1 ALCOHOL TREATMENT USE IN HIV INFECTED INDIVIDU ALS WITH HAZARDOUS DRINKING : UTILIZA T I ON, DRINKING OUTCOMES, AND SURVIVAL By XINGDI HU 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 2015
2 Â© 2015 X ingdi Hu
3 To my wife, Yue Zhong, and to my parents, Meisheng Qian and Jianwei Hu
4 ACKNOWLEDGMENTS I have been incredibly fortunate to receive support and encouragement from many people during my graduate study in the University of Florida. I would like to express my deep appreciation and gratitude to my mentor and chair of my di ssertation committee, Dr . Robert L. Cook, for his guidance and support at every progress I made along the way. To me, he is not just a mentor, but also a role model for life and future research career. I am also appreciative to Dr . Jeffrey S. Harman for his insightful suggestions on the study design, results interpretation, and long term support since my transitio n in to public health area. I am also grateful to Dr. Almut G. Winterstein for her constructive comments on my work and Dr. Richa rd D. Rheingans for his help to address challenges during the model development. The ir advice s ha ve significantly improved my wo rk. I would a lso like to thank my wife, Yue Zhong , for her enormous support and encouragement throughout years and my parents for their faith in me and big sacrifice made to support my study in the U nited S tates .
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 LIST OF ABBREVIATIONS ................................ ................................ ............................. 9 ABSTRACT ................................ ................................ ................................ ................... 10 C H A P T E R 1 INTRODUCTION AND LITERATURE REVIEW ................................ ..................... 12 Alcohol Consumption Definition ................................ ................................ .............. 13 Prevalence of Alcohol Use in HIV positive Individuals ................................ ............ 15 Drinking and Health of HIV positive Patients ................................ .......................... 16 Alcohol Treatment ................................ ................................ ................................ ... 21 HIV Disease Simulation Model ................................ ................................ ............... 25 Conclusion ................................ ................................ ................................ .............. 27 2 SPECIFIC AIMS AND HYPOTHESES ................................ ................................ .... 28 Aim1 Utilization of Alcohol Treatment ................................ ................................ ..... 28 Aim2 Treatment Effect on Drinking Outcomes ................................ ........................ 28 Aim3 Survival Impact of Hazardous Drinking and Alcohol Treatment ..................... 29 3 UTILIZATION OF ALCOHOL TREATMENT AMONG HIV POSITIVE WOMEN WITH HAZARDOUS DRINKING ................................ ................................ ............. 31 Introduction ................................ ................................ ................................ ............. 31 Method ................................ ................................ ................................ .................... 33 Study Sample ................................ ................................ ................................ ... 33 Measurements ................................ ................................ ................................ .. 33 Hazardous drinking ................................ ................................ .................... 33 Alcohol treatment use ................................ ................................ ................ 34 Predisposing factors ................................ ................................ .................. 34 Enabling factors ................................ ................................ ......................... 34 Need factors ................................ ................................ ............................... 35 Statistical Analysis ................................ ................................ ............................ 35 Results ................................ ................................ ................................ .................... 36 Discussion ................................ ................................ ................................ .............. 38
6 4 RECEIVING ALCOHOL TREATMENT IS ASSOCIATED WITH LESS ALCOHOL USE AMONG HIV POSITIVE WOMEN WITH HAZARDOUS DRINKING ................................ ................................ ................................ .............. 48 Introduction ................................ ................................ ................................ ............. 48 Method ................................ ................................ ................................ .................... 50 Study Sample ................................ ................................ ................................ ... 50 Dependent Variables ................................ ................................ ........................ 51 Primary Independent Variables ................................ ................................ ........ 51 Other Covariates ................................ ................................ .............................. 52 Statistical Analysis ................................ ................................ ............................ 53 Results ................................ ................................ ................................ .................... 53 Discu ssion ................................ ................................ ................................ .............. 54 5 ESTIMATING SURVIVAL IMPACT OF HAZARDOUS DRINKING AND ALCOHOL TREATMENT USE AMONG PEOPLE WITH HIV INFECTION ............ 61 Introduction ................................ ................................ ................................ ............. 61 Method ................................ ................................ ................................ .................... 63 Model Overview ................................ ................................ ................................ 63 HIV Disease Progression ................................ ................................ ................. 64 Modeling Hazardous Drinking and Alcohol Treatment ................................ ..... 66 Results ................................ ................................ ................................ .................... 69 HIV Progression Model ................................ ................................ ..................... 69 Survival Impact of Hazardous Dinking and Alcohol Treatment ......................... 70 Discussion ................................ ................................ ................................ .............. 71 6 CONCLUSIONS ................................ ................................ ................................ ..... 88 Summary ................................ ................................ ................................ ................ 88 Study 1 Utilization of Alcohol Treatment ................................ ................................ . 88 Study 2 Alcohol Treatment and Drinking Outcomes ................................ ............... 91 Study 3 Survival Impact of Hazardous Drinking and Alcohol Treatment ................. 91 Strengths and Limitations ................................ ................................ ....................... 93 Implications for Public Health and Future research ................................ ................ 94 APPEND IX WIHS QUESTIONS ABOUT ALCOHOL USE AND ALCOHOL TREATMENT .............. 96 LIST OF REFERENCES ................................ ................................ ............................... 97 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 111
7 LIST OF TABLES Table page 3 1 Characteristics of 474 HIV positive WIHS women with recent hazardous drinking (in the past 6 months) ................................ ................................ ........... 44 3 2 Recent alcohol treatment services utilization among HIV infected women with hazardous drinking ................................ ................................ ............................. 45 3 3 Correlates of using any alcohol treatment among HIV positive WIHS women with hazardous drinking (n=474) ................................ ................................ ........ 46 4 1 Characteristics of 343 HIV positive WIHS women with recent hazardous drinking ................................ ................................ ................................ .............. 58 4 2 The association of alcohol treatment with no unhealthy drinking and abstinence among HIV infected women with hazardous drinking ....................... 59 5 1 Summary of input parameters ................................ ................................ ............ 76 5 2 Quarterly probability of deaths by age, CD4 count, viral load, and hazardous drinking ................................ ................................ ................................ .............. 78 5 3 Estimates of mean survival and percentage of non AIDS causes of deaths ...... 79 5 4 Survival years estimated by increasing the chance of receiving any alcohol treatment ................................ ................................ ................................ ............ 80 5 5 Survival years without incorporating of treatment effect on relapse prevention .. 81
8 LIST OF FIGURES Figure page 4 1 treatment drinking visit ................................ ................................ ................................ ........ 60 5 1 Impact of alcohol treatment on both AIDS related and non AIDS related mortality ................................ ................................ ................................ .............. 82 5 2 Estimated mean survival years at age 20 and 35 (gray bar) from base model, and estimates of same ages from the Antiretroviral Treatment Cohort Collaborati on (ART CC) ................................ ................................ ..................... 83 5 3 Trend of survival years depending on utilization rate of alcohol treatment ......... 84 5 4 Marginal survival years by increasing the utilization rate ................................ .... 85 5 5 Trends of survival years by doubling alcohol treatment effect ............................ 86 5 6 Marginal survival years by doubling alcohol treatment effect ............................. 87
9 LIST OF ABBREVIATIONS AA Alcoholics Anonymous AUD Alcohol Use Disorders AIDS Acquired Immune D eficiency Syndrome ART Antiretro viral T herapy AIDS Acquired Immune D eficiency Syndrome AUDIT Alcohol Use Disorder Identification Test ART CC Antiretroviral Treatment Cohort Collaboration CBT C ognitive B ehavior T herapy CDC Center for Disease Control and Prevention CES D Epidemiological Studies Depression Scale DSM Diagnostic and Statistical Manual of Mental Disorders HIV Human Immunodeficiency Virus FDA Food and Drug Administration HAART Highly Active Antiretroviral Therapy HCV Hepatitis C Virus MSM Men Who Have Sex with Men NIAAA National Institute on Alcohol Abuse and Alcoholism NESARC National Epidemiological Survey on Alcohol and Related Conditions STD Sexually Transmitted Diseases WIHS
10 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 TREATMENT USE IN HIV INFECTED INDIVIDUALS WITH HAZARDOUS DRINKING: UTILIZATION , DRINKING OUTCOMES , AND SURVIVAL By Xingdi Hu May 2015 Chair: Robert L . Cook Major: Epidemiology C ommon among HIV infected individuals , hazardous alcohol consumption is associated with increased comorbidities and rapid disease progression . Receiving t reatment to reduce alcohol use may result in substantial benefits in health and well being. However, little is known about alcohol treatment use in HIV infected individuals . Focusing on women with HIV infection who had hazardous drinking, this dissertation had three objectives: (1) examine the utilization of alcohol treatment and correlates of utilization ; (2) assess whether receiving alcohol treatment was associated with im proved drinking outcomes; (3) estimate the survival impact of hazardous drinking and alcohol treatment. W e used the data from WIHS for study 1 and 2. In study 1, we performed multivariable logistic r egression to identify factors associated with utilization. Study 2 used longitudinal multivariable regression to assess the association of a lcohol treat ment with no unhealthy drinking and abstinence. In study 3, we developed an individual based, Monte Carlo simulation model to project the survival impact of hazardous drinking and alcohol treatment utilization among HIV infected individuals .
11 Study 1 found that almost one in five HIV positive women (19%) who had recent hazardous drinking reported utilization of any alcohol treatment; higher levels of alcohol consumption, lower income level s , having social support and rece nt illicit drug use wer e associated with receiving any alcohol treatment. Study 2 found that receiving any alcohol treatment was significantly associated with a 1. 58 greater odds of ac hieving no unhealthy drinking ( OR= 1. 58 , 95% CI = 1.05 2 . 37 ), and 2. 29 greater odds of achieving abstinence ( OR = 2. 29 , 95% CI = 1. 36 3 . 86 ). Study 3 found that hazardous drinking resulted in an average of 1.65 years reduction in projected survival years while receiving any alcohol treatment saved 0.18 years; the chance of treatment receipt and treatm ent effect on both drinking reduction and heavy drinking prevention played important roles in determining the survival impact of alcohol treatment. The findings of these three studies enriched our understanding of alcohol treatment use among HIV infected individuals. These findings further provide important evidence supporting increased efforts to enhance the access and utilization of alcohol treatme nt among HIV infected hazardous drinkers .
12 CHAPTER 1 INTRODUCTION AND LITE RATURE REVIEW First recognized in 1984, t he human immunodeficiency virus (HIV) is the virus that can progressively destroy h uman immune system ( e.g., CD4 cell ) and lead to death if untreated. T he acquired im munodeficiency syndrome (AIDS) is the en d stage of HIV infection in which people become particularly vulnerable to life threatening opportunistic infections (e.g., pneumonia, tuberculosis), some other complications (KrÃ¤mer et al., 2009) . HIV is not curable but can be managed by antiretroviral therapy (ART). The average survival after HIV infection without tre atment is from 9 to 11 years . The introduction of ART has substantially improved the life expectancy of HIV infected individuals which is now approaching that seen in the general populations in s ome high income countries (Samji et al., 2013; Smith et al. , 2014) . In the United States (US), over 1.1 million people are currently living with HIV/AIDS, with 55,000 to 60,000 people newly infected each year (Centers for Disease Control and Prevention, 2013; Hall et al., 2008) . Men who have sex with m en (MSM) remain the group most heavily affected by HIV. MSM represent 4% of all male population in the US but account for over three quarters of all infections in men and nearly two thirds of all new infections in 2010 (Centers for Disease Control and Prevention, 2013; Purcell et al., 2012) . During the past two decades, HIV infection has been increasingly expanded to include women. Today women have account ed for more than one quarter of HIV infected peop le and one fifth of new diagnose s of HIV infection in the US (Centers for Disease Control and Prevention, 2013) . High risk heterosexual contact is responsible for over 80% of newly diagnosed HIV infections i n women with
13 women of color disproportionally affected: African American women who comprise 12 % of female population account for over 60% of all new female HIV infections (Center for Disease Control and Pre vention, 2014) . Alcohol use, especially hazardous drinking, is a modifiable behavior that is associated with a wide array of adverse health outcomes among HIV infected people, such as increased risk of sexual risk behavior, poor adherence to ART, and r apid disease progression (Aza r et al., 2010; Hutton et al., 2012; Wu et al., 2011) . Treatment to reduce alcohol use may introduce substantial benefits by reducing the risk of alcohol related behavior al and health problems . As a result, providers should consistently screen for heavy drinking and refer those with heavy drinking to treatment (Neblett et al., 2011) . To serve as t he introduction section, C hapter 1 will describe important background knowledge, including definition for alcohol use, prevalence of alcohol use among HIV infected in dividuals, drinking and heath of HIV positive patient s , alcohol treatment, and HIV disease simulation model. Alcohol C onsumption Definition Alcohol use ranges from a wide spectrum from non e, never exceeding the limit (moderate drinking) , at risk or hazardous drinking , to alcohol use disorders ( alcohol abuse and dependence ) . Hazardou s drinking is usually defined as a pattern of alcohol use that is associated with increased risk of alcohol related or other health problems but may not necessarily be classified as alcohol use disorders (AUD) . According to the NIAAA (National Institute on Alcohol Abuse and Alcoholism), hazardous drinking is usually defined as exceeding the suggested drinking limits, including more than 3 drinks in a day or more than 7 drinks per week for women , and more than 4 drinks in a day or
14 more than 14 drinks per wee k for men (Willenbring et al., 2009) . Moderate drinking or low risk drinking is defined as having any drinking below hazardous drinking. AUD include two distinct diagnoses: Alcohol abuse and Alcohol dependence. The criteria of diagnosing AUD include the presence of alcohol related negative social, personal, legal, and health consequences as well as addiction manifestations ( craving, tolerance, and withdrawal ) . More specifically, the diagnosis of alcohol abuse ( harmful use of alcohol ) requires meeting one or more of four abuse criteria without meeting any one of dependence criteria. These four criteria include risk of bodily harm, relationship t rouble, role failure, and run ins with the law. On the other hand, t he diagnosis of alcohol dependence (alcoholism) requires meet ing at least three out of seven dependence criteria . These seven criteria for alcoholism include not able to stick to drinking limits, not able to cut down or stop drinking, spent a significa nt amount of time drinking, continued drinking despite problems, spent less time on other matters, sho wn an increase in tolerance, shown signs of withdrawal (Willenbring et al., 2009) . The more criteria met usually indicate the greater severity of dependence . However, recent evidence suggests the criteria of alcohol abuse and dependence often intertwine a mong patients diagnosed. Therefore, the most recent Diagnostic and Statistical Manual of Mental Disorders 5 th edition ( DSM V ) integrate s these two disorders into one single alcohol use disorder with mild, moderate, and se vere sub classifications (American Psychiatric Association, 2013) . These definitions for alcohol use (hazardous drinking, AUD) used in the g eneral population have also been widely used in studies among HIV infected population.
