care, health and development Child: Original Article bs_bs_banner
doi:10.1111/cch.12233
Victimization and depression among youth with disabilities in the US child welfare system K. L. Berg,* C.-S. Shiu,† M. E. Msall‡§ and K. Acharya¶** *The College of Public Health, Temple University, Philadelphia, PA, USA †Social Work, University of Washington, Seattle, WA, USA ‡Developmental & Behavioral Pediatrics at University of Chicago Medicine, University of Chicago Medicine, Chicago, IL, USA §Kennedy Research Center on Intellectual and Developmental Disabilities, Comer Children’s Hospital, Chicago, IL, USA ¶Department of Disability and Human Development and Pediatrics, University of Illinois-Chicago, Chicago, IL, USA, and **Department of Health and Human Services, Leadership Education in Neurodevelopmental and Related Disorders Training Program (LEND), Chicago, IL, USA Accepted for publication 10 January 2015
Keywords childhood disability, CWS, mental health, victimization Correspondence: Kristin L. Berg, PhD, The College of Public Health, Temple University, 1700 N. Broad Street, Philadelphia, PA 19122, USA E-mail: [email protected] Written permission has been obtained from all authors to submit this manuscript. Reuse of datasets: This study employs the second National Survey of Child and Adolescent Well-Being (NSCAW II), a longitudinal study sponsored by the Office of Planning, Research and Evaluation, Administration for Children and Families (ACF), US Department of Health and Human Services (DHHS). KLB and C-SS had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
© 2015 John Wiley & Sons Ltd
Abstract Aim This study aimed to examine the prevalence of victimization among a United States-wide cohort of youth with disabilities (YWD) investigated for maltreatment in the child welfare system (CWS) and their correlation with mental health. Methods Data were drawn from baseline interviews in the second National Survey of Child and Adolescent Well-Being, a national representative survey of youth involved in the CWS. Interviews took place between 2008 and 2009 and included 675 youth, 11–17 years old and residing with biological families across 83 counties nationwide. The sample consisted of 405 females (60.1%) and 270 males (39.9%), mean age = 13.5 years. We identified YWD if they reported one or more physical or neurodevelopmental health condition (n = 247). Reported victimization experiences and Children’s Depression Inventory (CDI) scores were analysed using weighted regression analyses. Results One-quarter of YWD in the CWS reported three or more victimizations during the prior year compared with 19% of youth without disabilities. The odds of YWD reporting a one-unit increase in level of victimization was 75% higher (P < 0.05) than youth without disabilities. Prevalence of clinical depression was significantly higher among YWD (14 vs. 5.5%; P < 0.05). Unlike youth without disabilities, the odds of clinical depression were 92% higher for every one-unit increase in victimization among YWD, controlling for covariates (P < 0.05). Of CWS-involved youth who reported three or more victimizations, 24.4% of YWD and 2.2% of non-disabled youth had CDI scores in the clinical range. Conclusion YWDs in the US CWS are at high risk of experiencing victimization and clinical depression. Our findings suggest that health professionals need to screen CWS-involved YWD for multiple forms of victimization, and develop and implement trauma-informed services that target the mental health sequelae that may jeopardize their independence in adulthood.
989
990 K.L. Berg et al.
Introduction Youth with disabilities (YWD) are overrepresented in the US child welfare system (CWS) (Heneghan et al. 2013). The results of several studies suggest that once within the CWS, YWD are more vulnerable to poor outcomes, including mental health disorders (Raviv et al. 2010; Heneghan et al. 2013). Among YWD in the general US population, the adverse impact of untreated mental illness are diverse and far reaching (Martorell et al. 2009; Einfeld et al. 2011; Mayes et al. 2011; Robb 2013). YWD who have co-morbid mental illness are less likely to be able to complete activities of daily living and to successfully transition to adulthood compared with YWD without mental illness (Einfeld et al. 2011; Robb 2013). Co-morbid mood disorders have a greater impact on the placement and developmental trajectories of YWD than the disability itself (Robb 2013). Among typically developing youth, data have linked victimization and violence exposure to a variety of mental health disorders such as depression and anxiety (Martinez & Richters 1993; Wise et al. 2001; Edwards et al. 2003). Specifically, studies have documented a linear dose–response relationship between exposure to violence and lifetime major depressive disorder (Wise et al. 2001; Edwards et al. 2003). However, the majority of studies of co-morbid mental illness among YWD have narrowly focused on the role of neurobiological factors and have not considered victimization and other trauma as contributing factors (Feinstein & Reiss 1996; Zimmerman et al. 1998). Research on the victimization experiences and mental health of YWD in the CWS is even more limited. Although several studies have linked disability status in the CWS to maltreatment re-report and parental physical assault, none have examined mental health effects (Dakil et al. 2011; Helton & Cross 2011). Available data on the mental health of CWS-involved youth have largely treated disability status as an additional risk factor for psychological distress and have not considered victimization as a contributing factor for this population (Raviv et al. 2010; Heneghan et al. 2013). Understanding the role of disability and victimization in the mental health of CWS youth is important because YWD make up a disproportionate percentage of those in child protective services; youth in the CWS typically experience a higher number of adversities and victimizations; and being repeatedly victimized is a robust predictor of mental illness and other adverse outcomes (Reading 2006; Turner et al. 2006; Finkelhor et al. 2007; Greeson et al. 2009; Sullivan 2009; Raviv et al. 2010; Tanaka et al. 2011). More significantly, effective treatment for psychological distress due to complex trauma differs from treatment approaches for mental illness with no history of complex
© 2015 John Wiley & Sons Ltd, Child: care, health and development, 41, 6, 989 –9 99
trauma (Griffin et al. 2011). The poor outcomes observed among YWD in the CWD may be driven, in part, by untreated multiple victimization and complex trauma, and its unique mental health sequelae. Our study is the first to explore prior year exposure to victimization among YWD in contact with the CWS and its association with their current mental health status. Our objectives were to (1) compare exposure to victimization during the previous 12 months among youth with and without disabilities in the US CWS; and (2) evaluate whether disability status moderates the relationship between level of victimization and clinical depression among CWS-involved youth.
