xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 002001120
007 cr mnu|||uuuuu
008 090429s2008 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0002594
The cumulative effects of victimization, community violence, and household dysfunction on depression and suicide ideation in a cohort of adolescent females
h [electronic resource] /
by Katherine Best.
[Tampa, Fla] :
b University of South Florida,
Title from PDF of title page.
Document formatted into pages; contains 147 pages.
Dissertation (Ph.D.)--University of South Florida, 2008.
Includes bibliographical references.
Text (Electronic dissertation) in PDF format.
ABSTRACT: Recent scholarly efforts have sought to examine the cumulative impact of deleterious adverse childhood exposures on various mental health outcomes. Lifetime prevalence rates for depressive disorders are approximately 20% among adolescents. Depression is ranked as the leading cause for disability and fourth leading contributor to the global burden of disease in the world. The purpose of this study was to determine the cumulative impact of adolescent adverse experiences on outcomes of depression, suicide ideation, and overall mental distress in a cohort of 125 adolescent girls receiving public assistance. The adverse exposures studied were personal victimization, household dysfunction, and community violence exposures. Across the three categories of exposures, adolescents reported that community exposures were the highest 92.8%, followed by household dysfunction 89.6 %, and lastly, personal victimization 80%.Over 40% reported experiencing more than seven adverse exposures. There was a doubling in the incidence of depression by the fourth year, and an almost ten percent increase in mental distress by the fourth year. Evidence of a significant direct association was found for those experiencing victimization with depression and suicide ideation. The total Adolescent Adverse Exposures (AAE) score was positively correlated with the CES-D scores in the last three years of the study, however not with suicide ideation. The cumulative impact or 'dose-response' relationship of such exposures on depression, suicide ideation, or change over time was not found. In contradiction with general beliefs and existing literature, a significant negative association was found with depression and having a parent incarcerated or experiencing the divorce of parents.This finding suggests given the homogeneity of this population, experiencing both poverty and high levels of exposure to victimization, that having an incarcerated parent or parental divorce may be potentially protective mitigating the stressful experiences of continued victimization. The results of this study offer evidence of high prevalence rates of adversity occurring in the lives of these already at risk adolescents. A call for efforts to reduce community violence and personal victimization in the context of poverty are needed to prevent the growing rates of depression and suicide ideation for these fragile families and adolescence.
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
Co-advisor: Jeanine Coreil, Ph.D.
Co-advisor: Roger A. Boothroyd, Ph.D.
x Community and Family Health
t USF Electronic Theses and Dissertations.
The Cumulative Effects of Victimizati on, Community Violence, and Household Dysfunction on Depression and Suicide Ideation in a Cohort of Adolescent Females by Katherine Best A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Community and Family Heath College of Public Health University of South Florida Co-Major Professor: Jeannine Coreil, Ph.D. Co-Major Professor: Roger A. Boothroyd, Ph.D. Martha Coulter, Ph.D. Randy Borum, Psy.D. Paul Greenbaum, Ph.D. Date of Approval: July 15, 2008 Key Words: Stressors, Cumulative Exposures, A dverse Events, Welfare, Mental Distress Copyright 2008, Katherine Best
Dedicated to my children Jason, Anna, Paul, and the memory of my little brother, Joe
ACKNOWLEDGEMENTS I would like to acknowledge all of the professors that have mentored during the completion of my doctorate studies. First, Dr Ellis Gesten, for s uggesting that I pursue graduate studies in the first place, and for his continual encouragement during the process. I will forever be indebted to Dr. Roger Boothroyd, my primary mentor, for remaining supportive through all of the years of our work together, and for always pushing me to see the positive in every situ ation. I would like to thank Dr. Jeannine Coreil for continuing to challenge me to impr ove and not to settle for anything but the best scholarly work. Finally, I thank Dr. Martha Coulter and Dr. Randy Borum for their dedication to the field of violence prevention and their depth of insight and wisdom. To Dr. Paul Greenbaum, I thank you for your patience.
i TABLE OF CONTENTS LIST OF TABLES ............................................................................................................. iv LIST OF FIGURES ........................................................................................................... vi ABSTRACT ...................................................................................................................... vii CHAPTER ONE: INTRODUCT ION AND STATEMENT OF THE PROBLEM .............1 Need for the Study ...................................................................................................4 Implications for Public Health ...............................................................................10 Overview of Study Methods ..................................................................................11 Definitions..................................................................................................14 Adolescent Victimization...............................................................15 Household Dysfunction .................................................................15 Community Violence .....................................................................15 Research Hypotheses .................................................................................16 Delimitations ..............................................................................................16 Limitations .................................................................................................17 CHAPTER TWO: REVIEW OF THE LITERATURE .....................................................18 Risk Factors ...........................................................................................................18 Effects of Cumulative Risk Factors ...........................................................20 Summary ........................................................................................20 Stress Process Model .............................................................................................21 Introduction ................................................................................................21 Status Strain ..................................................................................22 Variables Relative to Socioeconomic Position ..................22 Chronic Stressors of Household Dysfunction ................................24 Contextual Strain: Exposure s to Community Violence .................26 Acute Stressors: Victimization ......................................................29 Conclusions ................................................................................................31 The Cumulative Effects of Adverse Exposures .....................................................32 Multiple Stressors Across Domains ...........................................................34 Limitations of Curre nt Literature ...........................................................................36 Mechanisms of Influence .......................................................................................38 Pathways to Risk Amplification: Enabling and Reinforcing Factors ........38 Conclusions ............................................................................................................40
ii CHAPTER THREE: METHOD ........................................................................................42 Study Purpose ........................................................................................................42 Research Hypotheses .............................................................................................42 Overview of Study Design .....................................................................................43 Description of Welfare Reform: Adolescent Girls in Transition Data Set ................................................................................................44 The Sampling Frame and Process ..................................................44 Attrition Rates ....................................................................44 Strengths and Limitations of Study Sample .......................45 Original Data Collection Methods .................................................46 Instrumentation ..............................................................................46 Data Integrity .........................................................................................................47 Study Variable Selection........................................................................................47 Missing Data Strategies .............................................................................48 Behavioral and Demogra phic Control Variables .......................................51 Predictor Variables .....................................................................................54 Adolescent Victimization...............................................................57 Household Dysfunction .................................................................58 Community Violence .....................................................................58 Outcome Variables .....................................................................................61 Depression......................................................................................61 Suicide Ideation .............................................................................64 Mental Distress ..............................................................................64 Data Analysis .........................................................................................................65 Level One Analyses ...................................................................................65 Behavioral and Demogra phic Control Variables ...........................65 Predictor Variables .........................................................................66 Composite Variables ..........................................................67 Outcome Variables .........................................................................67 Level Two Analyses ..................................................................................68 Level Three Analyses ................................................................................71 Design Strengths and Limitations ..........................................................................72 Conclusions ............................................................................................................74 CHAPTER FOUR: RESULTS ..........................................................................................75 Level One Analyses ...............................................................................................75 Control Variables .......................................................................................75 Predictor Variables .....................................................................................76 Outcome Variables .....................................................................................79 Level Two Analyses ..............................................................................................81 Relationship Among Predictors .................................................................81 Relationship Among Socio-Demographics and Predictor Variables .........82 Relationship Among Socio-Demographics and Outcome Variables .........83 Odds Ratios of Individual It ems and Outcome Variables .........................85
iii Relationship of Socio-Demographi cs and Cumulative Score with Outcome Variables ...............................................................................91 Level Three Analyses ............................................................................................94 Contribution of Each Predictor on Outcomes ............................................94 Contribution of Cumulative Predictors on all Outcomes ...........................98 Summary of Findings ...........................................................................................102 CHAPTER FIVE: DISCU SSION OF FINDINGS ..........................................................108 PrevalenceÂ’s .........................................................................................................109 Relationships Between Adversities and Outcomes ..............................................110 Summary of Hypotheses ......................................................................................114 Limitations of Study ............................................................................................115 Contributions of the Study and Imp lications for Public Health ...........................116 Recommendations of Future Research ................................................................119 Conclusions ..........................................................................................................122 REFERENCES ................................................................................................................124 APPENDICES .................................................................................................................144 Appendix A Measures Included in the 2003 Adolescent Interview Protocol ......145 Appendix B Measures Included in the 2003 MotherÂ’s Interview Protocol .........147 ABOUT THE AUTHOR ....................................................................................... End Page
iv LIST OF TABLES Table 1. Follow-up Retention Rates ..............................................................................45 Table 2. Number of Cases Missing De pression Scores and Suicide Ideation Item ..................................................................................................................49 Table 3. Behavioral and Dem ographic Control Variables .............................................52 Table 4. Characteristics of the Mothers and Daughters .................................................53 Table 5. Items from ACE Study on Abuse ....................................................................56 Table 6. Items for Measurement of Exposure to Victimization .....................................56 Table 7. Items for Measurement of Household Dysfunction .........................................59 Table 8. Items for Measurement of Exposure to Community Violence ........................60 Table 9. Items Compri sing the CES-D Scale .................................................................62 Table 10. Definition and Prevalence of each Item by Category for Adverse Exposures .........................................................................................................78 Table 11. Number of Adverse Adol escent Experiences (AAE Score) ............................79 Table 12. Prevalence of Outcome Variables across Four Years ......................................80 Table 13. Correlation Matrix of Indi vidual Item Predictor Variables and Outcomes .........................................................................................................86 Table 14. Socio-Demographic Ch aracteristics by Outcomes ..........................................87 Table 15. Socio-Demographic Ch aracteristics by Outcomes ..........................................88 Table 16. Individual Factors In fluence on Any Suicide Ideation ....................................89
v Table 17. Individual Factors Influence on Any Depression ............................................90 Table 18. Bivariate Associations of AAE Scores with Outcomes across Time ..............93 Table 19. Summary of Re gression Analysis with Adverse Exposures on Any Mental Distress ................................................................................................96 Table 20. Influence of Adverse Exposure Categories on Outcomes ...............................97 Table 21. Summary of Logi stic Regressions for Suicide Ideation Controlling for Socio-Demographics ........................................................................................99 Table 22. Summary of Logi stic Regression Models fo r Depression Controlling for Socio-Demographics ..................................................................................99 Table 23. The Relationship of the AAE Score to Depression, Suicide Ideation, and Mental Distress ........................................................................................101 Table 24. Linear Regression Models for Total AAE SCORE and Change Scores .......102 Table 25. Summary of Bivariate Associ ations between Socio-Demographics, Individual Predictors and Outcomes ..............................................................104 Table 26. Summary of Bivariate Associa tions Between Familial Predictors and Outcomes .......................................................................................................105 Table 27. Summary of Bivariate Associ ations Between Community Predictors and Outcomes.................................................................................................106 Table 28. Summary of Signi ficance for Hypothesis ......................................................107
vi LIST OF FIGURES Figure 1. Conceptual Model for Tes ting Cumulative Impact of Adverse Exposures on Adolescent Mental Health Outcomes ........................................41 Figure 2. Operational Model for Testi ng the Cumulative Effects of Adverse Exposures on Depression and Suicide Ideation ...............................................70
vii THE CUMULATIVE EFFECTS OF VICTIM IZATION, COMMUNITY VIOLENCE, AND HOUSEHOLD DYSFUNCTION ON DEPRESSION AND SUICIDE IDEATION IN A COHORT OF ADOLESCENT FEMALES Katherine Best ABSTRACT Recent scholarly efforts have sought to examine the cumulative impact of deleterious adverse childhood exposures on va rious mental health outcomes. Lifetime prevalence rates for depressive disorders are approximately 20% among adolescents. Depression is ranked as the lead ing cause for disability and f ourth leading contributor to the global burden of di sease in the world. The purpose of this study was to determ ine the cumulative impact of adolescent adverse experiences on outcomes of depres sion, suicide ideation, and overall mental distress in a cohort of 125 adolescent girl s receiving public assi stance. The adverse exposures studied were personal victimi zation, household dysfunction, and community violence exposures. Across the three categories of exposures, adolescents reported that community exposures were the highest 92.8%, followed by household dysfunction 89.6 %, and lastly, personal victimization 80%. Over 40% repor ted experiencing more than seven adverse exposures. There was a doubling in the inciden ce of depression by the fourth year, and an almost ten percent increase in me ntal distress by the fourth year. Evidence of a significant
viii direct association was found for those expe riencing victimization with depression and suicide ideation. The total Adolescent Adverse Exposures (AAE) score was positively correlated with the CES-D scores in the last three years of the study, however not with suicide ideation. The cumulativ e impact or Â‘dose-respon seÂ’ relationship of such exposures on depression, suicide ideati on, or change over time was not found. In contradiction with general beliefs a nd existing literature, a significant negative association was found with depression and ha ving a parent incarcerated or experiencing the divorce of parents. This finding suggest s given the homogeneity of this population, experiencing both poverty and high levels of exposure to victimization, that having an incarcerated parent or parental divorce may be potentially protective mitigating the stressful experiences of continued victimization. The results of this study offer evidence of high prevalence rates of adversity occurring in the lives of these already at risk adolescents. A call for efforts to reduce community violence and personal victimizati on in the context of poverty are needed to prevent the growing rates of depression and su icide ideation for these fragile families and adolescence.
1 CHAPTER ONE INTRODUCTION AND STATEMENT OF THE PROBLEM Â“Kids are Â‘malleableÂ’ rather than Â‘resilie ntÂ’, in the sense that each threat costs them something.Â” Bruce Perry, M.D. Growing up with economic disadvantage is significantly associated with poorer health outcomes (Boothroyd & Olufokunbi, 2001; Boushey & Gundersen, 2001; Irwin, Burg, & Cart, 2002; Lichter & Crowley, 2000) Negative impacts include depressive symptoms (Aneshensel & Sucoff, 1996; Wight, Sepulveda & Aneshensel, 2004), dropping out of school (Haveman & Wolfe, 1995), teenage pregnancy (Kirby, 1997), and substance involvement (Frojd, Marttunen, Pelkonen, Pahlen & Kaltiala-Heino, 2006; Kirby & Fraser, 1997). The presence and persis tence of depressive symptoms (Dekovic, Buist, & Reitz, 2004) disproportionately aff ects those with low socioeconomic positions, as evidenced in the growing literature on hi gh rates of depression (Belle, 1990; Muntaner, Eaton, Miech, & OÂ’Campo, 2004; Ritchey, Gory, Fitzpatrick, & Mullis, 1990; Wight et al., 2004). The gravity of this disparity is that depression is ranked as the first and foremost leading cause for disability and four th leading contributor to the global burden of disease in the world (World Health Or ganization [WHO], 2001). The economic impact exceeds $63 billion per year in the United St ates, making it a significant public health problem (U.S. Department of Health a nd Human Services [USDHHS], 1999).
2 This problem is even more pervasiv e and insidious when considering the contribution of depression as a risk f actor for morbidity and mortality. Evidence documents a relationship between mental disorder and all of the first seven leading health indicators (e.g. physical activit y, overweight and obesity, toba cco use, substance abuse, high risk sexual behaviors, poor mental health, injury and vi olence) (National Institutes of Health [NIH], 2000). AdolescentsÂ’ lifetime prevalence rate s for any depressive disorder is approximately 20%, influencing nearly 6 million young people (USDHHS, 1999; Friedman, Best, Armstrong, Duchnowski, Ev ans, Hernandez et al., 2004; Tsuang & Tohen, 2002). Further investiga tion of the causes and the f unction of time and age on female adolescent depression is justified in light of the potential consequences of untreated depression (Dekovic et al., 2004): a 12-fold risk fact or for suicide in females, co-occurring disorders (USDHHS, 1999), subs tantial impairment in functioning across domains, high risk sexual behaviors, circum scribed lifetime opport unities and lower rates of employment due to depres sive symptoms (Friedman et al., 2004; Kalil, Born, Knuz, & Cuadill, 2001). Since 1966, federal efforts to address suicide have been championed by the National Institute of Mental Health (NIMH). The approach was that of identifying risk factors associated with suicide (USDHHS, 2001). By 1983, the Centers for Disease Control and Prevention (CDC) sought to bring to public attention the problem of teen suicides. In 1996, the World Health Organi zation established gui delines for national strategies for the prevention of suicide (USDHHS, 2001).
3 More recently, the Surgeon GeneralÂ’s Repor t on mental health and the objectives of Healthy People 2010 call for specific strate gies to evaluate suicide as a national agenda. Ranked among the top leading cause s of death since 1975, suicides accounted for 307,973 deaths in the U.S. between 1989-1998 (USDHHS, 2003). The economic burden of suicide in 19 95 was estimated at $111.3 billion, which does not take into account disability from attempted suicides, and the lack of data on cases of death that are uncertain (USDHHS, 2001). A child dies from suicide every two hours (American Foundation for Suicide Prev ention, 2003). Suicide rates for children 1014 years old increased by 100 perc ent from 1980-1996 (USDHHS, 2001). For adolescents aged 15-24, suicide is the 3rd leading cause of death (the overall suicide rate for this age group is approxi mately 11 deaths per 100,000); while males complete suicide at a greater rate, females attempt suicide three times more often (Institute of Medicine [IOM], 2002). A nati onal school based study conducted by Kann and colleagues (1998) found a one-year preval ence rate for suicidal ideation of 20.5 percent (as cited in IOM, 2002, p. 40). The literature demonstrates that adults e xposed to severe sexual or physical abuse in childhood are more suicidal than those not exposed are (Bryant & Range, 2001; Dube, Anda, Whitfield, Brown, Felitti, Dong, Giles et al., 2005). Th is study also suggests that severity and frequency of the sexual and physical abuse also contribute s to suicidality as victims report fewer reasons for living and fewer social concerns for committing suicide (Bryant & Range, 2001). Additional risk factors for suicidality as cited in the literature include; mental illness, substance abuse, conduct disorder, economic insecurity, and hopelessness due to
4 interpersonal losses, as well as other cont extual social factor s (DHHS, 2001; Group for the Advancement of Psychiatry [GAP], 1996; IOM, 2002). Familial antecedents for depression include: maternal depression, victim ization, criminality, and absentee fathers (IOM, 2002). Need for the Study In order to successfully face major public health challenges the IOM (2003) has suggested that the field embrace a model of understanding based upon the theoretical perspective of social eco logy. Specifically, research with the goal of designing interventions has to be derived or mapped onto the perceived model of explanation or determinants. The ecological model embraces a broad spectrum of linkages and causal pathways (Bronfenbrenner, 1989) that take in to consideration the complexities of the human condition and the dynamic, rather than static, nature of life and health (IOM, 2003). The need to study outcomes utilizi ng the social ecological model has been emphasized by researchers in the field for decades. For example, epidemiologist John Cassel (1964) who noted that Â‘rapid rates of change in any one of four linked open systemsÂ… the physiological, psychological, soci al, or cultural produc ed potential strains and possible breakdownÂ’. One of the overarching goals of Healthy Pe ople 2010 is the elimination of health disparities. Socioeconomic status (SES) is widely recognized as a fundamental causative factor in creating health disparities (G oodman, Adler, Kawachi, Frazier, Huang & Colditz, 2001). The economic insecurities and relative deprivation created by disadvantaged status contribute to a wide range of social conditions that have been identified as key causes of illness (Link & Phelan, 1995).
5 Current research on health disparities in the lives of infants, children and adults continues to focus on risk and protective factor s that identify a broad range of inequalities within there social and environmental contex ts (Aneshensel & Sucoff, 1996; Frojd, et., al, 2006; Goodman et al., 2001). However, in adol escent populations the graded relationship of SES and health is less clear (Goodman et al., 2001). Soci al disparities are typically operationalized through a variety of indicators such as income, education, occupation, social class or position, a nd perceived deprivation (Aneshensel & Sucoff, 1996; Cohen, 2002; Elstad, 1998; Frojd, et al., 2006). Despite the recognition that social structures impact mental health (Aneshensel & Sucoff, 1996; Frojd, et al., 2006; GAP, 1996) further understanding the relationship of social determinants and th e underlying mechanisms that contribute to mental disorder is needed (Cohen, 2002; Elstad, 1998; Irwin, et al., 2002). This is especially critical when considering the sensitive developmental period of adolescence and the impact of depressi on and suicide on this age group (IOM, 2002). It has been proposed that the link be tween psychological effects and social inequalities is increased expos ures to stress resulting in both health (Elstad, 1998) and mental health disparities (Aneshensel, Rutter, & Lachenbruch, 1991; Turner & Loyd, 1995). Systems of stratification based upon struct ural arrangements, economic class, race, or gender contribute to stressf ul life conditions through increa sed disparities in resources, opportunities, personal regard, and self -esteem (Link & Phelan, 1995; Pearlin, Lieberman, Meneghan, & Mullan, 1981; Pearlin, 1989). The literature suggests that structural arrangements seem to be mediated through perceptions of social status or position (Goodman et al., 2001). A persistent marginalized status in society, enduring conditions of loss, and inade quate resources or opportunities
6 result in numerous forms of psychologica l distress, withdrawal, substance abuse, depression, anxiety, hopelessne ss and decreased productivity (Aneshensel, et al., 1991; Cohen, 2002; Goodman, et al., 2001; Taylor & Turner, 2002; Turner & Loyd, 1995). In an effort to further understand the i ndirect psychosocial pathways of disease and disorder, recent scholarly efforts have sought to bridge community and individual level data into an ecodynamic analysis th at examines the cumulative impact of deleterious stress exposures on overall well being (Cohen, 2002; Felitti et al., 1998; Gilman, Kawachi, Fitzmaurice, & Buka, 2003; Taylor & Turner, 2002). Pearlin argues Â‘that stressful experiences arise not in a v acuum but may be traced to surrounding social structures and a personÂ’s location within such structuresÂ’ (Pearli n, 1989). The implication that overall health outcomes are the embodied expressions of the cumulative impact of stress exposures resulting from socioeconomic position and the perception of status in early life warrants further inves tigation (Aneshensel et al., 19 91; Elstad, 1998; Gilman et al., 2003; Goodman et al., 2001; Krieger, 2001; Taylor & Turner, 2002). Nevertheless, much of the current research has sought causal linkages through Â‘i nnate or individual characteristics versus imposed or societal constraintsÂ’ (A neshensel et al., 1991; Elstad, 1998; Gilman et al., 2003; Goodman et al., 2001 ; Krieger, 2001; Taylor & Turner, 2002). The role of cumulative adverse exposu res in the context of structural arrangements, and the perceptions of those stre ssors, is meaningful to explore in light of an increasing literature on social determ inants of overall health and well being (Aneshensel et al., 1991; Aneshensel & Sucoff, 1996; Elstad, 1998; Gilman et al., 2003; Goodman, 2001; Turn er & Lloyd, 1995).
7 Dong, Anda, Felitti, Dube, Williamson, Th ompson, and colleagues (2004) found that adults reporting any single adverse experience were likely to have been exposed to a multitude of adverse events in childhood suggesting the importance of understanding cooccurring traumas and the cumulative impact of stressful experiences. Wrongly assuming that one single type of expos ure or trauma is implicated in the development of overall psychological distress may lead to misdiagnos es, under assessment of distressing events, mistreatment of exposures, or mis-timed interventions (Chapma n, Whitfield, Felitti, Dube, Edwards, & Anda, 2004; Felitti, et al., 1998; Turner, Finkelhor & Ormrod, 2006; Whitfield, 1998). Further gaps and methodological deficienci es noted within the existing literature base for depression and suicidality are the lack of longitudinal studies and predictive models (IOM, 2002). Moreover, even in cases where the research suggests that there are a number of direct causal linkages, the confounding ecological factors offer further justification for context-oriented approaches as well as the investigation of processes that may be more effective in changing outcomes (Link & Phelan, 1995). A meta-analysis of socioeconomic inequa lity and depression by Lorant, Deliege, Eaton, Robert, Philippot, and Ansseau (2003) found compelling evidence for a causal relationship, but call for gr eater understanding of the c ourse of development of depression. Furthermore, a review of the lite rature disclosed limited studies examining the association of social position with de pression among women (Chapman et al., 2004; Muntaner, Eaton, Miech, & OÂ’Campo, 2004), desp ite evidence that females experience depression at twice the rate of males.
8 In addition, there is limited empirical rese arch seeking to understand the nature of these gender differences (Hazler & Mellin, 2004). A current ar gument in the literature is that female adolescents are generally more e xposed to multiple stressful events and that they are more distressed and r eactive to environmental stresso rs than males (Ge, Lorenz, Conger, Elder & Simons, 1994). The Institute of Medicine calls for Â“specific data from well-defined and characterized populations whose community leve l social descriptives are well-known,Â” in order to clarify the complex pathways fr om childhood experiences to mental illness and/or suicidality. Clarificat ion of the etiological pro cesses by assessing change longitudinally through identification of time of onset and the preceding temporal events, rather than simply specifying ri sk factors, also guides future development of interventions that are optimally timed and able to address multiple causal pathways (IOM, 2002; Kraemer, Stice, Kazdin, Offord, & Kupfer, 2001). An additional suggestion by the IOM (2002) is that, given that suicide is a relatively infrequent event, alternate endpoints, such as suicidal ideation, are required to add to the existing knowledge base. Further understanding of vari ous factors situated within the developmental transition of adoles cence and how these factors work in concert to either prevent or evoke suicidal behavior is needed (IOM, 2002). A number of scholars suggest a need for more research identifying not only specific acute st ressors but also the context of such events in continuing probl ems that produce chronic strain (Chapman et al., 2004; Felitti et al., 1998; Pearlin, 1989; Thoits, 1995; Turner et al., 2006).