15 In summary, hazardous drinking ( sometimes called at risk drinking or unhealthy drinking ) falls in the continuum of alcohol use between mode rate/light drinking and AUD . Hazardous drinkers are often identified only by their drinking levels (if levels exceeding weekly or daily limits of alcohol consumption ). Therefore, if no formal diagnosis is conducted, those with AUD who drink heavily often fall in the group of hazardous drinker . During the primary care encounters, w hen a high level of alc ohol consumption is reported, screen tool, such as Alcohol Use Disorder Identification Test (AUDIT) should be performed to differentiate levels of risk and seve rity. Those scoring in the high range (20 40) on the AUDIT should receive DSM based diag nosis assessment and, depending on severity of physical dependence, detoxification and other treat ment options . Even no DSM criteria is met, hazardous drinking represen ts a higher risk than the moderate/light/non drinkers for developing future alcohol dependence or other negative consequences. Therefore, interventions, such as brief intervention, are often recommended to motivate hazardous drinkers to cut down intake or abstain from drinking (Room et al., 2005) . P revalence of Alcohol Use in HIV positive Individuals According to the 201 3 Substance Abuse and Mental Health Services Administration (SAMHSA) data, 7 1 % of US adults reported alcohol use in the past year with more than half reported being current drinker (Substance Abuse and Mental Health Services Administration, 2013) . Men are more likely than women to be current drinkers and to have heavy drinking . Previous studies have suggested that 8% 44% of HIV positive individuals meet the criteria of hazardous drinking in different set tings, depending on the different measure s and different settings (Chander, 2011; Cook et al., 2009; Galvan et al., 2002; Rothlind et al., 2005; Stein et al., 2005; Tucker et al., 2003) .
16 Using data from the HIV cost and services utilization study ( HCSUS) , Galvan and colleagues found that more than half of HIV positive individuals in care reported any alcohol use in the preceding month. The prevalence of heavy drinking was almost twice that reported in the general population (Galvan et al., 2002) . In addition, they fo und men were more likely than women to use any alcohol ; factors associated with heavy drinking were drug use, less education (< high school vs. college) , and having no history of AID S defining illness . Based on a sample of 1957 HIV patients attending Unive rsity based HIV clinics, half reported using alcohol in the past 6 months , and 11% were identified as hazardous drinkers (Chander et al., 2006) . A nother study found that 19% of 219 HIV patients attending a HIV primary care clinic reported engaging in problem drinking (hazardous, heavy, or binge drinking), and 33% re ported mild to moderate amounts of alcohol consumption (Cook et al., 2001) . Among HIV positive subjects enrolled from the VA medical center, Rothlind and colleagues reported 44% me e t ing heavy drinking criteria (Rothlind et a l., 2005) . The use of alcohol in women with HIV infection has been extensively studied by Cook et al using a large multicenter cohort (n=2770) of HIV infected women (Cook et al., 2009) . Over the 11 year follow up period , half reported use of any alcohol in the past year with 14% 2 4% reported consumption at hazardous levels at any point of time over the study period. This study also suggested that being unemployed, having CD4 cell count of 200 500 cells/ml, HCV seropositive , and having high levels of depressive symptoms were associa ted with being more likely to hav e hazardous drinking . Drinking a nd Health of HIV positive Patients Although early studies conducted before the era of HAART did not find significant relationship between alc ohol consumption and HIV disease progression (Kaslow et al.,
17 1989) , many recent studies have demonstrated the significant association betwee n hazardous drinking and many negative health outcomes and behaviors in people with HIV infection. First, HIV infected people who have h azardous alcohol consumption were less likely to engage in HIV care (Chander et al., 2006) and to stay in care compared with ligh t or non drinkers (Cunningham et al., 2006) impression that heavy drinkers are unable to comply closely with strict regimen requirement (Spire et al., 2007) . Timely linkage to HIV outpatient care, prompt initiation of ART and prophylaxis for opportunistic infections when indicated, and subsequent adherence to prescribed medications are key foundation s for successful HIV disease management (Ulett et al., 2009) . The findings of negative relationship between heavy alcohol use and linkage to HIV clinical care provide one plausible explanation regarding why alcohol users are less likely to achieve undetectable viral load . Second, viral hepatitis (hepatitis B and hepatitis C) are common among individuals with HI V infection because they share similar transmission routes . In the US and w estern European, hepatitis C virus (HCV) infection has been found in 25 30% of HIV positive persons overall; 72 95% of injection drug user s; 1 12% of MSM ; 9 29% of heterosexuals (Alter, 2006) . HCV is also common among women with HIV infection. Among HIV infected pregnant wome n, the prevalence of HCV infection ranged from 17% to 54% (Thomas et al., 1998) . In the HAART era, ev idence has suggested that co infected HC V has influence on HIV disease progression. In the Swiss HIV cohort study, individuals co infected with HIV and HCV had faster progression to AIDS and slower increase in CD4 compared with those infected with HIV alon e (Greub et al., 2000) . This
18 finding is further supported by a meta analysis of CD4 cell counts after the initiation of HA ART comparing patients w ith HIV and HCV co infection with those with HIV infection alone (Miller et al., 2005 ) . HIV infection also affects HCV and HBV related liver diseases. In patients co infected with HIV and HCV or HBV alone, fibrosis rates are accelerated compared with those with HCV or HBV alone, leading to faster progression to the end stage of liver di seases (Benhamou et al., 1999) . Development of cirrhosis has been found 12 16 years earlier in patients co infected with HCV and HIV than in those infected with HCV alone (Soto et al., 1997) . Survival is poor among HCV infected patients w ith decompensated cirrhosis who are co infected with HIV ( estimated median survival of only 13 months ) (Pineda et al., 20 05) . Excessive alcohol consumption damage s liver tissues, causing alcoholic liver diseases (e.g, fatty liver, alcoholic hepatitis, and chronic hepatitis), and exacerbates e xisting viral hepatitis among HIV positive patients . In a study of the causes of deaths in patients with HIV infection, excessive alcohol consumption was the most common comorbidity, and was twice as high in those who died of liver diseases than in those who died of other causes (59% vs 24%) (Salmon Ceron et al., 2005) . Among women living with HIV/AIDS, HCV related liver disease has emerged as one of the most common cause of non AIDS related deaths (Cohen et al., 20 02) . Third, the fact that alcohol use is a risk factor for optimal adherence to antiretroviral medications has been consistently demonstrated in many studies using various samples of HIV infected individuals. Among 212 HIV positive persons attending two urban HIV clinics, researcher found that problem drinkers were more likely to take HIV medications off schedule during the previous week (45% vs 26%) ; problem drinkers
19 were more likely to report missing medication because of forgetti ng (48% vs 35%) , running out of medications (15% vs 8%) , and consuming alcohol or drugs (26% vs 3%) (Cook et al., 2001) . Analysis on a group of HIV positive i ndividuals who had a history of drinking problems found that poor adherence to prescribed antiretroviral medication was associated with any alcohol use in the past month compared with abstainer from alcohol (Samet et al., 2004) . This result suggests there may be no safe threshold for drinking in terms of antiretroviral adherence. Consis tent results were found in another study with 1711 HIV positive persons attending Uni versity based HIV clinics. The authors found that hazardous drinking w as significa ntly associated with decreased adherence to antiretroviral therapy compa red with non drinkers; hazardous drinkers reported had a lower odds of adherence than moderate drinkers, suggesting a possible dose response relationship between alcohol use and adherence to HIV medication (Chander, 2011) . Data from the Veteran Aging Cohort Study (VACS) suggested a possible temporal association between alcohol use and adherence: alcohol use on a specific day was s ignificantly associated with declined adherence on that day and two following days (S. Braithwaite et al., 2005) . A recent meta analysis based on 40 studies totaling 25 , 000 participants found that those who used alcohol or drank relatively more were more likely to be defined as being non adherent (Hendershot et al., 2009) . In fact, due to the concern of potential interplay of alcohol and ART, it appears some patient s intentionally skip their medications for drinking rather than limit their drinking for better adherence (Kalichman et al., 2013) . Fourth, hazardous drinking alters the immune system regulation and may directly result in rapid HIV disease progression. Studies based on animal models have
20 suggested the biologically deleterious effect of heavy alcohol consumption on the disease progression (Bagby et al., 2006) . Despite the documented relationship between alcohol use and poor antiretroviral adherence, the direct as sociation between alcohol use and HIV viral load and CD4 count has only been established for hazardous levels of alcohol consumption . Based on a large cohort of HIV positive patients in care, Chander et al suggested that hazardous drinking was associated w ith lower odds of achieving viral suppression compared to non drinking (AOR=0.72, 95% CI: 0.57 0.99), but null association was found between moderate drinking and achieving viral suppression (AOR=1 .00 , 95% CI: 0.84 1.2 0 ). A recent study using participant s enrolled from HIV treat ment settings found daily drinkers were associated with an approximately 4 fold increase in the odds of detectable viral load compared with reminders of the HAART users (AOR=3.81, 95% CI:=1.42 11.48), while inconclusive results w ere found among those not on HAART (Wu et al., 2011) . A prospective cohort study of 231 HIV positive patients added more evidence supporting deleterious effect of hazardous drinking on HIV disease progression . Researchers found that frequent alcohol users were 2.91 times more likely to present a decline of CD4 to <=200 cells/uL after controlling for antiretroviral use over time, suggesting the direct effect of alcohol on CD4 seemed to be independent of ART (Baum et al., 2010) . S tudies on such association are inconclusive for women with HIV infection . An analysis on 516 women in the HIV epidemiological Research Study (HERS) found insignificant association between alcohol use and CD4 cell count. Another large study of 1686 HIV there was no independent positive association between heavy alcohol use and time to
21 newly diagnosed AIDS defining illness or time to AIDS related death (Cook et al., 200 8) . However, a recent study using same cohort found heavy alcohol consumption was independently associated with increased risk of mortality (Neblett et al., 2011) Finally, alcohol consumption in individuals with HIV infection can be associated with a variety of adverse social, legal, occupational, psychological, and non AIDS related medical conditi ons (Friedmann, 2013) . For instance, alcohol alters some brain receptors and neurotransmitters, and several pharmacological effects of alcohol are likely to increase aggressive behavior, resulting in increased risk of violen t crime (Heather et al., 2001) . In summary, hazardous drinking represents an important healt h issue in HIV infected individuals in the post HAART era . Improved strategy to screen and treat hazardous drinking ma y have significantly beneficial implications for health and well being of HIV infected individuals . Alcohol T reatment A wide range of treatment options are available to a ddress hazardous drinking with different levels of risk and severity . The se treatments can be generally divided into three categories: brief intervention, specialized treatment programs, and mutual help groups. Brief interventions are int ended to provide prophylactic care before or soon after the onset of drinking problems. They are designed to motivate hazardous drinkers to moderate alcohol consumption, rather than to promote complete abstinence. They are often simply enough to be provide d by primary care physicians and are specially appropriate for hazardous drinking meeting some criteria for abuse rather than dependence (Rubinsky et al., 2013) . The cumul ative evid ence has suggested significant clinical effect on drinking behavior and related problems from receiving brief
22 intervention, while there is little evidence that brief interventions are equally effective for individuals with alcohol dependence . Specialized treatment refers to inte rventions directly managing alcohol withdrawal, preventing relapse to heavy drinking , and supporting social and psychological rehabilitation of the problem drinkers. Specialized treatment services consist of two components: programmatic settings (e.g., det oxification facilities, inpatient residential programs, outpatient clinics) and therapeutic approaches ( e.g., 12 step of Alcoholics Anonymous, behavioral therapies , and pharmacotherapies ) . Detoxification with the management of alcohol withdrawal is usually the first step for many alcohol dependent patients to relieve discomfort and to get prepared for rehabilitation. Non pharmacological detoxification can be done in outpatient setting and is appropriate for patients with mild to moderate dependence (Naranjo and Selle rs, 1986) . Pharmacological detoxification, often done in inpatient settings, is indicated for patients with serious medical illness, a past history of adverse withdrawal reactions, and current evidence of delirium tremens. Alcohol rehabilitation has typ ically been provided in residential settings (e.g. halfway house) lasting for a period of one month or more. Due to increasing cost concerns, outpatient management has become the dominant setting. Therapeutic approaches that are often employed in both in patient and outpatient settings include psychosocial treatments (e.g., cognitive behavior therapy [CBT], motivational interviewing [MI] , 12 step facilitation, family therapy ) , and pharmacotherapies . CBT focus es on teaching the relapse prevention skills and development of cognitive strategies to avoid clues to drinking. Twelve step facilitation is
23 designed to introduce patients to the principles of AA. MI is employed to increase three therapies had similar effectiveness , and more effective than confrontational and family therapies (Group, 1998; Heather et al., 2001; Ouimette et al., 1999) . The United States Preventive Services Task Force (USPST) recommends pharmacotherapies with concomitant psy chosocial therapies for treatment of patients with alcohol dependence (Willenbring et al., 2009) . Over the past decade, pharmacotherapies have begun to play a more important part in the treatment and rehabilitation of people with alcohol dependence. Disulfiram, naltrexone, acamprosate are three medications approved by the US Food and Drug Ad ministrati on (FDA) for treating alcohol dependence. Disulfiram is an alcohol aversion drug that causes discomfort when taking alcohol simultaneously . The few clinical trials on disfulfram indicated little evidence of its effectiveness because of poor medication comp liance (Fuller et al., 1986) . Naltrexone is an opioid antagonist to block endogenous opioids triggered by alcohol (Mason et al., 2002) . Therefore, naltrexone is hypothesized to reduce craving for alcohol and to help prevent relapse to heavy drinking by reducing the rewarding effect from alcohol consumption if drinking does occur (Ray et al., 2010) . Acamprosate is thought to promote abstinence by restoring the balance in t he gamma aminobutyric acid (GABA) and glutamate systems that is disrupted by chronic heavy drinking (RÃ¶sner et al., 2010) . A recent Meta analysis comparing naltrexone and acamprosate suggested that acamprosate was more efficacious in promoting abstinence while naltrexone was more efficacious in reducing heavy drinking and craving. Detoxif ication and a longer period of abstinence before