Methods Design and analytic sample We employed the second National Survey of Child and Adolescent Wellbeing (NSCAW II), a national representative, longitudinal survey of youth in the US CWS (Dolan et al. 2011). Using a two-stage combined stratification and cluster design, NSCAW II sampled 5873 children aged 0–17 years across 83 counties who had contact with the CWS within a 15-month interlude beginning in February 2008. Data collection involved in-person interviews and assessments with CWS caseworkers, caregivers and children (Dolan et al. 2011).
Survey design and assessment procedures For this study, the sample was restricted to all youth aged 11–17 years with a report of maltreatment and who remained in-home with their biological parent(s) (n = 675). This study excluded youth in all other care arrangements due to systematic differences between youth who remain with parents and those placed in out-of-home care (Barth et al. 2006). All study analyses were drawn from baseline interviews with youth and parents collected between March 2008 and September 2009 (Dolan et al. 2011). All procedures involved in analysing the NSCAW II were approved by the institutional review board at the Research Triangle Institute, and all statistical analyses using NSCAW II data were approved by the University of Chicago institutional review board.
Measures The following section describes key study variables. Refer to Appendix A1 for a more detailed overview of covariates.
Disability In NSCAW II, parents identified whether their child had a disability using a list of diagnostic medical conditions that included both
Victimization and depression among YWD
chronic and temporary disabilities. Using the diagnostic medical conditions, three possible disability categories were created based on clinical and epidemiological guidelines: (1) physical; (2) neurodevelopmental; and (3) learning/attention/behavioural health (Stein et al. 2013). Youth with physical/health disabilities were identified having one or more of the following conditions: severe persistent asthma, diabetes, heart problems including congenital heart disease, haematological disorders, cystic fibrosis, arthritis, joint problems, human immunodeficiency virus/ acquired immune deficiency syndrome and other complex health impairments. Youth with neurodevelopmental disabilities were identified as having one or more of the following conditions: visual impairment/blindness, hearing impairment/deafness, autism/autism spectrum disorder (ASD), Down syndrome, developmental delay/intellectual disability, cerebral palsy, muscular dystrophy and epilepsy/seizure disorder. Aware of the clinical overlap between neurodevelopmental and physical disabilities, we used the primary diagnosis to classify disability type. Because of the small sample size and the unequal distribution of diagnostic conditions across the sample, separate regression analyses by disability type were not feasible. In order to generate adequate statistical power, youth classified as having either a physical/health and/or neurodevelopmental disability were grouped together as YWD. Youth with only learning/attention/emotional–behavioural health disabilities were excluded from the operationalization of disability due to the high prevalence of learning/attention/ emotional–behavioural disabilities among youth in the CWS, potential confounding effects of emotional/behavioural disabilities on mental health outcomes and the significant evidence base that identifies learning/emotional/behavioural disabilities as relatively common sequelae to childhood maltreatment (Conroy & Brown 2004; Stahmer et al. 2005). As documented in the literature, the experience of victimization may contribute in a causal way to adolescent disability and not only its reverse. Given this challenge, disability was operationalized in such a way as to minimize confounds. Many of the disabilities selected for this study (ASD, mental retardation, etc.) are by definition developmental disabilities that are present at birth. Moreover, such disabilities, including asthma, arthritis and Down syndrome, are generally not considered part of the sequelae of child maltreatment/victimization. However, to ensure accuracy, a regression analysis was conducted including an additional dummy variable containing those youth with learning/attention/ emotional–behavioural disabilities. The results of the regression did not yield significant differences with regard to outcomes. Using the above criteria resulted in a sample of YWD (n = 247). For the distribution of diagnostic conditions
991
and estimated functional limitations in this population, see Appendix A2.
Victimization Assessment of victimization during the previous 12 months was based on youth self-report. Our victimization index consisted of eight items across four types of victimization: (1) physical abuse in the home; (2) emotional abuse in the home; (3) physical victimization outside of the home; and (4) threat or assault with a weapon. These items were selected based on the extensive data that link these types of victimization to mental health conditions and other negative outcomes among youth (Martinez & Richters 1993; Edwards et al. 2003; Reading 2006; Margolin et al. 2010). All victimization items included were dichotomous (yes/no) and recoded as ‘present’ (1) or ‘absent’ (0). Items from the Parent-Child Conflict Tactic Scale 2 (CTS 2; Cronbach’s α = 0.55) (Straus et al. 1998) were used to measure past year exposure to abuse in the home. Physical abuse was assessed using items from the Severe Assault Scale: a parent/ adult in the home (1) choked the youth; (2) beat/repeatedly hit the youth; or (3) scalded or burned him or her. Emotional abuse was measured using items from the Psychological Aggression Scale: a parent/adult in the home (1) called the youth dumb, lazy or some other name; or (2) swore at him or her. Outside of the home, victimization was assessed by the Child Health and Illness Profile-Adolescent Edition (CHIP-AE; Cronbach’s α > 0.70) (Starfield et al. 1995). The CHIP-AE is a self-administered questionnaire to assess injury and disease among youth age 11–17 years. This instrument measured past year exposure to victimization by asking whether youth were physically injured by a friend/classmate/neighbourhood kid/ stranger in either school, neighbourhood, mall, friend’s home or other setting. Two items assessed prior year exposure to threat or assault with a weapon: (1) the CTS-2 asked whether an adult with whom the youth lived threatened him or her with a knife or gun; (2) the CHIP-AE asked youth about injuries due to ‘a gunshot wound or stab wound’. For our analysis, responses to each of the eight items were recoded (0,1) and summed for each youth, with total scores ranging from 0 to 8. Based on the distribution of victimizations, we recoded our victimization measure into four levels of exposure over the previous year: (0) no victimizations; (1) one victimization; (2) two victimizations; (3) three or more victimizations. The distribution for our past year victimization variable was as follows: 39.5% of youth reported no victimization; 24.3% reported one; 15% reported two and 21.3%
© 2015 John Wiley & Sons Ltd, Child: care, health and development, 41, 6, 989 –9 99
992 K.L. Berg et al.
reported three or more victimizations. Different levels or dosages of victimization have been used effectively across research studies (Appleyard et al. 2005). Statistical analyses revealed that our victimization range (0, 1, 2, 3+) was clinically and statistically meaningful in terms of depression outcomes. For the distribution of types of victimization, see Appendix A3.
Mental health Youth depression was assessed by the Children’s Depression Inventory (CDI; Cronbach’s α = 0.71–0.86) (Kovacs 1984). The CDI is the most commonly used self-report measure of depression in youth, with normative data available from school, psychiatric and paediatric populations (Smucker et al. 1986). Based on CDI manual guidelines, youth were classified as clinically depressed if they scored ≥90th percentile for their age and sex in the normative population, a threshold that has been used in other epidemiological studies of youth depression (Smucker et al. 1986; Hyphantis et al. 2012). By these criteria, 9% of CWS-involved youth reported depression (Cronbach’s α for sample = 0.87).
Covariates Demographic variables in this study included child, family and community factors. Key demographic variables included youth age, race, gender, parent age, parent education, marital status, household poverty, neighbourhood social disorganization and urban/rural residence. Additional variables of interest were youth reported closeness to parent, parent-reported intimate partner violence, parental substance abuse and parent physical and mental health status. These variables were selected because research suggests that victimized youth are often in families and communities characterized by the aforementioned factors, which contributes to victimization risk and/or exacerbate its effects on the health/well-being of youth (Cicchetti 2004; Raina et al. 2005; Sidebotham & Heron 2006; Murphy et al. 2007; Turner et al. 2012; Emerson & Brigham 2014; see Appendix A1 for operational definitions).
Analyses Statistical procedures in STATA (version 12, College Station, TX, USA) were employed to account for complex survey design and reweighting. Because the sample was limited to youth residing with biological parents, domain analysis was applied to readjust for sample weights, permitting generalization to the
© 2015 John Wiley & Sons Ltd, Child: care, health and development, 41, 6, 989 –9 99
national sample of adolescents in the CWS who remain with parent(s). Weighted ordinal regression was utilized to estimate the odds of reporting level of victimization by disability status. This technique estimates the probability of experiencing the highest level of cumulative victimization (≥3) and all the lesser victimization levels. In this model, odds ratios were calculated in terms of unit change (the likelihood of a one-unit increase in victimization level). Weighted logistic regression analyses were employed to examine the effects of disability status, victimization level and other covariates in predicting clinical depression among CWS-involved youth. Logistic regression with the interaction term was utilized to test whether disability status moderated the relationship between level of victimization and clinical depression. Huber-White estimator was used to calculate standard error.
Results Population characteristics The average age of CWS-involved youth was 13.5 years. Approximately 60% were female and nearly half (44.7%) were White (Table 1). Over three-fourth resided in urban settings and 18% lived above 200% of the federal poverty level. Approximately 38% (n = 247) of youth in contact with the CWS had disabilities, YWD in the CWS did not differ significantly from youth without disabilities in terms of gender, age and poverty. However, as a group, YWD were more likely to reside in nonurban settings (30.3 vs. 19.8%; P ≤ 0.01), occupy neighbourhoods with moderate levels of social disorganization (46.9 vs. 32.6%; P < 0.05) and be cared for by primary parents with both higher levels of clinical depression (37 vs. 26.6%; P < 0.05), and lower levels of physical health (P < 0.05).
Victimization Among youth in the CWS, prior year exposure to victimization was relatively common. Although approximately 39.5% reported no victimization events over the prior 12 months, 25, 15 and 21.3% reported one, two or three or more victimization events over the same period, respectively. After incorporating covariates, the odds of reporting a one-unit increase in level of victimization was 75% higher for CWS-involved YWD (P < 0.05) versus youth without disabilities (Table 2). Lower odds of reporting a one-unit increase in level of victimization were associated with optimal closeness to the primary parent [odds ratio (OR) = 0.26; P < 0.01] and older parent age (OR = 0.95; P < 0.05) while higher odds of a one-unit increase was
Victimization and depression among YWD
993
Table 1. Description of study population Disability groups
Sample size Estimated population Level of victimization, % 0 1 2 3 or more Depression, yes % Demographic factors Age† Gender, female % Race/ethnicity % White Black Hispanic Others Residency, urban % Closeness to parent, optimal % Parent and family factors Age† Social support† Health status† Gender, female % Education level % HS Poverty level % 200% Partner status, partnered % Alcohol abuse, yes % Depression, yes % IPV % No 1–2 times >3 times Neighbourhood social disorganization % Low Median High
Population
No disability
Any disability
675
428
247
F-test
P = 0.617 39.45 24.26 15.04 21.25 8.60
41.50 24.18 15.28 19.04 5.50
36.05 24.38 14.66 24.92 13.80
P = 0.020
13.48 ± 3.2 60.17
13.41 ± 4.0 61.96
13.58 ± 5.0 57.20
P = 0.478 P = 0.398
44.67 17.90 27.67 9.76 76.26 50.50
45.96 17.11 27.32 9.61 80.23 48.10
42.51 19.22 28.25 10.01 69.66 54.70
P = 0.922 P = 0.020
37.72 ± 9.0 3.67 ± 2.0 46.00 ± 21.5 89.68
37.48 ± 10.7 3.70 ± 2.7 47.23 ± 23.2 88.02
38.10 ± 11.2 3.61 ± 2.2 44.04 ± 33.4 92.33
P = 0.253 P = 0.467 P = 0.036 P = 0.258
30.62 34.57 34.80
30.70 36.14 33.16
30.49 31.98 37.53
P = 0.701
19.67 31.23 25.93 17.74 38.60 6.69 30.56
17.23 31.79 26.43 18.34 38.75 6.52 26.64
23.73 30.30 25.10 16.75 38.35 6.95 37.07
P = 0.645
70.55 9.92 19.53
71.45 9.00 19.55
69.05 11.45 19.50
P = 0.738
43.17 37.98 13.70
44.99 32.62 15.44
40.16 46.88 10.81
P = 0.021
P = 0.007 P = 0.254
P = 0.941 P = 0.869 P = 0.018
†For continuous variables, data were presented as mean ± standard deviation. HS, high school; IPV, intimate partner violence.
correlated with older youth age (OR = 1.3; P < 0.01) and high neighbourhood social disorganization (OR 1.8; P < 0.05).
Clinical depression In the CWS, 9% of youth reported clinical depression. Results of bivariate analysis indicated that YWD were more than twice as
likely to report clinical depression in comparison with their typically developing peers (13.8 vs. 5.5%; P < 0.05). The relationship between disability and clinical depression persisted despite incorporating child and family factors. The odds of reporting clinical depression were four times higher for YWD, controlling for all other variables (OR 3.975; P < 0.01) (Table 3). Several additional covariates also significantly predicted clinical
© 2015 John Wiley & Sons Ltd, Child: care, health and development, 41, 6, 989 –9 99
994 K.L. Berg et al.
Table 2. Weighted ordinal logistic regression with level of victimization as the outcome Model 1 OR Disability Yes (vs. no) Children’s characteristics Age Gender Male (vs. female) Race/ethnicity Black (vs. white) Hispanic (vs. white) Others (vs. white) Closeness to parent Optimal (vs. non-optimal) Family characteristics Parent age Parent gender Parent alcohol abuse Yes (vs. no) Intimate partner violence 1–2 times (vs. 0) >2 times (vs. 0) Neighbourhood social disorganization Mid level (vs. low) High level (vs. low) Parent health status F-test P-value
1.288
Table 3. Disability, victimization level, selected covariates and clinical depression***,*
Model 2† SE 0.259
F(1, 72) = 1.58 P = 0.212
OR
Dichotomous depression† SE
1.749*
0.422
1.279**
0.086
0.978
0.206
1.652 1.451 1.448
1.138 0.547 0.720
0.257**
0.053
0.954* 2.124*
0.019 0.781
1.485
0.648
0.536 1.459
0.244 0.395
0.960 0.225 1.785* 0.446 1.006 0.007 F(27, 46) = 4.72 P < 0.000
*
doi:10.1111/cch.12233
Victimization and depression among youth with disabilities in the US child welfare system K. L. Berg,* C.-S. Shiu,† M. E. Msall‡§ and K. Acharya¶** *The College of Public Health, Temple University, Philadelphia, PA, USA †Social Work, University of Washington, Seattle, WA, USA ‡Developmental & Behavioral Pediatrics at University of Chicago Medicine, University of Chicago Medicine, Chicago, IL, USA §Kennedy Research Center on Intellectual and Developmental Disabilities, Comer Children’s Hospital, Chicago, IL, USA ¶Department of Disability and Human Development and Pediatrics, University of Illinois-Chicago, Chicago, IL, USA, and **Department of Health and Human Services, Leadership Education in Neurodevelopmental and Related Disorders Training Program (LEND), Chicago, IL, USA Accepted for publication 10 January 2015
Keywords childhood disability, CWS, mental health, victimization Correspondence: Kristin L. Berg, PhD, The College of Public Health, Temple University, 1700 N. Broad Street, Philadelphia, PA 19122, USA E-mail: [email protected] Written permission has been obtained from all authors to submit this manuscript. Reuse of datasets: This study employs the second National Survey of Child and Adolescent Well-Being (NSCAW II), a longitudinal study sponsored by the Office of Planning, Research and Evaluation, Administration for Children and Families (ACF), US Department of Health and Human Services (DHHS). KLB and C-SS had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
© 2015 John Wiley & Sons Ltd
Abstract Aim This study aimed to examine the prevalence of victimization among a United States-wide cohort of youth with disabilities (YWD) investigated for maltreatment in the child welfare system (CWS) and their correlation with mental health. Methods Data were drawn from baseline interviews in the second National Survey of Child and Adolescent Well-Being, a national representative survey of youth involved in the CWS. Interviews took place between 2008 and 2009 and included 675 youth, 11–17 years old and residing with biological families across 83 counties nationwide. The sample consisted of 405 females (60.1%) and 270 males (39.9%), mean age = 13.5 years. We identified YWD if they reported one or more physical or neurodevelopmental health condition (n = 247). Reported victimization experiences and Children’s Depression Inventory (CDI) scores were analysed using weighted regression analyses. Results One-quarter of YWD in the CWS reported three or more victimizations during the prior year compared with 19% of youth without disabilities. The odds of YWD reporting a one-unit increase in level of victimization was 75% higher (P < 0.05) than youth without disabilities. Prevalence of clinical depression was significantly higher among YWD (14 vs. 5.5%; P < 0.05). Unlike youth without disabilities, the odds of clinical depression were 92% higher for every one-unit increase in victimization among YWD, controlling for covariates (P < 0.05). Of CWS-involved youth who reported three or more victimizations, 24.4% of YWD and 2.2% of non-disabled youth had CDI scores in the clinical range. Conclusion YWDs in the US CWS are at high risk of experiencing victimization and clinical depression. Our findings suggest that health professionals need to screen CWS-involved YWD for multiple forms of victimization, and develop and implement trauma-informed services that target the mental health sequelae that may jeopardize their independence in adulthood.
989
990 K.L. Berg et al.
Introduction Youth with disabilities (YWD) are overrepresented in the US child welfare system (CWS) (Heneghan et al. 2013). The results of several studies suggest that once within the CWS, YWD are more vulnerable to poor outcomes, including mental health disorders (Raviv et al. 2010; Heneghan et al. 2013). Among YWD in the general US population, the adverse impact of untreated mental illness are diverse and far reaching (Martorell et al. 2009; Einfeld et al. 2011; Mayes et al. 2011; Robb 2013). YWD who have co-morbid mental illness are less likely to be able to complete activities of daily living and to successfully transition to adulthood compared with YWD without mental illness (Einfeld et al. 2011; Robb 2013). Co-morbid mood disorders have a greater impact on the placement and developmental trajectories of YWD than the disability itself (Robb 2013). Among typically developing youth, data have linked victimization and violence exposure to a variety of mental health disorders such as depression and anxiety (Martinez & Richters 1993; Wise et al. 2001; Edwards et al. 2003). Specifically, studies have documented a linear dose–response relationship between exposure to violence and lifetime major depressive disorder (Wise et al. 2001; Edwards et al. 2003). However, the majority of studies of co-morbid mental illness among YWD have narrowly focused on the role of neurobiological factors and have not considered victimization and other trauma as contributing factors (Feinstein & Reiss 1996; Zimmerman et al. 1998). Research on the victimization experiences and mental health of YWD in the CWS is even more limited. Although several studies have linked disability status in the CWS to maltreatment re-report and parental physical assault, none have examined mental health effects (Dakil et al. 2011; Helton & Cross 2011). Available data on the mental health of CWS-involved youth have largely treated disability status as an additional risk factor for psychological distress and have not considered victimization as a contributing factor for this population (Raviv et al. 2010; Heneghan et al. 2013). Understanding the role of disability and victimization in the mental health of CWS youth is important because YWD make up a disproportionate percentage of those in child protective services; youth in the CWS typically experience a higher number of adversities and victimizations; and being repeatedly victimized is a robust predictor of mental illness and other adverse outcomes (Reading 2006; Turner et al. 2006; Finkelhor et al. 2007; Greeson et al. 2009; Sullivan 2009; Raviv et al. 2010; Tanaka et al. 2011). More significantly, effective treatment for psychological distress due to complex trauma differs from treatment approaches for mental illness with no history of complex
© 2015 John Wiley & Sons Ltd, Child: care, health and development, 41, 6, 989 –9 99
trauma (Griffin et al. 2011). The poor outcomes observed among YWD in the CWD may be driven, in part, by untreated multiple victimization and complex trauma, and its unique mental health sequelae. Our study is the first to explore prior year exposure to victimization among YWD in contact with the CWS and its association with their current mental health status. Our objectives were to (1) compare exposure to victimization during the previous 12 months among youth with and without disabilities in the US CWS; and (2) evaluate whether disability status moderates the relationship between level of victimization and clinical depression among CWS-involved youth.
Methods Design and analytic sample We employed the second National Survey of Child and Adolescent Wellbeing (NSCAW II), a national representative, longitudinal survey of youth in the US CWS (Dolan et al. 2011). Using a two-stage combined stratification and cluster design, NSCAW II sampled 5873 children aged 0–17 years across 83 counties who had contact with the CWS within a 15-month interlude beginning in February 2008. Data collection involved in-person interviews and assessments with CWS caseworkers, caregivers and children (Dolan et al. 2011).
Survey design and assessment procedures For this study, the sample was restricted to all youth aged 11–17 years with a report of maltreatment and who remained in-home with their biological parent(s) (n = 675). This study excluded youth in all other care arrangements due to systematic differences between youth who remain with parents and those placed in out-of-home care (Barth et al. 2006). All study analyses were drawn from baseline interviews with youth and parents collected between March 2008 and September 2009 (Dolan et al. 2011). All procedures involved in analysing the NSCAW II were approved by the institutional review board at the Research Triangle Institute, and all statistical analyses using NSCAW II data were approved by the University of Chicago institutional review board.
Measures The following section describes key study variables. Refer to Appendix A1 for a more detailed overview of covariates.
Disability In NSCAW II, parents identified whether their child had a disability using a list of diagnostic medical conditions that included both
Victimization and depression among YWD
chronic and temporary disabilities. Using the diagnostic medical conditions, three possible disability categories were created based on clinical and epidemiological guidelines: (1) physical; (2) neurodevelopmental; and (3) learning/attention/behavioural health (Stein et al. 2013). Youth with physical/health disabilities were identified having one or more of the following conditions: severe persistent asthma, diabetes, heart problems including congenital heart disease, haematological disorders, cystic fibrosis, arthritis, joint problems, human immunodeficiency virus/ acquired immune deficiency syndrome and other complex health impairments. Youth with neurodevelopmental disabilities were identified as having one or more of the following conditions: visual impairment/blindness, hearing impairment/deafness, autism/autism spectrum disorder (ASD), Down syndrome, developmental delay/intellectual disability, cerebral palsy, muscular dystrophy and epilepsy/seizure disorder. Aware of the clinical overlap between neurodevelopmental and physical disabilities, we used the primary diagnosis to classify disability type. Because of the small sample size and the unequal distribution of diagnostic conditions across the sample, separate regression analyses by disability type were not feasible. In order to generate adequate statistical power, youth classified as having either a physical/health and/or neurodevelopmental disability were grouped together as YWD. Youth with only learning/attention/emotional–behavioural health disabilities were excluded from the operationalization of disability due to the high prevalence of learning/attention/ emotional–behavioural disabilities among youth in the CWS, potential confounding effects of emotional/behavioural disabilities on mental health outcomes and the significant evidence base that identifies learning/emotional/behavioural disabilities as relatively common sequelae to childhood maltreatment (Conroy & Brown 2004; Stahmer et al. 2005). As documented in the literature, the experience of victimization may contribute in a causal way to adolescent disability and not only its reverse. Given this challenge, disability was operationalized in such a way as to minimize confounds. Many of the disabilities selected for this study (ASD, mental retardation, etc.) are by definition developmental disabilities that are present at birth. Moreover, such disabilities, including asthma, arthritis and Down syndrome, are generally not considered part of the sequelae of child maltreatment/victimization. However, to ensure accuracy, a regression analysis was conducted including an additional dummy variable containing those youth with learning/attention/ emotional–behavioural disabilities. The results of the regression did not yield significant differences with regard to outcomes. Using the above criteria resulted in a sample of YWD (n = 247). For the distribution of diagnostic conditions
991
and estimated functional limitations in this population, see Appendix A2.
Victimization Assessment of victimization during the previous 12 months was based on youth self-report. Our victimization index consisted of eight items across four types of victimization: (1) physical abuse in the home; (2) emotional abuse in the home; (3) physical victimization outside of the home; and (4) threat or assault with a weapon. These items were selected based on the extensive data that link these types of victimization to mental health conditions and other negative outcomes among youth (Martinez & Richters 1993; Edwards et al. 2003; Reading 2006; Margolin et al. 2010). All victimization items included were dichotomous (yes/no) and recoded as ‘present’ (1) or ‘absent’ (0). Items from the Parent-Child Conflict Tactic Scale 2 (CTS 2; Cronbach’s α = 0.55) (Straus et al. 1998) were used to measure past year exposure to abuse in the home. Physical abuse was assessed using items from the Severe Assault Scale: a parent/ adult in the home (1) choked the youth; (2) beat/repeatedly hit the youth; or (3) scalded or burned him or her. Emotional abuse was measured using items from the Psychological Aggression Scale: a parent/adult in the home (1) called the youth dumb, lazy or some other name; or (2) swore at him or her. Outside of the home, victimization was assessed by the Child Health and Illness Profile-Adolescent Edition (CHIP-AE; Cronbach’s α > 0.70) (Starfield et al. 1995). The CHIP-AE is a self-administered questionnaire to assess injury and disease among youth age 11–17 years. This instrument measured past year exposure to victimization by asking whether youth were physically injured by a friend/classmate/neighbourhood kid/ stranger in either school, neighbourhood, mall, friend’s home or other setting. Two items assessed prior year exposure to threat or assault with a weapon: (1) the CTS-2 asked whether an adult with whom the youth lived threatened him or her with a knife or gun; (2) the CHIP-AE asked youth about injuries due to ‘a gunshot wound or stab wound’. For our analysis, responses to each of the eight items were recoded (0,1) and summed for each youth, with total scores ranging from 0 to 8. Based on the distribution of victimizations, we recoded our victimization measure into four levels of exposure over the previous year: (0) no victimizations; (1) one victimization; (2) two victimizations; (3) three or more victimizations. The distribution for our past year victimization variable was as follows: 39.5% of youth reported no victimization; 24.3% reported one; 15% reported two and 21.3%
© 2015 John Wiley & Sons Ltd, Child: care, health and development, 41, 6, 989 –9 99
992 K.L. Berg et al.
reported three or more victimizations. Different levels or dosages of victimization have been used effectively across research studies (Appleyard et al. 2005). Statistical analyses revealed that our victimization range (0, 1, 2, 3+) was clinically and statistically meaningful in terms of depression outcomes. For the distribution of types of victimization, see Appendix A3.
Mental health Youth depression was assessed by the Children’s Depression Inventory (CDI; Cronbach’s α = 0.71–0.86) (Kovacs 1984). The CDI is the most commonly used self-report measure of depression in youth, with normative data available from school, psychiatric and paediatric populations (Smucker et al. 1986). Based on CDI manual guidelines, youth were classified as clinically depressed if they scored ≥90th percentile for their age and sex in the normative population, a threshold that has been used in other epidemiological studies of youth depression (Smucker et al. 1986; Hyphantis et al. 2012). By these criteria, 9% of CWS-involved youth reported depression (Cronbach’s α for sample = 0.87).
Covariates Demographic variables in this study included child, family and community factors. Key demographic variables included youth age, race, gender, parent age, parent education, marital status, household poverty, neighbourhood social disorganization and urban/rural residence. Additional variables of interest were youth reported closeness to parent, parent-reported intimate partner violence, parental substance abuse and parent physical and mental health status. These variables were selected because research suggests that victimized youth are often in families and communities characterized by the aforementioned factors, which contributes to victimization risk and/or exacerbate its effects on the health/well-being of youth (Cicchetti 2004; Raina et al. 2005; Sidebotham & Heron 2006; Murphy et al. 2007; Turner et al. 2012; Emerson & Brigham 2014; see Appendix A1 for operational definitions).
Analyses Statistical procedures in STATA (version 12, College Station, TX, USA) were employed to account for complex survey design and reweighting. Because the sample was limited to youth residing with biological parents, domain analysis was applied to readjust for sample weights, permitting generalization to the
© 2015 John Wiley & Sons Ltd, Child: care, health and development, 41, 6, 989 –9 99
national sample of adolescents in the CWS who remain with parent(s). Weighted ordinal regression was utilized to estimate the odds of reporting level of victimization by disability status. This technique estimates the probability of experiencing the highest level of cumulative victimization (≥3) and all the lesser victimization levels. In this model, odds ratios were calculated in terms of unit change (the likelihood of a one-unit increase in victimization level). Weighted logistic regression analyses were employed to examine the effects of disability status, victimization level and other covariates in predicting clinical depression among CWS-involved youth. Logistic regression with the interaction term was utilized to test whether disability status moderated the relationship between level of victimization and clinical depression. Huber-White estimator was used to calculate standard error.
Results Population characteristics The average age of CWS-involved youth was 13.5 years. Approximately 60% were female and nearly half (44.7%) were White (Table 1). Over three-fourth resided in urban settings and 18% lived above 200% of the federal poverty level. Approximately 38% (n = 247) of youth in contact with the CWS had disabilities, YWD in the CWS did not differ significantly from youth without disabilities in terms of gender, age and poverty. However, as a group, YWD were more likely to reside in nonurban settings (30.3 vs. 19.8%; P ≤ 0.01), occupy neighbourhoods with moderate levels of social disorganization (46.9 vs. 32.6%; P < 0.05) and be cared for by primary parents with both higher levels of clinical depression (37 vs. 26.6%; P < 0.05), and lower levels of physical health (P < 0.05).
Victimization Among youth in the CWS, prior year exposure to victimization was relatively common. Although approximately 39.5% reported no victimization events over the prior 12 months, 25, 15 and 21.3% reported one, two or three or more victimization events over the same period, respectively. After incorporating covariates, the odds of reporting a one-unit increase in level of victimization was 75% higher for CWS-involved YWD (P < 0.05) versus youth without disabilities (Table 2). Lower odds of reporting a one-unit increase in level of victimization were associated with optimal closeness to the primary parent [odds ratio (OR) = 0.26; P < 0.01] and older parent age (OR = 0.95; P < 0.05) while higher odds of a one-unit increase was
Victimization and depression among YWD
993
Table 1. Description of study population Disability groups
Sample size Estimated population Level of victimization, % 0 1 2 3 or more Depression, yes % Demographic factors Age† Gender, female % Race/ethnicity % White Black Hispanic Others Residency, urban % Closeness to parent, optimal % Parent and family factors Age† Social support† Health status† Gender, female % Education level % HS Poverty level % 200% Partner status, partnered % Alcohol abuse, yes % Depression, yes % IPV % No 1–2 times >3 times Neighbourhood social disorganization % Low Median High
Population
No disability
Any disability
675
428
247
F-test
P = 0.617 39.45 24.26 15.04 21.25 8.60
41.50 24.18 15.28 19.04 5.50
36.05 24.38 14.66 24.92 13.80
P = 0.020
13.48 ± 3.2 60.17
13.41 ± 4.0 61.96
13.58 ± 5.0 57.20
P = 0.478 P = 0.398
44.67 17.90 27.67 9.76 76.26 50.50
45.96 17.11 27.32 9.61 80.23 48.10
42.51 19.22 28.25 10.01 69.66 54.70
P = 0.922 P = 0.020
37.72 ± 9.0 3.67 ± 2.0 46.00 ± 21.5 89.68
37.48 ± 10.7 3.70 ± 2.7 47.23 ± 23.2 88.02
38.10 ± 11.2 3.61 ± 2.2 44.04 ± 33.4 92.33
P = 0.253 P = 0.467 P = 0.036 P = 0.258
30.62 34.57 34.80
30.70 36.14 33.16
30.49 31.98 37.53
P = 0.701
19.67 31.23 25.93 17.74 38.60 6.69 30.56
17.23 31.79 26.43 18.34 38.75 6.52 26.64
23.73 30.30 25.10 16.75 38.35 6.95 37.07
P = 0.645
70.55 9.92 19.53
71.45 9.00 19.55
69.05 11.45 19.50
P = 0.738
43.17 37.98 13.70
44.99 32.62 15.44
40.16 46.88 10.81
P = 0.021
P = 0.007 P = 0.254
P = 0.941 P = 0.869 P = 0.018
†For continuous variables, data were presented as mean ± standard deviation. HS, high school; IPV, intimate partner violence.
correlated with older youth age (OR = 1.3; P < 0.01) and high neighbourhood social disorganization (OR 1.8; P < 0.05).
Clinical depression In the CWS, 9% of youth reported clinical depression. Results of bivariate analysis indicated that YWD were more than twice as
likely to report clinical depression in comparison with their typically developing peers (13.8 vs. 5.5%; P < 0.05). The relationship between disability and clinical depression persisted despite incorporating child and family factors. The odds of reporting clinical depression were four times higher for YWD, controlling for all other variables (OR 3.975; P < 0.01) (Table 3). Several additional covariates also significantly predicted clinical
© 2015 John Wiley & Sons Ltd, Child: care, health and development, 41, 6, 989 –9 99
994 K.L. Berg et al.
Table 2. Weighted ordinal logistic regression with level of victimization as the outcome Model 1 OR Disability Yes (vs. no) Children’s characteristics Age Gender Male (vs. female) Race/ethnicity Black (vs. white) Hispanic (vs. white) Others (vs. white) Closeness to parent Optimal (vs. non-optimal) Family characteristics Parent age Parent gender Parent alcohol abuse Yes (vs. no) Intimate partner violence 1–2 times (vs. 0) >2 times (vs. 0) Neighbourhood social disorganization Mid level (vs. low) High level (vs. low) Parent health status F-test P-value
1.288
Table 3. Disability, victimization level, selected covariates and clinical depression***,*
Model 2† SE 0.259
F(1, 72) = 1.58 P = 0.212
OR
Dichotomous depression† SE
1.749*
0.422
1.279**
0.086
0.978
0.206
1.652 1.451 1.448
1.138 0.547 0.720
0.257**
0.053
0.954* 2.124*
0.019 0.781
1.485
0.648
0.536 1.459
0.244 0.395
0.960 0.225 1.785* 0.446 1.006 0.007 F(27, 46) = 4.72 P < 0.000
*
