendix D), MAP males reported significantly (p<0.05) greater frequency than females on:

• Touching my private parts too much;
• Thinking about touching other people's private parts; and
• Having sexual feelings in my body.

By TSCC item (see Appendix D), MAP females reported significantly (p<0.05) greater frequency than males on:

• Not trusting people because they might want sex;
• Getting afraid or upset when I think about sex;
• Going away in my mind, trying not to think;
• Feeling dizzy;
• Bad dreams or nightmares;
• Remembering things that happened that I didn't like;
• Remembering scary things;
• Feeling scared of men;
• Remembering things that I don't want to remember;
• Feeling mean;
• Feeling like I hate people;
• Feeling mad;
• Feeling lonely;
• Feeling sad or unhappy;
• Crying;
• Wanting to hurt myself;
• Washing myself because I felt dirty inside;
• Feeling afraid something bad might happen;
• Getting scared all of a sudden and not knowing why;
• Being afraid of the dark; and
• Worrying about things.

These item-by-item findings indicate that male youths who have experienced maltreatment are more likely to have difficulties with sexual feelings, whereas females seem to experience more fear and anger-directed at both themselves and others. These findings suggest hypotheses related to a gendered experience of PTSD and may indicate future research directions. Also, the findings support the moderation approach in considering all analyses separately by gender.

Substance Use

At one-year follow-up testing, MAP youth (grade 7-12 students across Ontario) were administered the 2005 Ontario Student Drug Use Survey (OSDUS; Adlaf & Paglia-Boak, 2005) questionnaire package. This instrument collects information on different health risk behaviours, such as substance use (see Appendix E for a sample of the items).

Using the same instrument in this epidemiological study of youth involved in child welfare provides the opportunity to compare the prevalence of different health risk behaviours, such as substance use between youth involved with child welfare and youth from the general population. Details on the OSDUS/OSDUHS study and questionnaire are covered in the following section.

Ontario Student Drug Use Survey

Sample

The OSDUHS, which changed its name from the Ontario Student Drug Use Survey (OSDUS) in 2007, has been conducted every two years since 1977 and is funded by the Centre for Addiction and Mental Health (CAMH). The survey samples grade 7-12 students from 42 school boards and 137 schools across Ontario, using a full-probability two-stage (school, class), stratified (region and school type), cluster design. The resulting sample represents about 975,200 students, with those from Northern Ontario being over sampled. The school, class, and student participation rates remain at about 90%, 71%, and 72%, respectively. All aspects of the OSDUS research were approved by the University of Toronto and CAMH's joint Research Ethics Board. To maximize questionnaire coverage, the OSDUS uses random half-samples for selected questionnaire items.

The OSDUS questionnaire is administered to youth in their high school classrooms by staff of the Institute for Social Research, York University.

Although data were collected from a representative sample of grade 7-12 Ontario students (N exceeds 7,000 students), the present analysis includes only those youth aged 15 to 19 (M = 16.32, SD = 1.04) who indicated they had never been involved with the child welfare system, resulting in a sample size of 3,505, 53% of it female. This sample was used to allow for comparisons of same-age groupings in light of age fluctuations in substance use, and to have a clearer estimate of substance use among youth not involved with child welfare (as system involvement itself may affect use).

Measures

The OSDUS questionnaire covers several different health risk behaviours; however, the current focus is primarily on substance use practices. OSDUS bases its substance use items on other major surveillance initiatives, such as the American annual high school Monitoring the Future Study (www.monitoringthefuture.org; Johnston et al., 2007). Substance use items include age of substance use initiation, frequency, quantity per use, and the severity/problematic consequences of substance use behaviours. Substances of interest include alcohol and cannabis, as well as "other drugs," which, because of their low prevalence of use, were collapsed into a single category in the analyses below. "Other drugs" include glue and solvents (for sniffing), barbiturates, heroin, methamphetamines, stimulants without a doctor's prescription (other than cocaine), tranquilizers without a doctor's prescription, LSD, PCP, hallucinogens other than LSD or PCP, cocaine, crack cocaine, ecstasy, and methylphenidate (Ritalin) without a doctor's prescription.

Comparison between Samples

Both the MAP and the OSDUS used randomly selected epidemiological samples, with comparable demographics. The MAP Year-1 sample consisted of urban child welfare involved youth (61% female) 15 to 19 years of age (M = 16.80, SD = .99). The subsample from the 2005 OSDUS report consisted of non-child welfare involved youth (based on self-report; 53% female) in the same age range (M = 16.32, SD = 1.04). The substance use items compared in the current report are exactly matched across the two samples.

Analysis Plan

To assess the comparability between MAP and OSDUS youth on a single, given substance use item, relative risk ratios were calculated; the accompanying confidence intervals are provided. Relative risk is the risk of an event (or of developing a disease) relative to exposure. It is presented as a ratio of the probability of the event occurring in the exposed group versus the control (non- exposed) group. Relative risk is used frequently in the statistical analysis of binary outcomes where the outcome of interest has relatively low probability (Zhang & Yu, 1998). In a simple comparison between an experimental group and a control group, a relative risk (RR) of 1 means there is no difference in risk between the two groups. When relative risk presents as RR < 1, the event is less likely to occur in the experimental group than in the control group; RR > 1 means the event is more likely to occur in the experimental group than in the control group.

Whether a given relative risk can be considered statistically significant, that is, the confidence one has that a relative risk is not a consequence of chance, depends on the signal-to-noise ratio and the sample size. To this end, the 95% confidence intervals are provided for each relative risk computation. The inclusion of 1.00 within the confidence interval nullifies the relevance of the relative risk because it suggests that the relative risk is equally likely to be non-existent, or its effect is in the opposite direction.

To elucidate the likelihood of substance use when a particular maltreatment experience was reported (self- report or by a caseworker) in the MAP study, the relative risk of (a) having been abused in a particular way and (b) perceiving oneself as being abused in a particular way for (c) using particular substances, is presented.

This report compares gender differences in substance use between MAP and OSDUS youth in two ways. First, the relative risk of substance use for MAP youth in relation to OSDUS youth was calculated separately for each gender. Second, the relative risk of substance use for males in relation to females was calculated separately.

To calculate the likelihood of (a) substance use, dichotomized as ever used in lifetime, ever used in past 12 months, and whether a frequent user (yes/no) given, and (b) the cumulative maltreatment score (i.e., the total number of types of maltreatment experienced), logistic regressions were used. These provided the odds ratios (OR) for (a) substance use given, and (b) cumulative maltreatment. This same method was used when considering the association between (a) the cumulative substance use score (i.e., the total number of types of substances used), and (b) the type of maltreatment experienced.

Finally, to identify the gender-specific mechanism that mediates the relation between childhood maltreatment and substance use, a series of hierarchical multiple regression analyses were conducted to determine whether the hypothesized PTSD-mediated model fit Baron and Kenny's (1986) criteria for statistical mediation. However, to assess how child welfare youths' report of different maltreatment items on the CTQ cluster together, a principal component factor analysis with varimax rotation was conducted. This was deemed a necessary first step as the CTQ was not normed on child welfare adolescents and, as such, the reported 5-factor structure may not be valid. These results showed that for child welfare males, the CTQ subscales were relevant. However, for females, a 4-factor solution was obtained, the difference being that physical abuse and emotional abuse items loaded together (see Appendix E). Consequently, factor scores for maltreatment were used in these multiple regression analyses. We followed the multiple regression analyses testing mediation with further mediation tests - specifically, the Goodman test to conduct a formal assessment of statistical mediation.