9 Thus, the conceptual underpinnings for this study draw upon the classic stress process model developed by Pearlin and co lleagues. This perspective approaches outcomes through a socio-ecologic theoretical pe rspective-utilizing individual, familial, and environmental levels of mediating and moderating resources and stressors (Bronfenbrenner, 1989; P earlin et al., 1981). The propositions of social stress theory suggest that: multiple adverse exposures of perceived stressors arising from a constellation of contextu al stressors, chronic strain, and acute stressors contribute to poor me ntal health outcomes (Kirby & Fraser,1997; Krieger & Smith, 2004; Turner & Lloyd, 1995; Turner & Lloyd, 1999; Pearlin et al., 1981; Pearlin, 1989; Rudolph, Hammen, Burg e, Lindberg, Herzberg, & Daley, 2000; Rutter, 2005; Thompson, Mazza, Herti ng, Randell, & Eggert, 2005). Brooks-GunnÂ’s (1991) longitudinal study presented evidence of increased depre ssion in adolescents exposed to the stressors of numerous life even ts; it was not the novelty or type of events but the number of events that were significant. Building upon the broad framework of th e stress process theory (Pearlin, 1981) this study sought to replicate as closely as possible the analytic strategies utilized by Felitti and colleaguesÂ’ (1998) study of A dverse Childhood Experiences (ACE) conducted by the Kaiser Permanente Medical Group a nd the Centers for Disease Control and Prevention. To date the ACE study is a decade l ong ongoing collaboration assessing the impact of numerous interrelated ACEs on a wide variety of h ealth and behavioral outcomes in adults through a cross-sectio nal lens (Chapman et al., 2004; Dong, Anda, Felitti, Williamson, Dube, Brown et al ., 2005; Dube, Anda, Felitti, Chapman,
10 Williamson, Giles, 2001; Felitti et al., 1998). Specifically, these researchers have been assessing the relationship of childhood abuse and household dysfunction on adult morbidity and mortality. This study sought to add to the literatur e by validating the ACE study approach of investigating the cumulativ e effect of adverse chil dhood/adolescent exposures on outcomes of depression and suicide ideation in female adolescents utilizing data from a cohort of adolescent females bei ng raised on public assistance. Implications for Public Health In 1845, Friedrich Engels observed, Â“the sufferings of childhood are indelibly stamped on the adultsÂ” (Engels, 1845/1958; Krieger & Smith, 2004). This insight underpins current epidemiological studies that address the inequiti es accompanying a life of poverty and later health consequences (Link & Phelan, 1995; Lorant et al., 2003; Muntaner et al., 2004). The cumulative impact of chronic strain from both socioeconomic status and exposures to viol ence and victimization contri bute to poor mental health outcomes found in populations in poverty (C hapman et al., 2004; Felitti et al., 1998; Pearlin, 1989; Thoits, 1995; Turner et al., 2006). A growing number of researchers have examined the increased prevalence rates of disorder in children and adolescents over the past decade and the association with reported increased frequencies of childhood adve rsities in the form of victimization and household dysfunction (Chapman et al., 2004; Fo ege, 1998; Spat Widom, 1999; Turner et al., 2006). There is a call to more clearly de termine etiology of ment al disorder and to further understand the direct a nd indirect influences of a dversity and trauma on such outcomes (Thompson et al., 2005).
11 The public health significance of this study includes efforts to contribute to the knowledge base by: (1) increasing the understa nding of the cumulative impact or Â“dose responseÂ” relationship of adverse childhood/a dolescent exposures on depression, suicide ideation, and overall mental distress in a dolescents; (2) understa nding the impact of cumulative stressors across multiple domains within the context of poverty, utilizing a longitudinal lens; (3) providing valuable information regarding malleable stressors and resources for the purpose of developing spec ific interventions for female adolescents being raised in the context of poverty. For example, if specific adverse exposures in adolescence demonstrate a signifi cant contribution to a trajec tory of depression over time, then establishing meticulous scr eening efforts at the beginning of a school year or during primary care visits would help to inform interventions. It is hoped that this resear ch will contribute both con ceptually and empirically to our understanding of the pathways and the critical processe s and interactions between individual, family, and contextual stressors affecting the developmental period of female adolescence. Moreover, it is hoped that the resu lts of this study will help to provide even a brief a glimpse of the exposure to adversit ies that this populati on has experienced in order to guide future efforts towards interven tions to alleviate the burden of depression, suicide ideation, and overall mental distress for adolescent females. Overview of Study Methods This study examined secondary data from the Welfare Reform: Adolescent Girls in Transition Study (WRAGT) (Boothroyd, Ar mstrong, Gmez, Haynes, & Ort, 2003; Boothroyd, Armstrong, Gmez, Haynes, Ort, & Best, 2005). This study was a multi-
12 phase mixed methods longitudinal cohort study, which examined the impact of welfare reform on the future hopes and aspirations of female adolescents (Boothroyd et al., 2003). Ultimately, the projectÂ’s aim was to determine what factors differentiate adolescent girls whose lives follow a positive trajectory from those girls experiencing difficulties that are more serious. Participants in this study were part of a larger project funded by the Florida Agency for Health Care Administration (Contract #M0107) assessing the impact of welfare reform on the well-being and future aspi rations of adolescent gi rls. To be eligible for study participation, mothers had to be receiving Temporary Assistance for Needy Families Program (TANF) and have a daughter between the ages of 13 and 17 living at home at the start of the study. During th e course of the study, there was no further requirement for daughters to remain at home or for a mother to be receiving TANF. In addition, at the start of the study the families ha d to reside in a five county area in west central Florida. A sample of 125 mothers recei ving TANF (at the time of enrollment into the study) and their adoles cent daughters were identified from the 2000-2001 Florida Medicaid eligibility, data using the family identifier and other matching variables (such as gender, address, and last name). Recr uitment letters were mailed to 873 potential participants whereby 125 eligib le daughter/mother pairs were ultimately recruited for participation. Enrollment was stratified by race /ethnicity (i.e., white, black, Hispanic) and geographic location (i.e., urban versus rural). Each year of the study involved f ace-to-face interviews using various standardized measures with the 125 mothers and their daughters. These procedures were initially conducted in 2002, and then repeated in the three subsequent years, providing
13 four waves of data. Data for this study were drawn from all four waves collected from both the mothers and daughters. Justifications for utilizing a data set wher e participants are situated in poverty are based upon a number of suggestions. Kr aemer and colleagues (2001) call for disaggregating socioeconomic status to exam ine causal chains and the impact of the sequencing of events on mental health outco mes. As noted before the IOM (2002) calls for Â“well-defined and characterized popul ations whose community level social descriptives are well-known.Â” Within this study the term Â“welfareÂ” is not used as a vernacular term in reference to recipients of public assistance, but in reference to those involved with the federal welfare reform la ws receiving Temporary Assistance for Needy Families (TANF). This is discusse d more in depth in Chapter 2. The perceptions of poverty and the ac companying explanatory frameworks invariably assign blame or impute Â“ownersh ipÂ” of the problem, with the Â“burden of changeÂ” often discussed at the micro or indi vidual level. The contex t of poverty coupled with the stigmatizing labels associated with welfare in reference to social position is an emerging area of research as investigators begin to take into account the role of depression on coping mechanisms or planni ng alternative future s (Kalil et al., 2001; Wadsworth, Raviv, Compas, & Connor-Smith, 2005). The overarching assumptions for the study are based upon the theoretical lens of stress process theory, which pred icts that Â“large qualitative classes of events, that meet the criteria of being undesirable or unsche duled may contribute to stress and future depressionÂ” (Elstad, 1998; P earlin et al., 1981). Pearlin a nd colleagues (1981) note that this is a Â“reasonable approach when the goa l is to evaluate the total contribution of
14 classes of eventful experiences on stressÂ” a nd the resulting poor outcome. A small range of classes of stressful events have been selected for simplicity and manageability. Acute stressors refer to adolescent victimization or abuse; chronic stressors refer to household dysfunction; and c ontextual strain refers to community exposures of violence and perceived fear of the neighborhood context. While using the theoretical underpinnings of PearlinÂ’s concepts this study closely replicated the analytic methods of the ACE studies (Felitti et al., 1998) utilizing similar or identical items for the operationalization of adverse exposures and thus guided the selection of the independent variable s. It expands upon the ACE study model by capturing a wider range of a dverse exposures experienced during childhood/adolescence. In particular, community exposures to viol ence and perceived fear of the neighborhood context have been added as a repr esentation of contextual strain. The outcome variables of interest were depression, suicid e ideation, overall mental distress, and a change over time in all three variables. A score for each was assessed at 4 points in time a nd a change score computed by s ubtracting the score in year 1 from the score in year 4. The presence of symptoms of depression were measured by the CES-D and the presence of suicide idea tion were measured by thoughts of suicide reported by a single scaled item. Definitions The following is a description of the constructs to be used for the composite predictor variables. Each of the constructs have been opera tionalized and were developed by using categories of questions included under each class of events or exposures experienced by the adolescents. A description of each specific item follows in Chapter 3.
15 Adolescent Victimization The victimization of an adolescent was operationalized as a composite variable utilizing the following categories: (1) repor ts of being bullied, beaten up, sexually assaulted, robbed, stabbed, shot at, shot; (2) positive respons es to questions on physical, psychological, and sexual abuse; and (3) re ports of being sexually assaulted in the previous year. Household Dysfunction This construct was operationalized as a co mposite variable util izing the following categories of events: (1) family substance abuse; (2) maternal mental illness; (3) maternal victimization; (4) parental criminality; (5) parental separation/divorce; and (6) residential instability. In an effort to capture the contextual f actors in the lives of these adolescents, a measure of community violence has been adde d to the overall model to further expand the work of the ACE study and to assess the influence of such experiences on direct experiences of victimization or household dysf unction. Additional justif ication for this is found in recent concern that the promin ence given to child sexual abuse may be misplaced as Â‘carrying the weight of causalityÂ’ (Mullen, Marti n, Anderson, Romans, & Herbison, 1996) and that such factors may func tion as a marker for coexisting adversities impacting overall outcomes (Turner et al., 2006). Community Violence Community exposures to viol ence were operationalized as a composite variable based on reports of: (1) seeing someone sexuall y assaulted, robbed, sta bbed, shot at, shot, killed; (2) feeling afraid with adults, in neighborhood, or school; (3) knowledge of
16 weapons: knowing people who own a gun, bring a knife to school, bring a gun to school, or other weapons to school; (4) havi ng heard gunshots in the neighborhood. The purpose of this study was to dete rmine the cumulative impact or Â“dose responseÂ” of adverse experiences on depres sion, suicide ideation, and mental distress over time in a cohort of 125 adolescent girls raised on Welfare. The domains of adverse experiences investigated were adolescen t victimization, household dysfunction, and exposure to community violence. Research Hypotheses 1. Exposure to adolescent victimization w ill have a positive association with the presence of depression, suicide idea tion, and overall mental distress. 2. Household dysfunction will have a positive association with the presence of depression, suicide ideation, a nd overall mental distress. 3. Exposure to community violence will have a positive association with the presence of depression, suicide ideation, overall mental distress. 4. The number of adverse exposures w ill have a cumulative impact or Â“dose responseÂ” relationship with th e level of depression, su icide ideation, and mental distress. 5. The Adverse Adolescent Exposure Score wi ll be significantly related to change in depression, suicid e ideation, and mental distress over time. Delimitations Delimitations offer a description of the population to which results may be generalized (Locke, Wyric, Spirduso, & Silverman, 2000). The results from this dissertation may be generalized to the following populations:
17 1. Results are generalizble to female adolescents living with their Medicaid eligible mothers in 2002 w ithin the five county area of southwest Florida. Beyond that region of Florida, generalization to the remainder of the state is unclear. 2. Results are only generalizble to the population of female adolescents living with their Medicaid eligible mothers in 2002 within the five county area of west Florida who were not suffering fr om severe mental disorder. Limitations The limitations or restrictive weaknesses in the study design are as follows: 1. The small sample size (n=125) is a li mitation, restricting the number of variables that can be utilized in multivariate analyses. 2. The CES-D scale (Radloff, 1977) is lim ited to measurement of depressive symptomatology, not for meeting cr iteria for clinical depression. 3. The development of the composite variables for the AAE score were based on similar composites developed by Felitti and colleagues (1998) utilizing select questions from various scales not orig inally intended for these purposes. 4. Data were based on retrospective reca ll and may result in underreporting, given the evidence from other longitudi nal studies have demonstrated that retrospective self-reports of adverse exposures are lik ely to underestimate actual occurrence (Della, Yeager, & Lewis, 1990). 5. There may be mediators or moderators of the relationship between adverse exposures and depression othe r than the factors examined. 6. Age variations in the adol escent girls (13-17) in the first year of the study may have contributed to differences in percep tions and reports of adverse experiences.
18 CHAPTER TWO REVIEW OF THE LITERATURE The purpose of this literature review is to introduce the reader to the concept of adverse adolescent exposures and to explore the relationship between adverse exposures, depression, suicide ideation, and overall distre ss. Specifically, this review will present empirical evidence supporting th e conceptual framework guiding this study. In particular, a review of literature utilizing a cumulative or Â“dose responseÂ” approach of to the study of factors related to victimization, househol d dysfunction, community violence, and their association with depression, su icide ideation, or mental dist ress in adolescents will be presented. The global areas of risk discusse d are as follows: socioeconomic position, victimization, household dysfunction, and expos ures to community violence. The review will also emphasize the need fo r further research investigating the impact of such exposures and the complexities of adolescent experi ences of stressors and strains during transition to young adulthood. Risk Factors Broadly defined, risk factors refer to ant ecedent events that range from biological to environmental conditions that predispose, en able, or reinforce an onset, digression, or maintenance of a problematic outcome (K azdin, Kraemer, Kessler, Kupfer, & Offord, 1997; Kirby & Fraser, 1997). Sp ecifically, risk is defined as the probability that a
19 particular outcome is associated with exposur es to a certain characteristic, event, or environment over the base rate of the ge neral population (Kaz din et al., 1997). Kazdin and colleagues (1997) propose that exposure to one risk factor does not necessarily lead to a poor outco me, but with the presence of four or more risk factors, there is a 10-fold increase in ps ychological distress or disorder. In this regard, a review of the literature on specificity, (i.e ., that a particular risk factor is uniquely associated with a particular outcome) has found evidence to th e contrary, there are many different avenues to psychopathological outcomes, and simila r conditions lead to multiple outcomes (McMahon, Grant, Compas, Thurm, & Ey, 2003). The most common models for looking at risk factors are Additive (cumulative) Models, Interactive Models, or the Challe nge Model (Kirby & Fras er, 1997; Luthar & Zigler, 1991). Additive Models posit that ri sk and protective factors are polar opposites and lie along a continuum; incr eases in risk factors causes decreases in competence or coping (Kirby & Fraser, 1997). Inte ractive Models posit that protective factors exert an effect in the presence of risk factors. This is conceptualized as protective factors offering a buffering effect, an interruption of the risk chain, or that protec tive factors may prevent the initial occurrence of a risk factor (Kirby & Fras er, 1997). The third model, the Challenge Model posits that a curvilinear relati onship exits so that stressors actually lead to an enhanced competence (Luthar & Zigl er, 1991). Utilizing the over-arching socioecological perspective that allows for multi-sy stemic interactions enables a theoretical approach to assess and understand the rela tionship of risk factors on poor outcomes across life domains (Kirby & Fraser, 1997). The ju stification for this approach is that neither the additive, nor the challenge, nor th e interactive models of risk and protective
20 factors have been completely supported; how ever, research suggests that a balance of models may be helpful in assessing the presence of effects across systems (Kirby & Fraser, 1997). Effects of Cumulative Risk Factors The co-occurrence of multiple risk fact ors in the form of childhood abuse and household dysfunction has been labeled more recently as Adverse Childhood Experiences (ACE) (Felitti et al., 1998). Negative cumulative influences of multiple categories of ACEs on both physical and mental well bei ng have been found to be deleterious in numerous studies (Dong et al., 2004). Dong and colleagues found that adults reporting single adverse experiences were likely to ha ve been exposed to a multitude of adverse exposures in childhood, suggesting the im portance of understa nding co-occurring traumas and the cumulative impact of stressful experiences. Summary In summary, a host of factors have b een implicated in research studies investigating predisposing risk factors for the onset of depres sion, with varying levels of empirical support. Research that examines cu mulative adverse exposures contributing to poor mental health outcomes utilizes broad categories of adverse experiences across dimensions that include poverty, perceived deprivation, family disorganization or household dysfunction, victimization, and e xposures to community violence (Cohen, 2002; Dong et al., 2005; Felitti, 1998; Gilman et al., 2003; Taylor & Turner, 2002).
21 Stress Process Model Introduction The classic stress process model develope d by Pearlin and colleagues approaches outcomes through a socio-ecologic theoretical perspective utilizing individual, familial, and environmental levels of mediating and mo derating stressors and resources (Pearlin et al., 1981). Pearlin (1989) argues, Â“that stressful experiences arise not in a vacuum but may be traced to surrounding social struct ures and a personÂ’s location within such structuresÂ”. Systems of stra tification based upon social a nd economic class, race, or gender, contribute to stressful life conditions through increase d disparities in areas such as resources, opportunities, personal rega rd, and self-esteem (Brooks-Gunn, 1991; Gilman et al., 2003; Goodman et al., 2001; Link & Phelan, 1995; Pearlin et al., 1981; Pearlin, 1989; Rutter, 2005). Such structural arrangements exert a force of threatening or stigmatizing experiences creating stressors and strains re lative to perceptions of status within structural arrangements. Pearlin (1989) de fines stress as: Â“an exigency that people confront that is perceived as threatening or burdensomeÂ” (p.2 41). The events give rise to stress creating a stressor that has historically been divided into a life event, a recurring problem or chronic strain. The propositions of social stress theory suggest that perceived stressors arise from a constell ation of contextual stressor s, chronic strain, and acute stressors (Pearlin et al., 1981; Pearlin, 1989) The following section will discuss current literature in each of the domain s of the stress process model.
22 Status Strain The relationship between socioeconomic stat us (SES) and mental disorder is well established (Rutter, 2005), how ever, there is a lack of understanding of the underlying mechanisms between SES and overall outcomes (Goodman et al., 2001). Frequently, when utilizing SES as a risk factor, resear chers will utilize a proxy variable such as maternal education, welfare stat us, and receipt of reduced lunc h, income level, or parental occupation as indicators of SE S. A consistent finding in the literature is based on the status-attainment model; parentsÂ’ levels of education or occupation are associated with childrenÂ’s educational as well as occupa tional attainment (Jodl, Michel, Malanchuk, Eccles, & Sameroff, 2001). However, the auth ors note a primary limitation of the model, that is, Â“SES begets SESÂ”, a nd fails to offer any further e xplanation. Recently, there has been a shift in the empirical focus on SE S by further elaborating the process of socialization by parents. Variables Relative to Socioeconomic Position Numerous scholars concerned with the social causation of disorder emphasize the need to differentiate between the objective nature of social position, and the subject ive perception of placement in the social hierarchy (Goodman et al., 2001). Empirical ev idence demonstrates that depression is associated with social class or relative positi on, but is not consistent ly linked to SES (i.e., education or occupational pres tige) (Muntaner et al., 2004). An additional consideration is that adolescence is a critical developmental period often influenced by social desirability. An adolescentÂ’s emerging self-concept during this sensitive transitional time includes the crys tallization of perceived social status (Goodman et al., 2001). An illust ration of status strain at the individual level for an
23 adolescent may be seen in a youthsÂ’ desire to increase social stat us relative to their parents, Â“creating forms of adulthood that ar e different from those of their parentsÂ” (Thomson & Holland, 2002). Zaslow and colleaguesÂ’ (2001) review of various welfare reform initiatives implemented in numerous states, includi ng Florida, found a variety of negative adolescent outcomes. A growing body of ev idence from both cross-sectional and longitudinal studies have demonstrated evid ence that low SES (Gilman et al., 2003), as well as discrimination, is associated with el evated lifetime risk of depression (Dooley & Prause, 2002; Schlulz, Gravlee, Williams, Israel, Mentz, & Rowe, 2006; Taylor & Turner, 2002). The impact of the Personal Responsibil ity and Work Opportunity Reconciliation Act (PRWORA, P.L. 104-193) passed by C ongress in August 1996, not only dramatically altered welfare programs but also the lives of the children of recipients. The primary goal of the TANF Program was to assist people in becoming economically self-sufficient. While being raised on welfare increases the likelihood of future enrollment in welfare 80% of the daughters of welfare mother s, do not become dependent on welfare themselves. However, they are more likely to receive welfare (approximately 20%) compared to daughters of non-welfare mothers (about 3%) (Furstenberg, 1992). Gottschalk, McLanahan, and Sandefur (1994) at tribute this differential to the powerful effects of poverty and single parenthood. Taylor (2000) suggests that the powerful effects of poverty are not so easily teas ed apart, yet they seem to be more devastating for girls than boys. Research suggests that gender di fferences may be related to changes in reactivity to environmental stressors as well as the frequency of stressful occurrences
24 endured by girls (Ge et al., 1994) Such findings have resulte d in investigators stressing the need for early interventions to assist gi rls in Â“breaking the cy cleÂ” of poverty (Brooks & Buckner, 1996). In summary, while there is evidence that de pression is associated with both social class and relative position, th ere is little knowledge about distress in welfare samples. There is also little evidence on the role of chronic or acute stressors and the relationship between SES and depression or suicide ideati on, or what may mitigate the distress (Kalil et al., 2001). Chronic Stressors of Household Dysfunction One of the propositions of social stre ss theory suggests that chronic stress contributes to a wide range of poor outco mes (Pearlin et al., 1981; Pearlin, 1989). The explanations for the net e ffects of multiple complex childhood exposures ranges from shared and overlapping risk factors, co-m orbid patterns, and one factor creating an elevated risk for another (Dong et al., 2004; Kessler, Davis, & Kendler, 1997). Building upon this assumption, the broad concept of hous ehold dysfunction is conceptualized in the literature as family disorganization, in cluding a number of frequently co-occurring individual risk factors. Conditions su ch as parental psychopathology, depression, suicidality, substance abuse, criminal behaviors, child hood abuse and neglect, maternal victimization, maternal depression, parental separation and divor ce, and residential instability have been shown to place children at risk for poor outcomes (Dong et al., 2004; Kirby & Fraser, 1997). Studyi ng simply one or two for these exposures limits our understanding of the cumulative effect of such co-occurring stressful experiences.