24 medication initiation may suggest larger medication effect s for acamprosate and naltrexone, respectively (Maisel et al., 2013) . Prior observational and clinical trials in the general population have consistently demonstrated the participation in general alcohol treatment program is associated with improved outcomes in individuals with drinking problems (Dawson et al., 2006 ; Hasin et al., 2013; Moos and Moos, 2006, 2004; Timko et al., 2006; Ye and Kaskutas, 2009) . For example, b ased on 4422 individuals with prior to past year (PPY) onset of alcohol dependence obtained from the 2001 2002 NESARC data, Dawson et al demonstra ted that participation in general alcohol treatment program result ed in increased likelihood of any recovery, non abstinent recovery, abstinent recovery. In particular, individuals who participated in 12 step program in a ddition to formal treatment had bet ter recovery compared with those who r eceived formal treatment alone (Dawson et al., 2006) . Unlike ample evidence of the negative influence of hazardous drinking on many aspects of the life of people with HIV infection, i nformation is unexpectedly li mited for alcohol treatment among HIV patients who have drinking problems. Several studies examining the efficacy of alcohol treatment produced mixed results (Hasin et al., 2013; Samet et al., 2005; Velasquez et al., 2009) . A st udy of HIV infected men who had sex with men found that alcohol intervention resulted in reduction in both the number of drinking days and the number of heavy drinking days in the past 30 days over a 12 month follow up (Velasquez et al., 2009) . A recent s tudy based on 258 HIV infected individuals with recent hazardous drinking suggested alcohol treatment, such a s motivational interviewing had effect on reducing the number of drinks on drink days but such effects seemed to be modest and was dependent on the presence of alcohol
25 dependence (Hasin et al., 2013) . However, b ased on a sample of 151 HIV primary care patients with lifetime h istory of AUD, researchers found no significant alcohol treatment effect on reduction in drinking at both 6 month and 12 month follow up (Samet et al., 2005) . In addition, currently , there is no data on any pharmacotherapy for alcohol dependence in patients with HIV infection, altho ugh a number of trials are on going (Parry et al., 2014) . Given the profound negative impacts of hazardous alcohol consumption on the course of HIV, there is increasing interest in evaluating the long term benefits of alcohol treatments on HIV infected individuals. In addition, alcohol treatment should never be considered as a one time fix because most chronic drinkers, especially those who experience recurrent relapse, may require recurring stay i n treatment programs to eventually achieve sobriety. Disease simulation model may provide us a useful tool to project long term treatment effect in a timely and cost efficient way. HIV Disease Simulation Model Disease simulation modeling is one technique that allows long term health assessments based on the extrapolation of existing data (Briggs et al., 2006 ) . It can be easily expanded to address cost effectiveness questions if cost components are included. Because it is basically a computer based mathematical model, it is much more expedient and resource efficient compared to conducting longitudinal cohor t studies or lengthy clinical studies. Dependent on known clinical data, disease simulation model makes reasonable projections on long term outcomes that provide important evidence to support clinical decision making and public health policy making for therapeutic options for the management of many chronic conditions, like HIV (Drummond, 2005) .
26 In the past two decades, several influential studies employing HIV disease models have been published to inform vital guidelines on HIV disease management, such as cost effectiveness of HIV screening and the timing of initiating HAART (Braithwaite et al., 2005; Farnham et al., 2012; Freedberg et al., 2001; Pal tiel et al., 2005; Prabhu et al., 2011; Sanders et al., 2005; Walensky et al., 2013) . In 2001, Freedberg et al. published a mathematical model in NEJM to firstly evaluate the cost effectiveness of HAART (Freedberg et al., 2001) . The input parameters were derived from publications of big clinical trials from various countries, such as the AIDS clinical trial gro up 320 study, Johns Hopkins HIV clinic cohort study, and the Italy, Netherlands, Canada, and Australia (INCAS) trial. This model simulated 1 million hypothetical HIV patients and extrapolated the expected life years, development of opportunistic infections , and overall treatment costs. The results suggested that although the financial costs were significant, HAART could be generally considered as cost effective, at least compared with other therapeutic strategies generally used at that time . This finding co ntributed to the widespread adoption of HAART. In 2005, Sanders et al. reported a new disease simulation model in NEJM to demonstrate the cost effectiveness of HIV screening in the era of HAART (Sanders et al., 2005) . This model applied refined Markov chain model using Monte Carlo simulation (to gen erate random number) to account for real world uncertainties. The authors also incorporated a disease transmission model to measure additional benefits of transmission prevention caused by early detection and treatment . Solely focusing on health outcomes, one model developed by Braithwaite et al. made an attempt to quantify non AIDS related mortality in aging HIV patients (Braithwaite et al., 2005) . Coming into 2010, several new models
27 evolved to address new research challenges as an extension of previous models (Farnham et al., 2012; Prabhu et al., 2011) . For example, CDC published a HIV/AIDS model PATH (progression and transmission of HIV/AIDS), that investigated the cost effectiveness of HIV testing at diffe rent settings, like sexually transmitted diseases ( STD ) clinics, emergency room , and Inpatient (Prabhu et al., 2011) . Conclusion Epidemiological Research on alcohol use and HIV diseases has made significant progress in the past two decades on describing disease burden s and their complex etiological relationship . However, there a re still many areas are lacking. One example is the lack of understanding on alcohol treatment use among HIV infected hazardous drinkers. Alcohol treatment has been traditionally underutilized among people with drinking problems (Cohen et al., 2007) . Whether similar levels of utilization will be seen in HIV infected individuals is unknown. Furthermore, information on the profile of those who receive treatment improve current effort s to enhance treatment utilization if the utilization rate is found low . Similarly, better understanding of treatment effect is important because many people in need of treatment receive treatment due to the lack of sufficient knowledge of treatment and treatment effect (Dawson et al., 2012) . The rest of the dissertation will descri be efforts we have undertaken to make progress in this area. We will focus on 3 major aspects about alcohol treatment use in HIV infected individuals: utilization, treatment effect on drinking outcomes, and potential influence of treatment on survival.
28 CHAPTER 2 SPECIFIC AIMS AND HYPOTHESE S This project consists of three studies (presented in C hapter 3 5) to improve the understanding of alcohol treatment use among HIV infected individuals with hazardous drinking. Three specific aims and corresponding hypothesis are presented below. Aim1 Utilization of Alcohol Treatment Given the evidence that hazardous drinking is an important health problem in HIV infected individuals , provision of alcohol treatment services may result in substantial benefits in their health and well being. However, little is known about the alcohol treat ment use in HIV infect ed individuals , especially women with HIV infection. Using data from a large cohort of HIV infected women, t he objectives of this aim were to (1) describe utilization of alcohol treatment, and to (2) determine factors associated with utilization of any alc ohol treatment among HIV infected women with hazardous drinking. Hypothesis : We hypothesize that the util ization of alcohol treatment is low; we also hypothesize t hat enabling factor s and need factors may play important role s in determining the utilization of alcohol treatment. Aim2 Treatment Effect on Drinking Outcomes Information is quite limited on alcohol treatment effect among HIV infected people. A few studies examining the alcohol treatment effect in HIV patients have been limited by the small sampl e size or recruitment from a single setting. No study has ever examined the alcohol treatment effect with data from a large , multi center observational cohort of HIV patients. Using multicenter cohort data may improve the generalizability; using data from longitudinal cohort not originally designed to assess alcohol treatment
29 effect may also help reduce strong placebo effect which often occurs in clinical trials of alcohol treatment (Litten et al., 2013) . Moreover, almost all previous research has been conducted based on either men exclusive or men dominant population (Hasin et al., 2013; Samet et al., 2005) , yielding little data on women with HIV infection. The aim of the present study was to evaluate whether engaging in any alcohol treatment, irrespective of either formal treatment (e.g., detoxification, inpatient wards, outpatient specialty care), or informal treatment (e.g., Alcoholics Anonymous or other 12 step programs, halfway house s, family services agencies), was associated with less unhealthy drinking and more abstinence among female hazardous drinkers with HIV infection . Hypothesis: we hypothesize that receipt of alcohol treatment will be associated with improved drinking outcomes, such as no unhealthy drinking and abstinence , among H IV infected women . Aim3 Survival Impact of Hazardous Drinking and Alcohol Treatment Although AIDS related conditions remain the most common causes of deaths ,the aging population of HIV infected individuals is facing increasing risks of complications and mortality not direc tly attributable to HIV (Neblett et al., 2011; Samji et al., 2013; Smith et al., 2014; Wada et al., 2014; Weber et al., 2013) . Hazard ous alcohol consumption is a modifiable risky behavior associated with a variety of adverse health outcome and behaviors. All these associations may have important implication s for the health and even sur vival of HIV patients . Few studies have examined the relationship between hazardous drinking and survival in the HIV infected population. The first goal of this aim was to develop a computer simulation model to quantify the impact of hazardous drinking on years of survival among individuals with HIV
30 infecti on. This simulation differed from previously published work because it specifically modeled the dynamics of hazardous drinking behavior and effect of hazardous drinking on both HIV related and non HIV related deaths. Hypothesis : we hypothesize that hazardo us drinking is associated with substantial reduction in the survival among HIV infected individuals. Given the high prevalence of hazardous alcohol consumption, treatment to reduce alcohol use may have the potential to prolong life of HIV infected hazardou s drinking. Pr imary data analyses lack sufficient power to directly investigate this relationship in the current treatment era because deaths have become increasing ly rare with modern therapies, and few studies have longitudinal data the utilization of alc ohol treatment. The second goal of this aim was to expand the hazardous drinking simulation model to estimate the sur vival impact of alcohol treatment utilization among HIV infected individuals with hazardous drinking . Hypothesis : utilization of alcohol tr eatment may result in improvement in survival among HIV infected individuals with hazardous drinking.
31 CHAPTER 3 UTILIZATION OF ALCOHOL TREATMENT AMONG HIV POSITIVE WOMEN WITH HAZARDOUS DRINKING Introduction Alcohol use is common among women with HIV infection in the United States. Approximately half of HIV positive women reported any alcohol consumption in the past year with 14% 24% drinking at hazardous levels (Cook et al., 2009) . Recent studies have shown that hazardous drinking, often defined as >7 drinks per week or >3 drinks per occasion for women (Willenbring et al., 2009) , is associated with several adverse health behaviors and outcomes in HIV positive wome n. These include low engagement and retention in HIV care (Chander et al., 2006; Cunningham et al., 2006) , suboptimal adherence to HIV medication (S. Braithwaite et al., 2005; Chander et al., 2006; Cook et al., 2001; Hendershot et al., 2009) , increased HIV viral load (Chander et al., 2006; Samet et al., 2005) , increased risky sexual behaviors (Hutton et al., 2012) , and more rapid HIV disease progression (Wu et al., 2011) . In addition, heavy alcohol consumption impo ses greater risk for cirrhosis and hepatocellular carcinoma and worsens other liver conditions among HIV positiv e women with HCV co infection (Cainelli et al., 2001; Cohen et al., 2002; Joshi et al., 2011) . A variety of treatment options are available to address a wide spectrum of alcohol problems ranging from risky drinking to chronic alcohol dependence (Willenbring et al., 2009) . Prior studies in the general population have consistently found that participation in alcohol treatment programs, (e.g., detoxification, Alcoholics Anonymous or other 12 step programs, halfway houses) improves out comes in peo ple with AUD (Dawson et al., 2006; Hasin et al., 2013; Moos and Moos, 2006, 2004; Timko et al., 2006; Ye and Kaskutas, 2 009) . Nevertheless, community studies of treatment patterns
32 among non HIV infected population have repeatedly demonstrated substantial underutilization of alcohol treatment (Cohen et al., 2007; Harris an d Edlund, 2005; Mojtabai, 2005; Wu et al., 2003) . (Andersen, 2008) has been widely used to help better understand access and utilization of health care services (Babitsch et al., 2012; Dawson et al., 2012; Goldstein et al., 2010; Mojtabai, 2005) .This model classifies potential individual an d contextual determinants into 3 domains: predisposition affecting service use, such as demographics, social structure, and health beliefs. Enabling factors refer to vari ous characteristics influencing care delivery, such as financial or societal sources for care, availability of and access to care providers. Need factors include health status assessed by health professionals or perceived by patients themselves (e.g., seve rity of drinking problems, comorbid health conditions, and co existing mental health and substance abuse problems). Hazardous drinking is an important problem in HIV infected women, and provision of alcohol treatment services may result in substantial ben efits in their health and wellbeing. However, little is known about the utilization of alcohol treatment among women affected by both HIV infection and hazardous drinking. The objectives of this study were to describe utilization and, under the framework o Behavioral model, to determine factors associated with utilization of any alcohol treatment among HIV positive women with hazardous drinking.
33 Method Study Sample larg est, observational prospective cohort study of HIV disease in U.S. women (Bacon et al., 2005) . WIHS recruited HIV posit ive women from 6 study sites across the US: Brooklyn, New York; Bronx, New York; Chicago, Illinois; Los Angeles, California; San Francisco, California; and Washington, District of Columbia. To better reflect real world experience of HIV disease and managem ent, women were recruited via various venues, including HIV primary care clinics, HIV testing sites, hospital based programs, community outreach sites, and research programs (Barkan et al., 1998) . The questions assessing alcohol treatment utilization were implemented from 1994 to 2002 . The present study was a cross sectional study, in that information from each participant was based on a single visit in which she first reported hazardous drinking during this period. Written informed consent was obtained for all women participating in W IHS and the institutional review board of the University of Florida approved this study focusing on the existing WIHS data. Measurements Hazardous drinking Alcohol use questions assessed frequency and quantity, and weekly alcohol consumption was calculat ed to define hazardous drinking. The present study defined hazardous drinking as >=12 drinks per week (Gordon et al., 2001; Parsons et al., 2007) . In the absence of a formal AUD diagnosis in WIHS, using this relatively higher criterion than the NIAAA threshold (>7drinks per week or >3 drinks per day for women)
34 enhanced the specificity to better identify women who experienced alcohol related negative consequences and hence, were in need of alcohol treatment. Alcohol treatment use F ollowing the alcohol consumption questions, recent use (in the past 6 months) \ since your last visit, have you been in an alcohol treatment program? I am interested in any alcohol t reatment programs you may have been in, including inpatient and/or outpatient alcohol detoxification, halfway houses, Alcoholics Anonymous, and/or other alcohol treatment acknowledg ed the uptake, inquiring about specific sources of treatment. Because many women reported utilization of alcohol treatment from multiple resources in the past half year, we constructed a hierarchy of two mutually exclusive types of care (Dawson et al., 2012) : utilization of formal sources of care (inpatient detox or outpatient alcohol trea tment), irrespective of whether any informal sources of care were also used; utilization of informal sources of care (AA, halfway houses, etc) only. Predisposing factors Predisposing characteristics affecting care seeking behavior included age (18 35, 36 42, 43+), race/ethnicity (Hispanic any race, black non Hispanic, and White non Hispanic/other races), educational attainment (<=high school vs. >=college), employment stat us, and marital status (married/cohabitating vs. single/never married). Enabling factors Factors facilitating or acting as barriers to the utilization of alcohol treatment included family income (<=12K vs. >12K), health insurance coverage (yes/no), and perceived social support (yes/no). Strong support and aid from family members, friends
35 and social connections have been associated with greater likelihood of entry into treatment (Grant, 1997; Mowbray, 2014) compared with those who lack such support. Social support factors were determined by positive responses to all three dummy variables in the social support domain: whether receiving any help (yes/no); whether receiving encouragement from family members and/or friends (yes/no); whether family visit (yes/no). Need factors Need factors included the amount of weekly alcohol intake categorized into 3 levels: 12 15, 16 35, and 36+ drinks; presence of depressive symptoms measured as a score of the Center for Epidemiological Studies Depression S cale (CES D) equal or greater than 16 (Tandon et al., 2012) ; use of any illicit drugs defined as using the past 6 months. Overall healt h status was measured by a shortened version of the Medical Outcome Study (MOS) HIV health survey scoring from 0 to 100 (Liu et al., 2006) . HCV status was determined by antibody testing from blood collected at WIHS enrollment visit. Undetected HIV viral load was defined as a plasma HIV RNA viral load that was below the lower limit of detection of the quantitative HIV RNA assay. Statistical A nalysis Descriptive analyses were performed to describe recent use (in the past 6 months) of any alcohol treatment, specific types of treatment, as well as potential predictors. We performed unadjusted logistic regression on each predictor to determine the bivaria te association. Any variable statistically significant at an of 0.05 from the unadjusted analyses or predefined as an important predictor based on prior literature was retained in the final multivariable logistic regression to examine the independent
36 asso ciation with the odds of receiving any alcohol treatment. A penalized maximum likelihood estimation (PMLE), also called Firth bias correction method, was used to reduce small sample bias (King and Zeng, 2001) . In multivariable models requiring complete information for all included variables, even small amounts of missing data for each of the predictors can result in sizable sample reduction. Although the missing data for eac h variable were no more than 6%, the total amount of missing data in the final complete case model accounted for over 17% of the whole sample. To remediate this, we used mu ltiple imputation with a multivariate normal approach (Schafer, 1999) . We generated 5 copies of complete data sets with missing values imputed by random dr aws from the multiple imputation models. Multivariable logistic regression analysis was performed independently on each of the 5 imputed data sets. The resulting estimates for each variable were averaged to give a mean estimate and adjusted standard error (Rub in, 1987) . Because results were similar to those from complete cases analysis, we only reported adjusted associations from the imputed data. All analyses were conducted by SAS st atistical package, version 9.4. Results From 1994 to 2002, this study ident ified a total of 474 HIV positive women who reported meeting hazardous drinking criteria (>=12 weekly drinks) at least once during the study period. As shown in T able 3 1, 68.8% of the participants were 36 years of age or older. The majority of the women w ere Black non Hispanic (68.6%), had high school or less education (76.8%), and were single or never married (69.8%). Most women reported being unemployed (85.6%), having an average household income of no more than $12,000 (73.2%), having any kind of health insurance (75.7%), and around 60%
37 reported positive responses to all 3 social support questions. A median of 21 drinks per week was reported with approximately one fourth (23.4%) of the sample drinking at least 36 drinks per week over the past 6 months. D epressive symptoms (CES D>=16) (67.2%) and recent illicit drug use (71.9%) were common. More than half of this sample was HIV/HCV co infected. As shown in T able 3 2, among HIV positive women with a recent episode of hazardous drinking, 19% (n=90) reported recent alcohol treatment utilization. More specifically, 13.7% used at least one formal source of care and 5.3% used only informal sources of care. The most commonly reported type of treatment was AA (12.9%), followed by inpatient detoxification (9.9%) and outpatient alcohol treatment program (7.0%). Half (51%) of women who received any alcohol treatment reported using more than one type of alcohol treatment. Among women who had used at least one formal source of care, 61.5% also attended AA and 20% obtaine d care from other sources. Bivariate analyses indicated that those who used alcohol treatment have 2.79 times the odds to report a lower household income (Odds Ratio [OR] = 2.79, 95% Confidence Interval [CI] = 1.43 5.48), and being unemployed (OR=2.70, 95% CI=1.13 6.46) than those who did not use any alcohol treatment (Table 3 3). A significant dose response relationship was found between increasing weekly drinking amounts and odds of utilization of alcohol treatment: 16 35 drinks/week vs. 12 15 drinks /week, (OR=2.79, 95% CI=1.13 6.46); 36+ drinks/week vs. 12 15 drinks/week, (OR=4.54, 95% CI=2.11 9.76). Those who used alcohol treatment had 3 times the odds of acknowledging illicit drug use in the last 6 months (OR=3.00, 95% CI=1.58 5.72) than tho se who did not use any alcohol treatment. No significant association was found
38 between treatment utilization and other variables, including all predisposing factors, health insurance status, presence of depressive symptoms, HCV status, undetected HIV viral load, time of the 1 st hazardous drinking visit, and locations of WIHS site. Results from the multivariable model suggested those with household income <=$12,000 had greater odds of utilization of any alcohol treatment than those with higher household inco me (Adjusted Odds Ratio [AOR] = 2.61, 95% CI=1.23 to 5.56). Although insignificant in bivariate analysis, having social support approached significance with greater odds of obtaining alcohol treatment (AOR=1.78, 95% CI=1.03 to 3.08) after controlling for o ther covariates. The dose response association between levels of alcohol consumption and odds of utilizing any alcohol treatment persisted after multivariable adjustment. Compared with those who reported weekly drinking of 12 15 drinks, women who drank 16 35 drinks per week had 2.79 times greater odds of receiving any alcohol treatment (AOR=2.79, 95% CI=1.36 to 5.72), and the AOR escalated to 4.16 when the levels of drinking increased to more than 35 drinks per week (AOR=4.16, 95% CI=1.92 to 9.05). Finally, illicit drug users had greater odds of reporting alcohol treatment utilization (AOR=2.74, 95% CI=1.42 to 5.29) than non drug users. No other significant association was found. Discussion Among HIV positive women in WIHS reporting hazardous drinking, one in five reported recent utilization of alcohol treatment; half received multiple types of treatments among which AA was the most common; a mixture of enabling (income and social support) and need factors (drinking levels and drug use) proofed important in explaining utilization of alcohol treatment.
39 The rate of alcohol treatment reported by our study was relatively high. For instance, data from the NESARC suggest that 14.6% of individuals with lifetime AUD report ever engaging in alcohol treatment (Cohen et al., 2007) . This discrepancy may reflect various approaches assessin g treatment utilization (e.g. lifetime vs. past 6 months) as well as substantial differences in sample characteristics WIHS women have more comorbidities, poverty, mental health conditions, and many engaged in HIV care. Although higher than rates reporte d in NESARC, our result was consistent with findings from other studies: 2002 National Survey on Drug Use and Health (DSDUH) found that among individuals with combined AUD or SUD, the prevalence of past year treatment utilization for substance use disorder s (SUD) was as high as 22.1% (Mojtabai, 2005) . Indeed, studies on low socioeconomic class women with substance use disorders also revealed that the rates of past year substance abuse treatment could be as high as 44.0% (Hansen et al., 2004; Rosen et al., 2004) . Mojtabai suggested that the high levels of utilization of alcohol or substance abuse treatments may suggest more severe issues of AUD or SUD (Mojtabai, 2005) . This study found no independent effect of age and race on the odds of alcohol treatment utilization among HIV infected women with drinking problems. Previous studies have yielded variable results on the association of age with treatment seeking. Some studies have found tha t older individuals were more likely to receive treatment (Cohen et al., 2007; Grella et al., 1999; Tighe and Saxe, 2006; Weisner et al., 2002) while others have found that younger people were more likely to utilize treatment services (Ilgen et al., 2011; S and Bowe, 2008) . The resulting age trend of the present study suggests younger women may be more likely to receive alcohol treatment.
40 However, the lack of statistical significance may be results of moderate age effect or relatively small sample size with limited statistical power. In terms of race/ethnicity, our finding is consistent with prior work that found no racial difference in receipt of alcohol treatment (Hatzenbuehler et al., 2008) . The finding of the negative association of income with alcohol treatment utilization was generally consistent with prior observations among non HIV infected individuals with AUD (Cohen et al., 2007; Ilgen et al., 2011; Mojtabai, 2005) . The exact reason for observing such negative association among HIV infected women is unclear although it can be related to greater access to public services and public entitlements among low income individuals ( Weisner, 1993) . Substance abuse treatments have traditionally been closely affiliated with the criminal justice and welfare system (Weisner and Schmidt, 1995) . Individuals from lower socio economic classes are usually over represented in the criminal justice and welfare system and thus, may be more likely to receive treatment for their alcohol and drug abuse problems. For example, the most common form of treatment reported in the present st udy was participation in Alcoholic Anonymous, which is available at no cost. Our finding highlights social support as a potentially important predictor determining alcohol treatment utilization in HIV infected women. The effect of social support has been rarely investigated in prior alcohol treatment utilization studies as an independent variable (Ilgen et al., 2011) . Although a full scale social support measure, like the 27 item social support questionn aire , was not utilized, we found greater odds of receivin g alcohol treatment among HIV infected women with greater social support. Emotional support obtained from families and friends may help HIV infected women
41 overcome significant social stigma associated with both alcohol abuse and HIV infection an d motivate them to seek care to address both conditions. Also, vital ancillary services family services, may further enhance care seeking behavior . Lack of adequate supporting system has contributed substantially to low rates of alcohol treatment uptake (Ilgen et al., 2011) . Future alcohol treatment programs focusing on HIV infected women may consider the integration of specific social support strategies to facilitate care access and qua lity of care. The positive relationship of treatment utilization with drinking levels is self evident: the more alcohol is used, the greater odds that individuals will seek help for their AUD. The presence of depressive symptoms appears to play no import ant role in alcohol treatment utilization among HIV infected women. Women with depression may lack the energy or desire to seek care for drinking problems, which may explain the direction of effect found in the analysis. On the other hand, depressed women may seek care for their depressive symptoms, in which they may also receive screening or advice about their drinking problems (Dawson et al., 2012) . Both mechanisms may counterbalance each other, resulting in our finding of no effect. Future research should assess whether receiving treatment for depression increases the likelihood of seeki ng alcohol treatment, particularly since depression and hazardous drinking frequently co occur among individuals with HIV infection. In multivariable analyses controlling for predisposing, enabling and levels of alcohol consumption, recent illicit drug us e predicted the utilization of alcohol treatment. This positive association is consistent with prior research that illicit drug use is positively
42 associated with care seeking behaviors (Cohen et al., 2007; Dawson et al., 2012; Hasin, 1994; Ilgen et al., 2011; Kessler RC et al. , 2001; Mojtabai et al., 2002) . Cohen and colleagues (2007) suggested that comorbid substance use disorders had strong influence on alcohol treatment utilization. In the current study, co occurring drug use was reported by more than two thirds of the sa mple and may result in greater distress, increasing their motivation to seek care for both problems . Increased outreach to those with both alcohol and drug use problems could help target the population with the most need for treatment. This study has seve ral limitations. First, our findings of predictors of utilization of alcohol treatment should not be interpreted as a causal link because of the cross sectional study design. Second, the data on alcohol treatment utilization were based on self report quest ions with unknown validity and reliability. Additionally, our measure of care did not allow us to distinguish participants with voluntary utilization from those with coerced utilization or to distinguish those who were unable to get care from those who nev er sought care. Third, the small number of respondents who reported recent treatment utilization precluded our analysis to identify meaningful factors associated with different types of alcohol treatment. Fourth, WIHS is a convenient sample but intended to be representative of women with HIV. WIHS women had semiannual visits Therefore, our results may not be generalizable to all HIV infected women with drinking problems. Fifth, although the formal diagnosis of AUD is lacking for our sample, quantity and frequency questions of alcohol consumption have been widely used to identify people with drinking problems (Canagasaby and Vinson, 2005; Chung et a l., 2012;
43 Dawson et al., 2005; Rubinsky et al., 2013; Willenbring et al., 2009) . In addition, we choose a higher cut off point to define hazardous drinking, resulting in higher specificity in identifying subjects in need of treatment. Sixth, mental heal th disorders were not assessed in this study with the exception of depression. Comorbid personality disorders and anxiety disorders may also play important roles in determining alcohol treatment utilization. Despite the above limitations, this is the fir st study of which we are aware to examine predictors of utilization of alcohol treatment among HIV positive women. Our findings provide a unique profile of HIV infected women with hazardous drinking who received alcohol treatments. This information is impo rtant to improve the treatment service delivery to this population. Many HIV infected women who do not receive care for their AUD can potentially benefit from engaging in alcohol treatment. Improved screening for at risk drinking, referral to appropriate t reatment, and integrating treatment procedures tailored to the unique contexts for HIV infected women (e.g. combined treatment services to address drug and alcohol dependence, the availability of auxiliary services) could potentially help lessen the gap be tween need for and the utilization of alcohol treatment.
44 Table 3 1 . Characteristics of 474 HIV positive WIHS women with recent hazardous drinking (in the past 6 months) Characteristics % Predisposing Age (years) 18 35 36 42 43+ 31.2 38.2 30.6 Race/ethnicity Black non Hispanic Hispanic any race White non Hispanic/other 68.6 16.7 14.8 Education <=High school >=College 76.8 23.2 Marital status Married/cohabitating Single/never married 30.2 69.8 Enabling Income <=12K >12K 73.2 26.8 Employed 14.4 Any health insurance 75.7 Social support 60.4 Need Number of drinks per week 36+ 16 35 12 15 23.4 50.8 25.7 Depression (CES D>=16) 67.2 Any drug use 71.9 Quality of life (mean, SD) 58.0 (20.2) HCV 59.8 Undetected HIV viral load 9.1
45 Table 3 2 . Recent alcohol treatment services utilization among HIV infected women with hazardous drinking Type of alcohol treatment services All (n=474) % Formal sources of care (n=65) % Informal sources of care only (n=25) % Any alcohol treatment service a 19.0 Formal sources of care (inpatient detox or outpatient alcohol treatment) b 13.7 100.0 0.0 Informal sources of care only (AA, halfway houses, etc) 5.3 0.0 100.0 Inpatient alcohol detoxification 9.9 72.3 c 0.0 Outpatient alcohol treatment program 7.0 50.8 0.0 Alcoholic Anonymous (AA) 12.9 61.5 84.0 Other (halfway houses, etc.) 4.2 20.0 28.0 a Any alcohol treatment used in the past 6 months includes uptake of any formal (inpatient detoxification, outpatient alcohol treatment, etc.) or informal sources of care for alcohol problems (Alcoholic Anonymous, halfway house, etc.). b Irrespective of whether also used any informal sources of care . c Do not add up to 100% because some women used multiple sour ces of treatment
46 Table 3 3 . Correlates of using any alcohol treatment among HIV positive WIHS women with hazardous drinking (n=474) Bivariate analyses Multivariable analyses Characteristics Treated (n=90) % Untreated (n=384) % OR 95% CI P value AOR 95% CI P value Predisposing Age (years) 18 35 36 42 43+ 21.6 19.9 15.2 78.4 80.1 84.8 1.54 1.39 1.00 0.85 2.81 0.78 2.48 0.157 0.270 1.66 1.44 1.00 0.88 3.13 0.78 2.65 0.117 0.242 Race/ethnicity Black non Hispanic Hispanic any race White non Hispanic/other 17.8 20.2 22.9 82.2 79.8 77.1 0.73 0.86 1.00 0.39 1.37 0.39 1.87 0.331 0.699 0.64 0.86 1.00 0.33 1.26 0.37 2.00 0.199 0.730 Education <=High school >=College 19.2 18.2 80.8 81.8 1.07 1.00 0.62 1.86 0.806 Marital status Married/cohabitating Single/never married 17.3 19.3 82.7 80.7 0.87 1.00 0.52 1.47 0.605 Enabling Income <=12K >12K 22.0 9.2 78.0 90.8 2.79 1.00 1.43 5.48 0.003 2.61 1.00 1.23 5.56 0.014 Employment No Yes 20.7 8.8 79.3 91.2 2.70 1.00 1.13 6.46 0.025 0.55 1.00 0.22 1.35 0.188 Any health insurance Yes No 18.3 21.0 81.7 79.0 0.84 1.00 0.50 1.42 0.508
47 Table 3 3 . Continued Bivariate analyses Multivariable analyses Characteristics Treated (n=90) % Untreated (n=384) % OR 95% CI P value AOR 95% CI P value Social support Yes No 21.1 15.2 78.9 84.8 1.49 1.00 0.90 2.46 0.123 1.78 1.00 1.03 3.08 0.039 Need Number of drinks per week 36+ 16 35 12 15 28.8 19.9 8.2 71.2 80.1 91.8 4.54 2.79 1.00 2.11 9.76 1.36 5.72 0.005 <0.001 4.16 2.96 1.00 1.92 9.05 1.44 6.09 <0.001 0.003 Depression (CES D>=16) Yes No 18.2 20.0 81.8 80.0 0.89 1.00 0.54 1.47 0.646 Any drug use Yes No 22.9 9.0 77.1 91.0 3.00 1.00 1.58 5.72 0.001 2.74 1.00 1.42 5.29 0.003 Quality of life (mean, SD) 57.4 (20.9) 58.2 (20.2) 0.732 HCV Yes No 20.4 17.0 79.6 83.0 1.25 1.00 0.77 2.01 0.368 Undetected HIV viral load Yes No 21.9 19.5 78.1 80.5 1.16 1.00 0.53 2.53 0.709 AOR: Adjusted Odds Ratio; 95% CI: 95% Confidence Interval
48 CHAPTER 4 RECEIVING ALCOHOL TREATMENT IS ASSOCIATED WITH LESS ALCOHOL USE AMONG HIV POSITIVE WOMEN WITH HAZARDOUS DRINKING Introduction In the United States, women account for one quarter of people living with HIV and one fifth of new diagnoses of HIV infection (Center for Disease Control and Pre vention, 2014) . Women of color are disproportionately affected by the HIV epidemic. Afri can American women who comprise 12% of femal e population in the US account for nearly two thirds (64%) of all new HIV infections in women (Centers for Disease Control and Prevention, 2013) . HIV epidemic has increasingly expanded to women, and the re is increasing ne ed to address modifiable behavior attributable to excessive comorbi dities and mortality among women with HIV infection (Neblett et al., 2011) . Hazardous alcohol use is one such modifiable behavioral factor. Alcohol use is common among HIV positive women with the prevalence o f hazardous drinking ranging from 14% to 24% (Cook et al., 2009) . Ha zardous drinking is defined as a pattern of alcohol use associated with increased risk of alcohol related and other health issues but may not be necessarily classified as alcohol use disorders (Willenbring et al., 2009) . Epidemiological r esearch among people with HIV infection has consistently suggested that hazardous drinking is related with decreased engagement to HIV care and retention in care (Chander et al., 2006; Cunn ingham et al., 2007) , suboptimal adherence to ART (S. Braithwaite et al., 2005; Chander et al., 2006; Hendershot et al., 2009) , and increased risk of cirrhosis and hepatocellular carcinoma a mong those with hepatitis B or C co infection (Cohen et al., 2002; Joshi et al., 2011) . In addition, hazardous drinking is linked to poor virologic and immunologic responses to ART (Wu et al., 2011) as well as
49 greater risk of engaging in unprotected sex , anal sex, and having multiple sex partners (Hutton et al., 2012) , which may place individuals at increased risk for transmitting HIV. T reatment to reduce alcohol use may re sult in substantial benefits to health and well being of HIV patients with hazardous drinking . However, information is limited on alcohol treatment effect in this population . Only a few studies have assessed alcohol treatment effect among HIV infected individuals and they have been generally limited by small sample size or enrollment from a single setting (Hasin et al., 2013; Parsons et al., 2007; Samet et al., 2005; Velasquez et al., 2009) . Moreover, these studies have only focused on either men exclusive or men dominant sample s (>75%) , yielding findings less applicable to women with HIV infection. To improve the understanding of alcohol treatment effect on HIV infected women, the present study used data obtained from a large, multicenter cohort of HIV infected women to assess the association between engaging in any alcohol treatment, irrespective of treatment modality, and drinking outcomes (no unhealthy drinking and abstinence) in the nex t 6 months. Using data from the multicenter cohort tend s to improve the g eneralizability of our findings. Furthermore, these data are generated from the study cohort not originally designed to asse ss alcohol treatment. This may that have been a serious issue in many clinical trials of alcohol treatment (Litten et al., 2013) . We hypothesized that receipt of alcohol treatment would be positively rela ted with achieving n o unhealthy drinking and abstinence among HIV positive women with hazardous drinking.
50 Method Study S ample (WIHS), the largest, ongoing, prospective cohort study of HIV disease in U.S. women. Women participating WIHS were recruited from 6 study sites across the US via vari ous recruitment settings, such as HIV primary care clinics, HIV testing sites, hospital based program, community outreach programs and research programs. More details of study design of WIHS can been seen elsewhere (Bacon et al., 2005; Barkan et al., 1998) . Every 6 months, women came to WIHS sites to complete a structured questionnaire, receive physical examinations, and p rovide blood and specimens for repositories and lab tests. Information was obtained biannually on sociodemograhpics, overall health status, antiretroviral therapies, drug and alcohol use, sexual behaviors, health care utilization, and psychological status. The current study used WIHS data from 1994 to 2002 because alcohol treatment utilization was only assessed during this period. Similar to the definition employed in Chapter 3, h azardous drinking was defined if the self report weekly drinking amounts exce eded 12 drinks per week (Gordon et al., 2001; Parsons et al., 2007) . In the absence of a formal AUD diagnosis in WIHS, using a relatively higher criterion than the NIAAA threshold (>7drinks per week or >3 drinks per day for women) enhanced the specificity to better identify women who experienced alcohol related negative consequences and who were more likely to need alcohol treatment. To assess if alcohol use declined after receiving treatment, we only targeted hazardous drin king visits (Figure 4 1) . We only included hazardous drinking visits with out any alcohol treatment reported in the past 6 months as the baseline ( t 0 ) to
51 ensure no carry on influence from any pre existing and /or concomitant alcohol treatmen t . We then obtained alcohol treatment status at the visit ( t 1 ) next to the hazardous drinking visit to ensure that treatment occur red after the presence of hazardous drinking. Finally, we assessed the drinking outcomes at the visits (t 2 ) next to the treatment visit t o ensure outcome occurred after the treatment was initiated. T h is design set up a condition treatment outcome scenario using the existing cohort data to provide clear temporality for our analysis of alcohol treatment effect. We also excluded observations in which women initiated alcohol treatment within the 6 months before t 2 . W ritten informed consent was obtained for all women participating in WIHS and the institutional review board of the University of Florida approved this analysis of existing WIHS data. Dependent Variables Dependent variables were t wo dichotomous drinking variables reported at t 2 : no unhealthy drinking in the past 6 months (yes/no) and achieving abstinence from alcohol in the past 6 months (yes/no). To demo nstrate meaningful improvement in drinking outcomes , w e used lower cu t offs to define unhealthy drinking: >7 drinks per week or >3 drinks per day (Willenbring et al., 2009) . No unhe althy drinking could include both moderate/light drinking and abstinence. Abstinence was classified if no alcohol use was reported in the past 6 months . Primary Independent Variables The primary independent variable of interest was engagement in any alcohol treatment during the 6 months prior to t 1 , irrespective of treatment modality. Positive response was based on a single question asked in WIHS: the past 6 months, have you been in an alcohol treatment program? I am inte rested in
52 any alcohol treatment programs you may have been in, including inpatient and/or outpatient alcohol detoxification, halfway houses, Alcoholics Anonymous, and/or other Other Covariates In addition to age and race, we i ncluded covariates associated with receipt of alcohol treatment to control for potential confounding. As shown in Chapter 3, these variables included income, social support, levels of alcohol consumption, and illicit drug use. Having social support was det ermined by positive responses to all three dummy variables: whether receiving any help (yes/no); whether receiving encouragement from family members and/or friends (yes/no); and whether family members and/or friends the last visit (yes/no). Use of any illicit drugs was methadone, and other drugs in the past 6 months. We also included HCV status for our analysis . Awareness of HCV is associated wi th reduction in drinking among individuals co infected with HIV, and thus may confound the assessment of alcohol treatment effect (Tsui et al., 2007) . HCV status was determined by testing antibody by second generation or third generation enzyme linked immunoassay (Ortho Diagnostic Systems, Rochester, New York, USA), testing HCV RNA by HCV branched DNA (Quantiplex 2.0 branched chain DNA enhanced label amplification assay; Bayer Ve rsant Diagnostics, formerly Chiron Diagnostics, Emeryville, California, USA) and by reverse transcriptase polymerase chain reaction (COBAS Amplicor HCV Detection Kit, Roche Diagnostic Systems, Pleasanton, California, USA).
53 Statistical Analysis We performe d multivariable longitudinal regression to examine the association of alcohol treatment (t 1 ) wi th no unhealthy drinking (t 2 ) . Because each eligible subject may con tribute multiple observations for our analysis, generalized estimation e quation (GEE) was fit to adjust for correlations between these observations over time . Huber White sandwich estimator was specified to provide robust standard error estimates (Ballinger, 2004; Fitzmaurice et al., 2012; Zeger et al., 1988) . For this model, we adjusted for a series of covariates at t 0 , including age, race, income, soci al support, levels of alcohol consumption, illicit drug use, and HCV status . Except race, all other covariates were time varying. Same multivariable longitudinal model was fit for abstinence as the outcome. We imputed missing drinking data at t 2 with the n ext visit data if available, and imputed missing data at t 0 with the prior visit data if available . After imputation, 5.8% of observations were missing for one covariate or more. All analyses were conducted using SAS statistical package, version 9.4 (Cary, North Carolina, USA). Results Through 1994 to 2002, we identified a total of 3 43 eligible women with hazardous drinking who contributed a total of 865 observations for analysis . Among these women, 39 . 7% had 1 observation, 40 . 4% had 2 to 5 observations, and 19.9 % had more than 5 observations . Table 4 1 shows the characteristics of these HIV infected women at their 1 st hazardous drinking visit: mean age was 39.1 years old (SD=7.3 years) ; t he majority of the sample was Black (69.4%) ; more than two thirds (70%) reported having an average household income of no more than $12,000; approximately 60% reported having social supported in the past 6 months; one fifth (18.4%) had at least 36 drinks per week during the past 6 months and the median level was 21 drink s per week (IQR:
54 15 31.5 drinks/week) ; 68.8% reported illicit drug use; more than half was co infected with HCV (59.6%) . Table 4 2 shows the findings from the multivariable longitudinal models. We found receiving any alcohol treatment w as independently ass ociated with having no unhealthy drinking in the next 6 months (o dds ratio [OR] =1. 58 , 95% confidence i nterval [ 95% CI] = 1.05 2 . 37 ) controlling for age, race, income, social support, levels of alcohol consumption, illicit drug use and HCV+. Similarly, w e found receiving any alcohol treatment was independently associated with achieving abstinence in the next 6 months (OR = 2. 29 , 95% CI = 1. 36 3 . 86 ) after adjusting for covariates . T h e presence of a ny illic i t drug use was negatively associated with achieving abstinence (OR = 0. 46 , 95% CI = 0. 3 1 0. 68 ) while the awareness of HCV status was associated with increased odds of achieving abstinence (OR= 1.75, 95% CI= 1 . 04 2 . 95 ). Discussion Using a cohort of HIV positive women with hazard ous drinking, we found that receipt of alcohol treatment was associated with increased odds of achieving no unhealthy drinking and abstinence, respectively. These findings supported our hypothesis that engaging in alco hol treatment may improve drinking out comes among women with HIV infection . Studies in the general population have consistently suggested that participation in alcohol treatment services is associated with improved outcomes in people with drinking problems (Dawson et al., 2006; Moos and Moos, 2006, 2004; Timko et al., 2006; Ye and Kaskutas, 20 09) . Our findings in a cohort of HIV positive women with hazardous drinking appear to be consistent with these studies as well as findings from some prior research on HIV patients (Hasin et al., 2013; Velasquez et al., 2009) . A
55 study of 253 HIV infected men who have sex with men with alcohol use disorders found that alcohol intervention resulted in re duction in both the number of drinking days and the number of heavy drinking days over a 12 month follow up (Velasquez et al., 2009) . A recent study based on 258 HIV infected individuals with recent hazardous drinking suggested that alcohol treatment, such as motivational int erviewing (MI) , had an effect on reducing the number of drinks on drink days although such effects seemed to be modest and could be modified by the presence of alcohol dependence (Hasin et al., 2013) . In contrast, b ased on a sample of 151 HIV primary care patients with lifetime history of AUD, researchers found no significant alcohol treatment effect on reduction in drinking at both 6 month and 12 month follow up (Samet et al., 20 05) . However, only 30% of the whole sample actively engaged in hazardous drinking at the beginning of th is study, and more than 60% were in some substance abuse treatment program in the 6 months prior to the enrollment. These may result in substantial c eiling effect in this study the potential to demonstrate meaningful improvement relied on only 45 subjects who reported baseline hazardous drinking. In contrast, the present study o nly focused on visits in which women reported recent hazardous drinking (i n the past 6 months) and excluded those with concurrent or prior involvement in any alcohol tre atment . Hazardous alcohol consumption has been widely documented as an important behavioral barrier to optimal HIV related outcomes, such as good adherence to a ntiretroviral therapies (>=95%) and sustained HIV viral load suppression (Chander et al., 2006; Wu et al., 2011) . R eduction in drinking may have important health implications to HIV patients . However, research data are lacking for direct me asure of any effect of alcohol treatment on any measurable improvement for HIV outcomes among HIV -
56 infected individuals with drinking problems . Our findings of alcohol treatment effect on drinking outcome warrant future study to assess whether receipt of an y alcohol treatment has any effect on HIV outcomes, such as medication adherence , viral suppression or survival . This study has several limitations. First, alcohol treatment utilization was obtained by self report questions with u nknown validity and relia bility , which can be improved in future studies u sing insurance claim records or hospital d ischarge data . However, solely relying on administrative data may limit our ability to identify many therapeutic options for alcohol use disorders that are provided outside general medical care, like mutual help programs which are the most common ly used alcohol treatment services (Cohen et al., 2007) . Second, w e grouped alcohol treatment into one single dichotomous variable, assuming equivalent effects across different treatment s . However, prior studies have suggested different treatment modalities may have different efficacy on patients with different levels of severities (Dawson et al., 2012) . Per Chapter 3 findings, multiple treatment utilizatio n was prevalent in this sample, resulting in insufficient statistical power examining effect of ind ividual treatment modality . Third, our results may be subject to un observed variable bia s. For instance, women participating in alcoho l treatment programs may be highly motivated t o change their drinking behavior, and therefore more likely to reduce or stop alcohol consumption. Without such data in WIHS, we are unable to examine their influence on alcohol treatment effect. Future randomized controlled trials may provide more conclusive evidence regarding effect of alcohol treatment among HIV infected women.
57 Last but not least , o ur results may no t be generalizable to all HIV positiv e women with hazardous drinking because WIHS is a convenie nt sample of HIV infected women. However, WIHS is intended to be representative of women with HIV infection in US (Bacon et al., 2005) . Despite above limitations, our study pr ovides promising evidence suggesting engaging in alcohol treatment among women with HIV infection who h ave hazardous drinking was associated with less unhealthy drinking and increased abstinence. Considering high prevalence of hazardous drinking among women with HIV infection, improved screening of hazardous alcohol use and increased efforts to enhance util izati on of alcohol treatment may reduce complications caused by hazardous drinking and maximize benefits from ART to this population.
58 Table 4 1 . C haracteristic s of 3 43 HIV positive WIHS women with recent hazardous drinking a Characteristics N=3 43 Age ( mean, SD ) 39.1 (7.3) Black (%) 69.4 Income (%) <=12K >12K 70.0 30.0 Social support (%) 59. 6 Number of drinks per week (%) 12 15 16 35 36+ 27.4 5 4 . 2 1 8 . 4 Any drug use (%) 68. 4 HCV+ (%) 59. 6 a Characteristics on the 1 st hazardous drinking visit
59 Table 4 2 . T he association of alcohol treatment with no unhealthy drinking and abstinence among HIV infected women with hazardous drinking AOR: Adjusted Odds Ratio; 95% CI: 95% Confidence Interval ; a no unhealthy drinking includes moderate/light drinking as well as abstinence No unhealthy drinking a Abstinence AOR 95% CI P value AOR 95% CI P value Alcohol treatment Yes No 1. 5 8 1.00 1.0 5 2. 37 0.0 3 2. 29 1.00 1. 36 3 . 86 < 0.01 Age (every 5 years) 0. 9 8 0. 8 6 1. 1 1 0. 76 1 . 1 3 0. 92 1. 40 0. 25 Black vs . others 1. 1 4 0. 7 8 1. 6 9 0. 5 0 1 . 68 0.84 3 . 36 0. 1 4 Number of drinks per week 12 15 16 35 36+ 1. 41 1 . 0 3 1.00 0. 9 8 2 . 04 0. 7 5 1. 40 0. 0 7 0. 87 0. 91 0. 98 1.00 0. 5 1 1. 64 0. 63 1. 5 2 0. 76 0. 94 Income <=12K >12K 0.9 5 1.00 0.7 0 1. 29 0. 74 0.87 1.00 0. 4 8 1 . 60 0. 66 Social support Yes No 0. 8 6 1.00 0. 6 6 1.1 0 0. 2 3 0.9 0 1.00 0.6 8 1 . 21 0. 49 Any drug use Yes No 0. 7 5 1.00 0. 54 1.0 4 0. 09 0. 4 6 1.00 0. 31 0. 68 <0.0 1 HCV Yes No 0. 9 0 1.00 0. 6 0 1. 34 0. 59 1. 75 1.00 1 . 04 2 . 95 0. 04
60 Figure 4 1 . treatment drinking visit T 0 represents hazardous drinking visit s in which women receive no alcohol treatment in the past 6 months; t 1 represents visit next to t 0 where treatment status is obtained for the past 6 months; t 2 repre sents the visit next to t 1 in which drinking outcomes are reported for the past 6 months.
61 CHAPTER 5 ESTIMATING SURVIVAL IMPACT OF HAZARDOUS DRINKING AND ALCOHOL TREATMENT USE AMONG PEOPLE WITH HIV INFECTION Introduction Advances in antiretroviral therapy (ART) significantly prolonged the survival of people with HIV infection, shifting HIV fr om a progressive illness with a uniformly fatal outcome into a manageable chronic condition (Maartens et al., 2014; Volberding an d Deeks, 2010) . Although AIDS related conditio ns remain the most common cause of deaths, population of HIV infected individuals is aging and facing increased risks of complications and mortality not directly attributable to HIV (Neblett et al., 2011; Samji et al., 2013; Smith et al., 2014; Wada et al., 2014; Weber et al., 2013) . Thus, to identify and address mod ifiable behavioral factors that affect HIV diseases progression, comorbidities, a nd overall survival becomes increasing important for the delivery of high quality HIV care in the modern ART era (Aberg et al., 2014) . Hazardous alc ohol use is one such modifiable risk behavior. Hazardous alcohol use is often defined as >14 drinks per week or >4 drinks per day for men, and >7 drinks pe r week or >3 drinks per day for women (Willenbring et al., 2009) . Alcohol use is common among people living with HIV. The prevalence of hazardous alcohol consumption ranges from 8 % to 44 % among the HIV patients (Chander et al., 2006; Cook et al., 2009; Stein et al., 2005; Tucker et al., 2003) . Research has suggested that hazardous drinking is r elated with decr eased engagement and retention in HIV care (Chander et al., 2006; Cunningham et al., 2007) . It also impedes achieving optimal adherence to ART (S. Braithwaite et al., 2005; Chander et al., 2006; Hendershot et al., 2009) , result ing in poor virologic responses and more rapid progression to advanced stage (Wu et al., 2011) . Moreover, l iver disease is com mon in HIV infected individuals
62 because of high levels of co infection with viral hepatitis and hepato toxicity of some HIV medications (Joshi et al., 2011) . Chronic e xcessive alcohol consumption damages liver tissue, affects liver function, and lead to incr eased risk of cirrhosis and hepatocellular carcinoma (McGinnis et al., 2006) . Heavy drinking als o influences cognition function, and is associated with a variety of adverse social, legal, psychological or non medical consequences (Friedmann et al., 2001) . Few studies have examined the relationship between hazardous alcohol consumption and survival in the HIV infected population (Braithwaite et al., 2005; Cohen et al., 2002; Neblett et al., 2011) . The first goal of this study was to develop a computer simulation model to quantify the impact of hazardous drinking in terms of years of surviva l among individuals with HIV infection. This simulation model differed from previously published work (Braithwaite et al., 2005) because it sp ecifically modeled the dynamics of hazardous drinking behavior and its effect on both AIDS related and non AIDS related deaths. Studies in the general population have suggested that receipt of alcohol treatment had been associated with improved drinking outcome (Dawson et al., 2006; Moos and Moos, 2006, 2004; Timko et al., 2006; Ye and Kaskutas, 2009) . Some studies focusing on the HIV infected population have also suggested the effect of alcohol treatment on reducing number of drinks and drinking days over 60 days up to 12 month (Hasin et al., 2013; Velasquez et al., 2009) . Our study described in Chapter 4 further provides evidence suggesting protective effect of receiving alcohol treatment. Given the high prevalence of hazardous alcohol consumption among HIV patients , treatment to reduce hazardous drinking may have the pot ential to improve health and
63 extend life. Primary data analyses lack sufficient statistical power to investigate this relationship because deaths have become increasingly rare events with modern therapies, and few prospective cohorts have the measure of al cohol treatment utilization . Therefore, t he second purpose of the study was to estimate potential survival impact of receiving alcohol treatment among HIV infected hazardous drinkers . Moreover , we specifically explored the association of important treatment characteristics, utilization rate and relapse prevention, with survival endpoints. Our results may provide important information to health policy makers and HIV caregivers on planning and development of effec tive treatment strategies to address hazardous drinking among HIV patients . Method Model O verview We developed a computer simulation model to estimate the effect of hazardous drinking and alcohol treatment use on life expectancy among HIV infected people. Our model was an individual based, Monte Carlo simulation model using Microsoft Excel version 2010 (Microsoft, Redmond, WA) and Crystal Ball 11.2 (Oracle, Redwood City, CA). The model evaluated a cohort of 20,000 antiretroviral naÃ¯ve individuals with new ly diagnosed HIV in fection. Individuals entered the model with an mean age of 35 years, initial median CD4 counts of 364 cells/ÂµL and a median HIV viral load of 13,000 copies/ml (Govender et al., 2014) . We used a lifetime horizon to project survival. The length of each time cycle was three months, according to the guideline recommended frequency of monitoring of CD4 and viral load (DHHS, 2013) . Each patient proceeded through the model with a distinct c linical trajectory, mirroring the variability and heterogeneity that are present in actual HIV population. Patients were at risk for a
64 variety of clinical events, such as failure of a regimen, viral rebound, CD4 counts decline, and death. In the next secti ons, we described how the base model portrayed the HIV disease progression with the administration of HAART. Based on the validated base model, we, later, described how we incorporated the component s of hazardous drinking, alcohol treatment, and analysis p erformed to examine their influence on survival. HIV Disease Progression Table 5 1 summarizes the values of key input parameters and their corresponding sources. We assumed the HAART regimen initiated at the start of the simulation based on the most recen t guideline recommended use of ART for all HIV infected adults after diagnosis (DHHS, 2013) . We modeled a maximum of four HAART regimens (80% successful rate to achieve viral suppression if initiated) followed by a salvage regimen (Prabhu et al., 201 1) . Successful treatment regimen was associated with suppressed viral replication, recovery of CD4 cell counts, and prolonged life years (Maartens et al., 2014) . Once successfully suppressed, HIV viral load remained const ant at 48 copies/ml and CD4 began to increase (Kieffer et al., 2004) . CD4 counts were assumed to increase at the highest rates during the first 2 quarters following the success ful regimen (68 cells/ ÂµL/quarter), at a slower rate from 3 rd to 12 th quarter, and at 0 thereafter (Gras et al., 2007) . The maximum level of CD4 recovered was based on the baseline CD4. For instance, a patient whose initial CD4 count s was between 50 cells/ ÂµL and 200 cells/ ÂµL could return to a maximum of 410 cells/ ÂµL if viral suppression sustained. We obtained the probability of viral rebound for the first round of HAART regimen based on a Swiss HIV Cohort Study of 1,866 patients (Vo et al., 2008) , and we
65 assumed patients with each subsequent regimen would have increased rebound risk (Prabhu et al., 2011) . We converted annual rebound risk reported into quarterly probabilities by assuming the transitional rates were constant over time and the probabilities could be modeled by an exponential distribution , where p was the probability and r was the transitional rate (Briggs 2006). Similarly, all probabilities or risk obtained in other time frames were converted into quart erly transitional probability. When a treatment regimen ceased to be effective, vi ral lo ad rebounded and CD4 counts began to decline (S anders et al., 2005) . Upon exhaustion of all four suppressive regimens, a salvage regimen or non suppressive therapy was used so that the viral replication was no longer suppressed and CD4 counts continued to decline. We assumed that CD4 decrement rate was greater at a higher HIV viral load according to results reported by Mellors et al (Mellors et al., 1997) . We used risk of AIDS related mortality and non AIDS related morality reported among 3545 HIV infected individuals in Collaborations in HIV Outcomes Research US (CHORUS) (S. Braithwaite et al., 2005) . The Risk of deaths attributable to AIDS was a func tion of age, CD4 count, and viral load ( T able 5 2). Given limited death data, we assumed that the risk of AIDS related deaths for patients whose age > 60 equaled to that of those age 50 59. Non AIDS related mortality risk varied by different age groups an d disease categories . We considered top 4 leading causes of non AIDS related deaths commonly reported among HIV infected people: cardiovascular conditions, non AIDS cancers, liver diseases , and others (Smith et al., 2014) . For each non AIDS cause of death , we made a conservative assumption for HIV infected people by
66 assuming their mortality risks in age groups 70 79, 80 89, 90 99, and 100+ were the same with those in the general population (Murphy et al., 2013) . Before adding the components of hazardous drinking and alcohol treatment, we tested the model sensitivity to key survival predictors by comparing mean survival years from 6 combinations of baseline age, CD4 counts, and viral loads. In addition, we checked the model validity by comparing our estimates of survival for general HIV patients at age 20 and age 35 with findings reported by the largest publishe d analysis of mortality in the era of current therapies the Antiretroviral Treatment Cohort Collaboration (ART CC), consisted of 14 cohorts and 18,587 patients initiating therapy (Antiretroviral Therapy Cohort Collaboration, 2008) . Modeling Hazardous Drinking and Alcohol T reatment To estimate potential influence on survival, we assumed hazardous drinking had impacts on both AIDS related deaths and non AIDS related deaths (Figure 5 1) . The form influences on viral suppression. According to the study based on 1711 HIV positive individuals in treatment , researchers found that hazardous drinking was associated with decreased odds of achieving viral suppression compared with moderate drinkers and non drinkers (Chander et al., 2006) . We obtained the probability of hazardous drinking status by the following two formulas (Briggs et al., 2006) : (5 1)
67 deno tes odds ratio of viral suppression by hazardous drinking status; denotes the quarterly probability of achieving viral suppression among hazardous drinkers, and denotes the quarterly probability of viral rebound. ( 5 2) denotes rebound risk among general HIV infected individuals, represents the prevalence of hazardous drinking among HIV infected indi viduals and we assumed for the computation of . Hazardous alcohol consumption has been found to be associated with non HIV related deaths (e.g., liver diseases, alcohol related cancers) (Joshi et al., 2011; Kushi et al., 2006) . We modeled the impact of hazardous drinking on non HIV related deaths by using data of relative risks repo rted from a nationwide prospective mortality study with 490,000 people (Thun et al., 1997) . The risk of non AIDS related deaths by hazardous drinking status was obtained by formulas described below (Clarke et al., 2010) . Table 5 2 shows the resulting mortality risk for non AIDS causes by hazardous drinking status. Taking death risk by liver diseases as an example: ( 5 3) denotes quarterly mortality risk by liver diseases in general HIV infected individuals; denotes the quarterly mortality risk by liver diseases among hazardous drinkers ( 5 4) denotes relative risk of deaths of liver diseases by hazardous drinking status.
68 It is essential to include the dyna mic nature of hazardous drinking behavior for estimating impact on survival (Hahn and Samet, 2010) . In this model, hazardous drinking was measured by a time varying dichotomous variable updated every 3 months. W e used WIHS data to estimate the occurrence of hazardous drinking . Based on data of 474 hazardous drinkers identified in Chapter 3 , the estimated quarterly probabilities of discontinuing hazardous drinking were 40% for those receivi ng alcohol treatment , and 31% for those receiving no treatment. Fur ther focusing on those discontinued hazardous drinking we calculated the quarterly relapse rate (probability for relapsing to hazardous drinking) for those who were in alcohol treatment (7%) and for those without alcohol treatment (11%). We assumed the rel apse rate (7%) for successful treatment (remission of hazardous drinking) lasting for a year and then returne d to same relapse rate (11%) for those who previously recovered without the assistance of treatment . To estimate survival impact of hazardous drin king, we c ompared the mean survival years by whether hazardous drinking was present or not. To estimate survival impact of alcohol treatment, we first assumed a 10% quarterly utilization rate or a 10% probability of receiving any alcohol treatment. This pr obability was derived from the Chapter 3 finding that recent alcohol treatment use was reported among 19% of those with recent hazardous drinking. Then we compared the result with that estimated when no treatment was present. We conducted sensitivity anal yses to evaluate the impact of key treatment characteristics on survival years. Firstly, we examined influences of varying probabilities of receiving treatment (from 0% to 80%) on survival. Secondly, we projected treatment
69 effect on survival by assuming th at no treatment effect on subsequent relapse. Finally, we simulated the survival years ba sed on a hypothetical treatment , given the effectiveness of which were two times greater than the current treatment: following the treatment, twice the chance of achie ving non hazardous drinking in the next quarter (40% >80%), and half the chance of relapsing to hazardous drinking (7% >3.5%) for the next 4 quarters following the treatment. Results HIV Progression Model The results from the HIV progression model suggested that the estimated mean survival depended on age, baseline viral load and CD4 count in each age stratum (table 3). For 20 year old patients with 800 cells/ÂµL CD4 count, and 500 copies/ml viral loads, the mean survival were 45.45 years. In contrast, for patients at age 50 with the same baseline CD4 count and viral load, the mean survival were 26.27 years. Generally, the estimated mean survival increased with higher CD4 counts and decreased with higher vira l loads. The projected percentage of non AIDS deaths increased with age and varied substantially with basel ine CD4 counts and viral load (T able 5 3). In general, patients with less advanced stage of HIV disease (lower viral load and higher CD4 counts) were projected to have higher risk of non AIDS related deaths. For patients at age 20, the proportion probability of non AIDS related deaths was 51% a mong those with favorable prognostic indicators (e.g., CD4 count of 800 cells/ÂµL, and viral loads of 500 copies/ml). In contrast, this proportion reduced to only 32% among those with CD4 count of 150 cells/ÂµL and viral loads of 100000 copies/ml. For age 50 , the majority of deaths (>70%)
70 were attributable to non AIDS related causes regardless of baseline HIV prognostic factors. Based on a cohort of individuals with age, CD4 counts, and viral load obtained from typical newly diagnosed HIV patients (Govender et al., 2014) , the projected mean survival was 41.01 years at age 20, and 33.62 years at age 35. Both were no more than +/ 2 years compared to estimates from ART CC fo r the same ages (Antiretroviral Therapy Cohort Collaboration, 2008) (F igure 5 2 ), demonstrating the ability of our base model in projecting survival of general HIV infected individuals. Surviva l Impact of Hazardous Dinking and Alcohol T reatment The projected mean survival years for a newly diagnosed HIV patient were 34.62 years in the absence of hazardous drinking and 32.97 years in the presence of hazardous drinking but no alcohol treatment, su ggesting h azardous drinking independently resulted in an average of 1.65 years reducti on in projected survival. In the presence of alcohol treatment with a 10% quarterly chance of utilization , the projected mean survival was 33.15 years, indicating an aver age of 0.18 years gained by utilization of alcohol treatment. Increasing the utilization rate, or the chance of receiving any alcohol treatment, was associat ed with better survival outcome (Figure 5 3 ). For example, if the chance of getting any alcohol tre atment per quarter increased to 20%, death was delayed by 0.31 years, suggesting a 72% increase compared to the 10% utilization rate (Table 5 4). However, the trend of increasing survival years gradually leveled off with increasing utilization rate and pla teaued at around 0.68 years, accounting for 41.2% of overall loss in survival due to hazardous drinking . This also represented the maximum level of improvement in survival years by increasing the chance of using treatment up to 80%
71 given the default treatm ent effect . Figure 5 4 shows the trend of marginal survival years by increasing the chance of receiving treatment. The marginal survival years decreased sharply from 0.18 years at 10% chance to around 0.08 years at 30% chance. Continuing decrease in margin al gain was observed by increasing the chance of utilization. Increasing the chance of receiving any alcohol treatment was associated with increasing survival without considering treatment effect on relapse prevention (Table 5 5). However, life years saved in this case declined greatly. For example, at the utilization chance of 20%, the survival increase was 0.10 years, accounting for only 32.3% of that when treatment effect of relapse prevention was considered (0.31 years) . Doubling the treatment effect r esulted in substantial increases in survival years across all utilization chance (Figure 5 5 ). Given a 10% chance of receiving alcohol treatment, the survival years gained was 0.42 years which was 2.4 times as great as that gained without doubling the trea tment effect (Table 5 4). Almost one year life year (0.9 years) saved was observed when the chance of receiving treatment increased to 30%. Similarly, t he marginal survival yea rs gained decreased as the chance of receiving treatment increased (Figure 5 6 ). Discussion Our simulation model suggested that h azardous drinking am ong HIV infected individuals result ed in decreased survival but receiving alcohol treatment could avert this process . We found th at hazardous drinking among HIV patients resulted in an average of 1.65 years reduction in mean survival years. The presence of alcohol treatment showed modest survival benefits, although our sensitivity analyses suggested
72 that both the chance of receiv ing alcohol treatment and treatmen t effect may play important roles in determining the treatment impact on survival. Our estimate of decrement in survival due to hazardous drinking was lower than estimates from previous studies. In a study projecting alcohol use and survival in HIV patien ts, researchers found that daily hazardous drinkers was 4 6.4 years shorter in overall life expectancy compared with abstainers (Braithwaite et al., 2005) . Our estimates favoring a small survival impact may re flect that fact that our model allows the changing drinking behavior over time . Rather than assuming a consistent drinking behavior such as haz ardous drinking every day , we considered the case that hazardous drinkers can resolve by themselves from heavy drinking (Sobell et al., 2000, 1996, 1993) and h ave certain chance to relapse to hazardous drinking again in the future (Booth et al., 2001; Dawson et al., 2006; Moos and Moos, 2006) . We applied the probability of relapse obtained from the WIHS data, which was 11.4 % per quarter, or 38.4% per year. This means that i t takes an average of 2.6 years for a self resol ved hazardous drinker to experience another episode of hazardous drinking. Our data on relapse and natural recovery all came from the longitudinal cohort of HIV patients, representing the better representatio only unchanged drinking patterns. Also, if consistent life long hazardous drinking was considered in our model, the estimated mean survival reduction was 5.0 years which was close to estimates of daily hazardou s drinkers mentioned above . Significant underutilization of alcohol treatment among people with drinking problems has been consistently reported in previous literature (Cohen et al., 2007; Ilgen et al., 2011) . Our sensitivity analyses tested the association between changing
73 utilization rate and survi val. Our findings support their positive relationship although the strengths of the association are strongly contingen t on the treatment effect . For instance, our estimate of survival years gained at the 30% of utilization rate was a half year in base case and was almost one year if the treatment effect was doubled . In addition, our results suggest survival increase is not completely linear with increasing utilization rate. The marginal gain for every 10% increase in utilization rate gradually declined and remained flat after approaching 30%, suggesting a potential optimal point to reach for any new intervention or public health strategy encouraging access and utilization of alcohol treatment. Our data suggest any potential survival impact of alcohol treatm ent, if measurable, is strongly dependent on helping patients maintain recovery while in remission. Without considering treatment effect on relapse prevention, our model indicates that the treatme nt impact on survival is small even when the utilization rat e is high. These findings suggest the critical role of relapse prevention in the development and evaluation of alcohol treatment. Psychosocial therapies, including some self help programs, brief interventions, motivational interviewing (MI), and cognitive behavioral therapy (CBT), have shown certain effectiveness on reducing drinking and maintain sober among patients with low dependent on alcohol et al., 1999) . Several medications have also been demonstrated to be effective on relapse prevention among individuals with alcohol dependence (Anton et al., 2003; Maisel et al., 2013; Mason et al., 2014; Spanagel et al., 2014) . Although few has been investigated in the HIV infected population, early and tim ely provision of these treatment
74 to HIV infected individuals with the history of drinking problem or in remission but at high risk of relapse may demonstrate benefits in the long term. There are several limitations that should be considered. The main lim itation of this study is related to those inherent in any simulation study. O ur model, like any model, is at best an approximation of the reality and the effect of any misspecification could be amplified due to the extremely long simulation period (Briggs et al., 2006; Clarke et al., 2010) . Due to the lack of data on alcohol treatment among men with HIV infection, we used the data from a cohort of HIV infected women to derive information for drinking patterns and treatment effect. Men are generally more likel y to drink, drink heavily, and engage in alcohol specific treatment than women (Cohen et al., 2007) . If similar gender differences were observed in t he HIV infected population, our estimates on survival impact of both hazardous drinking as well as alcohol treatment tend to be underestimated. In addition as discussed above, our model assumed that hazardous drinkers who were newly diagnosed would be link ed to care immediately, and to initiate HAART to the same degree as those without hazardous drinking. This assumption may be too strong given previous studies suggest hazardous drinkers are less likely to engage and remain in HIV care (Chander et al., 2006; Cunningham et al., 2007) , resulting in potential underestimated survival impact of hazardous drinking. Finally, we modeled the survival impact of the general alcohol treatment without distinguishing individual treatment modality patients may encounter. Different types of treatm e nt have different efficacy and different mechanism of effect. For example, N altrexone can be used for AUD patients with concurrent alcohol use to reduce craving but a camprosate are general ly used to prevent relapse after detoxification is completed or abst inence
75 from alcohol has already been achieved (M aisel et al., 2013) . Although modeling specific type of treatment is beyond the scope of this study, future research can update the current mod el with data from high quality randomized trial to provide projection of long term impact for any particular t reatment. In summary, hazardous drinking leads to considerable negative implications on survival of HIV infected individuals. Receipt of alcohol treatment has the potential to avert this process to some degree which depends greatly on the chance of treatme nt entry and treatment effect. Greater efforts are needed to improve the delivery of alcohol treatment services and the development of new interventions more potent because of their potential survival benefits for HIV infected individuals with hazardous dr inking.
76 Table 5 1 . Summary of input parameters Variable Base Case Range Source CD4 cell count when diagnosed (cells/ÂµL) 364 238 542 (Govender et al., 2014) HIV viral load at diagnosis (log 10 copies/ml) 4.1 3.1 5 (Govender et al., 2014) CD4 cell count decline (cells/ÂµL/quarter) as a function of HIV viral load stratum (log 10 copies/ml) (Mellors et al., 1997) 9.1 7.6 10.6 2.7 3.3 11.2 9.8 12.6 3.3 4.0 13.8 12.7 14. 9 4.0 4.6 16.2 14.9 17.3 19.1 17.6 20.7 Suppressed HIV viral load (log 10 copies/ml) 1.7 1.0 2.7 (Kieffer et al., 2004) Rebound HIV viral load (log 10 copies/ml) 4.4 3.6 4.6 (Sanders et al., 2005) Maximum number of HAAR T regimens 4 (Prabhu et al., 2011) HIV viral load at salvage regimen 4.9 3.1 5.8 (Sanders et al., 2005) Probability of virologic suppression in HAART regimens 1 4 0.80 (Farnham et al., 2012; Prabhu et al., 2011) Quarterly increase in CD4 cell count during HIV viral load suppression (cells/ÂµL/quarter) (Gras et al., 2007) Quarters 1 2 68 Quarters 3 12 40 Quarters 12+ 0 Maximum CD4 cell count achieved based on CD4 cell count at initiation of HAART (cells/ÂµL) (Gras et al., 2007) < 50 410 50 200 548 201 350 660 351 500 780 > 500 870 Quarterly p robability of HIV viral load rebound % for general HIV infected individuals (Vo et al., 2008) ; (Prabhu et al., 2011) Regimen1 0.0190 Regimen2 0.0225 Regimen3 0.0268 Regimen4 0.0320
77 Table 5 1 . Continued Variable Base Case Range Source Quarterly p robability of HIV viral load rebound % for HIV infected individuals without hazardous drinking (HD a ) (Chander 2006) ; (Vo et al., 2008) ; (Prabhu et al., 2011) Regimen1 0.0179 0.0166 0.0190 Regimen2 0.0212 0.0196 0.0225 Regimen3 0.0252 0.0234 0.0268 Regimen4 0.0301 0.0280 0.0320 Quarterly p robability of HIV viral load rebound % for HIV infected hazardous drinkers (Chander 2006) ; (Vo et al., 2008) ; (Prabhu et al., 2011) Regimen1 0.0237 0.0190 0.0288 Regimen2 0.0278 0.0225 0.0340 Regimen3 0.0333 0.0268 0.0404 Regimen4 0.0397 0.0320 0.0481 Relative risk of deaths for non HIV conditions (HD vs. NHD b ) (Thun et al., 1997) Cardiovascular conditions 0.9 0.8401 1 . 0 924 Non AIDS Cancers 1.13 1.1275 1.1428 Liver diseases 4.00 3.0265 4.4429 Other causes 1.2 0.9701 1.5041 Quarterly p robability of achieving NHD The WIHS Treated 0.40 Untreated 0.31 Quarterly r elapse to HD The WIHS Treated 0.07 Untreated 0.11 Alcohol treatment utilization 0.10 a HD: hazardous drinking; a NHD: no hazardous drinking
78 Table 5 2 . Quarterly probability of deaths by age, CD4 count, viral load, and hazardous drinking a Age 20 29 30 39 40 49 50 59 60 69 70 79 80 89 90 99 100+ Probability of death attributable to AIDS Viral load b CD4 <4.47 0 49 0.010721 0.010721 0.024189 0.035858 0.035858 0.035858 0.035858 0.035858 0.035858 50 199 0.000750 0.000750 0.001679 0.002459 0.002459 0.002459 0.002459 0.002459 0.002459 200 349 0.000275 0.000275 0.000601 0.000876 0.000876 0.000876 0.000876 0.000876 0.000876 350 499 0.000100 0.000100 0.000225 0.000325 0.000325 0.000325 0.000325 0.000325 0.000325 500+ 0.000025 0.000025 0.000050 0.000100 0.000100 0.000100 0.000100 0.000100 0.000100 >4.47 0 49 0.021509 0.021509 0.049623 0.075156 0.075156 0.075156 0.075156 0.075156 0.075156 50 199 0.001503 0.001503 0.003316 0.004860 0.004860 0.004860 0.004860 0.004860 0.004860 200 349 0.000525 0.000525 0.001177 0.001704 0.001704 0.001704 0.001704 0.001704 0.001704 350 499 0.000200 0.000200 0.000450 0.000651 0.000650 0.000650 0.000650 0.000650 0.000650 500+ 0.000050 0.000050 0.000125 0.000175 0.000175 0.000175 0.000175 0.000175 0.000175 Probability of death attributable to non AIDS causes Overall Cardio 0.000050 0.000170 0.000214 0.000578 0.001311 0.001705 0.004807 0.013221 0.220000 Cancer 0.000027 0.000093 0.000117 0.000315 0.000715 0.000929 0.002622 0.007211 0.120000 Liver 0.000050 0.000170 0.000214 0.000578 0.001311 0.001705 0.004806 0.013221 0.220000 Other 0.000100 0.000341 0.000429 0.001157 0.002628 0.003409 0.009613 0.026442 0.440000 NHD c Cardio 0.000051 0.000174 0.000219 0.000589 0.001338 0.001739 0.004904 0.013491 0.224489 Cancer 0.000026 0.000096 0.000114 0.000307 0.000696 0.000906 0.002555 0.007029 0.116959 Liver 0.000031 0.000106 0.000134 0.000361 0.000819 0.001065 0.003004 0.008263 0.137500 Other 0.000097 0.000328 0.000412 0.001112 0.002527 0.003278 0.009243 0.025425 0.423077 HD d Cardio 0.000046 0.000156 0.000197 0.000530 0.001204 0.001565 0.004414 0.012142 0.202041 Cancer 0.000030 0.000102 0.000129 0.000347 0.000787 0.001024 0.002887 0.007942 0.132164 Liver 0.000125 0.000426 0.000536 0.001445 0.003279 0.004262 0.012016 0.033053 0.550000 Other 0.000115 0.000393 0.000495 0.001335 0.003033 0.003934 0.011092 0.030510 0.507692 a Death risk above are quarterly probabilities and obtained from (Braithwaite et al., 2005) ; b Viral load is termed as log10 copies ; c NHD denotes no hazardous drinking; d HD denotes hazardous drinking ; c
79 Table 5 3 . E stimates of mean survival and percentage of non AIDS causes of deaths Age (years) CD4 count (cells/ÂµL) Viral load (copies/ml) Mean Survival (years) Percentage of non AIDS causes (%) 20 800 500 45.45 51 150 100000 38.47 32 35 800 500 36.56 72 150 100000 31.96 50 50 800 500 26.27 89 150 100000 23.88 71
80 Table 5 4 . Survival years estimated by increasing the chance of receiving any alcohol treatment Receiving alcohol treatment Doubling alcohol treatment effect a Utilization rate Survival years Survival years saved Survival years Survival years saved Difference between survival years saved 0% 32.97 0.00 32.97 0.00 0.00 10% 33.15 0.18 33.39 0.42 0.24 20% 33.28 0.31 33.74 0.77 0.46 30% 33.36 0.38 33.87 0.90 0.52 40% 33.43 0.46 34.02 1.08 0.59 50% 33.49 0.52 34.12 1.15 0.62 60% 33.56 0.58 34.21 1.23 0.65 70% 33.61 0.64 34.21 1.24 0.60 80% 33.65 0.68 34.22 1.24 0.56 a Doubling alcohol treatment effect means twice the chance of reducing hazardous drinking (40% to 80%) and half the chance of relapse (7% to 3.5%)
81 Table 5 5 . S urvival years without incorporating of treatment effect on relapse prevention Receiving alcohol treatment No effect on relapse prevention Utilization rate Survival years Survival years saved Survival years Survival years saved Difference between survival years saved 0% 32.97 0.00 32.97 0.00 0.00 20% 33.28 0.31 33.07 0.10 0.21 40% 33.43 0.46 33.14 0.17 0.29 60% 33.56 0.58 33.17 0.20 0.39 80% 33.65 0.68 33.22 0.25 0.43
82 Figure 5 1 . Impact of alcohol treatment on both AIDS related and non AIDS related mortality
83 Figure 5 2 . E stimated mean survival years at age 20 and 35 (gray bar) from base model , and estimates of same ages from the Antiretroviral Treatment Cohort Collaboration (ART CC)
84 Figure 5 3 . Trend of survival years depending on utilization rate of alcohol treatment
85 F i gure 5 4 . M arginal survival years by increasing the utilization rate
86 Figure 5 5 . T rends of survival years by doubling alcohol treatment effect
87 Figure 5 6 . Marginal survival years by doubling alcohol treatment effect
88 CHAPTER 5 CONCLUSIONS Summary The overarching aim of this dissertation project was to improve the understanding of alcohol treatment use among HIV infected individuals with hazardous drinking. We conducted 3 different types of studies to help better address this goal. In the first study, we focused on the utilization. Using data from a large cohort of HIV infected women in the US, we first described the utilization of alcohol treatment among those with hazardous drinking, and then we identified factors correlated with the (Andersen, 1995) . The second study focused on examining alcohol treatment effect on drinking outcomes ( achieving no unhealthy drinking and abstinent ) . The third study made an attempt to project potential survival implication of hazardous drinking and alcohol treatment use among HIV infected peop le. The individual based, