Citation
Sex, Drugs, and Violence: A Longitudinal Analysis of the SAVA Syndemic among Women Involved in the Criminal Justice System

Material Information

Title:
Sex, Drugs, and Violence: A Longitudinal Analysis of the SAVA Syndemic among Women Involved in the Criminal Justice System
Creator:
Acheampong, Abenaa
Publisher:
University of Florida
Publication Date:
Language:
English

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Epidemiology
Committee Chair:
COTTLER,LINDA B
Committee Co-Chair:
STRILEY,CATHERINE L
Committee Members:
GERKE,TRAVIS A
WHITEHEAD,NICOLE ENNIS

Subjects

Subjects / Keywords:
criminal
drugs
hiv
justice
sava
violence
women

Notes

General Note:
This dissertation enhances the understanding of the SAVA syndemic, which is the intersection of substance abuse (SA), violence (V), and HIV/AIDS (A) among a population that has been significantly underrepresented in this area of research-women of color in the criminal justice system. The focus of the research is on an NINR-funded randomized control study, Sisters Teaching Options for Prevention project (STOP) (N=319) (R01NR09180 PI: Cottler). This dissertation examines SAVA and its association with the Courtroom Behavior Checklist (CRBCL), an assessment that quantified court behaviors, as well as the use of novel and sophisticated analytical techniques to identify correlates of changes in SAVA and criminal offenses over time. Results showed that 1) the CRBCL may have added utility in identifying female offenders with recent substance use and risky sexual behaviors, 2) drug courts are associated not only with decreases in substance use, but also with HIV/AIDS risk behaviors and violence experienced over time, 3) crack/cocaine users with or without the mutual reinforcing issues of SAVA were less likely to change over time compared to those who did not use crack/cocaine, 4) recent crack/cocaine use and SAVA significantly increased the odds of offending with misdemeanors/municipal violations, however not felonies. Interventions aimed to reduce offenses in similar populations should consider periodic assessment of substance use- especially crack/cocaine use- and offer additional support for substance users. The results of this dissertation also suggest the need for targeted interventions tailored to crack/cocaine users, as well as a wide-spread need for trauma-informed interventions among females involved in the criminal justice system.

Record Information

Source Institution:
UFRGP
Rights Management:
All applicable rights reserved by the source institution and holding location.
Embargo Date:
12/31/2018

Downloads

This item has the following downloads:


Full Text

PAGE 1

SEX, DRUGS, AND VIOLENCE: A LONGITUDINAL ANALYSIS OF THE SAVA SYNDEMIC AMONG WOMEN INVOLVED IN TH E CRIMINAL JUSTICE S YSTEM By ABENAA ACHEAMPONG JONES A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FL ORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2016

PAGE 2

2016 Abenaa Acheampong Jones

PAGE 3

To my husband, William Jones

PAGE 4

4 ACKNOWLEDGMENTS Though the journey to obtain my doctorate was a difficult journey, it has been made possible by a support system comprised of family, funding agencies, mentors, friends, and well wishers. I first want to thank my absolutely wonderful husband William Jones for be ing an amazing husband, friend, prayer partner, and my biggest cheerleader. When I was involved in a serious accident in 2015, William my then fianc took an extended leave from his job in Washington, DC to come and be by my side. His love and commitment were so evident that my physical therapist, and many others, often reminded me that even if I never walked correctly again, my traumatic brain symptoms never resolved, and I remained in chronic pain, I had something that many people in my situation did not someone who would still marry them and love them regardless. Without his unwavering love, patience, and support through the trials of this PhD journey, I would not be where I am today. Next, I would like to thank my mother for instilling in me the value of education and her encouragement and prayers through all stages of my educational pursuit. I came from a background that was atypical from students generally considered to have bright futures. In spite of having good grades and being in advanced classes all throughout my formative years, I was often overlooked. However, my mother gave my sister and me no excuses as to why we could not have futures like the affluent students. So despite being told by a high school teacher that the only thing I should envi sion certainly not as a graduate from any type of institution of higher education I was able to graduate from my high school and was one of a few students in my demographic to attend college. Though it

PAGE 5

5 seems surreal at times, that I, a first generation college student whose grandparents could neither read nor write, would go on to have the highest educational degree possible, to my mother, this was a given. I also want to thank my sister Adjwoa, my pe rsonal vigilante, and my best friend, for always being by my side; as well as all my family as they all supported and pulled together all their resources for many, many, years so that I would be able to be in this position today. My deepest gratitude also goes to the Florida Education Fund (FEF), which provided me with the McKnight Doctoral Fellowship. At my first annual McKnight jacketing ceremony for the newly graduated McKnight Fellows, I listened as the new graduates many in tears stated the pivotal rol e FEF had in their success. I soon came to learn that FEF was not just a funding agency, but rather an invaluable family. Dr. Lawrence Morehouse, Charles Jackson, Phyllis Reddick, Lyra Logan, and all at FEF who go above the call of duty to make sure that t he fellows had everything they need to succeed from personal encouragement, seminars and workshops, travel awards, etc. I am also forever grateful for the supportive role FEF played when I was advised to either medically withdraw from school for some time or significantly reduce my school load following my accident. Upon hearing this news, FEF leadership along with Dr. Tyisha Hathorn and Sarah Perry from the Office of Graduate Minority Programs and Dr. Cindy Prins from my department, assisted me with my me dical withdrawal application and securing additional funding. Their assistance was invaluable as I had a traumatic brain injury that left me with slow mental processing for months and even left me unable to speak in complete sentences for several weeks, ma king the simplest of tasks never

PAGE 6

6 ending. Furthermore, I received periodic emails from Charles Jackson inquiring on my well being and recovery, which always made me smile. I would like to express gratitude to Dr. Linda Cottler, my research mentor and chair of my dissertation committee, and one of the most well known, productive, and seasoned experts in the field of substance use epidemiology. Though it is fairly rare for someone to be accepted into a PhD program straight from an undergraduate degree program, she chose me to be her student and I am very grateful for the time she invested in helping me throughout my time in this program. Her effort and investment in me has afforded many wonderful opportunities to develop professionally and as a researcher. More over, I have had the special opportunity to have access to rich data from her numerous NIH funded studies; these studies perfectly aligned with my research interests and gave me an advantage in productivity. In the 4 years under her tutelage, and although I lost a significant amount of time recovering from my accident, I have been able to write 8 first authored publications/manuscripts and 19 research presentations as a result of this advantage. I am also very appreciative of Dr. Travis Gerke; whose sheer brilliance motivates me to continue learning. Dr. Gerke consistently availed himself to answer any questions I had and I am especially grateful that though he left the University of Florida, he was steadfast in continuing his role as the expert statisticia n on my dissertation committee. I am very grateful to Dr. Catherine Striley for all her assistance and feedback not only throughout this dissertation process, but also on other manuscripts and presentations. I also have profound gratitude for Dr. Nicole Wh itehead for all her support, encouragement, feedback, and unwavering belief in my ability to succeed.

PAGE 7

7 Over the years there have been several individuals who have been integral to my journey that I would also like to acknowledge. The first being Dr. Typhany e Dyer, my former academic advisor who simply became family. I am very appreciative for her consistent follow up with me over the years as well as her invitations for me to collaborate on several of her manuscripts as a co author. Most importantly, her enc ouragement, her steadfast belief in my abilities and future success, willingness to journey easier. Secondly, I would like to acknowledge Dr. Fern Webb, a member of the McKn ight Doctoral Fellow family, for her support, kind words, and her readiness to assist me in any way possible. I would like to acknowledge my current academic advisor Dr. Lusine Yaghjyan as well as the PhD program director Dr. Cindy Prins for all of their a ssistance and encouraging words over the years. Special thanks also goes to the academic support team in the Department of Epidemiology: Tamara Millay, Betsy Jones, Erica Boyd, and Yolanda Rutledge. Thank you all so very much for your patience, assistance, kindness, and encouragement throughout the years. Aside from family and academic life, there are also those whose presence enriched my experience here in Gainesville. I especially would like to acknowledge my adopted family Kevin and Denyce Switzer and al l the children (Sydney, Alexa, Van, Carson, Danielle, Gabe, Levi, and Camille) whom all have brought me such joy. I will miss them terribly. I acknowledge my unbelievably selfless and amazing friend Marian Ankomah, who, though a graduate student herself, c ooked all my meals and took such good care of me during my accident and other times of profound sadness and

PAGE 8

8 frustration. I want to express gratitude to my dearly beloved former labmate Dr. Sonam Lasopa, whose kindness towards me during the entirety of my p rogram I will not forget and the staff at Starbucks on Archer Road who became an unusual but constant source of encouragement. I would also like to thank Kevin and Mati Gardner, Mekut Archibong, Kiara Pino, Abundant Grace Community Church, the staff at the Starbucks on H Street (NW DC), Vicki Osborne, senior McKnight Fellows: Dr. OluKemi Akintewe, Dr. Danielle Tolson, Dr. June Carrington, and a host of many friends and family who were an incredible support system and motivation during my journey. I also wou ld like to acknowledge all the participants and all those involved in the Sisters Teaching Options for Prevention study the data source for my dissertation Lastly I would be remiss as to not acknowledge key individuals during my undergraduate education who laid a foundation that either led me the to the PhD route or made it possible. First, I want to thank Dr. Ellington Graves then the assistant director of the Center on Race and Social Policy whom I worked under as a volunteer on his state funded p roject aimed to assess health disparities among African American Virginians. Working with Dr. Graves was truly mind changing as I learned about the impact of social inequality manifest in the disproportionate burden of disease among minority populations. S uddenly, all the issues I saw growing up made logical sense and this epiphany solidified my interest in research. Secondly, I would like to sincerely thank Dawit Lemma from University of Maryland's Office of Student Financial Aid. I was forced to transfe r to the University of Maryland College Park from Virginia Tech due to financial reasons. At UMD, these

PAGE 9

9 issues persisted and I was told by financial aid representatives that all available financial aid had been exhausted that I had no other choice but to drop out of college. Fortunately, I found Dawit, who was compassionate and found ways to help a rather distraught student find the financial resources to stay enrolled. Had I been forced to drop out of college, as I was told was my only option, I would not be in this position today. Thirdly, I would like to acknowledge the McNair Scholars program along with my faculty mentors Dr. Carolyn Voorhees and Dr. Stacey Daughters, for providing me with incredible undergraduate exposure to research as well as fundin g for national and international research. Fourthly, I will always be grateful for my amazing core group of friends and fellow McNair scholars at UMD who were an immense form of social support during my undergraduate experience. I specifically want to tha nk (Dr.) Dara Winley, Golda Amlalo, Matilda Decker, Dr. Mosopefoluwa Lanlokun, (Dr.) Delores Quasi Woode, Nasreen Jones, and Adeola Ajibola. During my final year at UMD, I could no longer afford housing, my family lived too far away, and I did not have rel iable transportation. These young women did not view me as a burden, but rather opened up their dorms and apartments to me so that it was possible for me to finish my remaining courses at UMD. I am always filled with gratitude when I remember Delores, who gave me her twin size dorm bed and instead slept on the floor night after night because she utterly refused to have me sleep on the floor. The selflessness of these young women was and still is humbling today.

PAGE 10

10 Last but certainly not least, I am forever gr ateful for the early mentorship of Dr. Mia Smith Bynum, who made me feel as if I could fly, that I could do all things and be all things, and that I was not simply adequate, but I was among the best. During some of the darkest days of self doubt and feelin gs of defeat during my PhD journey, I would always remember her words and belief in me; there are only very few days that I do not In conclusion, I would like to thank all those who played a par t in my success whether big or small. Without the academic and social support, I would not be where I am today.

PAGE 11

11 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ .......... 16 LIS T OF FIGURES ................................ ................................ ................................ ........ 17 LIST OF ABBREVIATIONS ................................ ................................ ........................... 18 ABSTRACT ................................ ................................ ................................ ................... 19 CHAPTER 1 INTR ODUCTION ................................ ................................ ................................ .... 21 The Criminal Justice System in the United States ................................ .................. 21 The Criminalization of Use and Possession Substances ................................ ........ 21 The Disparities in the Criminal Justice System ................................ ....................... 22 The Emergence of Drug Courts ................................ ................................ .............. 23 Drug Courts and Services for at Risk Women ................................ ........................ 25 Theoretical Framework ................................ ................................ ........................... 25 Syndemic Theory and SAVA Among Women in the Criminal J ustice System .. 25 The Social Ecological Model ................................ ................................ ............ 27 A Case for Gender Based Behavioral Interventions ................................ ............... 28 The Courtroom Behavior Checklist ................................ ................................ ......... 29 The Gaps in Knowledge ................................ ................................ .......................... 30 Public Health Significance and Speci fic Aims ................................ ......................... 31 Summary of Significance ................................ ................................ ........................ 33 2 GENERAL METHODS ................................ ................................ ............................ 35 Approac h ................................ ................................ ................................ ................ 35 Overview of the Sisters Teaching Options for Prevention (STOP) Study ......... 35 Outreach and Recruitment ................................ ................................ ............... 36 Interview and Intervention ................................ ................................ ................ 37 Measures ................................ ................................ ................................ ................ 38 The Courtroom Behavior Check List (CRBCL) ................................ ................. 38 The Washington University Risk Behavior Assessment (WU RBA) ................. 39 The Violence Exposure Questionnaire (VEQ) ................................ .................. 40 Criminal Justice Data ................................ ................................ ....................... 40 Definitions of Main Variables of Interest ................................ ........................... 41 Exposure to Violence ................................ ................................ ....................... 41 HIV/AIDS Risk Behavior ................................ ................................ ................... 41 Unprotected sex acts ................................ ................................ ................. 42 Multiple sex part ners ................................ ................................ .................. 42

PAGE 12

12 Risky partner characteristics ................................ ................................ ...... 43 Substance Use ................................ ................................ ................................ 43 SAVA ................................ ................................ ................................ ................ 43 Perceptions of Risky Sexual and Drug Using Behaviors ................................ .. 44 Criminal Charges ................................ ................................ .............................. 44 Analytical Techniques ................................ ................................ ............................. 45 Bivariate Analyses ................................ ................................ ............................ 45 Multivariable and Multivariate Analyses ................................ ........................... 45 Regression analyses using GEE ................................ ................................ 45 Latent Transitional Analyses ................................ ................................ ...... 46 Intent to Treat ................................ ................................ ................................ ... 46 Loss to Follow Up ................................ ................................ ............................. 47 Retrospective Power Analyses ................................ ................................ ......... 48 3 ORDER IN THE COURT? THE ASS OCIATION BETWEEN SUBSTANCE USE, EXPOSURE TO VI OLENCE, RISKY SEXUAL BEHAVIORS AND OBSERVED COURT BEHAVIORS AMONG FEMALE OFFENDERS ................................ ........ 59 Introduction ................................ ................................ ................................ ............. 59 Gender Differences in Population Growth of the Criminal Justice System ....... 59 Drug Courts and the Development of the CRBCL ................................ ............ 59 SAVA Syndemic and the Potential Correlation with CRBCL ............................ 60 Current Analysis ................................ ................................ ............................... 61 Methods ................................ ................................ ................................ .................. 62 Measures ................................ ................................ ................................ ................ 63 The Washington University Risk Behavior Assessment (WU RBA) ................. 63 The Violence Exposure Q uestionnaire (VEQ) ................................ .................. 63 The Court Room Behavior Check List ( Main Outcome Observed Court Behaviors) ................................ ................................ ................................ ..... 63 Main Exposure SAVA in the pas t 4 months Prior to Baseline .......................... 64 Violence ................................ ................................ ................................ ..... 64 HIV/AIDS risk behaviors ................................ ................................ ............ 64 Substance use (number of uses in the past 30 days). ............................... 65 SAVA Criterion ................................ ................................ ........................... 65 Covariates ................................ ................................ ................................ .. 65 Analysis ................................ ................................ ................................ .................. 66 Results ................................ ................................ ................................ .................... 66 Socio Demographic Characteristics ................................ ................................ 66 Exposure to Violence, HIV/AIDS Risk, and Substance Use at Baseline .......... 67 SAVA Among the Sample at Baseline ................................ .............................. 68 Severi ty of Substance Use, Violence Experienced, and HIV/AIDS Risk at Baseline ................................ ................................ ................................ ........ 69 Multivariable Models Assessing Behavior Specific Correlates of CRBCL Scores ................................ ................................ ................................ ........... 69 Multivariable Models Assessing SAVA and CRBCL Scores ............................. 70 Discussion ................................ ................................ ................................ .............. 71 Strengths and Limitations ................................ ................................ ....................... 72

PAGE 13

13 Conclusion ................................ ................................ ................................ .............. 73 4 SEX, DRUGS, AND VIOLENCE: A LONGITUDINAL ANALYSIS OF THE SAVA SYNDEMIC AMONG FEMALE OFFENDERS ................................ ........................ 86 Introduction ................................ ................................ ................................ ............. 86 SAVA Syndemic among Women in the Criminal Justice System ..................... 86 Syndemic Theor y and SAVA ................................ ................................ ............ 87 Gender Based Behavioral Interventions for Women in the Criminal Justice System ................................ ................................ ................................ .......... 88 Gaps in Knowledge ................................ ................................ .......................... 88 Methods ................................ ................................ ................................ .................. 89 Sisters Teaching Options for Prevention and a Case Management Intervention ................................ ................................ ................................ ... 89 Outreach and Recruitment ................................ ................................ ............... 90 Main Exposures ................................ ................................ ................................ 91 Main Outcome SAV A Over time (Baseline, 4 month and 8 month Follow Ups) ................................ ................................ ................................ .............. 91 Violence ................................ ................................ ................................ ..... 91 HIV/AIDS Risk ................................ ................................ ............................ 91 Substance Use ................................ ................................ ........................... 92 SAVA Criteria ................................ ................................ ............................. 92 Covariates ................................ ................................ ................................ .. 92 Analysis ................................ ................................ ................................ .................. 93 Multiple Imputations ................................ ................................ ......................... 93 Analysis Technique ................................ ................................ .......................... 93 Results ................................ ................................ ................................ .................... 94 Socio Demographic Characteristics ................................ ................................ 94 Exposu re to Violence, Substance Use, and HIV/AIDS Risk Over Time Among Sample ................................ ................................ .............................. 94 SAVA Among the Sample ................................ ................................ ................ 95 Multivariate Poisson Regressions ................................ ................................ .... 96 Discussion ................................ ................................ ................................ .............. 99 Strengths and Limitations ................................ ................................ ..................... 101 Conclusion ................................ ................................ ................................ ............ 101 5 ONE STEP AT A TIME: A LATENT TRANSITIONAL ANALYSIS ON CHANGES IN SUBSTANCE USE, EXPOSURE TO VIOLENCE, AND HIV/AIDS RISK BEHAVIORS AMONG FEMALE OFFENDERS ................................ .................... 113 Introduction ................................ ................................ ................................ ........... 113 Females in the Criminal Justice Syst em and Syndemic Theory ..................... 113 Trans theoretical Model Stages of Changes ................................ .................. 114 Gaps in Knowledge ................................ ................................ ........................ 115 Methods ................................ ................................ ................................ ................ 117 Sisters Teaching Options for Prevention ................................ ........................ 117 Outreach and Recruitment ................................ ................................ ............. 117

PAGE 14

14 Measures ................................ ................................ ................................ ........ 118 Main Exposures ................................ ................................ .............................. 118 Main Outcome Substance Use, Violence, and HIV/AIDS R isk (Indicator Items) ................................ ................................ ................................ .......... 119 Violence ................................ ................................ ................................ ... 119 HIV/AIDS Risk ................................ ................................ .......................... 119 Substance Use ................................ ................................ ......................... 120 Analysis ................................ ................................ ................................ ................ 120 Analytical Technique ................................ ................................ ...................... 120 Loss to Follow Up ................................ ................................ ........................... 121 Results ................................ ................................ ................................ .................. 121 Descriptive Statistics of Sample and Indicator Items ................................ ...... 121 Model Fit ................................ ................................ ................................ ......... 122 Item response probabilities ................................ ................................ ............ 122 Latent Status Transitions from Baseline to 4 Month Follow Up ...................... 123 Latent Status Transitions from 4 Month to 8 Month Follow Ups ..................... 124 Correlates of Latent Statuses at Baseline ................................ ...................... 124 Discussion ................................ ................................ ................................ ............ 125 Limitations and Strengths ................................ ................................ ..................... 128 Conclusion ................................ ................................ ................................ ............ 129 6 SUBSTANCE USE, VICTI MIZATION, HIV/AIDS R ISK, AND RECIDIVISM AMONG FEMALES IN A T HERAPEUTIC JUSTICE P ROGRAM ......................... 135 Introduction ................................ ................................ ................................ ........... 135 Substance Use Among Females in Criminal Justice System ......................... 135 The Emergence of Drug Courts ................................ ................................ ...... 135 Recidivism ................................ ................................ ................................ ...... 136 Gaps in Knowledge ................................ ................................ ........................ 137 Methods ................................ ................................ ................................ ................ 139 Outreach and Data Collection ................................ ................................ ........ 139 Comprehensive Criminal Justice Records and Recidivism ............................. 140 Recent Use of Crack/Cocaine and Other Substances: ................................ ... 141 Violence ................................ ................................ ................................ .......... 141 HIV/AIDS Risk Behavior ................................ ................................ ................. 141 SAVA Criteria ................................ ................................ ................................ 142 Covariates ................................ ................................ ................................ ...... 142 Analyses ................................ ................................ ................................ ............... 142 Results ................................ ................................ ................................ .................. 142 Recidivism ................................ ................................ ................................ ...... 142 Socio Demographic Characteristics ................................ ............................... 143 Adjusted Multinomial Regression ................................ ................................ ... 144 Discussion ................................ ................................ ................................ ............ 146 Limitations and Strengths ................................ ................................ ..................... 149 Conclusion ................................ ................................ ................................ ............ 150 7 GENERAL DISCUSSION ................................ ................................ ..................... 157

PAGE 15

15 Scope, Significance, and Aims of Dissertation ................................ ...................... 157 Baseline Correlates of Baseline C ourtroom Behaviors ................................ ......... 160 Correlates of Changes in SAVA Over Time ................................ .......................... 160 Latent Status of Women and Discrete Changes Over Time ................................ 161 Crack/Cocaine, SAVA, and Re offenses Over Time ................................ ............. 162 Implications of Findings ................................ ................................ ........................ 163 LIST OF REFERENCES ................................ ................................ ............................. 165 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 177

PAGE 16

16 LIST OF TABLES Table page 2 1 D efinition of SAVA in Analyses ................................ ................................ ........... 55 3 1 Description and Distribution of CRBCL and Scores ................................ ............ 74 3 2 Sample Characteristics of Participant s at Baseline (N=264) .............................. 75 3 3 SAVA Among the Sample at Baseline (N=264) ................................ .................. 78 3 4 Severity of Substance Use, Violence Experienced, and HIV Risk by SAVA Groups ................................ ................................ ................................ ................ 83 3 5 Adjusted Multivariable Negative Binomial Regression Assessing Number of SAVA Criterion Met and Baseline CRBCL Scores (N=264) ................................ 84 4 1 Baseline Socio Demographic Characteristics of Sample (N=319) .................... 102 4 2 Longitudinal Assessment of SAVA Components Over Time ............................ 104 4 3 Longitudinal Assessment of SAVA Component Criterion by Intervention Group ................................ ................................ ................................ ............... 108 4 4 Correlates of Substance Use, Violence, HIVAIDS, and SAVA Over Time ........ 109 4 5 Correlates of Substance Use, Violence, HIVAIDS, and SAVA Over Time Using Complete Data ................................ ................................ ....................... 111 5 1 Socio demographic Characteristics of Participants at Baseline (N=317) .......... 131 5 2 Descriptive Statistics of Variables in Latent Transition Analysis (LTA) ............. 132 5 3 Model fit information used in selecting the LTA model ................................ ..... 132 5 4 Item Response Probabilities of Indicator Items ................................ ................ 133 5 5 Transitional Probabili ties of Latent Statuses ................................ ..................... 133 5 6 Predictors of Latent Statuses (N=317) ................................ .............................. 134 6 1 Socio Demographic Characteristics of the Sample by C harges 8 Months Post Baseline (N=317) ................................ ................................ ...................... 152 6 2 Adjusted Multinomial Regressions Predicting Offense Patterns by 8 months (N=317) ................................ ................................ ................................ ............ 153 6 3 Correlates of Number of Felonies and Misdemeanors / Municipal Violations by 8 months (N=317) ................................ ................................ ............................ 155

PAGE 17

17 LIST OF FIGURES Figure page 1 1 Social Ecological Framework on the SAVA Syndemic ................................ ....... 34 2 1 Sex, Drugs, and Violence: A Longitudinal Analysis of the SAVA Syndemic among Female Offenders ................................ ................................ ................... 50 2 2 Sample Source for the STOP Study N=1,150 ................................ .................... 51 2 3 Recruitment Flow Chart for the STOP Study ................................ ...................... 51 2 4 Conceptual Model of Pap er 1 ................................ ................................ ............. 52 2 5 Conceptual Model of Paper 2 ................................ ................................ ............. 53 2 6 Conceptual Model of Paper 3 ................................ ................................ ............. 53 2 7 Conceptual Model of Paper 4 ................................ ................................ ............. 54 4 1 Flow Log on Attrition by Intervention Group (N=319) ................................ ....... 102 6 1 Patterns of Offenses 8 months Post Baseline (N=317) ................................ .... 151 6 2 Types of Offenses 8 months Post Baseline (N=317) ................................ ........ 151

PAGE 18

18 LIST OF ABBREVIATIONS C DIS Computerized Diagnostic Interview Sc hedule CRBCL Courtroom Behavior Check List DSM IV Diagnostic and Statistical Manual IVth Edition GEE Generalized Estimating Equations HIV Human Immuno Deficiency Virus LCA Latent Class Analysis LTA Latent Transitional Analysis NI AAA National Institu te on Alcohol Abuse and Alcoholism NIDA National Institute on Drug Abuse NINR National Institute of Nursing Research PPCMI Peer Partnered Case Management Intervention SAVA Substance Abuse, Violence, HIV/AIDS risk behaviors SI Standard Intervention ST I Sexually Transmitted Infections STOP Sisters Teaching Options for Prevention project STS Sister to Sister WTW Women Teaching Women WU RBA Washington University Risk Behavior Assessment VEQ Violence Exposure Questionnaire

PAGE 19

19 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy SEX, DRUGS, AND VIOLENCE: A LONGITU DINAL ANALYSIS OF TH E SAVA SYNDEMIC AMONG WOMEN INVOLVED IN THE CRIMINAL JUSTIC E SYSTEM By Abenaa Acheampong Jones December 2016 Chair: Linda Cottler Major: Epidemiology This dissertation enhances the understanding of the SAVA syndemic, which is the intersection of substance abuse (SA), violence (V), and HIV/A IDS (A) among a population that has been significantly underrepresented in this area of research women of color in the criminal justice system The focus of the research is on an NINR funded randomized control study, Sisters Teaching Options for Prevention project (STOP) (N=319) (R01NR09180 PI: Cottler) This dissertation examines SAVA and its association with the Courtroom Behavior Checklist ( CRBCL ), an assessment that quantified court behaviors, as well as the use of novel and sophisticated analytical tec hniques to identify correlates o f changes in SAVA and criminal offenses over time. Results showed that 1) the CRBCL may have added utility in identifying female offenders with recent substance use and risky sexual behaviors 2) drug courts are associated not only with decreases in substance use, but also with HIV/AIDS risk behaviors and violence experienced over time, 3) crack/cocaine users with or without the mutual reinforcing issues of SAVA were less likely to change over time compared to those who did not use crack/cocaine 4) recent crack/cocaine use and SAVA

PAGE 20

20 significant ly increased the odds of offending with mi sdemeanors/municipal violations, however not felonies. Interventions aimed to reduce offenses in similar populations should consider perio dic assessment of substance use especially crack/cocaine use and offer additional support for substance users The results of this dissertation also suggest the ne ed for targeted interventions tailored to crack/cocaine users, as well as a wide spread need fo r trauma informed interventions among females involved in the criminal justice system.

