Citation
Alcohol Consumption among Persons Living with HIV: Patterns and Determinants of Use and Impact on Subclinical Cardiovascular Disease

Material Information

Title:
Alcohol Consumption among Persons Living with HIV: Patterns and Determinants of Use and Impact on Subclinical Cardiovascular Disease
Creator:
Chichetto, Natalie E
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:
COOK,ROBERT L
Committee Co-Chair:
SHEPS,DAVID SAMUEL
Committee Members:
CHEN,XINGUANG
WHITEHEAD,NICOLE ENNIS
PLANKEY,MICHAEL

Subjects

Subjects / Keywords:
alcohol
atherosclerosis
cardiovascular
hiv
longitudinal

Notes

General Note:
Cardiovascular disease (CVD) is the most prevalent non-HIV related cause of death among those persons living with HIV (PLWH) who are adherent to antiretroviral treatment (ART). After adjusting for CVD-related risk factors, those with HIV infection have a higher rate (11.13/1,000 vs. 6.98/1,000 person-years) and earlier onset for CVD, up to two times the odds for acute myocardial infarction, heart failure, and coronary heart disease, and over five times the risk for stroke, compared to uninfected populations. The higher adjusted risk among PLWH suggests that there are indicators outside of the traditional risk factor framework that are important contributors to cardiovascular health, such as alcohol consumption. Alcohol use is common among PLWH and is reported among 39-81%. Prevalence of hazardous drinking has been reported in as much as 25-45% of PLWH, with alcohol dependence ranging from 5.5-10%. While there is an established J-curve relationship between alcohol consumption and cardiovascular health in the general population, little is known about how long-term drinking behavior effects subclinical cardiovascular disease, also known as atherosclerosis, among PLWH.Using data from the Multicenter AIDS Cohort Study and the Womens Interagency HIV Study, we characterized patterns of alcohol consumption among PLWH from 2004-2013 by gender and assess the association between time-stable and varying clinical factors of long-term heavy and moderate alcohol consumption. We described the association between 10-year patterns of alcohol use and the prevalence and incidence of subclinical atherosclerosis, measured by B-mode carotid artery ultrasound imaging. Last, we assessed the longitudinal association between past 10-year and current patterns of alcohol use and carotid IMT progression. This study addressed an important gap in the literature regarding the possible J-curve association between alcohol consumption and cardiovascular health that has been found in the general population. The results of this study will help inform identification of CVD risk among PLWH, and has the potential to highlight the importance of tailored interventions that can better address alcohol use issues.

Record Information

Source Institution:
UFRGP
Rights Management:
All applicable rights reserved by the source institution and holding location.
Embargo Date:
5/31/2018

Downloads

This item has the following downloads:


Full Text

PAGE 1

ALCOHOL CONSUMPTION AMONG PERSONS LIVING WITH HIV: PATTERNS AND DETERMINANTS OF USE AND IMPACT ON SUBCLINICAL CARDIOVASCULAR DISEASE By NATALIE ELIZABETH CHICHETTO A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2017

PAGE 2

2017 Natalie Chichetto

PAGE 3

To my husband and p arents for all of your love and support.

PAGE 4

4 ACKNOWLEDGMENTS I thank Dr. Bob Cook for mentoring me over the past 5 my training and profess ional development has helped me to persevere through this exciting and trying process. I thank Dr. Michael Plankey for his persistent guidance and faith in me as a Multi center AIDS Cohort Study (MACS) and cohorts was incredibly helpful in forming my research questions and implications. I thank Dr. David Sheps for guiding me in the study of cardiovascular epidemiology. He selflessly dev oted his time to me to help fill my gap in knowledge and understanding of the complexities of cardiovascular health and behavior. I thank the rest of my dissertation committee members, Drs. Jim (Xinguang) Chen and Nicole Ennis Whitehead, who encouraged me to the finish line with effortless support and devotion to my training and career goals. I also thank the Department of Epidemiology for their support throughout the past 5 years. I thank Drs. Linda Cottler and Catherine Woodstock Striley for encouraging m e to pursue pre doctoral training and continuing to be supp orts since my days as a Master of Social W ork student This work was supported by the National Institute of Alcoholism and Alcohol Abuse of the National Institutes of Health (Kelso, F31 AA024064). Data for this dissertation were collected by the MACS and WIHS I thank the staff at the Data Coordinating Centers at both cohorts for their help obtaining the data for this dissertation. I thank the MACS and WIHS participants for their time and informatio n. I also thank Drs. Robert Kaplan, Wendy Post, and Alison Abraham for all of the time and service to me as senior investigators. I have benefited greatly from their knowledge of the cohorts and expertise in epidemio logy and cardiovascular health.

PAGE 5

5 The MACS is funded through the following: U01 AI35042, U01 AI35039 U01 AI35040, U01 AI35041 UM1 AI35043. The MACS is f unded primar ily by the NIAID, with additional co fund ing from the NCI. The WIHS is funding by the following : U01 AI 035004, U01 AI 031834, U01 A I 034993), U01 AI 034994 U01 AI 034989 U01 AI 042590, U01 HD 032632 The WIHS is funded primarily by the National Institute of Allergy and Infectious Diseases (NIAID Other funding sources for this study include R01 HL 083760, R01 HL 095 140, and R21 HL 120394 to R K aplan., R01 HL 095129 to W P ost and P30 AI 051519 to the Einstein Monte fiore Center for AIDS Research. I thank my parents, Mark and Teri Kelso and Ramona Murnan, for always encouraging me to follow my ambitions and for sacrificing all that th ey could to ensure that I may make my dreams come true. and they talked me off some ledges. I thank my siblings, Maria, Mike, Anthony, and Ryan for always being there for me, allow ing me to decompress, and reminding me that no matter what, I am their sister I thank my in laws, Dale and Frank, who took me in and became my family before it was ever official and for the constant support, and to my brother in laws, Mark, Pete, and Luke for all of the encouragement. I thank Juliette, Denny, Okafor, and Amal, who through shared experience one way or another helpe d me keep my eyes on the prize. I thank Mike, Megan, Kyle, Katie, and Chelsea for being my family while in Gainesville, FL and home away from home. I thank my husband, John. Because of his constant love and support I got through this monster of an experience. His motivation through the hardest of times is what kept me climbing this hill and now I can think back on all of the t imes he kept me going. He has been my rock, and for that no words can express my gratitude. I thank my dogs, Cosmo and Knoxville for allowing

PAGE 6

6 me to benefit from the (evidence based) stress relieving snuggles. Lastly, I thank God for always providing me the guidance I desperately sought in good time and in struggle.

PAGE 7

7 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF TABLES ................................ ................................ ................................ ......................... 10 LIST OF FIGURES ................................ ................................ ................................ ....................... 12 LIST OF ABBREVIATIONS ................................ ................................ ................................ ........ 13 ABSTRA CT ................................ ................................ ................................ ................................ ... 15 CHAPTER 1 SCOPE OF THE PROBLEM, LITERATURE REVIEW, AND DISSERTATION AIMS ................................ ................................ ................................ ................................ ...... 17 Introduction ................................ ................................ ................................ ............................. 17 Alcohol Consumption Definitions ................................ ................................ .......................... 17 Alcohol Consumption among Persons Living with HIV ................................ ........................ 18 Alcohol Consumption and Cardiovascular Risk Factors ................................ ........................ 19 Alcohol Consumption and Cardiovascular Disease ................................ ............................... 21 Alcohol Consumption and Atherosclerosis ................................ ................................ ............ 22 Gender and Atherosclerosis ................................ ................................ ................................ .... 24 Limitations of the Current Literature ................................ ................................ ...................... 26 Dissertation Significance ................................ ................................ ................................ ........ 26 Summary ................................ ................................ ................................ ................................ 27 Dissertation Aims ................................ ................................ ................................ ................... 27 2 GENERAL MATERIALS AND METHODS ................................ ................................ ........ 31 Study Setting, Selection, and Inclusion Criteria ................................ ................................ ..... 31 Measures ................................ ................................ ................................ ................................ 32 Independent Predictor ................................ ................................ ................................ ...... 32 Depend ent Outcome ................................ ................................ ................................ ........ 33 Covariates of Interest ................................ ................................ ................................ ....... 33 Data Analysis ................................ ................................ ................................ .......................... 35 IRB review ................................ ................................ ................................ .............................. 36 3 ASSOCIATION BETW EEN ALCOHOL CONSUMPTION TRAJECTORIES AND CLINICAL PROFILES AMONG MEN AND WOMEN LIVING WITH HIV .................... 37 Introduction ................................ ................................ ................................ ............................. 37 Methods ................................ ................................ ................................ ................................ .. 40 Study Design and Participants ................................ ................................ ......................... 40 Data Collection ................................ ................................ ................................ ................ 41

PAGE 8

8 Main outcome measure ................................ ................................ ............................ 41 Independent variables ................................ ................................ ............................... 41 Data Analyses ................................ ................................ ................................ .................. 43 Results ................................ ................................ ................................ ................................ ..... 44 Alcohol Consumption Trajectories ................................ ................................ .................. 44 Multivariate Analysis of Clinical Factors on Alcohol Consumption among Men .......... 45 Multivariate Analysis of Clinical Factors on Alcohol Consumption among Women ..... 45 Discussion ................................ ................................ ................................ ............................... 46 Limitations ................................ ................................ ................................ ....................... 48 Conclusions ................................ ................................ ................................ ............................. 48 4 THE IMPACT OF LONG TERM MODERATE AND HEAVY ALCOHOL CONSUMPTION PATTERNS ON SUBCLINICAL ATHEROSCLEROSIS AMONG PERSONS LIVING WITH HIV ................................ ................................ ............................ 57 Introduction ................................ ................................ ................................ ............................. 57 Materials and Methods ................................ ................................ ................................ ........... 59 Study Setting, Selection, and Inclusion Criteria ................................ .............................. 59 Data Col lection ................................ ................................ ................................ ................ 61 Main outcome measure ................................ ................................ ............................ 61 Independent variable ................................ ................................ ................................ 62 Covariat es ................................ ................................ ................................ ................. 62 Data Analyses ................................ ................................ ................................ .................. 63 Group based trajectory models ................................ ................................ ................ 63 Multivariable logistic regression models ................................ ................................ 64 Results ................................ ................................ ................................ ................................ ..... 64 Prevalent Subclinica l Atherosclerosis ................................ ................................ ............. 65 Incident Subclinical Atherosclerosis ................................ ................................ ............... 66 Discussion ................................ ................................ ................................ ............................... 67 Limitations ................................ ................................ ................................ ....................... 68 Conclusions ................................ ................................ ................................ ............................. 69 5 THE IMPACT OF 10 YEAR PAST AND CURRENT ALCOHOL CONSUMPTION PATTERNS ON EARLY PROGRESSION OF ATHEROSCLEROSIS AMONG PERSONS LIVING WITH HIV ................................ ................................ ............................ 80 Introduction ................................ ................................ ................................ ............................. 80 Methods ................................ ................................ ................................ ................................ .. 81 Study Setting, Selection, and Inclusion Criteria ................................ .............................. 81 Data Collection ................................ ................................ ................................ ................ 82 Main outcome measure ................................ ................................ ............................ 83 Independent variable ................................ ................................ ................................ 83 Covariates ................................ ................................ ................................ ................. 84 Data Analyses ................................ ................................ ................................ .................. 85 Group based trajectory models ................................ ................................ ................ 85 Generalized estimating equations ................................ ................................ ............. 85 Results ................................ ................................ ................................ ................................ ..... 86

PAGE 9

9 Alcohol Consumption and Baseline CCA IMT ................................ .............................. 87 Alcohol and CCA IMT over Time ................................ ................................ .................. 87 Discussion ................................ ................................ ................................ ............................... 88 Limitations ................................ ................................ ................................ ....................... 90 Conclusions ................................ ................................ ................................ ............................. 90 6 CONCLUSIONS ................................ ................................ ................................ .................. 100 Accomplishments of the Dissertation ................................ ................................ ................... 100 Public Health Recommendations and Future Directions ................................ ...................... 102 Advantages and Challenges of the Current Research ................................ ........................... 103 APPENDIX A ADDITIONAL CHAPTER 3 TABLES ................................ ................................ ............... 105 B ADDITIONAL CHAPTER 4 TABLES ................................ ................................ ............... 106 C ADDITIONAL CHAPTER 5 TABLES ................................ ................................ ............... 111 LIST OF REFERENCES ................................ ................................ ................................ ............. 116 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ....... 132

PAGE 10

10 LIST OF TABLES Table page 3 1 Baseline characteristics of persons living with HIV by cohort ................................ .......... 50 3 2 Multivariable analysis of associated factors of the moderate alcohol use compared to the low/abstinent alcohol use among men living with HIV ................................ ............... 53 3 3 Multivariable analysis of associated factors of th e moderate alcohol use compared to the low/abstinent alcohol use among women living with HIV ................................ .......... 54 3 4 Multivariable analysis of pred ictors of the heavy alcohol use compared to the abstinent/low alcohol use among men living with HIV ................................ .................... 55 3 5 Multivariable analysis of predictors of the heavy alcohol use compared to the abstinent/low alcohol use among women living with HIV ................................ ................ 56 4 1 Sample characteristics of women and men living with HIV by cohort ............................. 73 4 2 Age adjusted prevalence and incidence of subclinical atherosclerosis by alcohol consumption patterns ................................ ................................ ................................ ......... 74 4 3 Association between 10 year alcohol consumption patterns and prevalent subclinical atherosclerosis, overall and by cohort ................................ ................................ ................ 76 4 4 Association between 10 year alcohol consumption patterns and incident subclinical atherosclerosis, overall and by cohort ................................ ................................ ................ 78 5 1 Baseline characteristics of persons living with HIV by cohort ................................ .......... 94 5 2 Crude associations between 10 year alcohol consumption patterns, current alcohol consumption, and carotid artery intima media thickness, by cohort ................................ 96 5 3 As sociation between 10 year alcohol consumption patterns, current alcohol consumption, and carotid artery intima media thickness, controlling for the time and alcohol use interactions, by cohort ................................ ................................ ..................... 99 A 1 Factors associated with missing data in Aim 1 ................................ ................................ 105 B 1 Full model estimates of covariates and prevalent subclinical atherosclerosis, by cohort ................................ ................................ ................................ ............................... 106 B 2 Full model estimates of covariates and prevalent subclinical atherosclerosis, overall .... 107 B 3 Full model estimates of covariates and incident subclinical atherosclerosis, by cohort .. 108 B 4 Full model estimates of covariates and incident subclinical atherosclerosis, overall ...... 109

PAGE 11

11 B 5 Sample Characteristics between those with and without follow up in Aim 2 ................. 110 C 1 Full model estimates of covariates on carotid artery intima thickness among women ... 112 C 2 Full model estimates of covariates on carotid artery intima thickness among men ........ 113 C 3 Full model estimates of covariates on carotid artery intima thickness among men and women, adjusting for time by alcohol use interactions ................................ .................... 114 C 4 Sample Characteristics between those with and without follow up in Aim 3 ................. 115

PAGE 12

12 LIST OF FIGURES Figure page 1 1 Factors associated with the the relationship between alcohol consumption on atherosclerosis. ................................ ................................ ................................ ................... 30 3 1 Trajectories of alcohol consumption among 1,123 HIV+ women in th Interagency HIV Study ................................ ................................ ................................ ...... 51 3 2 Trajectories of alcohol consumption among 597 HIV+ men in the Multicenter AIDS Cohort Study. ................................ ................................ ................................ ..................... 52 4 1 Conceptual model for the association between alcohol consumption and atherosclerosis and confounding factors ................................ ................................ ............ 71 4 2 Timeline for 10 year trajectory models prior to prevalent and incident subclinical atherosclerosis ................................ ................................ ................................ .................... 72 4 3 10 year alcohol consumption trajectories by cohort. ................................ ......................... 75 5 1 Confounding factors associated with the association between alcohol consumption and early atherosclerosis development. ................................ ................................ ............. 92 5 2 Timeline for 10 year trajectory models prior to baseline carotid artery ultrasound .......... 93 5 3 10 year alcohol consumption trajectories by cohort. ................................ ......................... 95 5 4 Interaction of alcohol consumption patterns and time on change in CCA IMT m among men. ................................ ................................ ................................ ........................ 97 5 5 Interaction of alcohol consumption patterns and time on change in CCA IMT m among women ................................ ................................ ................................ .................... 98

PAGE 13

13 LIST OF ABBREVIATIONS AIC Akaike information c riteria AIDS Acq uired immunodeficiency s yndrome AOR Adjusted odds r atio ART Antiretroviral therapy AUDIT C Alcohol Use Disorders Identification Test BIC Bayesian information c riterion BMI Body Mass Index CACS Coronary Artery Calcium Score CCA Common c arotid a rtery CCA IMT m Common carotid a rter y intima medial t hickness micrometer CD4+ Cluster of differentiation 4 CES D Center for Epidemiological Studies Depression CI Confidence interval cIMT Carotid intima medial t hickness CVD Cardiovascular disease ELISA Enzyme linked immunosorbent assay GBTM Group based trajectory model GEE Generalized estimating equations HAART Highly Active Antiretroviral Therapy HDL High density lipoprotein HIV Human immunodeficiency virus

PAGE 14

14 ICA Internal carotid artery ICC Internal carotid circulation IQR Interquartile range IRB Institutional review board MACS Multicenter AIDS Cohort Study MI Myocardial Infarction MSM Men who have sex with men NHL BI National Heart, Lung, and Blood Institute NIAAA National Institute on Alcohol Abuse and Alcoholism NIH National Institutes of Health OR Odds r atio PLWH Persons living with HIV PP Propensity probability HIV RNA Human immunodeficiency virus Ribonucleic acid SD Standard Deviation WIHS

PAGE 15

15 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 ALCOHOL CONSUMPTION AMONG PERSONS LIVING WITH HIV: PATTERNS AND DETERMINANTS OF USE AND IMPACT ON SUBCLINICAL CARDIOVASCULAR DISEASE By Natalie Elizabeth Chichetto May 2017 Chair: Robert L Cook Major: Epidemiology Cardiovascular disease (CVD) is the most prevalent non HIV related cause of death among persons living with HIV (PLWH) who are adherent to antiretroviral treatment. After adjusting for CVD related risk factors, PLWH have a higher rate (11.13/1,000 vs. 6.98/1,000 person years) and earlier onset for CVD, up to two times the odds for acute myocardial infarction, heart failure, and coronary heart disease, and over five times the risk for stroke, compared to uninfected populations. The highe r adjusted risk among PLWH suggests that there are important indicators outside of the traditional CVD risk factor framewor k, such as alcohol consumption. Alcohol use is common among PLWH and is reported among 39 81%. Prevalence of hazardous drinking has b een reported in as much as 25 45% of PLWH, with alcohol dependence ranging from 5.5 10%. While there is an established J curve relationship between alcohol consumption and cardiovascular health in the general population, l ittle is known about how long term drinking behavior effects subclinical cardiovascular disease, also known as atherosclerosis, among PLWH.

PAGE 16

16 Study, we characterized patterns of alcohol consumption among PLWH f rom 2004 2013 by gender and assess ed the association between time stable and varying clinical factors of long term heavy and moderate alcohol consumption. We described the association between 10 year patterns of alcohol use and the prevalence and incidenc e of subclinical atherosclerosis, measured by B mode carotid artery ultrasound imaging. Last, we assessed the longitudinal association between past ( 10 year ) and current (6 month) patterns of alcohol use and non plaque carotid intima media thickness progr e ssion. This study addressed an important gap in the literature regarding the possible J curve association between alcohol consumption and cardiovascular health among PLWH, that has been found in the general population. The results of this study will help inform identification of CVD risk among PLWH, and has the potential to highlight the importance of tailored interventions that can better address alcohol use issues.

PAGE 17

17 CHAPTER 1 SCOPE OF THE PROBLEM LITERATURE REVIEW, AND DISSERTATION AIMS Introduction With the introduction of antiretroviral therapies (ART), the life expectancy of persons living with HIV (PLWH) has been substantially prolonged (Wandeler et al., 2016) Thus, prevention and management of age related chron ic illnesses are important targets for HIV specialists and researchers. Cardiovascular disease (CVD) is immerging as one of the most common comorbidities and causes of death in PLWH (Smith et al., 2014) Further, CVD is the most prevalent non HIV related c ause of death among those who are adherent to ART ( Rodger et al., 2013 ) After adjusting for CVD related risk factors, those with HIV infection have a higher rate (11.13/1,000 vs. 6.98/1,000 person years) and earlier onset for CVD (Triant et al. 2007) up to two times the odds for acute myocardial infarction (MI; Freiberg et al., 2013; Durand et al., 2011; Triant et al., 2007 ) heart failure (Butt et al., 2011) and coronary heart disease (Freiberg et al., 2011) and over five times the risk for stroke ( Wa lker et al., 2013 ; Durand et al., 2012 ) compared to uninfected populations. Further, PLWH with normal blood pressure had 1.28 times the odds for an acute MI, compared to uninfected controls (Armah et al., 2014) The higher adjusted risk among PLWH suggest s that there are indicators outside of the traditional risk factor framework that are important contributors to cardiovascular health, such as alcohol consumption. Alcohol Consumption Definitions Alcohol consumption has traditionally been described in terms of level. Moderate alcohol consumption refers to having up to 1 standard drink (12 ounces of beer; 8 ounces of malt liquor; 5 ounces of wine; 1.5 ounces of 80 proof distil led spirits or liquor) per day ( 0 to 7 standard drinks per week) for women and up to 2 standard drinks per day (0 to 14 drinks per week) from men (Center for Disease Control and Prevention [CDC] 2016). Heavy drinking is considered

PAGE 18

18 consumption > 7 drinks per week for women and > 14 drinks per week for men (CDC, 2016; Reid et al., 1999 ). Hazardous drinking is considered 14 drinks per week drinks per week for men and is a level that has been associated with increased risk for adverse health events (Reid, 2016). 5 drinks for men in a 2 hour period (CDC, 2016) An alcohol use disorder (AUD) is prob lematic drinking that meets 2 of the 11 criteria of the 5 th edition of the Diag nostic and Statistical M anual of Mental Disorders (DMS 5) in a 12 month period (National Institute on Alcohol Abuse and Alcoholism [NIAAA] 2016). An AUD is classified as mild (2 3 symptoms) moderate (4 5 symptoms) or severe (6 or more symptoms) depending on the number of crit eria met. The DMS 5 criteria include symptoms characteristic of alcohol dependence or alcohol abuse (NIAAA, 2016). Alcohol Consumption among Persons Living with HIV Alcohol use is common among PLWH and is reported among 39 81% ( Bilal et al., 2016; Monroe e t al., 2016 ; W andeler et al., 2016; Sullivan et al. 2011; Conen et al., 2009 ) Prevalence of hazardous drinking has been reported in as much as 25 45% ( Deiss et al., 2016; Monroe et al., 2016; Kader et al., 2014 ) of PLWH, with alcohol dependence ranging from 5.5 10% ( Jolley et al., 2016; Malbergier et al., 2015; Surah et al., 2013; Sullivan et al., 2011 ) Alcohol consumption, in general, is negatively associated with completing the steps of the HIV care continuum (Vagenas et al., 2015) and hazardous alcoh ol use is associated with poor retention in HIV care and lower visit adherence, compared to those that do not drink (Monroe et al., 2016) Likewise, hazardous alcohol consumption is associated with decreased ART adherence ( Pellowski et al., 2016; Malbergie r et al., 2015 ; Kalichman, et al., 2014 ; Tran et al., 2014 ) lower CD4+ T cell count (Kahler et al., 2015 ; Malbergier et al., 2015 ) and increased viral load (Deiss et al., 2016)

PAGE 19

19 Aside from the relationship between alcohol consumption and ART adherence, a lcohol abuse has been linked specifically to HIV progression through alteration of viral infectivity, inflammatory biomarkers, immune response, and tissue injury ( Monnig et al., 2016; Molina et al. 2014 ) Hazardous drinking in this population is also asso ciated with engagement in risky health behavior, such as cigarette smoking (B raithwaite et al., 2016; Pacek et al. 2014) and substance use (Parsons et al. 2014) which can lead to other chr onic illnesses. Some studies have found that PLWH who use alcohol have increased chronic comorbidity (Bilal et al., 20 16; Jolley et al., 2016; Kelso et al., 2015) while other studies have found no such association ( Kelly et al., 2016; Tsui et al., 2016; Wandeler et al., 2016 ; Fuster et al., 2013) Because many of the a forementioned studies are cross sectional, it is unclear whether moderate and hazardous drinking leads to chronic illness or if alcohol use is a coping response to such illness. Further, e merging evidence suggests that PLWH may be more affected by the harm ful sequela of alcohol use when compared to similar or lower levels of use among uninfected groups (Justice et al., 2016; McGinnis et al., 2016 ; Rentsch et al., 2016) Alcohol Consumption and Cardiovascular Risk Factors The mechanism by which alcohol consumption is thought to effect cardiovascular health among PLWH is compl ex, not well understood, and of great practical importance given the widespread global co nsumption of alcohol (Freiberg and So Armah, 2016). Biological and behavioral mechanisms are thought to explain the higher burden of CVD among PLWH, shown in our conceptual model (Figure 1 1 ). H eavy alcohol consumption and CVD are affected by demographic and psychosocial factors, including age, education, race/ethnicity, and socioeconomic status ( Conen et al., 2009 ; Galvan et al., 2002 ). Alcohol use is also associated with other risky health behaviors, such as

PAGE 20

20 tobacco use (Cook et al., 2013). injection drug use (Chitsaz et al., 2013 ; Conen et al., 2009 ) and depressive symptoms (Sullivan et al., 201 1). A lcohol use is significantly associated with traditional CVD risk factors, including dyslipidemia (high triglyceride and low density lipoprotein cholesterol levels [LDL]; Hejazi, et al., 2013 ; Mguez Burbano et al., 2009 ; Hadigan et al., 2001 ), and ins ulin resistance (type II diabetes ; Justice et al., 2006 ; Hadigan et al., 2001 ). Specifically, different levels of alcohol consumption tend to have a differential effect on important cardiovascular risk factors and biomarkers. For example, compared to non d rinkers, moderate alcohol consumption (1 drink per day in women or up to 2 drinks per day in men) is associated with a 10% increase in HDL cholesterol (Brinton, 2012). Heavy alcohol consumption is associated with an even greater increase in HDL, but is par adoxically associated with increases in triglyceride, LDL, and total cholesterol levels (Khanh et al., 2016; Brinton, 2012). Further, compared to never drinkers, those who consumed 2 7 drinks per week were less likely to have increases in damaging cardiac biomarkers (i.e., high sensitivity cardiac troponin T and N terminal pro B type natriuretic peptide); however, those who consumed 15 or more drinks per week were more likely to have incident increases in these same biomarkers (Lazo et al., 2016). A lcohol c onsumption may increase the development of CVD through HIV related factors. HIV infection alone increases systemic inflammation ( Bahrami et al., 2016 ; Shrestha et al., 2014 ) and immune activation ( Maniar et al., 2013; Neuhaus et al., 2010; Hsue et al., 200 9; Strategies for Management of Antiretroviral T herapy Study Group et al., 2006 ) which are pathophysiologic responses that contribute to the risk for CVD (Bahrami et al., 2016, Hsu et al., 2016, Hansson, 2005) Use of ART results in reduced inflammation through HIV RNA viral load suppression; however, alcohol consumption is associated with decrease d ART adherence, thus

PAGE 21

21 resulting in higher HIV RNA viral load and lower CD4 + T cell count ( Hendershot et al., 2009 ) Chronic inflammation and immun e activation can lead to the breakdown of the endothelial walls of the gastrointestinal tract, a process that leads to microbial translocation which triggers further immune and pro inflammatory responses t al., 2013 ) Previously mentioned, l ow level consumption of alcohol may have favorable lipid or antithrombotic effects. Paradoxically, low levels of alcohol use have also been shown to increase systemic inflammation, as well as risk for microbial transloc atio n (Brenchley and Douek, 2012). Through this pro inflammatory process, the endothelial wall loses structural integrity, allowing microbial material and bacteria to enter the bloodstream and cause buildup within arteries, leading to cardiovascular compli cations (Klatt et al., 2013; Brenchley et al., 2012; Freiberg and Kraemer, 2010 ) Additionally, heavy alcohol consumption is associated with frequent switching off treatment, HIV duration, and Hepatitis C co infection (Conen et al., 2009) Alcohol C onsumpt ion and C ardiovascular D isease M oderate alcohol consumption may be protective against CVD, with over moderate use being a risk factor in the general population (Mukamal et al., 2003a; Reynolds et al., 2003; Corrao et al., 2000; Sacco, et al., 1999; McElduf f and Do dson, 1997 ) Several studies have indicated significant crude positive associations between any alcohol use ( Twagirumukiza et al., 2007 ) heavy alcohol use ( Longo Mbenza et al., 2011 ) history of alcohol abuse or dependence (Durand et al., 2012) and clinical CVD among PLWH Additionally, heavy alcohol use (Freiberg et al., 2010; Corral et al., 2009 ) and abuse/dependence (Freiberg et al., 2010; Justice et al., 2008) are cross sectionally associated with increased odds fo r CVD, after adjusting for CVD and HIV related risk factors. Some studies among PLWH found moderate alcohol consumption to be associated with lower adjusted hazard ratio for CVD, compared to alcohol abstainers ( Wandeler et al., 2016; Carrieri et al., 2012) The current state of the literature is limited to mostly cross

PAGE 22

22 sectional methods and/or investigation of vague measures of alcohol use (e.g., any alcohol use, alcohol abuse/dependence history) to characterize risk among majority male HIV infected participants (Kelso et al., 2015) Further, most studies utilize only the medical record to classify diagnosis of clinical CVD, and do not assess early stages of disease development such as atherosclerosis. Alcohol C onsumption and A therosclerosis Subclinical CVD, also known as a theroscler osis is characterized by arterial plaques that may narrow the lumen, decrease blood flow and consequently predispose individuals to acute thrombotic events (National Institutes of Health [NIH] 2011 ) This process is precursory to CVD, and can lead to ser ious events, such as MI and stroke ( NIH 2014 ) Because atherosclerosis is a symptomatic and not typically assessed in clinical settings, it is difficult to know the prevalence and incidence of this subclinical disease. Most studies of atherosclerosis focus on prevalence and incidence of atherosclerosis in certain clinical populations. One recent study of men and women aged 40 54 years found the prevalence of atherosclerosis to be 63% (71% in men, 48% in women; Fernandez Friera et al., 2015). The Multi Ethni c Study of Atherosclerosis detected prevalent and incident atherosclerosis, measured by a positive coronary calcium score (CAC S ), in 48% and 20%, respectively (Pandey, 2014) Atherosclerosis is an independent risk factor for clinical cardiovascular events. One study found that those with atherosclerosis at baseline had twice the risk for incident CVD events, compared to those without atherosclerosis (25.8% vs. 12.2%; Robinson et al., 2009) Further, plaque formation from baseline to follow up was significant ly associated with incident CVD events (Hazard Ratio 1.22, CI 1.05 1.42, p<.01) with new plaque formation adding significant predictive value (+8%) in Receiver Operating Characteristic Curve Analysis (Benedetto et al., 2008). If atherosclerosis is detecte d and depending on the severity of disease and present risk factors treatment could

PAGE 23

23 include lifestyle modification (i.e., heart healthy eating, weight management, stress management, physical activity, and smoking cessation) medication (to lower cholester ol and/or blood pressure or to regulate blood sugar), or medical procedures for more severe atherosclerosis (i.e., coronary angioplasty, coronary artery bypass grafting, carotid endarterectomy; National Heart, Lung, and Blood Institute [ NHLBI ] 2016). Whil e many clinicians identify those at risk for CVD by assessing the presence of traditional CVD risk factors alone (including age, sex, total cholesterol, HDL cholesterol, smoking status, systolic blood pressure, and use of blood pressure medications), asses sing the presence of subclinical CVD is a more sensitive representation of those who may be at risk for CVD (George and Movahed, 2008) Non invasive tests used to identify subclinical CVD have been shown to detect those at high risk for clinical CVD who wo uld have otherwise been considered lower risk according to the traditional risk factors (Church et al., 2007) Non invasive assessment of subclinical CVD is most often by use of high resolution B mode carotid artery ultrasound or by computed tomography. Th e carotid artery ultrasound detects carotid intima media thickness ( cIMT; the thickness of the inner and middle layers of the carotid artery), carotid stiffness, and presence of arterial plaque/lesions (Stein et al., 2008) Computed tomography i s used to c alculate a CACS and detects presence of calcification of coronary arteries (Lester et al., 2009) While both tests characterize subclinical CVD, these tests measure different aspects of asymptomatic disorder, and thus are not strongly correlated ( Oei, et a l., 2002; Davis et al., 1999) Furthermore, research has suggested that carotid artery ultrasound has higher sensitivity, and can detect subclinical CVD among those with a low CACS (Lester et al., 2009; Davis et al., 1999) Further, use of cIMT and plaque information measured by carotid artery ultrasound has been shown to significantly improve coronary heart disease risk prediction

PAGE 24

24 over traditional risk factors ( by 3.8%, Polak et al., 2015; by 23%, Nambi et al., 2010; by 20% in men and 1% in women, Nambi et al., 2012 ) Further, cIMT was significantly associated with incident CVD events after adjusting for traditional CVD risk factors (Hazard R atio 1.63, CI 1.12 2.37 ; Polak and 2015 ) There is little research investigating the association between alcohol consumption and atherosclerosis. In the general population, moderate alcohol use has been cross sectionally associated with 55% lower risk for carotid artery plaque ( 95% CI 0.29 0.68, p <.001; Kohsaka et al., 2011) and statistically significantly lower arterial stiffness (Hougaku et al., 2005), compared to abstinence. Heavy alcohol use, however, has been associated with a significant increase in cIMT ( Zyriax et al., 2010) and stiffness (H ougaku et al., 2005), consistent with a J curved association found in the literature (Xie et al. 2012; Mukamal et al., 2003). Similarly, a longitudinal study of 20 year drinking patterns found that consistent heavy use was associated with a significant in crease in cIMT, compared to consistent moderate use (Britton et al., 2016). Other studies have found no significant association between alcohol consumption and cIMT or presence of carotid plaques ( Kim et al., 2014; Bauer et al., 2013; Zureik e t al., 2004 ). Some studies have found significant associations between alcohol consumption and subclinical disease in men, but not in women (Zyriax et al., 2010; Lee et al., 2009; Schminke et al., 2005 ). Gender and Atherosclerosis With acute cardiac syndromes (myocardi al infarction, stroke angina pectoris) on the rise i n women (Izadnegahdar et al., 2014; Sozzi et al., 2007), research on sex (biological) and gender (social) disparities in CVD has grown in rec ent years. Even with growing interest in such disparities, gen der differences have not been adequately investigated and significant gaps in the literature exist regarding possible differences in biological and behavioral mechanisms of disease. With regard to sex differences, the onset of CVD generally occurs 10 years later in

PAGE 25

25 premenopausal women than in men, with MI occurring 20 years later (Mathur et al., 2015). While men have traditionally been at higher risk for CVD, a fter menopause women have 10 times the risk for CVD, while men have a 4.6 times increase compared to the same age groups (Duvall, 2003) In fact, a contributing role may be the observed low LDL cholesterol and high HDL cholesterol up until menopause (Mathur et al., 2015). Other studies have found that traditional CVD risk factors may be more biological ly detrimental in women, compared to men. For instance, in large, longitudinal epidemiologic studies smoking behavior (Njolstad et al., 1996) and diabetes ( Stokes, et al., 1987) have to shown to be greater risk factors for CVD in women than men Another bi ological sex difference is size of arteries. Women tend to have smaller carotid arteries (Schulz and Rothwell, 2001; Krejza et al., 2006) less plaque in these arteries but more significant stenosis (Iemolo, et al., 2004), compared to men. Further, athero sclerosis in women is more likely to present as microvascular coronary disease, rather than plaque development and narrowing of the large coronary arteries (Vaccarino and Bremner, 2016). Therefore, it is possible that moderate and heavy drinking are associ ated with early progression of CVD, but in the small arteries and vessels of the coronary arteries. Gender differences have also been described in more recent literature. For example, neighborhood socioeconomic status and high professional status was found to be inversely associated with cIMT in women, but not in men (Grimaud et al., 2013). In the longitudinal Multi Ethnic Study of Atherosclerosis, educational status was also found to be linked to significantly slower stiffening of the carotid artery in wom en, but not in men (Stern et al., 2015). Psychological factors, s uch as chronic stress, trauma history, and depressive symptoms have also been associated with poor cardiovascular outcomes, but exponentially more so in women (Xu et al., 2015; Vaccarino et al., 2014; Rich Edwards et al., 2012; Korkeila et al., 2010 ) For example,

PAGE 26

26 i n a longitudinal study of nearly 8,000 adult men and women in the United States, women with depression or a history of attempted suicide had 3.20 (C I 1.12 9.17) and 14.57 (CI 2.65 80.10) times higher risk for CVD and ischemic heart disease, respectively, while the corresponding risk for men was 2.37 (CI 0.85 6.58) for CVD and 3.52 (CI 1.05 11.76) for ischemic heart disease (Shah et al., 2011). Because of these sex and gender differences in CVD and risk it is important to consider the association between alcohol consumption and atherosclerosis separately for men and women. L imitations of the C urrent L iterature To our knowledge, only three studies exist for which the main objective was to assess the relationship between alcohol consumption and CVD among PLWH ( Wandeler et al., 2016; Carrieri et al., 2012; Freiberg et al., 2010 ) and only one study assessed the cross sectional association between alcohol an d atherosclerosis (Hanna, et al., 2015). Given the high mortality and morbidity associated with clinical CVD, identifying those with subclinical atherosclerosis and modifiable risk factors is a high priority for primary and secondary prevention strategies Dissertation S ignificance This dissertation responds to the National HIV/AIDS Strategy for the United States to increase access to care and improve health outcomes for PLWH (The White House Office of National AIDS Policy, 2010) as well as the NIH objectives to advance discovery of therapeutic strategies to prevent HIV comorbidities across the lifespan and to determine the link between HIV and associated comorbidities ( NIH Office of AIDS Research, 2014) The research also aims to address the NIAAA objective to understand how alcohol use influences mortality among PLWH ( NIAAA 2014) and research focus of the NHLBI on the contribution of HIV related risk factors on the development of CVD ( NHLBI 2014) The current dissertation seeks to advance scient ific knowledge of risk factors for subclinical CVD that extend beyond cross sectional

PAGE 27

27 measurement of traditional risk factors among PLWH. We aimed to assess the J curve association between alcohol consumption and cardiovascular health among PLWH, and to pr ovide evidence that addresses the specific relationship s between moderate and heavy use on subclinical atherosclerosis. We had the unique opportunity to carry out our study aims using the data from 2,149 participants from the Cardiovascular Substudies of t HIV Study (WIHS) and the Multicentered AIDS Cohort Study (MACS). We use d this longitudinal data to describe 10 year alcohol consumption patterns by using self reported quanti ty and frequency of use. We use d state on the art methodolo gy to identify hypothesized patterns of alcohol consumption through group based trajectory analysis. Further, we have a sensitive measure of subclinical CVD, through use of B mode carotid artery ultrasound to detect the presence of carotid artery plaques a nd non plaque cIMT progression. Summary In summary, f ew studies have focused on the impact of alcohol use on cardiovascular health among PLWH. Among these studies, none have assess ed the impact of long term alcohol consumption patterns and how these patter ns effect the subclinical development of CVD. The results of this study will have implications for more effective identification of PLWH with high CVD risk outside of the traditional CVD risk framewo rk, affecting clinical practice This research also has i mplications for better recommendations for cl inical prevention services and has the potent ial to highlight the importance of tailored interventions that can better address alcohol use issues that are specific to PLWH Dissertation Aims AIM 1: To determine 10 year alcohol consumption patterns among HIV infected men and women and to identify factors associated with alcohol consumption patterns. Because of limitations of the current literature, a gap exists regarding whether alcohol use

PAGE 28

28 behaviors change over time among PLWH. Further, it is unclear if there are significant clinical factors associated with long term moderate and heavy alcohol consumpt ion by gender. Associated factors of alcohol consumption patterns would provide clinicians with the m eans to identify those with the greatest need for early intervention and alcohol abuse treatment. The goals of Aim 1 were to 1) describe patterns of alcohol consumption among PLWH from 2004 2013 by gender and 2) assess the association between time stable a nd varying clinical factors and long term heavy and moderate alcohol consumption. By utilizing reported number of drinks per week, we hypothesized that distinct patterns would emerge that are descriptive of stable (i.e., consistent abstinent, consistent m oderate, and consistent heavy) and changing alcohol use behavior (i.e., abstinent to moderate or heavy drinking; heavy to moderate or abstinence) over time. We also hypothesized that clinical factors would be identified, specifically by gender, as importan t predictors of long term moderate and heavy alcohol consumption. Specifically, we hypothesized that those with poor clinical profiles would be more likely to be heavy or moderate drinkers, compared to those who are abstinent or low drinkers. While clinica l associations of longitudinal alcohol consumption were the main focus of this analysis, the biopsycho social theoretical framework ( Engel GL 1977 ) was used to conceptualize other non clinical factors needed for analytical adjustment. AIM 2: To determine t he effect of 10 year alcohol consumption patterns on prevalent a n d incident subclini cal atherosclerosis among PLWH. The objective of Aim 2 was to assess the association between 10 year patterns of alcohol use and the prevalence and incidence of subclinical atherosclerosis, measured by B mode carotid artery ultrasound imaging. Specifically, we aimed to 1) test the association between long term moderate and heavy alcohol use and subclinical atherosclerosis among PLWH, and 2) to explore whether the relationshi ps

PAGE 29

29 appeared to differ by gender and between prevalent and incident disease. We hypothesized that long term moderate and heavy alcohol use would be significantly associated with increased risk for prevalent and incident subclinical atherosclerosis. AIM 3 : T o determine the effect of past ( 10 year ) and current (6 month) alcohol consumption patterns on the early development of subclinical atherosclerosis among PLWH. Specifically, we aimed to 1) assess the relationship between past (10 year) and current (6 month ) alcohol consumption patterns and non plaque cIMT p rogression among PLWH, and 2) explore whether the relationships appeared to differ by gender. We hypothesized that past alcohol consumption patterns would be more significantly associated with increases i n cIMT versu s current alcohol consumption. Specifically, we expected that 10 year patterns of moderate consumption would be associated with a protective effect and heavy consumption would be associated with a harmful effect on cIMT.

PAGE 30

30 Figure 1 1. Factors associated with the the relationship between alcohol consumption on atherosclerosis. Images taken from Kumar V et al. (2007).

PAGE 31

31 CHAPTER 2 G ENERAL MATERIALS AND METHODS Study Setting, Selection, and Inclusion C riteria The proposed study is a secondary ana lysis of the Cardiovascular Substudies of the multicenter longitudinal cohort studies of WIHS (Bronx, Brooklyn, Chicago, District of Columbia, Los Angeles, and San Francisco) and MACS (Baltimore/District of Columbia, Chicago, Los Angeles, and Pittsburgh). We use d data that h ad been collected from 1994 2014 to test the specific aims of the current dissertation The MACS ( Dudley et al., 1995; Detels et al., 1992, Kaslow et al., 1987 ) and WIHS ( Bacon et al., 2005; Barkan et al., 1998 ) are well established, nat ional multicenter cohorts of men who have sex with men (MSM) and of women, respectively, living with or at risk for HIV infection. The MACS recruited MSM across three waves, in 1984 1985 (n=4954), 1987 1991 (n=668), and 2001 2003 (n=1350). Women were recru ited in WIHS across two waves, in 1994 1995 (n=2625) and 2001 2002 (n=1141). The data were collected through structured interviews, and standardized physical, psychological, and laboratory assessments. HIV status was assessed by enzyme linked immunosorbent assay (ELISA) with Western blot for confirmation at baseline for HIV+ participants, and semi annually for HIV participants. HIV sero conversion was confirmed by testing HIV participants at each semi annual visit. Written informed consent was obtained pr ior to each semi annual assessment for both cohorts. The questionnaires are available online for MACS at http://aidscohortstudy.org and for WIHS at https://sta tepi.jhsph.edu/wihs/wordpress/. The WIHS Cardiovascular Substudy consists of 1,321 HIV infected women aged 25 to 60 years, with no history of heart surgery or coronary angioplasty/stent placement before HIV infection. Mean age ranged from 40.4 42.2 years (Parrinello et al., 2012) 60.4% of the sample is African American, 22.6% White, and 17% o ther. A bout 29% of the sample was of Hispanic

PAGE 32

32 ethnicity. The mean CD4 + T cell count was between 377 462, and mean HIV RNA log 10 copies/mL was between 2.5 3.7 (Parrinello et al., 2012) The MACS Cardiovascular Substudy consists of 828 HIV infected MSM over 40 year s of age, under 300lbs, with no history of heart surgery or coronary angioplasty/stent placement before HIV infection (Miller et al., 2014) Mean age ranged from 49.2 56.7 years, 61.4% of the sample was White, 32.2% African American, and 6.4% other. Regard less of race, 10.9% of the sample was of Hispanic ethnicity. The mean CD4 + T cell count was between 597 628, and mean HIV RNA log 10 copies/mL was between 2.4 3.2 (Miller et al., 2014) At the first subclinical CVD assessment, 10.3% of HIV infected women (C rystal et al., 2011) and 23.5% of men were identified as cases with subclinical CVD (Monroe et al., 2012) Also at first assessment, drinking ranged from abstinence (women=54%, men=20%), 1 2 drinks per week (women=35%, men= 56%), and over 2 drin ks per week (women=11%, men=24%; Kaplan et al., 2008) In addition to the standard data collection for the MACS and WIHS, participants in the cardiovascular sub studies underwent high resolution B mode carotid artery ultrasounds of 6 locations in the right carotid ar tery (the near and fall walls of the common carotid artery [CCA], carotid bifurcation, and internal carotid artery [ICA]; Hodis et al., 2001) using a standardized protocol across study sites (Kaplan et al., 2008). Quality control and reliability of the ca rotid artery ultrasound measurement was performed among a subset of WIHS and MACS participants and was found to have high intraclass correlations (ICC) in both WIHS (variation coefficient = 1.8%; ICC = 0.98) and MACS (variation coefficient = 1.0%; ICC=0.99 ; Kaplan et al., 2008 ). Measures Independent P redictor Alcohol c onsumption The WIHS and MACS collected data on alcohol consumption semiannually. Participants were asked how many days on average they consumed alcohol

PAGE 33

33 (frequency) and how many standard units of alcohol were consumed on a drinking day (quantity) The average number of alcoholic beverages consumed per week was calculated by multiplying the frequency by the quantity of use at each semiannual visit Dependent O utcome Atherosclerosis Atherosclerosis was measured using B mode carotid artery ultrasound between 2004 2013 P resence of an arterial lesion or plaque, which was a focal carotid intima media thickness over 1.5mm (Stein et al., 2008), was measured up to 2 times from 2004 2013 Ca rotid intima media thickness at the far right common carotid artery (CCA IMT) was measured up to 4 times in WIHS and up to 3 times in MACS from 2004 2013 The CCA IMT was assessed using the B mode carotid artery ultrasounds by automated computerized edge d etection of the images. Development of atherosclerosis was defined in three ways. First, lesions or plaques present at baseline assessment were considered prevalent cases. Second, participants that screened negative for lesions or plaques at the baseline a ssessment, but were screened positive at the follow up were considered incident cases. Third, as change in CCA IMT from the baseline to each subsequent follow up assessments. Cova riates of I nterest Conceptual model with covariates are shown in Figure 1 1. Demographics factors reported date of birth. Race was self reported and categorized as white, black, and Asian/Pacific Islander or Native American/Alaskan. Annual income was self reported at each visit and categorized, based on natural cut offs in the data, as < $10,000, $10,00 0 Behavioral factors Self reported smoking was assessed in number of packs smoked using standardized categories: less than half a pack per day; at le ast half a pack but less than one

PAGE 34

34 pack per day; at least one but less than two packs per day; two or more packs per day. Cumulative pack years were calculated to determine the average pack, multiplied by 0.5 (half a year to reflect the timeframe of the sem inannual visits) and summed across the years up to the baseline and follow up assessments. Self reported illicit drug use was dichotomous and measured by asking if participants used any of the following: crack or any form of cocaine; uppers (including cry stal, methamphetamines, speed, ice); heroin or other opiates. Depressive symptoms were assessed at each semi annual visit with the Center for Epidemiology Studies Depression Scale (CES D Radloff, 1977). Some research has found that utilizing the score of 16 or greater may inflate the rate of depression among PLWH, due to the overlapping somatic symptoms that may be present due to HIV infection (Kalichman et al. 2000). Therefore, a score of 23 or greater was considered probable depression. HIV related fact ors Plasma HIV RNA viral load and CD4+ T cell count was measured using standard laboratory techniques. HIV RNA viral load was categorized as suppressed (< 200 AIDSinfo 2016); CD4+ T cell count was categorized 500 cells/mm3), or low (< 300 cells/mm3). Cumulative ART use was calculated by adding the weighted ART use variable to reflect years of ART use by the end of the 10 year follow up period. Optimal ART adherence was de previously been associated with sustained viral suppression (Low Beer et al., 2000; Paterson et al., 2000 ). Cardiovascular related factors Cardiovascular risk factors incl uded body mass index (BMI ; underweight <18.5 kg/m 2 normal 18.5 24.9 kg/m 2 overweight >24.9 kg/m 2 ),

PAGE 35

35 have hypertension) and diabetes (dichotomized as having bee n diagnosed with diabetes versus no history of diabetes). The Framingham Risk Score (Wilson et al., 1998) was calculated, using the gender based algorithms, including the following variables: age, total cholesterol, high density lipoprotein (HDL) cholester ol, systolic blood pressure, and smoking status. Therefore, we did not adjust for these variables outside of this risk score in analyses where the Framingham Risk Score was used The Framingham Risk Score values range from negative to positive, with negati ve values indicating low risk and positive values indicating high risk. Data A nalysis Throughout all aims of the dissertation, Group based Trajectory Modeling (GBTM) was used to categorize alcohol consumption patterns and warrants discussion There are several situations in which GBTM is appropriate and needed. First, GBTM is particularly important when a predictor or outcome is one that often changes overtime. When we have time dependent data, it is m ore accurate to characterize that chang e, a s opposed to using one time point and no assessment of change over time. Second, GBTM also allows us to identify individual variability of a mean population trend. We can better assess the qualitative dimensions of changes that occur over time or with age in a particular behavior, such as alcohol use. Therefore, GBTM allows us to take a more person based approach to analyze change, and to identify distinct patterns of change that are substant ively meaningful, and that do not assume a general mean change pat tern of an entire sample. Third, GBTM allows us to identify distinctions between important subgroups of a population of interest that is more representative of the natural setting, as most populations are heterogeneous in nature. Fourth, GBTM are very hel pful when there is more than linear change over time. Because the analysis allows us to identify linear, quadratic, and cubic patterns of a particular behavior, we can more accurate assess how change occurs that

PAGE 36

36 is beyond a simple linear representation. La stly, a benefit of the GBTM includes the a bility to estimate proportion of the sample that are characterized in a specific trajectory or subgroup There are limitations to GBTM that should be considered. The GBTM is a semi parametric and probabilistic mode l that estimates grouped trajectories of the most similar individual patterns. Therefore, each trajectory group does not fully describe the individual level patterns contained within them and should not be considered absolute. Further, the validity of GBTM relies heavily on the professional judgment of the investigator This is specifically regarding the number of trajectories and the type of change (linear, quadratic, etc.) to specify. Therefore, it is of high importance to assess the existing literature o n the variable in question to understand the patterns that are already known to exist. This method is also not suitable for variables that change rarely over time, such as chronic illnesses that are quite stable. IRB review All MACS and WIHS participants p rovided written informed consent for overall study and sub study participation. This specific analysis was approved by MACS and WIHS and the Institutional Review Board at the University of Florida

PAGE 37

37 CHAPTER 3 ASSOCIATION BETWEEN ALCOHOL CONSUMPTION TRAJECTORIES AND CLINICAL PROFILES AMONG MEN AND WOMEN LIVING WITH HIV Introduction Alcohol use is common among persons living with HIV (PLWH) and is reported among 39 81% ( Bilal et al., 2016; Conen et al., 2009 ; Monroe et al., 2016; Wandeler et al., 2016 ; Sullivan et al., 2011 ). Prevalence of heavy drinking has been reported in as much as 25 45% ( Deiss et al., 2016; Monroe e t al., 2016; Kader et al., 2014 ) of PLWH, with alcohol dependence ranging from 5.5 10% ( Jolley et al., 2016; Malbergier et al., 2015; Surah et al., 2013; Sullivan et al., 2011 ). Alcohol consumption, in general, is negatively associated with completing the steps of the HIV care continuum (Vagenas et al., 2015) and heavy alcohol use is associated with poor retention in HIV care and lower visit adherence, compared to those who do not drink Likewise, heavy alcohol consumption is associated with decreased antiretroviral (ART) adherence ( Pellowski et al., 2016; Malbergier et al. 2015; Kalichman et al., 2014; Tran et al., 2014 ) lower CD4+ T cell count (Kahler et al., 2015; Malbergier et al., 2015 ), and increased viral load (Deiss et al., 2016 ). Aside from the relationship between alcohol consumption and ART adherence, alcohol abuse has been linked specifically to HIV progression through alter ation of viral infectivity, inflammatory biomarkers, immune response, and tissue injury ( Monnig et al., 2016; Molina et al. 2014 ). Heavy drinking in this population is also associated with engagement in risky health behavior, such as cigarette ( B raithwai te et al., 2016; Pacek et al., 2014) and substance use (Parsons et al., 2014 ), which can lead to other chronic illnesses. Some studies have found that PLWH who use alcohol have increased chronic comorbidity ( Balal et al., 20 16, Jolley et al., 2016; Willia ms et al., 2016; Kelso et al., 2015 ), while other studies h ave found no such association ( Kelly et al., 2016; Tsui et al., 2016; Wandeler et al., 2016 ; Fuster et al., 2013) Others have found a J curve association between alcohol consumption and risk for c hronic

PAGE 38

38 illness. For example, Wan deler et al. (2016) found that, among PLWH, low and moderate drinkers had significantly lower risk for cardiovascular disease events or death, compared to non drinkers. Because many of the aforementioned studies are cross se ctional, it is unclear whether moderate and heavy drinking leads to chronic illness or if alcohol use is a coping response to such illness. While similar findings have been shown in the general population, emerging evidence suggests that PLWH may be more a ffected by the harmful sequela of alcohol use when compared to similar or lower levels of use among uninfected groups ( Justice et al., 2016; McGinnis et al., 2016 ; Rentsch et al., 2011 ). Longitudinal studies of alcohol use patterns among PLWH have focused primarily on dichotomous measures of hazardous or heavy alcohol use, which limit the variability of alcohol consumption and can result in stagnated patterns over time ( Marshall et al., 2015a ; Jacob et al., 2013 ) These studies have also been limited by re latively short follow up (6 months 2 years) ( Marshall et al., 2015 a ; Mguez Burbano et al., 2014) or have synthesized longitudinal patterns by using lifetime recall of a lcohol use phases (Jacob et al., 2013 ). To our knowledge, there is limited research on levels of alcohol use aside from hazardous/heavy use over long periods of fo llow up. Cook et al. (2013) conducted a group based trajectory model (GBTM) of self reported number of drinks consumed per week to inform emerging patterns among HIV+ women in t 2006. This study found five trajectories of drinking, three of which described changing drinking behavior over time. These data, however, describe drinking patterns in the first half of the 20 year cohort s tudy and may not be relevant to drinking behavior in the post HAART era. Las tly, Marshall et al (2015 b ) conducted a longitudinal analysis of patterns of the Alcohol Use Disorder Identification Test Consumption questionnaire (AUDIT C ; Surah et al., 2013 ) sc ore among HIV+ men who have

PAGE 39

39 sex with men (MSM) from 2002 four stable trajectories, perhaps due to the use of the somewhat prescriptive AUDIT C score, which is used to identify alcohol use disorders (score ranging from 0 to 12) and has lower variability than that of self reported number of drinks per week, thus limiting the detection of change in drinking. While these two studies were conducted in different populations and examined inconsistent predi ctors of heavy alcohol use, illicit drug use was distinctly associated with heavy consumption. Because of limitations of the current literature, a gap exists regarding alcohol use changes over time among PLWH. Further, it is unclear if there are significan t clinical factors associated with long term moderate and heavy alcohol consumption by gender. Associated factors of alcohol consumption patterns would provide clinicians with the means to identify those with the greatest need for early intervention and al cohol abuse treatment. The goals of this analysis are to 1) describe patterns of alcohol consumption among PLWH from 2004 2013 by gender and 2) assess the association between time stable and varying clinical factors and long term heavy and moderate alcoho l consumption. By utilizing reported number of drinks per week, we hypothesized that distinct patterns will emerge that are descriptive of stable (i.e., consistent abstinent, consistent moderate, and consistent heavy) and changing alcohol use behavior (i.e ., abstinent to moderate or heavy drinking; heavy to moderate or abstinence) over time. We also hypothesized that clinical factors would be identified, specifically by gender, as important predictors of long term moderate and heavy alcohol consumption. Spe cifically, we hypothesized that those with poor clinical profiles would be more likely to be heavy or moderate drinkers, compared to those who are abstinent or low drinkers. While clinical associations of longitudinal alcohol consumption were the main focu s of this analysis, the biopsycho social theoretical

PAGE 40

40 framework (Engel, 1977 ) was used to conceptualize other non clinical factors ne eded for analytical adjustment. Methods Study Design and Participants The Multicenter AIDS Cohort Study (MACS) ( Dudley et al., 1995; Detels et al., 1992; Kaslow et al., 1987 Bacon et al., 2005 ; Barkan et al., 1998 ) are well established, national multicenter cohorts of men who have sex with men (MSM) and of women, respectively, living with or at risk for HIV infection. Participants from MACS were recruited from the following metropolitan areas: Baltimore, MD; Washington, DC; Chicago, IL; Pittsburgh, PA; Los Angeles, CA. Participants from WIHS were recruited from the following metropoli tan areas: Brooklyn and Bronx, NY; Washington, DC; Chicago, IL; Los Angeles and San Francisco, CA. The MACS recruited MSM across three waves, in 1984 1985 (n=4954), 1987 1991 (n=668), and 2001 2003 (n=1350). Women were recruited in WIHS across two waves, in 1994 1995 (n=2625) and 2001 2002 (n=1141). The data from these studies were collected from structured interviews, and standardized physical, psychological, and laboratory assessments. HIV status was assessed by enzyme linked immunosorbent assay (ELISA) with Western blot for confirmation at baseline for HIV+ participants, and semi annually for HIV participants. Sero conversion was confirmed by testing HIV participants at each semi annual visit using the aforementioned tests. Written informed consent was obtained prior to each semi annual assessment for both cohorts. The questionnaires are available online for MACS at www.statepi.jhsph.edu/macs/forms.html and for WIHS at https://statepiaps.jhsph.edu/wihs/index forms.htm. The current study utilized data fr om participants of the cardiovascular sub studies of the MACS and WIHS, to understand the associations between clinical profiles including cardiovascular disease risk factors (i.e.,

PAGE 41

41 Framingham risk score, BMI, diabetes) and alcohol consumption prior to car diovascular disease or related events. All MACS and WIHS participants provided written informed consent for overall study participation, and this specific analysis was approved by the Institutional Review Board at the University of Florida. Data Collection The cardiovascular sub study enrolled a subset of HIV+ WIHS participants (n=1,321), aged 25 60 years and the MACS enrolled a subset of HIV+ MSM (n=828), over 40 years of age and under 300 lbs. Participants who seroconverted during the study and those with less than 4 alcohol consumption assessments were excluded (WIHS n=198; MACS n=231). The median person years of follow up between 2004 2013 were 6.2 years [interquartile range (IQR): 6.0 7.5 years] for WIHS participants and 8.5 years (IQR: 8.0 10. 0 years) for MACS participants. Main outcome measure Alcohol consumption was measured via self report by asking about the average frequency (number of days per week) and quantity (number of drinks per drinking day) of use. The average number of drinks per week was calculated by multiplying the frequency by the quantity; consumption was categorized as abstinence to low (<1 drink per week), moderate (1 7[14] drinks per week for women [men]), or heavy use (> 7[14] dr inks per week for women [men]). Independent variables Clinical and b iological reported date of birth. Use of ART was reported at each visit and weighted by t he reported adherence of ART (Shoptaw et al., 2012 ). Cumulative ART was calculated by adding the we ighted ART use variable to reflect years of ART use by the end of the 10 year follow up period. Optimal ART

PAGE 42

42 has previously been associated with sustained viral suppression ( Low Beer et al., 2000; Paterson et al., 2000) Plasma HIV RNA viral load and CD4+ T cell count s were measured, semi annually, using standard laboratory techniques. HIV RNA viral load was subsequently categorized as undetectable (< 200 copies/ mL) or d CD4+ T 500 cells/ mm3), or low (< 300 cells/mm3). Diabetes was dichotomized as having ever been diagnosed with diabetes at any time durin g follow up versus no history of diabete s. The Framingham Risk Score (Wilson et al., 1998 ) was calculated, using the gender based algorithms, including the following variables: age, total cholesterol, high density lipoprotein (HDL) cholesterol, systolic bl ood pressure, and smoking status. Therefore, we did not adjust for these variables outside of this risk score. The Framingham Risk Score values range from negative to positive, with negative values indicating low risk and positive values indicating high ri sk. Body mass index was based on weight and height, and participants were categorized as being underweight (BMI <18.5 kg/m2), normal (18.5 24.9 kg/m2), and overweight (> 24.9 kg/m2). Psychological Depressive symptoms were assessed at each semi annual visi t with the Center for Epidemiology Studi es Depression Scale (CES D, Radloff, 1977 ). Some research has found that utilizing the score of 16 or greater may inflate the rate of depression among PLWH, due to the overlapping somatic symptoms that may be present du e to HIV infection (Kalichmann et al., 2000 ). Therefore, a score of 23 or greater was considered probable depression. Self reported illicit drug use was dichotomous and measured at each visit by asking if participants used any of the following: crack or any form of cocaine; uppers (including crystal, methamphetamines, speed, ice); heroin or other opiates.

PAGE 43

43 Social Race/ethnicity was self reported and categorized as non Hispanic white, non Hispanic black, and other races (Hispanic; Asian/Pacific Islander; Native American/Alaskan). Annual income was self reported at each visit and categorized, based on natural cut offs in the data, as < $10,000, $10,000 To adjust for missing data related to unmeasured variables, percentage of mi ssing follow up was calculated by summing the number of eligible visits missed, divided by the total number of eligible visits for each individual; wave of enrollment was included in the multivariable models. Data Analyses Univariate and bivariate analyses were conducted to assess frequencies and proportions of clinical factors and covariates. To describe patterns of alcohol consumption over time, we conducted group based trajectory modeling (GBTM). In the first modeling step, we assessed linear patterns of 3 5 groups, as suggested by previous research ( Marshall e t al., 2015a; Marshall et al., 2015b; Cook, et al., 2013 ) Goodness of fit was assessed at each step using the Akaike information criteria (AIC) and Bayesian information criteria (BIC; smaller the v alues, better the probability that any one group based trajectory adequately captures the individual patterns. Therefore, an individual pattern was assigned into the group pattern with the highest probability of group membership. Models with PP and/or model entropy values <0.7 were rejected ( Andruff et al., 2009 ). The 95 % confidence intervals (CI) of the resulting patterns were used to qualitatively assess the stability of the trajectories. Models with small CIs of trajectories were favored over wide CIs.

PAGE 44

44 Using repeated measures of alcohol consumption and clinical factors multivariate generalized estimating equations (GEE) were conducted to assess longitudinal associations between clinical factors and moderate (1 7 [14] drinks per week for women [men]) and heavy (>7[14] drinks per week for women[men]) alcohol consumption compared to abstinent/low use (<1 drinks per week), stratified by gender. Clinical factors were considered significantly associated with alcohol consumption at the p<.05 level. All statistical analyses were conducted using SAS 9.4 (SAS Institute, Inc., Car y, NC). Results Alcohol Consumption Trajectories Baseline characteristics by cohort are presented in Table 3 1 A five group trajectory model emerged as the be st f itting model for women (Figure 3 1 ; model entropy 0.89). Alcohol consumption patterns were la consumption throughout 10 low/abstinence years), and years). A five group trajectory model was the best fitting model among men (Figure 2 44% PP 0.98, very little to no alcohol consumption throughout 10 PP 0.98, moderate consumption throughout 10 10 years). While similar patterns emerged across gender, some differences became apparent. For instance, while women were less likely to drink, in general, membership in the increasers group

PAGE 45

45 was slightly greater among women than men (WIHS 15% vs. MACS 11%). Men were also more likely to be in the decreasers group tha n women (WIHS 7% vs. MACS 14%). Multivariate Analysis of Clinical Factors on Alcohol Consumption among Men Moderate drinking Bivariate and multivariab le analyses are shown in Table 3 2. In multivariate analysis, illicit drug use was associated with 2.21 times the odds for moderate drinking (CI 1.44 3.39, p<.001), compared to abstinent/lo w use. Those with diabetes had 53% lower odds for moderate drinking (CI 0.30 0.73, p<.001), and those with CD4 count < 300 had 0.93, p<.05). Heavy drinking Bivariate and multivariab le analys es are shown in Table 3 4. In multivariable analysis, longitudinal illicit drug use was associated with 2.28 (CI 1.16 4.49, p=.02) times higher odds for heavy drinking. Men diagnosed with diabetes had 67% (CI 0.11 1.01, p=0.05) lower odds for heavy drinkin g, compared to abstinent/low use. Multivariate Analysis of Clinical Factor s on Alcohol Consumption among Wom en Moderate drinking Bivariate and multivariab le analyses are shown in Table 3 3. In multivariate analysis, those with illicit drug use had nearly 3 times higher odds (CI 1.97 4.42, p<.001) for moderate drinking, compared to abstinent/low use. Suboptimal adherence to ART and detectable viral load were associated with 1.21 (CI 1.02 1.44, p<.05) and 1.37 (CI 1.10 1.69, p<.01) times the odds, respective ly for moderate drinking. Each unit increase in the Framingham risk score was associated with a 7% increase in odds for moderate drinking (CI 1.50 1.10, p<.001). Heavy drinking Bivariate and multivariabl e analyses are shown in Tables 3 5. In multivariable analysis, longitudinal illicit drug use was associated with over 6 times the odds (CI 3.56 13.4, p<.001) for heavy drinking, compared to abstinent/low use. Women with detectable viral load had 1.55 times the odds for heavy drinking (CI 1.04 2.32, p<0.05). Each increased

PAGE 46

46 year of ART use was associated with 7% lower odds for heavy drinking (CI 0.88 0.98, p<.01). Each unit increase in the Framingham risk score was associated with 12% increase in odds (CI 1.05 1.20, p<.001) for heavy drinking, compared to abst inent/low use. Discussion We aimed to describe alcohol consumption trajectories over time and to assess the longitudinal associations between clinical factors and moderate and heavy alcohol consumption. While several alcohol patterns characterized a stable level of consumption (i.e., low/abstinence, moderate, heavy), some patterns also featured shifts in drinking overtime. These shifting patterns add to the existing literature, as other researchers have found mainly stable consumption patterns, likely due t o utilizing dichotomous measures of hazardous or heavy drinking ( Marshall et al., 2015a; Jacob et al., 2013 ). This indicates that alcohol consumption should be measured longitudinally to accurately depict exposure. While women tended to drink less, in gene ral, they had slightly higher membership in the increasers trajectory and lower membership in the decreasers trajectory than men. Because these trajectories were increasing to or decreasing from the moderate consumption group, some may consider these resul ts without consequence. However, given the fact that it is relatively unknown whether moderate use confers health benefits or harms among PLWH, these results could suggest that women are a target for prevention/intervention strategies. This is specifically important when considering the lower threshold of number of dr inks needed for intoxication (McGinnis et al., 2016 ) and given the evidence that only 30 drinks per month (ie., moderate use) is associated with increased risk for physical injury and death in this population (Justice et al., 2016 ), far exceeding risk compared to the 70 drinks per month needed for similar impact among HIV individuals. Also of significance is the lack of a decreasing trajectory from the heavy pattern across both men and women,

PAGE 47

47 s uggesting that once heavy consumption becomes relatively common, this behavior remains overt ime. Results from the multivariate GEE models suggest that there are significant longitudinal clinical associations of moderate and heavy consumption that may help distinguish individuals for prevention and/or early intervention. The most significant associated factors of moderate and heavy alcohol consumption, across both MACS and WIHS cohorts, was longitudinal illicit drug use. This is con sistent with cross section al (Parsons et al., 2014) and longitudinal (Ruggles et al., 2016 ) research on the concordance of substance and alcohol use among PLWH. The Framingham risk score was associated with increased odds for moderate and heavy alcohol consumption among women. Diab etes, however, was associated with decreased odds for moderate and heavy drinking among men, which may be due to recommendations from care providers to reduce or stop drinking due to declining health or risk for clinical illness. Conversely, this associati on could also be indicative of a protective effect of alcohol consumption on diabetes, described in research among the general population ( Knott et al., 2015; Pietraszek et al., 2010; Carlsson et al. 2005 ) Among women, sub optimal ART adherence was associ ated with increased odds for moderate alcohol consumption. Furthermore, among women and controlling for ART adherence, having a HIV RNA viral load of 200 or greater was associated with increased odds for membership in the moderate and heavy consumption pat terns. This is consistent with previous research indicating that alcohol abuse is linked to HIV progression through al teration of viral infectivity (Deiss et al., 2016 ), inflammatory biomarkers, immune response, and tissue injury (Monnig et al., 2016 ; Molina et al., 2014 ). Conversely, men with lower CD4 count were less likely to be moderate drinkers, compared to the abstinent/low group, suggesting a protective effect.

PAGE 48

48 Limitations The readers should consider some limitations of the current study. First, alcohol consumption quantity and frequency were assessed via self report and is subject to recall and social desirability biases, which likely resulted in underestimated reports of alcohol consumption. Second, there are significant demographic differences between the WIHS and MACS cohorts, making direct comparisons of stratified GBTM analyses difficult. It is possible that any differences found may be due to differences in social factors between these cohorts. Third, we restricted our analyses to participan ts with at least 4 alcohol consumption assessments in order to estimate stable trajectory models. Therefore, it is possible that different trajectories could have emerged had we not excluded these participants. Fourth, those with heavy drinking and comorbi dities may have been more likely to drop out of the study or die. This could have affected the results relating to alcohol consumption and clinical conditions, making heavy consumption seem less common among those with diabetes or progressed HIV infection, when, in fact, there may have been a true positive association. Lastly, GBTM is a semi parametric and probabilistic model that estimates grouped trajectories of the most similar individual patterns. Therefore, each trajectory does not fully describe the i ndividual level patterns contained within them and should not be considered absolute. Conclusions In summary, the current study added to existing literature on the proportion of HIV+ persons who consume alcohol at specific levels, particularly moderate and heavy consumption. Because alcohol consumption patterns were not limited to characterize only heavy use, and rather were allowed to describe the course of difference levels of use, this study provided information regarding changes in alcohol consumption o ver time. The results also reveal characteristics that can be used to identify those at risk for moderate and heavy consumption.

PAGE 49

49 The U.S. Preventive Services Task Force recommends that clinicians assess all adults aged 18 years and older for alcohol misuse and to provide support to reduce risky alcohol consumptio n (US Preventive Services Task Force, 2 013 ). Further, several screening and brief intervention tools have been developed specifically for clinical use in the general and s pecif ic clinical populatio ns (Saitz et al., 2016 ). In line with these recommendations, clinicians should consider screening all patients for alcohol consumption, particularly if patients report current and past illicit drug use, suboptimal ART adherence, and if patients have detect able viral load. Clinicians could also consider assessing moderate alcohol consumption, as this study found detrimental associations of moderate use on adherence and viral load, particularly among women.

PAGE 50

50 Table 3 1. Baseline characteristics of persons liv ing with HIV by cohort WIHS (N=1123) MACS (N=597) Baseline Characteristics Frequency (Column Percentage) Race White African American/Black Other 248 (22) 676 (60) 199 (18) 311 (52) 225 (38) 61 (10) Age (continuous), mean (SD) 45.0 (7.6) 56.9 (7.7) Annual Income < $10,000 $10,000 $30,000 551 (51) 339 (31) 189 (18) 150 (31) 126 (26) 211 (43) Probable depression No Yes 895 (80) 228 (20) 514 (86) 83 (14) Illicit drug use No Yes 990 (92) 81 (8) 400 (78) 112 (22) Ever diagnosed with diabetes No Yes 829 (74) 297 (26) 388 (65) 209 (35) Body Mass Index < 18.5 18.5 24.9 25.0 487 (43) 225 (20) 411 (37) 133 (22) 228 (38) 236 (40) HIV RNA Viral Load < 200 copies/mL 200 copies/mL 628 (56) 495 (44) 417 (70) 180 (30) CD4+ T cell count 500 cells/mm 3 300 500 cells/mm 3 < 300 cells/mm 3 443 (39) 323 (29) 357 (32) 264 (44) 157 (26) 176 (30) HIV ART Adherence < 95% 514 (46) 609 (54) 163 (32) 353 (68) Cumulative ART exposure, mean (SD), years 12.1 (5.0) 9.4 (4.0) Framingham Risk Score, mean (SD) 8.4 (6.0) 11.1 (3.3)

PAGE 51

51 Figure 3 1 Study The solid lines represent predicted probabilities of alcohol consumption; the dotted lines represent the actual probabilities of alcohol consumption.

PAGE 52

52 Figure 3 2 Trajectories of alcohol consumption among 597 HIV+ men in the Multicenter AIDS Cohort Study The solid lines represent predicted probabilities of alcohol consumption; the dotted lines represent the actual probabilities of alcohol consumption.

PAGE 53

53 Table 3 2. Multivariable analysis of associated factors of the moderate alcohol use compared to the low/abstinent alcohol use among men living with HIV Characteristics OR 95% CI P Value AOR 95% CI P Value Race (Ref= White) African American/Black Other 0.39 0.23 0.27 0.58 0.11 0.11 <.001 <.001 0.53 0.27 0.32 0.90 0.12 0.64 .02 <.01 < $10,000 $10,000 $30,000 0.50 0.55 0.32 0.76 0.35 0.84 <.01 <.01 0.78 0.69 0.46 1.30 0.43 1.11 .33 .13 BMI status (Ref = Normal) Underweight Overweight 0.29 0.83 0.21 0.41 0.59 1.17 <.001 .30 0.86 0.86 0.58 1.28 0.59 1.26 .46 .45 Diabetes (Ref = No) 0.49 0.33 0.73 <.001 0.47 0.30 0.73 <.001 Probable depression (Ref = No) 1.00 0.71 1.42 .98 0.85 0.58 1.24 .40 Illicit drug use (Ref = No) a 1.80 1.22 2.64 <.01 2.21 1.44 3.39 <.001 0.79 0.56 1.11 .18 1.01 0.68 1.50 .95 CD4+ T 300 500 < 300 0.79 0.26 0.59 1.06 0.19 0.38 .11 <.001 0.80 0.61 0.57 1.11 0.40 0.93 .18 .02 Detectable viral load (Ref <200) 0.88 0.64 1.22 .44 0.93 0.64 1.36 .72 Age, years 0.97 0.94 0.99 <.01 0.99 0.95 1.04 .84 Cumulative ART exposure, years 1.11 1.05 1.16 <.001 0.97 0.90 1.05 .53 Framingham risk score 0.95 0.90 1.01 .10 0.98 0.90 1.07 .70

PAGE 54

54 Table 3 3 Multivariable analysis of associated factors of the moderate alcohol use compared to the low/abstinent alcohol use among wo men living with HIV Characteristic OR 95% CI P Value AOR 95% CI P Value Race (Ref= White) African American/Black Other 1.09 0.77 0.74 1.61 0.46 1.30 .65 .33 0.87 0.75 0.59 1.30 0.43 1.30 .51 .31 < $10,000 $10,000 $30,000 1.10 1.06 0.76 1.59 0.73 1.52 .60 .76 0.90 1.01 0.62 1.29 0.70 1.47 .56 .94 BMI status (Ref = Normal) Underweight Overweight 0.69 0.83 0.48 1.00 0.58 1.18 .05 .30 1.06 0.94 0.64 1.76 0.65 1.35 .80 .73 Diabetes (Ref = No) 0.83 0.58 1.18 .29 0.86 0.59 1.25 .43 Probable depression (Ref = No) 1.61 1.26 2.06 <.001 1.17 0.91 1.51 .21 Illicit drug use (Ref = No) 3.93 2.71 5.70 <.001 2.95 1.97 4.42 <.001 1.06 0.88 1.28 .52 1.21 1.02 1.44 .02 CD4+ T 300 500 < 300 1.13 0.76 0.56 1.03 0.56 1.03 .31 .08 1.01 0.95 0.79 1.28 0.68 1.33 .95 .78 Detectable viral load (Ref <200) 2.20 1.78 2.73 <.001 1.37 1.10 1.69 <.01 Age, years 0.98 0.96 1.00 .12 0.96 0.93 0.98 <.01 Cumulative ART exposure, years 0.98 0.96 1.01 .32 0.97 0.94 1.00 .07 Framingham risk score 1.05 1.02 1.07 <.001 1.07 1.04 1.10 < .001

PAGE 55

55 Table 3 4 Multivariable analysis of predictors of the heavy alcohol use compared to the abstinent/low alcohol use among men living with HIV Characteristics OR 95% CI P Value AOR 95% CI P Value Race (Ref= White) African American/Black Other 0.26 0.41 0.10 0.64 0.13 1.29 <.01 .13 0.37 0.71 0.10 1.31 0.19 2.57 .12 .60 < $10,000 $10,000 $30,000 0.17 0.24 0.05 0.54 0.09 0.60 <.01 <.01 0.19 0.23 0.05 0.75 0.09 0.59 .02 <.01 BMI status (Ref = Normal) Underweight Overweight 0.48 0.67 0.22 1.03 0.34 1.31 .06 .24 1.39 0.53 0.62 3.10 0.25 1.10 .42 .09 Diabetes (Ref = No) 0.30 0.11 0.79 .01 0.33 0.11 1.01 .05 Probable depression (Ref = No) 1.08 0.55 2.12 .82 1.07 0.53 2.17 .85 Illicit drug use (Ref = No) 2.12 1.12 3.99 .02 2.28 1.16 4.49 .02 Sub 0.91 0.47 1.76 .78 0.82 0.38 1.79 .62 CD4+ T 300 500 < 300 1.17 0.53 0.68 2.02 0.25 1.09 .56 .08 1.08 0.91 0.59 1.95 0.44 1.87 .81 .80 Detectable viral load (Ref <200) 0.92 0.46 1.81 .80 1.03 0.47 2.28 .94 Age, years 0.97 0.92 1.01 .15 1.01 0.93 1.09 .83 Cumulative ART exposure, years 1.05 0.95 1.16 .33 0.91 0.81 1.02 .12 Framingham risk score 0.98 0.89 1.08 .71 1.10 0.94 1.28 .24

PAGE 56

56 Table 3 5 Multivariable analysis of predictors of the heavy alcohol use compared to the abstinent/low alcohol use among women living with HIV Characteristic OR 95% CI P Value AOR 95% CI P Value Race (Ref= White) African American/Black Other 0.99 0.38 0.52 1.89 0.14 1.04 .66 .06 0.99 0.29 0.34 1.27 0.09 0.91 .21 .03 < $10,000 $10,000 $30,000 0.93 0.89 0.50 1.73 0.46 1.69 .82 .71 0.54 0.74 0.30 1.00 0.38 1.42 .05 .36 BMI status (Ref = Normal) Underweight Overweight 0.62 0.54 0.32 1.18 0.27 1.08 .14 .08 0.85 0.72 0.31 2.35 0.34 1.50 .75 .38 Diabetes (Ref = No) 0.92 0.49 1.74 .80 1.06 0.52 2.17 .88 Probable depression (Ref = No) 1.90 1.19 3.02 <.01 1.05 0.68 1.64 .81 Illicit drug use (Ref = No) 9.19 5.19 16.2 <.001 6.91 3.56 13.4 <.001 Sub 1.39 0.97 1.99 .07 1.11 0.85 1.44 .45 CD4+ T 300 500 < 300 0.99 0.89 0.58 1.69 0.50 1.59 .97 .70 0.80 0.97 0.47 1.35 0.50 1.85 .40 .92 Detectable viral load (Ref <200) 2.90 1.95 4.32 <.001 1.55 1.04 2.32 .03 Age, years 1.01 0.97 1.05 .71 0.96 0.90 1.01 .15 Cumulative ART exposure, years 0.95 0.90 1.00 .05 0.93 0.88 0.98 <.01 Framingham risk score 1.12 1.06 1.18 <.001 1.12 1.05 1.20 < .001

PAGE 57

57 CHAPTER 4 THE IMPACT OF LONG TERM MODERATE AND HEAVY ALCOHOL CONSUMPTION PATTERNS ON SUBCLIN ICAL ATHERO SCLEROSIS AMONG PERSONS LIVING WITH HIV Introduction M oderate alcohol consumption may be protective against CVD, while heavy use is associated with increased risk for CVD in the general population ( Mukamal et al., 2003a, Reynolds et al., 2003; Corrao et al., 2000; Sacco et al., 1999; McElduff and D obson, 1997 ) However, t he relationship between alcohol consumption a nd CVD has not been sufficiently examined among PLWH. Heavy alcohol consumption among PLWH is nearly twice the rate compared to uninfected populations (Galvan et al., 2002) Hazardous drinking was reported in 58% of PLWH who received HIV care and consumed alcohol in the past 6 months (Stein et al., 2005) Among HIV infected veterans, 20% and 33% screened positive ly for hazardous and binge drinking, respectively, with 32% having been diagnosed with an alcohol use disorder (Conigliaro et al., 2003) There is little research investigating the association between alcohol consumption and subclinical cardiovascular disease, also known as atherosclerosis. A therosclerosis is characterized by arterial plaque s that may narrow the lumen, decreas e blood flow and consequ ently predispose individuals to acute thrombotic events (National Institutes of Health, 2011) In the general population, light and moderate alcohol use has been cross sectionally associated with lower risk for carotid artery plaque (Kohsaka et al., 2011) and stiffness (Hougaku et al., 2005), compared to abstinence. Heavy alcohol use, however, has been shown to significantly increase carotid intima medial thickness (cIMT; Zyriax et al., 2010) and stiffness (Hougaku et al., 2005), consistent with a J curved association found in the literature (Xie et al. 2012; Mukamal et al., 2003 b ). Similarly, a longitudinal study of 20 year drinking patterns found

PAGE 58

58 that consistent heavy use was associated with a significant increase in cIMT, compared to consistent moderate use (Britton et al., 2016). Other studies have found no significant association between alcohol consumption and cIMT or presence of carot id plaques (Bauer et al., 2013; Zureik et al., 2004 ). Further, some studies have found significant associations betwee n alcohol consumption and subclinical disease in men, but not in women (Zyriax et al., 2010; Lee et al., 2009; Schminke et al., 2005 ). The mechanism by which alcohol consumption is thought to effect cardiovascular health is not well understood, and of grea t practical importance given the widespread global consumption of alcohol (Freiberg and So Armah, 2016) Biological and behavioral mechanisms have been proposed to account for the higher burden of CV D among PLWH. First, heavy alcohol consumption and CVD ar e affected by demographic and psych osocial factors, including age race/ethnicity, and socioeconomic status (Conen et al., 2009; Galvan et al., 2002 ), all of which also tend be associated with HIV infection risk Second, alcohol use is significantly associ ated with traditional CVD risk factors, including insulin resistance (type II diabetes ; Mguez Burbano et al., 2009 ), tobacco use (Cook et al., 2013) and illicit drug use ( Chitsaz et al., 2013; Cook et al., 2013; Conen et al., 2009 ) Third, HIV infection alone increases systemic inflammation (Bahrami et al., 2016; Shrestha et al., 2014 ) and immune activation (Maniar et al., 2013; Neuhaus et al., 2010; Hsue et al., 2009; Strategies for Management o f Antiretroviral Therapy Study Group et al., 2006 ), pathophy siologic responses that contribute to the risk for CVD (Bahrami et al., 2016, Hsu et al., 2016, Hansson, 2005) Chronic inflammation and immune activation can lead to the breakdown of the endothelial walls of the gastrointestinal tract, a process that lead s to microbial translocation which triggers further immune and pro inflammatory Klatt el al., 2013 ; Maniar et al., 2013 ) While

PAGE 59

59 low level consumption of alcohol may have favorable lipid or antithrombotic effects, low levels of alcohol use have been shown to increase systemic inflammation, as well as risk for microbial translocation (Brenchley and Douek, 2012). A recent systematic r eview found the current state of the literature to be limited to mostly cross sectional studies and/or investigation of vague measures of alcohol use (e.g., any alcohol use, alcohol abuse/dependence history) to characterize risk among majority male HIV inf ected participants (Kelso et al., 2015) While these studies help us begin to understand the importance of alcohol consumption on cardiovascular health, the study participants were majority male (78 100 %) Further, most studies utilize only the medical rec ord to classify diagnosis of clinical CVD, and do not assess early stages of disease development such as atherosclerosis. In this study, we assess ed the presence of subclinical atherosclerosis (George and Movahed, 2008) by n on invasive carotid artery ultrasound tests The objective of the current analysis was to assess the association between 10 year patterns of alcohol use and the prevalence and incidence of subclinical atherosclerosis, measured by B mode carotid artery ultrasound imag ing Specifically, w e aimed to 1) test the association between long term moderate and heavy alcohol use and subclinical atherosclerosis among PLWH, and 2) to explore whether the relationships appeared to differ by gender and between prevalent and incident disease. We hypothesized that long term moderate and heavy alcohol use would be significantly associated with increased risk for prevalent and incident subclinical atherosclerosis. Materials and Methods Study Setting, Selection, and Inclusion Criteria The Multicenter AIDS Cohort Study (MACS; Dudley et al., 1995; Detels et al., 1992, Kaslow et al., 1987 tudy (WIHS; Bacon et al., 2005; Barkan et al., 1998 ) are well established, national multicenter cohorts of men who have sex wit h men

PAGE 60

60 (MSM) and of women, respectively, living with or at risk for HIV infection. Participants from MACS were recruited from the following metropolitan areas: Baltimore, MD, Washington, DC, Chicago, IL, Pittsburgh, PA, Los Angeles, CA. Participants from WI HS were recruited from the following metropolitan areas: Brooklyn and Bronx, NY, Washington, DC, Chicago, IL, Los Angeles and San Francisco, CA. The MACS recruited MSM across three waves, in 1984 1985 (n=4954), 1987 1991 (n=668), and 2001 2003 (n=1350). W omen were recruited in WIHS across two waves, in 1994 1995 (n=2625) and 2001 2002 (n=1141). The data were collected through structured interviews, and standardized physical, psychological, and laboratory assessments. HIV status was assessed by enzyme linke d immunosorbent assay (ELISA) with Western blot for confirmation at baseline for HIV+ participants, and semi annually for HIV participants. Written informed consent was obtained prior to each semi annual assessment for both cohorts. The questionnaires are available online for MACS at http://aidscohortstudy.org and for WIHS at https://statepi.jhsph.edu/wihs/wordpress/ The WIHS cardiovascular sub study recruited women aged 25 to 60 years, with no history of hea rt surgery or coronary an gioplasty/stent placement before HIV infection (n=1,321); The MACS cardiovascular sub study recruited men over 40 years of age, under 300lbs, and with no histo ry of heart surgery or coronary angioplasty/stent placement before HIV infection (n=828). The cur rent study focused on those with HIV sero prevalence at baseline and excluded those who sero converted (WIHS: n=118; MACS: n=216). For both cohorts, those in the 2001 2002 (MACS 2003) wave of enrollment were not included in the prevalence analysis ( taking place in 2004 2006), since there was not opportunity to measure 10 year alcohol consumption at that point reducing the sample of baseline assessment by n=403 in WIHS and n=264 in MACS However, those that were in the 2001 2003 wave of enrollment who had a baseline (and no

PAGE 61

61 prevalent disease) and follow up assessment were included for the incidence analysis (taking place in 2011 2013), since 10 year al cohol consumption data were available by that time point. Therefore, those in the prevalence and incident an alysis are not directly comparable. The final sample sizes were n=800 in WIHS and n=348 in MACS for prevalent carotid lesions (at least one ultrasound assessment) and n=512 in WIHS and n=324 in MACS for incident lesions (two ultrasound assessments). The m e dian person years of follow up between 1994 2014 was 16.7 years (interquartile range [IQR] : 1 6.0 18 .5 years) for WIHS participants and 12.4 years (IQR: 10.0 16.5 years) for MACS participants All MACS and WIHS participants provided written informed consent for overall study participation, and this specific analysis was approved by the Institutional Review Board at the University of Florida Data Collection In addition to the standard data collection for the MACS and WIHS, participants in the cardiovascular sub studies underwent high resolution B mode carotid artery ultrasounds of 6 locations in the right carotid artery (the near and fall walls of the common carotid artery [CCA], carotid bifurcation, and internal carotid artery [ICA] ; Hodis et al., 2001 ) using a standardized protocol across study sites (Kaplan et al., 2008) Quality control and reliability of the carotid artery ultrasound measurement was perf ormed among a subset of WIHS and MACS participants and was found to have high intraclass correlations (ICC) in both WIHS (variation coefficient = 1.8%; ICC = 0.98) and MACS (variation coefficient = 1.0%; ICC=0.99 ; Kaplan et al., 2008 ). Main outcome measure Subclinical atherosclerosis was defined as the presence of an arterial lesion or plaque, which was a focal carotid intima media thickness over 1.5mm (Stein et al., 2008) and was measured up to 2 times from 2004 2013 Two outcomes of interest were consider ed: prevalent and incident subclinical atherosclerosis. Lesions or plaques present at baseline assessment were

PAGE 62

62 considered prevalent cases. Participants that screened negative for lesions or plaques at the baseline assessment, but positive at the follow up were considered incident cases, therefore those w ho had prevalent disease or who did not have 2 carotid ultrasound assessments were not included in the assessment of incident atherosclerosis. Independent variable Alcohol consumption was self reported by as king about the average frequency (number of days per week) and quantity (number of drinks per drinking day) of use. The average number of drinks per week was calculated by multiplying the frequency by the quantity. Number of drinks per week were capped at 14 for women and 21 for men in accordance with the definition of hazardous alcohol use (Reid et al., 1999). For example, women that reported > 14 drinks per week were given a value of 14. Covariates All covariates were chosen based on our conceptual model (Figure 4 1 ) and were measured at the time of baseline and follow up assessments. Age was assessed in years, using reported date of birth. Self reported race was categorized as white, black, and Asian/Pacific Islander or Native American/ Alaskan. Self reported smoking was assessed in number of packs smoked using standardized categories: less than half a pack per day; at least half a pack but less than one pack per day; at least one but less than two packs per day; two or more packs per day Cumulative pack years were calculated to determine the average pack, multiplied by 0.5 to reflect the semi annual visits, and summed across the years prior to the carotid artery ultrasound assessments. Self reported illicit drug use included any of the f ollowing: crack or any form of cocaine; uppers (including crystal, methamphetamines, speed, ice); heroin or other opiates and was dichotomized. Plasma HIV RNA viral load was measured using standard laboratory techniques. HIV RNA viral load was categorized as < 200 copies/mL

PAGE 63

63 or hypertension) and diabetes (dichotomized as having been diagnosed with diabetes versus no history of diabetes). Data Analyses Group based trajectory models To describe patterns of alcohol consumption over time, we conducted group based trajectory models (GBTM) In the first modeling step, we assessed linear pat terns of 3 5 groups, as suggested by previous research (Marshall et al., 2015a; Marshall et al., 2015b; Cook et al., 2013 ). Goodness of fit was assessed at each step using the Akaike information criteria and Bayesian information criteria (smaller the value s, better the model), group posterior probabilities summed PP/number of groups) The PP estimate is the probability that any one group based trajectory adeq uately captures the individual patterns. Therefore, an individual pattern was assigned into the group pattern with the highest probability of group membership. Models with PP and/or model entropy values <0.7 were rejected (Andruff et al., 2009). The 95% co nfidence intervals (CI) of the resulting patterns were used to qualitatively assess the stability of the trajectories. Models with small CIs of trajectories were favored over wide CIs. Separately for each cohort, a group based trajectory modeling was condu cted to describe 10 year drinking patterns prior to the measurement of prevalent and incident atherosclerosis (Figure 4 2 ). Trajectories that described weekly drinking of 1 7 drinks for women and 1 14 drinks for men when applicable, and those that described consistent Age

PAGE 64

64 adjusted prevalence and incidence was calculated stratified by gender and combined, using standardized popula tion data (United States Census Bureau, 2000) Multivariable logistic regression models Crude and adjusted associations between alcohol consumption patterns and the prevalence and incidence of subclinical atherosclerosis were conducted using separate logis tic regression models, first stratified by gender and then with cohorts combined. Multivariable adjustment of the aforementioned covariates was conducted in all logistic regression models in a stepwise fashion, with demographic factors (i.e., age, race/eth nicity, gender when applicable) being added first, followed by substance use (i.e., pack years, illicit drug use), CVD related risk factors (BMI, hypertension, diabetes, hepatitis C), and HIV RNA viral load. Compared to the lowest alcohol consumption patte rn, other alcohol consumption patterns were considered statistically significantly associated with prevalent and incident subclinical atherosclero sis at the p value < .05 level. All statistical analyses were conducted using SAS 9.4 (SAS Institute, Inc., Ca ry, NC). Results Sample characteristics by cohort at baseline and follow up are presented in Table 4 1. There were 800 HIV+ women and 348 HIV+ men at baseline. Median age at baseline was 45.8 (IQR 40.4 50.5) years in women and 56.1 (IQR 50.2 61.4) years in men. Women were more likely to be of black race than men (58% vs. 25%). Men were more likely to report illicit drugs use (19% vs. 9%) and had higher cumulative pack years (3.9 vs. 2.6) than women. Body mass index was higher among women (28.1 vs. 25.6), wh ile hypertension (40% vs. 24%) and diabetes (34% vs. 26%) was more likely in men than women. Hepatitis C co infection was higher among women than men (27% vs. 13%). At baseline, men were more likely to have suppressed viral load ( 84% vs. 52%) compared to w omen.

PAGE 65

65 There were 512 HIV+ women and 324 HIV+ men who met stud y criteria at follow up (Table 4 1 ). Median age (years) at follow up was about 51 in women and 65 in men. Women were more likely to be of black race than men (63% vs. 43%). Men were more likely t o report illicit drugs use (16% vs. 7%). Cumulative pack years (women 2.6 [IQR 0.0 4.1]; men 2.1 [IQR 0.0 2.1]) and Hepatitis C co infection (women 15%; men 16) were similar by gender Body mass index was higher among women (30.7 vs. 26.6), while hypertens ion (45% vs. 30%) and diabetes (37% vs. 26%) was more likely in men than women. At follow up, men were more likely to have suppressed viral load (89% vs. 72%) compared to women. Prevalent Subclinical Atherosclerosis Prevalent subclinical atherosclerosis wa s identified in 16% (n=125; age adjusted prevalence 14.1/100) of women and 26% (n=90; 25.0/100) of men. The prevalence for subclinical atherosclerosis (Table 4 2) was the highest among men in the abstinent/low alcohol use pattern (31.1/100), followed by he avy use (28.4/100) and moderate use (24.3/100). Among women, prevalence was highest in the abstinent/low alcohol use pattern (15.6/100), followed by moderate use (15.0/100) and heavy use (12.8/100). When the WIHS and MACS cohorts were combined, the highest prevalence for subclinical atherosclerosis was in the heavy alcohol use group (16.3/100), followed by the abstinent/low use group (15.9/100), and the moderate use group (15.2/100). In analysis of alcohol use patterns at baseline, a four group trajectory m odel emerged as the be st fitting model for women (Figure 4 3 Panel A ; model entropy 0. 9 4 ) and men (Figure 4 3 Panel B ; model entropy 0. 97 ). Alcohol consumption patterns included women 35 %; men 16%, little to no consumption throughout 10 women 5 9%; men 76%, moderate consumption throughout 10 women 6%; men 8% heavy consumption throughout 10 years).

PAGE 66

66 Crude and adjusted odds ratios (AOR) of alcohol consumption patterns on prevalent subclinical atherosc lerosis are presented in Table 4 3 Long term heavy alcohol use was not statistically significantly associated with increased risk for prevalent subclinical atherosclerosis in women (AOR 0.86, CI 0.38 1.97, p=.73) or men (AOR 1.16, CI 0.38 3.54, p=.79) co mpared to abstinence While moderate alcohol use tended toward a protective effect on prevalent subclinical atherosclerosis in women (AOR 0.71, CI 0.45 1.10, p=.12), a similar association was not found in men (AOR 1.08, CI 0.51 2.28, p=.84), compared to ab stinence and after controlling for race, age, pack years, illicit drug use, BMI, hypertension, diabetes, hepatitis c co infection, and suppressed viral load. In analysis that combined both women and men, heavy alcohol consumption was not statistically sign ificantly associated with prevalent subclinical atherosclerosis (AOR 0.99, CI 0.52 1.88, p=.97), while moderate alcohol consumption tended toward a protective effect, with 22% lower odds for prevalent subclinical disease (CI 0.54 1.13, p=.20), compared to abstinent/low use. Incident Subclinical Atherosclerosis Incident subclinical atherosclerosis was detected in 12% (n=61; incidence 13.0/100) of women and 18% (n=58; incidence 16.6/100) of men. The incidence for subclinical atherosclerosis (Table 4 2) was the highest among men in the abstinent/low alcohol use pattern (36.6/100), followed by heavy use (24.0/100) and moderate use (13.4/100). Among women, incidence was highest in the abstinent/low alcohol use pattern (15.3/100), followed by heavy use (1 1.6/100) and moderate use (8.6/100). When the WIHS and MACS cohorts were combined, the highest incidence for subclinical atherosclerosis was in the abstinent/low alcohol use group (15.9/100), followed by the heavy use group (15.0/100), and the moderate use group (10.1/100). A three group trajectory model emerged as the best fitting model for women (Figure 4 3 Panel C; model entropy 0.98 ) and men (Figure 4 3 Panel D ; model entropy 0.98 ). Alcohol

PAGE 67

67 %; MACS 19%, very little to no consumption throughout 10 years), 45%; MACS 72%, moderate consumption throughout 10 %; MACS 9%, heavy consumption throughout 10 years) alcohol consumption. Crude and adjusted odds ratios of alcohol consumption patterns on incident subclinical atherosclerosis are presented in Table 4 4. Long term heavy alcohol use was not statistically significantly associated with increased risk for incident subclinical atherosclerosis in women (AOR 1.08, CI 0.39 3.01, p=.87) and men (AOR 1.40, CI 0.46 4.26, p=.55). While moderate alcohol use leaned toward a protective effect on incident subclinical atherosclerosis in men (AOR 0.53, CI 0.25 1.13, p=.10), a similar association was not found in women (AOR 1.05, CI 0.57 1.92, p=.88), after controlling for race, age, pack years, illicit drug use, BMI, hypertension, diabetes, hepatitis c co infection, and suppressed viral load. In analysis that combined both cohorts, heavy (AOR 1.28, CI 0.63 2.62, p=.49) and modera te (AOR 0.79, CI 0.50 1.27, p=.34) alcohol consumption were not statistically significantly associated with incident subclinical atherosclerosis, compared to abstinent/low use. Discussion We aimed to assess the association between 10 year patterns of alcoh ol use and the prevalence and incidence of atherosclerosis, measured by B mode carotid artery ultrasound. Specifically, w e aimed to test the general assumption that moderate alcohol consumption is protective to cardiovascular health among PLWH. Contrary to our hypothesis, long term heavy alcohol use was not statistically significantly associated with prevalent or incident atherosclerosis. These results are contrary to previous studies that hav e demonstrated an association between heavy alcohol use and subcl inical atherosclerosis in the general population ( Xie et al., 2012; Zyriax et al., 2010; Hougaku et al., 2005 ; Mukamal et al., 2003 b ) and extent

PAGE 68

68 literature finding positive associations between heavy alcohol use and clinical cardiovascular disease among PLWH ( Freiberg et al., 2010 ; Corral et al., 2009 ; Justice et al., 2008 ) However, this finding is consistent with recent studies finding no association between heavy drinking and cardiovascular disease among PLWH ( Kelly et al., 2016; Wandeler et al. 2016 ; Womack el al., 2014) and subclinical atherosclerosis in the general population ( Kim et al., 2014; Bauer et al., 2013; Zureik et al., 2004 ). It is possible that heavy alcohol use is a contributor to risk for clinical cardiovascular disease, but less predictive at the developmental stages of atherosclerosis. An additional explanation could be that abstinent/low drinkers had a history of heavy drinking that is not captured in the 10 year patterns. If past heavy drinking conferred increased risk more th an 10 years later, then the effect of current 10 year patterns of heavy drinking could be attenuated when compared to the current abstinent/low drinkers. We found that long term moderate alcohol use leaned toward lower odds for prevalent (in women) and inc ident (in men) atherosclerosis. While these estimates did not meet statistical significance, this finding is consistent with other research indicating a J curved relationship between alcohol consumption and subclinical atherosclerosis in the general popula tion (Kohsaka et al., 2011; Hougaku et al., 2005) and extent research on clinical disease in the general population ( Mukamal et al., 2003a, Reynolds et al., 2003; Corrao et al., 2000; McElduff and Dobson, 1997, Sacco et al., 1999 ) and among PLWH (Wandeler et al., 2016, Carrieri et al., 2012, Schminke et al., 2005). Limitations The readers should consider some limitations of the current study. First, alcohol consumption quantity and frequency were assessed via self report and is subject to recall and social desirability biases, resulting in underestimated reports of alcohol consumption. However, this method has been established as a reliable and valid approach to alcohol use assessment (Del

PAGE 69

69 Bo ca and Darkes, 2003). Second, because we used carotid artery ultras ound, these results can only be generalized to disease within the carotid artery and may not extend to of plaque or lesions outside of this area. Carotid artery ultrasound also does not capture all mechanisms that are involved in atherothrombotic events, b eing most strongly associated with blood pressure levels (Sander et al., 2000) and bearing little relationship with coagulation (Sosef et al., 1994) or platelet activity (De Luca G et al., 2010), for example. However, research has found presence of plaque or lesions within the carotid artery to be highly correlated to subclinical disease in other vascular territories when compared to other methods that detect low to no disease (Lester et al., 2009, Davis et al., 1999). While we did not aim to investigate cl inical CVD, subclinical atherosclerosis is a proximal indicator of later clinical manifestations of CVD, such as myocardial infarction and stroke. Third, there are significant demographic differences between the WIHS and MACS cohorts, making direct compari sons of stratified analyses difficult Because of these differences, we carefully controlled for variables related to cardiovascular risk. Fourth, GBTM is a semi parametric and probabilistic model that estimates grouped trajectories of the most similar ind ividual patterns. Therefore, each trajectory group does not fully describe the individual level patterns contained within them and should not be considered absolute. Further we restricted our analyses to participants with at least 4 alcohol consumption a ssessments in order to estimate trajectory models. Therefore, it is possible that different trajectories could have emerged had we not excluded these participants Conclusions In summary, the current study adds to the literature on the effect of longitudin al alcohol consumption on CVD among PLWH, by focusing on early subclinical manifestations of atherosclerosis. This study provides important new information regarding the specific effect of moderate alcohol use, and thus helps fill the gap in this area of a lcohol research. Future research

PAGE 70

70 should continue to investigate the effect of alcohol use on atherosclerosis and interactions with other significant factors, such as mental health issues, social support, and antiretroviral treatments. It is possible, that through this continued research, we may be able to identify subgroups of drinkers that are a greater risk for car diovascular disease, by which tailored interventions can target. Further, v alidation of the J curve association could focus on the proposed risk mechanisms between alcohol consumption and cardiovascular health. For example, if moderate consumption is protective, we would expect decr ease in pro inflammatory and cardiac biomarkers.

PAGE 71

71 Figure 4 1. Conceptual model for the association between a lcohol consumption and atherosclerosis and confounding factors

PAGE 72

72 Figure 4 2 Timeline for 10 year trajectory models prior to prevalent and incident subclinical atherosclerosis

PAGE 73

73 Table 4 1 Sample characteristics of women and men living with HIV by cohort Baseline Follow up WIHS (N=800) MACS (N=348) WIHS (N = 512) MACS (N=324) Race White Black Other (Asian, Native American, etc.) 139 (17) 468 (58) 193 (24) 229 (68) 87 (25) 32 (9) 101 (20) 322 (63) 89 (17) 139 (43) 140 (43) 45 (14) Age, mean (IQR) 45.8 (40.4 50.5) 56.1 (50.2 61.4) 50.7 (45.6 55.0) 65.1 (60.5 70.0) Smoking pack years, mean (IQR) 2.7 (0.0 4.5) 3.9 (0.0 4.9) 2.6 (0.0 4.1) 2.1 (0.0 2.1) Illicit drug use No Yes 729 (91) 71 (9) 282 (81) 66 (19) 478 (93) 34 (7) 273 (84) 51 (16) Body Mass Index, mean (IQR) 28.1 (22.9 31.3) 25.6 (22.9 27.4) 30.7 (26.4 34.5) 26.6 (23.3 29.0) Hypertension No Yes 608 (76) 192 (24) 208 (60) 140 (40) 360 (70) 152 (30) 179 (55) 145 (45) Diabetes No Yes 596 (74) 204 (26) 229 (66) 119 (34) 378 (74) 134 (26) 205 (63) 119 (37) Hepatitis C status Negative Positive 583 (73) 217 (27) 304 (87) 44 (13) 433 (85) 79 (15) 271 (84) 53 (16) HIV RNA Viral Load < 200 copies/mL 200 copies/mL 418 (52) 382 (48) 291 (84) 57 (16) 368 (72) 144 (28) 289 (89) 35 (11) Cross sectional alcohol consumption Abstinence Low (<1 drink/week) Moderate (1 7 drinks/week) Heavy (>7 drinks/week) 513 (64) 217 (27) 70 (9) 213 (61) 117 (34) 18 (5) 405 (79) 88 (17) 19 (4) 248 (76) 64 (20) 12 (4) 10 year alcohol consumption pattern Abstinent Low Moderate Heavy 279 (35) 472 (59) 49 (6) 55 (16) 265 (76) 28 (8) 236 (46) 226 (45) 47 (9) 63 (19) 233 (72) 28 (9) Prevalent lesions, baseline No Yes 675 (84) 125 (16) 258 (74) 90 (26) Incident lesions, follow up No Yes 451 (88) 61 (12) 266 (82) 58 (18)

PAGE 74

74 Table 4 2. Age adjusted p revalence and incidence of subclinical atherosclerosis by alcohol consumption patterns Alcohol use pattern N (Prevalence) N (Incidence) MACS Abstinent/low Moderate Heavy 52 (31.1/100) 253 (24.3/100) 28 (28.4/100) 63 (36.6/100) 233 (13.4/100) 28 (24.0/100) WIHS Abstinent/low Moderate Heavy 279 (15.6/100) 406 (15.0/100) 49 (12.8/100) 236 (15.3/100) 226 (8.6/100) 47 (11.6/100) Total Abstinent Moderate Heavy 334 (15.9/100) 737 (15.2/100) 77 (16.3/100) 299 (15.9/100) 459 (10.1/100) 75 (15.0/100) Study

PAGE 75

75 Figure 4 3. 10 year alcohol consumption trajectories by cohort. Panel A: Alcohol consumption patterns in women prior to baseline carotid artery ultrasound; Panel B: Alcohol consumption patterns in men prior to baseline carotid artery ultrasound; Panel C: Alcohol consumption patterns in women prior to follow up carotid artery ultrasound; Panel D: Alcohol consumption patterns in men prior to follow up carotid a rtery ultrasound

PAGE 76

76 Table 4 3 Association between 10 year alcohol consumption patterns and prevalent subclinical atherosclerosis, overall and by cohort Multicenter AIDS Cohort Study (MACS) Crude Odds Ratio (95% CI) P value Adjusted Odds Ratio a (95% CI) P value Crude Odds Ratio (95% CI) P value Adjusted Odds Ratio a (95% CI) P value Alcohol Consumption Patterns Abstinent REF REF REF REF Low Moderate 0.78 (0.52 1.17) .23 0.71 (0.45 1.10) .12 0.99 (0.51 1.93) .98 1.08 (0.51 2.28) .84 Heavy 1.39 (0.66 2.92) .38 0.86 (0.38 1.97) .73 1.39 (0.51 3.77) .52 1.16 (0.38 3.54) .79 a Controlled for race, age, illicit drug use, pack years of cigarette use, body mass index, diabetes, hypertension, hepatitis c co infection, and suppressed viral load

PAGE 77

77 Table 4 3 Continued Combined Crude Odds Ratio (95% CI) P value Adjusted Odds Ratio a (95% CI) P value Alcohol Consumption Patterns Abstinent REF REF Low Moderate 0.97 (0.69 1.35) .84 0.78 (0.54 1.13) .20 Heavy 1.54 (0.86 2.75) .14 0.99 (0.52 1.88) .97 a Controlled for gender, race, age, illicit drug use, pack years of cigarette use, body mass index, diabetes, hypertension, hepatitis c co infection, and suppressed viral load

PAGE 78

78 Table 4 4. Association between 10 year alcohol consumption patterns and incident subclinical atherosclerosis, overall and by cohort Multicenter AIDS Cohort Study (MACS) Crude Odds Ratio (95% CI) P value Adjusted Odds Ratio a (95% CI) P value Crude Odds Ratio (95% CI) P value Adjusted Odds Ratio a (95% CI) P value Alcohol Consumption Patterns Abstinent/Low REF REF REF REF Moderate 0.95 (0.54 1.68) .86 1.05 (0.57 1.92) .88 0.50 (0.26 0.98) .04 0.53 (0.25 1.13) .10 Heavy 1.30 (0.52 3.18) .56 1.08 (0.39 3.01) .87 1.17 (0.43 3.18) .75 1.40 (0.46 4.26) .55 a Controlled for race, age, body mass index, diabetes, hypertension, illicit drug use, pack years of cigarette use, and suppressed viral load

PAGE 79

79 Table 4 4 Continued Combined Crude Odds Ratio (95% CI) P value Adjusted Odds Ratio a (95% CI) P value Alcohol Consumption Patterns Abstinent/Low REF Moderate 0.86 (0.57 1.32) .50 0.79 (0.50 1.27) .34 Heavy 1.45 (0.76 2.77) .26 1.28 (0.63 2.62) .49 a Controlled for gender, race, age, body mass index, diabetes, hypertension, illicit drug use, pack years of cigarette use, and suppressed viral load

PAGE 80

80 CHAPTER 5 THE IMPACT OF 10 YEAR PAST AND CURRENT ALCOHOL CONSUMPTION PATTERNS ON EARLY PROGRESSION OF ATHEROSCLEROSIS AMONG PERSONS LIVING WITH HIV Introduction Persons living with HIV (PLWH) have higher odds for subclinical atherosclerosis, compared to uninfected c ontrols ( Hanna et al., 2015; Hsue et al., 2012; Grunfeld et al., 2009 ), after adjusting for standard metabolic and HIV related risk factors. This suggests that there are indicators outside of the traditional risk factor framework that are important contrib utors to cardiovascular health, such as alcohol consumption. A J curve has been observed in epidemiologic studies of alcohol use and cardiovascular disease (CVD) morbidity and mortality in the general population ( Mukamal et al., 2003a; Reynolds et al., 200 3; Corrao et al., 2000; McElduff and Dobson, 1997; Sacco et al., 1999 ) Previous research has focused predominantly on the association between alcohol consumption and clinical endpoints (myocardial infarction, stroke, heart attack) and less on intermediate measures of cardiovascular disease development or subclinical atherosclerosis, such as carotid intima media thickness (cIMT). Carotid intima media thickness measured by high resolution B mode ultrasound, is the thickness (mm) of the inner and middle laye rs of the carotid artery. Further t he relationship between alcohol consumption and subclinical atherosclerosis has not been sufficiently examined among PLWH. The studies that do examine the relationship between alcohol consumption and subclinical atherosclerosis among the general population yield inconsistent findings. Some studies have found a protective effect of moderate alcohol consumption on subclinical atherosclerosis, compared to abstinence (Kohsaka et al., 2011; Hougaku et al., 2005), while others find alcohol consumption at any level to be a risk factor (Zyriax et al., 2010), or not associated with subclinical atherosclerosis (Tofferi et al., 2004). Further, some studies have

PAGE 81

81 found significant associations between alcohol consumption and s ubclinical disease in men, but not in women (Kim et al., 2014 ; Zyriax et al., 2010; Lee et al., 2009 ; Schminke et al., 2005 ). Aside from the general lack of research assessing the association between alcohol consumption and subclinical atherosclerosis amon g PLWH, there are no studies investigating the effects of past longitudinal and current alcohol consumption patterns on progression of cIMT as an indicat or of CVD disease development. The objective of the current analysis was to assess the association betw een patterns of alcohol consumption and cIMT progression among PLWH. Specifically, we aimed to 1) assess the relationship of past (10 year) and current (6 month) alcohol consumption patterns to cIMT progression among PLWH, and 2) to explore whether the rel ationships appeared to differ by gender. We hypothesized that past long term alcohol exposure would be more significantly associated with cIMT progression versus current short term alcohol exposure In general however we expect ed that moderate alcohol consumption would be associated with a protective effect (lower cIMT level over time) and heavy alcohol consumption would be associated with a harmful effect (higher cIMT lever over time) on cIMT. Methods Study Setting, Selection, and Inclusion Criteria Th e Multicenter AIDS Cohort Study (MACS; Dudley et al., 1995 ; Detels et al., 1992; Kaslow et al., 1987 ; Barkan et al., 1998 ) are well established, national multicenter cohorts of men who have sex w ith men (MSM) and of women, respectively, living with or at risk for HIV infection. Participants from MACS were recruited from the following metropolitan areas: Baltimore, MD, Washington, DC, Chicago, IL, Pittsburgh, PA, Los Angeles, CA. Participants from WIHS were recruited from the following metropolitan areas: Brooklyn and Bronx, NY, Washington, DC, Chicago, IL, Los

PAGE 82

82 Angeles and San Francisco, CA. The MACS recruited MSM across three waves, in 1984 1985 (n=4954), 1987 1991 (n=668), and 2001 2003 (n=1350). Women were recruited in WIHS across two waves, in 1994 1995 (n=2625) and 2001 2002 (n=1141). The data were collected through structured interviews, and standardized physical, psychological, and laboratory assessments. HIV status was assessed by enzyme lin ked immunosorbent assay (ELISA) with Western blot for confirmation at baseline for HIV+ participants, and semi annually for HIV participants. HIV sero conversion was confirmed by testing HIV participants at each semi annual visit. Written informed consen t was obtained prior to each semi annual assessment for both cohorts. The questionnaires are available online for MACS at http://aidscohortstudy.org and for WIHS at https://sta tepi.jhsph.edu/wihs/wordpress/. The WIHS cardiovascular sub study recruited wome n aged 25 to 60 years, with no history of hea rt surgery or coronary angioplasty/stent placement before HIV infection ; The MACS cardiovascular sub study recruited men over 40 years of age, under 300lbs, and with no histo ry of heart surgery or coronary angio plasty/stent placement before HIV infection For the current analysis we excluded those who seroconverted during the study and those with less than 4 alcohol consumption assessments. The m edian person years of follow up between 1994 2014 were 16.7 years (interquartile range [IQR] : 1 6.0 18 .5 years) for WIHS participants and 12.4 years (IQR: 10.0 16.5 years) for MACS participants All MACS and WIHS participants provided written informed consent for overall study participation. The Institutional Revie w Board at the University of Florida approved this specific analysis Data Collection In addition to the standard data collection for the MACS and WIHS, participants in the cardiovascular sub studies underwent high resolution B mode carotid artery ultrasou nds of 6 locations in the right carotid artery (the near and fall walls of the common carotid artery [CCA],

PAGE 83

83 carotid bifurcation, and internal carotid artery [ICA] ; Hodis et al., 2001 ) using a standardized protocol across study sites (Kaplan et al., 2008) Quality control and reliability of the carotid artery ultrasound measurement was performed among a subset of WIHS and MACS participants and was found to have high intraclass correlations (ICC) in both WIHS (variation coefficient = 1.8%; ICC = 0.98) and MA CS (variation coefficient = 1.0%; ICC=0.99 ; Kaplan et al., 2008 ). Main outcome measure Carotid intima media thickness at the far right CCA (CCA IMT) was measured up to 4 times in WIHS and up to 3 times in MACS from 2004 2013 The CCA IMT was assessed using the B mode carotid artery ultrasounds by automated computerized edge detection of the images. The outcome of interest was change in CCA IMT from the baseline to the follow up assessments. Independent variable Past 10 year alcohol consumption was self repo rted by asking about the average frequency (number of days per week) and quantity (number of drinks per drinking day) of use. The average number of drinks per week was calculated by multiplying the frequency by the quantity. We capped the maximum number of drinks per week to 14 for women and 21 for men in accordance with the definition of hazardous alcohol use (Reid et al., 1999). For example, women that reported > 14 drinks per week were given a value of 14. Current alcohol consumption was calculated by mu ltiplying the frequency by the quantity to yield the average number of drinks per week at baseline and all follow up assessments. Current alcohol er week for men), and heavy (>7 drinks per week for women o r >14 drinks per week for men).

PAGE 84

84 Covariates Covariates of interest were chosen based on our conceptual model of confounding factors in the association between alcohol consumption and early atheroscl erosis development (Figure 5 1 reported date of birth. Race was self reported and categorized as white, black, and Asian/Pacific Islander or Native American/Alaskan. Self reported smoking was assessed i n number of packs smoked using standardized categories: less than half a pack per day; at least half a pack but less than one pack per day; at least one but less than two packs per day; two or more packs per day. Cumulative pack years were calculated to de termine the average pack, multiplied by 0.5, and summed across the years up to the baseline and follow up assessments. Self reported illicit drug use was dichotomous and measured by asking if participants used any of the following: crack or any form of coc aine; uppers (including crystal, methamphetamines, speed, ice); heroin or other opiates. Depressive symptoms were assessed at each semi annual visit with the Center for Epidemiology St udies Depression Scale (CES D, Radloff, 1977). Some research has found t hat utilizing the score of 16 or greater may inflate the rate of depression among PLWH, due to the overlapping somatic symptoms that may be present due to HIV infection (Kalichman et al., 2000). Therefore, a score of 23 or greater was considered probable d epression. Plasma HIV RNA viral load was measured using standard laboratory techniques. HIV RNA viral load was categorized as <200 related comorbidities included body mass index (BMI), hypertension (blood press they have hypertension) and diabetes (dichotomized as having been diagnosed with diabetes versus no history of diabetes).

PAGE 85

85 Data Analyses Group based trajectory models To describe patterns of alcohol consumption over time, we conducted group based trajectory models (GBTM) In the first modeling step, we assessed linear patterns of 3 5 groups, as suggested by previous r esearch (Marshall et al., 2015 a ; Marshall et al., 2015 b ; Cook et al., 2013 ). Goodness of fit was assessed at each step using the Akaike information criteria and Bayesian information criteria (smaller the values, better the model), group posterior probabilities summed PP/number of groups) The PP estimate is the probability that any one group based trajectory adequately captures the individual patterns. Therefore, an individual pattern was assigned into the group pattern with the highe st probability of group membership. Models with PP and/or model entropy values <0.7 were rejected (Andruff et al., 2009). The 95% confidence intervals (CI) of the resulting patterns were used to qualitatively assess the stability of the trajectories. Model s with small CIs of trajectories were favored over wide CIs. For both cohorts, a group based trajectory model was conducted to describe 10 year drinking patterns prior to the baseline carotid artery ultrasound assessment. Current alcohol consumption was me asured and included in the models at the time of baseline and follow up assessments (Figure 5 2). All time varying covariates were measured at the time of baseline and follow up assessments. Generalized estimating equations Crude and adjusted associations between past and current alcohol consumption patterns and the change outcome of CCA IMT were conducted using generalized estimating equations, stratified by gender. We developed the models in a stepwise fashion by 1) assessing the effect of time on CCA IMT 2) assessing time, past alcohol consumption, and current alcohol consumption, 3) adjusting for aforementioned covariates, and 4) including interactions between

PAGE 86

86 past alcohol consumption and time and current alcohol consumption and time. Past and current a lcohol consumption patterns were considered statistically significantly associated with change in CCA IMT at the p value <.05 level. All statistical analyses were conducted using SAS 9.4 (SAS Institute, Inc., Cary, NC). Results Sample characteristics by co hort are presented in Table 5 1. Me an age (years) at baseline was about 45 in women (n=1181) and 70 in men (n=395). Women were more likely to be of black race than men (60% vs. 30%). Men were more likely to report illicit drugs use (20% vs. 8%) and had hig her cumulative pack years (3.2 vs. 2.0) than women. Probable depression (24% vs. 13%) and body mass index was higher among women (28.3 vs. 25.5), while hypertension (52% vs. 20%) and diabetes (33% vs. 22%) was more likely in men than women. At baseline, me n were more likely to have suppressed viral load (65% vs. 51%) compared to women. A four group trajectory model emerged as the best fitting model for women (Figure 5 3, Panel A; model entropy 0.88) and men (Figure 5 3, Panel B; model entropy 0.92). Alcohol consumption patterns (WIHS 30%; MACS 59%, low to moderate [1 2 (3) drinks/week for women (men)] throughout 10 4 (6 8) drinks/week for women (men)] and throughout 10 years). Current alcohol consumption at baseline was comparable by cohort (Abstinence: WIHS 55%, MACS 59%; Moderate WIHS 37%, MACS 36%; He avy: WIHS 8%, MACS 5%). The mean CCA IMT (m) in WIHS at baseline was 731.2 (standard deviation [SD] 116.9), 734.8 (SD 115.1) at follow up 1, 733.2 (SD 115.8) at follow up 2, and 750.3 (SD 124.1) at follow up 3. The mean CCA IMT (m) in MACS at baseline wa s 742.4 (SD 124.7), 756.4 (SD 124.9) at follow up 1, and 791.7 (SD 126.4) at follow up 2.

PAGE 87

87 Crude Associations between Alcohol Consumption and CCA IMT Mean CCA IMT by past and current alcohol consumption patterns and crude estimates of alcohol consumption pa tterns on CCA IMT are presented in Table 5 2. Among women, baseline CCA IMT was highest among those with past heavy ( mean 7 5 2.6 95% CI 732.4 772.9 ) and current heavy ( mean 749.4, 95% CI 739.3 759.5 ) alcohol consumption. C ompared to abstinence. p ast low ( 5.7 95% CI 17.6 6.2 p=. 35 ) 9.4, 95% CI 20.3 1.5, p= .09 ) and heavy 95% CI 12.6 31.1, p= .41 ) consumption was not statistically significantly associated with CCA IMT. While there was no statistical significant crude association between current moderate consumption 95% CI 3.4 6.0 p= .56 ) and CCA IMT, heavy consumption was associated with increase CCA 95% CI 2.1 19.8 p= .01 ). The baseline CCA IMT was highest in men with past moderate ( mean 7 79.8, 95% CI 744.2 815.6 ) and current moderate ( mean 759.7, 95% CI 746.6 772.7 ) alcohol consumption. Compared to abstinence, there were no statistically significant crude associations between past (Low: 11.6 95% CI 41.3 95% CI 25.5 62.6, p=.41; Heavy: 14.7 95% CI 38.7 95% CI 8.0 8.7, p=.94; Heavy: 8.9 95% CI 26.5 8.5, p=.32) alcohol co nsumption patterns and CCA IMT. Adjusted Associations between Alcohol Consumption and CCA IMT Adjusted associations between past and current alcohol consumption patterns are shown in Table 5 3. Among women, most past alcohol consumption patterns were not statistically significantly associated with CCA 6.8, 95% CI 20.4 6.8, p=. 33 20.2, 95% CI 46.1 5.7, p=.1 3 ). Past moderate consumption was associated with decre ased CCA IMT level 13.2, 95% CI 26.4 0.3, p=.0 5 ). While current moderate 6.4

PAGE 88

88 11.7, p= .57 ), current heavy consumption was associated with increased CCA IMT level 21.8, 95% CI 6.4 37.2, p= .01 ). Among men, past alcohol consumption patterns were not statistically significantly associated with CCA IMT ( Low: 2.7, 95% CI 29.9 35.3 p=. 87; Moderate: 24.9, 95% CI 28.6 78.3 p=. 36; Heavy: 7.6, 95% CI 72.5 57.3 p=. 82 ). While c urrent moderate consumption was not statistically significantly associated with CCA IMT, c ompared to 95% CI 23.5 27.5, p=.88), current heavy consumption was associated with a clinically significant increase in CCA IMT lev el ( 33.5, 95% CI 3.4 70.4 p=. 07). Interaction terms between past alcohol consumption and time and current alcohol consumption and time were statistically significant in men (Figure 5 4 ), but not in women (Figure 5 5 ). Men in the 10 year abstinent alcoh ol consumption group tended to decrease in CCA IMT level over time and had the greatest decrease in CCA IMT by the end of follow up 15.9, 95% CI 26.2, 5.7, p<.01). All other 10 year consumption groups increased in CCA IMT level over time (Time* 95% CI 1.0, 30.8, p=.07; Time*Hea ). Discussion We aimed to assess the association between past (10 year) and current (6 month) patterns of alcohol consumption a nd CCA IMT, measured by B mode carotid artery ultrasound. Specifically, we aimed to test the effect of heavy and moderate alcohol consumption on early atherosclerosis development among PLWH. While those in the 10 year heavy alcohol use group tended to have the highest CCA IMT, membership in this group was not statistically significantly associated increased CCA IMT in men and women, compared to the 10 year abstinent group.

PAGE 89

89 Men and women seemed to have contradictory results regarding the effect of 10 year mo derate alcohol use, with moderate drinking women having lower CCA IMT, whereas low and moderate drinking men had higher CCA IMT. Other studies have found inconsistent findings of alcohol effects on subclinical atherosclerosis between women and men ( Zyriax et al., 2010; Lee et al., 2009 ; Schminke et al., 2005 ). These differences may be a result of the gender differences in risk factors and clinical presentation of CVD. For example, CVD in women is more likely to present as microvascular coronary disease, rat her than plaque development and narrowing of the large coronary arteries (Vaccarino and Bremner, 2016). Therefore, it is possible that moderate and heavy drinking are associated with early progression of CVD, but in the small arteries and ve ssels of the co ronary arteries. Current heavy alcohol use was associated with statistically significant increase in CCA IMT in women, and clinically relevant increases in men. This finding is consistent with research in HIV uninfected populations that found heavy alcohol use to significantly increase CCA IMT (Zyriax et al., 2010) and carotid artery stiffness (Hougaku et al., 2005) Longitudinal effects of past and current alcohol consumption were found in men, but not women. Men in the past abstinent alcohol consumption g roup tended to decrease in CCA IMT over time and had the greatest decrease in CCA IMT by the end of follow up, while all other 10 year consumption groups increased in CCA IMT over time. Therefore, we did not find any level of long term alcohol use to be be neficial to cardiovascular health in men, but rather the opposite. Interestingly, current heavy alcohol consumption was associated with a significant decrease in CCA IMT over time This is likely due to the fact that current heavy users had the highest baseline CCA IMT, leaving greater opportunity to decrease over time than any other group. After adding the time and alcohol use interactions, a protective association of past moderate

PAGE 90

90 alcohol consumption on CCA IMT baseline level remained in women and curr ent heavy use was associated with higher baseline levels of CCA IMT in women and men. These findings are consistent with a J curved association found in the literature (Xie et al., 2012; Kohsaka et al., 2011; Hougaku et al., 2005). Limitations The readers should consider some limitations of the current study. First, alcohol consumption quantity and frequency were assessed via self report and is subject to recall and social desirability biases. These potential biases likely result in underestimation of alcoh ol consumption. However, this method has been established as a reliable and valid approach to alcohol use asses sment (Del Boca and Darkes, 2003). Second, because we used carotid artery ultrasound to measure non plaque CCA IMT, these results can only be gen eralized to the early developmental stages of atherosclerosis in the carotid artery. Research has found CCA IMT to be highly correlated to subclinical disease in other vascular territories when compared to other methods that detect low to no disease (Davis et al., 1999 ; Lester et al., 2009 ). Third, there are significant demographic differences between the WIHS and MACS cohorts, making direct comparisons of stratified analyses difficult. Because of these differences, we carefully controlled for confounding v ariables related to socio demographic status and cardiovascular risk. Fourth, GBTM is a semi parametric and probabilistic model that estimates grouped trajectories of the most similar individual patterns. Therefore, each trajectory group does not fully des cribe the individual level patterns contained within them and should not be considered absolute. Conclusions In summary, the current study adds to the literature on the effect of longitudinal and current alcohol consumption on the early development of subc linical atherosclerosis among PLWH, by focusing on changes in non plaque CCA IMT. This study provides important

PAGE 91

91 information regarding the specific effect of long term and short term moderate and heavy alcohol use among PLWH, and thus helps fill the gap in this area of alcohol research. It is possible that alcohol consumption has harmful effects in some person, but not others, Therefore, f uture research should continue to investigate the effect of alcohol use on the early development of atherosclerosis and i nteractions with other significant factors, such as mental health issues, social support, and antiretroviral treatments. Further research could also focus on the risk mechanism to validate the protective affect of moderate consumption by assessing the effe ct of alcohol consumption at different levels on pro and anti inflammatory and cardiac biomarkers.

PAGE 92

92 Figure 5 1. Confounding f actors associated with the association between alcohol consumption and early atherosclerosis development.

PAGE 93

93 Figure 5 2 Timeline for 10 year trajectory models prior to baseline carotid artery ultrasound

PAGE 94

94 Table 5 1. Baseline characteristics of persons living with HIV by cohort WIHS MACS Baseline Characteristics N (Column %) Mean (Standard Deviation) CCA IMT m Baseline (WIHS N=1181; MACS N=395) Follow up 1 (WIHS N=671; MACS N=329) Follow up 2 (WIHS N=499; MACS N=237) Follow up 3 (WIHS N=542) 731.2 (116.9) 734.8 (115.1) 733.2 (115.8) 750.3 (124.1) 742.4 (124.7) 756.4 (124.9) 791.7 (126.4) N/A Race White African American/Black Other 256 (22) 713 (60) 212 (18) 252 (64) 120 (30) 23 (6) Age (continuous), mean (SD) 45.0 (7.6) 69.7 (5.6) Probable depression No Yes 902 (76) 279 (24) 345 (87) 50 (13) Pack years (continuous), mean (SD) 2.0 (2.9) 3.2 (6.7) Illicit drug use No Yes 086 (92) 95 (8) 316 (80) 79 (20) Ever diagnosed with diabetes No Yes 921 (78) 260 (22) 263 (67) 132 (33) Hypertension No Yes 947 (80) 234 (20) 190 (48) 205 (52) Body mass index, mean (SD) 28.3 (7.3) 25.5 (3.7) HIV RNA Viral Load < 200 copies/Ml 200 copies/Ml 599 (51) 582 (49) 258 (65) 137 (35) 10 year alcohol consumption pattern Abstinence Low Moderate Heavy 420 (36) 351 (30) 334 (28) 76 (6) 59 (15) 235 (59) 71 (18) 30 (8) Current alcohol consumption Abstinence Moderate Heavy 651 (55) 441 (37) 89 (8) 234 (59) 142 (36) 19 (5)

PAGE 95

95 Figure 5 3. 10 year alcohol consumption trajectories by cohort Panel A: Alcohol consumption patterns in women prior to baseline carotid artery ultrasound measurement; Panel B: Alcohol consumption patterns in men prior to baseline carotid artery ultrasoun d measurement

PAGE 96

96 Table 5 2. Crude a ssociation s between 10 year alcohol consumption patterns, current alcohol consumption, and carotid artery intima media thickness, by cohort Multicenter AIDS Cohort Study CCA IMT, m Mean (SD) CCA IMT, m Mean (SD) 10 year Alcohol Consumption Patterns Abstinence 7 43.4 ( 733.9 752.9 ) REF 761.3 ( 735.6 787.1 ) REF Low 737.7 (727.4 747.9 ) 5.7 ( 17.6 6.2 ) 749.7 ( 735.0 764.5 ) 11.6 ( 41.3 18.1 ) Moderate 734.0 (724.2 743.8 ) 9.4 ( 20.3 1.5 ) 779.8 ( 744.2 815.6 ) 18.5 ( 25.5 62.6 ) Heavy 75 2.6 (732.4 772.9 ) 9.2 ( 12.6 31.1 ) 776.0 ( 729.2 822.9 ) 14.7 ( 38.7 68.2 ) Current Alcohol Consumption Abstinence 73 8.4 ( 731.4 745.4 ) REF 759.3 ( 746.6 772.1 ) REF Moderate 73 9.7 (732.5 747.0 ) 1.3 ( 3.4 6.0 ) 759.7 (746.6 772.7 ) 0.4 ( 8.0 8.7 ) Heavy 749.4 (739.3 759.5 ) 11.0 ( 2.1 19.8 )** 750.4 (731.2 769.5 ) 8.9 ( 26.5 8.5 ) All analyses controlled for time, age, race, pack years of cigarette use, illicit drug use, probable depression, hypertension, diabetes, and body mass index, suppressed HIV RNA viral load. Gender was controlled for in the combined analysis.

PAGE 97

97 Figure 5 4. Interaction of alcohol consumption patterns and time on change in CCA IMT m among men.

PAGE 98

98 Figure 5 5. Interaction of alcohol consumption patterns and time on change in CCA IMT m among women

PAGE 99

99 Table 5 3. Association between 10 year alcohol consumption patterns, current alcohol consumption, and carotid artery intima media thickness, controlling for the time and alcohol use interactions, by cohort Multicenter AIDS Cohort Study 10 year Alcohol Consumption Patterns Abstinence REF REF Low 6.8 ( 20.4 6.8 ) 2.7 ( 29.9 35.3) Moderate 13.2 ( 26.4 0.3 )* 24.9 ( 28.6 78.3) Heavy 20.2 ( 46.1 5.7 ) 7.6 ( 72.5 57.3) Current Alcohol Consumption Abstinence REF REF Moderate 2.6 ( 6.4 11.7) 2.0 ( 23.5 27.5) Heavy 21.8 (6.4 37.2 )** 33.5 ( 3.4 All analyses controlled for time, age, race, pack years of cigarette use, illicit drug use, probable depression, hypertension, diabetes, and body mass index, suppressed HIV RNA viral load.

PAGE 100

100 CHAPTER 6 CONCLUSIONS Accomplishments of the Dissertation This dissertation sought to advance scien tific knowledge of the associated factors of long term heavy and moderate alcohol consumption and to fill the gap in knowledge of the effect of long term moderate and heavy alcohol consumption on the development of atherosclerosis Specifically, w e aimed t o challenge or confirm the general assumption that moderate alcohol consumption is protective to cardiovascular health, and provide d evidence that addresses this relationship among PLWH. W e characterized patterns of alcohol consumption among PLWH from 2004 2013 by gender and assess ed the association between time stable and varying clinical factors of long term heavy and moderate alcohol consumption. We also described the association between 10 year patterns of alcohol use and the prevalence and incidence o f subclinical atherosclerosis, measured by B mode carotid artery ultrasound imaging. Lastly we assessed the longitudinal association between past 10 year and current patterns of alcohol use and carotid IMT progression. Taken together, this research addres sed an important gap in the literature regarding the possible J curve association between alcohol consumption and cardiovascular health among PLWH, that has been found in the general population. The results of this study have implications for clinical prac tice and identifica tion of PLWH with high CVD risk outside of the traditional risk framework. This research also underlines the need for consistent recommendations of alcohol consumption related to CVD risk and has the potential to highlight the importance of tailored interventions that can better address alcohol use issues that are specific to PLWH. We expected the first objective to provide new knowledge describing alcohol consumption trajectories over time and longitudinal associations between clinical f actors and

PAGE 101

101 moderate and heavy alcohol consumption among PLWH While several alcohol patterns characterized a stable level of consumption (i.e., low/abstinence, moderate, heavy), some patterns also featured shifts in drinking overtime. This indicates that alcohol consumption should be measured longitudinally to accurately depict exposure. Results also suggested that there are significant longitudinal clinical associations of moderate and heavy consumption that may help distinguish individuals for prevention and/or early intervention. The most significant associated factors of moderate and heavy alcohol consumption, across both men and women, was longitudinal illicit drug use. The Framingham risk score was associated with increased odds for moderate and heavy alcohol consumption among women. Also among women, sub optimal ART adherence was associated with increased odds for moderate alcohol consumption. Furthermore, having a HIV RNA viral load of 200 or greater was associated with increased odds for moderate an d heavy consumption in women. In our second objective, we expected to identify longitudinal drinking patterns that were significantly associated with prevalent and incident subclinical atherosclerosis. A particular advantage to this study was the ability t o identify those who did not have subclinical atherosclerosis at baseline that later developed this outcome, years later. With this information, we were able to identify 10 year alcohol consumption patterns prior to disease development. Contrary to our hyp othesis, we found that heavy alcohol consumption was not statistically significantly associated with increased risk for prevalent or incident subclinical atherosclerosis in women or men compared to abstinence. Moderate consumption was associated with 29% l ower odds of prevalent disease in women and 47% lower odds of incident disease in men. While moderate alcohol consumption was not statistically significantly associated with subclinical

PAGE 102

102 atherosclerosis, it is considered clinically significant and is consis tent with a protective effect found in other studies. In our third objective, we highlight the relationship between past (10 year) and current (6 month) alcohol consumption patterns and progression in non plaque CCA IMT among PLWH. With up to 4 repeated me asurements of CCA IMT, we were able to test the association between past and current alcohol use patterns and baseline CCA IMT, as well as assess the longitudinal effect of these drinking patterns on change in CCA IMT over time. In this analysis, we found that while past moderate alcohol consumption was moderately associated with a protective effect on CCA IMT in women, no level of alcohol use was found to be protective in men. Interactions between past and current alcohol consumption and time were statisti cally significant in men, but not in women. Compared to abstinence, men in a ll other 10 year consumption groups increased in CCA IMT over time. After adding the interactions, a protective tendency of 10 year moderate use on baseline CCA IMT level remained in women and current heavy use was associated with higher CCA IMT in women and men. Public Health Recommendations and Future Directions The U.S. Preventive Services Task Force recommends that clinicians assess all adults aged 18 years and older for alcohol misuse, and to provide support to reduce risky alcohol consumption (US Preventive Services Task Force, 2013). Further, several screening and brief intervention tools have been developed specifically for clinical use in the general and specific clinical po pulations (Saitz et al., 2016). In line with these recommendations, clinicians should consider screening all patients for alcohol consumption, particularly if patients report current and past illicit drug use, suboptimal ART adherence, and if patients have detectable viral load. Clinicians could also consider assessing moderate alcohol consumption, as this study found

PAGE 103

103 detrimental associations of moderate use on adherence and viral load, particularly among women. Future research should continue to investigate the effect of alcohol use on atherosclerosis and interactions with other significant factors, such as mental health issues, social support, and antiretroviral treatments. We particularly need more research conducted that confirms the protective effect of moderate alcohol use on CVD and all cause mortality among PLWH, as it is not clear whether PLWH should consume alcohol in moderation to reduce cardiovascular risk. se who quit drinking due to declining health. Often times, this group features high risk for poor outcomes due to past heavy drinking, while current ly being in the non use group, skewing results toward the null. Therefore, future research should focus on c haracterizing those in the sick quitter group and making efforts to separate this sub population from th e abstinent or low alcohol use group. Advantages and Challenges of the Current Research Utilizing the MACS and WIHS datasets to complete the objectives of the current dissertation come with many advantages and some challenges. The MACS and WIHS cohorts are incredibly rich sources of clinical, psychological, and behavioral information from those living with HIV and those at high risk for future HIV acquisi tion. The amount of information available for this dissertation really allowed us to explore several areas of potential cofounding factors in the relationship between alcohol consumption and subclinical atherosclerosis. Further, because these cohorts are l ong standing and collect data on a semi annual basis, we had the unique opportunity to assess trends in alcohol use over time, and use these patterns to more accurately understand the risk or benefits of specific long term alcohol use patterns on cardiovas cular health. An additional advantage of these two cohorts is that many of the data

PAGE 104

104 collection methods used are the same or similar in terms of us ing standardized questionnaires when available (i.e., using the CE S D to assess depressive mood). Some challen ges did arise, however, when using both cohorts for this dissertation research. Of note, the cohorts are of two different populations (women and men who have sex with men), with disparate recruitment methods. While neither cohort can be considered national ly representative of PLWH in general, each wave of enrollment for both cohorts had specific goals in mind that did not coincide. For example, the WIHS has been specific in recruiting women who are racially/ethnically diverse, of lower socioeconomic status, and of whom generally engage in high risk behaviors (i.e., substance use). Conversely, initial MACS participant recruitment was based centrally on those with HIV or at high risk during the early history of the HIV epidemic mainly affecting white MSM of h igher socioeconomic status. Later, MACS purposively over sampled racially/ethnically diverse men for this reason. These demographic differences by cohort make gender comparisons difficult, if not sometimes impossible, as we cannot rule out the possibility that gender effects are not merely the effect of population or recruitment strategy differences. An additional challenge in using both cohorts is the fact that not all data collection methods are the same, with many questions being asked differently or lab oratory tests using different cut offs. Similarly, because these are two different coh orts and datasets, the variable names within these datasets that are describing the same indicator are also not centralize which resulted in significant additional time for data cleaning.

PAGE 105

105 APPENDIX A ADDITIONAL CHAPTER 3 TABLES Table A 1. Factors associated with missing data in Aim 1 WIHS MACS Characteristics Percent Missing P Percent Missing P <10 742 (66) 10+ 381 (34) <10 377 (63) 10+ 220 (37) Race White Black Other 151 (61) 453 (67) 138 (69) 97 (39) 223 (33) 61 (31) .12 200 (64) 139 (62) 38 (62) 111 (36) 86 (38) 23 (38) .83 Income < $10,000 $10,000 $30,000 375 (68) 239 (70) 125 (66) 176 (32) 100 (29) 64 (34) .56 105 (70) 86 (68) 146 (69) 45 (30) 40 (32) 65 (31) .95 Probable Depression No Yes 532 (65) 160 (70) 313 (35) 68 (30) .15 324 (63) 53 (64) 190 (37) 30 (36) .88 Illicit Drug Use No Yes 693 (70) 44 (54) 297 (30) 37 (46) <.01 284 (71) 68 (61) 116 (29) 44 (39) .04 BMI Status Underweight Normal Overweight 335 (69) 125 (56) 282 (69) 152 (31) 100 (44) 129 (31) .001 51 (38) 165 (72) 161 (68) 82 (62) 63 (28) 75 (32) <.001 Diabetes No Yes 527 (64) 215 (72) 299 (36) 82 (28) <.01 232 (60) 145 (69) 156 (40) 64 (31) .02 CD4+ T cell count 3 300 500 cells/mm 3 < 300 cells/mm 3 319 (72) 219 (68) 204 (57) 124 (28) 104 (32) 153 (43) <.001 189 (72) 103 (66) 85 (48) 75 (28) 54 (34) 91 (52) <.001 Suppressed Viral load No Yes 324 (65) 418 (66) 171 (34) 210 (33) .70 113 (63) 264 (63) 67 (37) 153 (37) .90 Age, years 44.9 (7.4) 45.3 (7.9) .48 57.4 (7.5) 56.2 (8.0) .07 FRS 8.4 (6.1) 8.4 (5.9) .99 11.3 (3.0) 10.7 (3.7) .03 Cumulative ART use, years 13.6 (4.4) 9.3 (4.9) <.001 10.5 (3.8) 7.7 (3.8) <.001 Alcohol Use Abstinent Moderate Heavy 569 (65) 121 (71) 52 (70) 310 (35) 49 (29) 22 (30) .20 237 (61) 120 (68) 20 (62) 152 (39) 56 (32) 12 (37) .25

PAGE 106

106 APPENDIX B ADDITIONAL CHAPTER 4 TABLES Table B 1. Full model estimates of covariates and prevalent subclinical atherosclerosis, by cohort Multicenter AIDS Cohort Study (MACS) Crude Odds Ratio (95% CI) P value Adjusted Odds Ratio a (95% CI) P value Crude Odds Ratio (95% CI) P value Adjusted Odds Ratio a (95% CI) P value Race (Ref= White) African American/Black Other 0.71 (0.44 1.13) 0.29 (0.15 0.57) .15 <.001 0.73 (0.42 1.23) 0.34 (0.17 0.68) .23 .002 0.38 (0.20 0.73) 0.40 (0.15 1.09) .004 .07 0.40 (0.19 0.82) 0.57 (0.20 1.62) .01 .28 Age 1.09 (1.06 1.12) <.001 1.09 (1.06 1.12) <.001 0.99 (0.96 1.03) .72 1.01 (0.98 1.05) .28 Pack years 1.10 (1.04 1.16) <.001 1.08 (1.02 1.15) .009 1.07 (1.04 1.11) <.0 01 1.07 (1.03 1.11) <.001 Illicit drug use (Ref = No) 2.52 (1.46 4.38) <.001 1.83 (0.99 3.38) .05 0.73 (0.38 1.39) .34 0.60 (0.29 1.25) .17 BMI status 0.95 (0.92 0.98) <.001 0.97 (0.94 1.00) .06 0.94 (0.88 1.00) .06 0.93 (0.87 1.00) .05 Diabetes (Ref = No) 1.47 (0.97 2.22) .07 1.45 (0.91 2.32) .12 1.24 (0.75 2.03) .40 1.37 (0.78 2.40) .27 Hypertension (Ref = No) 1.41 (0.92 2.15) .11 0.92 (0.57 1.51) .75 1.52 (0.94 2.46) .09 1.60 (0.95 2.72) .08 Hepatitis C Co Infection (Ref = No) 2.03 (1.36 3.02) <.001 1.37 (0.88 2.13) .16 1.99 (1.03 3.86) .04 1.69 (0.81 3.52) .16 (Ref = < 200 copies/mL) 1.32 (0.90 1.94) .15 1.29 (0.85 1.97) .23 0.87 (0.46 1.65) .68 0.93 (0.46 1.89) .85 Alcohol Consumption Patterns (Ref = Abstinence) Low Moderate Heavy 0.78 (0.52 1.17) 1.39 (0.66 2.92) .23 .38 0.71 (0.45 1.10) 0.86 (0.38 1.97) .12 .73 0.99 (0.51 1.93) 1.39 (0.51 3.77) .98 .52 1.08 (0.51 2.28) 1.16 (0.38 3.54) .84 .79

PAGE 107

107 Table B 2. Full model estimates of covariates and prevalent subclinical atherosclerosis, overall Crude Odds Ratio (95% CI) P value Adjusted Odds Ratio a (95% CI) P value Gender (Ref = Female) 1.88 (1.39 2.56) <.001 0.74 (0.47 1.15) .18 Race (Ref= White) African American/Black Other 0.71 (0.44 1.13) 0.29 (0.15 0.57) .15 <.001 0.54 (0.37 0.80) 0.31 (0.18 0.55) .002 <.001 Age 1.09 (1.06 1.12) <.001 1.06 (1.03 1.08) <.001 Pack years 1.10 (1.04 1.16) <.001 1.07 (1.04 1.11) <.001 Illicit drug use (Ref = No) 2.52 (1.46 4.38) <.001 1.13 (0.71 1.79) .60 BMI status 0.95 (0.92 0.98) <.001 0.96 (0.93 0.99) .005 Diabetes (Ref = No) 1.47 (0.97 2.22) .07 1.40 (0.98 1.99) .06 Hypertension (Ref = No) 1.41 (0.92 2.15) .11 1.27 (0.89 1.80) .18 Hepatitis C Co Infection (Ref = No) 2.03 (1.36 3.02) <.001 1.46 (1.01 2.11) .04 1.32 (0.90 1.94) .15 1.30 (0.92 1.85) .14 Alcohol Consumption Patterns (Ref = Abstinence) Low Moderate Heavy 0.97 (0.69 1.35) 1.54 (0.86 2.75) .84 .14 0.78 (0.54 1.13) 0.99 (0.52 1.88) .20 .97

PAGE 108

108 Table B 3 Full model estimates of covariates and incident subclinical atherosclerosis, by cohort Multicenter AIDS Cohort Study (MACS) Crude Odds Ratio (95% CI) P value Adjusted Odds Ratio a (95% CI) P value Crude Odds Ratio (95% CI) P value Adjusted Odds Ratio a (95% CI) P value Race (Ref= White) African American/Black Other 0.83 (0.43 1.61 ) 0.70 (0.29 1.70 ) .58 .43 0.59 (0.28 1.22 ) 0.36 (0.17 1.21 ) .15 11 0.82 (0.45 1.50 ) 0.61 (0.23 1.58 ) .52 31 0.78 (0.37 1.65 ) 0.91 (0.32 2.58 ) .52 86 Age 1.09 (1.05 1.13 ) <.001 1.07 (1.03 1.12) <.001 1.03 (0.99 1.07 ) .15 1.05 (1.00 1.10 ) 04 Pack years 1.08 (1.02 1.15 ) <.0 1 1.05 (0.98 1.13 ) 14 1.10 (1.04 1 .16 ) <.0 01 1.12 (1.06 1.20 ) <.001 Illicit drug use (Ref = No) 1.65 (0.65 4.16 ) .29 1.12 (0.38 3.31 ) 84 1.14 (0.54 2.44 ) .73 1.51 (0.64 3.58 ) 35 BMI status 0.92 (0.87 0.96 ) <.001 0.91 (0.86 0.96 ) .001 1.01 (0.95 1.07 ) .77 0.99 (0.92 1.06 ) 76 Diabetes (Ref = No) 1.71 (0.97 3.00 ) .0 6 1.30 (0.69 2.43 ) 41 1.27 (0.71 2.27 ) .4 2 1.54 (0.80 2.99 ) 20 Hypertension (Ref = No) 1.77 (1.02 3.07 ) 04 1.53 (0.79 2.97 ) 21 1.98 ( 1.11 3.52 ) .0 2 1.65 (0.87 3.12 ) 13 Hepatitis C Co Infection (Ref = No) 2.42 (1.30 4.51 ) <.0 1 1.44 (0.70 2.94 ) 32 1.25 (0.60 2. 6 0 ) .55 0.79 (0.34 1.86 ) 59 (Ref = < 200 copies/mL) 1.40 (0.79 2.47 ) 24 1.82 (0.95 3.47 ) 07 0.40 (0.12 1.35 ) .14 0.36 (0.10 1.29 ) 12 Alcohol Consumption Patterns (Ref = Abstinence) Low Moderate Heavy 0.95 (0.54 1.68 ) 1.30 (0.52 3.18 ) .86 56 1.05 (0.57 1.92 ) 1.08 (0.39 3.01 ) .88 .87 0.50 (0.26 0.98 ) 1.17 (0. 43 3.18 ) .04 75 0.53 (0.25 1.13 ) 1.40 (0.46 4.26 ) .10 55

PAGE 109

109 Table B 4 Full model estimates of covariates and incident subclinical atherosclerosis, overall Crude Odds Ratio (95% CI) P value Adjusted Odds Ratio a (95% CI) P value Gender (Ref = Female) 1.61 (1.09 2.38) .02 0.54 (0.29 1.01 ) .05 Race (Ref= White) African American/Black Other 0.73 (0.48 1.12) 0.59 (0.32 1.12) .15 .11 0.63 (0.38 1.04 ) 0.61 (0.30 1.21 ) .07 .15 Age 1.05 (1.03 1.07) <.001 1.06 (1.03 1.09 ) <.001 Pack years 1.09 (1.05 1.13) <.001 1.10 (1.05 1.14 ) <.001 Illicit drug use (Ref = No) 1.46 (0.81 2.61) .20 1.11 (0.58 2.12 ) .75 BMI status 0.94 (0.91 0.97) <.001 0.94 (0.90 0.98 ) <.01 Diabetes (Ref = No) 1.55 (1.04 2.32) .03 1.40 (0.90 2.18 ) .13 Hypertension (Ref = No) 1.98 (1.34 2.93) <.001 1.66 (1.06 2.60 ) .03 Hepatitis C Co Infection (Ref = No) 1.81 (1.13 2.91) .01 1.07 (0.63 1.83 ) .80 0.92 (0.57 1.48) .72 1.16 (0.67 2.00 ) 59 Alcohol Consumption Patterns (Ref = Abstinence) Low Moderate Heavy 0.86 (0.57 1.32) 1.45 (0.76 2.77) .50 .26 0.79 (0.50 1.27 ) 1.28 (0.63 2.61 ) .34 .49

PAGE 110

110 Table B 5. Sample Characteristics between those with and without follow up in Aim 2 WIHS (n=800 minus 125 [those with Prev disease]; n=675) MACS (n=348 minus 90 [those with Prev disease]; n=258) No FU (N=387) FU (N=288) P value No FU (N = 63) FU (N=195) P value Race White Black Other ( Asian, Native American, etc ) 74 (19) 217 (56) 96 (25) 34 (12) 172 (60) 82 (28) .03 55 (87) 6 (10) 2 (3) 102 (52) 68 (35) 25 (13) <.00 1 Age, mean (95% CI) 45.3 (44.6 46.1) 44.5 (43.7 45.4) .63 51.3 (50.1 52.4) 57.8 (56.7 59.0) <.00 1 Smoking pack years, mean (95% CI) 2.6 (2.3 2.9) 2.4 (2.0 2.7) .32 4.1 (2.5 5.7) 2.5 (1.8 3.3) .06 Illicit drug use No Yes 359 (93) 28 (7) 266 (92) 22 (8) .84 48 (76) 15 (24) 158 (81) 37 (19) .40 Body Mass Index, mean (95% CI) 28.3 (27.5 29.0) 28.8 (27.9 29.6) .37 25.4 (24.6 26.2) 26.1 (25.4 26.7) .18 Hypertension No Yes 295 (76) 92 (24) 225 (78) 63 (22) .56 38 (60) 25 (40) 123 (63) 72 (37) .69 Diabetes No Yes 297 (77) 90 (23) 214 (74) 74 (26) .46 45 (71) 18 (29) 128 (66) 67 (79) .39 Hepatitis C status Negative Positive 290 (75) 97 (25) 218 (76) 70 (42) .82 56 (89) 7 (11) 175 (90) 20 (10) .85 HIV RNA Viral Load < 200 copies/mL 200 copies/mL 203 (52) 184 (48) 157 (55) 131 (45) .60 39 (62) 24 (38) 178 (91) 17 (9) <.001 10 year alcohol consumption pattern Abstinent Low Moderate Heavy 135 (35) 226 (58) 26 (7) 96 (33) 180 (62) 12 (4) .29 6 (15) 53 (27) 4 (21) 35 (18) 145 (74) 15 (79) .24

PAGE 111

111 APPENDIX C ADDITIONAL CHAPTER 5 TABLES Table C 1. Cross tabulation of past and current alcohol consumption patterns at baseline carotid artery ultrasound WIHS Current Alcohol Consumption MACS Cur r ent Alcohol Consumption Abstinent Moderate Heavy Abstinent Moderate Heavy Past Alcohol Consumption Abstinent Low Moderate Heavy 391 (93.1) 177 (50.4) 75 (22.5) 8 (10.5) 28 (6.7) 165 (47.0) 223 (66.8) 25 (32.9) 1 (0.2) 9 (2.6) 36 (10.8) 43 (56.6) 59 (100) 154 (65.5) 17 (23.9) 4 (13.3) 0 (0) 79 (33.6) 52 (73.2) 11 (36.7) 0 (0) 2 (0.85) 2 (2.82) 15 (50.0)

PAGE 112

112 Table C 2 Crude and adjusted model estimates of covariates on carotid artery intima thickness among women Crude P value Adjusted P value Time 8.43 (6.83, 10.0 ) <.001 8.23 ( 10.7, 5.79) <.001 Race (Ref= White) African American/Black Other 40.0 (23.4, 56.6) 6.3 ( 25.6, 13.0) <.001 .52 34.2 (20.1, 48.2) 7.01 ( 23.4, 9.39) <.001 .40 Age 6.92 (6.03, 7.81) <.001 6.51 (5.66, 7.42) <.001 Pack years 4.62 (2.03, 7.20) <.001 4.46 (2.09, 6.83) <.001 Illicit drug use (Ref = No) 4.72 ( 2.83, 12.3) .22 2.32 ( 5.71, 10.3) .57 Probable Depression (Ref = No) 3.50 ( 1.65, 8.66) .18 3.18 ( 1.90, 8.26) .22 BMI status 0.68 ( .008, 1.36) .05 0.99 (0.30,1.68) <.01 Diabetes (Ref = No) 37.1 (20.9, 53.2) <.001 23.2 (9.22, 37.1) <.001 Hypertension (Ref = No) 10.8 (5.44, 16.2) <.001 5.71 (0.22, 11.2) .04 copies/mL) 1.13 ( 3.44, 5.70) .63 1.04 ( 3.55, 5.63) .66 Past Alcohol Consumption Pattern (Ref = Abstinence) Low Moderate Heavy 5.71 ( 17.6 6.18 ) 9.40 ( 2 .3 1.54 ) 9.25 ( 12.6 31.1) .35 .09 .41 3.30 ( 14.8, 8.16) 10.8 ( 21.8, 0.19) 12.5 ( 34.2, 9.1) .57 .05 .26 Current Alcohol Consumption Pattern (Ref = Abstinence) Moderate Heavy 1.32 ( 3.42, 6.06 ) 11.0 (2.11, 19.8 ) .56 .01 2.44 ( 2.49, 7.38) 12.2 (3.13, 21.3) .33 <.01

PAGE 113

113 T able C 3 Crude and Adjusted model estimates of covariates on carotid artery intima thickness among men Crude P value Adjusted P value Time 23.1 (19.6, 26.6) <.001 6.10 ( 14.5, 2.26) .15 Race (Ref= White) African American/Black Other 28.4 (1.64, 55.1) 12.8 ( 55.4, 29.7) .04 .55 57.9 (30.7 85.0) 39.1 ( 3.71, 82.0) <.001 .07 Age 6.85 (4.82, 8.89) <.001 7.75 (5.51, 9.98) <.001 Pack years 2.21 (0.61, 3.80) <.01 1.87 (0.29, 3.45) .02 Illicit drug use (Ref = No) 3.00 ( 8.03, 14.0) .59 2.22 ( 8.93, 13.4) .70 Probable Depression (Ref = No) 6.48 ( 2.61, 15.6) .16 9.18 ( 0.13, 18.5) .05 BMI status 1.03 ( 0.24, 2.31) .11 Diabetes (Ref = No) 14.5 ( 11.0, 40.1) .26 0.06 ( 23.7, 23.8) .99 Hypertension (Ref = No) 5.88 ( 12.7, 0.91) .09 3.09 ( 9.81, 3.64) .37 3.17 ( 11.5, 5.13) .45 3.87 ( 12.4, 4.65) .37 Past Alcohol Consumption Pattern (Ref = Abstinence) Low Moderate Heavy 11.6 ( 41.3, 18.1) 18.5 ( 25.5, 62.6) 14.7 ( 38.7, 68.2) .44 .41 .59 25.2 ( 3.01, 53.4) 51.2 (6.91, 95.5) 32.3 ( 21.9, 86.6) .08 .02 .24 Current Alcohol Consumption Pattern (Ref = Abstinence) Moderate Heavy 0.40 ( 8.00, 8.68 ) 8.96 ( 26.5, 8.55) .94 .32 0.31 ( 9.02, 8.39) 13.3 ( 32.1, 5.54) .94 .17

PAGE 114

114 T able C 4 Full model estimates of covariates on carotid artery intima thickness among men and women, adjusting for time by alcohol use interactions Study Multicenter AIDS Cohort Study Adjusted P value Adjusted P value Time 9.11 ( 12.4, 5.79) <.001 15.9 ( 26.2, 5.68) <.01 Race (Ref= White) African American/Black Other 34.1 (20.1, 48.2) 6.91 ( 23.3, 9.50) <.001 .41 56.4 (29.2, 83.6) 39.0 ( 3.6, 81.6) <.001 .07 Age 6.54 (5.67, 7.42) <.001 7.78 (5.56, 9.99) <.001 Pack years 4.46 (1.99, 6.94) <.001 1.76 (0.19, 3.33) .03 Illicit drug use (Ref = No) 2.15 ( 5.92, 10.2) .60 3.83 ( 7.46, 15.1) .51 Probable Depression (Ref = No) 2.96 ( 2.10, 8.01) .25 9.42 ( 1.01, 19.8) .08 BMI status 0.98 (0.35, 0.29) <.01 0.97 ( 0.33, 2.27) .14 Diabetes (Ref = No) 23.2 (9.24, 37.1) <.001 0.21 ( 23.9, 23.5) .99 Hypertension (Ref = No) 5.76 (0.22, 11.3) .04 4.27 ( 11.0, 2.49) .21 1.10 ( 3.53, 5.72) .64 2.85 ( 11.2, 5.47) .50 Past Alcohol Consumption Pattern (Ref = Abstinence) Low Moderate Heavy 6.80 ( 20.4, 6.78) 13.2 ( 26.4, .03) 20.2 ( 46.1, 5.72) .33 .05 .13 2.67 ( 29.9, 35.3) 24.9 ( 28.6, 78.3) 7.60 ( 72.5, 57.3) .87 .36 .82 Current Alcohol Consumption Pattern (Ref = Abstinence) Moderate Heavy 2.63 ( 6.40, 11.7) 21.8 (6.40, 37.2) .57 <.01 1.98 ( 23.5, 27.5) 33.5 ( 3.43, 70.4) .88 .07 Past Alcohol Consumption Pattern X Time Low Moderate Heavy 1.85 ( 2.46, 6.17) 1.23 ( 3.18, 5.63) 3.61 ( 4.01, 11.2) .40 .58 .35 12.5 (4.77, 20.2) 14.9 ( 1.02, 30.8) 20.1 (1.57, 38.7) <.01 .07 .03 Current Alcohol Consumption Pattern X Time Moderate Heavy 0.03 ( 3.68, 3.75) 4.06 ( 11.1, 3.01) .98 .26 1.55 ( 14.3, 11.2) 23.2 ( 41.1, 5.36) .81 .01

PAGE 115

115 Table C 5 Sample Characteristics between those with and without follow up in Aim 3 WIHS MACS Baseline Characteristics No FU 485 (41) FU 696 (59) p value No FU 51 (13) FU 344 (87) p value Race White Black Other 116 (24) 277 (57) 92 (19) 140 (20) 436 (63) 120 (17) .15 36 (70) 13 (25) 2 (9) 216 (63) 107 (31) 21 (6) .53 Probable Depression No Yes 356 (73) 129 (27) 546 (78) 150 (22) .04 44 (86) 7 (14) 301 (88) 43 (12) .81 Illicit Drug Use No Yes 439 (91) 46 (9) 647 (93) 49 (7) .13 42 (82) 9 (18) 274 (80) 70 (20) .65 Diabetes No Yes 381 (79) 104 (21) 540 (78) 156 (22) .69 35 (69) 16 (31) 228 (66) 116 (34) .74 Suppressed Viral load No Yes 233 (48) 252 (52) 349 (50) 347 (50) .48 22 (43) 29 (57) 115 (33) 229 (67) .17 Age, years 45.4 (8.1) 44.8 (7.2) .17 69.1 (5.5) 69.8 (5.6) .44 BMI Status 27.6 28.8 <.01 25.6 (4.3) 25.5 (3.7) .76 Pack years 2.2 (3.1) 1.8 (2.8) .03 2.2 (5.1) 3.3 (6.9) .17 Past Alcohol Use Abstinent Low Moderate Heavy 177 (36) 144 (30) 131 (27) 33 (7) 243 (35) 207 (30) 203 (29) 43 (57) .84 6 (12) 32 (63) 11 (22) 2 (4) 53 (15) 203 (59) 60 (17) 28 (8) .58 Current Alcohol Use Abstinent Moderate Heavy 268 (55) 183 (38) 34 (7) 383 (55) 258 (37) 55 (8) .84 26 (51) 22 (43) 3 (6) 208 (60) 120 (35) 16 (5) .44 Intima Medial Thickness 740.7 (118.1) 724.7 (115.7) 0.02 741.1 (123.4) 742.6 (125.1) .93

PAGE 116

116 LIST OF REFERENCES AIDSinfo. (2016). Virologic Failure: Adult and Adolescent ARV Guidelines. Retrieved from https://aidsinfo.nih.gov/guidelines/html/1/adult and adolescent arv guidelines/15/virologic failure Andruff H Carraro N Thompson A Gaudreau P & Louvet B. (2009). Latent class growth modeling: A tutorial. Tutor Quant itative Methods for Psychology 5, 11 24. Armah K A Chang C C Baker J V Ramachandran V S Budoff M J Crane H M ., Freiberg M S ; Veterans Aging Cohort Study (VACS) Project Team. (2014). Prehypertension, hypertension, and the risk of acute myocardial infarction in HIV infected and uninfected veterans. Clinical Infectious Diseases 58(1), 121 129. Bacon M C von Wyl V Alden C Sharp G., Robison E. Hessol N., Young M. A. (2005) the bench. Clinical and Diagnostic Laboratory Immunology 12(9), 1013 1019 Bahrami H Budoff M Haberlen S A Rezaeian P Ketlogetswe K Tracy R ., Post WS. (2016). Inflammatory markers associated with subclinical coronary artery disease: The multicenter AIDS cohort study. Journal of the American Heart Association 5(6) Barkan S E Melnick S L Preston Martin S Weber K., Kalish L. A., Miotti P., Feldman J. (1998). Epidemiology Cambridge Mass 9(2), 117 125. Bauer M Delaney J A C Mohlenkamp S Jockel K H Kronmal R A Lehmann N McClelland R L ; Multi Ethnic Study of Atherosclerosis and the Investigator Group of the Heinz Nixdorf Recall Study. (2013). Comparison of factors associated with carotid intima media thickness in the multi ethnic study of atherosclerosis (MESA) and the Heinz Nixdork Recall Study (HNR). Jou rnal of the American Society of Echocardiography 26(6), 667 673. Benedetto F. A., Tripepi G., Mallamaci F., & Zoccali C. (2008). Rate of atherosclerotic plaque formation predicts cardiovascular events in ESRD. J Am Soc Nephrol 19(4), 757 763. Bilal U., L au B., Lazo M., McCaul M. E., Hutton H. E., Sulkowski M. S., Chander G. (2016). Interaction between alcohol consumption patterns, antiretroviral therapy t ype, and liver fibrosis in persons living with H IV. AIDS Patient Care and STDs 30(5), 200 207. Brait hwaite R. S., Fang Y., Tate J., Mentor S. M., Bryant K. J., Fiellin D. A., & Justice A. C (2016). Do alcohol misuse, smoking, and depression vary concordantly or sequentially? A longitudinal s tudy of HIV infected and matched uninfected veterans in c a re. AIDS and Behavior 20(3), 566 572. Brenchley J M & Douek D C. (2012). Microbial translocation across the GI tract. Annual Review of Immunology 30, 149 173.

PAGE 117

117 Brinton E. A. (2012). Effects of ethanol intake on lipoproteins. Current Atherosclerosis Rep ort 14, 108 114. Britton A Hardy R Kuh D Deanfield J Charakida M & Bell S (2016) Twenty year trajectories of alcohol consumption during midlife and atherosclerotic thickening in early old age: findings from two British population cohort studies. BMC Medicine 14(1), 111. Butt A A Chang C C Kuller L Goetz M B Leaf D Rimlan d D Freiberg MS. (2011). Risk of heart failure with human immunodeficiency virus in the absence of prior diagnosis of coronary heart disease. Archives of Internal Medicine 171(8), 737 743. Carlsson S Hammar N & Grill V. (2005). Alcohol consump tion and type 2 diabetes Meta analysis of epidemiological studies indicates a U shaped relationship. Diabetologia 48(6) 1051 1054. Carrieri M P Protopopescu C Le Moing V Reboud P Raffi F Mahy S Leport C ; S CO8 APROCO COPILOTE Study Gro up. (2012). Impact of immunodepression and moderate alcohol consumption on coronary and other arterial disease events in an 11 year cohort of HIV infected patients on antiretroviral therapy. British Medical Journal Open 2(6) e001155 Center for Disease Prevention and Control (2016). Alcohol & Public Health. Retrieved at https://www.cdc.gov/alcohol/faqs.htm Chitsaz E Meyer J P Krishnan A Springer S A Marcus R Zaller N Altice F L. (2013). Contribution of substance use disorders on HI V treatment outcomes and antiretroviral medication adherence among HIV infected persons entering jail. AIDS and Behavior 17(Suppl 2), S118 S127. Church T S Levine B D McGuire D K Lamonte M J Fitzgerald S J Cheng Y J Nichaman M Z. (2007). Coronary artery calcium score, risk factors, and incident coronary heart disease events. Atherosclerosis 190(1), 224 231. Conen A Fehr J Glass T R Furrer H Weber R Vernazza P. Battegay M ; Swiss HIV Cohort Study. (2009). Self reported alcohol consumption and its association with adherence and outcome of antiretroviral therapy in the Swiss HIV Cohort Study. Antiviral Therapy 14(3), 349 57. Conigliaro J Gordon A J McGinnis K A Rabeneck L & Justice A C ; Veterans Aging Cohort 3 Site Study. (2003). How harmful is hazardous alcohol use and abuse in HIV infection: do health care providers know who is at risk? Journal of Acquired Immune Deficiency Syndromes 33(4), 521 525. Cook R. L., Zhu F., Belnap B. H., Weber K. M ., Cole S. R., Vlahov D., Cohen M. H. (2013). Alcohol consumption trajectory patterns in adult women with HIV infection. AIDS and Behavior 17(5) 1705 1712.

PAGE 118

118 Conen A Wang Q Glass T R Fux C. A., Thurnheer M. C., Orasch C., Fehr J. (2013). Association of alcohol consumption and HIV surrogate markers in participants of the swiss HIV cohort study. Journal of Acquired Immune Deficiency Syndromes 64(5), 472 478. Corral I Quereda C Moreno A Prez Elas M J Dronda F Casado J L Moreno S. (2009). Cerebrovascular ischemic events in HIV 1 infected patients receiving highly active antiretroviral therapy: incidence and risk factors. Cerebrovascular Diseases 27, 559 563. Corrao G Rubbiati L Bagnardi V Zambon A & Poikolainen K. (2000). Alcohol and coronary heart disease: a meta analysis. Addiction 95(10), 1505 1523. Crystal H. A., Weedon J., Holman S., Manly J., Valcour V., Cohen M., Kaplan R. C. (2011). Associations of cardiovascular variables and HAART with cognition in middle aged HIV infected and uninfected women. Journal of NeuroVirology 17, 469 476. Zingaropoli M A Oliva A D'Agostino C Al Moghazi S De Luca G Vullo V. (2014). Immune activation, immunosenescence, and osteoprotegerin as m arkers of endothelial dysfunction in subclinical HIV associated atherosclerosis. Mediators of Inflammation 2014, 192594. Davis P H Dawson J D Mahoney L T & Lauer R M. (1999). Increased carotid intimal medial thickness and coronary calcification are related in young and middle aged adults. The Muscatine study. Circulation 100(8), 838 842. Deiss R G Mesner O Agan B K Ganesan A., Okulicz J. F., Bavaro M., Macalino G. E (2016) Characterizing the association between a lcohol an d HIV virologic failure in a military cohort on antiretroviral t herapy. Alcoholism: Clinical and Experimental Research 40(3), 529 535. De Luca G Venegoni L Iorio S Secco G G Cassetti E Verdoia M Marino P ; Novara Atherosclerosis Study Group. (2010) Platelet distribution width and the extent of coronary artery disease: results from a large prospective study. Platelets 21(7), 508 14. Del Boca F K & Darkes J. (2003). The validity of self reports of alcohol consumption: state of the science and challenges for research. Addiction 98 Suppl( 2 ), 1 12. Detels R Phair J P Saah A J Rinaldo C R Murioz A & Kaslow R A. (1992). Recent scientific contributions to un derstanding HIV/AIDS from the Multicenter AIDS Cohort Study. J ournal of Epidemiol ogy 2(2 s up), 11 19. Dudley J Jin S Hoover D Metz S Thackeray R & Chmiel J. (1995). The Multicenter AIDS Cohort Study: retention after 9 1/2 years. American Journal of Epidemiology 142(3), 323 330.

PAGE 119

119 Durand M Sheehy O Baril J G Lelorier J & Tremblay C L. (2011). Association between HIV infection, antiretroviral therapy, and risk of acute myocardial infarction: a cohort and nested case Journal of Acquired Immune Deficiency Syndromes 57(3), 245 253. Du rand M Sheehy O Baril J G LeLorier J & Tremblay C L. (2012). Risk of spontaneous intracranial hemorrhage in HIV infected individuals: a population based cohort study. Journal of Stroke and Cerebrovascular Diseases 22(7), e34 e41. Duvall W. L. (2 003). Cardiovascular disease in women. Mt Sinary J Med 70(5), 293 305. Engel GL. (1977). The need for a new medical model: a challenge for biomedicine. Science 196(4286), 129 136. Fernandez Friera L., Penalvo J. L., Fernandez Ortiz A., Ibanez B., Lopez Melgar B., Laclaustra M., Fuster V. (2015). Prevalence, vascular distribution, and multiterritorial extent of subclinical atherosclerosis in a middle aged cohort: The PESA (Prog ression of Early Subclinical Atherosclerosis) Study. Circulation 131(24), 2104 2113. Freiberg M S Chang C C Kuller L H Skanderson M Lowy E Kraemer K L Justice A C. (2013). HIV infection and the risk of acute myocardial infarction. JAMA Internal Medicine 173(8), 614 622. Freiberg M S Chang C C Skanderson M McGinnis K Kuller L H Kraemer K L Justice A C ; Veterans Aging Cohort Study. (2011). The risk of incident coronary heart disease among veterans with and wit hout HIV and hepatitis C. Circulation: Cardiovascular Quality and Outcomes 4(4), 425 432. Freiberg M. S. & Kraemer K. L. (2010). Focus on the heart: Alcohol consumption, HIV infection, and cardiovascular disease. Alcohol Research & Health 33(3), 237 246. Freiberg M S McGinnis K A Kraemer K Samet J. H., Conigliaro J., Curtis Ellison R Justice A C ; VACS Project Team. (2010) The association between alcohol consumption and prevalent cardiovascular diseases among HIV infected and HIV uninfect ed men. Journal of Acquired Immune Deficiency Syndromes 53(2), 247 253. Freiberg M S & So Armah K. (2016). HIV and cardiovascular disease: We need a mechanism, and we need a plan. Journal of the American Heart Association 5(3) e003411. Fu ster D Tsui J I Cheng D M Quinn E K Bridden C Nunes D Samet J H (2013). Impact of lifetime alcohol use on liver fibrosis in a population of HIV infected patients with and without hepatitis C coinfection. Alcoholism: Clinical and Experimental Rese arch 37(9), 1527 1535 Galvan F H Bing E G Fleishman J A London A S Caetano R Burnam M A Shapiro M. (2002). The prevalence of alcohol consumption and heavy drinking among people with HIV in the United States: results from the HIV Co st and Services Utilization Study. Journal of Studies on Alcohol 63(2), 179 86.

PAGE 120

120 George A & Movahed A. (2008). Coronary artery calcium scores: current thinking and clinical applications. The Open Cardiovascular Medicine Journal 2, 87 92. Grimaud O., Lapostolle A., Berr C., Helmer C., Dufouil C., Kihal W., .Chauvin P. (2013). Gender differences in the association between socioeconomic status and subclinical atherosclerosis. PLOS ONE 8(11), e80195. Grunfeld C Delaney J A Wanke C Currier J S Scherzer R Biggs M L Kronmal RA. (2009). Preclinical atherosclerosis due to HIV infection: carotid intima medial thickness measurements from the FRAM study. AIDS 23(14), 1841 9. Hadigan C., Jeste S., Anderson E. J., Tsay R., Cyr H., & Grinspood S. (2001). Modifiable dietary habits and their relation to metabolic abnormalities in men and women with human immunodeficiency virus infection and fat redistribution. Clinical Infectious Diseases 33(5), 710 717. Hanna D B Post W S D eal J A Hodis H N Jacobson L P Mack W J Kingsley L A. (2015). HIV infection is associated with progression of subclinical carotid atherosclerosis. Clinical Infectious Diseases 61(4), 640 50. Hansson G K. (2005). Inflammation, atherosclerosis, and coronary artery disease. The New England Journal of Medicine 352(16), 1685 95. Hendershot C. S., Stoner S. A., Pantalone D. W., & Simoni J. M. (2009). Alcohol use and antiretroviral adherence: review and meta analysis. Journal of Acquired Immune Deficiency Syndromes 52(2), 180 202. Hejazi N., Rajikan R., Choong C. L., & Sahar S. (2013). Metabolic abnormalities in adult HIV infected population on antiretroviral medication in Malaysia: a cross sectional survey. BMC Public Health 13, 758. Hodis H N Mack W J Lobo R A Shoupe D Sevanian A Mahrer P R Azen S P ; Estrogen in the Prevention of Atherosclerosis Trial Research Group. (2001). Estrogen in the prevention of atherosclerosis. A randomized, double b lind, placebo controlled trial. Annals of Internal Medicine 135, 939 53. Hougaku H Fleg J L Lakatta E G Scuteri A Earley C J Najjar S Metter E J. (2005). Effect of light to moderate alcohol consumption on age associated arterial stiffening. American Journal of Cardiology 95, 1006 1010. Hsu D C Ma Y F Hur S Li D Rupert A Scherzer R Hsue P Y. (2016). Plasma IL 6 levels are independently associated with atherosclerosis and mortality in HIV infected individuals on suppressive antiretroviral therapy. AIDS 30(1 3), 2065 74. Hsue P Y Hunt P W Schnell A Kalapus S C Hoh R Ganz P Deeks S. G. (2009). Role of viral replicatio n, antiretroviral therapy, and immunodeficiency in HIV associated atherosclerosis. AIDS 23(9), 1059 1067.

PAGE 121

121 Hsue P Y Scherzer R Hunt P W Schnell A Bolger A F Kalapus S C Deeks M D. (2012). Carotid intima media thickness progression in HIV infected adults occurs preferentially at the carotid bifurcation and is predicted by inflammation. Journal of the American Heart Association 1(2). Iemolo F., Mertiniuk A., Steinman D. A., Spence J. D. (20 04). Sex differences in carotid plaque and stenosis. Stroke 35(2), 477 481. Izadnegahdar M., Singer, J., Lee M. K., Gao M., Thompson C. R. Kopec J., Humphries K. H. (2014). Do younger women fare worse? Sex differences in acute myocardial infarction hospit alization and early mortality rates over ten years. J Womens Health (Larchmt) 23(1), 10 17. Jacob T., Blonigen D. M., Upah R., & Justice A. (2013). Lifetime Drinking Trajectories Among Veterans in Treatment for HIV. Alcoholism: Clinical and Experimental R esearch 37(7), 1179 1187. Jolley S. E., Alkhafaf Q., Hough C., & Welsh D. A. (2016). Presence of an alcohol use disorder is associated with greater pneumonia severity in hospitalized HIV infected patients. Lung 194(5), 755 62. Justice A. C., Lasky E., Mc Ginnis K. A., Skanderson M., Conigliaro J., Fultz S. L., Bryant K.; VACS 3 Project Team. (2006). Medical disease and alcohol use among veterans with human immunodeficiency infection: A comparison of disease measurement strategies. Medical Care 44(8 Suppl 2), S52 S60. Justice A. C., McGinnis K. A., Tate J. P., Braithwaite R. S., Bryant K. J., Cook R. L., Fiellin D. A (2016). Risk of mortality and physiologic injury evident with lower alcohol exposure among HIV infected compared with uninfected men. Drug and Alcohol Dependence 161, 95 103. Justice A C Zingmond D S Gordon K S Fultz S L Goulet J L King J T Jr, Mattocks K M ; Veterans Aging Cohort Study Project Team. (2008). Drug toxicity, HIV progression, or comorbidity of aging: does tipranavir use increase the risk of intracranial hemorrhage? Clinical Infectious Diseases 47, 1226 1230. Kader R Seedat S Govender R Koch J R & Parry C D. (2014). Hazardous and harmful use of alcohol and/or other drugs and health status among South African patients attending HIV clinics. AIDS and Behavior 18(3), 525 534. Kahler C. W., Wray T. B., Pantalone D. W., Mastroleo N. R., Kruis R. D., Mayer K. H., & Monti P. M (2015). Assessing sexual motives for drinking alcohol among HIV positive men who have sex with men. Psychology of Addictive Behaviors 29(1), 247 253. Kalichman S. C., Grebler T., Amaral C. M., McNer ney M., White D., Kalichman M. O., Eaton L (2014). Viral suppression and antiretroviral medication adherence among alcohol using HIV positive adults. International Journal of Behavioral Medicine 21(5), 811 820.

PAGE 122

122 Kalichman S C. Rompa D Cage M. (2000). Distinguishing between overlapping somatic symptoms of depression and HIV disease in people living with HIV AIDS. Journal of Nervous and Mental Disease 188, 662 670 Kaplan R C Kingsley L A Gange S J Benning L Jacobson L P Lazar J Hodis H N. (2008). Low CD4+ T cell count as a major atherosclerosis risk factor in HIV infected women and men. AIDS 22, 1615 24. Kaslow R A Ostrow D G Detels R Phair J P Polk B F & Rinaldo C R. (1987). The Multicenter AIDS Cohort St udy: rationale, organization, and selected characteristics of the participants. American Journal of Epidemiology 126(2), 310 318. Kell y S. G., Plankey M., Post W. S., Li X., Stall R., Jacobson L. P., Palella F. J. Jr. (2016). Associations between To bacco, Alcohol, and Drug Use with Coronary Artery Plaque among HIV Infected and Uninfected Men in the Multicenter AIDS Cohort Study. PloS One 11(1), e0147822. Kelso N E Sheps D S & Cook R L. (2015). The association between alcohol use and cardiovascular disease among people living with HIV: a systematic review. The American Journal of Drug and Alcohol Abuse 41(6), 479 88 Kim M K Shin J Kweon S S Shin D H Lee Y H Chun B Y & Choi B Y. (2014). Harmful and beneficial relationships between alcohol consumption and subclinical atherosclerosis. Nutrition, Metabolism and Cardiovascular Diseases 24, 767 776. Klatt N R Funderburg N T & Brenchley J M. (2013). Microbial translocation, i mmune activation, and HIV disease. Trends in Microbiology 21(1), 6 13. Knott C Bell S & Britton A. (2015). Alcohol consumption and the risk of type 2 diabetes: A systematic review and dose response meta analysis of more than 1.9 m illi on individuals from 38 observational s tudies. Diabetes Care 38(9), 1804 1812. Kohsaka S Jin Z Rundek T Homma S Sacco R L & Di Tullio M R. (2011). Alcohol consumption and atherosclerotic burden in the proximal thoracic aorta. Atherosclerosis 219(2), 794 8. K orkeila J., Vahtera J., Korkeila K., Kivimaki M., Sumanen M., Koskenvuo K., & Koskenvuo M. (2010). Childhood adversities as predictors of incident coronary heart disease and cerebrovascular disease. Heart 96(4), 298 303. Krejza J., Arkuszewski M., Kasner S. E., Weigele J., Ustymowicz A., Hurst R. W., .Messe S. R. (2006). Carotid artery diameter in men and women and the relation to body and neck size. Stroke 37(4), 1103 1105. Kumar V., Abbas A. K., & Aster J. C. (2007). Robbins Basic Pathology (8 th Edi tion). Philadelphia, PA. Saunders/Elsevier.

PAGE 123

123 Lazo M., Chen Y., McEvoy J. W., Ndumele C., Konety S., Ballantyne C. M., Selvin E. (2016). Alcohol consumption and cardiac biomarkers: The Atherosclerosis Risk in Communities (ARIC) Study. Clinical Chemistr y 62(9), 1202 1210. Lee Y H Shin M H Kweon S S Choi S W Kim H Y Ryu S Y Choi J S. (2009). Alcohol consumption and carotid artery structure in Korean adults aged 50 years and older. BMC Public Health 9, 358. Lester S J Eleid M F Khandheria B K & Hurst R T. (2009). Carotid intima media thickness and coronary artery calcium score as indications of subclinical atherosclerosis. Mayo Clinic Proceedings 84(3), 229 233. Lohse N., Hansen A. B., Pedersen G., Kronborg G., Gerstof t J., Srensen H. T., ., Obel N. (2007). Survival of persons with and without HIV infection in Denmark, 1995 2005. Annals of Internal Medicine 146(2), 87 95. Longo Mbenza B., Mashi M. L., Tshikwela M. L., Mokondjimobe E., Gombet T., Ellenga Mbolla B., Fuele S. M. (2011). Relationship between younger age, autoimmunity, cardiometabolic risk, oxidative stress, HAART, and ischemic stroke in Africans with HIV/AIDS. ISRN Cardiology 2001, 897908. Low Beer S Yip B V Hogg R S & Montaner J S. (2000). Adherence to triple therapy and viral load response. Journal of Acquired Immune Deficiency Syndromes 23(4), 360 361. Malbergier A Amaral R A do, & Cardoso L D. (2015). Alcohol dependence and CD4 cell count: is there a relation ship? AIDS Care 27(1), 54 58. Maniar A Ellis C Asmuth D Pollard R & Rutledge J. (2013). HIV infection and atherosclerosis: evaluating the drivers of inflammation. European Journal of Preventive Cardiology 10(5), 720 728. Marshall B D L Operario D Bryant K J Cook R. L Edelman E. J., Gaither J. R., .Fiellin D. A. (2015). Drinking trajectories among HIV infected men who have sex with men: a cohort study of United States veterans. Drug and Alcohol Dependence 148, 69 76. Marshall B D L ., Shoveller J. A., Kahler C. W., Koblin B. A., Mayer K. H., Mimiaga M. J., Operario D. (2015). Heavy drinking trajectories among men who have sex with men: a longitudinal, group based analysis. Alcoholism: Clinical and Experimental Research 39(2), 380 389. Mathur P., Ostadal B., Romeo F., Mehta J. L. (2015). Gender related differences in atherosclerosis. Cardiovasc Drugs Ther 29(4), 319 327. McElduff P & Dobson A J. (1997). How much alcohol and how often? Population based case control st udy of alcohol consumption and risk of major coronary event. British Medical Journal 314(7088), 1159 1164.

PAGE 124

124 McGinni s K. A., Fiellin D. A., Tate J. P., Cook R. L., Braithwaite R. S., Bryant K. J., Justice A. C.; Veterans Aging Cohort Study. (2016). status and viral load in m en. AIDS and Behavior 20(3), 504 511. Mguez Burbano M J Espinoza L Vargas M & LaForest D. (2014). Mood disorders and BDNF relationship with alcohol drinking trajectories among PLWH receiving c are. Journal of Alcoholism & Drug Dependence 2(2), 148. Mguez Burbano M J Lewis J E & Malow R. (2009). Alcohol and race/ethnicity elicit differen t changes in lipid profiles in HIV infected individuals receiving highly active antiretroviral therapy. Journal of the Association of Nurses in AIDS Care 20(3), 176 183. Miller P E Budoff M Zikusoka M Li X Palella F Jr. Kingsley L A Po st W S. (2014). Comparison of racial differences in plaque composition and stenosis between HIV positive and HIV negative men from the Multicenter AIDS Cohort Stu dy. American Journal of Cardiology 114(3), 369 375. Molina P E Bagby G J & Nelson S. ( 2014). Biomedical consequences of alcohol use disorders in the HIV infected host. Current HIV Research 12(4), 265 275. Monni g M. A., Kahler C. W., Cioe P. A., Tucker L., Monti P. M., Mayer K. H., & Ramratnam B. (2016). Alcohol use predicts elevation in inflammatory marker soluble CD14 in men living with HIV. AIDS Care 28(11), 1434 40. Monroe A. K., Dobs A. S., Xu X., Palella F. J., Kingsley L. A., Post W. S., Brown T. T. (2012). Low free testosterone in HIV infec ted men is associated with subclinical cardiovascular disease. HIV Medicine 13, 358 366. Monr oe A. K., Lau B., Mugavero M. J., Mathews W. C., Mayer K. H., Napravnik S., Chander G (2016). Heavy alcohol use is associated with worse retention in HIV c are. Journal of Acquired Immune Deficiency Syndromes 73(4), 419 425. Mukamal K J Conigrave K M Mittleman M A Camargo C A Jr, Stampfer M J Willett W C & Rimm E B. (2003). Role of drinking pattern and type of alcohol consumed in coronary h eart disease in men. The New England Journal of Medicine 348(2), 109 118. Mukamal K J Kronmal R A Mittleman M A H Polak J F Cushman M & Siscovick D S. (2003). Alcohol consumption and carotid at herosclerosis in older adults: T he Cardiovascular Health Study. Arteriosclerosis, Thrombosis, and Vascular Biology 23(12), 2252 9. Nambi V., Chambless L., He M., Folson A. R., Mosley T., Boerwinkle E., & Ballantyne C. M. (2012). Common carotid artery intima media thickness is as good as carotid intima media thickness of all carotid artery segments in improving prediction of coronary heart disease risk in the Atheroclerosis Risk in Communities (ARIC) study. European Heart Journal 33, 183 190.

PAGE 125

125 Nambi V., Chambless L., Folsom A. R., He M., Hu Y., Mosley T., .Ballantyne C. M. (2010). Carotid intima media thicknes and presence or absence of plaque improves prediction of coronary heart disease risk. The ARIC (Atherosclerosis Risk in Communities) Study. Journal of the American College of Car diology 55(15), 1600 1607. National Heart, Lung, and Blood Institute. (2014). HIV Related Research in the Division of Cardiovascular Sciences (DCVS). Retrieved at www.nhlbi.nih.gov/research/funding/aids/about/hiv research dcvs.htm National Heart, Lung, a nd Blood Institute. (2016). How is Atherosclerosis Treated? Retrieved at https://www.nhlbi.nih.gov/health/health topics/topics/atherosclerosis/treatment National Institute on Alcohol Abuse and Alcoholism. (2014). Strategic Plan. Retrieved at www.niaaa.nih .gov/about niaaa/our work/strategic plan. National Institute on Alcohol Abuse and Alcoholism. (2016). Alcohol Use Disorder: A Comparison Between DSM IV and DSM 5. Retrieved at https://pubs.niaaa.nih.gov/publications/dsmfactsheet/dsmfact.htm National Insti tute of Health. (2011). National Heart, Lung, and Blood Institute : What is Atherosclerosis? Retrieved at http://www.nhlbi.nih.gov/health/health topics/topics/atherosclerosis/ National Institutes of Health Office of AIDS Research. (2014). FY 2014 Trans NIH Plan for HIV Related Research. Priority: Improving Disease Outcomes for HIV Infected Individuals. Retrieved at www.oar.nih.gov/strategicplan/fy2014/pdf/FY2014_improving_disease.pdf. Neuhaus J Jacobs D R Jr, Baker J V Calmy A Duprez D La Rosa A Neaton J D. (2010). Markers of inflammation, coagulation, and renal function are elevated in adults with HIV infection. The Journal of Infectious Diseases 201(12), 1788 1795. Njolstad I., Arnesen E., Lund Larson P. D. (1996). Smoking, serum lipid s, blood pressure, and sex differences in myocardial infarction. A 12 year follow up of the Finnmark study. Circulation 93(3), 450 456. Oei H. H., Vliegenthart R., Hak A. E., Iglesias del Sol A., Hofman A., Oudkerk M., Witteman J. C. (2002). The associati on between coronary calcification assessed by electron beam computed tomography and measures of extracoronary atherosclerosis: the Rotterdam Coronary Calcification Study. Journal of the American College of Cardiology 39(11), 1745 1751. Pacek L R Harrel l P T & Martins S S. (2014). Cigarette smoking and drug use among a nationally representative sample of HIV positive individuals. The American Journal on Addictions 23(6), 582 590.

PAGE 126

126 Palella F J Jr, Baker R K Moorman A C Chmiel J S Wood K C Brooks J T & Holmberg S D ; HIV Outpatient Study Investigators. (2006). Mortality in the highly active antiretroviral therapy era: changing causes of death and disease in the HIV outpatient study. Journal of Acquired Immune Deficiency Syndromes 43(1), 27 34. Pandey A. K., Blaha M. J., Sharma K., Rivera J., Budoff M. J., Blankstein R., .Nasir K. (2014). Family history of coronary heart disease and the incidence and progression of coronary artery calcification: Multi Ethnic Study of Atheroscler osis (MESA). Atherosclerosis 232(2), 369 376. Parrinello C. M., Landay A. L., Hodis H. N., Gange S. J., Norris P. J., Young M., Kaplan R. C. (2012). Association of subclinical atherosclerosis with lipid levels amongst antiretroviral treated and untr eated HIV HIV Study. Atherosclerosis 225(2), 408 411. Parsons J T Starks T J Millar B M Boonrai K & Marcotte D. (2014). Patterns of substance use among HIV positive adults over 50: Implications for trea tment and medication a dherence. Drug and Alcohol Dependence 139, 33 40 Paterson D L Swindells S Mohr J Brester M Vergis E. N., Squier C., Singh N (2000). Adherence to protease inhibitor therapy and outcomes in patients with HIV infection. Annals of Internal Medicine 133(1), 21 30. Pellowski J A Kalichman S C Kalichman M O & Cherry C. (2016). Alcohol antiretroviral therapy interactive toxicity beliefs and daily medication adherence and alcohol use among people living with HIV. AIDS Care 28(8), 963 970. Pietraszek A Gregersen S & Hermansen K. (2010). Alcohol and type 2 diabetes. A review. Nutrition, Metabolism and Cardiovascular Diseases NMCD 20(5), 366 375. Polak J. F. & detected common carotid artery intima media thickness and incident coronary heart disease n the multi ethnic study of atherosclerosis. Journal of the American Heart Association 4, e001492. (2015). Associations of coronary heart disease with common carotid artery near and far wall intima media thickness (IMT): the Multi Ethnic Study of Atherosclerosis. Journal of the American Society of Echocardiography 28(9), 1114 1121. Radloff L S. (1977 ). The CES D s cale : A self report depression scale for research in the general p opulation. Applied Psychological Measurement 1(3), 385 401. Reid M C Fiellin D A & G. (1999). Hazardous and harmful alcohol consumption in primary care. Arch ives of Internal Medicine 159(15), 1681 1689.

PAGE 127

127 Rentsch C Tate J P Akgn K M Crystal S., Wang K. H., Ryan Greysen S., Rimland D (2016). Alcohol related diagnoses and all cause hospitalization among HIV infected and uninfected p a tients: A longitudinal analysis of united states v eterans from 1997 to 2011. AIDS and Behavior 20(3), 555 564. Reynolds K Lewis B Nolen J D Kinney G L Sathya B & He J. (2003). Alcohol consumption and risk of stroke: A meta analysis. Journal of the Ameri can Medical Association 289(5):579 588. Rich Edwards J W Mason S Rexrode K Spiegelman D Hibert E Kawachi I Wright R J. (2012). Physical and sexual abuse in childhood as predictors of early onset cardiovascular events in women. Circulat ion ,126(8), 920 927. Robinson J. G., Fox K. M., Bullano M. F., Grandy S., & the SHIELD Study Group. (2009). Atherosclerosis profile and incidence of cardiovascular events: a population based survey. BMC Cardiovascular Disorders 9(46). Rodger A. J., Lodwi ck R., Schechter M., Deeks S., Amin J., Gilson R., Phillips A.; INSIGHT SMART, ESPRIT Study Groups. (2013). Mortality in well controlled HIV in the continuous antiretroviral therapy arms of the SMART and ESPRIT trials compared to the general populati on. AIDS 27(6), 973 979. R uggles K. V., Fang Y., Tate J., Mentor S. M., Bryant K. J., Fiellin D. A., Braithwaite R. S (2016). What are the patterns between depression, smoking, unhealthy alcohol use, and other substance use among individuals receiving medical care? A longitudinal study of 5479 p articipants. AIDS and Behavior Epub ahead of print. Sacco R L Elkind M Boden Albala B Lin I F Kargman D E Hauser W A Paik M C. (1999). The protective effect of moderate alcohol consumption on ischemic stroke. Journal of the American Medical Association 281(1), 53 60. Saitz R Saxon A & Hermann R. (2016). Screening for unhealthy use of alcohol and other drugs in primary care. Retrieved at https://www.uptodate.com/contents/screening for unhealthy use of alcohol and other drugs in primary care/print. Sander D Kukla C Klingelhofer J Winbeck K & Conrad B. (2000). Relationship between circadian blood pressure patterns and progression of early carotid atherosclerosis: A 3 year follow up study. Circulation 102, 1536 1541. Schminke U Luedemann J Berger K Alte D Mitusch R Wood W G Kessler C. (2005). Association between alcohol consumption and subclinical carotid atherosclerosis: The study of hea lth in Pomerania. Stroke 36, 1746 1752. Schulz U. G. & Rothwell P. M. (2001). Sex differences in carotid bifurcation anatomy and the distribution of atherosclerotic plaque Stroke 32(7), 1525 1531.

PAGE 128

128 Shah A J Veledar E Hong Y Bremner J D & Vaccarino V. (2011). Depression and history of attempted suicide as risk factors for heart disease mortality in young individuals. Arch Gen Psychiatry 6 8 (11), 1135 1142. Sho ptaw S., Stall R., Bordon J., Kao U., Cox C., Li X., Plankey MW (2012). Cumulative exposure to stimulants and immune function outcomes among HIV positive and HIV negative men in the Multicenter AIDS Cohort Study. International Jo urnal of STD & AIDS 23(8), 576 580. Shrestha S Irvin M R Grunfeld C & Arnett D K. (2014). HIV, inflammation, and calcium in atherosclerosis. Arteriosclerosis, Thrombosis, and Vascular Biology 34(2), 244 50. Smith C. J., Ryom L., Weber R., Morlat P., Pradier C., Reiss P., Lundgren J. D.; D:A:D Study Group. (2014). Trends in underlying causes of death in people with HIV in 1999 to 2011 (D:A:D): a multicohort collaboration. Lancet 384(9939), 241 8. Sosef M N Bosch J G van Oostayen J Visser T Reiber J H Rosendall F R. (1994). Relation of plasma coagulation factor VII and fibrinogen to carotid artery intima media thickness. Thrombosis and Haemostasis 72(2), 250 4. Sozzi F. B., Danzi G. B., Foc o L., Ferlini M., Tubaro M., Galli M., Mannucci P. M. (2007). Myocardial infarction in the young: a sex based comparison. Coron Artery Dis 18(6), 429 431. Stein J H Korcarz C E Hurst R T Lonn E Kendall C B Mohler E R Post W S ; American Society of Echocardiography Carotid Intima Media Thickness Task Force. (2008). Use of carotid ultrasound to identify subclinical vascular disease and evaluate cardiovascular disease risk: A consensus statement from the American Society of Echoc ardiography Carotid Intima Media Thickness Task Force. Endorse by the Society for Vascular Medicine. Journal of the American Society of Echocardiography 21(2), 93 111. Stein M Herman D S Trisvan E Pirraglia P Engler P Anderson B J. (2005). Alcohol use and sexual risk behavior among human immunodeficiency virus positive persons. Alcoholism: Clinical and Experimental Research 29(5), 837 43. Stern R., Tattersall M. C., Gepner A.D., Korcarz C. E., Kaufman J., Colangelo L. A., .Stein J. H. (2015). Sex differences in predictors of longitudinal changes in carotid artery stiffness: The Multi Ethnic Study of Atherosclerosis (MESA). Arterioscler Thromb Vasc Biol 35(2), 478 484. importance of selected risk factors for various manifestations of cardiovascular disease among men and women from 35 to 64 years old: 30 years of follow up in the Fra mingham Study. Circulation 75(6 Pt 2), V65 73.

PAGE 129

129 Strategies for Management of Antiretroviral Therapy (SMART) Study Group, El Sadr W M Lundgren J Neaton J D Gordin F Abrams D Rappoport C. (2006). CD4+ count guided interruption of antiretro viral treatment. The New England Journal of Medicine 355(22), 2283 2296. Sullivan L E Goulet J L Justice A C & Fiellin D A. (2011). Alcohol consumption and depressive symptoms over time: a longitudinal study of patients with and without HIV infection. Drug and Alcohol Dependence 117(2 3), 158 163. S (2013). Use of the Alcohol Use Disorders Identification Test (AUDIT) to determine the prevalence of alcohol misuse among HIV infected individuals. International Journal of STD & AIDS 24(7), 517 521. Tofferi J K Taylor A J Feuerstein I M G. (2 004). Alcohol intake is not associated with subclinical coronary atherosclerosis. American Heart Journal 148(5), 803 9. Tran B X Nguyen L T Do C D Nguyen Q L & Maher R M. (2014). Associations between alcohol use disorders and adherence to ant iretroviral treatment and quality of life amongst people living with HIV/AIDS. BMC Public Health 14, 27. Triant V A Lee H Hadigan C & Grinspoon S K. (2007). Increased acute myocardial infarction rates and cardiovascular risk factors among patients with human immunodeficiency virus disease. The Journal of Clinical Endocrinology & Metabolism 92(7), 2506 2512. Tsui J I Cheng D M Quinn E ., Bridden C., Merl in J. S., Saitz R., Samet J. H (2016). Pain and Mortality Risk in a Cohort of HIV Infected Persons with Alcohol Use Disorders. AIDS and Behavior 20(3), 583 589. Twagirumukiza M., Nkeramihigo E., Seminega B., Gasakure E., Boccara F., & Barbaro G. (2007). Prevalence of dilated cardiomyopathy in HIV infected African patients not receiving HAART: a multicenter, observational, prospective, cohort study in Rwanda. Current HIV Research 5(1): 129 137. US Census Bureau. (2000). National Population Projections. Re trieved at https://www.census.gov/population/projections/data/national/ US Preventive Services Task Force. (2013). Final Update Summary: Alcohol Misuse: Screening and Behavioral Counseling Interventions in Primary Care. Retrieved at https://www.uspreventi veservicestaskforce.org/Page/Document/UpdateSummaryFinal/alc ohol misuse screening and behavioral counseling interventions in primary care Vaccarino V & Bremner J D (2016) Behavioral, emotional and neurobiological determinents of coronary heart disease risk in women. Neurosci Biobehav Rev Epub ahead of print.

PAGE 130

130 Vaccarino V., Shah A. J., Rooks C., Ibeanu I., Nye J. A., Pimple P., .Raggi P. (2014). Sex differences in mental stress induced myocardial ischemia in young survivors of an acute myocardial infarction. Psychosom Med 76(3), 171 180. Vagenas P Azar M M Copenhaver M M Springer S A Molina P E Altice F L. (2015). The impact of alcohol use and related disorders on the HIV continuum of care: A lcohol and the HIV continuum of c are. Current HIV/AIDS Reports 12(4), 421 436. Vu K. N., Ballantyne C. M., Hoogeveen R. C., Nam bi V., Volcik K. A., Boerwinkle E., & Morrison A. C. (2016). Causal role of alcohol consumption in an improved lipid profile: The Atherosclerosis Risk in Communities (ARIC) Study. PLoS ONE 11(2): e0148765. Walker R W Jusabani A Aris E Gray W K Unwin N Swai M Mugusi F. (2013). Stroke risk factors in an incident population in urban and rural Tanzania: a prospective, community based, case control study. The Lancet Global Health 1(5), e282 e288. Wandeler G Kraus D Fehr J Conen A., C almy A., Orasch C., Furrer H.; Swiss HIV Cohort Study (2016). The J curve in HIV: Low and moderate alcohol intake p redicts mortality but not the occurrence of major cardiovascular e vents. Journal of Acquired Immune Deficiency Syndromes 71 (3), 302 309. The White House Office of National AIDS Policy. (2010). National HIV/AIDS Strategy for the United States. Retrieved at www.whitehouse.gov/sites/default/files/uploads/NHAS.pdf. Williams E C Hahn J A Saitz R Bryant K Lira M C & Samet J H. (2016). Alcohol u se and Human Immunodeficiency Virus (HIV) Infection: Current knowledge, implications, and f utur e d irections. Alcoholism: Clinical and Experimental Research 40(10), 2056 2072 Wilson P W B Levy D Belanger A M S ilbershatz H & Kannel W B. (1998). Prediction of coronary heart disease using risk factor categories. Circulation 97(18), 1837 1847. Womack J A Chang C H So Armah K A Alcorn C Baker J V Brown S T Freiberg M S. (2014). HIV infection and cardiovascular disease in women. Journal of the American Heart Association 3(5), e001035. Xie X Ma Y T Yang Y N Fu Z Y Ma X Huang D Gao X. (2012). Alcohol consumption and car otid atherosclerosis in China: T he Cardio vascular Risk Survey. European Journal of Preventive Cardiolog y 19(3), 314 21. Xu X Bao H Strait K Spertus J A Lichtman J H . Krumholz H M. (2015). Sex differences in perceived stress and early recovery in young and middle ag ed patients with acute myocardial infarction. Circulation 1 31(7), 614 623.

PAGE 131

131 Zureik M Garipy J Courbon D Dartigues J F Ritchie K Tzourio C Ducimetire P. (2004). Alcohol consumption and carotid artery structure in older French adults: the Three City Study. Stroke 35(12):2770 5. Zyriax B C Lau K Klhn T Boeing H Vlzke H & Windler E. (2010). Association between alcohol consumption and carotid intima media thickness in a healthy population: d ata of the STRATEGY study (Stress Atherosclerosis and ECG Study). European Journal of Clinical Nutrition 64, 1199 1206.

PAGE 132

132 BIOGRAPHICAL SKETCH Natalie Chichetto graduated suma cum l aude from the University of Missouri in St. Louis with a Bachelor of Art degree in Psychology and Trauma Studies in 2009 and received her Master of Social Work degree from Washington University in St. Louis with a focus in Mental Health and Research Specia lization in 2011. She entered the PhD program in Epidemiology at the University of Florida in 2012. During her time as a PhD student, she was awarded the Graduate School Fellowship, which provided up to 4 years of full funding. In 2015, Natalie was awarded an F31 Individual Research Fellowship ( F31 AA024064 ) through the National Institute for Alcohol Abuse and Alcoholism to fund the last two years of her studies and dissertation work. Her research interests are in behavioral cardiology and chronic disease prevention among vulnerable populations, as well as capacity building at the policy, health care, and community levels to effectively treat and reduce chronic diseases