25 Nevertheless, previous studies of these ri sk factors have independently predicted an increased lifetime risk fo r depression. Research has addr essed parental separation and divorce in childhood (Gilman et al., 2003; McMahon et al., 2003) parental substance abuse, aggressive parenting styles, and maternal victimi zation (Barnett, Miller-Perrin & Perrin, 1997; Stein, Leslie & Nyamathi, 2002) with greater frequency and duration of exposure to abuse (Favorini, 1995; Nurc o, Blatchley, Hanlon, OÂ’Grady, 1999); In addition, research has demonstrated that th ere is a greater risk for poor outcomes in adolescents exposed to parental psychopat hology (Barnett et al., 1997; Hammen, Henry, & Daley, 2000). The mechanisms hypothesized to explain the link between parental psychopathology and child outcomes range from genetic transmission, parental modeling, poor parenting, parental in carceration, and overall family dysfunction (Dougherty, Klein, & Davila, 2004; Hammen et al ., 2000; Kirby & Fraser, 1997). Until recently, researchers have neglected to fully separate areas of household dysfunction from violence exposures, and have focused simply on demonstrating the cooccurrence of these two variables. However, Felitti and colleagues (1998) began a series of studies with the Kaiser Permanente H ealth Plan that distinguished between the constructs of adverse childhood exposures and various health and mental health outcomes. As noted in Chapter 1, the Adverse Childhood Experiences (ACE) study provides important direction for the research proposed here. Felitti and associates operationally defi ne the construct of household dysfunction using the following variables: 1) parental separation or divorce, 2) a household member with substance abuse, 3) mental illness, 4) incarceration, and 5) maternal victimization. Residential mobility was added to expand the model for household dysfunction in 2005
26 (Dong et al., 2005). The justifi cation for including relocation, generally assumed to be a major life event, comes from empirical evid ence supporting the relationship between high residential instability and the onset of de pression in cohort studies (Dong et al., 2005; Gilman et al., 2003). Approximately 8% of the general population moves in a six-month period. Residential mobility in families leaving TANF is 42% within a six-month period (Sard, 2002). Best and colleagues (2006) f ound a significant relationship between the number of moves an adolescent experienced and level of depression in a TANF sample. The interpretation of the increased depression is that multiple moves may interrupt work schedules, jeopardizing employment for olde r adolescents, advers ely affect a youth's educational progress, and lead to the loss of social connections because of changes in peer groups (Gilman et al., 2003; Magdol, 2002). In summary, despite the differences in defining household dysfunction, the cumulative effects of multiple complex chil dhood exposures range from factors that frequently co-occur; (Dong et al., 2004) may re main persistent, or increase the risk of another traumatic exposure (P earlin et al., 1981; Pearli n, 1989; Turner & Lloyd, 1999). Moreover, there is evidence that adverse ch ildhood exposures in the form of household dysfunction (i.e., maternal de pression, family member in carceration, family discord) predispose an adolescent female to be more likely to experience depression when exposed to acute stressors than those not exposed to household dysfunction (BrooksGunn, 1991; Hammen et al., 2000). Contextual Strain: Exposure s to Community Violence One of the propositions of social stress th eory is that perceived stressors arise from a constellation of contextual stressors, ch ronic strain, and acute stressors (Pearlin et
27 al., 1981; Pearlin, 1989). Cichet ti and Rogosch (2002) called for a Â“developmental analysis of distal influences and the relationships to more proximal causes for psychopathology in youthÂ”. Specifically, these aut hors noted that contextual factors such as community violence can exert significan t effects on youth and that a developmental analysis would enable unders tanding of both the progressi on of experience and the trajectories followed. Over the past twenty-five years, the pr oblem of community violence within the urban areas of our nation has become a grow ing concern. Prevalence estimates for youth exposures to community violence range from 75% overall (National Center for Children Exposed to Violence [NCCEV], 2006) to 88% of inner city yout h (Mazza & Reynolds, 1999). Evidence of outcomes to exposures to co mmunity violence range from increased risk of homicide (Centers for Disease Contro l and Prevention [CDC], 2002), altering the developmental trajectories of youth, exacer bating distress reactions, strengthening perceptions of danger and hope lessness (Duckworth, Hale, Cl air, & Adams, 2000; Gerard & Buehler, 2004; Howard, Feigelman, Li, Cr oss, & Rachuba, 2002; Luthar & Goldstein, 2004; White, Bruce, Farrell, & Kliewer, 1998) and increased suicide risk (Mazza & Reynolds, 1999; Vermeiren, Ruchkin, Leckma n, Deboutte, & Schwab-Stone, 2002). Duckworth and colleagues (2000) tested the utility of a proximity model versus the chronic threat model with three forms of community exposure. The goal was to explain the separate and additi ve contributions of various forms of community violence exposures on psychological distress in adol escents. The investigators defined community violence using three dimensions: direct victimization, witnessing violence, and
28 community chaos (i.e., criminal activity, drug dealing, and the sound of gunfire near home or school). Multiple linear regression analysis found significant support for both the proximal model and the chronic threat model. Moreover, these researchers found that community chaos mediates the relationship between direct victimization and stress reactions. Perceptions of community safety, witnessi ng, and hearing of vi olence, and direct exposures are associated with various form s of psychological distress in both crosssectional and longitudinal st udies (Duckworth et al., 2000; Gutman & Sameroff, 2004; Howard et al., 2002; OÂ’Donnell, Schwab-S tone, & Muyeed, 2002; Scarpa, Heruley, Sumate, & Haden, 2006; White, Bruce, Farre ll, & Kliewer, 1998). In the Philadelphia Family Management Study, Gutman and colleag ues (2004) investigated the processes in successful adolescent development by asse ssing pathways from neighborhoods to families and the developmental course of adolescents. Evidence for a relationship between social environment and depression in longitudinal studies was demonstrated. Neighborhood problems (e.g. assaults, muggings drug use and dealing, and vandalism) were significantly associated with later adol escent depression but did not demonstrate an impact in early adolescence. These findings suggested that young adult females may be more vulnerable to contextual surroundings as they enter childbearing roles than their younger peers. In summary, findings in the literature support the hypothesis that the chronic stress of contextual exposures to chao s and community violence contribute to psychological distress in youth over time. Th ese exposures range from hearing about violence near oneÂ’s home or school (Scarpa et al., 2006) to witnessing or experiencing
29 personal victimization (Howard et al., 2002) Evidence of a grad ed relationship of cumulative exposures to community violence and psychological distress over time has been demonstrated (Gerard et al., 2004), yet discrepancies in the literature remain on desensitization to violence e xposures over time, (especiall y in males) and on buffering effects for psychological distress (White et al., 1998). Additional concerns stem from a lack of consistency in defini tion; many researchers include di rect victimization with the ambient exposures of chaos, witnessing and h earing of the violence (White et al., 1998). These researchers suggest that direct victimization should be analyzed independently of community violence. Acute Stressors: Victimization Acute stressors are part of the constella tion of stressors proposed by social stress theory (Pearlin et al., 1981; Pearlin, 1989). The determination that a stressor is acute is based on the level of threat, the quality of th e event according to desi rability and the level of disruption. Recently, there has been an increa sed interest in the impact of exposure to violent victimization during the developmen tal stages of child hood and adolescence in terms of morbidity and mortal ity across the life sp an (Felitti et al., 1998). Approximately 22% of children with learning disabilities acquired their disab ility as a result of severe child maltreatment (Karr-Morse & Wiley, 1997). Exposure to trauma occurring during a cri tical developmental period as in the example of intrauterine trauma or toxi c substance exposure often interrupts the development of the brain resulting in neurological deficits (Perry, 2001). The implications of trauma and maltreatment on children during sensitive developmental stages are often damaging due to active deve lopment of the cortex, which may result in
30 emotional, behavioral, and cognitive delays and impairment (Perry, 2001). Nevertheless, exposures to maltreatment during the developm ental stage of adolescence should not be minimized, as it is still considered a developm entally sensitive time. It is further proposed by Perry (2001) that chronic stimulation of the stress res ponse systems as in chronic maltreatment cases; (i.e., hypothalamic-pitu itary-adrenal, central nervous system, noradrenergic-dopaminergic systems) may re sult in alteration in functioning across emotional, behavioral, and cognitive domains. The developmental literature points to the pivotal tasks and vulnerabilities of brain fo rmation and regulation processes that if exposed to severe or chronic disruption during neurobiological maturation may be permanently altered (Perry, 2001; Rudolph et al., 2000). One example of such alteration after sexual abuse is the hast ening or premature onset of puberty (Finkelhor & Hashima, 2001). A large number of health/mental health outcomes have been associated with victimization, including: depr ession, anxiety, post traumatic stress disorder, suicidality, substance abuse, violence, teen pregnancy a nd risky sexual behavior s (Dube, et al., 2005; Finkelhor & Hashima, 2001; Howard et al., 2002; Jong, Mulham, & Kam, 2000; KendallTackett et al.,1993; Spat Wido m, 1999; Stevens, Murphy, & McKnight, 2003; Turner et al., 2006). In 2001, the national prevalence estimates of violent victimization for youth ages 16-19 was 5.6 percent or 1 out of 18 experien cing victimization (Child Trends, 2003). The ratio for sexual assault and rape for adol escents age 12-17 is 1.5 times greater than adults (Finkelhor & Hashima, 2001). These authors proposed that the reason for higher levels of victimization in youth is due to their dependenc y status. The distribution of
31 maltreatment of all youth range from 54% fo r confirmed cases of neglect, 22% for physical abuse, 8% for sexual abuse, 4% fo r emotional maltreatment, and 12% for other forms of maltreatment (Na tional Committee to Prevent Child Abuse, 1998). In a national representative sample of 2,030 youth, cumulative exposures to multiple forms of victimization exposures remained a significant predictor of adverse mental health outcomes (Turner et al., 2006). In a cross-sectional study of 2,603 youth with a mean age of 13 years, investig ators found that sexual abuse was both independently and directly associated with suicide attempts (Bergen, Martin, Richardson, Allison, & Roeger, 2003). Utiliz ing sequential logistic regression modeling these researchers found depression both mediated suic ide risk and directly impacted suicide attempts. In summary, the impact of victimi zation during the sensitive period of adolescence has an independent impact on mu ltiple health and mental health outcomes but is mediated and moderated by the type of event, injury, a nd relationship to the perpetrator. Cumulative measures of victimi zation appear to mediate antecedent stressors and strains. Conclusions While it is hardly novel to suggest that traumatic or adverse experiences have a significant impact on mental health outcomes, there is a concern that misattributing the potency of one risk factor (i .e., child sexual abuse) as an antecedent event that may predispose or maintain a problematic outcome across the life course may risk giving that event undue prominence while ignoring othe r factors (Felitti et al., 1998; Menard, Bandeen-Roche, & Chilcoat, 2004; Turner & Lloyd, 1995).
32 Even within the body of child maltreatment literature there is a focus on a number of overlapping risk factors that frequently co-occur, suggesting the multifinality of stressors (i.e., similar stressors associated wi th different outcomes) and equifinality (i.e., different stressors are associated with the sa me outcome in different people) (Menard et al., 2004). Social stress theory suggests that pe rceived threats arise fr om a constellation of various stressors and strains (P earlin et al., 1981; Pearlin, 198 9). It has been argued that adolescents are at greater risk in regards to social structures, expectations, community environments, and family experiences due to de velopmental sensitivities. An adolescentÂ’s emerging self-concept during this transitional tim e includes the crystallization of multiple systems, but also the simultaneous challenge of multiple adaptive experiences (Ge et al., 1994). When considering the cumulative impact, or in the rubric of epidemiology, dose response relationship, of exposures to house hold dysfunction, community violence, or personal victimization (Menard et al., 2004), a magnitude of mental health problems in adolescents, including depr ession, anxiety, post traumatic stress disorder, suicidality, substance abuse, violence, teen pregnanc y and risky sexual behaviors, have been identified in the literature (Felitti, 1998; Ge, 1994; Kessler et al., 1997; Turner & Lloyd, 1995). The Cumulative Effects of Adverse Exposures More recently, a conflict in the field exists regarding whether it is more valuable to pursue specificity of risk f actors, or cumulative risk models that mirror the real world of social organization. Researchers embracing cumulative approaches to risk posit that a primary danger in pursuing individu al risk constructs is that it may perpetua te stereotypes
33 about a persistently marginalized group in our society (as in the se xually abused) (Felitti, 1998; Kessler et al., 1997; Molnar Buka & Kessler, 2001; Schneider & Ingram, 1993; Turner & Lloyd, 1995). Thus, there is a risk of increasing personal burdens while limiting future opportunities in this population. While some researchers argue that because the field is still developing, a lack of evidence of specificity within adolescent literature exploring particular risk factors and the relationship to unique mental health out comes (a perspective of the socio-medical model or more precisely here developm ental psychopathology) can be expected (McMahon et al., 2003). Other researchers embrace the sociological paradigm that supports a broader category of psychological or emotional distress so as not to exclude mental health consequences (Aneshensel et al., 1991). A basic premise of social stress theory is that the effects of stress are non-specific, as evidenced by the empiri cal data of an array of disorder s that occur after an exposure to a stressor. The danger of this approach as argued by Aneshensel and colleagues (1991) lies in the misclassification of persons who ar e seemingly non-disorder ed due to specific categorization and thereby missing other manifestations of stress exposures. Nevertheless, studies vary in how out comes of psychological or emotional distress are defined. Some incl ude symptoms of emotional distress that impact daily functioning such as in school performance for youth, symptomatology as evidenced by various symptom check lists that include m easures of depression, anxiety, post traumatic stress disorder, or suicide ideation (Howard et al., 2002; Turner & Lloyd, 1995) and/or diagnostic evidence of psychiatric disorder. Symptoms associated with depression during
34 adolescence include: interpers onal and academic dysfunction, helplessness, anger, eating disorders, sexual promiscuity, runni ng away, and substance abuse. During the past two decades a substantia l amount of evidence on the association of chronic adversities and si ngle events of adverse expos ures and poor mental health functioning in adolescents has emerged (Berge n et al., 2003; Finkelhor & Hashima, 2001; Garmezy & Masten, 1994; Howard et al ., 2002; Johnson, Kotch, Ca tellier, Winsor, Dufort, Hunter, Amaya-Jackson, 2002; Jong et al., 2000; Kendall-T ackett et al., 1993; McMahon et al., 2003; Spat Widom, 1999; St evens et al., 2003; Turner & Lloyd, 1999; Turner et al., 2006). It is ar gued that higher rates for female mental disorder may be due to the relationship between chr onic stress and the role of care giving (the high cost of care giving hypothesis) and that s ubstance abuse and anti-social behavior in males are typically not included in epidemiological surveys (Aneshensel et al., 1991). Nevertheless, it remains that the third leading cause of death for adolescents between 15-24 years of age is suicide (Hazler & Melli n, 2004). Specific symptoms associated with suicide ideation are increas ed anxiety, depression, stress, hopelessness, and loss of self-esteem (H azler & Mellin, 2004). Multiple Stressors Across Domains Literature on multiple stressors across th ree or more domains is extremely limited. The majority of studies util izing cumulative approaches have looked at one or two domains, primarily in the areas of victimi zation or community violence, and have been cross-sectional in design. The fo llowing is the state of the lite rature that supports findings with respect to cumulative impact of a dversities across more than two domains.
35 The Dunedin WomenÂ’s Study conducted in New Zealand investigated the longterm impact of child physical, emotional, and sexual abuse acro ss the life span in a randomly selected community sample of 497 women (Mullen et al ., 1996). Results found that abuse was not randomly distributed but was more prevalent in disturbed and disrupted homes, with excessive moves contributing to poor outcomes. The researchers report that combined abuse variables expl ain a modest proporti on (1-5%) of overall distress within the context of others stressors. The National Comorbidity Survey (Kessl er et al., 1997) included twenty-six adversities that encompassed loss events, pa rental factors, and personal adversities. Multivariate analyses revealed additive effects on probability of disorder, where the cumulative effect of two or more adversities ha d a greater impact than the effect of one. It was noted that after the sixth adversity was added into the equation the researchers were unable to detect an effect. Moreover, these authors caution against interpreting results of a single-adversity as a causal in dicator of a single di sorder based upon the clustering of adversities in their results. A retrospective cohort study with 18,175 me mbers began a series of studies targeting the interplay between multiple stressors (Chapman et al., 2004; Dube et al., 2001; Dong et al., 2005; Felitti et al., 1998) The relationship of adverse childhood experiences and various health and mental h ealth outcomes were investigated utilizing multivariate logistic regression. A dose re sponse relationship was found with ACEÂ’s in any category increasing the risk of attempted suicide 2-5 fold (p<.001). A recent longitudinal study with 2-waves of data collection by Menard and colleagues (2004) with 1,715 participants f ound evidence that multiple stressors tend to
36 occur in clusters and that there is a co-o ccurring pattern of adversities. The domains studied were socio-demographic characteristics, including welfare stat us, 3 categories of child abuse (sexual, physical and emotional), parental warmth, parentÂ’s mental illness, suicidal behavior, household member substa nce abuse and incarce ration, and non-nuclear family structure. These investigators utilized both multiple regression analysis and latent variable modeling, adding to the existing li terature with a focus on person-centered variables. A longitudinal study from 1975 to 1993 with a community sample of 659 families found that maladaptive parenting and chil dhood maltreatment were associated with elevated risk for suicide attempts and inte rpersonal difficulties during middle adolescence (Johnson, Cohen, Gould, Kasen, Brown & Brook, 2002). Domains of adversities included poverty, critical life events, househol d dysfunction, and childhood abuse. A commonality of all of these studies is th at adversities tend to occur in clusters, and that identifying one event as causal of poor mental heal th outcomes could result in misspecification. Evidence for the cumulative eff ect of two or more adversities having a greater impact than the effect of one demonstrates a dose re sponse relationship with poor outcomes. Limitations of Current Literature The following section is an overview of the major gaps and limitations noted within the literature. Broadly, the gaps include: (1) a lack of theoretical testing; (2) a lack of longitudinal research; (3) a lack of examination of th e association between various contextual factors and social position with depression among fema le adolescents; (4) a need for the study of suicidal ideation (IOM, 2002) (i.e., given th at suicide is a relatively
37 infrequent event); (5) and finally, the n eed for the exploration of mediating and moderating influences on the impact of depression. The first significant limitation is the paucity of longitudinal studies on adolescents investigating depression and su icide ideation utilizing a theore tical framework (i.e., social stress theory that proposes that stressors arise from a constella tion of contextual stressors, chronic strain, and acute stressors (Pearlin et al ., 1981; Pearlin, 1989). Although noteworthy, the Kaiser Permanente studies have investigated distin ct adverse exposures on mental health outcomes, yet these studies ar e still restrictive in terms of elucidating clear mechanisms of influence. Data are often truncated into a single variable and therefore analysis neglects to consider the longitudinal nature, namely, chronicity on the developmental trajectory of mental health outcomes. A meta-analysis of socioeconomic ine quality and depression by Lorant and colleagues (2003) found compelling evidence for a causal relationship, but call for greater understanding of the course of devel opment of depression. Furthermore, a review of the literature disclosed th at there were no known studies examining the association of social position with depression among fema le adolescents (Mut aner et al., 2004). The Institute of Medicine (2002) calls for Â“specific da ta from well-defined and characterized populations whose community leve l social descriptives are well knownÂ”, in order to clarify the complex pathways fr om childhood experiences to mental illness and/or suicidality. An additional suggestion by the IOM (2002) is that given that suicide is a relatively infrequent even t, alternate endpoints, such as suicidal ideation, are required to add to the existing knowledge base.
38 There is a need for clarification of etiological processe s by assessing change longitudinally through identification of time of onset and the preceding temporal events in order to further understand the sequencing of events on me ntal health outcomes and to further the examine a causal chain (Thoits, 1995 ). Such information would guide future development of interventions that are optimally timed and able to address multiple causal pathways (IOM, 2002; Kraemer et al., 2001). Methodological limitations include the wi de variation in sample populations, variation in operationalization of constructs, and the exclus ion of various contextual factors. Study designs employed with a dolescents are typically cross sectional assessments, although longitudinal designs are pr eferable given the desire to establish causality. The studies noted above included a long interval of time between initial exposures and recall of experiences and outcome s. One last limitation of current research is the relatively little exploration of medi ating and moderating influences on depression and the potential for a unique combination of factors that place female adolescents at greater risk for suicide id eation (Hazler & Mellin, 2004). Mechanisms of Influence As stated above, there is a need for future research to test the impact of mediating and moderating factors in adolescents on depr ession and suicide ideation. Such evidence would help to inform prevention efforts and to further understand how these factors work in concert to evoke poor outcomes. Pathways to Risk Amplification: Enabling and Reinforcing Factors Intermediate factors along the causality c ontinuum are considered to either further enable or buffer the impact of negative influences; thereby co ntributing to risk
39 amplification, or mitigating against poor outcomes. An illustration of this are factors that contribute to poorer outcomes for child sexual abuse such as: age of abuse, the duration or frequency of abuse, type of sexual activ ity, child/perpetrator re lationship, number of perpetrators, victim gender, force or physic al injury, multiple forms of abuse, and perceptions of abuse (Barnett et al., 1997). The risk factors associated with sexual abuse are female gender, prepubescent age, passivi ty, divorced home, motherÂ’s employment, or disability, stepfather or boyf riends presence in the home, a nd maternal history of sexual abuse. Across all dimensions of maltreatment a nd stressful life events, the literature notes that gender, age, and socioeconomic di sadvantage are primary risk factors (Ge at al., 1994; Barnett et al., 1997). Detailed analyses have shown that dist al risk factors of poverty, while increasing the likelihood of pr oximal family dysfunction and contextual exposures to violence, have an indirect effect on depression yet remain part of the causal chain of direct effects through pr oximal factors (Rutter, 2005). As this study uses an all female samp le with socioeconomic disadvantage, age will be further discussed and explored as a mechanism of influence on outcomes. The literature discusses two prominent age range s for victimization through child abuse. These groups fall between 0 to 5 year olds ( 51%) to the adolescent group of 12 to 17 year olds representing (35%) of those reported (Barnett et al., 1997). Investigations on the experience of stressful life events have f ound that females between the ages of 13 and 15 report a dramatic increase in stressful exposur es followed by a slight decline in negative events after 15 (Ge at al., 1994).
40 Further illustration of mechan isms of influence contribu ting to child victimization is found in the study of household dysfunction parental substance abuse (Miller, Smyth, and Mudar, 1999; Murphy, Jellinek, Quinn, Sm ith, Poitrast, & Goshko, 1991). In another study, mechanisms of influence (the presence of both injury and clos e relationship to the perpetrator) were significant to the dose relationship between victimization and suicidality. The contribution of injury and relationship of perpet rator was (6.6%) in nonideators, and (58.9%) in those with su icidal behaviors (Simon et al., 2002). Conclusions This study aimed to contribute to the existi ng knowledge base in three ways: first, by investigating the prevalence of stressors across multiple domains in the lives of female adolescents raised in the cont ext of welfare. Secondly, to capitalize on longitudinal data to clarify the cumulative impact of exposur es to adverse events and the change of depression, suicide ideation, and mental dist ress over time; and lastly, to explore the mechanisms of cumulative exposures of adverse exposures on depression, suicide ideation, and overall mental distress. Th e conceptual framework for testing the independent effect of the constructs of personal victimization, household dysfunction, and exposure to community violence as well as the additive impact of these cumulative measures of adverse exposures on depression, suicide ideation, mental distress and the change in theses outcomes is presented in Figure 1. The empirical evidence presented above provides justification fo r each of these pathways. A more detailed conceptual model with each category deve loped by dimensions is presented in chapter three.
41 Figure 1 Conceptual Model for Testing Cumulative Impact of Adverse Exposures on Adolescen t Mental Health Outcomes Control Variables Mental Distress Depression Suicide Ideation + Adverse Adolescent Exposures Household Dysfunction Personal Victimization Community Violence + +
42 CHAPTER THREE METHOD This chapter describes the methods used to conduct this study. It is divided into the following sections: (1) study purpose; (2) research hypotheses; (3 ) overview of study design; (4) description of the Welfare Reform : Adolescent Girls in Transition data set; (5) study sample; (6) original data collection met hods and instrumentation; (7) study variable selection; (8) data analysis; and (9) design strengths and limitations. Study Purpose The purpose of this study was to dete rmine the cumulative impact or Â“dose responseÂ” effect of adverse experiences on depression, suicide ideation, and overall mental distress over time in a cohort of 125 adolescent girls raised on public assistance. The adverse experiences studied were pers onal victimization, hous ehold dysfunction, and community violence exposures. The devel opment and change in depression, suicide ideation, and mental distress ov er the four points of time dur ing the study period served as the outcome variables. Research Hypotheses 1. Exposure to adolescent victimization w ill have a positive association with the presence of depression, suicide idea tion, and overall mental distress. 2. Household dysfunction will have a positive association with the presence of depression, suicide ideation, a nd overall mental distress.
43 3. Exposure to community violence will have a positive association with the presence of depression, suicide idea tion, and overall mental distress. 4. The number of adverse exposures w ill have a cumulative impact or Â“dose responseÂ” relationship with th e level of depression, su icide ideation, and overall mental distress. 5. The Adverse Adolescent Exposure Score wi ll be significantly related to change in depression, suicid e ideation, and mental distress over time. Overview of Study Design This study was based upon secondary analys is of data from the Welfare Reform: Adolescent Girls in Transition (WRAGT) (Boothroyd et al., 2003; Boothroyd et al., 2005), a mixed method longitudinal study examini ng the impact of welfare reform on the future hopes and aspirations of female adolescen ts. Participants in the original study were part of a larger project funded by the Flor ida Agency for Health Care Administration (Contract #M0107). Data for the current study were drawn from all four waves of interviews conducted with the mothers and their daughters. Both univariate and bivariate analyses were performed to assess completeness of da ta and associations between individual predictors, composite variables, and outcome variables. A combination of multivariate analyses were used including ordinary l east square (OLS) regression and logistic regression models developed to explore the impact of adverse experiences on depression, suicide ideation, overall mental di stress and change over time.
44 Description of Welfare Reform: Adol escent Girls in Transition Data Set The Sampling Frame and Process Mothers receiving TANF (at the time of enrollment into the study) and their adolescent daughters were identified from the 2000-2001 Florida Medicaid eligibility data using the family identifier and other ma tching variables (such as gender, address, and last name). To be eligible for study pa rticipation, mothers had to be receiving TANF and have a daughter between the ages of 13 and 18 living at home. Additionally, the families had to reside in a five county area in west central Florida. Recruitment letters were mailed to 873 potential participants, whereby 125 eligible daughter/mother pairs were recruited for participation. Enrollment wa s stratified on race/ethnicity (i.e., white, black, Hispanic) and geographic loca tion (i.e., urban versus rural). Attrition Rates. Given that subject attrition pr esents significant methodological challenges in longitudinal studies, attrition ra tes were assessed in planning the current study. Retention rates for both mothers and daughters at each of the follow-up interviews are summarized in Table 1. In order to de scribe those who left the study, additional information regarding barriers to participati on were gathered. Barri ers for participants who were not re-interviewed included, unr esponsiveness, failure to keep multiple interview appointments, maternal involvement in a drug rehabilitation program, incarceration, maternal deat h, disavowal of previous participation in the study, outstanding warrants for arrest, and placement in foster care where permission could not be obtained for an interv iew (Boothroyd et al., 2005).
45 Table 1 Follow-up Retention Rates Year Phase 1 Mothers Daughters N % N % 2002 125 100% 125 100% 2003 113 90.4% 116 92.8% 2004 107 85.6% 111 88.8% 2005 1131 90.4% 1152 92.0% Note. Reprint from Source: (Boothroyd et al., 2005). Welfare reform: Adolescents girls in transition Â– A three-year follow-up study. 1 Three mothers were deceased. 2 One daughter was deceased. Strengths and Limitations of Study Sample. 1. A primary strength of this cohort sample is that this is a homogeneous sample in regards to exposure to welf are status, a proxy for poverty. 2. The sample was selected using random sampling procedures from an enrolled population then stratified on race/ethnicity and geogra phic location (i.e., urban versus rural). 3. Another general strength of a cohort st udy sample is that there is stronger evidence for exposure-disease associations given the homogeneity of the groups experiences. Cohort studies allow for greater determin ation of risk of exposures over other sampling stra tegies (Friis & Sellers, 1999). 4. An additional strength of the study sample is that trained interviewers conducted face-to-face interviews in participantÂ’s homes.
46 5. A common limitation of longitudinal cohor t studies is loss of data due to attrition. However, as noted above, ther e were high rates of retention for the entire length of the study thus creating a minimal limitation for this sample. 6. A final limitation is the constraint of a small sample size to support the complex analyses. 7. Age variations in the girls (13-17) in the first year of the study may have contributed to differences in perceptions and reports of adverse experiences. Original Data Collection Methods Each year of the study, face-to-face interviews were conducted by trained interviewers in the homes of participants us ing various standardized measures with 125 mothers who were receiving TANF. These proc edures were initially conducted in 2002, and repeated in each of the three followi ng years providing four waves of data. All procedures and protocols were reviewed a nd approved by the UniversityÂ’s Institutional Review Board prior to th e initiation of the study. Instrumentation The WRAGT was a comprehensive study ut ilizing 35 psychomet rically tested measures including respondent self-report health, mental health and substance abuse status measures (See Appendix A & B for a full list of measures). In addition, information on respondent demographics, family characteristics, and living situations were included in both the mother and daughter protocols. The original study assessed demographic data using the Standards fo r Mental Health Decision Support Systems developed by Leginski and colleagues (National Institute of Mental Health [NIMH], 1989). The instrumentation and questions pertin ent to this study are reviewed below,
47 organized by constructs within the concep tual framework for te sting the effect of stressors and strains on depr ession and suicide ideation. Data Integrity Preliminary data analyses were pe rformed on the original WRAGT 2002 to 2005 data set to ascertain the feasibility of conducting the current study. Specifically, to determine whether the data set provided appropr iate variables to test the current studyÂ’s proposed objectives of replicati ng the ACE study with adolescence. Study Variable Selection The proposed theoretical framework requi red that variable s address adverse adolescent experiences that ranged from personal victimization, household dysfunction, and community violence exposures as well as the outcomes of inte rest: the development or change in depression and suicide ideati on over time. Evidence from the preliminary review of the data demonstrated sufficient data to construct the categories of interest and sufficient variability within the sample making the study feasible. To narrow the focus of this study, only variables closely aligned with the approach utilized by Felitti and colleague sÂ’ ACE (1998) study were included. The exceptions are adolescent behavioral variables selected as control variables and variables relative to the operationalization of community violence. Justification for these additions to the ACE model were based on the need for accommodating adolescent female behaviors that impact the transition into a dulthood (e.g. pregnancy and school status) and to include contextual exposures of comm unity violence to further understand the influence these exposures may exert upon outcomes.
48 Missing Data Strategies Strategies for handling missing data range from ignoring the missingness to imputing data with predictive models (Buhi, Goodson, & Neilands, 2008; McKnight, McKnight, Sidani & Figueredo, 2007). The foll owing section discusses the classification of missingness as one of three probable mech anisms, and further defines and examines strategies to approach th e current data set. The first classification of missingness is mi ssing at random (MAR), or data that is missing not as a function of the item but as a function of some other observed variable (Buhi et al., 2008; McKnight, 2007) An example of where this may occur in this study would be variables regarding teen pregnanc y. The section of the protocol regarding sexual experiences may have been skipped if th e interviewer was told that the adolescent had not yet had a sexual encount er. This could be attributed to the adolescentÂ’s age and could be considered MAR, and therefore, ignorable missingness. Missing completely at random (MCAR), or missingness no t resulting or related to an observed variable or incomplete data is also considered ignorab le. MCARÂ’s may result from cases where a participant is lost to follow up due to deat h or illness (Buhi et al., 2008). Not missing at random (NMAR) data, or missingness due to syst ematic influences, such as incriminating questions or missing due to unobs erved data is considered to be non-ignorable as it produces biases in the results that require interpretation. Given that there is complete baseline data for all cases across demographic questions, indices of adverse e xposures, and outcome variables, those cases that were lost to follow up were assumed to be MCAR, thus the data for subsequent years were considered as ignorable and not in need of imputation, deletion, or direct estimation
49 (White, Carpenter, Evans, & Schroter, 2004). Nevertheless, justific ation for not deleting cases from the baseline year was due to the small size of the samp le, and as correlation, and prevalence rates were available on trauma tic exposures and outcome scores in the first year of the study each of the cases were retained in or der to preserve sample size. Missing data for the outcome variables of depression and suicide ideation are presented below. In the first year of the study, 125 adolescents completed the CES-D. In at least one of the four years of the st udy there were 13 cases overall (10.4%) missing CES-D scores in one of the four years. There were nine cases (7.2%) missing CES-D scores during two year of the four years. There was only one case w ith complete baseline data that was lost to follow up in the three subsequent years of th e study. The results are presented in Table 2. Table 2 Number of Cases Missing Depression Scores and Suicide Ideation Item Frequency Percent Complete Data 102 81.6 Missing 1 Year 13 10.4 Missing 2 Years 9 7.2 Missing 3 Years 1 .8 Total 125 100.0 To insure data integrity a review of each item of the CES-D was conducted at each point in time during the study. This analys is revealed one or less items missing per year on three cases. An imputation technique presented in the literature for the CES-D
50 has been used to retain those cases. Specifica lly, that a missing item be assigned the mean value of the completed items (Garris on, Addy, Jackson, McKeown, & Waller, 1991). An analysis of the correlation betwee n observed items of the CES-D and the imputed items data set resulted in a perfect (1) correlation at (p = .01). For missing cases found during the follow up years, a regressi on analysis was performed to develop a predictive equation base d upon previous year(s) depression sc ore. These scores were then individually reviewed and then imputed into the final variable in order to retain the cases and preserve power for further analyses. Each of the predictor variables is a composite index. Due to the small sample size, deletion techniques were not used in this study. It is important to note that the central assertion of the entire study is that each of the predictor indices (i.e., victimization, household dysfunction, and commun ity violence) are measuring the cumulative effects of each of the relevant dimensions comprising the overall variable. Therefore, the same strategies utilized in the ACE studies for missing information on an exposure to trauma for an adolescent were utilized. Specificall y, those items not endorsed within a category of adverse exposure and counted as missing were dummy coded as Â“0Â”; thus assuming that the adolescent did not have that experience. While the results may underestimate exposures; misclassifying youth as unexposed and creating a bias towards the null, this is a minimal concern for this study given the preliminary analysis of the three adverse exposure composite variables revealed prevalence rates ranging from (80% 92.8%). As this is not a study of severity or chronicity but of exposure, there seems to be sufficient data for the purposes of this study.
51 Further rationale for this approach is in the distinction between data and information. Specifically, data on one exposure to parental divorce, criminality, or personal sexual assault provides sufficient information to be included and counted as an exposure to a phenomenon (McKnight et al., 2007) Thus, a score of Â“1Â” in the dimension of a criminal household member (a data point ) which at any point dur ing the 4 years still counts as a score of Â“1Â” whethe r a participant reported exposur e in the other three years, or did not complete the questi on in the other three years. As this is not a study of severity or ch ronicity but of exposure that data point provides us with information on the exposure to such an occurrence. This approach eliminates exhaustive efforts of dealing with a missing data point on a single item during the entire course of the study give n the infrequency of missing items. Behavioral and Demographic Control Variables Demographic and behavioral data were selected to statistically control for potential confounding influences. Control vari ables selected include daughtersÂ’ age, ethnicity, teen pregnancy, and school status. The speci fic items asked during the interviews along with response op tions are listed in Table 3. The descriptive statistics for the above control variables, as well as selected demographic characteristics for the sample at the time of enrollment in the WRAGT study are presented in Table 4.
52 Table 3 Behavioral and Demographic Control Variables Variable Item (s) Response Option Daughters Age What is your da te of birth? Continuous Daughters Race /Ethnicity How would you describe your race/ethnicity? 1=White 2= Black/African American 3= Hispanic 4=Other Teen Pregnancy Have you ever been pregnant? 0=No 1=Yes School status Demographic Questions Are you still in school? Have you had any college, business or technical school? Reason not in school? 0=No 1=Yes Qualitative
53 Table 4 Charactistics of the Mothers and Daughters Characteristics Mothers 2002 (n = 125) Daughters 2002 (n = 125) Age: Mean SD Range 38.4 4.99 30 53 15.5 .99 13 17 Race/Ethnicity: White Black/African American Hispanic 40.7% 38.2% 21.1% 33.6% 40.8% 25.6% Marital status: Married or living as married Divorced, Separated, or Widowed Never married 12.8% 54.4% 32.8% 0% 0% 100% Education: Dropped out of school Completed high school Teen Pregnancy: Ever pregnant 50.4% 49.6% 20.8% NA 15% Length of time on TANF: Less than 6 months Six months to 1 year 1 to 2 years Over 2 years Not on TANF 15.4% 18.7% 23.6% 42.3% 0% None
54 Predictor Variables This section presents variables selected from the larger data set based upon a demonstrated linkage in one or more previ ous studies as contribu ting to depression or suicide ideation. Furthermore, the variables were selected to closely replicate the ACE studies (Felitti et al., 1998) with the additi on of exposures of community violence. The contextual exposures were added to further assess the contribution of community violence on depression and su icide ideation as well as the potential interaction with household dys function and adolescent victim ization. Previous research suggests that exposures to neighborhood violence augment the effects of maternal depression on child outcomes, exerting a sort of double jeopardy thr ough this interaction (Silverstein, Augustyn, Cabral, & Zudkermean, 2006) It further suggests a need to assess the role of maternal victimization, suggesting such trauma may explain a motherÂ’s lack of responsiveness to he r childÂ’s exposure to violence (Simon, Anderson, Thompson, Crosby & Sacks, 2002). For the purposes of this study, each cat egory of exposure was a composite variable constructed with each of the indi cators of adverse exposures. Justification offered by previous research indicates that defining abuse or traumatic experiences by any adverse event during childhood is wrought w ith difficulties as we try to delineate from undesirable mundane exposures to events that are more critical. The current literature suggests that givi ng weight to one form of traumatic exposure minimizes the powerful additive eff ects of multiple exposures and even the multiplicative effect or ripple effect overtime of such exposures.
55 There is a greater consensus in the literature that abuse and victim ization is not randomly distributed but is instead found predominan tly in disrupted and dysfunctional families with economic and social disa dvantage (Mullen et al., 1996). The specific items selected for each categor y of this study mimic the constructs of the ACE study and as closely as possible replic ate, the line of questioning in regards to exposures to physical and se xual trauma and household dysf unction. It is important to note that one of the underlying reasons for th e differences in the line of questioning regarding constructs of abuse is that the ACE was completed by adults participating in a health care plan through a mailed surv ey with an average age of 56. In contrast, the WRAGT was administered face to face with youth of an average age of 15. The potential ethical dilemmas for the research team of the original WRAGT study and the implications for youth reporti ng abuse by a parent or adult household member would have made approval by the In stitutional Review Board (IRB) virtually impossible as well as limiting parental consent for participation in the study in general. Thus, there were no questions within the WR AGT in regards to emotional abuse by a parent figure. There are questions regarding physical and sexual assault. The questions from the ACE study regarding exposure to abus e are presented in Ta ble 5. For ease of comparison, the items selected from the WRAGT on exposure to victimization are presented in Table 6.
56 Table 5 Items from ACE Study on Abuse Abuse by Category Item Emotional Did a parent or ot her adult in the householdÂ… a. Often or very often, swear at you, insult you, or put you down? b. Often or very often, act in a way that made you afraid that you might be physically hurt? Physical Did a parent or ot her adult in the householdÂ… a. Sometimes, often or very ofte n push, grab, slap, or throw something at you? b. Ever hit you so hard that you had marks or were injured? Sexual Did an adult person at least 5 years older everÂ… a. Touch or fondle you in a sexual way? b. Have you touched their body in a sexual way? c. Attempt oral, anal, or vagi nal intercourse with you? d. Actually have oral, anal, or vaginal intercou rse with you?Â” Table 6 Items for Measurement of Exposure to Victimization Variable Item Response Options Adolescent Victimization Have you ever beenÂ…(or in the past year) a. bullied/pushed around? b. beaten up? c. sexually assaulted? d. robbed? e. stabbed? f. shot at? g. shot? 0=No 1=Yes Abused Have you ever been abused by anyone, verbally, physically, sexually or psychologically? 0=No 1=Yes Sexually assaulted In the past year have you experienced being sexually assaulted? 0=No 1=Yes
57 The literature suggests that there is a limitation in the non-specific or general question of sexual assault; typically, results pr oduce a significantly lo wer rate of response than questions that are specific, thus the re sulting data is likely to be a conservative estimate of potential exposures to sexual a buse (Mullen et al., 1996) Once again, it is important to note that a strength of the methods utilized in data collection by the WRAGT study team are that the questions we re asked in face to face interviews with trained interviewers in contrast to the ACE studies which utilized only mailed surveys to gather data. Adolescent Victimization This composite variable was operationalized as: (1) first year reports of ever being bullied, beaten up, sexually assaulted, robbed, st abbed, shot at, shot; (in subsequent years this question reads Â“in the past year have you been bullied, etc.) (2) positive responses to questions on physical, psychological, and se xual abuse; (3) repor t of being sexually assaulted in the past year. Justification for including this question again is due to documented underreporting of sexual assaults. This particular question is found in a different location of the protocol embedde d in the Life Events Inventory (Monaghan, Robinson, & Dodge, 1979). A second comparison category from the ACE study is from the category of household dysfunction regarding mental illness: (1) Â“Was a household member depressed or mentally ill? (2) Did a hous ehold member attempt suicide?Â” In contrast, the items in this study for the same category of parental mental illness are: ( 1) Â“In the past year have you experienced your parent having emotional /psychiatric problems?; (2) (Mother) In the past month how often have you felt like hurting or killing yourself?Â” As in the ACE
58 study (Felitti et al., 1998) respondents in this st udy are defined as expos ed to a category if they responded Â“yesÂ” to 1 or more of the questions within the category. Household Dysfunction This composite variable was operationaliz ed as: (1) family substance abuse; (2) maternal mental illness; (3) maternal victimiza tion; (4) parental criminality; (5) parental separation/divorce; (6) reside ntial instability. The questi ons and response options are presented in Table 7. Community Violence This composite was operationalized as a variable based on re ports of: (1) seeing someone sexually assaulted, robbed, stabbed, shot at, shot, kill ed; (2) feeling afraid with adults, in the neighborhood, or at school; (3) knowledge of weapons: people who own a gun, bring a knife to school, br ing a gun to school, or othe r weapons to school; (4) the hearing of gunshots in the neighborhood. The questions and response options for this category are presented in Table 8.
59 Table 7 Items for Measurement of Household Dysfunction Variable Item Response Options Family Substance Abuse (Mother) Have you ever had a drinking or other drug problem? Have any of your family members ever had a drinking or other drug problem? 0= No 1=Yes Maternal Mental Illness In the past year you have experienced your parent having emotional/psychiatric problems? 0=No 1=Yes In the past month how often, have you felt like hurting or killing yourself? 1=At least every day 2=Several times a week 3=Several times during month 4=Once during month 5=Not at all Maternal Victimization Have you ever been abused by anyone either verbally, physically, sexually or psychologically? 0=No 1=Yes Parental Criminality In the past year you have experiencedÂ….. a. your parent going to jail/prison for a year or more? b. your parent going to jail/prison for 30 days or less? 0=No 1=Yes Parental Separation/ Divorce In the past year you have experienced: a. separation of your parents? b. divorce of your parents? 0=No 1=Yes Residential Instability How many places have you lived in the past year? [include current residence] Continuous
60 Table 8 Items for Measurement of Expos ure to Community Violence Variable Item Response Options Exposure to violence Have you ever seen someoneÂ…. a. sexually assaulted? b. robbed? c. stabbed? d. shot at? e. shot? f. killed? 0=No 1=Yes Feeling Afraid Do you feel afraid Â… a. with adults? b. outside in neighborhood? c. at school? 1=Very 2=Somewhat 3=Not very Presence of weapons Do you know kids whoÂ… a. own a gun? b. bring a knife to school? c. bring a gun to school? d. bring other weapons to school? 0=No 1=Yes Presence of gun shots in the neighborhood How often, if ever, do you hear gunshots in your neighborhood? 1=Almost every day/night 2=Once or twice a week 3=Once or twice a month 4=Once or twice a year 5=Have never heard gunshots
61 Outcome Variables Depression Depression was measured utilizing th e Center for Epidemiology Studies Depression Scale (CES-D) (Radloff, 1977), a se lf-report 20-item meas ure of depression. The CES-D is designed to measure depressi ve symptomatology in the general population for both adolescents and adults. With adoles cents, the CES-D has demonstrated positive predictive value in measuring major depres sion, dysthymia, and ps ychiatric disorder (Garrison et al., 1991). The line of questioni ng instructs respondents to indicate the frequency they have experienced a feeli ng during the past week on a 4-point scale. Response options include 1= Rarely or none of the time (less than 1 day); 2 = Some or a little of the time (1-2 days); 3 = Occasionally or a little of the time (3-4 days); 4 = Most or all of the time (5-7 days). In order to score the CES-D the four positively worded items are reverse coded in order to f it the direction of responses for all items. Each item is then recoded on a 4-point scale to provide a range of zero to three; O for Â“Rarely or none of the timeÂ” to a score of 3 for Â“Most or all of the timeÂ”. The scores are computed by summing the recoded items. The sum of the 20 items provides a range from zero to 60 with scores greater than or equa l to established cut-offs indicating potential depression. The items for the CESD are presented in Table 9.
62 Table 9 Items Comprising the CES-D Scale Please tell me which answer best describes how often you felt or behaved this way in the past weekÂ…. a. I was bothered by things that usually donÂ’t bother me. b. I did not feel like eating; my appetite was poor. c. I felt that I could not shake off the blues even with help from my family or friends. d. I felt that I was just as good as other people. e. I had trouble keeping my mind on what I was doing. f. I felt depressed. g. I felt that everything I did was an effort. h. I felt hopeful about the future. i. I thought my life had been a failure. j. I felt fearful. k. My sleep was restless. l. I was happy. m. I talked less than usual. n. I felt lonely. o. People were unfriendly. p. I enjoyed life. q. I had crying spells. r. I felt sad. s. I felt that people disliked me. t. I could not get Â“goingÂ”.
63 The CES-D scale is deemed highly relia ble for both adult and adolescent populations with internal reliability rangi ng from .82 (Taylor & Turner, 2002) to .85 (Radloff, 1977). There is evidence of reliab ility for using the CES-D for adolescents ranging in age from 12 to 18, the alpha values obtained for groups of both male and females ranged from 0.87 (Garrison et al ., 1991) to .91 (Reifman & Windle, 1995). The commonly accepted cut-off score for Â“clinical ca senessÂ” in adults is a score of 16 or above (Kalil et al., 2001). For adolescents Garrison and colleagues (1991) determined that the optimal screening cut points are 22 or above for females and 12 or above in male adolescents (12-15 year s old) or grades 7th and 8th. For older adolescents (16 to 18 years old) or grades 9th through 12th these authors recommend a cut off score of 24 or above fo r females and 22 in males (Garrison et al., 1991). In this study, given the ages range from 13 to 18 at the star t of the study both cut point scores for the two different groups of young adolescents and older adolescents have been implemented. Specifically, from 12 up to 15 years of age have a cut off score of 22 and 16 to 18 years of age cut off score of 24. In the last two years of the study, those ranging from 18 years old and above are scor ed at the cut-off score for Â“clinical casenessÂ” in adults a score of 16 or above (K alil et al., 2001). In the first year of the study, the internal reliability for th e CES-D obtained an alpha of 0.83. In this study depression scores were used both as continuous and nominal variables. As a continuous variable, the depression score was used in the preliminary analyses in establishing correlations acro ss each of the four years of the study. As a dichotomized variable, scored as (1 = the presence of depression) and (0 = absence of depression), it was used in selected logi stic regression models to distinguish between
64 groups. In order to assess change in depres sion from Year 1 to Year 4 of the study a change score was developed from the CE S-D continuous score using the following equation: (CESD Year 4 Â– CESD Year 1 = CESD). Suicide Ideation Suicide ideation is measured using one item taken from the Pediatric Symptom Index (Jellinek, Murphy & Burns, 1986). Res pondents were asked during the past month (How often have you felt like hurting or ki lling yourself?). Response options were (1= often); (2 = sometimes); or (3 = never). For scoring the entire instrument, these responses were reversed scored to (2 = often); (1 = sometimes); (0 = never). As a continuous variable, it was used in the preliminar y analysis to esta blish correlations. As dichotomized, this variable was colla psed and recoded to Â“0Â” = (Never) and Â“1Â” (Any Suicide Ideation). In order to assess change in su icide ideation from Year 1 to Year 4, a change score was developed for the suicide id eation (continuous) variable utilizing the following equation: (SUICIDE Year 4 Â– SUICIDE Year 1 = SUICIDE IDEATION). Mental Distress Due to small sample size a final outcome va riable was also constructed by combining those scoring above the cut off score on the CE SD in each year with those with suicide ideation in the same year. These scores we re developed in each year as continuous variables and then dichotomized and recoded in to a final variable that was scored as Â“0Â” = (Never) and Â“1Â” (Any Mental Distress) fo r ever having experienced mental distress during the entire study.
65 Data Analysis A logistic regression analysis was em ployed to adjust for the potential confounding effects of age, race, educationa l attainment, and teen pregnancy on the relationship between the number of adve rse exposures and depression and suicide ideation. To test for the dose response relati onship of adolescent adverse exposures the number of exposures by category were entere d as a cumulative dichotomous variable (0, 1, 2, 3, 4 Â…) for each dependent variable. The measure of exposure was simply the sum of exposures across domains; the number of exposures ranged from Â“0Â” (unexpos ed) to Â“12Â” (exposed to all categories). A description for the univariate, bivariate, and multivariate analyses follows. Level One Analyses To prepare the data for analysis, frequency distributions were generated to test for variability, to assess the degree of missing data and to assess prevalence for each of the variables. Where indicated response opti ons may have been collapsed based on distributions. For continuous variables, m eans, standard deviations, skewness, and kurtosis were generated. Behavioral and Demographic Control Variables Demographic and behavioral variables used to control for confounding effects included: (1) daughters age, (2) daughter race /ethnicity, (3) educational status, and (4) teen pregnancy. The educational status vari able was constructed utilizing items from across all four years of the study that specifically asked about enrollment in school, reason for not being enrolled, whether a high school diploma or GED was attained and
66 whether or not the respondent had participat ed in any post secondary studies. Once again, due to limited sample size, two of th e demographic variables were recoded. The race/ethnicity variable was recode d (0 = white and, 1 = non-white) for logistic models and age was recoded into (0 = 15 years old and, 1= to 16 years old) reflecting the two different cut off scores fo r the CES-D of 12 to 15 years of age and 16 to 18 years of age suggested in the literatur e (Garrison et al., 1991; Kalil et al., 2001). Predictor Variables Data representing any of the adverse adol escent experiences within each category of exposure were re-coded as dichotom ous responses of Â“0Â” (no event) or Â“1Â” (yes, an occurrence of th e event). The four exceptions requiring that items be recoded are as follows: (1) (Mother) In the past month how often have you felt like hurting or killing yourself? (1=At least every day, 2 =Several times a week, 3=Several times during the month, 4=Once during the month 5=Not at all). This item was recoded to (0= Not at all, 1= Any positive response in options 1-4). (2) How often, if ever, do you hear gunshots in your neighborhood? Responses have been recoded into a ca tegorical variable: (0=Never, 1= one or more times in past year). (3) Do you feel afraid Â…a.) with adults; b.) outside in neighborhood; c.) at school? (1=Very, 2= Somewhat, 3= Not Very). The item has been recoded into (0= Not Very and /somewhat, / 1= Very). (4) How many places have you lived in the past year? This was originally a continuous variable. The response option wa s recoded such that any response of
67 more than one move during the four year study has been recoded into a (0 = no / 1 = moved). Composite Variables. Respondents were defined as exposed to a category of event if they responded Â“yesÂ” to one or more of the questions within the category. These variables were constructed ba sed on data across the four obs ervations. First, a variable was computed to determine the number of tim es an event was reported across the four points of time. The resulting variable could range from zero if an adolescent reported no exposure during any of the interviews to f our if they reported exposure during each interview. Next, each of these continuous e xposure variables, for every event category, were recoded into different dichotomous va riables of 0 = not exposed to 1 = exposed. Specifically, a zero was given to adolescents who reported no exposure across the four points of time or a one if they reported a ny exposure at any point during the study. If a participant reported an event at one point in time, the event was counted as occurring. If they reported the adverse expos ure at two or more times, th ey still only received a score of Â“1Â”. This approach offers an index of exposure but not the level of severity or chronicity of an exposure. Outcome Variables The dependent variables (CES-D scores, suicide ideation, and mental distress) were assessed across all four points of time. Continuous outcome variables for each year were preserved for correlations and linea r regressions and then dichotomized for purposes of logistic models as well as th e development of an overall mental distress outcome variable.
68 Level Two Analyses The second level of analysis tested bivari ate associations for each of the control, predictor, and outcome variables. Just as w ith the univariate analysis, separate analyses were conducted for each level of variable. In order to establish co rrelation coefficients, Pearson correlation matrices were computed for each of the individual items used to construct the cumulative scores for each ca tegory of adverse exposure. Measures of association examined the relationships betw een each socio-demographic variable (i.e., age, race/ethnicity, and educa tion), individual predictor item s, and outcome variables. As this study seeks to replicate previ ous adverse childhood exposures studies, as in the ACE studies, the same methods were employed for the overall construction of the cumulative indices within each of the categ ories of adverse adolescent exposures: (1) adolescent victimization; (2) household dys function; and (3) community violence. A cumulative variable for each of the observed experiences has been constructed based on dichotomous variables described in th e previous section. The use of an overall cumulative measure of traumatic events or Â“adverse adolescent exposuresÂ” has been recommended to provide a more accurate indi cator of stress leading to poor outcomes (Smith, Leve, & Chamberlain, 2006). A model for testing the cumu lative effect of adverse events on depression and suicide ideation is depicted in Figure 2. The three major constructs of concer n (e.g. victimization, household dysfunction, and community violence) were operationalized by the items included in the developed composite scores. A sum of each of these three scores created the overall or total cumulative score of adolescent adverse expos ures experienced. A respondent was defined as exposed to an event if they respond Â“yes Â” to one or more of the questions within a
69 specific category. For each respondent, the number of adverse adolescent exposures (AAE) was summed to create an AAE score, which ranged from Â“0Â” (unexposed) to Â“12Â” (exposed to all categories). The final AAE score was an interval variable used as a summary measure for the cumulative effect of multiple exposures to: (1) adolescent victimization; (2) household dysfunction; and (3) community violence. In summary, the construction of the predic tor variables was a three-step process: Step One Â– A case count for each item in 12 dimensions reported across four points in time: (0 = No Exposure) (4 = Exposure Each Year). Step Two Â– Each continuous adverse exposure was recoded in to a dichotomized variable: (0 = No Exposure) (1= Exposed). Step Three Â– Computed AAE Score : A sum of exposures across 12 dimensions (range 0 to 12).
70 Figure 2 Operational Model for Testing the Cumulative Effects of Adverse Exposures on Depr ession and Suicide Ideation Changes in Outcomes Mental Distress (Yr4-Yr1) Depression CES-D Score Suicide Ideation (1) Scaled Adverse Adolescent Exposu res Composite Score Household Dysfunction (6 ) Family Substance Abuse (2) Maternal Mental Illness (2) Maternal Victimization (1) Parental Criminality (2) Parental Separa tion/Divorce (2) Residential Instability (1) Adolescent Victimization (2 ) Experience of abuse (3) Ever been assaulted, shot, or stabbed, etc. (6) Community Violence (4) Seeing someone, sexually assaulted, shot, stabbed, or killed, etc. (6) Fear with adults, in neighborhood/school (3) Presence of weapons (4) Hearing gun shots (1) + + Control Variables (4) Daughters Age Daughters Race/Ethnicity School status Teen Pregnancy
71 Level Three Analyses To test the first three hypot heses, the individual impact of exposures per category, the adjusted odds ratios (ORs) and 95% c onfidence intervals (CIs) from logistic regression models were used. In order to adjust for confounders, socio-demographic variables of age, race, educational status, a nd teen pregnancy were entered into each of the models first to isolate the relationships of adverse exposures on the presence of depression, suicide ideation, and overall mental distress. To test hypotheses four and five, the imp act of cumulative adverse exposures to the level of depression, suicide ideation, a nd overall mental distress the number of exposures were entered as a single ordinal variable (0, 1, 2, 3, 4 Â…) variable into multivariate regression models with continuous outcome variables of depression, suicide ideation, mental distress, and computed cha nge scores. Due to small sample size this procedure was repeated with a grouped AAE scor e. The first cut off score of four adverse exposures was based upon the general consensus within the risk fact or literature, which states that having 4 or more risk factors increases poor outcomes. Specifically, Kazdin and colleagues (1997) found that exposure to one risk factor did not necessarily lead to a poor outcome, but that the presence of four or more risk factors, resulted in a 10-fold increase in psychological distress. This wa s validated by research ers investigating the enduring impact of adverse exposures across the lifespan in childre n resulting in both mental and physical distress whereby persons wi th greater than or equal to 4 exposures were at greater risk of poor outcomes (Anda, Felitti, Bremmer, Walker, Whitfield, Perry, Dube, & Giles, 2006).
72 The decision to establish another cut o ff score at seven or more was based upon a previous study conducted by Dube and colle agues (2001) where these researchers found that having a score of at leas t seven adverse exposures incr eased suicide attempts by over 50 fold in adolescents and 30 fold in a dults. Kessler and colleagues (1997) included twenty-six adversities that encompassed lo ss events, parental factors, and personal adversities. A multivariate analyses reveal ed additive effects on the probability of disorder, however, it was noted in their study th at after the sixth adversity was added into the equation the researchers were unable to detect an effect. Design Strengths and Limitations A primary strength of the study is that th is is a cohort sample, which by design provides stronger evidence for exposure-diseas e associations than in other sampling strategies as it permits direct determination of risk (Friis & Sellers, 1999). An additional strength of the study is that interviews were conducted face to face with trained interviewers. In regards to study sample limitati ons, the most common limitation of longitudinal cohort studies is loss of data due to attrition. However, there were high rates of retention for the entire length of the study as noted previously, t hus creating a minimal limitation for this sample. The data for this st udy were originally colle cted to monitor the risk and protective processes across developmental, familial, and environmental domains on the general well being or resiliency of a cohort of adolescent girls growing up in families receiving assistance. Although this study proposed to look at similar risk and protective processes, the outcomes under inves tigation were depression and suicidality.
73 Another constraint within this data set is that there were no specific instruments incorporated in the protocols screeni ng for suicidal though ts or behavior. Another important issue in secondary da ta analysis is access to documentation and coding. A strength of this study is that though I was not part of the original conceptualization of the study, I have worked closely with the quantitative data and have permission from the principal investigator Dr. Roger Boothroyd to use the data. Another strong point is that I have access to the research team to address questions that may arise regarding the database and th e original codebook and transcripts. The limitations or restrictive weaknesses in the study design are as follows: 1. The small sample size (n=125) is a limita tion, restricting the number of variables that can be utilized in analytic models. 2. Assessment of depression using the CESD scale (Radloff, 1977) is limited to measurement of depressive symptomatology, not for meeting crit eria for clinical depression. 3. The development of the composite variables for the AAE score is based on similar composites developed by Felitti and colleagues (1998 utilizing select questions from various scales not orig inally intended for these purposes. 4. Data are based on retrospective recall a nd may result in biases toward the null, given the evidence for longitudinal studi es that have demonstrated that retrospective self-reports of adverse exposures are lik ely to underestimate actual occurrence (Della et al., 1990). 5. There may be mediators or moderators of the relationship between adverse exposures and depression othe r than the factors examined.
74 6. Age variations in the adol escent girls (13-17) in the first year of the study may have contributed to differences in percep tions and reports of adverse experiences. Conclusion Although growing up in poverty is associated with increased likelihood of teenage pregnancy, academic failure, substance abuse, an d suicidality (Martin, Andersen, Lynch & Kupper, 1999), this presents an incomplete picture of all the mechanisms at play when considering the accompanying stressors that ma y not befall an adolescent with a more advantaged background (Musick, 1993). The in dependent relationshi p of welfare and depression has been established in crosssectional studies (D ooley & Prause, 2002). There is also literature de monstrating a strong, graded relationship between adverse exposures and depression as well as suicide attempts in adult women. However, little attention has b een devoted to the role of a dverse exposures as antecedents for depression or as a factor that may infl uence suicide ideation over time in a welldefined adolescent population. In summary, this study adds to the l iterature by broadening the spectrum of adverse exposures often studied for adolescen ts. The indicators in cluded were: personal victimization, living with a household member that had been incarcerated, mentally ill, victimized, or using substances, knowing or witnessing the victimi zation of another, reporting fear of going out in the neighborhood, and the presence of weapons or gunfire. The paucity of literature addressing the cumulative effect of multiple stressors and multiple agents influencing adolescents tran sitioning into adulthood in the context of poverty highlights the need for fu rther investigation to assess the potential ripple effect of such exposures on depressi on and suicide ideation.
75 CHAPTER FOUR RESULTS The results are reported by level of anal yses. Level one provides the descriptive statistics of both the predictor and outcome variables. Level two pr ovides the results of analyses exploring associati ons and significance levels for the predictor and the outcome variables. The final section, level three, pres ents the results of the logistic and multiple regression analyses used to examine each hypothesis. Level One Analyses Frequency distributions have been used to assess the variability, the degree of missing data, and the prevalence for each of the variables. Where indicated response options have been recoded based on distributi ons or for specific an alytic procedures. Control Variables Demographic and behavioral variables used to control for confounding effects were: (1) daughtersÂ’ age, (2) da ughtersÂ’ race/ethnicit y, (3) educational status, and (4) teen pregnancy. The mean age at the start of the study was 16 (range: 14 to 18 years of age, SD=1.055). The racial distribution of the adolescents was 32% white, 41.6% black, 23.2% Hispanic, and 3.2% other. In regards to educational status, by the end of the entire study 27.2% of the daughters had dropped out of school, 16% were still enrolled in high school, 11.2% had a high school diploma or GED, and 45.6% had some post secondary
76 education (e.g. college, technical, or busin ess school). By the end of the study, 48.8% (61) of the girls had experi enced at least one pregnancy. Predictor Variables Definitions and prevalence of each of the 33 items used to construct the predictor composite variables across four points in time are presented in Table 10. As stated above these percentages are based upon the report of one adverse expos ure per item across twelve dimensions during the four years of the study. The results at each step of the development for the predictor variables follo w the same procedure. In step one, the predictor items reported across four points in time ranged fr om 0 = No Exposure to 4 = Exposure Each Year. In step two, each continuous adverse exposure was recoded into a dichotomized variable: 0 = No Exposure to 1= Exposed. Th e prevalence rates presented are based on the report of at least one exposure for any item across the four years. Prevalence estimates for individual items across the four years ranged from 1.6% (2) adolescents being shot to 63.2% (79) of the adolescents know ing kids who owned a gun. The next highest rate of exposure was in residential mobility where reports of having moved more than one time 61.6% (77) during the four years ranged from 0 to 13 moves. Within the category of victimization, 49.6% (62) re ported abuse with 75.2% (94) experiencing direct victimiza tion of being bullied, beaten up, robbed, stabbed, shot at, or shot. Within the category of household dys function, overall exposur e was reported at 89.6% (112). Over 61% reported hi gh levels of residential mobility, followed by familial substance abuse (52.8%), and maternal abuse (43.2%). Within the category of community
77 violence 92.8% reported exposures. The hi ghest category of exposure was knowledge of weapons among peers; 72.8% (91). Witnessing the victimization of another during their lifetime followed this dimension. The highe st rate reported within witnessing the victimization of another were adolescents having seen someone being robbed (42.4%). Both categories of feeling af raid and hearing gunshots in the neighborhood were reported at the same rate across the four years (36%). Overall, across the three categories of adverse exposures, adolescents reported th at community exposures were the highest 92.8% or (116), followed by household dysf unction 89.6 % (112), and lastly, personal victimization 80% (100). The final step in the development of th e AAE score and the results are presented below. Specifically, the level of positive re sponses for 12 dimensi ons within the three categories of victimization, household dys function, and community violence were summed providing exposures ranging from: 0 to 12. The results of adolescents reporting exposures, presen ted in Table 11, found that no adolescent reported exposure to all 12 of the categories. Two of the adolescents reported zero exposures and two adolescents reported the highest score of 11 exposures. A total of 5.6% or (7) ad olescents reported experiencing more than ten adverse exposures. The mean number of advers e adolescent exposures was 5.86 events (SD=2.493). There was a bi-modal distribution of exposures for those receiving a score of 5 and 6 exposures.
78 Table 10 Definition and Prevalence of each Item by Category for Adverse Exposures Category of Adolescent Adverse Exposures (12) Percent N=125 Victimization: (2) 80.0% (100) 1) Abuse/Maltreatment 49.6 % (62) Have you been verbally, physically, sexually or psychologically abused? 44.8 % (56) Have you ever been sexually assaulted? 30.0% (38) In the past year have you experienced being sexually assaulted? 11.2% (14) 2) Have you ever beenÂ… (in the past year have you been) 75.2% (94) a. bullied/pushed around b. beaten up d. robbed e. stabbed f. shot at g. shot 54.9% 26.4% 35.2% 6.4% 9.6% 1.6% (68) (33) (44) (8) (12) (2) Household Dysfunction: (6) 89.6% (112) 1) Family Substance Abuse (Mother) 52.8% (66) Have you ever had a drinking or other drug problem? 23.2% (29) Have any of your family members ever had a drinking/drug problem? 48.0% (60) 2) Maternal Mental Illness 36.0% (45) Have you experienced parent having emotional/psychiatric problems? 34.4% (43) In past month how often have you felt like hurting or killing yourself? 3.2% (4) 3) Maternal Abuse: Any physical, sexual or psychological abuse 43.2% (54) 4) Parental Criminality: Any parental incarceration during study. 28.8% (36) 5) Parental Separation/Divorce: In the past year. 25.6% (32) 6) Residential Mobility: More than one move during entire study. 61.6% (77) Community Violence : (4) 92.8% (116) 1) Have you ever seen someone beingÂ… 68.8% (86) a. sexually assaulted b. robbed c. stabbed d. shot at e. shot f. killed 20.0% 42.4% 27.2% 40.8% 31.2% 20.8% (25) (53) (34) (51) (39) (26)
79 Category of Adolescent Adverse Exposures (12) Percent N=125 2) Do you feel afraid Â… 36.0% (45) a. with adults b. outside in neighborhood c. at school 4.0% 15.2% 20.8% (5) (19) (26) 3) Presence of Weapons: Do you know kids whoÂ…. 72.8% (91) a. own a gun b. bring a knife to school c. bring a gun to school d. bring other weapons to school 63.2% 37.2% 10.4% 16.0% (79) (47) (13) (20) 4) Do you hear gunshots in your neighborhood? 36.0% (45) Table 11 Number of Adverse Adolescent Experiences (AAE Score) (AAE SCORE) Frequency (N=125) Percent 0 2 1.6 1 3 2.4 2 9 7.2 3 10 8.0 4 11 8.8 5 19 15.2 6 19 15.2 7 17 13.6 8 14 11.2 9 14 11.2 10 5 4.0 11 2 1.6 Outcome Variables The dependent variables were assesse d across all four points of time. For adolescents ranging in age from 12 to 15 years of age were a cut-off score for depression was computed at 22 based on the literature (Garrison et al., 1991). Those adolescents ranging in age from 16 to 18 received a cut-o ff score of 24. In the last two years of the
80 study those ranging from 18 years old and above were scored at the cut-off score for Â“clinical casenessÂ” for adults as done in previous studies with a score of 16 or above (Kalil et al., 2001). The overall prevalence rates for the outco me variables are presented in Table 12. This includes depression scores above the cr iterion cut-off as computed using the above ages and scores at each point in time, the pr evalence of suicide ideation at each point in time, and a combined variable constructed fo r overall mental distress due to small cell sizes. Those cases meeting both criteria for CE SD depression score and reporting suicide ideation were only given a score of Â“1Â” for ove rall mental distress, the remainder of the cases were independent of each other. The ove rall distribution of suicide ideation scores remained remarkably consistent for the f our-year period. Where there was a doubling in the incidence of depression by the fourth year and an almost ten percent increase in mental distress by the fourth year. Table 12 Prevalence of Outcome Variables across Four Years Year 1 Year 2 Year 3 Year 4 % (n) % (n) % (n) % (n) Depression Above Criterion* 14.4% (18) 20.0% (25) 20.0% (25) 29.6%(37) Suicide Ideation 5.6% (7) 5.6% (7) 4.8 % (6) 5.6% (7) Any Mental Distress 22.6% (17) 24.0% (30) 22.4 %(28) 31.2%(39) Note. Cut-off Score adjusted for adolescentsÂ’ age.
81 Level Two Analyses Bivariate techniques demonstrating the associations found between individual items comprising the composite variables developed for each of the categories and the AAE score, socio-demographics and outcomes follow in this secti on. Correlations among the individual items used to develop the three composite predictor variables as well as the dichotimized outcome variables were examined and are presented in Table 13. Relationship Among Predictor Variables Amongst the individual item predictors th e strongest correlations were found between childhood exposure to physical, psycho logical, or sexual abuse and exposure to other forms of personal victimization ( r = .347), parental mental illness ( r = .423), maternal experience of abuse (r = .233), witnessing victimization ( r =. 323), and knowledge of weapons ( r = .319). All of these items we re positively correlated at a significance level at less than the p < .01. Moderate correlations at less than p < .05 were found between the adolescentÂ’s experience of abuse and havi ng an incarcerated parent ( r = .182), parental divorce or separation ( r =. 188), and both the primary outco me variables of depression ( r = .204) and suicide ideation ( r = .152). Exposure to personal vi ctimization was significantly associated at p < .01 level with parental mental illness ( r = .272), maternal experience of abuse ( r = .239), knowledge of weapons ( r = .232, and the outcome of depression ( r = .246). The relationship between pe rsonal victimization and witnessing community violence were si gnificant at the level of p <. 05 ( r =. 173). Family substance abuse is significantly correlated at p <.01 levels with maternal experience of abuse ( r = .275), having a parent incarcerated ( r = .354), and residential
82 moves ( r =. 308). Items significantly co rrelated with parental me ntal illness below the .01 level are maternal abuse (r =. 288), incarcerated parent ( r = .259), and knowledge of weapons ( r = .309). Witnessing violence ( r = .181) and the outcome of suicide ideation ( r = .193) were significant at less than .05 levels. Maternal exposure to abuse is correlated at the p < .01 level with knowledge of weapons ( r =. 243), and the outcome of suicide ideation ( r = .246). Other correlates positively associ ated with family substance abuse at p < .05 are incarcerated parent ( r = .194) and witnessing community violence ( r = .169). Additional significant relations hips with an incarcerated parent included parental separation or divorce ( r = .153, p <. 05). At the ( p < .01) level residential moves ( r = .248), witnessing violence ( r = .276), and the outcome variable of depression we re found to have a ne gative correlation ( r = .212). Parental separation and divorce also was negatively correlated with depression ( r = -. 224). This finding suggests that these adolescents may have experienced some protective mechanisms with the se paration of a family member. As would be expected witnessing viol ence and knowledge of weapon carrying in the community were significantly related at ( p < .01) ( r =. 325). Being afraid was associated with the outcome variables of suicide ideation ( r = .240, p < .01) and depression ( r = .159, p < .05). Knowledge of weapons demo nstrated an association with suicide ideation at ( r = .159 p < .05). Relationship Among Socio-Demographics and Predictor Variables The socio-demographic characteristics by the number of adverse exposures reported are presented in Table 14. While not statistically significan t, those reporting the majority of adverse exposures tended to be 16 years or old at the time of the first
83 interview. Yet, 50% of those 15 and 18 year s old at the time of the first interview reported experiencing greater than or equal to seven or more adverse exposures. The race variable was significant for differe nces in experience of exposures x2 = (6, N=125) = 14.129, p =. 028. When recoded into a dichotomi zed categorical vari able of white/nonwhite, non-whites were significantly more likel y to have an experience of any adverse exposure x2 (2, N=125) = 7.052, p =. 03. However, those with a score of seven or more adverse exposures were more lik ely to be white adolescents. Neither educational status nor pregnancie s were significant in regards to adverse exposures. However, interestingly, those who have experienced over seven adverse exposures had dropped out of high school at some point du ring the study or to the contrary had been enrolled in post high sc hool education. This finding is in alignment with two theories of risk. Specifically, that greater adversity is either indicative of poor outcomes (Kazdin et al., 1997) or that stressors actually lead to an enhanced competence (Luthar & Zigler, 1991) through a compensato ry effect, which increases striving. Relationship Among Socio-Demographics and Outcome Variables The prevalence of outcomes for suic ide ideation and depression by sociodemographic characteristics are presented in Table 15. Within the bivariate analysis of associations between socio-demographic va riables and outcome variables, only age ( r = .180, p .04) and pregnancy ( r =. 231, p = .01) demonstrated a pos itive association that was significant with depression. Specifically, for 14 year olds 3 out of 10 reported suicide ideation, while of the 18 year olds 7 out of 10 scored above criterion score for depression. This finding is consistent with the literature in that female adolescen tÂ’s depressed feelings begin to increase from about age 13 upward unt il they stabilize; the explanations remain
84 unclear, but vary from hormonal influence, eco nomic insecurity, to greater sensitivity to peer disruption as during fre quent moves (Ge et al., 1994). Adolescents reporting ever being pregna nt were 2.6 times more likely to experience an elevated depression score (95% CI 1.3 Â– 5. 3) ( p = .01). As expected, the remainder of the socio-demographic variable s were not significantly associated with either of the outcome variables. There was a notable visual difference in the proportion of Hispanic youth reporting suicide ideation versus whites and blacks, however not significantly different. Nevert heless, a separate analysis was run with only those reporting Spanish origin with suicide id eation to assess difference; no significant relationship was found x2 = (1, N=125) = .840, p =. 359. A possible explanation for the increased suicide ideation found with time on TANF, may be found in the stressvulnerability model proposed by Rich and Bonner in 1987 (as cited in the GAP Report No. 140, 1996). Specifically, if one assumes that there is a combination of factors that may contribute to the stress and vulnerability of an adolescent female, such as limited resources and high levels of residential mobility, combined with exposure to victimization, then a possibility of increased risk for suic ide ideation seems plausible. The findings of a decrease in depression with time on TANF are in line with previous findings of studies of female adolescentÂ’s perceptions of parent al economic stress and depression (Frojd et al., 2006). An adoles centsÂ’ positive adjustment to the economic circumstances of their mothers receiving publ ic assistance may be a function of their
85 increasing age and ability to develop persona l resources and behavi oral options (Yagub, 2002). Odds Ratios of Individual Items and Outcome Variables The odds of individual items predicting su icide ideation (Table 16) or depression (Table 17) confirm the Pears on correlations found. The predic tors increasing the odds of reporting suicide ideation were a buse OR 3.4 (95% CI 1.2 Â– 10.1) ( p = .023), parental mental illness OR 2.9 (95% CI 1.1 -7.9) ( p = .031), maternal abuse OR 3.4 (95% CI 1.2 Â– 9.8) ( p = .020), and fear of the outsi de OR 3.7 (95% CI 1.4 Â– 10.5) ( p = .007). Knowledge of weapons in the hands of peers approached significance with the odds of developing suicide ideation at 3.7 (95% CI 0.8 16.9) ( p = .076). The odds of adolescents having been victimized and scoring above the cut off score for depression are OR 3.5 (95% CI 1.4 Â– 8.4) ( p = .005).
86 Table 13 Correlation Matrix of In dividual Item Predictor Variables and Outcomes Note. **Correlation is significant at the 0.01 level. *Correlation is significant at the 0.05 level. 1 2 3 4 5 6 7 8 9 10 11 12 1. Abuse ___ 2. Victim .347** ___ 3. Family Substance Abuse -.024 -.061 ___ 4. Parental Mental Illness .423** .276** .008 ___ 5. Maternal Abuse .233** .239** .275** .288** ___ 6. Parental Incarceration .182* .079 .354** .259** .194* ___ 7. Parental Sep/Divorce .188* .125 -.143 .095 .118 .153* ___ 8. Residential Moves -.006 .080 .308** .112 .058 .248** -.065 ___ 9. Witnessed Violence .323** .173* .124 .181* .169* .276** .078 .143 ___ 10. Afraid Outside .056 .083 .008 .028 .086 -.109 -.134 -.025 .001 ___ 11. Knowledge of Weapons .319** .232** -.002 .309** .243** .111 .111 .035 .326** -.028 ___ 12. Heard Gun Shots -.044 .122 .075 .028 -.048 -.072 .057 .078 .145 .097 .009 ___ Any Suicide Ideation .204* .140 -.046 .193* .216** -.122 -.095 -.124 .093 .240** .159* -.132 Any Depression .152* .246** -.037 .025 .114 -.212** -.224** .120 .118 .159* .074 .092
87 Table 14 Socio-Demographic Characteristics by Outcomes Number of Adverse Adolescent Exposures (N=125) 0 Â– 4 AAE 5 Â– 6 AAE 7 Â– 11 AAE Total in Class Age at first Interview 14 40.0% 10.0% 50.0% 8.0% 15 20.8% 29.2% 50.0% 19.2% 16 30.4% 26.1% 43.5% 36.8% 17 28.6% 42.9% 28.6% 28.0% 18 20.0% 30.0% 50.0% 8.0% Race White 12.5% 37.5% 50.0% 32.0% Black 40.4% 26.9% 32.7% 41.6% Hispanic 31.0% 20.7% 48.3% 23.2% Other 0.0% 75.0% 25.0% 3.2% Education Dropped Out 20.0% 32.4% 47.1% 27.2% In high school 30.0% 30.0% 40.0% 16.0% High School or GED 50.0% 35.7% 14.3% 11.2% Post High School 26.3% 28.1% 45.6% 45.6% Ever Pregnant 26.2% 32.8% 41.0% 48.8% Length of Time on TANF: Less than 6 Months 14.3% 57.1% 28.6% 23.3% Six Months to 1 Year 0.0% 50.0% 50.0% 20.0% Over 1 Year 17.6% 29.4% 52.9% 56.7% All Participants 28.0% 30.0% 42.0% 100.0% Note. Rows within factors sum to 100.
88 Table 15 Socio-Demographic Characteristics by Outcomes Characteristic Any Suicide Ideation Any Depression No Yes No Yes Age at first Interview 14 70.0% 30.0% 90.0% 10.0% 15 95.8% 4.2 % 50.0% 50.0% 16 82.6% 17.4% 52.2% 47.8% 17 85.7% 14.3% 51.4% 48.6% 18 80.0% 20.0 % 30.0% 70.0% Race White 85.0% 15.0% 55.0% 45.0% Black 86.5% 13.5% 51.9% 48.1% Hispanic 79.3% 20.7% 51.7% 48.3% Other 100.0% 0.0% 50.0% 50.0% Education Dropped Out 76.5% 23.5% 44.1% 55.9% In High School 95.0% 5.0% 65.0% 35.0% High School or GED 85.7% 14.3% 50.0% 50.0% Post High School 86.0% 14.0% 54.4 % 45.6% Ever Pregnant 85.2% 14.8% 41.0% 59.0% Time on TANF Less Than 6 Months 100.0% 0.0% 42.9% 57.1% Six Months to 1 Year 83.3% 6.7% 50.0% 50.0% Over 1 Year 76.5% 23.5% 64.7% 35.3% Note. Rows within factors sum to 100%
89 Table 16 Individual Factors Influen ce on Any Suicide Ideation AAE Category % Any Suicide Ideation During Study P value Odds Ratio 95% Confidence Interval 1. Abuse 22.6 3.4 (1.2 -10.0) .023* 2. Victim 18.1 3.2 (.70 14.7) .118 3. Family Substance Abus e 13.6 .77 (.29 2.1) .607 4. Parental Mental Illness 24.4 2.9 (1.1 -7.9) .031* 5. Maternal Abuse 24.1 3.4 (1.2 9.8) .020* 6. Parental Incarceration 8.3 .42 (.12.52) .185 7. Parental Sep/Divorce 9.4 .50 (.14 1.8) .295 8. Residential Moves 11.7 .50 (.19 1.4) .171 9. Witnessed Violence 17.4 1.9 (.57 5.9) .300 10. Afraid Outside 26.7 3.8 (1.4 -10.5) .007** 11. Knowledge of Weapon s 18.7 3.7 (.80 16.9) .076 12. Heard Gun Shots 8.9 .42 (.13 1.4) .140 Note. **Correlation is significant at the 0.01 level (One-tailed). *Correlation is significant at the 0.05 level (One-tailed).
90 Table 17 Individual Factors Influ ence on Any Depression AAE Category % Any Depression During Study P value Odds Ratio 95% Confidence Interval 1. Abuse 54.8 1.9 (.91 3.8) .091 2. Victim 54.3 3.4 (1.4 8.4) .005** 3. Family Substance Abuse 45.5 .86 (.43 1.7) .679 4. Parental Mental Illness 48.9 1.1 (.54 2.3) .779 5. Maternal Abuse 53.7 1.6 (.78 3.2) .205 6. Parental Incarceration 30.6 .38 (.17 .86) .016* 7. Parental Sep/Divorce 28.1 .34 (.14 .80) .011* 8. Residential Moves 51.9 1.7 (.79 3.4) .179 9. Witnessed Violence 51.2 1.8 (.78 3.6) .189 10. Afraid Outside 57.8 2.0 (.93 4.1) .077 11. Knowledge of Weapons 49.5 1.4 (.63 3.1) .411 12. Heard Gun Shots 53.3 1.5 (.71 3.1) .304 Note. **Correlation is significant at the 0.01 level (One-tailed). *Correlation is significant at the 0.05 level (One-tailed).
91 Conversely, there was a negative associ ation with depression for those adolescents not reporting havi ng a parent incarcerated OR .38 (95% CI 0.2 Â– 0.9) ( p = .016), nor experiencing a parental separati on or divorce OR .34 (95% CI .14 .80) ( p = .011). Relationships of Socio-Demographics and Cu mulative Score with Outcome Variables The bivarite relationships between the control variables, the predictor categories and the cumulative AAE score with each out come variable are presented in Table 18. The CES-D scores at each year, as well as the outcome variable s of suicide ideation, mental distress, and change scores permit th e exploration of the impact of exposures across time. Age recoded as (0 = below 16, 1 = 16 or over) is significantly associated with change in suicide ideation over time ( r =. 155, p < .05) and change in overall mental distress ( r = .217, p < .01). The race variable was not si gnificant with a ny of the outcome variables at this level of analysis. Educati on recoded as (0 = enro lled or graduated, 1 = dropped out of high school) is significan tly associated with suicide ideation ( r = .182, p < .05) and any mental distress ( r = .151, p < .05) during the study. Having experienced any pregnancy was significantly associat ed with CES-D scores in year 4 ( r =. 156, p < .05), change in CES-D score ( r = .152, p < .05), and any mental distress during the study ( r = .161, p < .05). Further bivarate analyses were conduc ted to examine both the cumulative AAE score, and the sum of scores from the cat egories of victimization, household dysfunction, and community violence. Victimization was positively associated with CES-D scores. In Year 1 ( r = .156, p < .08), Year 2 ( r = .312, p < .001), Year 3 ( r = .358, p < .001), and Year 4 ( r = .233, p = .009), any suicide id eation across study ( r = .215, p =
92 .016), and any mental distress ( r = .273, p = .002). While the other correlations were not significant in this category, the results across the four years demonstrate an increase in strength of association for victimization and depression. Household dysfunction was not associated as a category with any of the outcome variables. Community violence exposures, however, demonstrated a strong significant relationship for all four years of depression scores in Year 1 ( r = .238, p = .008), Year 2 ( r = .305, p =. 001), Year 3 ( r = .244, p = .006), Year 4 ( r = .282, p =. 001), and overall mental distress ( r = .225, p = .012). There was no significant relationship with suicide ideation or change in outcomes. The total AAE score was significantly asso ciated with CES-D scores at Year 2 ( r = .241, p = .007), Year 3 ( r = .247, p = .005), Year 4 ( r = .183, p = .041). However, there was no significant relationship found between th e total AAE score and mental distress or suicide ideation or with a ny of the change scores.
93 Table 18 Bivariate Associations of AAE Scor es with Outcomes across Time Note. **Correlation is significant at the 0.01 level (One-tailed). *Correlation is significant at the 0.05 level (One-tailed). CESD Year 1 CESD Year 2 CESD Year 3 CESD Year 4 Change in CESD Any Suicide Ideation Change in Suicide Any Mental Distress Change Mental Distress Control Age .019 -.102 -.048 -.056 -.079 .007 .155* .134 .217** Race .103 .074 .100 .103 .009 .004 .102 .024 .030 Education .101 .123 .002 .133 .042 .182* -.059 151* -.005 Any Pregnancy .016 .123 .147 .156* .152* -.012 -.102 .161* .125 Predictors Victimization Household Community .156* .005 .238** .312** .019 .305** .358** .013 .244** .233** -.074 .282** .095 .095 .108 .215* .009 .140 .092 -.010 .056 .273** -.035 .225* .048 -.091 -.030 Total AAE SCORE .166 .241** .247** .183* .017 .166 .032 .162 -.067
94 Level Three Analyses Contribution of Each Predictor on Outcomes Separate, individual regression models were examined to test the contribution of each category of adverse exposure on the out comes of depression, suicide ideation, and any mental distress at any point during the study. Odds ratios were calculated using separate logistic models both controlling fo r and not controlling for socio-demographics. When all 12 adverse exposures were entere d into the model, 29.8% of the variance in experiencing any mental distress during the co urse of the study was explained. In seeking the best-fit model for the data, a forw ard maximum likelihood ratio model was determined. The final forward logistic regr ession model presented in Table 19, explained 24% of the variance by reta ining four variables: abuse OR 2.5 (CI, 1.07 6.01) ( p = .036); parental incarceration, OR .28 (CI .113 -.71) ( p =. 007); parental separation and divorce, OR .30 (CI .12 .78) ( p = .013); and personal victim ization, OR 3.4. (1.3 9.2) ( p = .014). The first three hypotheses called for an exam ination of the relative contribution of the composite categories of exposures to victimization, household dysfunction, and community violence to depression and suicid e ideation. For each hypot hesis, a series of logistic regression models were analyzed at first without controlling for sociodemographics, the results are presented in Table 20. Specificall y, those experiencing personal victimization were over three times more likely to experience mental distress than those adolescents not w ho had not been victimized.
95 However, while the results infer that personal victimization is the critical exposure in predicting overall mental distress in these adolescents, it is important to note that across the three categorie s of adverse exposures there is very little variance suggesting that community exposures we re virtually a constant and household dysfunction exposures were almost at 90%. Sp ecifically, these adoles cents reported that community exposures were the highes t 92.8% or (116), followed by household dysfunction 89.6 % (112), and lastly, persona l victimization 80% (100). The lack of variance in familial and contextual exposures is the most plausible explanation for the lack of evidence supporting the respective hypotheses. Those experiencing exposure to victim ization were 2.35 times more likely to report suicide ideation, while those experiencing community violence exposures were only 1.469 times more likely to report suic ide ideation. Household dysfunction did not play a significant role in the report of suic ide ideation. Only the cat egory of victimization was significant in predicting depression OR 4.7 (CI .63 Â–13.5) ( p = .004). Those exposed to household dysfunction were over two times more likely to deve lop depression while those exposed to the category of community violence were 1.87 times more likely to develop depression. Those exposed to direct victimization were over four times more likely to experience mental distress OR 4.2 (CI 95% 1.55 Â– 11.41) ( p = .005). The results of the full logistic regressi on models for suicide ideation (Table 21) and any depression (Table 22) by each cumula tive category of adverse exposure indicate only slight attenuations in odds ratios.
96 Table 19 Summary of Regression Analysis with Ad verse Exposures on Any Mental Distress Nagelkerke R2 B SE P value Odds Ratio 95% Confidence Interval Step 1 .080 Victim 1.194 .447 .008 3.3 (1.4 -7.9) Step 2 .153 Incarcerated Parent -1.154 .432 .007 0.3 (.14 .74) Victim 1.353 .464 .004 3.9 (1.6 9.6) Step 3 .200 Incarcerated Parent -1.071 .442 .015 0.3 (.14 .82) Parental Sep/Divorce -1.013 .460 .028 0.4 (.15 .89) Victim 1.516 .479 .002 4.6 (1.8 11) Step 4 .241 Abuse .928 .442 .036 2.5 (1.1 6.0) Incarcerated Parent -1.266 .468 .007 0.3 (.11 -.71) Parental Sep/Divorce -1.209 .487 .013 0.3 (.12 .78) Victim -1.236 .502 .014 3.4 (1.3 9.2)
97 Table 20 Influence of Adverse Exposure Categories on Outcomes AAE Category Any Suicide Ideation During Study (n= 19) P-value Percent (n) Odds Ratio 95% Confidence Interval Victimization 89.5% (17) 2.4 (0.5 10.9) .224 Household Dysfunction 89.5% (17) 1.0 (0.2 4.8) .984 Community Violence 94.7% (18) 1.5 (0.3 18.1) .723 Any Depression During Study (n = 59) Victimization 91.5% (54) 4.7 (0.6 13.5) .004* Household Dysfunction 93.2 % (55) 2.2 (0.6 7.5) .218 Community Violence 94.9% (56) 1.9 (0.5 7.8) .387 Any Mental Distress Du ring Study (n = 63) Victimization 90.5 % (57) 4.2 (1.6 Â– 11.4) .005* Household Dysfunction 92.1% (58) 1.7 (0.5 -5.6) .367 Community Violence 93.7 % (59) 1.3 (0.3 -5.1) .717 Note. *Correlation is signifi cant at the 0.05 level. After adjusting for age, race, education, a nd pregnancy for each of the categorical exposures only those experiencing victimiza tion were over two times more likely to report suicide ideation OR 2.3 (CI 95% .45 Â– 11.5) ( p = .371). The findings demonstrate that being pregnant is protective in regards to suicide ideation, but not for depression as seen in Table 22. Adolescents who had dropped out of school had greater odds of reporting suicide ideation OR 2.6 (CI 95% .86 Â– 7.9) ( p = .092). Adolescents who were 16 years of age or
98 older at the start of the study were more li kely to report suicid e ideation than their younger peers OR 1.7 (CI 95% .49 Â– 6.0) ( p = .398). Significant odds ratios for depression were found for those experienci ng a pregnancy OR 2.3 (CI 95% 1.03 5.2) ( p = .043); and for those reporting exposure to vi ctimization, they were almost five times more likely to score above the cut-off scor es for depression OR 4.99 (CI 95% 1.6 15.5) ( p = .005). Contribution of Cumulative Predictors on All Outcomes To test the fourth hypothesis, the odds ratios associated with increasing levels of exposure to adverse experiences on the presen ce of depression, suic ide ideation, and any mental distress were examined. These are summarized in Table 23. The results do not support the presence of a Â‘dose responseÂ’ re lationship as hypothesi zed would exist in relationship to increasing levels of adverse exposures to the risk of developing depression given none of the odds ratios were significant. In terms of suicid e ideation, ag ain there was no evidence found to support a dose response relationship.
99 Table 21 Summary of Logistic Regressions for Suicide Ideation Cont rolling for SocioDemographics B SE Pvalue Odds Ratio 95% Confidence Interval Controls Age .539 .638 .398 1.7 (0.5 -5.9) Race .193 .566 .733 1.2 (0.4 3.7) Education .954 .567 .092 2.6 (0.9 7.9) Any Pregnancy -.476 .568 .402 0.6 (0.2 -1.9) Predictors1 Victimization .821 .827 .321 2.3 (0.5 -11.5) Household Dysfunction -.333 .891 .709 0.7 (.13 4.1) Community Violence .421 1.157 .716 1.5 (0.2 14.7) Note 1All control variables entered into each Logis tic Regression with each predictor category. Table 22 Summary of Logistic Regression Models for Depression Controlling for SocioDemographics B SE P value Odds Ratio 95% Confidence Interval Age .289 .452 .523 1.3 (0.6 3.2) Race .444 .426 .297 1.6 (0.7 3.6) Education .166 .444 .712 1.2 (0.5 2.8) Any Pregnancy .835 .410 .043* 2.3 (1.03 5.2) Predictors1 Victimization 1.608 .557 .005* 4.9 (1.6 15.5) Household Dysfunction .278 .725 .701 1.3 (.32 5.5) Community Violence .293 .844 .728 1.3 (0.3 7.01) Note. 1Control variables entered into each Logistic Regression with predictor categories. Correlations significant at the 0.05 level.
100 However, significant odds ratios were found associated with more than five adverse exposures and more than six advers e exposures to advers e events. Adolescents with more than 5 exposures were 4.94 times (p < .05) more likely to report suicide ideation compared to adolescen ts exposed to 5 or less adverse events. None of the odds ratios associated with the risk of mental distress were significant. Interestingly, though statistically not si gnificant, any exposure greater than one demonstrated over three times OR 3. 74 (CI 95% .41-34.47) the likely hood of experiencing depression. However, as advers e exposures increased in number the likely hood for reporting depressive symptoms did not suggesting that other mechanisms may be at play. This result is c ounterintuitive, that more ri sk may inoculate one towards depression. What is unknown is whether othe r adverse outcomes developed other than depression, such as anxiety or substance abuse. The cumulative AAE scores showed an incr ease in the risk for developing suicide ideation when an adolescent had scored over three exposures. Little to no evidence was found to support the relationship between the AAE score and mental distress. To test hypothesis five, linear regressi on models were examined controlling for socio-demographic variables. These are presen ted in Table 24. Like wise, little evidence was found to support hypothesis five. When cont rolling for socio-demographics the direct effects of the cumulative impact of the AAE score disappeared for change scores across all outcomes. The results do not support the pr esence of a Â‘dose responseÂ’ relationship as hypothesized would exist in rela tionship to increasing levels of adverse exposures to the change in developing depression given none of the odds ratios were significant.
101 Table 23 The Relationship of the AAE Score to Depre ssion, Suicide Ideation, and Mental Distress AAE SCORE N Depression Suicide Idea tion Mental Distress % OR (CI95%) % OR (CI95%) % OR (CI95%) > 0 (123) 50.4 1.02 (.06-16.62) 14.6 .17 (.010 -2.866) 50.4 1.02 (.06 Â–16.62) > 1 (120) 48.3 3.74 (.41-34.47) 14.2 .25 (.038 -1.59) 50.8 1.55 (.25 -9.62) > 2 (111) 50.5 3.73 (.99-14.11) 15.3 1.08 (.22 -5.29) 53.2 2.84 (.84 Â– 9.59) > 3 (101) 51.5 2.58 (.98-6.75) 16.8 2.23 (.48 Â– 10.37) 54.5 2.39 (.94 Â– 6.09) > 4 (90) 52.2 2.10 (.93-4.72) 18.9 3.84 (.84 Â– 17.60) 55.6 2.12 (.95 Â– 4.72) > 5 (71) 50.7 1.39 (.68-2.83) 22.5 4.94 (1.36 Â–17.98)* 54.9 1.53 (.745 Â– 3.10) > 6 (52) 51.9 1.38 (.68-2.83) 23.1 2.83 (1.03 Â– 7.78)* 55.8 1.45 (.71 Â– 2.96) > 7 (35) 51.4 1.26 (.58-2.77) 20.0 1.63 (.582 4.54) 54.3 1.24 (.57 Â– 2.72) > 8 (21) 47.6 1.02 (.40-2.61) 23.8 2.01 (.635 Â– 6.35) 52.4 1.10(.43 Â– 2.81) > 9 (7) 28.6 .43 (.08-2.30) 14.3 .93 (.105 Â– 8.16) 28.6 .374 (.07 Â– 2.00) > 10 (2) 0.0 NA 50.0 5.83 (.349 Â– 97.527) 50.0 .98 (.06 Â– 16.09) Note. *p < 0.05 level.
102 Table 24 Linear Regression Models for To tal AAE SCORE and Change Scores SE Fa R2 P -value Change in Depression During Study 8.904 1.107 .044 .361 Change in Suicide Ideation .386 1.736 .068 .132 Change in Mental Distress .604 1.440 .057 .215 Note. 1Model adjusted for age, race, education, and pregnancy: df=5. Summary of Findings Although the data did not support all of th e hypotheses in this study, there were a number of significant findings. While commun ity exposures to violence were the most reported, exposure to personal victimization wa s the largest contributor to outcomes of depression, suicide ideation, and overall distress. Evidence of a direct association was f ound for those experiencing victimization with suicide ideation, mental di stress, and CES-D depression sc ores at all four points in time. However, victimization did not pred ict change in any of the outcomes. The relationship of the total AAE score was positively correlated with the CES-D scores across time as well as with any mental di stress experienced duri ng the study. Summaries of all of the bivariate asso ciations are presented in Ta ble 25, Table 26, and Table 27.
103 Multivariate analyses exploring the relative contribution of the cumulative impact of exposures found that there was a significan t relationship for the presence of suicide ideation after five exposures. However, ther e was no evidence supporting that having an elevated AAE score increased the risk for developing depression or mental distress. Only school dropout remained a significant predictor for suicid e ideation. In the models for depression and mental distress only age remained after controlling for sociodemographics, any significant contributi on of cumulative expos ures to adversity disappeare. A final summary table of suppor t for all of the study hypotheses follows in Table 28.
104 Table 25 Summary of Bivariate Associations between Socio -demographics, Individual Predictors and Outcomes Predictors Significance Level Interpretation Socio-Demographics Age S P < .01 Adolescents age 16 and over report signifi cantly more days of suicide ideation and depressive symptoms. Race NS While non-whites reported more sympto ms of mental distress, whites were more likely to experience over 7 adverse experiences. Education S P < .05 Dropping out of high school has a positive significant association with suicide ideation and greater mental distress. Pregnancy S P < .05 Experiencing an adolescent pregnancy has a positive significant association with depression and greater mental distress. Adverse Exposures Victimization S P = .005 Experiencing at least one form of vi ctimization has a positive significant association with outcomes of depres sion and greater mental distress. Abuse S P = .02 Experiencing abuse has a positive significant associat ion with household dysfunction, community violence, and out comes of suicide ideation and greater mental distress. Victim S P =. 005 Experiencing victimization has a pos itive significant association with household dysfunction, community violence, and w ith outcomes of depression and greater mental distress.
105 Table 26 Summary of Bivariate Associations Between Familial Predictors and Outcomes Predictors Significance Level Interpretation Household Dysfunction NS The category of household dysfuncti on was not associated with distress. Family Substance S P < .01 Family substance abuse has a positive significant association with other dimensions of household dysfunction. Parental Mental Illness S P < .01 Parental mental illness has a posi tive significant association with other dimensions of household dysfunction, comm unity violence, and suicide ideation. Maternal Abuse S P = .02 Maternal abuse has a positive significant association with other dimensions of household dysfunction, community vi olence, and suicide ideation. Parental Incarceration S P < .01Having a parent incarcerated has a positiv e significant association with other dimensions of household dysfunction, comm unity violence, and has a negative significant association with the outcome of depression. Parental Sep/Divorce S P < .01Parental separation or divorce has a positive significant association with abuse, an incarcerated parent, and has a negative significant association with the outcome of depression. Residential Moves S P < .01Residential moves have a positive signifi cant association with family substance abuse and having a parent incarcerated.
106 Table 27 Summary of Bivariate Associations Be tween Community Predictors and Outcomes Predictors Significance Level Interpretation Community Violence S P < .01 The category of exposure to community violence has a positive significant association with the outc ome of depression at all points in time and overall mental distress. Witnessed Violence S P < .01 Witnessing violence is has a positive association with various forms of household dysfunction and knowledge of weapons in the community. Afraid Outside S P = .007 Being afraid out in the community has a positive signi ficant association with the outcome of depressi on and with suicide ideation. Knowledge of Weapons S P <. 01 Knowledge of weapons amongst peer s has a positive significant association with the outcome victimization, matern al abuse, parental mental illness, witnessing violence. Heard Gun Shots NS Hearing gunshots in th e community was not a ssociated with other exposures or mental distress.
107 Table 28 Summary of Significance for Hypothesis Level Hypothesis Support Category of Exposure 1. Exposure to adolescent victimization will ha ve a positive association with depression, suicide ideation, and mental distress. Yes 2. Household dysfunction will have a positiv e association with depression, suicide ideation, and mental distress. No 3. Exposure to community violence will have a positive association with depression, suicide ideation, and mental distress. Partial Cumulative Exposures 4. The number of adverse exposures will have a cumulative impact or Â“dose responseÂ” relationship with the level of depression, su icide ideation, and overa ll mental distress. No 5. The Adverse Adolescent Exposure Score wi ll be significantly related to change in depression, suicide ideation, a nd mental distress over time. No
108 CHAPTER FIVE DISCUSSION OF FINDINGS This investigation represents a departur e from previous studies examining single forms of risk in relationship to advers e mental health outcomes, to assessing the cumulative impact of community, family, and personal exposures to adversity. The results bring evidence of extreme levels of contextual risk that in one population would create even more dire results, yet in this group unseen mechanisms of resilience and protective factors were present. The re sults imply that caution must be taken in examining generally assumed risk factors, su ch as family characteristics of divorce or incarceration, as in this population these factors we re protective. The important implication is that social and family characte rizes of assumed risk appears to be relative to the population and their contex tual experiences of exposures to adversity and violence. The purpose of this study was to determine the cumulative impact of adverse experiences on depression and suicide ideation over time in female adolescents. In addition, this study expanded upon the ACE st udy model by capturing a wider range of adverse exposures from the community. The contextual variables included exposures ranging from witnessing violence, being af raid in the community, knowledge of the presence of weapons, to hearing gunshots. In as much as this study attempted to replicate the analytic strategies of the ACE studies, th e results for this study did not demonstrate a
109 cumulative impact, even though the exposure rate s for adversity were three times national rates. Prevalences Within this cohort of adolescent female s raised on public assistance, 92.8% had been exposed to community violence. This leve l of contextual adversity was greater than those reporting being raised in familie s with household dysfunction (89.6 %), and personal victimization (80%). Yet, while the overall category of co mmunity exposures to violence were the most highly reported, result s from exposure to personal victimization made the greatest contribution to outcom es of depression an d suicide ideation. Consistent with other studi es, the rates of exposure in this population were widespread, however, slightly higher than those reported in st udies of urban youth exposures in the general popul ation which ranged from 85% for community exposures to 70% for interpersonal victimi zation (Duckworth et al., 2000). One of the most disturbing findings was that over 72% of the adolescents had experienced five or more adverse exposures and 89% experienced over three adve rse events, almost triple the exposures reported by Kessler and colleagues (1997) in the National Co -Morbidity Study. Prevalence estimates for individual items across the four years ranged from 1.6% of adolescents being shot, to 63.2% of th e adolescents knowing kids who owned a gun. The next highest rate of exposure was in residential mobility where reports of having moved more than one time 61.6% during the f our years ranged from zero to 13 moves. The highest category of exposure 72.8% fo r an individual item was knowledge of weapons among peers. Witnessing the victim ization of another during their lifetime followed this dimension. The highest rate repor ted within witnessing the victimization of
110 another were adolescents having seen someone being robbed (42.4%). Both categories of feeling afraid and hearing gunshots in the ne ighborhood were reported at the same rate across the four years (36%). The distribution of suicide ideation scores remained remarkably consistent for the 4-year period ranging from 4.8% to 6%. Th ere was a doubling in the incidence of depression (14% to 29.6%) by the fourth year and an almost ten percent increase in mental distress (22.6% to 31.2%) was found by the fourth year. These last findings suggests that time and age are important in studying the outcomes of depression or suicide ideation, in that there may be a decayi ng effect in regards to suicide ideation, and a persistence in the effect of adversity on depression. Relationship Between Adversity and Outcomes As anticipated, and interrelatedness of adverse exposures within this sample validates previous findings in the literature of the common co-occurrence of such events in the lives of adolescents and that such a dverse experiences are not isolated events. In general individual adverse exposures studied were consistent with a number of recent studies suggesting that childhood exposures typically overlap and co-occur (Bergen et al., 2003; Finkelhor & Hashima, 2001; Howard et al., 2002; Johnson et al., 2002; Jong et al., 2000; Kendall-Tackett et al., 1993; Kessl er et al.,1997; McMahon et al., 2003; Spat Widom, 1999; Stevens et al., 2003; Turn er & Lloyd, 1999; Turner et al., 2006). The average age at the start of the study for those reporting multiple exposures was 16 years old. By the end of the study, almost half of the girls had experienced at least one pregnancy. While those of non-white origin experienced, in general, more adverse
111 exposures, those experiencing over seven advers e exposures were more likely to be white adolescents. Although not reaching levels of significance, adolescents of Hispanic origin in this population reported slightly higher levels of suicide ideation th an did their peers. There is a conflict in the lit erature concerning those who re port suicide ideation versus the evidence of suicide completers. To date, female adolescent completers tend to be white; however, recent national data suggests th at ideators and attempters tend to be adolescent Hispanic females (Eaton, Ka nn, Crosby, & Flores, 2007). Explanations offered by the CDC research team for the differences are family characteristics, acculturation, and socio-cultural differences. Nevertheless, further exploration of how family characteristics contribute to these differences is needed. By the end of the entire study over 27 % of the daughters had dropped out of school. Just over 27% were still enrolled in or had completed high school, and over 45% had some post secondary education. Interes tingly, those with the highest cumulative exposures, over seven, either reported having dropped out of school or having gone onto to post secondary. These results may be expl ained by the Challenge Model for risk and protective factors found in the risk literature which posits that a curvilinear relationship exits such that stressors can actually lead to an enhanced competence (Luthar & Zigler, 1991). This is in contrast to the additive or cumulative model of risk that proposes that exposure to one risk factor does not necessa rily lead to a poor outcome, but with the presence of four or more risk factors, ther e is a 10-fold increase in psychological distress or disorder (Kazdin etal.,1997; Kirby & Fraser, 1997; Luthar & Zigler, 1991).
112 Not surprisingly, evidence was found for the significant direct association between adverse exposures to victimization a nd being afraid outside to both the outcomes of depression and suicide ideation. Parental mental illness, maternal abuse, and knowledge of weapons amongst peers were signifi cantly associated with suicide ideation. A possible alternative explan ation for the increased su icide ideation found with time on TANF, may be found in the stress-vul nerability model proposed by Rich and Bonner in 1987 (as cited in the GAP Report No. 140, 1996). Specifically, that there is a transaction between social-e motional alienation, cognitive distortions, and inadequate adaptive abilities providing a predispositi on for suicidal thoughts and behaviors. Economic insecurity and peer disruption seems pa rticularly relevant to this sample given all of the participants at the start of the study were receivi ng public assistance and that they also experienced on average higher leve ls of residential mobility; up to 13 moves over the four year study. If one assumes that there is a combinati on of factors that may contribute to the stress and vulnerability of an adolescent fema le, such as limited resources and high levels of residential mobility, combined with exposur e to victimization, then a possibility of increased risk for suicide ideation seems pl ausible. The findings of a decrease in depression with time on TANF are in line with previous findings of studies of female adolescentÂ’s perceptions of parental economic stress and de pression (Frojd et al., 2006). The lack of persistence in depression in relationship to time on TANF, may be both conceptually and theore tically explained by the stress process model proposed by Pearlin (1981); that an adaptation to ch ronic stressors is a mechanism of our physiological need to return to homeostasis. A further interpretation of the results of an
113 adolescentsÂ’ positive adjustment to the economic circumstances of their mothers receiving public assistance ove r a period of time may be a fu nction of their increasing age. Developing personal resources and beha vioral options for changing the course of their own lives through further education or additional suppo rtive relationships may be the most reasonable explanation (Yagub, 2002). In addition to the main results, there we re a number of important findings that were contradictory to the current assumption that divorce and parental incarceration is a risk factor and not protective. Specificall y, depression was negatively associated with having an incarcerated parent a nd parental separation and divo rce. Investigations of the broader concepts of victimiza tion in general, personal victim ization is thought to be a function of the dependency status of both wo men and children (Finklehor et al., 2001). Along these lines, perceptions of whether even ts are depriving or liberating may mitigate stressful experiences as demonstrated in this study where having a pa rent incarcerated or experiencing the divorce of parents was potenti ally protective in the face of victimization. As mentioned earlier, a prev ious review of this data set, and the qualitative responses by the daughters and mothers on the relationship of the perpetrators to the daughter found that of those who had been vi ctimized (47%) reported that the perpetrator was a family member or father figure, incl uding a motherÂ’s boyfri end. These results are similar to national data where half ( 51%) of childhood sexual victimizations are perpetrated by parents as found in the comp rehensive review of national databases conducted by Finkelhor and Ha shima (2001). Once again, though not fully investigated in this study, a large number of adolescents in this data set re ported running away suggesting a need to further understa nd the experiences of these youth.
114 Summary of Hypotheses Overall, the results offer partial supp ort the hypotheses of the study. The most profound results were demonstrated with expos ure to adolescent victimization. There was a significant and positive association between adolescent victimization and depression, suicide ideation, and mental distress which fully supported the first hypothesis. Those reporting exposure to victimiza tion were almost five times more likely to score above the cut-off scores for depression. No support for the second hypothesis was found for the overall category of household dysfunction contributing to any of th e outcome variables. The lack of findings on the cumulative index of household dysfunc tion may be due to lack of variance, however, it is important to note that the indi viduals reporting parental mental illness or maternal abuse were over three times more likely to repor t suicide ideation. Partial support was found for the third hypot hesis, which proposed that exposure to community violence would have a positive association with depression, suicide ideation, and mental distress. Evidence wa s found for the significant association of community violence to CES-D scores at each of the four points in time and to overall mental distress across the study. The fourth hypothesis was not supported which proposed that the number of adverse exposures would have a cumulative im pact or Â“dose respons eÂ” relationship with the level of depression, suicide ideation, a nd overall mental distress. No support was found for the fifth hypothesis. The total AAE score did not make a contribution to the change in depression, suicide ideat ion, and overall mental distress.
115 Limitations of Study There were a number of limitations or re strictive weaknesses in the study, which may have contributed in the lack of suppor t for the proposed hypot heses. Methodological issues included: sample size, measurement, retrospective recall, and design issues. The small sample size constrained the analyses a nd restricted the power with which to draw conclusions. Another weakness, and perhaps th e most profound was the lack of variance in the study due to the high rates of exposures However, the lack of variance in familial and contextual exposures is the most plausi ble explanation for the constrained support of the respective hypotheses. A furt her explanation would be theo retical, as implied by the socio-ecological framework that proximal th reats or stressors are more important than familial and distal factors experienced in th e broader community in regards to mental health outcomes. An additional limitation was in the area of measurement. The instrument for measuring depression, the CES-D scale (Radlo ff, 1977) is a measurement of depressive symptomatology not a measure to meet criteria for clinical depressi on. In regards to the outcome of suicide ideation, only one question for suicide ideation was used and may have constrained the results versus having a more comprehensive instrument assessing suicide gestures, hopelessness, and suicide attempts. The development of the composite variables for the AAE score were based on similar composites developed by Felitti and co lleagues (1998) utilizi ng select questions from various scales not originally intende d for these purposes. In this regard, age variations in the adolescent girls (13-17) in the first year of the study may have contributed to differences in perceptions and reports of a dverse experiences. Moreover,
116 the data were based on retrospective reca ll and may have resulted in underreporting. Current evidence from other longitudinal studi es demonstrates that retrospective selfreports of adverse exposures are likely to underestimate actual occu rrence (Della et al., 1990; Kessler et al., 1997). Finally, as in most studies, there may al so be mediating or moderating variables that were not examined that play a role in the relationship between adverse exposures and the outcomes of depression and suicide ideation. This is certainly suggested by the results only explaining a small proportion of the va riance for the outcomes under examination. An array of alternative expl anations for the results of this study could include the relationship of the perpetrato r to the adolescent victim, ch ronicity of exposures, and narrowness of the outcomes studied. Contributions of this Study and Implications for Public Health The study aimed to contribute to the existi ng knowledge base in three ways: first, by investigating the prevalence of stressors across multiple domains in the lives of female adolescents raised in the c ontext of welfare. Secondly, by capitalizing on longitudinal data to clarify the Â‘dose res ponseÂ’ or cumulative impact of exposures to adverse events and the developmental trajectory of depr ession and suicide ideation; and lastly, by exploring the impact of exposures across time and the change in scores for depression, suicide ideation, and mental distress over time. One of the most poignant findings of this study sample of adolescent girls growing up in poverty is that there was a significant level of exposure (89%) having experienced over three adverse events, almost three times the national rates (Kessler et al., 1997). In another recent na tional study on the relationship between suicide attempts
117 and childhood adverse exposures reported ap proximately 64% of both males and females experienced multiple adverse exposures (Dube et al., 2001). Other studies conducted in the United States estimate the prevalence of violence at 75% for those who have been exposed to at least one violent act in their lifetime (Barnett et al., 1997). In examining the greater report of suicid e ideation in the younge st group, previous research on pre-adolescents and adolescents found that youth with families of low control, low cohesiveness, and high conflict were more likely to report suicidal thoughts and behaviors (GAP Report No. 140, 1996). An in terpretation of this finding may be that this younger group is more vulnerable to the stre ssors they have been exposed to and lack adequate resources to respond to the e xposures versus their older peers. This disturbing evidence adds to the issu es of concerns for this population of Â“at risk girlsÂ” who are over exposed to adve rsity suggesting a need for interventions to protect or inoculate these adolescents from such extensive exposures. The implications for both public health prevention policy and prac tice for risk and protec tive factors are far reaching, given the level of depression seen in populations receiving public assistance. Both preventative strategies and interventi on strategies that are age and gender-specific are warranted at the family and community le vel offering support for victims of domestic and community violence. First points of en try into any public health systems including schools or medical settings should develop trauma sensitive screenings for adverse exposures. Given the importance of school environm ents and that drop out was directly associated with increased suicide ideation, evidence based strategies for fostering high school completion and reinforcing protective fact ors such as social and peer support, are
118 needed. The development of family services th at would offer interventions addressing the needs of these fragile families and the level of violence within the families is also critical. Given the relationship of maternal abuse and maternal illness on poor outcomes in daughters, interventions of therapeutic s upport seem warranted. Lastly, widespread preventative approaches aimed at reducing victimization in both families and communities are perhaps most important. It is also noteworthy, that the majori ty of studies on TANF recipients and depression have been on mothers from 20 to 40 years old. In contra st, existing studies on adolescents receiving public assistance have primarily focused on poor behavioral outcomes such as school attendance, pregnancy, and substance abuse. Researchers interested in social charac teristics have cautioned that focusing on disorderspecific outcomes may be utilizing an over-determini stic theoretical perspective and that there is a need to focus on broader mental health outcomes of both distress and resilience that would allow for greater understanding of the pathways and the mechanisms that may be influencing such outcomes (Aneshensel et al., 1991;Gennetian, Duncan, Knox, Vargas, Clark-Ka uffman, London, 2002; Yaqub, 2002). The results of this study s uggest that factors other th an poverty are at play for such outcomes and that sensitivity to differe nces in social group characteristics and vulnerabilities are needed to further explor e alternative explanati ons for depression and suicide in such Â‘at riskÂ’ populations. Specifically, gi ven the results that parental incarceration and divorce were protective for these adolescents offe rs a cautionary note that generalizing constructs of risk and protec tive factors of such soci al characteristics is dangerous.
119 Recommendations for Future Research There are multiple recommendations that include theoretical, conceptual, and methodological issues that have emerged from this study for future research, but also implications for policy changes. There re mains a paucity of longitudinal studies on adolescents investigating the onset of de pression and suicide ideation utilizing a theoretical framework. It has been argued that adolescents are at greater risk in regards to threats and strains as they transition into a dulthood and are exposed to a constellation of various stressors imposed on them from so cial structures, expectations, community environments, family experiences and issues of dependency. A theoretical consideration for future research would be to develop more inclusive or non-specific outcomes for studying th e impact of stressors on mental health. The results of this study supports the need to further explore the implications of multiple, co-occurring stressors and rela ted determinants for poor mental health outcomes. A basic premise of social stress theory is that the effects of stress are nonspecific, as evidenced by the empirical data of an array of disorder s that occur after an exposure to a stressor. The danger of misclassification of persons who are seemingly non-disordered due to specific categorization and thereby missing other manifestations of stre ss exposures is an approach argued by Aneshensel and colleagues (1991). Specifically, symptoms associated wi th depression during adolescence may manifest in various ways that do not meet a diagnostic category. Such outcomes could include: interpersonal and academic dysfunction, helplessness, anger, eating disorders, sexual promiscuity, running away, and substan ce abuse, all of whic h were not directly assessed in this study. Neither were specific symptoms associated with suicide ideation
120 assessed, such as increased anxiety, depres sion, stress, hopelessness, and loss of selfesteem (Hazler & Mellin, 2004). Moreover, give n that suicide is a relatively infrequent event, but a tragic and preventable outcome in the lives of many adolescents, there is a need for further exploration of mediating and moderating influences on other alternate endpoints such as suicide gestures, suic ide attempters, and suicide completers. Studies are also needed to assess the im pact of screening efforts for specific adverse exposures in adolescents, existing interventions and the timeliness of such interventions to circumvent poor outcomes and to increase protective processes for adolescents. Studies utilizing longitudinal methodologies that can more clearly assess the effects of time on the trajectory of depressi on over time, and where intervention points for specific ages should be placed to change a negative trajectory woul d be helpful in the future. Such information would allow for the examination for mechanisms of influence and risk amplifiers for poor mental hea lth outcomes among adolescents exposed to adversity. In addition, studies that further expl ore protective mechanisms that would potentially moderate exposures creating more malleable and flexible outcomes would be an important contribution to public health po licy, as there seems to be powerful unnamed variables present in the live s of some adolescents that encourage resilient outcomes, despite extreme exposures to adversities. The literature on cumulative measures of household dysfunction, while recently incorporating residential mobility/instability as a lifetime score, have not considered running away as part of the construct. Current studies define mobility by the number of moves and do not explore differences in the nature of the move. The preceding, distal,
121 and proximal factors of adoles cent residential inst ability present a troublesome gap in the literature, considering that r unning away in adolescents is associated with extremely poor outcomes such as psychological distress, substance abuse, victimization, and youth homelessness, further investigation is wa rranted (Kingree, Braithwaite, & Woodring, 2001). This study suggests that futu re research assessing the potentially liberating effect from stressful exposures in the home envir onment through parental divorce or parental incarceration is needed. While the results found here th at parental divorce or incarceration is protective is counter-intuitive, furthe r knowledge of adolescents attribution of adverse experiences with parent al figures is clearly needed to understand the buffering effect for mental health outcome s that parental separa tion or incarceration may offer in the face of perceived threats, family chaos, witnessing domestic violence, and child abuse. Finkelhor and colleagues (2001) suggest th at the primary theoretical concern for the child or adolescent victim of an adult pe rpetrator is that of dependency issues, which certainly implies the need for clearer policy ar ound these issues but a deeper investigation of runaway youth as well and the reasons fo r the behaviors. Youth, in general, are required to live with adults a nd have little choice of where they want to live, who they want to live with, or where they want to go to school. They are typically not free to leave or are financially unable to leave hostile home environments, hostile siblings, or even exposures to school violence or street crime. To this end, an important consideration for future studies would be to u tilize qualitative methods that wo uld offer greater explanatory power to further understand and identify the Â‘c ritical momentsÂ’ and the micro-processes
122 (Thomson & Holland, 2002) that occur betwee n social realities and the individual biography of each adolescent that define the choices to runaway or remain dependent. Conclusions In summary, findings from this study cons idered the cumulative impact, or in the rubric of epidemiology, the Â‘dose responseÂ’ relationship, of exposures to household dysfunction, community violence, and persona l victimization on depression and suicide ideation over time. The results found that expo sure to adolescent victimization was the primary predictor for depression, suic ide ideation, and me ntal distress. It is hoped that this rese arch has contributed both conceptually and empirically to our understanding of the pathways and the critical processe s and interactions between individual, family, and contextual stresso rs impacting the developmental period of adolescence. This study was conducted based on the assumption that these results will help to bring to awareness the prevalence of adverse exposures occurring in the lives of adolescents females and hopefully guide future efforts towards trauma sensitive and gender specific interventions that may help to alleviate the burden of depression and suicide ideation. It is argued by some that Â“the sufferings of childhood are indelibly stamped on the adultsÂ” (Engels, 1845/1958; Kr ieger & Smith, 2004). The higher rates of depression in females living in poverty may be due to th e relationship of such indelibly stamped exposures. The results of this study offer ev idence on the high prevalence rates of adversity occurring in the lives of these adol escents, and the cumulative impact of such exposures on depression and suicide ideati on. This study found that those experiencing
123 victimization were almost five times more likely to report suicid e ideation, and over two times more likely to develop depression. Public health goals of reducing depre ssion and suicide will fail without greater considerations and interventions to address the fundamental facts of exposures of youth to an array of adversities. Efforts towards reducing community violence and personal victimization as well as developing trauma sensitive interventions that would buffer household dysfunction may play an important role in preventing depression, and suicide ideation.
124 REFERENCES American Foundation for Suicide Preventi on (AFSP). (2003). Retrieved February 11, 2004 from http://www.afsp.org/index-1.htm Aneshensel, C.S., Rutter, C.M., & Lachenbruch, P.A. (1991). Social st ructure, stress, and mental health: Competing con ceptual and analytic models. American Sociological Review, 56 166-178. Aneshensel, C.S. & Sucoff, C.A. (1996). The neighborhood context of adolescent mental health. Journal of Health and Social Behavior, 37 293-310. Anda, R.F., Felitti, V.J., Bremmer, J. D., Walker, J.D., Whitfield, C.L., Perry, B. D., Dube, S.R., Giles, W.H. (2006). The enduri ng effects of abuse and related adverse experiences in childhood: A convergen ce of evidence from neurobiology and epidemiology. European Archives of Psychiatr y Clinical Neuroscience, 256 174186. Barnett, O., Miller-Perri n, C., & Perrin, R. (1997). Family Violence Across the Lifespan. Thousand Oaks, CA: Sag Becker, E. (2001). Adversity and its outcomes: The measurement of childhood trauma. In Frane y, K., Geffner, R., Falconer, R. (Eds). The Cost of Child Maltreat ment: Who Pays? We All Do (pp.53-69). San Diego, CA: Family Violence & Sexual Assault Institute. Belle, D. (1990). Poverty and womenÂ’s mental health. American Psychologist 45 385389.
125 Bergen, H.A., Martin, G., Richardson, A.S ., Allison, S. L., Roeger, L. (2003). Sexual abuse and suicidal behavior: A model cons tructed from a large community sample of adolescents. Journal of American Academy of Child Adolescent Psychiatry, 42 (11), 1301-1309. Best, K.A., Armstrong, M.I., Boothroyd, R.A., Gomez, A., Haynes, D., Ort, R., Chen. (2004, January). Achieving the Vision: The Future Goals and Aspirations of Adolescent Girls Living in Families Receiving TANF. Paper presented at the 16 th Annual Conference of the S outheast Evaluation Associat ion, Tallahassee, FL. Best, K.A., Boothroyd, R.A., Armstrong, M., Haye s, D., Chen, R., Ort, R. Gomez, A. (2006). Rethinking Female Adolescent Depre ssion in the Context of Poverty. Conference proceedings for the 17th Annual Research Conference: A System of Care for ChildrenÂ’s Mental Health: E xpanding the Research Base, Tampa, FL: University of South Florida Press. Boothroyd, R. A., Armstrong, M. I., Gmez, A. Haynes, D., Ort, R. & Best, K. (2005). Welfare reform: Adolescent girls in tr ansition Â– A Three-year follow-up study. Tampa, FL: University of South Florida, Louis de la Parte Fl orida Mental Health Institute. Boothroyd, R. A., Armstrong, M. I., Gm ez, A. Haynes, D., & Ort, R. (2003). Welfare reform: Adolescent Girls in transit ion Â– A One-year follow-up study. Tampa, FL: University of South Florida, Louis de la Parte Florida Mental Health Institute. http://www.fmhi.usf.edu/institute /pubs/pdf/ahca/2003-boothroyd-welfaretransition-girls.pdf
126 Boothroyd, R. A., & Olufokunbi, D. (2001). Leav ing the welfare rolls: The health and mental health status of current and former welfare recipients. Mental Health Services Research 3 (3), 119-128. Boushey, H., & Gundersen, B. (2001). When working just isnÂ’ t enough: Measuring the hardships faced by families af ter moving from welfare to work Briefing Paper [on-line]. Economic Policy Institute. Retrieved on December 17, 2003 from http://www.epinet.org/content. cfm/briefingpapers_hardships Brooks-Gunn, J. (1991). How stressful is the tran sition to adolescence for girls? In M.E. Colten & S. Gore (Eds.), Adolescent Stress: C auses and Consequences (pp. 1311490). New York: Aldine de Gruyter. Brooks, M. G., & Buckner, J. C. (1996). Work history and welfare: J ob histories, barriers to employment, and predictors of wo rk among low-income single mothers. American Journal of Orthopsychiatry 66 526-537. Bronfenbrenner, U. (1989). Ecological systems theory. Annals of Child Development, 6 187-249. Bryant, S.L. & Range, L.M. (2001). Type and severity of child abuse and college students lifetime suicidality. In Franey, K., Geffner, R., Falconer, R. (Eds). The Cost of Child Maltreatmen t: Who Pays? We All Do (pp.53-69). San Diego, CA: Family Violence & Sexual Assault Institute. Buhi, E.R., Goodson, P., & Neilands, T.B. (2008). Out of sight, not out of mind: Strategies for handing missing data. Ameri can Journal of Health Behavior, 32(1); 83-92.
127 Cassel, J. (1964). Social science theory as a source of hypotheses in epidemiological research. American Journal of Public Health 54 (9), 1482-1488. Centers for Disease Control and Prevention. National Center for Injury Prevention and Control. Retrieved 6/17/2002 from http://www.cdc.gov/ncipc/duip/duip.htm Chapman, D.P., Whitfield, C.L., Felitti, V.J., Dube, S. R., Edwards, V.J. & Anda, R.F. (2004). Adverse childhood expe riences and the risk of depressive disorders in adulthood. Journal of Affective Disorders. 82, 217-225. Child Trends (2003). Child Trends Data bank Indicator. Retreived 7/9/2007 from http://childtrendsdataba nk.org/PDF/Violence.pdf Cohen, C. I. (2002). Social inequality and h ealth: Will psychiatry assume center stage? Psychiatric Services, 53 (8): 937-939.. Dallam, S.J. (2001). The long-term medical consequences of childhood maltreatment. In The Cost of Child Maltreat ment: Who Pays? We all do Family Violence and Sexual Assault Institute, San Diego, CA. Della, Femina, D., Yeager, C.A., & Lewis, D. O. (1990). Child abuse: Adolescent records vs. adult recall. Child Abuse & Neglect, 14, 227-231. Dekovic, M., Buist, K.L., & Reitz, E. (2004). Stability and changes in problem behavior during adolescence: Latent growth analysis. Journal of Youth and Adolescence, 33, (1). 1-12. Dooley, D. & Prause, J. (2002). Mental hea lth and welfare transitions: Depression and alcohol abuse in AFDC women. American Journal of Community Psychology, 30 (6), 787-813.
128 Dong, M., Anda, R.F., Felitti, V. J., Dube, S. R., Williamson, D. F., Thompson, T.J., Loo, C.M., Giles, W.H. (2004). The interr elatedness of multiple forms of childhood abuse, neglect, and household dysfunction. Child Abuse & Neglect, 28. 771-784. Dong, M., Anda, R.F., Felitti, V. J., Williams on, D. F., Dube, S. R., Brown, D.W., & Giles, W.H. (2005). Childhood residential mobility and multiple health risks during adolescence and adulthood: Th e hidden role of adverse childhood experiences. Archives of Pedi atric Medicine, 159 1104-1110. Dougherty, L.R., Klein, D.N., & Davila, J. (2004) A growth curve analysis of the course of dysthymic disorder: The effects of chronic stress and moderation by adverse parent-child relationshi ps and family history. Journal of Consulting and Clinical Psychology. 72 (6), 1012-1021. Dube, S.R., Anda, R.F., Felitti, V.J., Ch apman, D.P., Williamson, D. F Giles, W.H. (2001). Childhood abuse, household dysfunc tion, and the risk of attempted suicide throughout the lifesp an: Findings from the a dverse childhood experiences study. Journal of the American Medical Association, 286 (24), 3089Dube, S.R., Anda, R.F., Whitfield, C.L., Br own, D.W., Felitti, V.J., Dong, M., Giles, W.H. (2005). Long-term consequences of childhood sexual abuse by gender of victim. American Journal of Preventive Medicine, 28 (5), 430438. Duckworth, M.P., Hale, D.D., Clair, S.D ., & Adams, H.E. (2000). Influence of interpersonal violence and community ch aos on stress reactions in children. Journal of Interpe rsonal Violence, 15 (8), 806-826. Eaton, D.K., Kann,L., Crosby, A., & Glenn, F. (November, 2007). Do associations of suicide ideation and suicide attempts with health risk behaviors vary by
129 race/ethnicity among 9th-12th grade fema le students? Paper presented at the 135th American Public Hea lth Association Annual Me eting. Washington, DC. Elstad, J.I. (1998). The psycho-social pers pective on social inequalities in health. Sociology of Health & Illness, 20 (5), 598-618. Engels F. (1958). The Condition of the Working Class in England (W.O. Henderson & W.H. Chaloner, Trans.). Stanford, CA: Stan ford University Press. (Original work published in 1845). Favorini, A. (1995). Concept of codependenc y: blaming the victim or pathway to recovery? Social Work. 40, 827-831. Felitti, V. J., Anda, R. F., Nordenberg, D., Williamson, D.F., Spitz, A.M., Edwards,V., Koss, M.P., Marks, J.S. (1998). Relationship of childhood abuse and household dysfunction to many of the leading caus es of death in adults: The Adverse Childhood Experiences (ACE) Study. American Journal of Preventative Medicine, 14 (4), 245-258. Finkelhor, D. & Hashima, P.Y. (2001). Th e victimization of children and youth: A comprehensive overview. In K. White (Eds). Handbook of Youth and Justice New York: Plenum Publishers. Foege, W.H. (1998). Adverse childhood expe riences: A public health perspective. American Journal of Preventative Medicine, 14 (4), 354-355. Friedman R.M., Best, K. A., Armstrong M.I., Duchnowski A.J., Evans M.E., Hernandez M., Hodges, S., & Kutash, K.B. (2004). Child mental health policy. In: B. L. Levin, J. Petrila, & K. Hennessy (eds). Mental Health Services: A Public Health Perspective, 2nd Edition. New York: Oxford University Press.
130 Friis, R.H. & Sellers, T.A. (1999). Epidemiology for Public Health Practice Maryland: Aspen Publishers. Frojd, S., Marttunen, M., Pelkonen, M., Pahlen, B. & Kaltiala-Heino, R. (2006). Perceived financial difficulties and ma ladjustment outcomes in adolescence. European Journal of Public Health, 15 (5), 542-548. Furstenberg, F. (1992). The next generation: The children of teenage mothers grow up. In M. Rosenbaum and M.Testa (Eds.), Early Parenthood and Coming of Age in the 1990s New Brunswick, NJ: Rutgers University Press. Garmezy, N. & Masten, A. (1994). Chronic Adve rsities. In M. Rutter, E. Taylor, & L. Herson (Eds.), Child and Adolescent Psychiatry. Blackwell Scientific Publications, Boston. Garrison, C., Addy, C., Jackson, K., McKeow n, R., Waller, J. (1991). A longitudinal study of suicidal ideat ion in young adolescents. Journal of the American Academy of Child and Adolescent Psychiatry. 30, 597-603. Ge, X., Lorenz, F.O., Conger, R.D., Elder, G.H., Simons, R.L, (1994). Trajectories of stressful life events and depressi ve symptoms during adolescence. Developmental Psychology, 30 (4), 47-483. Gennetian, L. A., Duncan, G. J., Knox, V. W., Vargas, W. G., Clark-Kauffman, E., London, A. S. (2002, May). How Welfare and Work Policies for Parents Affect Adolescents: A Synthesis of Research. New York: Manpower Development Research Corporation.
131 Gerard, J.M. & Buehler, C. (2004). Cu mulative environmental risk and youth maladjustment: The role of youth attributes. Child Development, 75 (6), 18321849. Gilman, S.E., Kawachi, I., Fitzmaurice, G.M. & Buka, S.L. (2003). Socio-economic status, family disruption and residential stability in childhood: relation to onset, recurrence and remission of major depression. Psychological Medicine, 33 : 13411355. Glaser, D. (2002). Emotional abuse and neglect (psychologi cal maltreatment: a conceptual framework. Child Abuse & Neglect. 26 pp. 697-714. Goodman, D., Adler, N.E., Kawachi, I., Fraz ier, A.L., Huang, B., & Colditz, G.A. (2001). AdolescentsÂ’ perceptions of social stat us: development and evaluation of a new indicator. Pediatrics, 108 (2), 1-8. Retrieve d August 17, 2006 on line at: www.pediatrics.org/cgi/content/full/108/2/e31 Gottschalk, P., McLanahan, S., & Sandefur, G. (1994). The dynamics and intergenerational transmission of poverty and welfare participation. In S. H. Danziger, G. D. Sandefour, a nd D. H. Weinberg, (Eds.), Confronting Poverty: Prescriptions for Change Cambridge, MA: Harvard University Press. Group for the Advancement of Psychiatry (GAP). (1996). Adolescent Suicide Committee on Adolescence, Report No. 140: Washingt on, DC: American Psychiatric Press, Inc. Gutman, L. M. & Sameroff, A.J. (2004). Contin uities in depression from adolescence to young adulthood: Contrasting ecologi cal influences. Developmental Psychopathology, 16 (4), 967-984.
132 Hammen, C., Henry, R., & Daley, S.E. (2000). Depression and sensitization to stressors among young women as a function of childhood adversity. Journal of Consulting and Clinical Psychology, 68 (5), 782-787. Haveman, R., & Wolfe, B. (1995). The determin ants of children's attainments: A review of methods and findings. Journal of Economic Literature 32 1829-1878. Hazler, R.J. & Mellin, E.A. (2004). The deve lopmental origins and treatment needs of female adolescent with depression. Journal of Counseling & Development 82 18-24. Howard, D.E., Feigelman, S., Li, X., Cross, S., & Rachuba, L. (2002). The relationship among violence victimization, witnes sing violence and youth distress. Journal of Adolescent Health. 31, 455-462. Institute of Medicine. (2002). Reducing Suicide: A National Imperative. Goldsmith, S.K., Pellmar, T. C., Kleinman, A. M., Bunne y, W. E. (Eds.). National Academies Press. Retrieved February 10, 2004 from http://www.nap.edu/books/0309083214/html/ Institute of Medicine. (2003). Summary. In Who Will Keep the Public Healthy? Educating Public Health Professionals for the 21st Century. (pp.3-26). Washington, DC: National Academies Press. Irwin, C.E., Burg, S.J., & Cart, C.U. (2002) AmericaÂ’s adolescents: Where have we been, where are we going? Journal of Adolescent Health 31, 91-121. Jellinek, M. S., Murphy, J. M., & Burns, B. J. (1986). Brief psychosocial screening in outpatient pediatric practice. The Journal of Pediatrics 109 371-378.
133 Jodl, K.M.,, Michel, A., Malanchuk, O., Eccles, J.S., Sameroff, A. (2001). ParentsÂ’ roles in shaping early adolescent sÂ’ occupational aspirations. Child Development, 72 4, 1247-1265. Johnson, J.G., Cohen, P., Gould, M.S., Kase n, S., Brown, J., & Brook, J.S. (2002). Childhood adversities, interper sonal difficulties, and risk for sick attempts during late adolescence and early adulthood. Archives of General Psychiatry, 59 741749. Johnson, R.M., Kotch, J.B., Catellier, D.J., Wi nsor, J.R., Dufort, V., Hunter, W., AmayaJackson, L. (2002). Adverse behavior al and emotional outcomes from child abuse and witnessed violence. Child Maltreatment. 7 (3), 179-186. Jong, K. de., Mulham, M. & Kam, S. van de r. (January 11, 2000). Assessing Trauma in Sierra Leone: Psychosocial Questionnaire: Freetown Survey Outcomes. A Doctors without Borders: Medecen s Sans Frontieres Special Report. Kalil, A., Born, C.D., Knuz, J., & Cuadill, P.J. (2001). Life stressors, social support, and depressive symptoms among fi rst-time welfare recipients. American Journal of Community Psychology 20 355-369. Karr-Morse, R. & Wiley, M.S. (1997). Ghosts from the Nursery: Tracing the Roots of Violence. New York, NY: Atlantic Monthly Press. Kazdin, A., Kraemer, H.C., Kessler, R.C., Kupf er, D., & Offord, D. (1997). Contribution of risk factor research to developmental psychopathology. Clinical Psychological Review,14, (4), 375-406.
134 Kendall-Tackett, K. A., Williams, L. M., & Finkelhor, D. (1993). Impact of sexual abuse on children: A review and synthe sis of recent empirical studies. Psychological Bulletin 113 164-180. Kessler, R.G., Davis, C.G., & Kendler, K. S. (1997). Childhood adversity and adult psychiatric disorder in the US National Comorbidity Survey. Psychological Medicine, 27 1101-1119. Kingree, J.B., Braithwaite, R., Woodring, T. (2001). Psychosocia l and behavioral problems in relation to recent experi ence as a runaway among adolescent detainees. Criminal Justice and Behavior 28 (2), 190-205. Kirby, D. (1997). No easy answers: Research findings on programs to reduce teen pregnancy Washington, DC: National Campai gn to Prevent Teen Pregnancy, Task Force on Effective Programs and Research. Kirby, L.D. & Fraser, M.W. (1997). Risk and Resilience in Childhood: An Ecological Perspective. Washington, DC: NASW Press. Kraemer, H.C., Stice, E., Kazdin, A., Offo rd, D., & Kupfer, D. (2001). How do risk factors work together? Mediators, mode rators, and independent, overlapping, and proxy risk factors. American Journal of Psychiatry. 158 (6), 848-856. Krieger, N. (2001). Theories of social epidemiology in the 21st century: An ecosocial perspective. International Journal of Epidemiology 30, 668-677. Krieger, N. & Smith, G.D. (2004). Â“Bodi es Count,Â” and body counts: Social epidemiology and embodying inequality. Epidemiologic Reviews 26, 92-103. Leginski, Croze, Driggers, Dumpman, G eertsen, Kamis-Gould, Namerow, & Lincoln (1989).
135 Lichter, D. T., & Crowley, M. C. (2000). Poverty in American: Beyond welfare reform. Population Bulletin 57 (2), 1-36. Link, B.G. & Phelan, J. (1995). Social condit ions as fundamental causes of disease. Journal of Health and Social Behavior, (Extra Issue): 80-94. Locke, L.F.,Wyric Spirduso, W., & Silverman, S.J. (2000). Proposals that Work: A Guide for Planning Dissertations and Grant Proposals (4th ed.). Thousand Oaks: Sage Publication, Inc. Lorant, V., Deliege, D., Eaton, W., Robert A., Philippot, P., & Ansseau, M. (2003). Socioeconomic inequalities in depression: A meta-analysis. American Journal of Epidemiology 157 98-112. Luthar, S.S. & Goldstein, A. (2004). Ch ildrenÂ’s exposure to community violence: Implications for understandi ng risk and resilience. Journal of Clinical Child and Adolescent Psychology, 33 (3), 499-505. Luthar, S.S. & Zigler, E. ( 1991). Vulnerability and competen ce: A review of research on resilience in childhood. American Journal of Arthopsychiatry 61 (1). 6-22. Magdol, L., (2002). Is Moving Gendered? The effects of residential mobility on the psychological well-being of men and women. Sex Roles, 47 553-560. Martin, S.L., Anderson, K.L., Lynch, S.R., & Kupper, L.L. (1999). Vi olence in the lives of pregnant teenage women: associations with multiple substance use. American Journal of Drug and Alcohol Abuse, 25, 425-440. Mazza, J.J. & Reynolds, W.M. (1999). E xposure to violence in young inner-city adolescents: Relationships with suic idal ideation, depression, and PTSD symptomatology. Journal of Abnormal Child Psychology, 27 (3), 203-213.
136 McMahon, S.D., Grant, K. E., Compas, B.E ., Thurm, A.E., Ey, S. (2003). Stress and psychopathology in children and adolescents : is there evidence of specificity? Journal of Child Psychology and Psychiatry, 44 (1), 107-133. McKnight, K.M., McKnight, P.E., Si dani, S. & Figueredo, A.J. (2007). Missing Data: A gentle introduction Guilford Publishing. Menard, C., Bandeen-Roche, K.J., Chilcoat, H.D. (2004). Epidemiology of multiple childhood traumatic events: child abuse, parental psychopathology, and other family-level stressors. Social Psychiatric Epidemiology, 39 857-865. Miller, B.A., Smyth, N.J., and Mudar, P.J. (1999). MothersÂ’ alcohol and other drug problems and their punitiveness toward their children. Journal of Studies on Alcohol, 60, 632-642. Molnar, B.E., Buka, S.L., Kessler, R.C. ( 2001). Child sexual abuse and subsequent psychopathology: Results from the National Comorbidity Survey. American Journal of Public Health 91, 753-760. Monaghan, J. H., Robinson, J. O., & Dodge, J. A. (1979). The ChildrenÂ’s Life Events Inventory. Journal of Psychosomatic Research 23 63-68. Mullen, P.E., Martin, J. L., Anderson, J.C ., Romans, S.E., Herbison, G.P. (1996). The long-term impact of the physical, emoti onal, and sexual abuse of children: A community study. Child Abuse Neglect, 20 (1), 7-21. Muntaner, C., Eaton, W.E., Miech, R., & OÂ’ Campo, P. (2004). Socioeconomic position and major mental disorders. Epidemiologic Reviews 26, 53-62.
137 Murphy, J.M., Jellinek, M., Quinn, D., Smit h, G., Poitrast, F.G., & Goshko, M. (1991). Substance abuse and serious child mistreat ment: Prevalence, risk, and outcome in a court sample. Child Abuse and Neglect, 15, 197-211. Musick, J. (1993). Young, Poor, and Pregnant: The Psychology of Teenage Motherhood Yale University Press: New Haven and London. National Institute of Mental Health. Series FN No.10, Data Standards for Mental Health Decision Support Systems, by Leginski, W.A.; Croze, C.; Driggers, J.; Dumpman, S.; Geertsen, D.; Kamis-Gould, E.; Namerow, M.J.; Patton, R.E.; Wilson, N.Z.; and Wurster, C.R. DHHS Pub. No. ( ADM)89-1589. Washington, DC: Supt. of Docs., U.S. Govt. Print. Off., 1989. DHHS Publications No. (ADM)89-1589 Printed 1989 National Center for Children Exposed to Violence (NCCEV). (2006). Child Study Center, Yale University. Re trieved February 10, 2007 from http://www.nccev.org/violence/community.html National Committee to Prevent Child Abuse. (1998). Child Abuse and Neglect Statistics. Retrieved March 2006 from http ://www.childabuse.org/facts National Institutes of Health (NIH) and Subs tance Abuse and Mental Health Services Administration. (2000). Healthy People 2010. Retrieved March 25, 2004 from http://www.healthypeople.gov/Documen t/HTML/Volume2/18Mental.htm#_Toc4 86932699 Nurco, D.N., Blatchley, R. J., Hanlon,T.E., and OÂ’Grady, K.E. (1999). Early deviance and related risk factors in the children of narcotic addicts. American Journal of Drug and Alcohol Abuse, 25, 25-45.
138 OÂ’Donnell, D. A., Schwab-Stone, M.E., Muyeed, A. Z. (2002). Multidimensional resilience in urban children exposed to community violence. Child Development, 73 ( 4), 1265-1282. Pearlin, L.I. (1989). The soci ological study of stress. Journal of Health and Social Behavior, 30 (3), 241-256. Pearlin, L. I., M. A. Lieberman, E. G. Me neghan, and J. T. Mullan. (1981). The stress process. Journal of Health and Social Behavior, 22, 337-356. Pearlin, L.I. & Schooler, C. ( 1978). The structure of coping. Journal of Health and Social Behavior, 19 2-21. Perry, B. D. (2001). The neuroarc heology of childhood maltreatment: The neurodevelopmental costs of adverse childhood events. The Cost of Child Maltreatment: Who Pays? We All Do. Family Violence & Sexual Assault Institute. San Diego, CA. Personal Responsibility and Work Opportuni ty Reconciliation Act of 1996, P.L. 104-193, 110 Stat. 2105. Radloff, L. S. (1977). The CE S-D Scale: A self-report depr ession scale for research in the general population. Applied Psychological Measurement 1 385-401. Reifman, A. & Windle, M. (1995). Adolescen t suidical behaviors as a function of depression, hopelessness, alcohl use, and social support: A longitutdianl investiagion. American Jounal of Community Psyhcoloy, (23), 3, 329 Â– 354. Ritchey, F., Gory, M., Fitzpatrick, K., & Mu llis, J. (1990). A comparison of homeless, community-wide, and selected distressed samples on the CES-Depression Scale. American Journal of Public Health 80 1384-1386.
139 Rudolph, K.D., Hammen, C., Burge, D., Lindberg, N., Herzberg, D., & Daley, S.E. (2000). Toward an interpersonal lifestress model of depression: The developmental context of stress generation. Development and Psychopathology, 12 215-234. Rutter, M. (2005). Environmentally mediat ed risks for psychopathology: Research strategies and findings. Journal of American Academy Child Adolescent, Psychiatry, 44, 1, 3-8. Sard, B. (2002). A housing perspective on TANF reauthorization and support for working families. Center on Budget and Policy Priorities [Retrieved on line June 22, 2004] from http://www.cbpp.org/3-12-02hous.htm Scarpa, A. Hurley, J.D., Shumate, H.W. & Haden, S. C. (2006). Lifetime prevalence and socioemotional effects of hear ing about community violence. Journal of Interpersonal Violence, 21 (1), 5-23. Schneider,A. & Ingram, H. (1993). Social construction of target populations: Implications for politics and policy. American Political Science Review, 87 (2), 334-347. Schultz, A.J., Gravlee, C.C., Williams, D.R., Israel, B.A., Mentz, G., Rowe, Z. (2006). Discrimination, symptoms of depressi on, and self-rated health among African American women in Detroit: Resu lts from a longitudinal Analysis. American Journal of Public Health, 96 (7), 1265-1270. Silverstein, M., Augustyn, M., Cabral, H., & Zudkermean, B. (2006). Maternal depression and violence e xposure: Double jeopardy for child school functioning. Pediatrics, 118 (3): 792-800.
140 Simon, T.R., Anderson, M., Thompson, M.P.,Cro sby, A., & Sacks, J.J. (2002). Assault victimization and suicidal ideation or beha vior within a national sample of U.S. adults. Suicide and Life-Threatening Behavior, 32 (1), 42-50. Smith, D.K., Leve, L.K., Chamberlain, P. (2006). Adolescent GirlsÂ’ Offending and Health-Risking Sexual Behavior: Th e predictive role of trauma. Child Maltreatment, 11 (4), 346-353. Spat Widom, C. P.D. (1999). Post-traumatic stress disorder in abused and neglected children grown up. American Journal of Psychiatry 156 1223-1229. Stein, J.A., Leslie, M.B., Nyamathi, A. ( 2002). Relative contributions of parent substance use and childhood maltreatment to chronic homelessness, depression, and substance abuse problems among home less women: mediating roles of selfesteem and abuse in adulthood. Child Abuse & Neglect 26, 1011-1027. Stevens, S. J., Murphy, B. S., & McKni ght, K. (2003). Traumatic stress and gender differences in relationship to substance a buse, mental health, physical health, and HIV risk behavior in a sa mple of adolescents enrolled in drug treatment. Child Maltreatment 8 46-57. Taylor, J. & Turner, R.J. (2002) Perceived discrimination, so cial stress, and depression in the transition to adu lthood: Racial contrasts. Social Psychology Quarterly, 65 (3): 213-225. Thoits, P. (1995). Stress, coping, and social support processes: Where are we? What next? Journal of Health and Social Behavior (Extra Issue): 53-79.
141 Thomson, R. & Holland, J. (June, 2002). Final Report: Inventing Adulthoods: Young Peoples Strategies for Transition. Retrieved April 18, 2005 from http://www.lsbu.ac.uk/ahs/ff/reports/finalreport.shtml Thompson, E.A., Mazza, J.J., Herting, J.R., Randell, B.P., & Eggert, L.L. (2005). The mediating roles of anxiety, depressio n, and hopelessness on adolescent suicidal behaviors. Suicide and Life-threatening Behavior, 35 (1), 14-34. Tsuang, M, T. & Tohen, M. (Eds). (2002). Textbook in Psychiatric Epidemiology 2nd Ed. New York, NY: John Wiley & sons, Inc. Turner, H.A., Finkelhor, D., Ormrod, R. (2006) The effect of lifetime victimization on the mental health of children and adolescents. Social Science & Medicine, 62 1327. Turner, R.J. & Lloyd, D. A. (1995). Lifetime tr aumas and mental health: The significance of cumulative adversity. Journal of Health and Social Behavior, 36 (4). 360-376. Turner, R.J. & Lloyd, D. A. (1999). The stress process and the social distribution of depression. Journal of Health and Social Behavior, 40 (4), 374-404. United States Department of Hea lth and Human Services. (1999). Mental Health: A Report of the Surgeon General. Rockville, MD: U.S. Department of Health and Human Services. United States Department of Hea lth and Human Services. (2001). National Strategy for Suicide Prevention: Goals and Objectives for Action. Rockville, MD: US Public Health Service. United States Public Health Service Of fice of the Surgeon General. (2003). The Presidents New Freedom Commission on Mental Health. Achieving the Promise:
142 Transforming Mental Health Care in America Final Report. Rockville, MD: Department of Health and Human Serv ices. US Public Health Service. Vermeiren, R., Ruchkin, V., Leckman, P.E., Deboutte, D., Schwab-Stone, M. (2002). Exposure to violence and suicide risk in adolescents: A community study. Journal of Abnormal Child Psychology, 30 (5), 529-537. Wadsworth, M.E., Raviv, T., Compas, B.E., Connor-Smith, J.K. (2005). Parent and Adolescent Responses to Poverty Rela ted Stress: Tests of Mediated and Moderated Coping Models. Journal of Child and Family Studies 14 (2), 283 298. White, K.S., Bruce, S.E., Farrell, A.D., & K liewer, W. (1998). Impact of exposure to community violence on anxiety: A longitudi nal study of family social support as a protective factor for urban children. Journal of Child and Family Studies, 7, (2), 187-203. White, I.R., Carpenter, J., Evan s, S., & Schroter, S. (2004). Eliciting and using expert opinions about dropout bias in randomized controlled trials. ESRC Research Methods Programme: Â‘Missing data in multi-level modelsÂ’. Retrieved September 10th on line at http://www.lshtm.ac.uk/ms u/missingdata/papers/IMpriors7.pdf Whitfield, C.L. (1998). Adverse childhood experiences and trauma. American Journal of Preventive Medicine, 1 4 (4), 361-364. Wight, R.G., Sepulveda, J.E., & Aneshensel, C.S. (2004). Depressive symptoms: how do adolescents compare with adults? Journal of Adolescent Health 34, 314-323.
143 World Health Organization (2001). The World Health Report 2001, Mental Health: New Understanding, New Hope Geneva, Switzerland. [Ret rieved On-line November 2004] at http://www.who.int/msa/mnh/ems/dalys/intro.htm#unmet Yaqub, S. (2002). Poor children grow into poor adults: Harmful mechanisms or overdeterministic theory? Journal of International Development 14, 1081-1093. Zaslow, M. J., Moore, K. A., Brooks, J. L., Morris, P. A., Tout, K., Redd, Z. A., & Emig, C. A. (2001). Experimental studies of welfare reform and children. The Future of Children 12 79-95.
145 Appendix A Measures Included in the 2003 A dolescent Interview Protocol Domains Source Client Demographics Leginski, W. A., Croze, C., Driggers, J., Dumpman, S., Geertsen, D., Kamis-Gould, E., Namerow, J. J., & Lincoln, Y. S. (1989). Data Standards for Mental Health Decision Support Systems (ADM89-1589). Rockville, MD: National Institute of Mental Health. Health Status (SF-12) Keller, S. D., Kosinski, M., & Ware, J. E. (1996). A 12-Item Short-Form Health Survey (SF-12). A construction of scales and preliminary tests of reliability and validity. Medical Care 32 (3), 220-223. Mental Health Status ( i.e. symptomatology [PSC] and depression [CES-D]) Jellinek, M. S., Murphy, J. M., & Burns, B. J. (1986). Brief psychosocial screening in outpatient pediatric practice. The Journal of Pediatrics 109 371-378. Radloff, L. S. (1977). The CES-D Scal e: A self-report depression scale for research in the general population. Applied Psychological Measurement 1 385-401. Substance Abuse Status and Use (SSI) Winters, K. C., & Zenilman, J. M. (1994).The Simple Screening Instrument for Alcohol and Other Drug Abuse and Infections. (TIP #11). Rockville, MD: Substance Abuse and Mental Health Services Administration, Center for Substance Abuse Treatment. Functioning Substance Abuse and Mental Health Services (1996). Managed care multisite study. Quality of Life with: Living Situation, Family Relationships, Finances, Work & School, & Health Lehman, A., (1988). A quality of life interview for the chronically mentally ill. Evaluation and Program Planning 11 51-62. Life Events Monaghan, J. H., Robinson, J. O., & Dodge, J. A. (1979). The ChildrenÂ’s Life Events Inventory. Journal of Psychosomatic Research 23 63-68. Religiousness/Spirituality Fetzer Institute & National Institute on Aging (1999). Brief mulitdimensional measurement of religiousness/spirituality for use in health research Kalamazoo, MI: Fetzer Institute. Self-Efficacy Connolly, J. (1989). Social self-effi cacy in adolescence: Relations with self-concept, social adjustment, and mental health. Canadian Journal of Behavioural Science 21 258-269. Bandura, A. (2001). Guide for Constructing Self-Efficacy Scales. Stanford, CA: Stanford University. Self-Esteem Rosenberg, M. (1989) Society and the Adolescent Self-Image (rev. ed.). Middletown, CN: Wesleyan University Press. Social Supports Harter, S. (1985). Manual for the Social Support Scale for Children Denver, CO: University of Denver. Locus of Control Nowicki, S., & Strickland, B. R. (1973). A locus of control scale for children. Journal of Consulting and Clinical Psychology 40 148-154. Service Needs and Use Self-developed.
146 Domains Source Hopes and Aspirations Self-developed. High Risk Behaviors Goodenow, C. (1999). Massachusetts Youth Risk Behavior Survey Massachusetts Department of Education. http://www.doe.mass.edu/lss/yrbs99/acknowledge.html Hess, J. C., & Rothgeb, J. M. (1999). Measuring the impact of welfare reform: Issues in designing the survey of program dynamics questionnaire. Washington, DC: US Census Bureau. Family Relationships (FAD Version 3) Epstein, N. B., Baldwin, L. M., & Bishop, D. S. (1983). The McMaster family assessment device. Journal of Marital and Family Therapy 9: 171180. Attitudes Toward Marriage Johnson, C. A., Stanley, S. M., Glenn, N. D., Amato, P. R. (2001). Marriage in Oklahoma: 2001 Baseline Statewide Survey on Marriage and Divorce. Stillwater, OK: Oklahoma State University.
147 Appendix B Measures Included in the 2003 Mo therÂ’s Interview Protocol Domains Source Client Demographics Leginski, W. A., Croze, C., Driggers, J., Dumpman, S., Geertsen, D., Kamis-Gould, E., Na merow, J. J., & Lincoln, Y. S. (1989). Data Standards for Mental Health Decision Support Systems (ADM89-1589). Rockville, MD: National Institute of Mental Health. Health Status (SF-12) (about daughter, few general questions about self) Keller, S.D., Kosinski, M., & Wa re, J. E. (1996). A 12-Item Short-Form Health Survey (SF12). A construction of scales and preliminary tests of reliability and validity. Medical Care 32 (3), 220-223. Mental Health Status ( i.e. symptomatology [PSC] (about daughter) and depression [CES-D] (about self) Jellinek, M. S., Murphy, J. M., & Burns, B. J. (1986). Brief psychosocial screening in outpatient pediatric practice. The Journal of Pediatrics 109 371-378. Radloff, L. S. (1977). Th e CES-D Scale: A self-report depression scale for research in the general population. Applied Psychological Measurement 1 385-401. Substance Abuse Status and Use Adapted from other studies. Functioning (about daughter) Substance Abuse and Mental Health Services (1996). Managed care multi-site study. Quality of Life with:Living Situation, Family Relationships, Finances, Work & School Heal th (about self) Lehman, A., (1988). A quality of life interview for the chronically mentally ill. Evaluation and Program Planning 11 51-62. Life Events (about daughter) Monaghan, J. H., Robinson, J. O., Dodge, J. A. (1979). The ChildrenÂ’s Life Events Inventory. Journal of Psychosomatic Research 23 63-68. Religiousness/Spirituality (about self) Fetzer Institute & National Institute on Aging (1999). Brief mulitdimensional measurement of religiousness/spirituality for use in health research Kalamazoo, MI: Fetzer Institute. Service Needs and Use Self-developed. Hopes and Aspirations Self-developed. (about daughter) High Risk Behaviors (about daughter) Goodenow, C. (1999). Massachusetts Youth Risk Behavior Survey Massachusetts Department of Education. http://www.doe.mass.edu/lss/yrbs99/acknowledge.html Hess, J. C., & Rothgeb, J. M. (1999). Measuring the Impact of WelfareReform: Issues in Designing the Survey of Program Dynamics Questionnaire. Washington, DC: US Census Bureau. Family Relationships (FAD Version 3) Epstein, N. B., Baldwin, L. M., & Bishop, D. S. (1983). The McMaster family assessment device. Journal of Marital and Family Therapy 9: 171-180. Attitudes Toward Marriage Johnson, C. A., Stanley, S. M., Glenn, N. D., Amato, P. R. (2001). Marriage in Oklahoma: 2001 Baseline Statewide Survey on Marriage and Divorce. Stillwater, OK: Oklahoma State University.
ABOUT THE AUTHOR After a decade of work in South East Asia, teaching classical ballet and working for the 1988 Olympic Games as a liaison, Kather ine Best returned to the States and earned a BachelorÂ’s degree in Psychology from the University of South Florida. By 2002, she had completed the dual Masters in Soci al Work and Masters in Public Health. Winning an award for her outstanding thesis on costs and length of stay for substance exposed neonates she began working at the Florida Mental Health Institute under the direction of Dr. Robert Friedman and Dr. Rodger Boothroyd. Her areas of research have included risk and protective fact ors, resilience, ethics in re search, trauma exposures in children, and social needs for adults living in poverty. In 2004, as a licensed clinical social worker, Katherine founded the Encour agement Institute, a collaborative group practice in Sarasota treating individuals and families.