PAGE 21

21 CHAPTER 1 INTRODUCTION The Criminal Justice System in the United States I n recent years the need to reform the United States (US) criminal justice system has become ap parent. Currently, the US has the highest incarceration rates and correctional supervision (i.e. probation and parole) rates in the world (Austin et al. 2013; Tripodi & Pettus Davis, 2013). In 1971, the number of those incarcerated in state and federal pr isons wa s approximately 198,000; by 2014, this number increased to 1.6 million (Kaeble, Glaze, Tsoutis, & Minton, 2015 ; Tripodi & Pettus Davis, 2013). Moreover, when the nearly 750,000 individuals incarcerated in local jails are included, there are over 2. 2 million individuals incarcerated in the US (Kaeble, Glaze, Tsoutis, & Minton, 2015 ). Specifically, from 1980 to 2011 there has been a 370% increase of inmates in prison, 255% increase of probationers, 287% increase of parolees and 304% increase of jailed perso ns altogether translating to 7 million people involved in the criminal justice system by 2011 (Austin et al. 2013). Even though policy changes aimed to reduce the number of incarcerated individuals began in 2010, still 1 in 100 adult residents of the US are imprisoned and 1 in 36 are under some form of correctional supervision (Kaeble et al. Austin et al. 2013) Of notable importance, the annual cost per person in prison totals over $31,000, a sum which places a substantial financial burden on s tate and federal budgets (Henrichson & Delaney, 2012). The Criminalization of Use and Possession Substances The rise of incarceration in the US has been mainly attributed to delibe rate policies such as the War on Drugs, which began in 1971 (Tripodi & Pett us Davis, 2013). The War on Drugs instituted the criminalization of addiction and distribution of illicit

PAGE 22

22 substances in an effort to reduce substance use (Austin et al. 2013; Tripodi & Pettus Davis, 2013). The National Survey on Drug Use and Health report s that the rate of substance abuse and dependence was 35% among probationers while only 8% among those not on probation (DHSS, 2013). The Arrestee Drug Abuse Monitoring Program II (ADAM II), which tests arrestees for the presence of illegal subst ances amon g male arrestees, also corroborates the hig h prevalence of substance use among those in the judicial system (ONDCP, 2014). Findings from the 2013 report show that the proportion of arrestees testing positive for an y of the substances assessed ranged from 6 3% to 83%. Among female offenders, it has been estimated that 66% were arrested for substance use related crimes or for committing crimes while under the influence of substances (Tripodi & Pettus Davis, 2013). Overall, incarcerated women have elevated odds of having a substance use problem and are more likely than men to be incarcerated because of substance use related problems (Strathdee et al. 2015). The Disparities in the Criminal Justice System The increase in criminalization has also brought about d isparities in arrests and incarceration (Wendel, Drucker, Ostermann, DeWitt, & Clear, 2015; Austin et al. 2013; Gallagher et al. 2015). For example, affecte d populations often come from populations with low socioeconomic status, high rates of poverty and unemployment, under achieving schools, and subpar healthcare to name a few (Austin et al. 2013; Gallagher et al. 2015; ONDCP, 2014; Tripodi et al. 2013). Gender differences in rising incarceration and correctional supervision rates are also evident T hough only 7% of the prison population and 18% of those under correctional supervision are women, they are clearly be com ing the fastest growing population of inmates with their rate of increase nearly twice the rate of men (Rivera &

PAGE 23

23 Vesey, 2015; Tr ipodi & Pettus Davis, 2013). W omen are also more likely than men to be incarcerated in state judicial systems for a drug related crime (Rivera & Vesy, 2015). In 1971, there was a total of 6,329 women incarcerated in federal or state prisons; however, in 2010 the number rose to 112,797 (Tripodi & Pettus Davis, 2013). Racial disparities in arrests and incarcerati on have also resulted from the dramatic increase in incarceration and correctional supervision. Those most disproportionately affected by the criminal just ice system are under represented minorities (mainly African Americans), though studies show they are either less likely or as likely to use substances as their white counterparts (Austin et al. 2013). For example, in analyzing drug policy change in New Yo rk, Wendel and colleagues (2015) found that despite similar substance use rates among blacks and whites, 27% of the increase in the African American prison population was attributed to substance use while only 14% of the rise in the Caucasian prison popula tion was attributed to substance use. Other researchers have also f ound that the odds of arrests for African Americans at ages 17, 22, and 27 were a round 13%, 83%, and 235% greater than that for Caucasians -significant disparities which cannot be explain ed by drug use rates (Mitchell & Caudy et al. 2015). The criminalization of substance use and subsequent disproportionate enforcement of such laws, highlights the excessively punitive and racially charged criminal justice system and increases social inequ ality, leads to resentment and racial tensions, and question s the effectiveness and the legitimacy of the US judicial system (Austin et al. 2013 ; Pinard 2015 ). The Emergence of Drug Courts The failure of the War on Drugs, which brought about various soci o economic consequences, has led to the increased support of judicial diversion programs such as

PAGE 24

24 drug courts (Austin et al. 2013; Gallagher et al. 2015). Drug courts are an increasingly popular criminal justice intervention program aimed to better assist drug addicted populations and reduce incarceration rates by availing treatment to offenders as an alternative to incarceration (Austin et al. 2013; Gallagher et al. 2015; Tripodi & Pettus Davis, 2013). Research by Subramanian and Moreno (2014) found that between 2009 and 2013, 14 states passed legislation to expand drug courts and drug treatments. When compared with traditional methods of handling substance use related crimes (incarceration and correctional supervision such as probation), drug courts have been shown to be highly effective (Gallagher, 2013b). A part of therapeutic justice interventions, such as drug courts, allows offenders to recognize and acknowledge their illegal act and their need for treatment as well as realize how treatment can deter future criminal behaviors and potential dangerous behaviors (Lamb et al. 2014). Offenders must meet the legal requirements of the program, including regular attendance, agreeing to substance abuse treat ment, periodic progress reports, maintaining sobriet y, and other s (Lamb, Weinberger, & Gross, 2014). Specifically, Omjarrh et al. (2012) evaluated 92 drug courts and found that participants of drug courts had recidivism rates of around 38% to 50% less than non participants. Moreover, drug courts have been f ound to be cost effective when compared to the traditional mode of imprisonment. According to the National Association of Drug Court Professionals, every $1 invested in drug courts saves $3.36 in costs specifically related to the criminal justice system W hen including victimization and healthcare costs, research shows that savings could be up to $27 per $1 invested (NADCP, 2015).

PAGE 25

25 Drug Courts and Services for at Risk W omen Because women are one of the fastest growing prison populations and are more likely to face legal ramification s for substance use and possession, more focus should be placed on addressing the ir needs (Rivera & Vesey, 2015; Tripodi & Pettus Davis, 2013). Women in drug courts and women in the judicial system also have high rates of varied m ental health issues, including addiction, major dep ressive disorder, and posttr aumatic stress disorder (PTSD) (Harner & Riley, 2013). Though drug courts are shown to reduce criminal activity, individuals with mental health issues with co occurring substanc e use are less likely to successfully complete the program and achieve substance use abstinence (Sevigny et al. 2013; Mendoza et al. 2013; Peters et al. 2012). It has been found that that nearly 73% of women in prison have mental disorders and only 11% of those in need of substance abuse treatment actually receive care (Tripodi & Pettus Davis, 2013; Stevens, 2012). These issues highlight the great need for attention Theoretical Framework Syndemic Theory and SAVA A mong Women in the Criminal Justice Syst em Drug courts present a significant population of women with a history of substance use childhood and adult exposure to violence, and sex trading (Fulkerson, Keena, & 20 12). The intersection of substance abuse (SA), violence (V) and HIV/AIDS (A), intertwined, and mutually reinforcing health and social problems of substance use, violence, studies on the syndemic theory have well documented the link between illicit drug use

PAGE 26

26 and higher rates of risky sexual behaviors (Dyer et al. 2013; Islam et al. 2013; Chorba et al. 2012; McHugh et al. 2012; Patrick et al. 2012). For example, female substance users have been shown to have 2.72 times the odds of having concurrent partnerships, and nearly 10 times the odds of engaging in sex trading than non substance using women (Ad imora, Schoenbach, Taylor, Khan, Schwartz, 2011; Tetrault et al. 2010, Martin et al. 2010). Furthermore, these risks are exacerbated in women who are low income, homeless, and lack financial and social support (Blankenship, Reinhardt, Sherman, El Bassel, 2015; Peters et al. 2012). Women involved in the criminal justice system are known to have significantly elevated rates of SAV A compared to the general population of women (Meyer et al. 2011; Elkington et al. 2008; Harner & Riley, 2013; Roth et al. 201 2). In addition to significant disparities in health and risk behaviors among women in the general population, e ven in the criminal justice system, significant disparities in health and risk behaviors exist between men and women. Compared to men in the ju dicial system, women are not only more likely to have a history of substance abuse and dependence but also are more likely to have a history c hildhood trauma and homelessness. The latter factors have also been linked with SAVA related items such as multip le sex partners, increased unprotected sex, exposure to violence, and sex trading (Senn, Carey, & Coury Doniger, 2011; Messina, Grella, Cartier, & Torres, 2010; Klein, Elfison, and Sterk, 2008). For example, in a study of female sex trade rs, Surratt et al. found that age of first abuse (physical or sexual) preceded first sex trade in 70% of the women (Surratt, Kurtz, Chen, & Moss, 2012). Overall, SAVA has been shown to be a significant barrier to HIV/AIDS prevention particularly in women since women

PAGE 27

27 compris e 31% of new HIV infections transmitted through heterosexual contacts (Meyer, Springer, & Altice, 2011). The Social Ecological Model Another framework that explains SAVA among female offenders is the social ecological model (McLeroy, Bibeau, Steckler, & G lanz, 1988). According to this model, there are 4 levels that interact with or influence behaviors: the individ ual level (personal characteristics and beliefs), the interpersonal level (individuals and their relationships with family, peers, and other s), t he community level (how community factors can impact individuals and their behaviors), and the societal level (social policies that impact individuals and behavior) (Figure 1.1) Regarding the individual level personality characteristics such as sensatio n seeking, poor impulse control, and risk taking have all been linked with SAVA, especially in those involved in the criminal justice system (Weir & Latkin, 2014). At the interpersonal level the criminal justice system may contribute to HIV risk behaviors by disrupting stable social networks and economic situations, and by destabilizing intimate relationships that are known to decrease high risk sexual behaviors, such as concurrent and multiple sexual partners (Epperson et al. 2010; Freudenberg, 2009; Ple fieger et al. 2013; Sharpe et al. 2012). Community level factors such as social norms, peer groups, and high intensity drug use areas are also linked with higher rates of drugs and crimes all of which have been linked to HIV/STI clusters and their subse quent risk behaviors (Jennings et al. 2013; Sharpe et al. 2012; Tripodi et al. 2013). Specifically, researchers have found that high intensity drug areas, which are marked by everyday violence, have created

PAGE 28

28 especially deleterious conditions for women as they restrict movement and perpetuate subordination to drug subculture (Blakenship, Reinhard, Sherman, El Bassel, 2015). Lastly, societal level factors such as policies that lead to residential segregation (and subsequently to the concentration of poverty in various geographical districts), the War on Drugs, and the targeted marketing of psychoactive drugs, have led to a dramatic increase of incarcerated individuals, and SAVA (e.g. easily accessible illicit drugs, crime and victimization, prostitution and risky sexual networks) (Adimora et al. 2005; Jennings et al. 2013; Sharpe et al. 2012; Tripodi et al. 2013). Studies of SAVA, in the contex t of the alternative to incarceration programs such as drug courts may provide a special opportunity to examine th e effect of policies aiming to mitigate the negative effects of the War on Drugs. A Case for Gender Based Behavioral Interventions Gender based research studies and interventions, specifically those for women with comorbid HIV risk and drug use, are spars e. For example, in a systematic review examining HIV structural interventions among female injection drug users, Blankenship et al. found that of 25 studies that examined the risk environment, only 7 considered gender and these mainly consisted of includin g the variable in a multivariable regression (Blankenship et al. 2015). Gender based research and interventions are especially needed in the criminal justice system as it contain s a subpopulation of high risk women different from the general population of women and their male counterparts. Because incarcerated females report higher rates of drug and sexual related risk behaviors as well as earlier initiation of these behaviors (multiple sex partners, sex while under the influence, and sex trading) than in carcerated males, effective behavioral interventions are needed among this population However, drug and behavioral

PAGE 29

29 interventions in the judicial systems are mainly male oriented ; consequently, female recidivism have been linked with the lack of female ori ented intervention s (Stewart & Gobeil, 2015; Tripodi & Pettus Davis, 2013; Messina et al. 2010). In addition, t here is also a need for gender specific health behavior interventions that provide social support and access to social services for female offen ders, especially considering the significantly higher rates of negative life events (e.g. lifetime trauma, homelessness) found in this population (Blankenship et al. 2015; Binswanger et al. 2010; Messina et al. 2010). Prevention studies need to address the often ignored STI prevention services amongst drug court enrollees (Robertson, St. Lawrence, & McCluskey, 2012). In addition, interventions providing practical support such as transportation, skills acquisition, a nd other support are usually lacking f or drug court enrollees (Stewart & Gobeil, 2015; Peters et al. 2012). A qualitative study of former drug court enrollees found that the women believed that social support especially from other females who formerly used drugs, were vital to successful com pletion of the drug court program (Fischer, Geiger, & Hughes, 2007). The Courtroom Behavior Checklist Success in drug court is based on court officials and partnered health ve behaviors, such as abstaining from illicit drugs and criminal behaviors. To date, few assessments exist that serve as a proxy for conforming to norms, a provision which can be useful in delineating women in drug c ourt who may be at high risk for recidiv ism and termination and may need a more specified and intense intervention for behavior ch ange The Courtroom Behavior Checklist (CRBCL), an assessment developed for

PAGE 30

30 STOP and used over 7,000 times in the Municipal Court system, assessed readiness for court in women at Drug Court. The CRBCL assessment measured items that occurred before or has documents about progress that occurred during court suc for change. A prior analysis evaluating the association between high CRBCL scores (indicating unacceptable behavior) and recidivism (defined as a new misdemeanor felony arrest, or new municip al violation ) found that the women with higher CRBCL scores had 2.84 higher odds of re arrest than women with lower scores (Reingle et al. 2012). Since many female offenders are arrested for drug related charges and often have high rates of victimization, risky sexual behavior, and mental health problems that include substance related issues the CRBCL could capture awareness of how to gui de women and researchers for future drug and STI interve ntions in high risk populations. The Gaps in Knowledge Because drug courts are becoming increasingly popular, more research on drug courts and other interventions aimed to reduce substance use and its co occurring behaviors are needed. In a recent review of the literature, El Bassel and Strathradee (2015) identified several specific questions that need to be addressed in the field: How prevalent is drug use among certain subpopulations? How prevalent are co occurring disorders such as HIV, violence, etc. among women who use drugs? How do the institutions of race, e thnicity, and class interact with gender to heighten risk among drug using women? What interventions will reduce HIV incidence for women who use drugs and have co occurring issues such as violence, etc? Specifically, they called for epidemiologic studies among women in alternative to incarceration programs and incarcerated women along with epidemiological studies on

PAGE 31

31 the SAVA syndemic (El Bassel & Strathdee, 2015). Deering and colleagues (2014) also argued for epidemiological studies on the prevalence and correlates of violence among sex workers s pecifically, on the magnitude, type, and severity of violence Public Health Significance and Specific Aims Each year in the United States, illicit drug use costs $193 billion in costs related to crime, producti vity, and about 19 million people are diagnosed with HIV/STIs (CDC, 2013; NIH, 2015). Moreover, nearly 7 million residents of the US are either incarcerated or under correctional supervision, which further constrains budgets on all levels of government (Au stin et al. 2013; Gallagher et al. 2015; Wendel et al. 2015). An analysis of the economic costs of the criminal justice system found that among the 40 states in which complete data was available, tax payers spent nearly 40 billion dollars, with annual c osts per inmate ranging as high as $50,000 $60,000 in states with higher costs of living ( Henrichson, & Delaney, 2012). Along with the insurmountable costs, t he traditional mode of imprisonment for addiction and risky sexual behaviors such as sex trading, rather than treatment for addiction and improved access to social services for impoverished women, have been Peters, Kremling, Bekman, & Caudy, 2012). Moreover, s ubstance us ers are known to have higher odds of committing other crimes which is a threat to public safety (Zhou et al. 2012; Vaughn et al. 2010). Longitudinal research on SAVA among women in the criminal justice system is paramount to not only advancing the field of substance use and it s co occurring health and social consequences but also for the development of cost effective interventions that will improve public safety (Tetrault et al. 2010). Thus, the specific aims of this dissertation were to:

PAGE 32

32 1. Assess the as sociation between self reported SAVA leading up to baseline and and controlling for socio demographic characteristics. 2. E valuate the longitudinal trends of the SAVA syndemic over a n 8 month period among female offenders a) Evaluate the association between a peer partnered case management intervention (PPCMI) and a standard intervention (SI) on the likelihood of reduction of SAVA over time b) Assess the strength of relationships between violence, substance use, and HIV/AIDS risk by assessing the effect of the initial prevalence of these issues on longitudinal outcomes. c) Determine the effect of race, markers of socio economic status such as education and stable housing, and observed court behaviors at baseline on SAVA over time. 3. Explore latent status of women based on substance use, exposure to violence, and risky sexual behaviors at baseline, 4 months, and 8 months a) Identify the proportion of individuals in each latent status at the base line, 4 month follow up, and 8 month follow up and the probabilit y of transitioning to lower status over time. b) Asse ss the effect of the PPCMI intervention versus the SI on latent status transitions. c) Evaluate differences in the association between socio de mographic chara cteristics, child sexual abuse, perception of drug use and initial latent status membership. 4. Examine the association between crack/cocaine use at baseline and any felony, misdemeanor, and municipal violations at an 8 month follow up. a) Asse ss the association between SAVA (any substance use, being exposed to violence, and having HIV/AIDS risk behaviors) at baseline and any felony, misdemeanor, and municipal violations at an 8 month follow up. b) Determine the association between crack/cocaine u se at baseline and the number of felony, misdemeanor, and municipal violations at an 8 month follow up. c) Explore the association between SAVA at baseline and the number of felony, misdemeanor, and municipal violations at an 8 month follow up.

PAGE 33

33 Summary of S ignificance 1. Currently, the US has the highest incarceration rates and correctional supervision (e.g. probation and parole) rates in the world (Austin et al. 2013; Tripodi & Pettus Davis, 2013). 2. The failure of the War on Drugs, which brought about various socio economic consequences has led to the increased support of judicial diversion programs such as drug courts (Austin et al. 2013; Gallagher et al. 2015). 3. Women are one of the fastest growing prison population and are more likely to face legal ramifica tion for substance use and possession (Rivera & Vesey, 2015; Tripodi & Pettus Davis, 2013). 4. mutually reinforcing health and social problems of substance use, violence, and H 5. When compared to men in the judicial system, women have been shown to have a greater burden of addictions as well as childhood trauma and abuse, and homelessness which has also been linked SAVA (Senn et al. 2011 ; Messina et al. 2010; Klein et al. 2008). 6. The Courtroom Behavior Check List (CRBCL) which is used to record the behaviors of the women in court may be used as a proxy for conforming to norms (Reingle et al. 2012). 7. Since many female offenders are arre sted for drug related charges and often have high rates of victimization and risky sexual behavior, SAVA may be associated with the CRBCL in women in Drug Court. 8. Studies examining the prevalence of SAVA and correlates of change s in SAVA over time among wom en in the criminal justice is needed.

PAGE 34

34 Figure 1 1. Social Ecological Framework on the SAVA Syndemic

PAGE 35

35 CHAPTER 2 GENERAL METHODS Approach Overview of the Sisters Teaching Options for Prevention (STOP) Study The Sisters Teaching Options for Prevention (S TOP) study was derived from two prior intervention studies, Women Teaching Women (WTW, N=501, DA 11622 PI: Cottler) funded by the National Institute on Drug Abuse (NIDA) and Sister to Sister (STS, N=348, AA12111 PI: Cottler) funded by the National Institut e on Alcohol Abuse and Alcoholism (NIAAA). These studies aimed to evaluate and change substance use and risky sexual behaviors among community recruited women ( Nurutdinova, Abdallah, Bradford, O'Leary, & Cottler, 2011 ; Osborne & Cottler, 2012; Johnson, Cot tler Abdallah, ; Vaddiparti, Striley, & Cottler, 2016 ; Johnson, Cottler, Ben Abdallah, & O'Leary, 2012 ) intravenous drug use or a positive urine drug test indicating recen t use of cocaine, opiates, or amphetamines. The STS study was comprised of women who were heavy drinkers or had problem drinking as determined by the Alcohol Use Disorders Identification test (AUDIT) ( Bohn, Babor Kranzler, 1995 ). Additionally, recruited w omen for either study must not have undergone substance or alcohol abuse treatment in the past 30 days prior to study enrollment. All women in these studies received the NIDA HIV pre and post zed to receive a Well Woman Examination with or without 4 sessions of risk reduction counseling. The results of these studies suggested that among these women, offenders were the least likely to change their behaviors. In response, a gender specific behav ioral

PAGE 36

36 intervention aimed at female offenders was developed using a peer partnered approach, whereas peer partners, who were previously in the court system themselves, can serve as positive examples. The specific aims f or the National Institute of Nursing R esearch (NINR) funded STOP study (R01NR09180 PI: Cottler) were as follows: 1. Adapt a culturally relevant, gender specific, community based, theoretically driven Peer Partnered behavioral intervention for a randomized clinical trial (RCT) to reduce HIV risk b ehaviors, and facilitate access to needed services and research protocols. 2. Reach a difficult to recruit population of women female offenders in need of HIV/STD testing, counseling, and medical and behavioral interventions. 3. Enroll these women into this RCT and retain them with high response rates, comparing a standard intervention to a Peer Partnered Case Management Intervention (PPCMI). 4. Assess the effectiveness of the PPCMI at 4 and 8 months to facilitate access to needed services, to reduce barriers to s ervice access, to reduce high risk behaviors, to increase knowledge of HIV and other STDs, and to improve trust in and understanding of research involvement. Outreach and Recruitment For the STOP study research staff members were present at the Municipal Court System of St. Louis, Missouri and completed a CRBCL for every woman seen in court (Figure 2.1) The sample source of STOP participants came from 1,150 individuals, mainly from city drug courts (78%), state drug courts (12%), and 10% came from other types of courts or centers. Of these individuals, 640 or 56% were eligible for the STOP study (Figure 2.2). To be eligible, women had to be at least 18 years of age must intend to remain in the St. Louis area for the next 12 months, and have no known cogn itive issues The total sample size consisted of 319 women who completed their baseline interviews and were randomized for the intervention (Figure 2.3).

PAGE 37

37 In terview and Intervention Interested, eligible women were scheduled for two baseline interviews : 1) a review of the informed consent, study commitment form, collection of urine for drug use, chlamydia, and gonorrhea testing, and an assessment of substance use and risky sexual behaviors. Next, women underwent pre HIV counseling with a short educational se ssion on how to reduce risky sexual behaviors and drug use. After that, blood samples were drawn for HIV, HCV, and syphilis. 2) Two weeks later, women returned for an assessment of psychiatric disorders, health services utilized, and history of exposure to violence After thi s round of assessments, women received the results of their STI testing along with posttest HIV counseling. Women were also given details on necessary treatment options and services they could access. After this, the women were randomiz ed to either the standard intervention (SI), in which nothing else was required, or the peer partnered case management intervention (PPCMI). The SI included the standard National Institute on Drug Abuse pre and posttest counseling. Out of 319 women, 155 we re randomized to the SI, while 164 were randomized to the PPCMI. In the PPCMI group, women could receive 40 hours of case management over a 10 week period with a peer partner. They also had the option to utilize assistance from the ir peer partner to apply for medical assistance, govern ment aid, parenting classes, or GED training. Peer partners also provided transportation to services assigned by the judge. While being transported, the women watched a series of DVDs in the van on safe sexual behaviors and ge neral education regarding health.

PAGE 38

38 Measures The Courtroom Behavior Check List (CRBCL) STOP research staff members, who were not members of court, were present at c ourt to complete the CRBCL, which was developed to evaluate the disposition o f drug offendin g women seen at drug c ourt. The assessment measured attendance, demeanor, and behavior in Drug Court. The scores ranged from 0 (indicating favorable b ehavior), to 35 (for an unexcused absence from the Court). The 12 items assessed are as follows : Items 1 6 were assess ed before court while items 7 12 were assessed during court. 1. Present at court: who were not present when summoned received the hi ghest score of 2. Under the influence of drugs or alcohol: Women who did not appear to be under the woman who received a comment f rom the court regarding being under the influence 3. Cell phones/pagers turned off : Women who were observed to have their cell phones devices turne d off received a score of 4. Has notes about progress: Women with notes about their progress received a score 5. Has documents about progress: Women were given vario us tasks to complete to assi st them in their rehabilitation, therefore documents attesting to their progress in completed these tasks were pivotal. Participants who had documents about their 6. Is alone: children to court were given a 7. Disruptive in court: those who received court comments on their disruptive talking received a score of

PAGE 39

39 8. Courteous to court staff: Women who were observed to be courteo us to the court staff received who were not courteous to court staff were 9. Interrupted the judge: Women who did not interrupt the judge while the judge was score of research staff) or 3 (if received a court comment). 10. Prepared to take notes on required tasks: Women who were prepared to take notes point. 11. Responded to judge appropriately: Women who responded appro priately to the women who did not respond to the judge appropriately ( as observed by STOP research staff ) women who received a comment by the c 12. Appeared to have confident demeanor: Women who appeared to have a confident demeanor (as observed by research staff ) did not appear confident rec Many of the items on the CRBCL could be objectively observed, such as having appropriate documentation. For items that were subjective, extens ive trainings were held among research staff to ensure uniformity in judgments and improve r eliability (Reingle et al. 2012). Moreover, since the women were in court more than once the CRBCL score prio r to the baseline was used for analysis, which is consistent to how the scores have been previously analyzed. Using the CRBCL score closest to wo baseline assessments was done in an attempt to capture the best representation of their observed court behaviors at the beginning of STOP. The Washington University Risk Behavior Assessment (WU RBA ) The WU ssment, assessed risky behavior including risky sexual and drug using behaviors, and demographic information (Shacham & Cottler 2010 ; Needle et al. 1995 ). Risky sexual behaviors were assessed

PAGE 40

40 for the past four months and they included: sex trading (for fo od, cash, or drugs), sex acts and number of times protected, and number of sexual partners. A study that examined the psychometric properties of this WU RBA found support for that this assessment was reliable (Shacham & Cottler 2010). All are described in detail below. The Violence Exposure Questionnaire (VEQ) The V iolence Exposure Q uestionnaire developed from the Conflict Tactics Scale, assess ed various forms of current and past violent behaviors and abuse including: child sexual abuse, holding weapons, and physical and emotional abuse (Strauss, 1979) Criminal Justice Data A rrest data were collected utilizing three Criminal Justice Data Banks. Access to these official data banks was made possible due to a partnership between the Washington University inv estigators and Judge James Sullivan, the Presiding Judge of the St. Louis City Municipal Court (Reingle et al. 2012) Summons and arrest files on local municipal violations were derived directly from the Regional Justice Information System (REJIS) which, in partnership with the government entity the Bureau of Justice Statistics, is a leading source of criminal justice data and is used for current criminal justice related analyses (Frandsen, Naglich, Lauver, & Lee, 2013 ; Reingle et al. 2012 ). The MULES (M issouri Uniform Law Enforcement System) arrest file database was used to garner information regarding misdemeanor and felony arrests occurring in Missouri ( the state in which the STOP study occurred). T he National Crime Information Center (NCIC) Interstate Identification Index (Triple I) was used to extract data on arrests that occurred outside the state of Missouri. Additional information ( e.g. demographic identifiers used to associate arrest records and individuals) w ere collected from the

PAGE 41

41 REJIS Municipal Court files and manual files which is maintained by the City of Saint Louis Probation Office. Furthermore, the Washington University research staff developed a data abstraction form which ensure d that arrest data were properly separated by follow up time (Reingle et al. 2012). Definition s of Main Variables of Interest Exposure to V iolence Overall, participants were considered as having experienced violence in the past 4 months following quest ions: 1. the past 4 months 2. the past 4 months has anyone pressured or forced you to participate in 3. the past 4 months has anyone abused you emotionally, that is, did or said 4. the past 4 months has anyone hurt you to the point that you had bruises, 5. the past 4 months has anyone attacked you w ith a knife, s tick, bottle, or HIV/AIDS Risk Behavior Overall, participants were considered as having HIV/AIDS risk behavior if they had a risky partner (an injection drug user or has other partners simultaneously) or multiple sex partner s with at least 1 unprotected sex act. This rigorous definition has been used in prior studies to capture those who are truly at risk for HIV/AIDS and other STIs (Meyer, Springer, & Altice, 2011). Unprotected sex acts, risky partners, and multiple sex part ners are defined below:

PAGE 42

42 Unprotected s ex acts The overall number of unprotected sex acts w as derived from the sum of all unprotected sex acts (any unprotected oral, vaginal, and anal) : 1. Unprotected Oral Sex: To assess unprotected oral sex, participants wer e asked : the past 4 months approximately how many times did you give oral sex, or a blow job, to a man, where he put his penis in your mouth? Participants were Number of unprote cted oral sex acts was calculated by subtracting the number of times a male condom was used from the number of times oral sex was given. For example, if a woman reported 4 instances of oral sex and in 3 of those instances a male condom was used, she was co nsidered to have 1 instance of unprotected oral sex. 2. Unprotected Vaginal Sex : The items used to assess unprotected vaginal sex the past 4 months approximately how many times did you have vaginal of unprotected vaginal sex acts with a man was derived from adding the total number of times protection was used and subtracting this number from the numb er of vaginal sex acts engaged in. For example, if a woman reported 7 instances of vaginal sex and in 3 of those instances a male condom was used and in 2 of those instances a female condom was used, she was considered to have 2 instances of unprotected va ginal sex. 3. the past 4 months how many times did you have sex where a man put his penis into your b utt, also condom during ana was calculated by subtracting the number of times protection was used from the total number of anal sex acts engaged in. For example, if a woman reported 3 instances of anal sex and in 2 of those instances a male condom was used and in 1 of those instances a female condom was used, she was considered to have no instance of unprotected anal sex. Multiple sex p artners Participants were as ked the past 4 months how many different people have Participants with 2 or more sex partners in the past 4 months were considered to have had multiple sexual partners.

PAGE 43

43 Risky partner c haracteristics Particip since (date 4 month the former question or 1+ for the latter question were considered to have had a risky partner. Substance Use Overall, substance use was the summation of the number of use s for all drugs as well as a dichotomous yes/no variable indicating any subs tance use in the past 30 days The number of drug uses in the past 30 days was assessed using the WU RBA. Number of uses of marijuana, stimulants (speed, amphetamines), crack, cocaine, and heroin was calculated for each participant by multiplying the numbe r of uses of specific drugs (how many days have you used (drug) by how many times a day each drug was used). For example, if a participant reported that she used crack/cocaine 3 days in the last 30 days, and on each of those days she used crack/cocaine tw ice, her total number of uses of crack/cocaine was 6. If this participant also reported that she used marijuana 7 days in the last 30 days, and on each of the days she used about once a day, her total number of uses of marijuana was 7. If she did not repor t the use of any other drug, her combined number of uses for any drug was 13 in the last 30 days. SAVA Though there we re a number of items in each SAVA component, these items were simplified (Table 2 1) To be considered to have experienced violence in the past 4 months participants must have reported experiencing at least one or more of the violence items within the given time frame. Having HIV/AIDS risk behavior was

PAGE 44

44 categorized as having a risky partner or having 2+ sex partners, AND having 1+ unprot ected sex acts within the given time peri od. Lastly, meeting the criterion for substance was defined as using any substance for one or more days in the past 30 days Overall, there were f our levels in the SAVA variable, component criteria met), Perceptions of Risky Sexual and Drug Using Behaviors The Social Ecological M odel states that individual level fac tors influence behavior; however, an understudied area regarding SAVA is the risk perceptions of high risk women (shown previously in Figure 1.1) In this analysis we have the opportunity to use informative and novel questions that assess the perceptions of combined sexual risk and drug using behaviors among high risk women who may also be involved in sex trading. The questions from the WU RBA that will be used are: have risky sexual b ehaviors that need changing and se behaviors tha t need changing. disagree. All of these questions were dichotomized as agree or do not agree/neutral Criminal Charges C omprehensive criminal justice records, along with the abstraction form, were used to evaluate whether women were charged with a felony, misdemeanor, or municipal violation in the period between base line and 8 month follow up. T his dissertation evaluates types of criminal charges (felony, misdemeanors, municipal violatio ns) and the number of charges.

PAGE 45

45 Analytical Techniques Bivariate Analyses To evaluate the bivariate associations between independent variables and outcomes of interest, either Wilcoxon signed rank sum test (for continuous variables) or Kruskal Wallis analys es (categorical variables) were used for count/ordinal outcome variables of interest (e.g. CRBCL scores) Wilcoxon signed rank sum test and Kruskal Wallis tests are non parametric tests which assume that the difference between two variables are ordinal (UC LA, 2016). For categorical independent and outcome variables of interest, chi square analyses, an appropriate measure of association for such variables, were used. Multivariable and Multivariate Analyses Regression analyses u sing GEE For the longitudinal analysis component, multivariate Poisson regressions, using generalized estimating equations (GEE) were used to determine any significan t changes in SAVA over time. GEE models were used for longitudinal analyses, using an apriori correlation matrix to adju st for dependency of observations over time, which allowed for various types of regressions on repeated measures (Liang & Zeger, 1986). The main assumption with GEE is selecting a correlation matrix, however, GEE is still fairly robust against choosing an incorrect correlation matrix. For this data, the working correlation matrix that was most feasible was autoregressive, which assumed that the correlation between repeated measurements decreased over time. GEE models first compute a standard regression, whi ch assumed independence of observation, then adjusted the coefficients based on the correlation. Since changes in behavior are often known to be pronounced short term and tend to wane as time progresses, the effect of

PAGE 46

46 time in these analyses were assessed n on linearly to account for this common phenomenon. Moreover, the fit statistics of the multivari ate model s was examined to ensure acceptability Latent Transitional A nalyses T he longitudinal extension of a latent class analysis (LCA), the latent transit ional analysis (LTA), w as used to quantify SAVA among latent subgroups of women This is a person centered approach for modeling behavior that classifies multiple dimensions of behavior to aggregate individuals with common behavioral patterns (Rhoades, Gre enberg, Lanza, & Blair, 2011). In this analysis, our indicator items reflect an overarching latent variable of conforming to socially normative behavior, as in the reduction of SAVA items. The appropriate number of latent s ubgroups was chosen using fit sta tistics the proportion of individuals within each latent status, and the interpretability of the statuses. In this model 3 sets of parameters were estimated : latent status membership probabilities, transitional probabilities, and item response probabiliti es. The latent status probabilities estimate d the proportion of individuals who were likely to belong to each latent status at each time interval Transitional probabilities estimate d the likelihood of changing one latent status to another latent status at the next follow up period. Lastl y, item response probabilities estimate d the agreement of the specific indicators of the latent variable and latent status membership. Intent to Treat The STOP study featured many strengths such as: an innovative interventi on, a relatively large sample size and minimal loss to follow up given t he sample population (Baseline n = 319, 4 month n= 261, 8 mo nth n = 282), a rich dataset with detailed measures of SAVA components, multiple time points, as well as the innovative CRBCL.

PAGE 47

47 However, this study is not without limitations. The purpose of the STOP study wa s to evaluate whether women receiving the PPCMI would be more likely to have positive outcomes, including a reduction in risky sexual and drug use behaviors than women receivi ng only the SI. A potential problem was that women randomized to the PPCMI group may not have utilized all possible intervention elements Though this is a main limitation for all randomized controlled studies, it may have be en especially true for studies in which the study sample is comprised of those with substance use related problems Because of this, our analyse s w ere based on intent to treat analyses which give a more conse rvative estimate of the effect of PPCMI, if statistically significant changes are observed. Loss to Follow Up Missing data is also common and usually an unavoidable problem that plagues analyses especially longitudinal study designs. The seriousness of missing data depends on the pattern of missingness. Regarding this dataset the re were both legitimate missing and illegitimate missing data. For example, there were several skip patterns, where, if a participant d id not endorse engaging a specific behavior, subsequent questions regarding that specific behavior were appropriately ski pped. However, there were some illegitimate skip s ranging from non responding to skip errors Due to frequent and regular quality control efforts, errors and missing data were rare. In fact, in the baseline assessments, there were only 2 instances of this type of missing data. The most common form of missi ngness in the STOP study was due to l oss to follow up, and missingness with the CRBCL scores. There were some missing CRBCL scores due to the fact that some women were never in court during the time STOP r esearchers were there, or they were excused from court by the judge.

PAGE 48

48 Common ways to deal with missing data include the deletion method which can lead to a loss of precision and power depending on the size of the sample. This can lead to larger p values, standard errors, and confidence intervals. Other methods of addressing missing data can be single imputation methods such as mean substitution and single regression imputation. Though these methods allow for the use of all data, they reduce variation in da ta thereby compromising the validity of the estimates. A more sophisticated approach to handling missing data is through multiple imputations, a statistical technique that substitutes missing value s with probable values that account for the uncertainty abo ut the precise value to impute (Yaun, 2010) This method uses all available data, thus preserves power and is a good estimation of standard errors provided by repeated imputation and f or this reason, multiple imputations will be used to address missing dat a in this dissertation Retrospective Power Analyses Our analytical methods include d several GEE based Poisson regressions. Using PASS 14, the minimum detectable difference for a Poisson regression of a dependent variable of counts on a normally distribut ed (mean=0, standard deviation=1) independent variable using a sample si ze of 319 observations, achieved 80% power at a .05 alpha to detect a response rate ratio of at least 1.115, with no correlation between the covariate of inter est and other covariates. H owever, even when there wa s a moderate level of correlation between covariates, (R Squared of .5), the minimum detectable difference wa s still low at 1.169 which suggest ed that the sample size of 319 was adequate. Similarly, the power analyse s suggested that there wa s adequate sample size to conduct a multiple logistic regression, o r a multinomial regression. The sample size of 319 achieved 80% power to detect an R Squared as low as .05

PAGE 49

49 attributed to 10 independent variables at the .05 significance level Lastly, though there are not formal power analyses designed for LTA models, and though the sample needed to effectively conduct LTA models could vary depending on the complexity of the model, the LTA developers recommend a sample size over 300 (Lanza et al. 2015)

PAGE 50

50 Figure 2 1. Sex, Drugs, and Violence: A Longitudinal Analysis of the SAVA Syndemic among Female Offenders Courtroom Behavior Check List : Behaviors of women in court R eferred through Municipal Court sys tem (N=319) Baseline Assessments Washington University Risk Behavior (WU RBA) : sociodemographic c haracteristics, drug use, risky sexual behaviors, and perceptions of risky behaviors Violence Exposure Questionnaire: past 4 month physical and sexual violence Courtroom Behavior Check List Scores (CRBCL) Randomization to standard intervention (SI) or SI with peer partnered case management intervention (PPCMI) SI (N=155) SI + PPCMI (N=164) 10 week, 40 hour case management 4 month Follow Up **For SI+PPCMI, 4 months starts after 1 0 week intervention** Assessments same as baseline Baseline Data SAVA (Substance use, violence, and HIV risk behaviors), Risk Perceptions, Socio Demographic Characteristics Interve ntion Group, Risk Perceptions, Socio Demographic Characteristics Changes in SAVA from baseline to 4 and 8 month follow ups 8 month Follow Up Assessments same as baseline Paper 2 Paper 3 Paper 1 Original Study Proposed Analyses SAVA Latent Statuses from baseline to 4 and 8 month follow ups I ntervention Group, Risk Perceptions, Socio Demographic Characteristics Paper 4 SAVA, Intervention Group, Socio Demographic Characteristics Misdemeanor, Municipal Violation, Felony Charges at 4 and 8 month follow ups

PAGE 51

51 Figure 2 2. Sample Source for the STOP Study N=1,150 Fig ure 2 3. Recruitment Flow Chart for the STOP Study

PAGE 52

52 Figure 2 4. Conceptual Model of Paper 1

PAGE 53

53 Figure 2 5. Conceptual Model of Paper 2 Figure 2 6. Conceptual Model of Paper 3

PAGE 54

54 Figure 2 7. Conceptual Model of Paper 4

PAGE 55

55 Table 2 1 Definitio n of SAVA in Analyses SAVA in the Past 4 Months Variables and SAS Codes VIOLENCE COMPONENTS Was threatened with a gun VEQ2D. During the past 4 months, has anyone attacked you with a gun? No=1 Yes=5 if veq2d= 5 then v_gun= 1 ; else v_gun= 0 ; Was pressured or forced to participate in sexual acts VEQ2B. During the past 4 months, has anyone pressured or forced to participate in sexual acts against your will? No=1 Yes=5 if veq2b= 5 then v_sexacts= 1 ; else v_sexacts= 0 ; Emotionally abused VEQ2A. During the p ast 4 months, has anyone abused you emotionally, that is, did or said things to make you feel very bad about your life? No=1 Yes=5 if veq2a= 5 then veabuse= 1 ; else veabuse= 0 ; Physically abused (hurt to the point of bruises, cuts, broken bones) VEQ2C. D uring the past 4 months, has anyone hurt you to the point that you had bruises, cuts, broken bones, or otherwise physically abused you? No=1 Yes=5 if veq2c= 5 then v_physbuse= 1 ; else v_physabuse= 0 ; Attacked with knife, stick, bottle, or other weapon VE Q2e. During the past 4 months, has anyone attacked with knife, stick, bottle, or other weapon? if veq2e= 5 then v_attack= 1 ; else v_attack= 0 ; Any Violence (1+violence components) If v_attack=1 or v_physabuse=1 or v_sexacts=1 or veabuse=1 or v_gun=1 then v_ any=1; else v_any=0

PAGE 56

56 Table 2 1 Continued SAVA in the Past 4 Months Variables and SAS Codes HIV/AIDS RISK BEHAVIOR COMPONENTS Unprotected vaginal sex SA7. In the past 4 months, approximately how many times did you have vaginal sex with a man ? SA8. Of these (SA7) times, how many times was a male condom used? SA9. Of these (SA7) times, how many times was a female condom used? Unprotected_vaginal= sa7 (sa8+sa9) Unprotected anal sex SA16. In the past 4 months how many times did you have sex where a man put his penis into your butt, also known as anal sex? SA17. Of these (SA16) times, how many times was he wearing a male condom during anal sex? SA18. Of these (SA16) times, how many times did you use a female condom in your butt? Unprotected_ anal= sa16 (sa17+sa18) Any unprotected sex act (vaginal, anal, or oral) Any_unprotected= unprotected_oral+ unprotected_vaginal+unprotected_anal Number of sex partners (2+ vs. less than 2) SA3.) In the past 4 months, how many different people have you had vaginal, oral, or anal sex? Risky partner (likely to be an IDU or have another partner) vs. None AS18.) Do you think your primary partner has had sex with anyone else in the last 30 days? No=1 Yes=5 SA4.) Thinking about the (# from SA3) people you had sex with since (date 4 months ago), how many were likely to have been drug injectors? If as18=5 or sa4 ge 1 then riskypartner=1; if as18=1 and sa4=0 then riskypartner=0. HIV/AIDS Risk B ehavior R isky partner OR 2+ se x partners AND 1+ unprotect ed sex act

PAGE 57

57 Table 2 1 Continued SAVA in the Past 4 Months Variables and SAS Codes Violence Components SUBSTANCE USE (NUMBER OF USES IN PAST 30 DAYS) Crack DUIA3. How many days have you used crack in any way in the last 30 days? DUIB3. During th ese (# in A) days, how many times a day did you usually use crack? crackuses=duia3*duib3 Cocaine DUIA4. How many days have you used cocaine in any way in the last 30 days? DUIB4. During these (# in A) days, how many times a day did you usually use coc aine? cocaineuses=duia4*duib4 Crack/Cocaine CC= Crackuses + cocaineuses Heroin DUIA5. How many days have you used heroin in any way in the last 30 days? DUIB5. During these (# in A) days, how many times a day did you usually use heroin? heroinuses =duia5*duib5 Stimulants DUIA2. How many days have you used stimulants in any way in the last 30 days? DUIB2. During these (# in A) days, how many times a day did you usually use stimulants? stimulantuses=duia2*duib2 Total number of drug uses in th e past 4 months Total_drug_uses=Mjuses+stimulantuses+ crackuses+ cocaineuses+ heroinuses

PAGE 58

58 Table 2 1 Continued SAVA in the Past 4 Months Variables and SAS Codes SAVA (SUBSTANCE USE+ VIOLENCE+ HIV/AIDS RISK BEHAVIORS) 0 None met 1 One SAV A crit erion met 2 Two SAV A criteria met 3 All three SAV A criteria met

PAGE 59

59 CHAPTER 3 ORDER IN THE COURT? THE ASSOCIATION BETWEEN SUBSTANCE USE, EXPOSURE TO VI OLENCE, RISKY SEXUAL BEHAVIORS AND OBSERVED COURT BEHAVIORS AMONG FEMALE OFFENDERS Introduction Gender D ifferences in Population Growth of the Criminal Justice System The proportion of those involved in the criminal justice system in the United States has increased dramatically over the past several decades (Blumstein, 2015; Dumont, Allen, Brockmann, Alexa nder, & Rich, 2013; Massoglia & Pridemore, 2015). Though the alarming figures of those incarcerated have led to small decreases in incarceration rates, gender differences in these rates are evident (Greiner, Law, & Brown, 2014; Minton & Golinelli, 2014). R esearch by Minton & Golinelli (2014) found that though the population of individuals in county and city jail decreased in recent years, the proportion of females in these jails increased by nearly 11%. Though men are still significantly more likely to be i nvolved in some form of correctional supervision, women are now the fastest growing population in the criminal justice system (Hall, Golder, Conley, & Sawning, 2013; Greiner, Law, & Brown, 2014; Minton & Golinelli, 2014). Drug Courts and the Development o f the CRBCL In recent years, therapeutic justice programs have emerged as a way to curb incarceration rates among offenders with mental health issues including addiction (Mitchell, Wilson, Eggers, & MacKenzie, 2012; Matusow et al. 2013; Sevigny, Pollack, & Reuter, 2013). Drug courts, a therapeutic justice program for offenders charged with drug offenses, have become increasingly popular. However, drug court completion rates

PAGE 60

60 are sub optimal (DeVall, Gregory, & Hartmann, 2015; Mitchell et al. 2012; Sevigny, Fuleihan, & Ferdik, 2013). Overall, the criminal justice system and related therapeutic justice programs comprise a population of females who are at risk for poor outcomes and termination mling, Bekman, & Caudy, 2012; Reingle et al. 2012; Sevigny, Fuleihan, & Ferdik, 2013). In response to the need for a good measure to predict female offenders who will be non compliant or change high risk behaviors while in therapeutic justice programs, i nvestigators developed the Courtroom Behavior Checklist (CRBCL) (Reingle et al. 2012). The CRBCL, an assessment that has been used over 7,000 times in the Municipal Court system, quantifies readiness for court and court behaviors among women at Drug Court and may also serve as a proxy for conformity to norms. In a prior analysis evaluating the association between high CRBCL scores (indicating unacceptable behavior) and recidivism (defined as new misdemeanor or felony arrests and new Municipal violations), Reingle and colleagues (2012) found that the women with higher CRBCL scores had up to three times the odds of re offending than women with lo wer scores SAVA Syndemic and the Potential Correlation with CRBCL Research has found that among women, initial in volvement in the criminal justice system and subsequent re offenses are linked with the SAVA syndemic, which is the con current and mutually reinforcing issues of substance abuse (SA), violence (V), and sexual behaviors that can lead to HIV/AIDS (A) such a s sex trading (Singer,1996; Fries, Fedock, & Kubiak, 2014; Scott, Grella, Dennis, & Funk, 2014; Tripodi & Davis, 2013; Messina, Grella, Cartier, & Torres, 2010; Steinberg et al. 2011; Klein, Elfison, Senn, Carey, & Coury Doniger, 2011). Since many female offenders are arrested for drug

PAGE 61

61 related charges and often have high rates of victimization and risky sexual behaviors, the CRBCL may be a way to identify women in Drug Court with these often co occurring issues. A way to conceptualize the possible link be tween observed court behaviors and its potential link with SAVA may lie in the most common framework used in interventions for offender populations, the therapeutic community treatment program. This framework views any type of negative behaviors (e.g. subs tance use), as a part of a larger behavioral disorder and suggest s that changes in unfavorable behaviors depends on adopti n g prosocial behavior others (Staton Tindall, Harp, Winston, Webster, & Pa ngburn, 2015). Using this linked with other issues such as the SAVA syndemic, which is known as the most common pathway into the correctional system for females. Curre nt Analysis The current analysis assesses the association between self reported SAVA measured using the CRBCL and controlling for socio demographic characteristics. We hypothesiz e that women who meet the criteria for the SAVA syndemic will have significantly higher baseline CRBCL scores compared to women who do not. The CRBCL may serve as a means to identify the women at Drug Court at highest risk for the SAVA syndemic, which may allow more targeted and intense interventions. In addition, prior research indicates that religion/spirituality has been associated with decreases in risky behaviors and increases in prosocial behaviors (Shariff, Willard, Andersen, & Norenzayan, 2016; Ach eampong, Lasopa, Striley, & Cottler, 2015; Cheney

PAGE 62

62 et al. 2014; Gmel et al. 2013; Sussman, Reynaud, Aubin, & Leventhal, 2011). We also aim to understand the possible role of religion/spirituality and other environmental and social risk factors such as var ious markers of social economic status, childhood experiences, perceptions of risky sexual and drug using behaviors, and social support in the association between SAVA and observed court behaviors. The criminal justice system lacks a good measure to predi ct who will be non compliant or change their drug using behavior and accompanying risky sexual behaviors. A recent systematic review of the HIV literature highlighted a lack of structural interventions in criminal justice settings to reduce HIV/AIDS transm ission (Shoptaw et al. 2013). Thus, results from these analyses may fill an important gap with the development of a much needed structural intervention. Methods Data for this cross sectional analysis comes from a National Institute of Nursing Research fun ded study Sisters Teaching Options for Prevention (STOP) (R01NR09180 PI: Cottler). The sample was comprised of 319 underserved women mainly recruited from a Municipal Drug Court System in Midwest, US. STOP research staff members were present at the courts and recruited women by handing out flyers outlining the details of the study. Interested, eligible, women (at least 18 years of age) were scheduled for an initial interview outside of the courtroom. It is important to note that though the majority of the STOP sample came from a drug court system, around 12% of the sample were recruited from areas other than court (e.g. community treatment centers). Of the 282 women recruited from the municipal court system, 6% (N=18) were excluded from this analysis as the y were granted an excused absence from court, thus, the sample size for this analysis was 264 women. A ll women underwent the same

PAGE 63

63 baseline processes and interviews about risky sexual behavior, substance use, violence, and socio demographic variables, at ba seline. This study was approved by the Institutional Review Board at Washington University of St Louis. Measures The Washington University Risk Behavior Assessment (WU RBA) Assessment, t he WU RBA was used to assess sexual behavior such as number of sex partners, condom use during sexual activities, and sex trading, as well as drug using behaviors such as quantity and frequency of licit and illicit drug use (Needle et al. 1995). WU RBA wa s also used to examine perceptions of sexual and drug using behaviors, and demographi c information of participants. A study that examined the psychometric properties of this WU RBA found that this assessment had good reliability (Shacham & Cottler 2010). The Violence Exposure Questionnaire (VEQ) The Violence Exposure Questionnaire, developed from the Conflict Tactics Scale (Strauss, 1979), assessed various forms of current and past violent experiences and abuse including: sexual abuse, being threatened or attacked with a weapon, and physical and emotional abuse. The Court Room Behavior Check List ( Main Outcome Observed Court Behaviors) STOP research staff were allowed into court to assess courtroom behaviors using the CRBCL; 13 items were measured, with items 1 7 assessed before court and items 8 13 assessed during court. The assessment measured attendance, items related to court readiness, demeanor, and behavior (e.g. disrupting the judge, being distracted) in drug court. The scores ranged from 0 (indica ting favorable behavior), to 35 points (for

PAGE 64

64 an unexcused absence from the Court). For items that were subjective, staff were trained to ensure they understood the intent. Moreover, since the women were seen in court more than once, the CRBCL score prior to the baseline was used for analysis, consistent with previous analyses (Reingle et al. 2012). The CRBCL recorded behaviors through an overall score that ranged from 0 (indicating optimal behavior), to 35 (for an unexcused absence from the Court) (Table 1) Main Exposure SAVA in the past 4 months Prior to Baseline Violence To assess exposure to violence, as in victimization, participants were asked months, has anyone pressured or forced you to participate in nths, has anyone hurt you to the point that you had bruises, cuts, participants were consid ered having experienced violence in the past 4 months if they HIV/AIDS risk behaviors Participants were considered to meet the criteria for HIV/AIDS risk behaviors if they reported having at least one p artner who was an injection drug user or had other partners simultaneously or the women had 2+ sex partners AND at least one reported unprotected sex acts (any unprotected vaginal, anal, or oral sex).

PAGE 65

65 Substance use (number of uses in the past 30 days). P subsequently how many times a day each drug was used, if they reported using a specific substance for 1 or more days in the past 30 days. The number of uses of marijuana, stimulants (speed, am phetamines), crack/cocaine, and heroin in the past 30 days was calculated for each participant by multiplying the number of days used by how many times a day each drug was used. Overall, recent substance use was the summation of the number of uses for all drugs in the past 30 days. Both the total number of uses for each drug, as well as the total number of uses for all drugs was used in the analysis. SAVA C riterion et al l three criteria (substance use, violence, and HIV/AIDS risk) were considered to have the SAVA syndemic. Covar iates Potential cofounders that were included in this analysis were: religion/spirituality (defined as viewing religion and spirituality as very important, attending religious services regularly, and seeking advice from religious leaders in the past 12 mon ths vs. no ), number of arrests greater than 25 th percentile of reported arrests in the sample (4+ life time arrests vs. 3 or less life time arrests), childhood experiences ( separated 6+ months from parents before the age of 15 vs. no or less than 6+ months from parents before the age of 15 ; experiencing child sexual abuse before the age of 15 years of age

PAGE 66

66 vs. no ), and social support (having someone who you can talk to and ask for favors vs. no social support). Socio demographic covariates included in this a nalyses were: education (at least a high school diploma vs. no high school diploma), age (18 29 years of age vs. 30+), unstable housing (living on the streets, with others, halfway house etc vs. live in own house or apartment) and race (black vs. non blac k). Analysis To evaluate the bivariate associations between variables of interest and CRBCL scores, Kruskal Wallis tests were used. Kruskal Wallis tests are used to compare categorical variables and ordinal dependent variables (UCLA, 2016). This was chose n as the most appropriate avenue as the CRBCL score is a count variable, therefore, it is appropriate to avoid the assumption of normality. Moreover, we used negative binomial regressions for our multivariable analyses to account for over dispersion in CRB CL scores. Overall, 2 multivariable analyses were conducted: 1) a negative binomial regression model assessing behavior specific correlates of baseline CRBCL scores and 2) a negative binomial regression model assessing the association between the number of SAVA criterion met and baseline CRBCL scores. Results Socio Demographic Characteristics Among our sample, 69% self identified as Black, 31 % self identified as non black -primarily White ( Table 3 2 ). Around a third of the women reported being married, wido wed, separated, or divorced, and less than 30 years of age, while near ly half (46 %) of the women reported having less than a high school diploma. Reg arding childhood experiences, 74 % of the women reported being separated 6+ months from at least one parent, wh ile around half of the women (50 %) reported being sexually abused before

PAGE 67

67 the age of 15. The vast majority of the women also reported having social support (someone to talk to and ask for favors) ( 78 %) but not stable housing (74%) Regarding belief syste ms, 43 % of the women reported that they had risky sexual behavior s that needed changing, while 48 % reported that they had drug using behaviors that needed changi ng. Approximately one forth (23 %) of the women reported regularly attending religious services, viewing religion/spirituality as important to them, and seeking advice from religious leaders in the past 12 months. Overall, our sample consisted of women who reported numerous crimina l justice involvement, around 71 % of the women reported 4 or more life time arrests. Of these socio demographic variables, religion/spirituality was significantly associated with decreased CRBCL scores, while believing that you had risky drug using behaviors that need changing lower education, being recruited from a city dru g court and 4+ arrests was significantly associated with increased CRBCL scores (p<0.05). Exposure to Violence, HIV/AIDS Risk, and Substance Use at Baseline Exposure to violence in the past 4 months was highly prevalent with nearly 60% of women reporting at least one instance of violence (Figure 3 3 ). The most commonly reported instance of violence was emotional abuse which was reported by 53 % of the women, followed by 19% for physical abuse (defined as being hurt to the point of bru ises, cuts, or broken b ones). Around 10% of women reported being pressured or forced to participate in sexual acts or being attacked with a knife, stick, bottle, or other weapon. A small percentag e of women (4 %) reported being threatened with a gun in the past 4 months. Of these violent experiences, only being threatened with a gun was significantly associated with decreased CRBCL scores.

PAGE 68

68 Sexual risk taking wa s also commonly reported (Table 3 3 ); 69% reported having at least one instance of unprotected sex of any kind in the pa st 4 months. Specifically, 64 % of the women reported at least one instance of unprotected vagi nal sex in the past 4 months, 49 % reported at least one instance of performing oral sex without any protection in the past 4 months, and 9% reported at least one instance of unprotected anal sex in th e past 4 months. Nearly half (55 %) of the women reported having 2 or more sex partners in the past 4 months, while a quarter of the women reported havin g at least one risky partner (25%). Overall, 48 % of the women met the criterion for having HIV/AIDS risk behaviors, defined as having 2 or more sex partners or at least one risky partner AND having one or more instance of an unprotected sex act. Illicit substance use in the past 30 days was reported by almost half of th e women (47%) ( Table 3 3 ). The most commonly used substances were crack/cocaine and mar ijuana, which was reported by 35% and 27 % of the women respectively. Only a small percentage of the women reported heroin (5 %) or stimulant use (1%). The use of crack/co caine and the composite variable combining any substance use were significantly associated with higher CRBCL scores. SAVA Among the Sample at Baseline Overall, 17 % of the women recently used an illicit substance, experienced at least one incident of viole nce, and met the criteria for HIV/AIDS risk behaviors in the past 4 months, which met the criteria for a SAVA syndemic ( Table 3 3 ). T he same percentage of women did not experience any type of violence or have HIV/AIDS risk behaviors in the past 4 months, a nd did not use an illicit drug in t he last 30 days, while, 33 % of th e women met 1 criterion or 2 SAVA criteria.

PAGE 69

69 Severity of Substance Use, Violence Experienced, and HIV/AIDS Risk at Baseline We also evaluated whether the severity of substance use, violence experienced, and the risky sexual behaviors differed among SAVA criterion (Table 3 4 ). The results showed that among those who had SAVA, the median number of times individuals used any type of subs tance in the past 30 days was 22 These individuals also h ad a median number of 3 sexual partners, 15 unprotect ed sex acts, and experienced 1 act of violence in the past 4 months. Those who did not have SAVA tended to have used substances fewer times than those with SAVA, however similar numbers of sex partners a nd violence experienced were evident among all groups. Multivariable Models Assessing Behavior Specific Correlates of CRBCL Scores We used a multivariable negative binomial regression model to assess behavior specific correlates of baseline CRBCL scores (T able 3 5 ). In the un adjusted model, women who used any substance in the past 30 days had significantly higher scores (RR 1.75, 95% CI 1.39, 2.16 ) respectively than women who d id not report any substance use; however, the strength of this association decrea sed in the adjusted model (RR 1.40, 95% CI 1.10 1.70) W omen who met the criteria for having HIV/AIDS risk also had significantly higher CRBCL scores in the adjusted model (RR 1.30 95% 1.02, 1.66 ). On the contrary, women who had experienced violence in t he past 4 months had significantly lower scores than women who were not exposed to violence (RR .71 95% .56, .91 ). In regards to socio demographic correlates of CRBCL scores, women who were less educated, had 4 or more arrests or were recruited from the city drug court had 30%, 33%, and 44% increased scores compared with women who had higher levels of education less than 4 arrests or were recruited from the county, family, or state drug

PAGE 70

70 courts In addition, w omen who were less religious/spiritual were significantly more likely to have higher CRBCL scores than women who were very religious/spiritual. In fact, religion/spirituality was of the strongest predictor of CRBCL scores in the adjusted model (RR 1.41 95% CI 1.06, 1.88 ). Race and age were not sign ificant correlates of CRBCL scores in the unadjusted or the adjusted models. Multivariable Models Assessing SAVA and CRBCL Scores Our multivariable negative binomial regression model assessing SAVA and baseline CRBCL scores yielded interesting and similar results to the be havior specific model (Table 3 5 ). In the unadjusted model, women who had SAVA ha d 73 % higher CRBCL scores than women wh o did not meet any SAVA criterion (RR 1.73, 95% 1.15, 2.61 ). However, after adjusting for socio demographic factors, th e number of SAVA component criterion met was no longer statistically significant ly associated with CRBCL scores. In this model, there was a trend for wome n with 4 or more arrests or lower education to have higher CRBCL scores than women who had less than 4 arrests or higher education however, these associations marginally missed significance (RR 1.31, 95% CI .98, 1.77; RR 1.21, 95% CI .95, 1.54). The association between religion/spirituality and CRBCL scores was statis tically significant and increased in strength in this model compared to the adjusted behavior specific model (RR 1.67 95% CI 1.26 2.23 ). Participants who agreed that they had risky drug using behaviors that needed changing were significantly more likely to have higher CRBCL scores than part ici pants who did not agree (RR 1.40 9 5% CI 1.08, 1.82 ).

PAGE 71

71 Discussion In this analysis, we aimed to evaluate the association between self reported measured using the CRBCL an d controlling for socio demographic characteristics. We hypothesized that women with SAVA would have significantly higher baseline CRBCL scores compared to women with who do not meet the criteria for SAVA. The results of our unadjusted negative binomial re gression model, which illustrated the significant increase in CRBCL scores among women with SAVA, supported this hypothesis. However, this association was attenuated when adjusted for socio demographic characteristics of the sample. However, women who used an illicit substance in the past 30 days or met the criteria for HIV/AIDS risk were more likely to have unfavorable court behaviors than non substance using women or women who did not meet the criteria for HIV/AIDS risk. On the contrary, exposure to viole nce was a significant correlate of lower CRBCL scores. Our analyses yielded further interesting results. In our unadjusted model assessing SAVA and CRBCL scores, women who met one SAVA component criterion had increased CRBCL scores, however, this was not true for women who met 2 component criteria. We believe that the descriptive statistics on the severity of substance use, violence, and risky sexual behaviors by the number of SAVA component criteria met provides some clarification. Women who met 2 SAVA c omponent criteria tended to experience more violence than women in the other groups. Since exposure to violence was linked with significantly decreased s cores, this may explain why meeting 2 component criteria was not significant in the unadjusted model, w hile meeting one all 3 were

PAGE 72

72 Our study did yield further results that were consistent with the known literature. Those who were less religious/spiritual, had less than a high school diploma, and had 4 or more arrests were significantly more likely to hav e unfavorable court behaviors. Prior studies have found that religion/spirituality has been linked with prosocial behaviors, which may offer an explanation as to why religious/spiritual women were more likely to have more favorable court behaviors (Shariff et al. 2016; Sussman, et al. 2011). In addition, religion/spirituality has also been shown to reduce the odds of substance use which was one of the strongest predictors of court behaviors in our analyses (Cheney et al. 2014). Reingle et al. (2012) fou nd that CRBCL scores predicted future criminal offenses, however, the results of these analyses suggest that the CRBCL also is associated with baseline demographics of the women such as prior arrest history. Lastly, women with lower education were signifi cantly more likely to have unfavorable court behaviors than women with more education. Lower education has been previously linked with poorer criminal justice outcomes (Mitchel l et al. 2012). However, in regards to court behaviors, it may be plausible tha t women with lower education levels may have been unaware of the expectations of court, and thus leading these women to having higher CRBCL scores. Strengths and Limitations There are several limitations in this study. First, our sample was not randomly s elected, meaning that the results of this study may not be generalizable to all females in drug court. Second, we relied on self report data on sensitive questions, which can lead to the social desirability bias. However, our study also has many strengths including a relatively large population of an under researched population of female offenders. Moreover, our study used a rich data set with detailed items on substance

PAGE 73

73 use, violence, risky sexual behaviors, and perceptions of these behaviors. The availabi lity of such data allowed a detailed analysis examining many variables which may not be available in other data sources. Moreover, to our knowledge, the CRBCL is the first assessment which quantifies court readiness and behaviors, providing a measurable va riable for analyzing these outcomes. Conclusion The CRBCL may have added utility in identifying female offenders with recent substance use, exposure to violence, and risky sexual behaviors. Further studies on other samples of offenders are needed to suppo rt these findings. Overall, our study was comprised of a high risk population of females who have experienced recent trauma, used substances, and have risky sexual behaviors. Our findings corroborate with researchers who advocate for gender specific interv entions for females due to high levels of substance use, risky sexual behaviors, and trauma experienced (Blankenship, Reinhard, Sherman, & El Bassel, 2015; Scott et al. 2014; Hall et al. 2014; Abram, Teplin, & McClelland, 2004). Future studies should eval uate the prevalence of substance use, exposure to violence, and risky sexual behaviors over time among the female offender population, as well as examine significant correlates of change in these issues.

PAGE 74

74 Table 3 1. Description and Distribution of CRBCL and Scores Item Values Present at court Present=0, Unexcused absence=35 Under the influence of drugs or alcohol No=0, Yes (observed by research staff) =5, Yes (court comment) =10 Cell phones/pagers turned off Yes=0, No=1 Has notes about progress Yes=0, No=1 Has documents about progress Yes=0, No=1, Falsified documents=5 Is alone or with an advocate With Advocate=0, Alone=1, With Children=3 Disruptive in court No Talking=0, Talking (observed by research staff) =1 Talking (court comment) =2 Not Eating=0, Eating=1 Not Rifling through Possessions=0, Ri fling through Possessions=1 Courteous to court staff Yes=0, No=1 Interrupted the judge No=0, Yes (observed by research staff) =1, Yes (court comment) =3 Prepared to take note s on required tasks Yes=0, No=1 Responded to judge appropriately Yes=0, No (observed by research staff) =1, No (court comment) =2 Appeared to have confident demeanor Yes=0, No=1 Distribution of CRBCL scores 25 th Percentile (3), Median (4), 75 th Percentile (5)

PAGE 75

75 Table 3 2 Sample Characteristics of Participants at Baseline ( N=264 ) Demographic Characteristics CRBCL Scores ( 0 2 ) N=50 (19% ) CRBCL Scores (3 ) N=64 (24% ) CRBCL Scores (4 ) N=77 (29% ) CRBCL Scores (5+ ) N=73 (27% ) Total N=264 (100% ) p value Race African American 34 (6 8%) 46 (7 2%) 57 (74 %) 56 (63 %) 183 (69 %) .55 All other races 16 (32 %) 18 (28 %) 20 (26 %) 27 (37 %) 81 (31 %) Marital Status Ever married 19 (38 %) 17 (27 %) 26 ( 34 %) 29 ( 40 %) 91 (34 %) .49 Never married 31 (62 %) 47 (73 %) 51 (66 %) 44 (60 %) 172 (66 %) Age Less than 30 years of age 1 0 (20 %) 24 (38 %) 20 (26 %) 19 (26 %) 73 (28 %) .88 30 years of age+ 40 (8 0%) 40 (63 %) 57 (74 %) 54 (74 %) 191 (72 %) Social Support Has social support 40 (80 %) 4 7 (73 %) 55 (7 1%) 63 (86 %) 205 (78 %) .24 No social support 10 (20 %) 17 (27 %) 22 ( 29 %) 10 ( 1 4%) 59 (22 %) Education Less than high school diploma 22 (44 %) 26 (41 %) 33 (43 %) 43 (59 %) 147 (46%) .03 High school diploma or higher 28 (56 %) 38 ( 59 %) 44 (57 %) 30 (41 %) 140 (53 %)

PAGE 76

76 Table 3 2 Continued Demographic Characteristics CRBCL Scores ( 0 2 ) N=50 (19% ) CRBCL Scores (3 ) N=64 (24% ) CRBCL Scores (4 ) N=77 (29% ) CRBCL Scores (5+ ) N=73 (27% ) Total N=264 (100% ) p value No 31 (62 %) 26 (42 %) 35 (45 %) 38 (52 %) 130 (50 %) .71 Yes 19 (3 8 %) 36 ( 58 %) 42 (55 %) 35 (48 %) 132 (50 %) Separation from Parents before the age of 15 (6+ mos.) No 13 (2 6%) 15 ( 2 3 %) 17 (22 %) 24 (33 %) 69 ( 26 %) .03 Yes 37 (74 %) 49 (77 %) 59 ( 7 8 % ) 49 (67 %) 1 94 (74 %) Arrest History Less than 4 arrests 76 (34 %) 20 (31 %) 24 (31 %) 15 (21 %) 76 (29 %) .05 More than 4 arrests 33 ( 66 %) 44 ( 6 9 %) 53 (69 %) 58 (79 %) 188 (71 %) Housing Unstable Housing 35 ( 7 0 %) 47 (73 %) 56 (7 3 %) 56 ( 77 %) 194 ( 74 %) .48 Stable Housing 15 (30 %) 17 (27 %) 21 (27 %) 17 (23 %) 70 (27 %) Religion/Spirituality No 3 5 ( 7 0 %) 48 (75 %) 59 (77 %) 62 (85 %) 204 (77 %) Yes 15 (30 %) 16 ( 2 5%) 18 (23 %) 11 (15 %) 60 (23 %) .03

PAGE 77

77 Table 3 2 Continued Demographic Characteristics CRBCL Scores ( 0 2 ) N=50 (19% ) CRBCL Scores (3 ) N=64 (24% ) CRBCL Scores (4 ) N=77 (29% ) CRBCL Scores (5+ ) N=73 (27% ) Total N=264 (100% ) p value Recruitment Site County, Family or State Drug Court 17 (34 %) 29 (45 %) 17 (22 %) 12 (16 %) 75 (28 %) .03 City Drug Court 33 (66 %) 35 (55 %) 60 (78 %) 61 ( 84 %) 189 (72 %) Perceived to have risky sexual behaviors that need changing No 29 (58%) 35 (55%) 46 (60%) 41 (56%) 151 (57%) .80 Yes 21 (42%) 29 (45%) 31 (40%) 32 (44%) 113 (43%) Perceived to have risky drug behaviors that need changing No 28 (56%) 38 (59%) 44 (57%) 28 (38%) 138 (52%) .01 Yes 22 (44%) 26 (41%) 33 (43%) 45 (62%) 126 (48%)

PAGE 78

78 Table 3 3 SAVA Amo ng the Samp le at Baseline (N=264 ) SAVA in the Past 4 Months CRBCL Scores ( 0 2 ) N=50 (19% ) CRBCL Scores (3 ) N=64 (24% ) CRBCL Scores (4 ) N=77 (29% ) CRBCL Scores (5+ ) N=73 (27% ) Total N=264 (100% ) p value Violence Components Was threatened with a gu n No 45 (90 %) 62 (97 %) 75 (97 %) 72 ( 9 8%) 254 (96 %) Yes 5 (10%) 2 (3 %) 2 (3 %) 1 ( 1 %) 10 (4 %) .03 Was pressured or forced to participate in sexual acts No 47 ( 9 4 %) 55 (86 %) 73 (95 %) 6 4 (8 8 %) 239 (91%) .64 Yes 3 (6 %) 9 (1 4%) 4 (5 %) 9 (12 %) 25 (9%) Emotionally abused No 27 (54 %) 29 (45 %) 31 (40 %) 37 (51 %) 1 24 ( 47 %) .93 Yes 23 (46 %) 35 (55 %) 46 (60 %) 36 (49 %) 140 (53 %) Physically abused (hurt to the point of bruises, cuts, broken bones) No 41 (82 %) 53 (83 %) 59 (77 %) 61 (8 4 %) 214 (81%) .96 Yes 9 (18 %) 11 (17 %) 18 (23 %) 12 (16 %) 50 (19%)

PAGE 79

79 Table 3 3 Continued Demographic Characteristics CRBCL Scores ( 0 2 ) N=50 (19% ) CRBCL Scores (3 ) N=64 (24% ) CRBCL Scores (4 ) N=77 (29% ) CRBCL Scores (5+ ) N=73 (27% ) Total N=264 (100% ) p value Attacked with knife, stick, bottle, or other weapon No 45 (90 %) 58 (91 %) 72 (94 %) 65 (89 %) 24 0 (91%) .87 Yes 5 (1 0 %) 6 (9 %) 5 (6 %) 8 (11 %) 24 (9%) Any Violence (1+violence components) N o 25 (5 0%) 26 (41 %) 29 (38 %) 35 (48 %) 138 (43%) .88 Yes 25 (50 %) 38 (59 %) 48 (62 %) 38 (52 %) 149 (56 %) HIV/AIDS Risk Behavior Components Unprotected oral sex (performed) No 26 (52 %) 35 (55%) 41 (53 %) 33 (45 %) 135 (51 %) .39 Yes 24 (4 8%) 29 (45 %) 36 (47 %) 40 ( 55 %) 129 (49 %) Unprotected vaginal sex No 16 (32 %) 24 (38 %) 30 (39 %) 25 (34 %) 95 (36 %) .97 Yes 34 ( 6 8%) 40 ( 63 %) 47 61 %) 48 (66 %) 1 69 (64 %)

PAGE 80

80 Table 3 3 Continued Demographic Characteristi cs CRBCL Scores ( 0 2 ) N=50 (19% ) CRBCL Scores (3 ) N=64 (24% ) CRBCL Scores (4 ) N=77 (29% ) CRBCL Scores (5+ ) N=73 (27% ) Total N=264 (100% ) p value Unprotected anal sex No 45 (90 %) 59 (92 %) 72 (94 %) 6 3 (86 %) 239 (91%) .55 Yes 5 (1 0 %) 5 (8 %) 5 ( 6% ) 10 (14 %) 25 (9%) Any unprotected sex act (vaginal, anal, or oral) No 13 (26 %) 22 (34 %) 27 (35 %) 19 (26 %) 81 (31%) .76 Yes 37 (74 %) 42 ( 66 %) 50 (65 %) 54 (74 %) 183 (69%) Number of sex partners (2+ vs. less than 2) No 29 ( 58 %) 26 (41 %) 33 (43 %) 40 (55 %) 1 2 8 ( 4 8 %) .08 Yes 21 (42 %) 38 (59 %) 44 (57 %) 33 (45 %) 136 (52 %) Risky partner (likely to be an IDU or have another partner) No 40 (80 %) 44 (69 %) 62 (81 %) 52 (7 1 %) 198 (75 %) .58 Yes 1 0 ( 20 %) 20 ( 3 1 %) 15 ( 19%) 21 (29 %) 66 (25 %)

PAGE 81

81 Table 3 3 Continued Demographic Characteristics CRBCL Scores ( 0 2 ) N=50 (19% ) CRBCL Scores (3 ) N=64 (24% ) CRBCL Scores (4 ) N=77 (29% ) CRBCL Scores (5+ ) N=73 (27% ) Total N=264 (100% ) p value HIV/AIDS ris k behavior (risky partner OR 1+ sex partners AND 1+ unprotected sex act) No 28 (56 %) 34 (53 %) 43 (56 %) 31 (42 %) 136 (52%) .14 Yes 22 (44 %) 30 (47 %) 34 (44 %) 42 (58 %) 128 (48%) Substance Use (Number of Uses in Past 30 Days) Marijuana No 3 8 (76 %) 49 (77 %) 47 (61 %) 14 (19 %) 193 (73 %) <. 01 Yes 12 (24 %) 15 (23 %) 30 (39 %) 59 (81 %) 71 ( 27 %) Crack/Cocaine No 36 (72 %) 46 (72 %) 50 (65 %) 40 (55 %) 1 72 (65 %) <.0 1 Yes 14 (28 %) 18 (28 %) 27 (35 %) 33 (45 %) 92 (35 % ) Heroin No 40 ( 9 8 %) 61 (95 %) 73 (95 %) 68 (93 %) 251 (95 %) 1 8 Yes 1 (2 %) 3 (5 %) 4 (5 %) 5 ( 7 %) 13 ( 5 %)

PAGE 82

82 Table 3 3 Continued Demographic Characteristics CRBCL Scores ( 0 2 ) N=50 (19% ) CRBCL Scores (3 ) N=64 (24% ) CRBCL Scores (4 ) N=77 (29% ) CRBCL Scores (5+ ) N=73 (27% ) Total N=264 (100% ) p value Stimulants No 50 (100 %) 62 (97 %) 77 ( 1 00 %) 73 (100 %) 262 (99%) .34 Yes 0 (0%) 2 (3 %) 0 (0%) 0 (0%) 2 (1%) Any drug use in the past 4 months No 3 2 (64 %) 39 (61 %) 3 6 (47 %) 34 (47 %) 141 (53 %) 01 Yes 18 (3 6%) 25 (39 %) 41 (53 %) 39 (53 %) 123 ( 47 %) SAVA No Criterion Met 1 2 (24 %) 11 (17 %) 11 (14 %) 10 ( 1 4%) 44 (17 %) .13 One SAVA Criterion M et 19 (38 %) 2 4 ( 3 8 %) 21 (27 %) 22 (10%) 8 6 (3 3 %) Two Sava Crit eria M et 11 (22 %) 18 (28 %) 33 (43 %) 26 (35 %) 88 (33%) All three SA VA Criteria M et 8 (16 %) 11 (17 %) 12 (16 %) 15 ( 2 1%) 46 (1 7 %)

PAGE 83

83 Table 3 4 Severity of Substance Use, Violence Experienced, and HIV Risk by SAVA Groups SAVA Patterns N (%) Median # of Times Used Substances in Past 30 Days Median # of Any Type of Unprotected Sex in Past 4 mo nths Median # of Sex Partners in Past 4 months Median # of Violent Experience s in Past 4 months No SAVA Criterion Met 44 (17 %) ----1 SAVA Criterion Met 86 (33 %) Violence Only 34 (13%) ---1 HIV Risk Only 20 (8 %) -9 3 -Subs tance Use Only 32 (11%) 10 ----2 SAVA Criteria Met 88 (33 %) Violence + HIV Risk 43 (16%) -16 3 2 Substance Use + Violence 26 (10%) 3 --1 HIV Risk+ Substance Use 19 (8 %) 24 15 2 -All 3 SAVA Criteria Met 46 (17 %) Sub stance Use + Violence+ HIV Risk 22 15 3 1 *All groups are mutually exclusive *Example Interpretation: Among those who m et al l 3 SAVA criteria, the median number of substance uses was 22, the median number of unprotected sex acts was 15, the median number of sex partners was 3, and the median number of violence experienced was 1

PAGE 84

84 Tab le 3 5 Adjusted Multivariable Negative Binomial Regre ssion Assessing Number of SAVA Cr iterion M et and Baseline CRBCL S cores (N=264 ) Variables Unadjusted Estimates RR (95% Wald Confidence Limits) Adjusted Estimates Substance Use, Violence, HIV/AIDS Risk RR (95% Wald Confidence Limits) Adjusted Estimates SAVA RR (95% Wald Confidence Limits) Race Non Black 1.0 ----------1.0 ----------1.0 ----------Black .84 (.64, 1.08 ) .95 (.74, 1.21 ) .93 (.73, 1.20 ) Age 30 years of age+ ----------1.0 ----------1.0 ----------18 29 years of age .84 (.64, 1.10) 1.07 (.80, 1.44 ) 1.01 (.74, 1.37 ) Education High school diploma+ 1.0 ----------1.0 ----------1.0 ----------Less than high school Diploma 1.54 (1.21, 1.95 ) 1.30 (1.03, 1.64 ) 1.21 ( .95,1.54 ) Recruitment Site County, Family or State Drug Court 1.0 ----------1.0 ----------1.0 ----------City Drug Court 1.99 (1.52, 2.62 ) 1.44 (1.01, 2.04 ) 1.50 (1.06, 2. 1 3 ) Arrest History Less than 4 arrests 1.0 ----------1.0 ----------1.0 ----------4+ arrests 1.80 (1.38, 2 34 ) 1.33 (.99, 1.79 ) 1.31 (.98, 1.77) Religion/Spirituality Yes 1.0 ----------1.0 ----------1.0 ----------No 1.68 ( 1.26 2.25 ) 1.41 (1.06, 1.88 ) 1.67 (1.26 2.23 )

PAGE 85

85 Table 3 5 Continued Variables Unadjusted Estimates RR (95% Wald Confidence Limits) Adjusted Estimates Substance Use, Violence, HIV/AIDS Risk RR (95% Wald Confidence Limits) Adjusted Estimates SAVA RR (95% Wald Confidence Limits) Perceived to Have Risky Drug Using Behaviors that Need Changin g No 1.0 ----------------------------1.0 ----------Yes 1.58 (1.26, 2.01 ) ------------------1.40 (1.08, 1.82 ) SAVA No criterion Met 1.0 ----------------------------1.0 ----------1 Criterion Met 1.51 (1.05, 1.90 ) ------------------1.24 (.86, 1.77 ) 2 Criteria Met 1.32 (.83, 1.90 ) ------------------1.01 (.70, 1.47 ) 3 Criteria Met 1.73 (1.15 2.61 ) ------------------1.09 (.71, 1.67 ) Violence No 1.0 ----------1.0 --------------------------Yes .76 (.59, 1.04 ) .71 (.56, .91 ) -----------------HIV/AIDS Risk No 1.0 ----------1.0 ---------------------------Yes 1.19 (. 9 3, 1.51 ) 1.30 (1.02, 1.66 ) -----------------Any Substance Use i n Past 30 Days No ---------------------------Yes 1.75 (1.39, 2.16 ) 1.40 (1.10, 1.79 ) -----------------

PAGE 86

86 CHAPTER 4 SEX, DRUGS, AND VIOLENCE: A LONGITUDINAL ANALYSIS OF THE SAVA SYNDEMIC AMONG FEMALE OFFENDERS Introduction SAVA Syndemic among Women in the Criminal Justice System Syndemics are defined as two or more inseparable epidemics which work together synergistically to produce excessive negative health and social consequences in affected populations (Singer, 1996; Singer, 2006; Singer, 2009). Th e intersection of substance abuse (SA), violence (V) and HIV/AIDS (A), otherwise known as the SAVA mutually reinforcing health and social problems of substance use, vio lence, and documented the link between substance use, risky sexual behaviors, and exposure to violence (Salas Wright, Olate, & Vaughn, 2015; Gilbert et al. 2015; Sulliva n, Messer, & Quinlivan, 2015; Illangasekare, Burke, McDonnell, & Gielen, 2013; Dyer et al. 2013; Islam et al. 2013; Russell, Eaton, Peterson Williams, 2013; Adimora, Schoenbach, Taylor, Khan, Schwartz, 2011; Meyer et al. 2011). Of special interest are women involved in the criminal justice system, who are known to have significantly elevated rates of the SAVA syndemic compared with the general population of women (Meyer et al. 2011; Elkington et al. 2008; Harner & Riley, 2013; Roth et al. 2012; Fulke Ferdik, 2013; Peters, Kremling, Bekman, & Caudy, 2012). Moreover, when compared to men in the judicial system, women face a greater burden of the SAVA syndemic (Senn, Carey, & Coury Doniger, 2011; Messina, Grella, Cartier, & Torres, 2010; Klein, Elfison,

PAGE 87

87 & Sterk, 2008). A better understanding of the SAVA syndemic among this population, along with understanding of interventions that can reduce SAVA among women in criminal justice settings are vital. Syndemi c Theory and SAVA In order to understand and properly design interventions to reduce the SAVA syndemic, it is imperative to delve in to the contextual factors of social, environmental, and political influences that propagate this issue, particularly among those involved in the criminal justice system (Russell, Eaton, & Petersen Williams, 2013; Singer, 1996; Singer, 2006; Singer, 2009). For example, the criminal justice system is known to contribute to HIV risk behaviors by disrupting stable social networks and economic situations (Khan et al. 2015). Such disruptions are known to destabilize intimate relationships and increases high risk sexual behaviors such as concurrent and multiple sexual partners, and increase the likelihood of trading sex for money and other resources (Khan et al. 2015; Epperson et al. 2010; Freudenberg, 2009; Plefieger et al. 2013; Sharpe et al. 2012). Environmental factors such as high intensity drug use areas and social norms and peer groups are also linked with higher rates of d rugs and crimes, which have all been linked to HIV/STI clusters and their subsequent risk behaviors (Jennings et al. 2013; Sharpe et al. 2012; Tripodi et al. 2013). Furthermore, policies such as the War on Drugs, policies that lead to residential segregat ion and subsequently to the concentration of poverty in various geographical districts, and the targeted marketing of psychoactive drugs, have led to a dramatic increase of incarcerated individuals, and subsequently SAVA (Adimora et al. 2005; Jennings et al. 2013; Sharpe et al. 2012; Tripodi et al. 2013).

PAGE 88

88 Gender Based Behavioral Interventions for Women in the Criminal Justice System Because women are now the fastest growing population in the criminal justice system and the SAVA syndemic is elevated in t his population, research and behavioral interventions tailored to the specific needs of this population are warranted (Tripodi & Pettus Davis, 2013; Blankenship et al. 2015; Binswanger et al. 2010; Messina et al. 2010). Gender based research and interve ntions are especially needed in the criminal justice system since female recidivism has been linked with a lack of female oriented drug and behavioral interventions (Tripodi & Pettus Davis, 2013; Messina et al. 2010). Moreover, there is also a need for ge nder specific health behavior interventions that provide social support and access to social services for female offenders, especially considering the significantly higher rates of negative life events found in this population (Blankenship et al. 2015; Bi nswanger et al. 2010; Messina et al. 2010). Gaps in Knowledge In a recent review of the literature on substance using women, authored by El Bassel and Strathradee, a lack of epidemiologic studies on SAVA, especially in criminal justice settings and alte rnatives to incarceration programs, was identified as a gap in the knowledge (El Bassel & Strathdee, 2015). Moreover, this review highlighted a need for studies that illustrate the prevalence of violence and risky sexual behaviors among vulnerable subpopul ations, along with studies that elucidate the effect of race and socio economic status on these issues among drug using women. Therefore, this current study aims to evaluate the longitudinal trends of the SAVA syndemic over time among female offenders to c ontribute to the literature in this area. Specifically, we aim to: 1) evaluate the association between a peer partnered case management intervention and

PAGE 89

89 decreases in the likelihood of SAVA over time; 2) assess the strength of relationships between violence substance use, and HIV/AIDS risk by assessing the effect of the initial prevalence of these issues on longitudinal outcomes; and 3) determine the effect of race, markers of socio economic status such as education and stable hou sing at baseline on SAVA ov er time. We hypothesize that: 1) the peer partnered case management intervention will be associated with decreases in the likelihood of SAVA over time, 2) the initial baseline prevalence of violence, substance use, and HIV/AIDS risk will be associated with the longitudinal outcomes of these issues, however baseline substance use would have the strongest effect on violence and HIV/AIDS risk over time, and 3) race, lower education, and unstable housin g at baseline will be associated with an increased likeliho od of SAVA over time. Methods Sisters Teaching Options for Prevention and a Case Management Intervention The data for this study comes from Sisters Teaching Options for Prevention project (STOP) (R01NR09180, PI: Cottler), a randomized controlled field stu dy which featured a gender based behavioral intervention to reduce high risk drug and sexual behaviors among female offenders. A major strength of STOP is that it addresses the often ignored STI prevention services amongst drug court enrollees (Robertson, St. Lawrence, & McCluskey, 2012). STOP also assessed the effectiveness of using a 40 hour Peer Partnered Case Management Intervention (PPCMI) to increase access and utilization of needed health services, compared to a standard intervention (SI) alone (John son et al. 2011). The SI, which all participants received, consisted of the National Institute on Drug Abuse (NIDA) standard pre and post HIV test counseling. In contrast,

PAGE 90

90 the PPCMI intervention provided practical support such as transportation and social support that is usually lacking for drug court enrollees (Peters et al. 2012). Outreach and Recruitment Participants in the STOP study were mainly recruited from a Municipal Drug Court System in the Midwest by research staff who disseminated information al flyers which provided details on the study. Interested and eligible women (at least 18 years of age) were then scheduled for their baseline assessments by research staff. All participants were interviewed using the Washington University Risk Behavior As sessment (WU RBA), the Violence Exposure Questionnaire (VEQ), and other assessments (Shacham & Cottler, 2010) as well as receiving the SI at their baseline session. The WU risky sexual and drug us ing behaviors, perceptions of risky sexual and drug using behaviors, and demographic information (Needle et al. 1995), while the VEQ, derived from the Conflict Tactic Scale (Strauss, 1979) assessed various forms of current and past violent experiences. Fo llowing their baseline session, the women were randomized into either the SI only group, in which nothing else was required of them, or were randomized in to the PPCMI where they were to complete the 10 week 40 hour peer partnered case management intervent ion. Participants were also interviewed using the same assessments at the 4 and 8 month follow up sessions. For women in the SI, their 4 month follow up intervention was scheduled immediately after baseline interviews, while the 4 month follow up was sched uled immediately after the 10 week PPCMI intervention for the PPCMI group. This study was approved by the Washington

PAGE 91

91 Main Exposures The main exposures in this study were the assigned intervention (PPCMI v s. SI), and socio demographic factors such as: race (black vs. non black), education (high school diploma or higher vs. no high school diploma), unstable housing (living on the streets, with others, shelters etc vs. liv ing in own house or apartment), and a ge (18 29 years of age vs. 30+). Main Outcome SAV A Over time (Baseline, 4 month and 8 month Follow Ups) Violence g the past 4 months, the past 4 months, has anyone hurt you to the point that you had bruises, cuts, broken bones, or otherwise least one of these instances were categorized as having experienced violence in the past 4 months. HIV/AIDS R isk To be considered as being at risk for HIV/AIDS, participants must have reported having at least one risky sexual partner (a partner who is a n injection drug user or has other sexual partners simultaneously) OR 2 or more sex partners in the past 4 months AND 1 or more reported unprotected sex acts in the past 4 months (any unprotected oral, vaginal, or anal sex).

PAGE 92

92 Substance Use To assess recen drug is used and if they reported using a specific substance one or more days in the past 30 days. Recent substa nce use is defined as using any substance (crack/cocaine, marijuana, stimulants, and heroin) at least one time in the past 30 days. SAVA Criteria Based on the above violence, HIV/AIDS risk and substance use variables, a four level variable was created to assess SAVA among the participants. This variable ranged component criteria met). Participan ts who m et al l three criteria (substance use, violence, and HIV/AIDS risk) were categorized as having the SAVA syndemic. Covariates Covariates included in the analysis were: social support (defined as having someone who you can talk to and ask for favors ), number of arrests greater than 25 th percentile of reported arrests in the sample ( 4+ life time arrests vs. less than 4 life time arrests), high religion/spirituality (defined as viewing religion and spirituality as very important, attending religious s ervices regularly, and seeking advice from religious leaders all in the past 12 months vs. low or no religion/spirituality), childhood parental separation (separated 6+ months from parents before the age of 15 vs. no or less than 6+ months from parents bef ore the age of 15 ), and child sexual abuse (CSA) (experienced child sexual abuse before the age of 15 vs. not).

PAGE 93

93 Analysis Multiple Imputations Of the 319 participants in the STOP study at baseline, 261 women completed the 4 month interview and 282 women c ompleted the 8 month interview. To address the issue of missing data, multiple imputation was used. This method generates a set of plausible values for the missing values and allows all available data to be used, thus preserves statistical power while also providing appropriate estimation of standard errors through repeated imputation (imputation number=10). Missing data analyses revealed that separation form at least one parent for 6 mon ths or more ( p value <.02) and arrest history ( p value <.01) were sign ificant predictors of missingness; however, intervention group was not related to missingness (Figure 4 1). Variables related to missingness were included in the imputation model to meet the missing at random (MAR) assumption of multiple imputation. All an alyses were conducted using SAS 9.4. Analysis Technique Multivariate Poisson regressions using generalized estimating equations (GEE), which specified a working correlation structure of autoregressive, was used to estimate the relative risks of correlate s of SAVA over time (meeting all three criteria vs. meeting less than 3 or none), along with individual criterion over time (substance use (any substance use vs. no use), violence (any exposure to violence vs. no exposure), and HIV/AIDS risk (yes vs. no )). To account for the fact that changes in behavior are often pronounced short term and tend to wane as time progresses, the effect of time in these analyses was assessed as non linear. An alpha correction was also implemented to control for multiple testing Because 4 multivariate regression analyses were conducted, only correlates significant at the .0125 level or less were considered significant. In

PAGE 94

94 addition, a sensitivity analyses, using only complete data, revealed negligible differences between regressi on estimates of imputed data and c omplete case analyses (Table 4 5 ). Results Socio Demographic Characteristics In our sample, 71% of the women were African American, nearly half had less than a high school diploma (46%), and around a third were between the ages of 18 to 29 (27%) or were ever married (36%) ( Table 4 1 ). In addition, the women reported a high percentage of unstable housing (76%), child sexual abuse before the age of 15 years of age (51%), and having 4 or more arrests (70%). Around half of the women were randomized to recei ve the SI+PPCMI intervention (51 %), while the others were assigned to the SI only intervention. Exposure to Violence, Substance Use, and HIV/AIDS Risk o ver Time among the Sample Almost half of the women reported using illicit substances in the past 30 days (47%) at baseline, with this percentage reducing to 38% by t he 8 month follow up (Table 4 2 ). The most commonly used substances were crack/cocaine (baseline: 34%, 8 month: 27%) and marijuana (baseline: 29%, 8 month: 22%). E xposure to violence in the past 4 months was consistently high in our sample, though decreases ov er time were evident (Table 4 2 ). The baseline incidence of violence was nearly 60%; however, this decreased to 40% by the 8 month follow up. The most commonly reported instance of violence was emotional abuse which was reported by 52% of the women but was reduced to 36% by the 8 month follow up.

PAGE 95

95 Physical abuse was also common with 19% of the women reporting this at baseline but also reduced at the 8 month follo w up to 10%. Regarding HIV/AIDS risk, nearly half of all the women were categorized as at risk, meaning that they had 2 or more sex partners or at least one risky partner AND had reported at least one instance of an unprotected sex act ( Table 4 2 ). However this number reduced to 27% by the 8 month follow up. The most common risky sexual behavior was unprotected sex acts, which was reported by nearly 70% of the women at baseline and reduced to 55% at the 8 month follow up. SAVA Among the Sample Overall, ab out 20% of the women recently used an illicit substance, experienced at least one incident of violence, and met the criteria for HIV/AIDS risk behaviors in the past 4 months, meaning that they were considered as having the SAVA syndemic; however, the perce ntage of women who were classified as having the SAVA syndemic dropped to 11% by t he 8 month follow up (Table 4 2 ). When we examined SAVA component criteria by intervention group, we found no statistically significant associations (Table 4 3 ). However, th ere was a trend for the women in the PPCMI group to report less exposure to violence than women in the SI. Moreover, though the percentage of people who used any substances slightly decreased among all the women, the frequency of use showed an interesting trend. While the median number of drug uses among those in the PPCMI group who used drugs remained stable or decreased at the 4 and 8 month follow ups, the median number of drug uses among those in the SI group who used drugs increased after baseline. Spec ifically, women in the PPCMI group who reported any substance use at 4

PAGE 96

96 and 8 month post baseline had a median number of use of 15 and 6 respectiv ely, however, this number was 23 and 15 among women in the SI group. Multivariate Poisson Regressions A multiva riate Poisson regression model was used to obtain relative risk estimates on correlates of substance use, violence, HIV/AIDS risk, and o verall SAVA over time (Table 4 4 ). Regarding the substance use model, women who believed at baseline that they had risky drug using behaviors that needed changing were significantly more likely to use substances over time (RR 1.76, 95% CI: 1.37, 2.26 ) However, there was a strong trend between baseline substance use perceptions and time, meaning that these same women tended to be more likely to change their substanc e use behavior over time (4 month RR .74 9 5% CI: .54, 1.01; 8 month RR .72 9 5% CI: .52, 1.02 ). Over all there were no signif icant changes or trends evident in substance use over time meaning that only women who believed they had risky drug using behaviors that needed changing had a trend of decrease d substance use over time Additionally, women who were not high ly religious/spiritual were significantly more likely to use substances over time compared with women who were highly religious/spiritual (RR 1. 3 7 95% CI: 1.07, 1.74 ). Our results show that aside from perceptions of drug using behaviors, women who continued to use substances over time were significantly more likely to be African American (RR 1.64 95% CI: 1.26, 2.14 ) and had greater number of lifetime arrests (RR 1.58 95% CI: 1 .22, 2.06 ) Women who were currently or formerly married had a trend of increased substance use over time (RR 1.24 95% CI: 1 .01, 1.52) however, i ntervention group was not statisti cally correlated with substance use over time.

PAGE 97

97 Though exposure to violence was high among our sample, the risk of experiencing violence over t ime decreased by nearly 20% at the 4 month follow up (RR .81, 95% CI: .71, 9 2 ) and over 30% by the 8 month follo w up (RR .67, 95% CI: .57, .78 ). Interestingly, the strongest correlate of exposure to violence over time was child sexual abuse (RR 1.52, 95% CI: 1.26, 1.83 ). In addition, women with 4 or m ore arrests at baseline (RR 1.30 95% CI: 1.06, 1.58 ) or women who met the criterion for HIV/AIDS risk at baseline (RR 1.49, 95% CI:1.25, 1.78 ) were significantly more likely to report experienced violence over time compared with women with fewer lifetime arrests and women who did not meet the criterion for HIV/AIDS risk at baseline A trend for women in the PPCMI intervention group to be less likely to experience violence over time was evident, however this trend missed statistical significance (RR .89, 95% CI: .77, 1.03 ). An interaction between intervention group and th e 4 and 8 month time points revealed that significant differences between intervention groups and follow up time were not evident. The li kelihood of meeting the criterion of HIV/AIDS risk decreas ed by 29 % at 4 month follow up (RR .71 95% CI: .60, .85 ) and over 40% by the 8 month follow up (RR .58, 95% CI: .48, .70 ). Participants who reported that they had risky sexual behaviors that needed changing were significantly more likely to be at risk for HIV/AIDS over time than women who did not believe they had r isky sexual behaviors that needed changing (RR 1.88, 95% CI:1.51 2.32) However, a significant interaction between sexual risk perception and time was not evident, meaning that women who believed they had risky sexual behaviors that needed changing at base line were just as likely to decrease their risk of HIV/AIDS over time as women who did not. Significant correlates of women who

PAGE 98

98 were at risk for HIV/AIDS over time included baseline exposure to violence (RR 1.36, 95% CI: 1.08 1.72 ). Moreover, there was a t rend for women who had unstable hous ing (RR 1.31, 95% CI: .99 1.75), baseline substance use (RR 1.24, 95% CI: 1.01 1.52 ) and younger women (RR 1.21 9 5% CI: .95 1. 51 ) to be more likely to be at risk for HIV/AIDS over time compared with women with stable h ousing, no substance use at baseline, and older women. I ntervention group was not associated with HIV/AIDS risk over time. When examining the initial baseline prevalence of violence, substance use, and HIV/AIDS risk on the longitudinal outcomes of these is sues, the results showed that women who experienced violence at baseline were more likely to meet the criterion for H IV/AIDS risk over time ( RR 1.36, 95% CI: 1.08 1.72 ). On the other hand, women who met the criterion for HIV/AIDS risk at baseline were more likely to experience violence over time (RR 1.51 95% CI: 1.27 1.79 ). However, baseline violence and HIV/AIDS risk were not significantly associated with substance use over time. Overall, the likelihood of having the SAVA syndemic (experiencing violence, using drugs, and meeting the criterion for HIV/AIDS risk), significantly decreased by nearly 40% by the 8 month follow up (RR .81, 95% CI: .68 .96 ) though a significant change was not evident at the 4 month follow up The likelihood of having the SAVA syn demic over time was greater among women who had experie nced child sexual abuse (RR 1.63 95% CI : 1.07 2.47 ), had 4 or more arrests (RR 1.81, 95% CI : 1.07 3.08 ), and women who believed they had sexual and drug using behaviors that need changi ng at baseline (RR 4.27, 95% CI : 1.61 11.30 ). Additionally, there was a trend for black women to have an increased risk of having the SAVA syndemic over time (RR

PAGE 99

99 1.49, 95% CI: .96 2 .30 ). However, intervention group were not significant correlates of SAVA over time. Discu ssion In this study, we aimed to evaluate the association between a peer partnered case management intervention and decreases in the likelihood of SAVA over time. Though the likelihood of SAVA significantly decreased by the 8 month follow up our results d id not support our hypothesis that the peer partnered case management intervention would be associated with decreases in the likelihood of SAVA over time. This suggests that the decreases in the likelihood of SAVA as well as decreases in HIV/AIDS risk and exposure to violence over time were as pronounced among those in the PPCMI group as it was for those in the SI group. However, there was a trend for women in the PPCMI group to experience less violence and to have a lower median number of times of total su bstance uses than women in the SI group. We did notice that the median number of total substance uses among those who used substances at each time point rose for the SI group following the baseline assessment. The most probable reason may be that women who used substances fewer times at baseline were able to abstain by the 4th and 8 month follow up, thus leaving the heavier using women in the sample. Furthermore, though women in the PPCMI group were to receive up to 40 hours of peer partnered case manageme nt, the vast majority of the women did not complete more than 20 hours of the intervention. Uptake of the intervention may have been made difficult because of the rigorous requirements of drug court. Moreover, the vast majority of the women faced harsh rea lities in fact, nearly 80% of the women did not have a stable place of their own to stay. Such factors, along with substance use and legal

PAGE 100

100 issues may all have contributed to the suboptimal uptake of the PPCMI intervention. These findings suggest that more research on intervention uptake on marginalized and vulnerable populations are needed. Had the women utilized all intervention elements current research supports the hypothesis of a reduction in the likelihood of SAVA. Research by Corsi et al. found that case management reduced the risky drug and sexual behavior in methamphetamine users; while others have found that former female substance users (the background of some pe er partners) and social services where vital to successful completion of drug court (Corsi et al. 2012; Fischer, Geiger, & Hughes, 2007). The second aim of this study was to determine the strength of relationships between violence, substance use, and HIV/ AIDS risk by assessing the effect of the initial prevalence of these issues on longitudinal outcomes. We hypothesized that the initial prevalence of violence, substance use, and violence will be associated with the longitudinal outcomes of these issues, ho wever baseline substance use would have the strongest effect on violence and HIV/AIDS risk over time. Our results found that baseline substance use was only marginally associated with HIV/AIDS risk over time, but was not associated with significant increas ed risk of violence over time. We also found that baseline violence was associated with HIV/AIDS risk over time and vice versa, however the association was slightly stronger in the relationship between baseline HIV/AIDS risk and violence over time. Baselin e violence and HIV/AIDS risk were not significantly associated with substance use over time.

PAGE 101

101 Our third aim was to determine the effect of race, markers of socio economic status such as education and stable housing on SAVA over time. Previous studies have shown that issues related to SAVA are exacerbated in women who are low income, homeless, and lack financial and social support (Blankenship, Reinhardt, Sherman, El Bassel, 2015; Peters et al. 2012; Jennes et al. 2011; Martin et al. 2010; Blankenship et al. 2015; Sharpe et al. 2012). Our hypothesis that race, lower education, and unstable housing at baseline would be associated with an increased likelihood of SAVA over time was largely unsupported. The lack of association may be due to the fact that a l arge proportion of our sample had these characteristics, limiting variability. The lack of association may also be attributed to the fact that these women were in drug court, a criminal justice intervention which offers additional support for vulnerable wo men. Strengths and Limitations Our main limitation in this study was the fact that our sample was not randomly selected, decreasing the generalizability of these results to all females in drug court. Our study also relied on self report data on sensitive q uestions, which can lead to the social desirability bias; specifically, this could lead to the underreporting of risky behaviors. However, our study features a relatively large population of an under researched population and a rich data set with detailed items on substance use, violence, risky sexual behaviors, and perceptions of these behaviors. Conclusion Overall, our findings suggest that involvement in a therapeutic justice program such as drug courts are associated with not only decreases in substanc e use, but also with HIV/AIDS risk behaviors and violence experienced over time. There may be added utility in additional interventions such as the PPCMI. Moreover, women who report that

PAGE 102

102 they have risky sexual and drug using behaviors that need changing ma y benefit from additional intensive interventions to assist in changing their behaviors. Future studies should examine the existence of heterogeneous subgroups of women within the female offender population and evaluate whether changes in drug use, sexual behaviors, and exposure to violence differ by such groups. Figure 4 1. Flow Log on Attrition by Intervention Group (N=319)

PAGE 103

103 Table 4 1. Baseline Socio Demographic Characteristics of Sample (N=319) Socio Demographic Characteristics at Baseline N (% ) African American 225 (71%) Ever married 114 (36%) 18 29 years of age 8 8 (2 8%) Has social support 247 (77%) Less than high school diploma 147 (46%) Child Sexual Abuse 163 (51%) Separated from Parents in Childhood (6+ mos.) 230 (72%) More than 4 arrests 224 (70%) Unstable Housing 243 (76%) High Religion/Spirituality 70 (22%) Recruited from Municipal Drug Court System 281 (88%) Perceived to Have Risky Sexual Behaviors that Need Changing 139 (44%) Perc eived to Have Drug Using Behaviors t hat Need Changing 146 (46%)

PAGE 104

104 Table 4 2 Longitudinal Assessm ent of SAVA Components Over Time SAVA in the Past 4 Months Baseline (N=319) 4 Month Follow Up (N=261) 8 Month Follow Up (N=282) p value 4 Month Follow Up p value 8 Month Follow Up Violence Components Was threatened with a gun No 308 (97 %) 257 (98 %) 278 (99 %) Yes 11 (3%) 4 (2 %) 4 (1 %) .15 .13 Was pressured or forced to participate in sexual acts No 291 ( 9 1 %) 239 (92 %) 272 (96 %) .96 <.01 Yes 28 (9 %) 22 (8 %) 10 (4 %) Emotionally abused No 152 (48 %) 155 (59 %) 181 (64 %) <.001 <.0001 Yes 167 (52 %) 106 (41 %) 101 (36 %) Physically abused (hurt to the point of bruises, cuts, broken bones) No 258 (81 %) 228 (87 %) 255 (90 %) .03 <.001 Yes 61 (19 %) 33 (13 %) 27 (10 %) Attacked with knife, stick, bottle, or other weapon No 290 (91 %) 247 (95 %) 276 (98 %) .07 <.001 Yes 29 (9 %) 14 (5 %) 6 (2 %)

PAGE 105

105 Table 4 2 Continued SAVA in the Past 4 Months Baseline (N=319) 4 Month Follow Up (N=261) 8 Month Follow Up (N=282) p value 4 Month Follow Up p value 8 Month Follow Up Any Violence (1+violence components) No 138 (43 %) 145 (56 %) 177 (63 %) <.001 <.0001 Yes 181 (57 %) 116 (44 %) 105 (37 %) HIV/AIDS Risk Behavior Components Unprotected oral sex (performed) No 163 (51 %) 167 (64 %) 182 (65 %) <.001 <.0001 Yes 156 (49 %) 94 (36 %) 100 (35 %) Unprotected vaginal sex No 118 (37 %) 117 (45 %) 143 (51 %) 02 <.001 Yes 201 ( 63 %) 144 ( 55 %) 139 (49 %) Unprotected anal sex No 291 (91 %) 243 (93 %) 263 (93 %) .34 .26 Yes 28 (9 %) 18 (7 %) 19 (7 %) Any unprotected sex act (vaginal, anal, or oral) No 99 (31 %) 107 (41 %) 126 (45 %) <.01 <. 0001 Yes 220 (69 %) 154 ( 59 %) 156 (55 %)

PAGE 106

106 Table 4 2 Continued SAVA in the Past 4 Months Baseline (N=319) 4 Month Follow Up (N=261) 8 Month Follow Up (N=282) p value 4 Month Follow Up p value 8 Month Follow Up Number of sex partners (2+ vs. less than 2) No 172 ( 54 %) 179 (69 %) 207 (73 %) <.0001 <.0001 Yes 147 (46 %) 82 (31 %) 75 (27 %) Risky partner (likely to be an IDU or have another partner) No 242 (76 %) 215 (82 %) 248 (88 %) .07 <.0001 Yes 77 ( 24 %) 46 (18 %) 33 ( 1 2 %) HIV/AIDS risk behavior (risky partner OR 1+ sex partners AND 1+ unprotected sex act) No 165 (52 %) 175 (67 %) 207 (73 %) <.0001 <.0001 Yes 154 (48 %) 86 (33 %) 75 (27 %) Substance Use (Number of Uses in Past 30 Days) Mariju ana No 227 (71 %) 199 (76 %) 219 (78 %) .05 .04 Yes 92 (29 %) 62 (24 %) 63 (22 %)

PAGE 107

107 p values we re generated from unadjusted models assessing time and SAVA component s Table 4 2 Continued SAVA in the Past 4 Months Baseline (N=319) 4 Month Follow Up (N=261) 8 Month Follow Up (N=282) p value 4 Month Follo w Up p value 8 Month Follow Up Crack/Cocaine No 211 (66 %) 185 (71 %) 205 (73 %) .06 .02 Yes 108 (34 %) 76 (29 %) 77 (27 %) Heroin No 306 ( 9 2 %) 252 (97 %) 272 (96 %) .62 .70 Yes 13 (4 %) 9 (3 %) 10 (4 %) Stimulants No 317 ( 99 %) 259 (99 %) 281 ( 1 00 %) .84 .64 Yes 2 (1 %) 2 (1 %) 1 (0%) Any drug use in the past 4 months No 169 (53 %) 157 (60 %) 176 (62 %) .01 <.01 Yes 150 (47 %) 104 (40 %) 106 (38 %) SAVA No Criterion Met 57 (18 %) 86 (33 %) 111 (39 %) 17 <.01 One SAVA Criterion met 98 (31 %) 81 ( 3 1 %) 86 (31 %) Two Sava Cr iteria met 105 (33 %) 57 (22 %) 33 (19 %) 3 All three SA VA C riteria met 59 (19 %) 37 (14 %) 12 (11 %)

PAGE 108

108 Table 4 3 Longitudinal Assessment of SAVA Component Criterion by Intervention Group *p values were generated from unadjusted models assessing intervention group and individual SAVA criterion over time Baseline 4 month Follow up Missing (58) 8 month Follow up Missing (37) Median # of Times % of Participants Median # of Times % of Participants Median # of Times % of Participants SI N=155 PPCMI N=164 SI PPCMI SI N=132 PPCMI N=129 SI PPCMI SI N=138 PPCMI N=144 SI PPCMI p value Violence 94 (61%) 87 (5 3%) -------68 (52%) 48 (37%) ------56 (41%) 49 (34%) -------.05 Number of Violence Ex periences -------1 1 --------1 1 --------1 1 ----HIV/AIDS Risk 80 (52%) 74 (45%) -------49 (37%) 37 (29%) -------39 (28%) 36 (25%) -------.16 Number of Unprotected Sex --------19 11 --------10 10 --------10 10 ----Number of Sex Partners --------3 3 --------3 2 --------2 2 ----Substance Use 72 (46%) 78 (48%) -------48 (36%) 56 (43%) -------51 (37%) 55 (38%) --------.70 Number of Times Substances used --------12 15 --------23 15 --------15 6

PAGE 109

109 Table 4 4 Correlates of Substance Use, Violen ce, HIVAIDS, and SAVA Over Time Substance Use Over Time Violence Over Time HIV/AIDS Risk Over Time SAVA Over time Variables RR (CI) RR (CI) RR (CI) RR (CI) Black 1.64 (1.26, 2.14 ) .96 (.82 1.13 ) 1.00 (. 82, 1.24 ) 1.49 (.96, 2.30 ) E ver Married 1.24 (1.0 1, 1.52 ) ------------18 29 years of age 1.03 (82, 1.35 ) 1.05 ( 89 1.25 ) 1.19 (.95, 1.48 ) 1.25 (.81, 1.92) Less than high school diploma .9 8 (.81, 1.19 ) .95 (.80 1.12 ) 1.00 (.82, 1.20 ) .89 (.63, 1 .27 ) Child Sexual Abuse ----1.52 (1.26, 1.83 ) ----1.63 (1.07, 2. 47 ) Separated from parents 6+mos .95 (.77, 1.17) 1.13 (.91, 1.41 ) ----1.16 (.72, 1.86 ) Arrested 4+ times 1.58 (1.22, 2.06 ) 1.30 (1.06, 1.58 ) 1.05 (.82, 1.35) 1.81 (1.07, 3.08 ) Intervention Group (PPCMI) .98 (.81, 1.19 ) .89 (.77 1.04 ) .94 (.78, 1.1 5 ) .86 (.59, 1 .25 ) Recruited from Court System .85 (.64, 1.16 ) .97 (.73, 1. 2 9 ) 1.25 (.89, 1.76 ) 1.06 (.64, 1.75 ) Low Religion/Spirituality 1. 3 7 ( 1.07, 1.74 ) ------------Unstable Housing --------1.30 (.99, 1.75 ) ----Baseline Violence 1.07 (.88, 1.3 3 ) ----1.36 (1.08, 1.72 ) ----Baseline Substance Use --------1.24 (1.01, 1.52 ) ----

PAGE 110

110 Table 4 4 Continued Substance Use Over Time Violence Over Time HIV/AIDS Risk Over Time SAVA Over time Variables RR (CI) RR (CI) RR (CI) RR (CI) Baseline HIV/AIDS risk ----1.49 (1.25 1.78 ) ----I have risky drug using behaviors that need changing 1.76 (1.37, 2.26 ) --------.82 (.41, 1 .66 ) I have risky sexual behaviors that need changing --------1.94 (1.55, 2.4 2) .79 (.36, 1.71 ) Time (4 month) 1.02 ( .80, 1.33 ) .81 (.72, 9 2 ) .71 (.60 .85 ) .82 (.60, 1.13 ) Time (8 month) .97 (.74, 1.28) .67 (.57, .78 ) .58 (.49 .70 ) .61 (.43, 87 ) Baseline risky drug use perceptions* Time (4 month) .74 (.54, 1.01 ) ------------Baseline risky drug use perceptions* Time (8 month) .72 (.52, 1.00 ) ------------Baseline risky drug use perceptions* Baseline risky sexu al behavior perceptions -----------4.27 (1.61, 11.30 ) *Table represents 4 separate multivariate models (substance use, violence, HIV/AIDS Risk, SAVA) ----denotes that a variable was not included in the multivariate model due to a lack of signific ance in unadjusted model

PAGE 111

111 Table 4 5 Correlates of Substance Use, Violence, HIVAIDS, and SAVA Over Time Using Complete Data Substance Use Over Time Violence Over Time HIV/AIDS Risk Over Time SAVA Over time Variables RR (CI) RR (CI) RR (CI) RR (CI) Bl ack 1.69 (1.30, 2.20 ) .96 (.82, 1.13 ) .99 (. 80, 1.22 ) 1.47 (.94, 2.30 ) E ver Married 1.21 (.98, 1.50 ) ------------18 29 years of age 1.05 (81, 1.36 ) 1.04 ( 88, 1.24 ) 1.20 (.95, 1.51 ) 1.26 (.81, 1.94) Less than high school diploma .9 7 (.80, 1.18 ) .93 (.79, 1.10 ) .99 (.82, 1.22 ) .90 (.62, 1 .30 ) Child Sexual Abuse ----1.54 (1.27, 1.86 ) ----1.74 (1.14, 2. 66 ) Separated from parents 6+mos .93 (.76, 1.15) 1.10 (.88, 1.37 ) ----1.13 (.70, 1.81 ) Arrested 4+ times 1 .63 (1.24, 2.13 ) 1.30 (1.06, 1.59 ) 1.03 (.80, 1.34) 1.77 (1.01, 3.08 ) Intervention Group (PPCMI) .95 (.78, 1.16 ) .90 (.77, 1.06 ) .94 (.77, 1.1 5 ) .82 (.57, 1 .17 ) Recruited from Court System .85 (.63, 1.14 ) .97 (.73, 1. 2 9 ) 1.25 (.89, 1.75 ) 1.03 (.62, 1.72 ) Low Religion/Spirituality 1.40 ( 1.09, 1.80 ) ------------Unstable Housing --------1.34 (1.00, 1.79 ) ----Baseline Violence 1.08 (.88, 1.3 3 ) ----1.42 (1.12, 1.81 ) ----Baseline Substance Use --------1.24 (1.01, 1.52 ) ----

PAGE 112

112 Table 4 5 Continued Substance Use Over Time Violence Over Time HIV/AIDS Risk Over Time SAVA Over time Variables RR (CI) RR (CI) RR (CI) RR (CI) Baseline HIV/AIDS risk ----1.56 (1.30, 1.86 ) ----I have ri sky drug using behaviors that need changing 1.79 (1.39, 2.30 ) --------.84 (.40, 1.72 ) I have risky sexual behaviors that need changing --------1.96 (1.57, 2.45 ) 80 (.36, 1 .74 ) Time (4 month) .97 ( .75, 1.24 ) .80 (.70, 9 1 ) .70 (.59, .8 2 ) .79 (.58, 1.08 ) Time (8 month) .98 (.75, 1.29) .67 (.57, .78 ) .55 (.46, .67 ) .60 (.42, .85 ) Baseline risky drug use perceptions* Time (4 month) .78 (.58, 1.06 ) ------------Baseline risky drug use perceptions* Time (8 month) .71 (.51, .99 ) ------------Baseline risky drug use perceptions* Baseline risky sexual behavior perceptions -----------4.65 (1.71, 12.65 ) *Table represents 4 separate multivariate models (substance use, violence, HIV/AIDS Risk, SAVA) ----denotes that a variable was not included in the multivariate model due to a lack of significance in unadjusted model

PAGE 113

113 CHAPTER 5 ONE STEP AT A TIME: A LATENT TRANSITIONAL ANALYSIS ON CHANGES IN SUBSTANCE USE, EXPOSURE TO VIOLENCE, AND HIV/AIDS RISK BEHAVIORS AMONG FEMALE OFFENDERS Introduction Females in the Criminal Justice System and Syndemic Theory Currently, females have emerged as the fastest growing prison population, yet are under represented in research (Welty et al. 2016; Millay, Satyanaray Crecelius, & Cottler, 2009). Females in the criminal justice system, including those in drug courts, have been shown to have comorbid issues such as exposure to violence and HIV/AIDS risk behaviors including multiple sex partners and unprotec ted sex (Festinger et al. 2016; Morse et al. 2015; DePesa, Eldridge, Deavers, & Cassisi, 2015; Cosden, Larsen, Donahue, & Nylund Gibson, 2015; Meyer et al. 2015; Messina, Calhoun, & Braithwaite, 2014; Saxena, Messina, & Grella, 2014, Torchalla, Nosen, R ostam, & Allen, 2012; Millay et al. 2009). Specifically, female offenders have been reported to have substance use related problems at a higher rate than male offenders and nearly 10 times higher than non offending women (Saxena, Messina, & Grella, 2014). Research by Cosden and colleagues (2015) found that a lifetime prevalence of comorbid trauma and substance use has been reported by 80 90% among in treatment individuals, with females more likely to report physical and sexual victimization and more likely to have severe symptoms of trauma when compared with males. In a recent meta analysis by Gilbert and colleagues (2015), the relationship between substance use and violence was described as intricate and bidirectional. Moreover, females involved in the cr iminal justice system have been shown to be up to 15 times more likely than non incarcerated females and almost twice as likely as

PAGE 114

114 incarcerated males to test positive for HIV (Meyer et al. 2015; Lichtenstein & Malow, 2010). The intersectionality of substa nce use, violence, and HIV/AIDS risk behaviors, termed the SAVA syndemic, are known to be synergistic and mutually re enforcing, and the most common pathway to initial criminal justice involvement among females, with the continuation of these issues linked with increased odds of recidivism in this population (Meyer et al. 2015; Abad et al. 2015; Meyer, Springer, & Altice, 2011; Lichtenstein & Malow, 2010; Singer, 2009; Singer, 2006). Trans theoretical Model Stages of Changes A theoretical framework that has been widely used in understanding behavior change and may also help understand changes in addictive and co occurring issues such as violence and HIV/AIDS risk behaviors is the Trans theoretical model, or simply known as the Stages of Change model (Sera fini, Shipley, & Stewart, 2016; Proeschold Bell et al. 2016; Gold et al. 2016; Abad et al. 2015; Prochaska, DiClemente, & Norcross, 1992). In this model, Prochaska, DiClemente, & Norcross (1992) suggest that there are 4 stages of change through which in dividuals move to make changes in their behaviors: 1) the pre contemplation stage, where individuals are unaware of the necessity to change 2) the contemplation stage, where individuals are aware of behavior that need to change but have not made any defini tive decision to change, 3) the action stage, where individuals are considerably modifying their behaviors for up to 6 months and 4) the maintenance stage where the prime focus is relapse prevention. Theories of health behavior change suggest that change o ccurs in stages; thus, new analysis techniques such as the latent transitional analysis (LTA), a longitudinal extension of a latent class analysis (LCA), can serve as a means to quantify discrete stages of behavior change, through the estimation of the pro bability of transitions

PAGE 115

115 among those in observed latent statuses (Corsden et al. 2015; Lanza & Collins, 2008; Lanza, Patrick, & Maggs, 2010). A person centered approach for modeling behavior profiles such as LTAs, classifies multiple dimensions of behavior to aggregate individuals with common because individuals may change membershi p in latent statuses over time, a contrast from the stagnant class memberships in LCA (Lanza, Patrick, & Maggs, 2010). Simply, this analysis is appropriate for answering questions regarding the behavioral profiles of individuals who are more likely to chan ge over time (Roberts & Ward, 2011). LTA is particularly useful as it can handle complex interactions among multiple dimensions, which allows for the identification of mutually exclusive classifications. Such analyses may aid in the development of targeted interventions for higher risk statuses. Gaps in Knowledge Though women in the criminal justice system have the highest rates of substance use related problems, this high risk group of women is often excluded from large scale epidemiologic studies (Welty e t al. 2016). Furthermore, within the literature on substance use and criminal justice involved individuals, studies often do not decipher differences in types of substances used though it is evident that different drugs have different etiologies, for exam ple marijuana and crack/cocaine (Welty et al. 2016). Prior analyses among a sample of women in the criminal justice system have reported Significant variability in the pre valence of the health and social issues such as violence

PAGE 116

116 and HIV/AIDS risk among substance users have also been shown to differ by type of drug used (Welty et al. 2016; Jones et al. unpublished). In addition, extant literature has shown that women with h istory of childhood traumas, such as CSA are at increased odds of victimization, substance use, and risky sexual behaviors in adulthood (Morse et al. 2015; Meyer, Springer, & Altice, 2011, Millay et al. eir risky drug using and sexual behaviors may also be indicative of their latent status memberships and should also be explored. Research by Robertson et al. (2012) found that Drug Court participants perceived that their risk of HIV/AIDS as low, though sub stance use and risky sexual behaviors are often co occurring and synergistic. Lastly, female substance users involved with additional interventions such as case management have been shown to significantly reduce their substance use over time (Corsi et al. 2012). Therefore, the aims of these analyses are to: 1) identify latent statuses of women based on substance use, exposure to violence, and risky sexual behaviors at baseline, 2) examine the proportion of women in each latent status at the baseline, 4 mon th follow up, and 8 month follow up and the probability of each transitioning to a lower risk status over time, 3) assess the effect of intervention status on latent status transitions and 4) evaluate the association between socio demographic characteristi cs, child sexual abuse, drug use perceptions, and initial latent status membership. We hypothesize that 1) several latent statuses of women will be identified, particularly a latent status characterized by a high probability of substance use, exposure to v iolence, and risky sexual behaviors, 2) individuals in latent statuses characterized by a high probability of crack/cocaine use will be less likely to transition to lower risk statuses over

PAGE 117

117 time compared to those in statuses with low probabilities of crack /cocaine use, 3) women randomized to a peer partnered case management intervention (PPCMI) group will be more likely to transition to a lower risk status than those not in a case management in group, but a standard intervention (SI) group, and 4) women who report that they have risky drug using behaviors, have experienced child sexual abuse, older women, and have more lifetime arrests will have elevated odds of being in latent statuses categorized by high probabilities of substance use, exposure to violence and HIV/AIDS risk compared to women without these characteristics. Methods Sisters Teaching Options for Prevention Participants in this analyses were from the Sisters Teaching Options for Prevention project (STOP) (R01NR09180, PI: Cottler), a randomized controlled field study which aimed to reduce high risk drug and sexual behaviors among female in drug court by using a two arm behavioral intervention. Specifically, each woman received standard intervention (SI) which consisted of the National Institute o n Drug Abuse (NIDA) standard pre and post HIV test counseling (Johnson et al. 2011). Half of the participants were randomized to the Peer Partnered Case Management Intervention (PPCMI) in which they received up to 40 hours of further assistance in accessi ng and utilizing needed health services, and had the social support of a peer mentor (Johnson et al. 2011). Outreach and Recruitment Participants in the STOP study, who were drug court enrollees at a Municipal Drug Court System in the Midwest, were recru ited through informational flyers which provided details on the study. In order to be eligible for STOP, participants had to be

PAGE 118

118 being present in court, be at least 18 years of age, have no known cognitive disability, and must have provided informed consent Women who were interested and met the eligibility criteria for STOP were then scheduled for their initial baseline assessments. All participants were interviewed regarding their substance use and sexual behaviors, as well as their exposure to violence in the past 4 months at baseline and the 4 and 8 month follow ups. Measures Validated measures such as the Washington University Risk Behavior Assessment (WU RBA) and the Violence Exposure Questionnaire (VEQ) were used to exposure to violence (Shacham & Cottler 2010). The WU behavior including risky sexual and drug using behaviors, perceptions of various sexual and drug using behaviors, and socio de mographic information (Shacham & Cottler 2010). The VEQ assessed various forms of current and past exposures to violence. Participants were also interviewed using the same assessments at the 4 and 8 month follow up sessions. This study was approved by the Washington University of St Louis Institutional Review Board. Main Exposures In this study, we assess factors such as the effect of the randomized intervention (PPCMI vs. SI) and socio demographic factors such as: race (black vs. non black), age (18 29 ye ars of age vs. 30+), child sexual abuse (yes vs. no), and number of lifetime arrests (categorized as 4+ life time arrests vs. less than 4 life time arrests). Additionally, we included a covariate of a potentially high risk group of women: those who believe d they had risky drug using behaviors that needed changing (believing you have risky

PAGE 119

119 drug using behaviors that need changing vs. no risky drug using behaviors that need changing). Main Outcome Substance Use, Violence, and HIV/AIDS Risk (Indicator Items) Vi olence 3) anyone hurt you to the point that you had bruises, cuts, broken bones, or otherwise instances were categorized as having experienced violence in the past 4 months; howev er, exposure to violence was categorized into two variables: 1) experienced emotional abuse and 2) being attacked with a weapon or experiencing physical and sexual abuse. These variables were created in order to assess whether latent statuses of women diff ered by types of violence experienced. HIV/AIDS R isk These items were: 1) having at least one risky partner, which is a partner who is an injection drug user or a recent partner who has other sexual partners simultaneously, 2) multiple sex partners, defi ned as having 2+ sex partners and 3) any unprotected sex (1+ reported unprotected sex acts), which included any unprotected oral, vaginal, or anal sex in the past 4 months.

PAGE 120

120 Substance Use Recent substance use was defined as using any substance (crack/cocai ne, marijuana, stimulants, and heroin) at least one time in the past 30 days. From prior analyses, we know that in this sample, virtually all participants who used any substances used either crack/cocaine or marijuana almost exclusively (Jones et al. unpu blished). Because crack/cocaine users have been shown to have worse outcomes in drug court, in this analysis, substance use was represented by two variables, 1) crack/cocaine use and 2) the use of other drugs (virtually all participants who used marijuana only). Analysis Analytical Technique LTA analyses work by first identifying key latent statuses of behavior from various multi dimensional items. In this analysis, our indicator items (substance use, violence, and HIV/AIDS risk variables from which latent statuses are derived) reflect an overarching latent theme of the SAVA syndemic. The LTA model estimates 3 sets of parameters: latent status membership probabilities, transitional probabilities, and item response probabilities (Lanza & Collins, 2008). The latent status probabilities estimate the proportion of individuals that is expected to belong to each latent status at each time period. The item response probabilities estimate the agreement of the specific indicators of the latent variable and latent sta tus membership, while the transitional probabilities estimate the probability of changing one latent status to another latent status at the next time period. Once latent statuses are established, predictors of behavior profiles are then used to further un derstand the phenomenon of the underlying behaviors studied and the

PAGE 121

121 individuals at highest risk (Lanza, Patrick, & Maggs, 2010). Multinomial logistic regression was used to predict latent statuses at baseline, while intervention status was used to predict latent status changes over time. Loss to Follow Up Of the 319 participants in the STOP study at baseline, 261 women completed the 4 month follow up interview, while 282 women completed the 8 month follow up interview. The Proc LTA procedure allows for miss ing values in indicator items and analyzes the data under the missing at random assumption. However, this procedure does not allow for missing values in covariates, therefore, two participants who refused to report whether they experienced child sexual abu se were excluded. Thus, our final sample size consisted of 317 women at baseline, 259 at the 4 month follow up, 280 at the 8 month follow up All analyses were conducted using SAS 9.4. Results Descriptive Statistics of Sample and Indicator Items Our sample consisted of a higher number of women who were black (71%) and were 30 years of age or older (73%) (Figure 5 1). The majority of the women in our sample experienced child sexual abuse (51%), were arrested 4 or more times (70%), while 46% believed they had risky drug using behaviors that needed changing. Half of the women were randomized to receive either the SI or the SI+PPCMI intervention. Descriptive analyses of our indicator items show that all items for risky sexual behaviors, violence, and substance u se were substantially prevalent at baseline, however all behaviors decreased over time (Table 5 1).

PAGE 122

122 Model Fit An initial LTA analysis, void of covariates, was modeled to examine the most appropriate number of latent statuses which produced the ideal model fit and parsimony (Table 5 2). We identified the number of latent statuses representing distinct behavioral profiles most appropriate by comparing model fit statistics such as the likelihood ratio G2, Akaike information criteria (AIC) (Akaike, 1974) and th e Bayesian information criteria (BIC) (Schwarz, 1978) of varying numbers of latent statuses. A model with 4 latent statuses produced the smallest AIC, BIC, G statistic and was also the most interpretable, suggesting that a model with 4 latent statuses was the most appropriate. In an attempt to avoid small cell sizes, latent statuses greater than 4 were not considered. Item response probabilities The item response probabilities (Table 5 3) identified 4 distinct behavioral profiles (statuses) in the sample a t baseline Those in Status 1 were characterized by a high probability of risky sexual behaviors such as unprotected sex (.86), multiple sex partners (.91), and moderate probability of risky partners (.47 ). Those in Status 1 also had a high probability of emotional abuse (.81) crack/cocaine use (.88), and a relatively high probability of experiencing violent acts (.46) and using drug s other than crack/cocaine (.50) compared to other statuses. Individuals in Status 2 w ere characterized by a very high probabi lity of crack/cocaine use (.95) and moderately high probability of unprotected sex (.53). However, since th os e in Status 2 had very low probabilities of multiple sex partners and risky partners, they were not considered as having risky sexual behaviors, th ese women were most likely married or were in monogamous relationships. Individuals in Status 3 were characterized by a high

PAGE 123

123 probability of unprotected sex (.77), being emotionally abused (.69), and a moderately high probability of having multiple sex part ners (.53). Women in this status also had a low probability of any substance use. Lastly, individuals in Status 4 were categorized by low probabilities of risky sexual behaviors, exposure to violence, and substance use. Regarding the proportion of partici pants in the latent statuses (Table 5 3), 21% were in Status 1 at baseline, followed by 18% in Status 2 37% in Status 3 and 24% in Status 4 The proportion of those in Status 1 reduced in size over time (17% at the 4 month follow up, 16% at the 8 month f ollow up), as well as the proportion of those in Status 3 (25% at the 4 month follow up, 14% at the 8 month follow up). The proportion of those in Status 4 increased in size over time (42% at the 4 month follow up, 56% at the 8 month follow up), however, t he proportion of women in Status 2 remained relatively stable over time (16% at the 4 month follow up, 14% at the 8 month follow up) (Figure 5 2). Latent Status Transitions from Baseline to 4 Month Follow Up The results from the transitional probabilities that is, the likelihood of a woman transitioning from one status to another at the next follow up time, revealed that the majority of women were likely to remain in their prior status (Table 5 3). Among those in Status 1 at baseline, 6 9% were likely to r emain in the same status at the 4 month follow up. Of those who changed latent statuses, 19% were likely to transition to Status 3 10% were likely to transition to Status 4 and 2% were likely to transition to Status 2 Of those in Status 2 at baseline, 6 7% were likely to remain in this status at the 4 month follow up, while those who transitioned were likely to transition exclusively into Status 4 (33%). Among those in Status 3 at baseline, 57% were likely to remain in this status at the 4 month follow up 30% were likely to transition to Status 4 8% were likely to

PAGE 124

124 transition to Status 2 and 5% were likely to transition to Status 1 Of those who were in Status 4 at baseline, 96% were likely to remain in this status at the 4 month follow up, while 4% were likely to transition to Status 1 Latent Status Transitions from 4 Month to 8 Month Follow Ups Regarding latent status transitions from the 4 month follow up to the 8 month follow up, the majority of all participants in the Status 1 were likely to remain in this status at the 8 month follow up (84%), with 5% and 11% being likely to transition into Status 3 and Status 4 respectively. Of those in Status 2 at the 4 month follow up, 71% were likely to remain in this status at the 8 month follow up, while 28% a nd 3% of individuals were likely to transition to Status 4 and Status 1 respectively Only 50% of the individuals in Status 3 at the 4 month follow up were likely to remai n in this status at the 8 month follow up. The vast majority (46%) were likely to tr ansition in to Status 4 while only 4% were likely to tra nsition to Status 1 Among those in Status 4 at the 4 month follow up, 91% were likely to remain in this status at the 8 month follow up, with the majority of those who transitioned being likely to tr ansition to Status 3 (5%) and Status 2 (6%) Correlates of Latent Statuses at Baseline Using a multinomial logistic regression, significant differences among individuals in the latent statuses were evident with the exception of race and arrests (Table 5 4) Results revealed that those in Status 1 and Status 2 were significantly older than those in Status 4 (AOR: Status 1 (.84), Status 2 (.26)) while those in Status 3 were nearly 2.23 times more likely to be younger than individuals in Status 4 Childhood t rauma also significantly differed by latent statuses, with those in Status 1 and Status 3 being 3.10 and 3.08 more likely to experience child sexual abuse than those in Status 4 Individuals

PAGE 125

125 in Status 2 were as likely as those in Status 4 to experience chi ld sexual abuse. Participants in all statuses were significantly more likely to believe they had risky drug using behaviors that needed changing compared to those in Status 4 (AOR: Status 1 (16.04), Status 2 (5.71), Status 3 (3.75 )). Lastly, though race an d lifetime arrests were not significant at the .05 level, trends were evident among the latent statuses. There was a trend for black women to be more likely in Status 2 and less likely to be in Status 1 and 3 then Status 4 while there was a trend for wome n with more lifetime arrests to be in Status 1 and Status 2 than Status 4 Discussion In this analysis, our first aim was to explore latent statuses of women based on substance use, exposure to violence, and risky sexual behaviors at baseline. We hypothesi zed that several latent statuses of women will be identified, particularly a latent status characterized by a high probability of substance use, exposure to violence, and risky sexual behaviors. The results of this analysis supported this hypothesis as dis tinct behavioral profiles indicating a latent status characterized by a high probability of substance use, exposure to violence, and risky sexual behaviors (Status 1), a latent status characterized by a high probability of crack/cocaine use only (Status 2) a latent status characterized by a moderately high probability of emotional abuse and risky sexual behaviors (Status 3), and a latent status characterized by a low probability of substance use, exposure to violence, and risky sexual behaviors at baseline were observed Our second aim of this analysis was to examine the proportion of individuals in each latent status at the baseline, the 4 month follow up, and the 8 month follow up and the probability of transitioning to lower risk statuses over time. We hypothesized that

PAGE 126

126 individuals in latent status characterized by a high probability of crack/cocaine use will be less likely to transition to lower risk statuses over time compared to those in status with a low probability of crack/cocaine use. Our results also supported our hypothesis, though the proportion of those in Status 3 substantially decreased at every follow up (37% at baseline to 14% at the 8 month follow up), such substantial decreases in the S tatus 1 and S tatus 2 were not evident. However, thou gh a sizeable amount of individuals in S tatus 1 at baseline transitioned to a lesser risk status, 84% of those in this latent status at the 4 month follow up remained at the 8 month follow up. This suggests that there may be a potential window of opportuni ty among those in Status 1 whereas if change does not occur fairly soon, these individuals may need significantly more time to modify their behaviors. In contrast, nearly 30% of those who were in Status 2 at baseline and 4 month follow ups transitioned, a nd virtually all transitioned into Status 4 The relatively stable proportion of individuals in Status 2 over time presumably reflects the proportion of individuals relapsing, thus transitioning into this status at the various follow up times. For example, nearly 10% of the women in the sizeable baseline Status 3 transitioned to Status 2 by the 4 month follow up. Moreover, t he addictive nature of crack/cocaine suggests that the large reduction seen among those in Status 3, the most transient status, may be attributed to their low probability of crack/cocaine use. These findings suggest that risky sexual behaviors may be easier modified than drug using behaviors in our sample. We also aimed to assess the effect of intervention status on latent status transit ions, however, there was no variation to be explained between the statuses, meaning that those in t he standard intervention only w ere just as likely to transition to

PAGE 127

127 the low risk status as those in the PPCMI intervention status. The lack of difference betw een the intervention groups may be attributed to suboptimal utilization of the case management intervention. Uptake of the intervention may have been made difficult by the demands of a rigorous justice intervention such as drug courts and the complexities of the day to day life that many of these women live. Additionally, research by DePesa (2015) and colleagues found that interventions to date tend to have small effects on risky sexual behaviors among female substance users. Our last aim was to evaluate di fferences in the association between socio demographic characteristics, child sexual abuse, drug use perceptions, and initial latent status membership. Our results supported our hypothesis that women who report that they have risky drug using behaviors, ha ve experienced child sexual abuse, and older age will have elevat ed odds of being in statuses with higher probabilities of substance use, exposure to violence, and risky sexual behaviors compared to women who did not believe they had risky drug using behav iors that needed changing, did not experience child sexual abuse, or were younger. Our analyses also suggest the need for trauma informed interventions among females involved in the criminal justice system as other studies have concluded (Cosden et al. 2015; Messina, Calhoun, Braithwaite, 2014; Saxena, Messina, & Grella, 2014; Torchalla et al. 2012). Prior research has found that among individuals in substance use treatment, higher relapse rates were evident among women with a history of trauma compare d to men who experienced trauma (Cosden et al. 2015). Trauma informed interventions allow individuals to learn about how to recognize and manage the impact of trauma in their lives while in substance use treatment (Cosden et

PAGE 128

128 al. 2015). In our sample, chi ld sexual abuse was a prime predictor of latent statuses characterized by risky sexual behaviors and victimi zation in adulthood, namely Status 1 and Status 3 The complexity of the issues of SAVA and the hurdles such as unemployment and unstable housing t hat disproportionately plague the everyday of women in drug court may also play a significant role in the sub optimal changes often seen (Jones, unpublished; Morse et al. 2015). In qualitative studies of drug court enrollees staff, community partners, as well as incarcerated females, participants consistently noted the difficulty for females to attain sobriety while facing issues related to domestic violence and the responsibilities of being the primary care takers of children (Morse et al. 2015, Millay et al. 2009). Limitations and Strengths Proper interpretation of the results of this study cannot be made without addressing associated limitations. The main limitation with this study is that participants were not selected at random, thereby limiting the generalizability of our results. In addition, reliance on self report data on sensitive topics such as SAVA and child sexual abuse may lead to the underreporting of such issues. However, there are several strengths of this study including: a relatively la rge sample size of a hard to reach and under represented population, longitudinal data, and detailed items on SAVA. To our knowledge, this is the first study to explore and quantify sequential changes in risky sexual behavior, exposure to violence, and sub stance use among women involved in the criminal justice system.

PAGE 129

129 Conclusion The results of our analyses showed distinct behavioral patterns among women in drug court ranging from a high probability of substance use, exposure to violence, and risky sexual b ehaviors to low probabilities of these factors. Though the proportion of women in the lowest risk status (Status 4) increased substantially over time, the proportion of women in latent statuses characterized by a high probability of crack/cocaine use (Stat us1 and Status 2) did not substantially decrease over time. Our analyses suggest a wide spread need for trauma informed interventions among females involved in the criminal justice system, as well as targeted interventions tailored to crack/cocaine users. Future studies should delve in the association between crack/cocaine use, the impact of SAVA syndemic, and criminal justice outcomes.

PAGE 130

130 Figure 5 1 Prevalence of Latent Statuses Over Time (N=317)

PAGE 131

131 Table 5 1. Socio demographic Characteristics of Participants at Baseline (N=317) Socio Demographic Characteristics at Baseline N (% ) Race Black 225 (70 %) Non Black 94 (30%) Age 18 29 Years of Age 8 7 (27 %) 30+ Years of Age 230 (73% Child Sexual Abuse No 154 (49%) Yes 163 (51%) Arrest History 4 or More A rrests 224 (71 %) Less Than 4 Arrests 93 (29%) Perc eived to Have Drug Using Behaviors t hat Need Changing No 172 (54%) Yes 145 (46%) Intervention Group SI 154 (49%) SI+PPCMI 163 (51%)

PAGE 132

132 Table 5 2 Descriptive Statistics of Variables in Latent Transition Analysis (LTA) Indicators Items of Latent Status Code Label Baseline N=317 Frequency (%) 4 months N=259 Frequency (%) 8 month N=280 Frequency (%) Risky Sexual Behavior Items Unprotected Sex 1 No 97 (31%) 105 (41%) 124 (44%) 2 Yes 220 (69%) 154 (59%) 156 (56%) Multiple Sex Partners 1 No 171 (54%) 178 (69%) 206 (74%) 2 Yes 141 (46%) 81 (31%) 74 (26%) Risky Partner 1 No 241 (76%) 213 (82%) 246 (88%) 2 Yes 76 (24%) 46 (18%) 33 (12%) Violence Items Emotionally Abused 1 No 151 (48%) 153 (59%) 179 (64%) 2 Yes 166 (52%) 106 (41%) 101 (36%) Violent Acts 1 No 235 (74%) 210 (81%) 247 (88%) 2 Yes 82 (26%) 49 (19%) 33 (12%) Substance Use Items C rack/cocaine Use 1 No 210 (66%) 184 (71%) 204 (73%) 2 Yes 107 (34%) 75 (29%) 76 (27%) Other Drug Use 1 No 223 (70%) 193 (75%) 210 (75%) 2 Yes 94 (30%) 66 (25%) 70 (25%) Table 5 3 Model fit information used in selecting the LTA model 2 3102.75 2097132 3140.75 3212.29 3 2929.74 2097116 2999.74 3131.52 4 2794.40 2097096 2904.40 3111.49

PAGE 133

133 Table 5 4 Item Response Probabilities of Indicator Items Item Response Probabilities Stat us 1 Status 2 Status 3 Status 4 Risky Sexual Behavior Items Unprotected Sex .86 .53 .77 .45 Multiple Sex Partners .91 .13 .53 .10 Risky Partner .47 .11 .28 .03 Violence Items Emotionally Abused .81 .31 .69 .16 Violent Acts .46 .12 .35 .01 Substance Use Items Crack/cocaine Use .88 .95 .00 .00 Other Drug Use .50 .46 .27 .10 Table 5 5 Transitional Probabilities of Latent Statuses Transitional Probabilities Status 1 Status 2 Status 3 Status 4 Baseline (Rows) 4 month Fol low Up (Columns) Status 1 .69 .02 .19 .10 Status 2 .00 .67 00 .33 Status 3 .05 .08 .57 .30 Status 4 .04 .00 .00 .96 4 months Follow Up (Rows) 8 month Follow Up (Columns) Status 1 .84 .00 .05 .11 Status 2 .03 .71 00 .26 Status 3 .04 00 .50 .46 Status 4 .00 .06 .05 .91

PAGE 134

134 Table 5 6 Predictors of Latent Statuses (N=317) p value Status 1 Status 2 Status 3 Status 4 (Reference) Race Black vs. Non Black .10 .63 1.83 .59 1.0 Age Less than 30 yrs. Vs. 30+ <.001 .84 .26 2.23 1.0 Childhood Trauma CSA vs. None <.001 3.10 1.07 3.08 1.0 Lifetime Arrests 4+ Arrests vs. Less than 4 .11 2.43 2.52 1.27 1.0 Perception of Drug Using Behavior Drug Using Behavior that Needs Changing <.001 16 .04 5.71 3.75 1.0

PAGE 135

135 CHAPTER 6 SUBSTANCE USE, VICTIMIZATION, HIV/ AIDS RISK, AND RECIDIVISM AMONG FEMALES IN A THERAPE UTIC JUSTICE PROGRAM Introduction Substance Use Among Females in Criminal Justice System With around 707 individuals per 100,000 residents incarcerated, the United States has the highest incarceration rate in the world (Welty et al. 2016; Meyer, Cepeda, Taxman, & Altice, 2015; Konecky, Cellucci, & Mochrie, 2016; Lichtenstein & Marlow, 2010). Furthermore with over 3 million women arrested each year, women are the fastest growing prison population in the United States (Hall, Golder, Conley, & Sawning, 2013; Scott, Grella, Dennis, & Funk, 2014; Abram, Teplin & McClelland, 2014; Greier, Law, & Brown, 2014; McGee, Baker, Davis, Muller, & Kelly, 2014). Of these women, many are substance users (Gearon, Kaltman, Brown, & Bellack, 2014; Abram, Teplin & McClelland, 2014; Staton Tindall, Harp, Winston, Webster, & Pangrum, 2015). Though substance use and other related problems are common among women in the criminal justice system, data regarding these issues among this population is minimal Cottler, 2009). The Emergence of Drug Courts Because the dramatic increase in cri minal justice involvement is fueled by drug related crimes, therapeutic justice interventions such as drug courts have emerged and are increasing in popularity (Festinger, Dugosh, Kurth, & Metzger, 2016; Morse, Silverstein, Thomas, Bedel, & Cerulli, 2015; Liang, Knottnerus, & Long, 2015; Konecky, Cellucci, & Mochrie, 2016). Drug courts, which have been shown to be effective in reducing substance use among participants, incorporate community based substance

PAGE 136

136 use treatment services, along with other social ser vices to help achieve sobriety and improve the social and economic well being of participants (Festinger et al. 2016; Morse et al. 2015). Drug courts also institute a penal system where non compliant participants are sanctioned with monitoring systems, s hort term incarceration, and other legal consequences, while compliant participants are given incentives as a means to elicit behavior change and compliance (Liang, Knottnerus, & Long, 2015; Morse et al. 2015; Mitchell, Wilso n, Eggers, & MacKenzie, 2012). Recidivism A prime goal of the crimin al justice system is to reduce recidivism among previous offenders; however, the same individuals often return to the criminal justice system female offenders return to prison within three years of release and 66% of incarcerated women are repeat offenders (McGee et al. 2014). Current research recognizes the important interplay of gender in criminal justice involvement, wherein pathways into th e criminal justice system differs by gender (Fries, Fedock, & Kubiak, 2014; Greiner, Law, & Brown, 2014; McGhee et al. 2014). While the most common pathways into the criminal justice system for men stems from using violence as a mean to control and associ ating masculinity with criminal behaviors, the most common pathways for women stem from substance abuse and often comorbid factors such as risky sexual behaviors such as sex trading and victimization (Scott, Grella, Dennis, & Funk, 2014; Greier, Law, & Bro wn, 2014; Fries, Fedock, & Kubiak, 2014). The often synergistic and mutually re enforcing factors of substance use, exposure to violence, and risky sexual behaviors that can lead to HIV/AIDS, termed the SAVA syndemic, is also prevalent among women in the c riminal justice system (Singer, 1996; Meyer, Springer, & Altice, 2011). Because

PAGE 137

137 of the varying pathways to the criminal justice system for men and women, gender specific interventions and services are needed (Fries, Fedock, & Kubiak, 2014). It is of specia l importance that we address factors associated with women in the criminal justice system, as maternal incarceration has been shown to be the strongest predictor et al. 2014) Gaps in Knowledge Though some studies have shown substance use to strongly predict recidivism more research in this area is still needed (Staton Tindall et al. 2015; Gallagher et al. 2015). Stanton Hill and colleagues (2015) argue that substance use, as with other mental health disorders, has been traditionally under represented in studies examining recidivism due to changes in use over time, difficulties regarding measurement, and the view of individual substance use as inferior to public safety and criminal risk. Drug courts may provide an avenue to conduct much needed research on this population. In a study of female drug court enrollees, DuBois et al. found that the vast majority of the women met the DSM IV criteria for cocaine abuse or dependence in the prior research has found that drug court participants who reported crack/cocaine as their drug of choice had nearly 2.5 times the odds of termination from the program than those who re ported other drugs (Gallagher et al. 2015). A prior analysis on female drug court enrollees found that subgroups of women characterized by high probabilities of substance use, victimization, and HIV/AIDS risk (the SAVA syndemic) or a high probability of r ecent crack/cocaine use alone were less likely to transition to a low risk group (characterized by low probabilities of substance use, victimization, and HIV/AIDS risk) over time

PAGE 138

138 compared to those who did not use crack/cocaine and did not have the SAVA sy ndemic (Jones, unpublished). Other studies have also shown that child sexual abuse, race, sex trading, socioeconomic status, age, prior arrest history, and unstable housing were also linked with significantly higher odds of recidivism in women (McGee et a l. 2014; Staton Tindall et al. 2015; Gallagher et al. 2015; Scott et al. 2014). Overall, a strong need for gender specific interventions to prevent recidivism and substance use is evident, however such interventions with criminal justice system populat ions are rare (Hall et al. 2014; McGhee et al. 2014). Because women involved in the criminal justice system who are recent crack/cocaine users or have the SAVA syndemic have been shown to be less likely to change their high risk behaviors than women wh o do not use crack/cocaine and do not have the SAVA syndemic, it is important to assess types of offenses correlated with crack/cocaine use and SAVA, as well as explore factors that may potentially lead to improved criminal justice outcomes among this popu lation. Thus, this analysis examines: 1) the association between crack/cocaine use at baseline (compared with no crack/cocaine use) and type of offense by an 8 month follow up (at least one felony charge, at least one misdemeanor or municipal violation bu t no felony charge, or no offense ), 2) the association between crack/cocaine use at baseline (compared with no crack/cocaine use) and the number of felonies and misdemeanor/municipal violations at an 8 month follow up, 3) the association between SAVA (any substance use, AND being exposed to violence, AND having HIV/AIDS risk behaviors) at baseline and type of offense at an 8 month follow up (at least one felony charge, at least one

PAGE 139

139 misdemeanor or municipal violation charge but no felony charge, or no offens e ), 4) the association between SAVA (compared with no SAVA) at baseline and the number of felonies and misdemeanor/municipal violations at an 8 month follow up. Additionally, we aim to explore other structural and social factors that have been traditionall y and distinctly linked with incarceration among women such as intervention group, race, unstable housing, and socioeconomic status (Fries, Fedock, & Kubiak, 2014). Methods Outreach and Data Collection Our sample of women were derived from the Sisters Teac hing Options for Prevention project (STOP) (N=319), a National Institute of Nursing Research (NINR) funded randomized field study (R01NR09180, PI: Cottler). STOP research staff recruited women mainly from a Municipal Drug Court System in the Midwest by dis tributing informational flyers which outlined details of the study. Women at least 18 years of age, present in court, with no cognitive disability, and interested were then scheduled to be interviewed by research staff. Around 12% of the sample were also r ecruited from community treatment centers and other judicial programs (e.g. half way houses). All STOP participants received a standard intervention (SI) which consisted of the National Institute on Drug Abuse (NIDA) standard pre and post HIV test counseli ng. Around half of the women in the STOP study were also randomized into a Peer Partnered Case Management Intervention (PPCMI), where they received up to 40 hours of case management in addition to the SI (Johnson et al. 2011). All participants were interv iewed using the Washington University Risk Behavior Assessment (WU RBA) (Shacham & Cottler 2010). The WU Assessment, assessed crack/cocaine use, and socio demographic information. This

PAGE 140

140 study was approved by the Washin gton University of St Louis Institutional Review Board. C omprehensive Criminal Justice Records and Recidivism Access to official criminal justice data was made possible due to a partnership between the Presiding Judge of the St. Louis City Municipal Court Judge James Sullivan, and the Washingt on University investigators (Reingle et al. 2012). C omprehensive criminal justice records were garnered from three Criminal Justice Data Banks Summons and arrest files on local municipal violations were derived dire ctly from the Regional Justice Information System (REJIS) which, in partnership with the government entity the Bureau of Justice Statistics, is a leading source of criminal justice data and is used for current criminal justice related analyses (Frandsen, Naglich, Lauver, & Lee, 2013; Reingle et al. 2012). Moreover, the REJIS Municipal Court files and manual files, which is maintained by the City of Saint Louis Probation Office provided d emographic identifiers used to associate arrest records a nd individu als The Missouri Uniform Law Enforcement System (MULES) arrest file database was used to garner information regarding misdemeanor and felony arrests occurring in Missouri (the state in which the STOP study occurred). Furthermore, the National Crime Inform ation Center (NCIC) Interstate Identification Index (Triple I) was used to extract data on arrests that occurred outside the state of Missouri. To assess recidivism a data abstraction form which was developed to ensure tha t offense data were properly se parated by follow up time, was used to identify felonies misdemeanor s or municipal violation s in the period between baseline and 8 month follow up. This analysis evaluates types of charges (felony, misdemeanors, municipal violations) a s well as the numbe r of charges. It is important to note that

PAGE 141

141 complete data on charges were unavailable for 2 participants who were deceased by the 8 month follow up Therefore, prior data on these women were excluded from the analyses thereby making the final sample size 3 17 women Recent Use of Crack/Cocaine and Other Substances have you used (crack/cocaine) in any way in the last 30 crack/cocaine at least one day were cl assified as crack/cocaine users. Recent use of any substance was defined as using any substance (crack/cocaine, marijuana, stimulants, and heroin) at least one time in the past 30 days. Violence Participants were considered having experienced violence if t st 4 months, has anyone abused you emotionally, that is, did or said things to make you feel very bad about your bruises, cuts, broken bones, or otherwise physically abused HIV/AIDS Risk Behavior Participants who reported having at least one at risk partner (defined as having a partner who is an injection drug user or a partner who has other partners simultaneously) or 2+ sex partners AND 1+ reported unprotected sex act (any unprotected vaginal, anal, or oral sex) in the past 4 months were considered to meet the criteria for HIV/AIDS risk behaviors.

PAGE 142

142 SAVA Criteria Pa rticipants who m et al l three criteria at baseline (any substance use in the past 30 days, any exposure to violence, and HIV/AIDS risk behaviors) were considered to criterion met component criteria met). Covariates Covariates assessed in this analysis include: social support (having someone to talk to and ask for favors vs. not having anyone), child sexual abuse b efore the age of 15 (yes vs. no), arrest history (4+ lifetime arrests vs. 3 or less lifetime arrests ), intervention group (PPCMI vs. SI), past family upbringing (separated 6+ months from parents in childhood vs. never separated that long). Socio demographi c variables in this analyses included: age (18 29 vs. 30+), race (black vs. non black), education (high school diploma+ vs. no high school diploma), unstable housing (living on the streets, with others, halfway house etc. vs. living in own house or apartme nt). Analyses C hi square analyses and m ultinomial logistic regressions determined the association between crack/cocaine use only, SAVA, and types of offenses while negative binomial regressions assessed correlates of number of offenses. All analy ses were conducted using SAS 9.4 (SAS Institute Inc., Cary, NC, USA). Results Re cidivism Aroun d 45% of the participants (N=143 ) were charged with at least one municipal violation, misdemeanor, or felony 8 months post baseline (Figure 6 1). The three most

PAGE 143

143 common ty pes of offenses were municipal violations alone (20%), felonies alone (11 %), and felonies and misdemeanors combined (4%). Due to the varying patterns of offenses a simplified offense variable that combined charges was created (Figure 6 2). In total, 174 p articipants (55%) had no offense 8 month post baseline, 25% had at least one misdemeanor or municipal violation but no felony charge, and 20% of participants had at least one felony charge alone or in conjunction with a misdemeanor and/or municipal violati on. Socio Demographic Characteristic s In our sample, 223 participants (70 %) self identified as black, nearly a third (2 8 %) we re 18 29 years of age almost half (46%) had less than a high school diploma, and around three quarters (76%) reported unstable hou sing (Table 6 1). The vast majority (77%) of the women in the sample reported that they had someone to confide in and ask for favors. Regarding childhood experiences, half of the women (51%) reported child sexual abuse before the age of 15, while nearly th ree quarters (72%) reported that they have been separated for at least 6 months from at least one parent during their childhood. Recent crack/cocaine use was also common among this sample, with one third of the women reporting past 30 day use of crack/coca ine. Nearly 20% of the women had all the SAVA criteria -any substance, exposure to at least one act of violence in the last 4 months, and met the criteria for having HIV/AIDS risk behaviors. The results of the chi square analyses showed that race, number of arrests, unstable housing, SAVA, and recent crack/cocaine use was associated with offenses at the 8 month follow up at the .05 significance level.

PAGE 144

144 Adjusted Multinomial Regression We used two multinomial regressions predicting type of offenses among the women (Table 6 2): 1) a model examining crack/cocaine use and type of offenses and 2) a model examining SAVA and type of offenses The results of the adjusted multinomial regression model revealed that women who used crack/cocaine at baseline had over two times the odds of having at least one misdemean or/municipal violation (AOR 2.21, CI 1.21 4.04 ) than women who did not use crack/cocaine at baseline; but, crack/cocaine users did not have a significant difference in odds of having a felony compared with the ir non using counterparts. Arrest history was the strongest correlate of misdemeanors/municipal viol ations (AOR 2.36, CI 1.15, 4.8 9) however, arrest history was not significantly associated with having a felony offense Though race was not significantly associated with a felony offense in the adjusted model, those who were black were significantly less likely to have at least one misdemean or/municipal violation (AOR 0.48, CI 0.26, 0.89 ) compared to women who were non black. Age was the only significant co rrelate of having a felony, with women between the ages of 18 29 at increased odds of a felony offense compared wit h women who were older (AOR 1.91, CI 1.03 3.62 ). In the second model that assessed SAVA and types of offenses women who met the criteria for SAVA had around 4 times the odds of misdemeanors/municipal violations than women who did not meet any component criterion. Women who met one or more component criteria for SAVA did not have significantly elevated odds of offenses compared to those who met none. Significant correlates of offenses were the same as those in the crack/cocaine model and similar in strength, with black race and arrest history predicting misdemeanors/municipal violations and younger age alone predicting felonies.

PAGE 145

145 Next, we examine d correlates of number of offenses among the women using negative binomial regressions (Table 6 3). B ecause descriptive statistics showed that the conditional mean of number of felonies was slightly different than the conditional variance, a negative binom ial regression was used to model the number of felonies in the past 8 months. An overall goodness of fit chi square analyses of model fit yielded a non significant p value (1.00), giving further support that the use of a negative binomial regression was an appropriate analytical approach. Results revealed no significant differences in incident rate ratio (IRR) of number of felonies among crack/cocaine users compared with those who did not use crack/cocaine. However, there was a consistent trend for those ca tegorized as having the SAVA syndemic to have substantially less felony charges than those who did not meet the criteria (IRR .37, CI .14 1.02). On the contrary, there was also a trend for those in the 18 29 age group to have a greater number of felony cha rges than those who were older (Model 1 IRR 1.62, CI .92, 2.84; Model 2 IRR 1.72, CI .95, 3.12). Because there was significant over dispersion in the number of misdemeanors/municipal violations, a negative binomial regression was also used to assess corre lates of misdemeanors/municipal violations. An overall goodness of fit chi square analysis of model fit also yielded a non significant p value (.99), suggesting that this analytical approach was appropriate. Results of the analyses show that prior arrest h istory was correlated with a greater amount of misdemeanors/municipal violations, while race was correlated with reduced amount of misdemeanors. Specifically, individuals with 4 or more life time arrests had nearly 5 times the incidence of

PAGE 146

146 misdemeanors/mun icipal violations than those who had less life time arrests (Model 1 IRR 4.56, CI 2.26, 9.27; 4.72, CI 2.37, 9.40). Discussion Through this study, we examined the association of recent crack/cocaine use, having the SAVA syndemic, and specific types of off enses Our results showed that by 8 months post baseline, 45% of the females in our sample were charged with at least one felony, misdemeanor, or municipal violation. Though details on specific crimes committed by participants were unavailable, commentary from a high ranking drug court official indicated that many of the offenses post baseline stemmed from the reinstatement of original offenses, a consequence for failing to comply with drug court rules and requirements. Moreover, increased number of offense s may also be attributed to companion cases, that is, multiple charges resulting from one incident. For example, the sheer fact that a drug court participant is charged with a new felony also meant that such individual faced charges related to the violatio n of their terms of parole or probation. Notably, felony charges usually stemmed from drug possession and small sales of drugs, while misdemeanor charges typically stemmed from prostitution related offenses. Municipal violations, however, stemmed from sm all local infractions, which have been specifically labeled by the Department of Justice in its investigation of the criminal justice system in the city of Ferguson (a city in the Greater St. Louis Metropolitan A rea ) as excessively punitive and discrimina tory against marginalized populations (Pinard, 2015 ) In this study, crack/cocaine use alone and SAVA was associated with offenses among our sample of females in drug court, this association was only significant for

PAGE 147

147 misdemeanors/municipal violations and not more s erious types of offenses. This suggests that the Municipal Drug Court system is capturing the intended population of substance users, non violent drug offenders, for these types of justice interventions. Our study also aimed to explore intervent ion group and other structural and social factors linked with incarceration among women such as race, unstable housing, and socioeconomic status (Fries, Fedock, & Kubiak, 2014; Gallagher, 2013; Krebs, Lindquist, Koetse, & Lattimore, 2007). The results show ed that intervention group was not significantly associated with offenses The lack of a significant association may have been attributed to the sub optimal uptake of the PPCMI. Though women could utilize up to 40 hours of the case management intervention, most women utilized less than 20 hours, which may have been attributed to the rigorous demands of drug courts and maintaining sobriety. Significant differences between the intervention groups may have been evident had all women in the PPCMI group utilized all 40 hours. In our sample of mainly women in drug court black women were significantly less likely to have a n offense 8 months post baseline compared to non black women. This corroborates with a previous analysis on this sample which found a signific antly lower self reported number of arrests for black women compared to non black women from the period between baseline and 4 month follow up (Johnson et al. 2011). However, other studies have found that those who were black tended to have poorer outcome s in therapeutic justice programs (Gallagher, 2013; Krebs, Lindquist, Koetse, & Lattimore, 2007). Other studies have found no racial disparities (Gallagher et al. 2015). A possible explanation may lie in the fact that blacks have been shown to have worse drug court outcomes in drug courts where minorities are underrepresented, while no

PAGE 148

148 racial disparities have been found in drug courts with relatively equal number of minorities and non minorities (Gallagher, 2013a; Gallagher et al. 2015). I n our study, bla cks comprised 70 % of our sample; since under representation in drug court has been shown to be linked with poorer outcome, this may explain why non blacks were significantly more likely to have an offense by the 8 month follow up This racial discrepancy m ay also be due to the fact that this study consisted of a female sample, which may suggest that there may be gender differences between black males and females and drug court outcomes. This may especially be profound as black males are significantly more l ikely to be involved in the criminal justice system than black f emales. T his racial discrepancy may be due to selection bias, as our sample of women involved in the criminal justice system was not randomly selected. A prior study by Cottler and colleagues (2013) found that blacks were more likely to participate in research compared to mem bers of other races, which may partly account for our large percentage of blacks. However, since it is well known that blacks are disproportionately affected by the crimin al justice system, especially in the St. Louis area, the large percentage of black women in our samp l e wa s not surprising. In the above mentioned investigation by the Department of Justice it was revealed that blacks comprised of 95% of all those charged charged with resisting arrests, 89% of those charged with failure to obey, and 92% of those char ged with disturbing the peace (DOJ 2015). Altogether Blacks comprised of 90% of all citations and 93% of all arrests in the Ferguson criminal justice system between the years 2012 2014 ( DOJ, 2015 ). Though the majority of the women in our sample were from the city of St. Louis which has a different police department then

PAGE 149

149 Ferguson, a city in St. Louis c ounty, the findings from the D epartment of Justice illustrate the disproportionate representation of blacks in the criminal justice system. However, i t was surprising that having an unstable parent/child relationship, unstable housing, lower education, an d child sexual abuse were not significantly associated with offenses as these factors have been shown to be associated with initial 2013; Degenhardt & Hall, 2012). The l ack of association in these variables may in part be attributed to the high level prevalence of these factors in this sample, thus leaving little room for significant variations. Also, proximal factors may have a stronger effect on present behavior than di stal factors and the pathways such as substance use may be important predictors. Other significant correlates included arrest history and age. In our sample, women who had 4 or more arrests were significantly more likely to have future misdemeanor/municipa l violations, while women who were younger were significantly more likely to have future felony charges. These results are expected as both arrest history and younger age are linked with deviant behaviors and have been linked with poorer outcomes in therap Gallagher et al. 2015). Arrest history was not associated with future felonies, suggesting that women with greater number of arrests were low level offenders. This maybe an important finding as it implicate s that even those with lengthy arrest histories in drug court are not violent. Limitations and Strengths As with all studies, this study has several limitations. The main limitation centers on the fact that our sample of female offenders were no t randomly chosen, therefore the

PAGE 150

150 results may not be generalizable to all females in drug court. Moreover, our study relied on self report data on crack/cocaine use and other sensitive topics such as child sexual abuse and current risky sexual behaviors. Th ese sensitive questions may lead to the social desirability bias. On the other hand, this study has several strengths including using official and complete criminal justice records to examine offenses having a respectable sample size, and more information regarding the association between crack/cocaine and other important correlates and offenses among an understudied and marginalized population of women. Conclusion In our sample of participants in a therapeutic justice program, crack/cocaine use and the SA VA syndemic was a significant predictor of future misdemeanors/municipal violations. Interventions aimed to reduce offenses in a similar population should consider periodic assessment of substance use, especially crack/cocaine use and offer additional supp ort for those who use substances. Although women who recently used crack/cocaine and or had the SAVA syndemic were more likely to reoffend than women who did not, they were not at higher odds of engaging in more serious offenses such as felonies, which cou ld be attributed to their participation in a therapeutic justice program. This study provides further support for the development of gender based substance use interventions to improve health and social outcomes of high risk women (Blankenship et al. 2015 ; Messina et al. 2010; Binswanger et al. 2010). Such interventions should be trauma informed and include HIV/AIDS prevention practices such as safe sex practices.

PAGE 151

151 Figure 6 1. Patterns of Offenses 8 months Post Baseline (N=317 ) Figure 6 2. Types of Offenses 8 months Post Baseline (N=31 7 )

PAGE 152

152 Table 6 1. Socio Demographic Characteristics of the Sample by Charg es 8 Months Post Baseline (N=317 ) Demographic Characteristics Felony Charge N=64 (20%) Misdeme anor or Municipal Violation N=79 (25%) No Charge N=174 (55%) Total N=317 (100%) p value Black 41 (64%) 48 (61 %) 134 (77%) 223 (70 %) .02 18 29 yrs. of a ge 25 (39%) 18 (23 %) 45 (26%) 8 8 (2 8 %) .07 Has social support 25 (86%) 62 (77%) 130 (75%) 247 (77%) .18 Less than high school diploma 31 (48%) 44 (56 %) 72 (41%) 147 (46%) .10 Child Sexual Abuse (before age 15) 31 (48 %) 43 (54 %) 88 (51%) 162 (51%) .7 4 Separated from parents 6+mos (before age 15 ) 52 (81%) 54 (69 %) 122 (70%) 228 (72%) .19 4+ Arrests 45 (70%) 67 (85%) 110 (63%) 222 (70%) <.01 Unstable Housing 43 (67%) 68 (86 %) 131 (75%) 242 (76%) .03 PPCMI Intervention 36 (56%) 43 (54 %) 83 (48%) 1 62 (51%) .40 All SAVA Criteria Met 5 (8%) 25 (32 %) 29 (17%) 59 (19%) .01 1 or 2 SAVA Criteria Met 43 (67%) 48 (61 %) 111 (64%) 202 (64%) Recruited from Community 7 (11%) 3 (4%) 27 (16%) 37 (12%) .03 Crack/Cocaine use 17 (27%) 39 (49%) 51 (29%) 107 (3 4%) <.01

PAGE 153

153 Table 6 2. Adjusted Multinomial Regressions Predicting Offense Patterns by 8 months (N=317 ) Felonies OR (95% CI) Misdemeanors/ Municipal Viola tions OR (95% CI) Felonies OR (95% CI) Misdemeanors/ Municipal Violations OR (95% CI) Crack/Cocaine Use Yes 1.02 (0.51, 2.05 ) 2.21 (1.21, 4.04 ) ------No 1.0 1.0 ------SAVA All 3 Criteria Met ------0.32 (0.10,1.02 ) 4.18 (1.4 5, 12.10 ) 1 or 2 Criterion Met ------0.74 (0.36, 1.50 ) 2.15 (0. 83 5.62 ) No Criterion Met ------1.0 1.0 Race Black 0.58 ( 0.30, 1. 0 9 ) 0.48 (0.26 0. 89 ) 0.61 (0.32, 1.15) 0.51 (0.28, 0.92 ) All Other Races 1.0 1.0 1.0 1.0 Arrest History Arrest 4+ 1.32 (0.68, 2.54) 2 .36 (1.15, 4.89 ) 1.43 (0.74, 2.77) 2.30 (1.12, 4.73 ) 3 or Less 1.0 1.0 1.0 1.0

PAGE 154

154 Table 6 2 Continued Felonies OR (95% CI) Misdemeanors/ Municipal Violations OR (95% CI) Felonies OR (95% CI) Misdeme anors/ Municipal Violations OR (95% CI) Recruitment Site Community .73 (.29, 1.81) .29 (.08, 1.03 ) .75 (.30, 1.86) .29 (.08, .99 ) Municipal Drug Court System 1.0 1.0 1.0 1.0 Age 18 29 yrs. of age 1.91 (1.03, 3.62 ) 1.15 (0.58, 2.28 ) 1.97 (1.06 3.69 ) 0.90 (0.46, 1.75 ) 30+ 1.0 1.0 1.0 1.0

PAGE 155

155 Table 6 3. Correlates of Number of Felonies and Misdemeanors/ Municipal Violations by 8 months (N=317 ) Felonies Misdemeanors/ Municipal Violations Model 1 (SAVA) IRR (95% CI) Model 2 (Crack/Cocaine) IRR (9 5% CI) Model 1 (SAVA) IRR (95% CI) Model 2 (Crack/Cocaine) IRR (95% CI) Crack/Cocaine Use Yes --1.16 (.64, 2.13) --.76 (.42, 1.37) No --1.0 --1.0 SAVA All 3 Criteria Met .37 (.14, 1.02) --.80 (.31, 2.07) --1 or 2 Criterion Met .94 (.47, 1.88) --.78 (.35, 1.69) --No Criterion Met 1.0 --1.0 --Race Black .91 (.51, 1.63) 0.90 (.50, 1.62) 0.50 (.27, .95) .54 (.29, 1.00) All Other Races 1.0 1.0 1.0 1.0 Recruitment Site Community .73 (.29, 1.81) 55 (.21, 1.44) .11 (.03, .38) .11 (.03, .38) Municipal Drug Court System 1.0 1.0 1.0 1.0

PAGE 156

156 T able 6 3 Continued Felonies Misdemeanors / Municipal Violations Model 1 (SAVA) IRR (95% CI) Model 2 (Crack/Cocaine) IRR (95% CI) Model 1 (SAVA) IRR (95% CI) Model 2 (Crack/Cocaine) IRR (95% CI) Arrest History Arrest 4+ .91 (.50, 1.65) 0.87 (.48 1.57) 4.56 (2.26, 9.27) 4.72 (2.37, 9.40) 3 or Less 1.0 1.0 1.0 1.0 Age 18 29 yrs. of age 1.62 (.92, 2.84) 1.72 (.95, 3.12) 1.35 (.72, 2.54) 1.26 (. 67, 2.35) 30+ 1.0 1.0 1.0 1.0

PAGE 157

157 CHAPTER 7 GENERAL DISCUSSION Scope, Significance, and Aims of Dissertation The intersectionality of substance use, violence, and HIV/AIDS risk behavior s, termed the SAVA syndemic, is not only synergistic and mutually re en forcing, but is also the most common pathway to initial criminal justice involvement among women (Meyer et al. 2015; Abad et al. 2015; Meyer, Springer, & Altice, 2011; Lichtenstein & Malow, 2010; Singer, 2009; Singer, 2006). Though women in the criminal justice system are disproportionately affected by the SAVA syndemic, this high risk group is often excluded from large scale epidemiologic studies (Welty et al. 2016). Th erefore, th is dissertation focused on the SAVA syndemic among a population that has b een significantly underrepresented in research women of color in the criminal justice system (Blankenship, Reinhard, & El Bassel, 2015; El Bassel & Strathdee, 2015; Meyer, Springer, & Altice, 2011; Shoptaw et al. 2013). Specifically, this dissertation foc used on several gaps in knowledge in the literature of substance use, violence, HIV/AIDS, criminal justice, and gender. First, a recent systematic review of the HIV literature highlighted a lack of structural interventions in criminal justice settings to r educe HIV/AIDS transmission (Shoptaw et al. 2013). Because of this, the Courtroom Behavioral Checklist (CRBCL), a measure which quantified court behaviors and has been linked with poorer criminal justice outcomes among women in drug court (Reingle et al. 2012), was evaluated to assess an association with SAVA. In addition, another review of the literature on substance using women authored by El Bassel and Strathradee (2015) also noted the lack of epidemiologic studies on

PAGE 158

158 SAVA, especially in criminal jus tice settings and alterna tives to incarceration programs Particularly, studies that illustrate the prevalence of violence and risky sexual behaviors and elucidate the effect of race and socio economic status on these issues among drug using women are scar ce I n addition, there i s a need to examine the existence of heterogeneous subgroups of women within the female offender population and evaluate whether changes in drug use, sexual behaviors, and exposure to violence differ by such groups. Because t he mos t common pathways into the criminal justice system for women stem from substance abuse and comorbid factors such as risky sexual behaviors and victimization (Scott, Grella, Dennis, & Funk, 2014; Greier, Law, & Brown, 2014; Fries, Fedock, & Kubiak, 2014) i t i s important to assess the impact of the SAVA syndemic on criminal justice outcomes. Stanton Hill and colleagues (2015) argue that issues pertaining to the SAVA syndemic have been traditionally under represented in studies examining recidivism due to the diff iculties regarding measurement, as well as the view of SAVA as inferior to public safety. It is of special importance that we identify and address factors associated with women in the criminal justice system, as maternal incarceration has been shown t o be the strongest predictor of future incarceration of et al. 2014). Based on the gaps in the literature identified the specific aims of this dissertation were to: 1. Assess the association between sel f reported SAVA leading up to baseline and and controlling for socio demographic characteristics. 2. E valuate the longitudinal trends of the SAVA syndemic over an 8 month period amon g female offenders

PAGE 159

159 a) Evaluate the association between a peer partnered case management intervention (PPCMI) and a standard intervention (SI) on the likelihood of reduction of SAVA over time b) Assess the strength of relationships between violence, substance use, and HIV/AIDS risk by assessing the effect of the initial prevalence of these issues on longitudinal outcomes. c) Determine the effect of race, markers of socio economic status such as education and stable housing, and observed court behaviors at baselin e on SAVA over time. 3. Explore latent status of women based on substance use, exposure to violence, and risky sexual behaviors at baseline, 4 months, and 8 months a) Identify the proportion of individuals in each latent status at the baseline, 4 month follow up, and 8 month follow up and the probabilit y of transitioning to lower status over time. b) Asse ss the effect of the PPCMI intervention versus the SI on latent status transitions. c) Evaluate differences in the association between socio demographic chara cteris tics, child sexual abuse, perception of drug use and initial latent status membership. 4. Examine the association between crack/cocaine use at baseline and any felony, misdemeanor, and municipal violations at an 8 month follow up. a) Assess the association be tween SAVA (any substance use, being exposed to violence, and having HIV/AIDS risk behaviors) at baseline and any felony, misdemeanor, and municipal violations at an 8 month follow up. b) Assess the association between crack/cocaine use at baseline and any f elony, misdemeanor, and municipal violations at an 8 month follow up. c) Determine the association between crack/cocaine use at baseline and the number of felony, misdemeanor, and municipal violations at an 8 month follow up. d) Explore the association between SAVA at baseline and the number of felony, misdemeanor, and municipal violations at an 8 month follow up.

PAGE 160

160 Baseline Correlates of Courtroom Behaviors The first aim of this dissertation was to evaluate the association between self reported SAVA leading up baseline, measured using the CRBCL and controlling for socio demographic characteristics. It was hypothesized that women with SAVA would have significantly higher baseline CRBCL scores compared to wo men with who do not meet the criteria for SAVA. Though the results of our bivariate analysis illustrated the significant increase in CRBCL scores among women with SAVA, this association was attenuated when adjusted for socio demographic characteristics of the sample. However, women who used an illicit substance in the past 30 days or met the criteria for HIV/AIDS risk were more likely to have unfavorable court behaviors than non substance using women or women who did not meet the criteria for HIV/AIDS risk. On the contrary, women who were exposed to violence were significant ly more likely to have lower CRBCL scores Moreover, the results revealed that who were less religious/spiritual, had 4 or more arrests, and less education were significantly more lik ely to have unfavorable court behaviors. Such findings are consistent with known literature as it has been shown that religion/spirituality is associated with prosocial behaviors, criminal justice involvement is correlated with court behaviors, and lower e ducation is associated with poorer outcomes in the criminal justice system (Reingle et al. 2012; Shariff et al. 2016; Sus sman, et al. 2011 ; Mitchel et al. 2012). Correlates of Changes in SAVA Over Time When we evaluated the longitudinal trends of the S AVA syndemic over the 8 month period, results showed that though the likelihood of SAVA significantly decreased by the 8 month follow up intervention group was not associated with a significant

PAGE 161

161 decrease in the likelihood of SAVA over time. This suggests t hat the decrease in the likelihood of SAVA as well as decrease in HIV/AIDS risk and exposure to violence over time were as pronounced among those in the PPCMI group as it was for those in the SI group. Nevertheless, there was a trend for women in the PPCMI group to experience less violence and to have a lower median number of times of total substance uses than women in the SI group. The analyses also revealed a suboptimal uptake of the PPCMI intervention. T he vast majority of women in the PPCMI group did no t utilize any more than 20 hours of the intervention out of the possible 40 hours Uptake of the intervention may have been made difficult due to the rigorous requirements of judicial diversion programs Moreover, the vast majority of the women faced harsh realities such as unstable housing Such factors, along with problems related to legal and substance use may all have contributed to the suboptimal utilization of the PPCMI intervention. It was also hypothesized that educational attainment, race, and unst able housing at baseline would be associated with an increased likelihood of SAVA over time however, this was largely unsupported. The lack of association may be due to the fact that a large proportion of our sample had these characteristics, limiting var i ability. The lack of association between social demographic variables may also be attributed to the fact the women were in justice intervention programs which offer additional support for vulnerable women Latent Status of Women and Discrete Changes Over Time When evaluating heterogeneous subgroups of women within this sample, four distinct behavioral profiles at baseline, referred to as statuses, emerged: 1) a latent status characterized by a high probability of substance use, exposure to violence, and

PAGE 162

162 ri sky sexual behaviors 2) a latent status characterized by a high probability of crack/cocaine use only, 3) a latent status characterized by a moderately high probability of emotional abuse and risky sexual behaviors and 4) a latent status characterized by a low probability of substance use, exposure to violence, and risky sexual behavior s The results also revealed that the proportion of women in latent statuses characterized by a high probability of crack/cocaine use (Status1 and Status 2) did not substa ntially decrease over time. Because of the known addictive nature of crack/cocaine the large reduction seen among those in Status 3, the most transient status, seems to be attributed to their low probability of crack/cocaine use. This finding suggest that in our sample, risky sexual behaviors may be easier altered than substance using behaviors. Other results suggested that women w ho perceived that they had risky drug using behaviors that needed changing, experienced child sexual abuse before the age of 15 and were older age will have elevat ed odds of being in statuses with higher probabilities of substance use, exposure to violence, and risky sexual behaviors compared to women who did not have those characteristics Crack/Cocaine, SAVA, and Re offenses Ov er Time Lastly we examined the association of recent crack/cocaine use, having the SAVA syndemic, and specific types of offenses by the 8 month follow up The results showed that by 8 months post baseline, 45% of the females in our sample were charged wit h at least one felony, misdemeanor, or municipal violation. However, crack/cocaine use alone and SAVA were only associated with lower level charges ( misdemeanors/municipal violations ) but not more serious charges (felonies) Commentary provided on types of crimes committed suggests that many of the offenses post baseline stemmed from the reinstatement of original offenses, a consequence for

PAGE 163

163 failing to comply with drug court rules and requirements, which was almost exclusively absconding. Additionally, felon y charges usually stemmed from drug possession and small sales of drugs, misdemeanor charges typically stemmed from prostitution related offenses, while municipal violations were small local infractions. Moreover, many of the women had multiple offenses by the 8 month follow up. The increased number of offenses may also be attributed to companion cases, that is, multiple charges resulting from one incident. Implications of Findings The implications of the findings from the various results of this dissertat ion are as follows: 1. Though f urther studies on other samples of o ffenders are needed to support are initial findings, t he CRBCL may have added utility in identifying female offenders with recent substance use, exposure to violence, and risky sexual behavior s. 2. I nvolvement in a therapeutic justice program such as drug courts are associated with not only decreases in substance use, but also with HIV/AIDS risk behaviors and violence experienced over time. 3. There may be added utility in additional interventions such as the PPCMI to reduce substance use, HIV/AIDS risk, and exposure to violence 4. There are distinct behavioral patterns among women in drug court in regards to substance use, exposure to violence, and risky sexual behaviors 5. W omen who report that they h ave risky sex and drug using behaviors that need changing may benefit from additional intensive interventions to assist in changing their behaviors. 6. T he Municipal Drug Court system are capturing the intended population of substance users, non violent dr ug offenders, for these types of justice interventions 7. Interventions aimed to reduce offenses in a similar population should consider periodic assessment of substance use, especially crack/cocaine use and offer additional support for those who use substan ces. 8. Our analyses suggest a wide spread need for trauma informed interventions among females involved in the criminal justice system, as well as targeted interventions tailored to crack/cocaine users.

PAGE 164

164 9. More gender based substance use interventions to impro ve health and social outcomes of high risk women are needed.

PAGE 165

165 LIST OF REFERENCES A systematic review of HIV and STI behavior change interventions for female sex workers in the United States AIDS and Behavior, 19 (9), 1701 1719. Abram, K. M., Teplin, L. A., & McClelland, G. M. (2003). Comorbidity o f severe psychiatri c disorders a nd subst ance use di ` sorders among women i n jail American Journal of Psychiatry, 160 (5), 1007 1 010 Acheampong, A. B., Lasopa, S., Striley, C. W., & Cottler, L. B. (2015). Gender Differences in the Association Between Religion/Spirituality and Simultaneous Polys ubstance Use (SPU). Journal of Religion and Health 1 11. Adimora, A. A., & Schoenbach, V. J. (2005). Social Contex t, Sexual networks, and racial disparities in rates o f sexually transmitted infections Journal of Infectious Diseases, 191 (Supplement 1), S115 S122. Adimora, A. A., Schoenbach, V. J., Taylor, E. M., Khan, M. R., & Schwartz, R. J. (2011). Concurrent partnerships, nonmonogamous partners, and substance use among women in the United States. American Journal of Public Health, 101 (1), 128 136. Akaike, H. (1974). A new look at t he statistical model identification IEEE Transactions on Automatic Control, 19, 716 723. Auerbach, J. D., & Smith, L. R. (2015). Theoretical foundations o f resea rch focused o n hiv prevention a mong subs tance involved women : A review of observational a nd intervention studies JAIDS Journal of Acquired Immune Defi ciency Syndromes, 69 S146 S154. Austin, J., Cadora, E., Clear, T. R., Dansky, K., Greene, J., Gupta, V., Mauer, M., Porter, N., Tucker, S.& Young, M. C. (2013). Ending mass incarceration: Charting a new justice reinvestment. Washington, DC: The Sentencing Project Binswanger, I. A., Merrill, J. O., Krueger, P. M., White, M. C., Booth, R. E., & Elmore, J. G. (2010). G ender differences in chronic medical, psychiatric, and substance dependence disorders among jail inmates American Journal of Public Health, 1 00 (3), 476 482. Blankenship, K. M., Reinhard, E., Sherman, S. G., & El Bassel, N. (2015). Structural interventions for HIV prevent ion among women who use drugs: A global perspective Journal of Acquired Immune Deficiency Syndromes, 69, S140 S145. Blumstein A. (2015). Racial disproportionality in prison In race and social problems (pp. 187 193). Springer New York.

PAGE 166

166 Bohn MJ, Babor RG, Kranzler H. (1995). The Alcohol Use Disorders Identification Test (AUDIT): Vali dation of a screening instrument for use i n me dic al settings Journal of Studies on Alcohol, 56 423 432. Calsyn, D. A., Cousins, S. J., Hatch Maillette, M. A., Forcehimes, A., Mandler, R., Doyle, S. R., & Woody, G. (2010). Sex u nder the influence of drugs or alcohol : Common for men in substance abuse treatment and associated with high risk sexual behavior The American Journal on Addictions, 19 (2), 119 127. Cheney, A. M., Curran, G. M., Booth, B. M., Sullivan, S. D., Stewart, K. E., & Borders, T. F. (2014). The religious and spiritual dimensions of cutting down and stopping cocaine use a qualitative exploration among African Amer icans in the sou th. Journal of Drug Issues, 44 (1), 94 113. Cohen, L. R., & Hien, D. A. (2014). Treatment outcomes for women with substance abuse a nd PTSD who have experienced complex trauma Psychiat ric S ervices. Corsi, K. F., Lehman, W. E., Min, S. J., Lance, S. P., Spee r, N., Booth, R. E., & Shoptaw, S. (2012). The feasibility of interventions to reduce HIV risk and drug use a mong heterosexual methamphetamine users Journal of AIDS & Clinical Research, (10). C ottler, L.B., McCloskey, D.J., Aguilar Gaxiola, S., Bennett, N .M., Strelnick, H., Dwyer & Striley, C. W. (2013). Community needs, concerns, a nd perceptions about health research : Findings f rom The Cl inical a nd Translational Science Award Sentinel Netw ork. American Journal of Public Health, 103 (9), 1685 1692. Compton W. M., Cottler L. B., Dorsey K. B ., Spitznagel E. L., Mager D. E. (1996). Comparing assessments o f DSM IV substance dependence disorders using CIDI SAM and SCAN. Drug Alcohol Dependence, 41 179 87. Cosden, M., Larsen, J. L., Donahue, M. T., & Nylund Gibson, K. (2015 ). Trauma symptoms for men and women i n substance abuse treatment : A latent transition analysi s. Journal of Substance Abuse Treatmen t, 50, 18 25. Deering, K. N Amin, A., Shovelle r, J., Nesbitt, A., Garcia Moreno, C., Duff, P., Argento, E. & Shannon, K. (2014). A systematic review of the correlates of violence against sex workers. American Journal of Public Health, 104 (5), e42 e54. Degenhardt, L., & Hall, W. (2012). Extent of illi cit drug use and dependence, and their contribution t o the global burden o f disease The Lancet, 379 (9810), 55 70. DePesa, N. S., Eldridge, G. D., Deavers, F., & Cassisi, J. E. (2015). Predictors of condom use i n women receiving court mandated drug a nd a lc ohol treatment: implications f or intervention AIDS C are, 27 (3), 392 400.

PAGE 167

167 DeVall, K. E., Gregory, P. D., & Hartmann, D. J. (2015). Extending recidivism monitoring for drug courts methods issues and policy implications International Journal o f Offender The rapy a nd Comparative Criminology, 0306624X15590205. attitudes of females i n drug court toward additional safeguards i n hiv prevention research Prevention Science, 10 (4), 345 352. Dumont, D. M., A llen, S. A., Brockmann, B. W., Alexander, N. E., & Rich, J. D. (2013). I ncarceration community health, a nd racial disparities Journal of Health Care for the Poor and Underserved, 24 (1), 78 88. Dyer, T. P., Regan, R., Wilton, L., Harawa, N. T., Wang, L., & Shoptaw, S. (2013). Differences in substance use, psychosocial characteristics and HIV related sexual risk behavior between black men who have sex with men only (BMSMO) and black men who have sex with men and women (BMSMW) in six us cities Journal of Ur ban Health, 1 13. El Bassel, N., & Strathdee, S. A. (2015). Women who use or inject drugs : An action agenda for women specific, multilevel, and combinatio n HIV prevention and research Journal of Acquired Immune Deficiency Syndromes, 69, S182 S190. Elkingt on, K. S., Teplin, L. A., Mericle, A. A., Welty, L. J., Romero, E. G., & Abram, K. M. (2008). HIV/Sexually transmitted infection risk behaviors in delinquent youth with psychiatric disorders : A longitudinal study Journal of the American Academy of Child & Adolescent Psychiatry, 47 (8), 901 911. Festinger, D. S., Dugosh, K. L., Kurth, A. E., & Metzger, D. S. (2016). Examining the efficacy of a computer facilitated HIV prevention tool i n drug court Drug and Alcohol Dependence, 162, 44 50. Frandsen, R. J., N aglich, D., Lauver, G. A., Lee, A. D., Regional Justice Information Service (REJIS), & United States of America. (2013). Background Checks for Firearm Transfers, 2010 Statistical Tables. N ational C riminal J ustice 238226. Fries, L., Fedock, G., & Kubiak, S P. (2014). Role of gender, substance use, and serious mental illness in anticipated post j ail homelessness Social Work Research, 38 (2), 10 7 116 success and nonsuccess in the drug cour t Int ernational Journal of Offender Therapy & Comparative Criminology, 57 (10), 1297 1316. doi:10.1177/0306624X12447774 Gallagher, J. R. (2013a). Drug court graduation rates : Implications f or policy a dvocacy a nd future research Alcoholism Treatment Quarterly, 3 1 (2), 241 253.

PAGE 168

168 Gallagher, J. R. (2013b). African American participant n drug court outcomes Journal of Social Work Practice in the Addictions 13, 143 162. Gallagher, J. R., Nordberg, A., Deranek, M. S., Ivory, E., Carlton, J., & Miller, J. W. (2015). Predicting termination from drug court and comparing recidivism patterns : Treating substance use disorders in criminal justice settings Alcoholism Treatment Quarterly, 33 (1), 28 43. Gearon, J. S., Kaltman, S. I., Brown, C., & B ellack, A. S. (2014). Traumatic life events a nd PTSD among women with substance use disorders and schizophrenia. Psychiatric Service s. Gilbert, L., Raj, A., Hien, D., Stockman, J., Terlikbayeva, A., & Wyatt, G. (2015). Targeting t he SAVA (substance abuse, violence, a nd aids) syndemic among w omen and girls : A global review of epidemiology a nd integrated interventions Journal of Acquired Immune Deficiency Syndromes, 69, S118 S127. Gmel, G., Mohler Kuo, M., Dermota, P., Gaume, J., Bertholet, N., Daeppen, J. B ., & Studer, J. (2013). Religion is good, belief is better: Religion, religiosity, and substance use am ong young Swiss men. Substance Use & Misuse, 48 (12), 1085 1098. Gold, M.A., Tzilos, G.K., Stein, L.A.R., Anderson, B.J., Stein, M.D., Ryan, C.M., Zuckoff A & DiClemente, C. (2016). A randomized controlled trial to compare computer assisted motivational intervention with didactic educational counseling to reduce unprotected sex in female adolescents Journal of Pediatric and Adolescent Gynecology 29 (1), 2 6 32. Greiner, L. E., Law, M. A., & Brown, S. L. (2014). Using dynamic factors to predict recidivism among women a four wave prospective study Criminal Justice and Behavior, 0093854814553222. Hall, M. T., Golder, S., Conley, C. L., & Sawning, S. (2013). D esigning programming and interventions for women i n the criminal justice system American Journal of Criminal Justice, 38 (1), 27 50. Harner, H. M., & Riley, S. (2013). The responses from women in a maximum s ecurity prison Qualitative Health Research, 23 (1), 26 42. Henrichson, C., & Delaney, R. (2012). The price of prisons: What incarceration costs taxpayers. Federal Sentencing Reporter, 25 (1), 68 80. Illangasekare, S. L., Burke, J. G., McDonnell, K. A., & Gi elen, A. C. (2013). The impac t o f intimate pa rtner violence, substance use, a nd HIV on depressive symptoms a mong abused low income urban women Journal of I nter personal V iolence, 28 (14), 2831 2848.

PAGE 169

169 Islam, M. M., Topp, L., Conigrave, K. M., Haber, P. S., Wh ite, A., & Day, C. A. (2013). Sexually transmitted infe ctions, sexual risk behaviours and perceived barriers to safe sex among drug user s Australian and New Zealand Journal of Public Health, 37 (4), 311 315. Jennings, J., Woods, S. E., & Curriero, F. C. (2 013). P3. 254 The spatial and temporal associations between neighbourhood drug markets and rates of sexually transmitted infections in an urban setting Sexually Transmitted Infections, 89 (Suppl 1), A228 A228. Johnson, J. E., O'Leary, C. C., Striley, C. W ., Abdallah, A. B., Bradford, S., & Cottler, L. B. (2011). effects of major depression on crack use a nd arrests among women i n drug court Addiction, 106(7), 1279 1286. Johnson, S. D., Cottler, L. B., Ben Abdallah, A., & O'Leary, C. (2012). risk factors fo r gun related behaviors a mong urban out of treatment substance using wome n. Substance Use & Misuse, 47 (11), 1200 1207. o f sexual trauma and recent HIV risk behaviors o f commu nity recruited substance using women AIDS and Behavior, 15 (1), 172 178. association of trauma and PTSD with the substance use profiles of alcohol and cocaine dependent out of tr eatment women The American Journal on Addictions, 19 (6), 490 495. Kaeble, D., Glaze, L., Tsoutis, A., & Minton, T. (2015). Correctional p opulations in the United States, 2014. Bureau of Justice Statistics Bulletin (NCJ 249513). Kamal, F., Flavin, S., Camp bell, F., Behan, C., Fagan, J., & Smyth, R. (2007). Factors affecting the outcome of methadone maintenance treatme nt in opiate dependence. Irish Medical Journal, 100 (3), 393 397. Khan, M. R., Golin, C.E., Friedman, S.R., Scheidell, J.D., Adimora, A.A., Ju don Monk, S., Hobbs, M.M., Dockery, G., Griffin, S., Oza, K.K. & Myers, D. (2015). STI/HIV sexual risk behavior and prevalent STI among incarcerated African American men in committed partnerships : The significance of poverty, mood disorders, and substance use AIDS and Behavior, 19 (8), 1478 1490. Klein, H., Elifson, K. W., & Sterk, C. E. (2008). Depression a nd HIV risk behavior practices a Women & Health, 48(2), 167 188. Konecky, B., Cellucci, T., & Mochrie, K. (2016). Predictors of pr ogram failure in a juvenile drug court progra m. Addictive Behaviors, 59, 80 83.

PAGE 170

170 Krebs, C. P., Lindquist, C. H., Koetse, W., & Lattimore, P. K. (2007). Assessing the long term impact o f drug court participatio n on recidivism w ith generalized estimating equa tions Drug and Alcohol Dependence, 91 (1), 57 68. Lamb, H. R., Weinberger, L. E., & Gross, B. H. (2014). Community treatment of severely mentally ill offenders under the jurisdiction of the criminal justice system: A review. Psychiatric services. Lanza, S. T., & Collins, L. M. (2008). A new SAS procedure for latent transition analysis: transitions in dating and sexua l risk behavior. Developmental P sychology, 44 (2), 446. Lanza, S. T., Dziak, J. J., Huang, L., Wagner, A., & Collins, L. M. (2015). PROC LCA & P ROC LTA users' guide (Version 1.3.2). University Park: The Methodology Center, Penn State. Lanza, S. T., Patrick, M. E., & Maggs, J. L. (2010). Latent transition analysis : Benefits of a latent variable approach to modeling transitions i n substance u se Jo urnal of Drug Issues, 40 ( 1), 93 120. Liang, K. Y., & Zeger, S. L. (1986). Longitudinal data analysis using generalized linear models Biometrika 13 22. Lichtenstein, B., & Malow, R. (2010). A critical review o f HIV related interventions for women prisoner s i n the United States Journal of the Association of Nurses in AIDS Care, 21 (5), 380 394. Martin, L., Hearst, M. O., & Widome, R. (2010). Meaningful d ifferences: Comparison o f ad ult women who first traded sex as a juvenile versus as a n adult Violence Aga inst Women 16 (11), 1252 1269. Massoglia, M., & Pridemore, W A. (2015). Incarceration and h ealth. Annual Review of Sociology, 41, 291 310. Matusow, H., Dickman SL, Rich JD, Fong C, Dumont DM, Hardin C, Marlowe D, & Rosenblum, A. (2013). Medication assiste d treatment i n us drug courts: results from a nationwide survey of availability, barriers a nd attitu des. Journal of Substance Abuse Treatment, 44( 5), 473 480. McGee, Z. T., Baker, S. R., Davis, B. L., Muller, D. J., & Kelly, A. B. (2014). Examining risk fa ctors for recidivism and disparities in treatment among female probationers Journal of Sociology, 2 (2), 219 232. McHugh, R., Weitzman, M., Safren, S. A., Murray, H. W., Pollack, M. H., & Otto, M. W. (2012). Sexual HIV risk behaviors in a treatment refract ory opioid dependent sampl e. Journal of Psychoactive Drugs, 44 (3), 237 242.

PAGE 171

171 Messina, N., Grella, C. E., Cartier, J., & Torres, S. (2010). A randomized experimental study o f gender responsive substance abuse treatme nt for women i n prison Journal of Substan ce Abuse Treatment, 38 (2), 97 107. Metsch, L. R., Feaster, D. J., Gooden, L., Schackman, B. R., Matheson, T., Das, M., Golden M.R., Huffaker S., Haynes L.F, Tross S., Malotte C.K, Douaihy A, Korthuis P.T., Duffus W.A, Henn S., Bolan R., Philip S.S, Castro J.G., Castellon P.C, McLaughlin G., Mandler R.N., Branson B., & Colfax, G. N. (2013 ). Effect of risk reduction counseling w ith rapid hiv te sting on risk o f acquiring sexually transmitted infections : The AWARE randomized clinical trial Journal of the Amer ican Medical Association, 310 (16), 1701 1710. Mendoza, N. S., Trinidad, J. R., Nochajski, T. H., & Farrell, M. C. (2013). Symptoms of Depression and Successful Drug Court Completion. Community Mental Health Journal, 49(6), 787 792. Meyer, J. P., Springer, S. A., & A ltice, F. L. (2011). Substance abuse, violence, and HIV in w omen: A literature review of the syndemic Journal of Women's Health 20 (7). Risky Business: Focus status analysis o f sexual behaviors, dru g use a nd victimiz ation among incarcerated women i n St. louis Journal of Urban Health, 86 (5), 810 817. Minton, T. D., & Golinelli, D. (2014). Jail Inmates at Midyear 2013 statistical Tables. Washington, DC: US Department of Justice, Office of Justice Programs, Bureau of Justice Statistics. Mitchell, O., Wilson, D. B., Eggers, A., & MacKenzie, D. L. (2012). Assessing the effectiveness of drug courts o n recidivism : A meta analytic review of traditional a nd non tr aditional drug courts Journal of Criminal Justice, 40 (1), 60 71. Mitchell, O., & Caudy, M. S. (2015). Examining racial disparities in drug arrests. Justice Quarterly, 32 (2), 288 313. Morse, D. S., Silverstein, J., Thomas, K., Bedel, P., & Cerulli, C. (201 5). Finding the loopholes: A cro ss sectional qualitative study o f system ic barriers to treatment access f or women drug court participants Health & Justice, 3 (1), 1 9 of maternal incarceration on the criminal justice involvement of adult offspring a research note Journal of Research in Crime and Delinquency, 0022427815593988. National Institute on Drug A buse. (2015). Tends and Statistics. http://www.drugabuse.gov/related topics/trends statistics National Association of Drug Court Professionals. (2015). Drug Courts Work. http://www.nadcp.org/learn/facts and figures

PAGE 172

172 Needle, R., Fisher, D.G., Weatherby, N., Chitwood, D., Brown, B., Cesari, H., Booth, R., Williams, M.L., Watters, J., A ndersen, M. and Braunstein, M. ( 1995 ) Reli ability o f self reported HIV risk behaviors o f drug users Psychology of A ddictive B ehaviors, 9 (4), p.242. Nurutdinova, D., Abdallah, A. B., Bradford, S., O'Leary, C. C., & Cottler, L. B. (2011). Risk factors associated w ith hepatitis c among f emale substa nce users enrolled i n community based HIV prevention studies BMC Research Notes 4 (1), 1. Ojmarrh, M., Wilson, D. B., Eggers, A., & MacKenzie, D. L. (2012). Assessing the effectiveness of drug courts on recidivism: A meta analytic review of traditional an d non traditional drug courts. Journal of Criminal Justice, 40, 60 71. Osborne, V., & Cottler, L. B. (2012). Subtypes of alcohol de pendence and their effect o n sexual behavior chang e. Substance Use & Misuse 47 (3), 318 328. Patrick, M., O'Malley, P., Johns ton, L., Terry McElrath, Y., & Schulenberg, J. (2012). HIV /AIDS risk behaviors and substance use by young adults in the United States. Prevention Science, 13 (5), 532 538. Peters, R. H., Kremling, J., Bekman, N. M., & Caudy, M. S. (2012). Co occurring disor ders in treat ment based courts: R esults of a national survey Behavioral Sciences & The Law, 30 (6), 800 820. doi:10.1002/bsl.2024 Pinard, M. (2015). Poor, b lack and 'wanted': Criminal j ustice in Ferguson and Baltimore. Howard Law Journal, 58 (3). Prochaska, J. O., DiClemente, C. C., & Norcross, J. C. (1992 ). In search o f how people change: applications t o addictive behaviors American Psychologist 47 (9), 1102. Proeschold Bell, R. J., Reif, S., Taylor, B., Patkar, A., Mannelli, P., Yao, J., & Quinlivan, E. B (2016). Substance use outcomes of an integrated HIV substance use treatment model implemented by social workers and HIV medical providers Health & Social Work, 41 (1), e1 e10. Reilly, K. H., Neaigus, A., Jenness, S. M., Hagan, H., Wendel, T., & Gelp Aco sta, C. (2013). High HIV prevalence among low income, black women in New York city w ith self reported HIV negative a nd unknown status Journal of Women's Health 22 (9), 745 754. ottler, L. B. (2013). Can courtroo m behavior predict recidivism? A n assessment of t he courtroom behavior check list f or women prese nting i n drug court American Journal of Criminal Justice 38 (4), 520 534. Rivera, L. M., & Veysey, B. M. (2015). Criminal ju stice system involvement and gender stereotypes: consequences and implications for women's implicit and explicit criminal identities. Albany Law Review, 78 (3), 1109 1128.

PAGE 173

173 Rhoades, B. L., Greenberg, M. T., Lanza, S. T., & Blair, C. (2011). Demographic and f amilial predictors o f early executive function development : Contribution of a person centered perspectiv e Journal of Experimental Child Psychology, 108 (3), 638 662. Roberts, T. J., & Ward, S. E. (2011). Using latent transition analysis in nursing research t o explore change over tim e. Nursing Research, 60 (1), 73. Robertson, A. A., St. Lawrence, J. S., & McCluskey, D. L. (2012). HIV/STI risk behavior of drug court participant s. Journal of Offender Rehabilitation, 51 (7), 453 473. Robins LN, Cottler LB, Buchol z KK, Compton WM, North CS, & Rourke KM. (2000). Diagnostic interview schedule for the DSM IV (DIS IV). Roth, A. M., Rosenberger, J. G., Reece, M., & Van Der Pol, B. (2012). A methodological approach to improve t he sexual health o f vulnerable female popula tio ns : Incentivized peer recruitment a nd field based S TD t esting. Journal of Health Care for The Poor a nd Underserved 23 (1), 367 375. Russell, B. S., Eaton, L. A., & Petersen Williams, P. (2013). Intersecting epidemics a mong pregnant women: alcoho l use, i nterpersonal violence a nd HIV i nfection i n South Africa Current HIV/AIDS Reports, 10 (1), 103 110. SAS 9.4. SAS Institute Inc., Cary, NC, USA Saxena, P., Messina, N. P., & Grella, C. E. (2014). Who Benefits f rom Gender Responsive Treatment? Accounting f or Abuse Hi story On Longitudinal Outcomes for Women i n Prison. Criminal Justice a nd Behavior, 41(4), 417 432. Schwarz, G. (1978). Estimating the dimension of a model. Annals of Statistics, 6, 461 464. Scott, C. K., Grella, C. E., Dennis, M. L., & Funk, R. R. (2014). Predictors o f recidivism over 3 years among substance using women released f rom jail Criminal Justice and Behavior, 0093854814546894. Senn, T. E., Carey, M. P., & Coury Doniger, P. (2011). Self defining a s sexua lly abused a nd adult sexual risk be havio r: Results from a cross sectional survey of women attending a n STD c linic. Child Abuse & Neglect 35 (5), 353 362. Serafini, K., Shipley, L., & Stewart, D. G. (2016). Motivation a nd substance use outcomes among adolescents i n a school based interventio n Addictive Behaviors, 53 74 79. Sevigny, E. L., Fuleihan, B. K., & Ferdik, F. V. (2013). D o drug courts reduce the use o f incarceration ? A meta analysis Journal of Criminal Justice 41 (6), 416 425.

PAGE 174

174 Sevigny, E. L., Pollack, H. A., & Reuter, P. (2013). Can drug courts help to reduce prison a nd jail population s? The ANNALS of the American Academy of Political and Social Science, 647 (1), 190 212. Shacham, E., & Cottler, L. (2010). Sexual behaviors among club drug users : Prevalence and r eliability Archives of Sexual Behavior, 39 (6), 1331 1341. Shariff, A. F., Willard, A. K., Andersen, T., & Norenzayan, A. (2016). Religious priming a meta analysis with a focus o n prosociality Personality and Social Psychology Review, 20 (1), 27 48. Shoptaw, S., Montgomery, B., Williams, C.T., El Bassel, N., Aramrattana, A., Metzger, D.S., Kuo, I., Bastos, F.I. & Strathdee, S. A. (2013). Not just the needle : The state o f hiv prevention science among substance users a nd future directions Journal of Acquired Immune Deficiency Syndromes (1999), 63 (0 2), S174. Simon, K., Benegal, A., Jamison, S., Lu, G., & Patel, K. (2016). Small scale analysis of t he municipal court system i n St. L ouis county Singer, M. (1996). A dose o f drugs, a t ouch of violence, a case o f AIDS : conceptualizi ng the SAVA s yndemic. Free Inquiry in Creative Sociology 24 (2), 99 110. Singer, M. (2006). A dose of drugs, a touch of violence, a case o f AIDS p art 2: Further conceptualizing t he SAVA syndemic Free Inquiry in Creative Sociology 34 (1), 39 54. Singer, M (2009). Introduction t o s yndemics: A critical systems approach to public a nd community health John Wiley & Sons. Staton Tindall, M., Harp, K. L., Winston, E., Webster, J. M., & Pangburn, K. (2015). Factors associated with recidivism a mong correction s ba sed treatment participants in rural a nd urban areas Journal of Substance Abuse Treatment 56, 16 22 Steinberg, J. K., Grella, C. E., Boudov, M. R., Kerndt, P. R., & Kadrnka, C. M. (2011). Methamphetamine use a nd high risk sexual behaviors a mong i ncarcera ted female adolescents w ith diagnosed STD Journal of Urban Health 88 (2), 352 364. Sterk, C. E., Elifson, K. W., & DePadilla, L. (2013). Neighbourhood structural characteristics a nd crack cocaine use : Exploring T he impac t o f perceived neighbourhood disord er on us e a mong African Americans International Journal of Drug Policy 25 (3), 616 623. Stevens, S. (2012). Meeting the s ubstance abuse treatment needs of lesbian, bisexual a nd t ransgender women : Implications from research t o practice Substance Abuse and Rehabilitation 3 ( Suppl 1), 27

PAGE 175

175 Stewart, L., & Gobeil, R. (2015). Correctional interventions for women offenders: a rapid evidence assessment. Journal of Criminological Research, Policy and Practice, 1 (3), 116 130. Straus, M. A. (1979). Measuring intrafam ily conflict a nd violence : The C onf lict T actics (CT) scales. Journal of Marriage and the Family, 75 88. Strathdee, S. A., West, B. S., Reed, E., Moazan, B., Azim, T., & Dolan, K. (2015). Substance use and HIV among female sex workers and female prisoners: risk environments and implications for prevent ion, treatment, and policies. Journal of Acquired Immune Deficiency Syndromes, 69 S110 S1 Subramanian, R., & Moreno, R. (2014). Drug War Dtente? A Review of State level Drug Law Reform, 2009 2013. Sussman, S. Reynaud, M., Aubin, H. J., & Leventhal, A. M. (2011). Drug Addiction, Love, and t he Higher Power. Evaluation & t he Health Professions, 34(3), 362 370. Tetrault, J. M., Fiellin, D. A., Niccolai, L. M., & Sullivan, L. E. (2010). Substance use in patients w ith sexually transmitted infections : Results from a n ational U.S. survey American Journal on Addictions, 19 (6), 504 509. Torchalla, I., Nosen, L., Rostam, H., & Allen, P. (2012). Integrated tr eatment programs for individuals w ith conc urrent substance use disorders a nd trauma experiences : A systematic rev iew a nd meta analysis Journal of Substance Abuse Treatment 42 (1), 65 77. Tripodi, S. J., & Pettus Davis, C. (2013). Histories of childhood victimization a nd subsequent mental h ealth problems, substance us e, and sexual victimization for a sample o f incarc erated women i n the US International Journal of Law & Psychiatry 36 (1 ), 30 40. UCLA: Statistical Consulting Group. (2016). S tatistical t ests in SAS. http://www.ats.ucla.edu/stat/sas/whatstat/whatstat.htm Unit ed States Department of Health and Human Services. (2013). Results from the 2012 National Survey on Drug Use and Health: Summary of national findings. Substance Abuse and Mental Health Services Administration : Rockville, MD, USA. Unit ed States Department of Justice Civil Rights Division. (2015 ). Investigation of the Ferguson Police Department. https://www.justice.gov/sites/default/files/opa/press releases/attachments/2015/03/04/ferguson_police_department_report.pdf Vaddiparti, K., Striley, C. W., & Cottler, L. B. (2016). Association between ga mbling and exposure to guns among cocaine using women Violence and Gender

PAGE 176

176 Vaddiparti, K., Bogetto, J., Callahan, C., Abdallah, A. B., Spitznagel, E. L., & Cottler, L. B. (2006). The effects of childhood trauma on sex trading i n substance using women Arc hives o f Sexual Behavior 35 (4), 451 459. Washburn, J., Teplin, L., Voss, L., Simon, C., Abram, K., & McClelland, G. (2008). Psychiatric disorders among detained youths: a comparison of youths processed in juvenile court and adult criminal court. Psychiatr ic Services, 59(9), 965 973. Welty, L. J., Harrison, A. J., Abram, K. M., Olson, N.D., Aaby, D.A., McCoy, K.P., Washburn, J.J. & Teplin, L. A. (2016). Health disparities in d rug and alcohol use disorders: A 12 year longitudinal study of youths after dete ntion American Journal of Public Health, 106 (5), 872 880. Wendel, T., Drucker, E., Ostermann, M., DeWitt, S., & Clear, T. (2015). A Natural Experiment in Reform: Analyzing Drug Policy Change In New York City. Werb, D., Debeck, K., Kerr, T., Li, K., Montan er, J., & Wood, E. (2010). Modelling crack cocaine use trends over 10 years in a Canadian setting. Drug and Alcohol Review 29 (3), 271. Yuan, Y. C. (2010). Multiple imputation for missing data: Concepts and new development (Version 9.0). SAS Institute Inc, Rockville, MD, 49.

PAGE 177

177 BIOGRAPHICAL SKETCH Abenaa Acheam pong Jones received her PhD in e pidemiology from the University of Florida in 2016 and a Bachelor of S cience degree in public and community h ealth at the University of Maryland College Park in 2012. Th roughout the years, Dr. Jones has engaged in several studies focusing on drug addiction, HIV/STIs, and risky sexual behavior among various marginalized populations. During her tenure at the University of Maryland, Dr. Jones was a scholar in the Ronald E. McNair Scholar Post Baccalaureate Program, which provided her with 3 years of funding to engage in undergraduate research in HIV/AIDS and substance use. Through these independent research projects, and though only an undergraduate then, Dr. Jones developed an Institutional Review Board proposal, create project designs, conduct pilot studies, implement study procedures, and wrote research papers that were published. During this time, Dr. Jones simultaneously was awarded the prestigious Benjamin Gilman Intern ational Scholarship, along with the University Study Abroad Consortium Grant and the Ronald E. McNair Ambassador Grant to study abroad at the University of Ghana Legon. In Ghana, she worked with the West African Project to Combat AIDS and STIs (WAPCAS) fac ilitating HIV prevention programs for sex workers, as well as with the University of Ghana School of Public Health researching the affect parental HIV status has on children. Through these experiences, she built upon what she had previously learned about r unning a successful research projects, but most importantly, she learned how to work with marginalized groups, either by working one on one with sex workers or facilitating focus groups at an HIV/AIDS support group. By the end of her time at the University of Maryland, Dr. Jones was a recipient of 12

PAGE 178

178 grants, honors, and awards, including the Martin Luther King Service and Scholarship recognition, where she was the sole student chosen to be highlighted on the University of Maryland webpage for excellence in service and scholarship. Dr. Jones then went on to the University of Florida as a McKnight Doctoral Fellow, a fellowship program that provided 5 years of funding for her doctoral studies under the tutelage of Dr. Linda Cottler. As a PhD student and a McKni ght Doctoral Fellow, she was a trainee at HealthStreet, a community based program that reaches underrepresented populations in North Central Florida with a goal to reduce health disparities in healthcare and research. Specifically, HealthStreet (founded by Dr. Linda Cottler), focuses on recruiting community members and linking them with needed health and social services as well as opportunities to engage in health research at the University of Florida. There, Dr. Jones gained more experience in STI and drug use research, primary data collection, culturally competent community engagement, and increased her analytical skills set by conducting secondary analysis on collected data. During her tenure at the University of Florida, Dr. Jones collaborated with inve stigators from the University of Maryland the University of California Los Angeles, and the University of Miami, wrote 8 first authored publications/manuscripts, along with 6 other peer reviewed publications and 19 original research presentations (13 firs t authored). Some of her projects included: assessing the association between prescription opioid and illicit drug use and STIs, using novel statistical software such as ArcGIS to determine clusters of drug use and STIs in Urban Gainesville, examining the protective effects of religion and spirituality on polysubstance use among a large sample of prescription opioid abusers, changes in drug using patterns and sexual risk in men

PAGE 179

179 who have sex with men (MSM) and men who have sex with men and women (MSMW), and more. Dr. Jones also leaves the University of Florida with 11 awards and honors including: the Primm Singleton Minority Travel Award for the 76 th College on Problems the NIDA Women & Sex/Gender Junior Investigator Travel Award for the 78 th College on Problems of Drug Dependence conference (CPDD), as well as an invitation from the their annual Social Determinants of Health Conference.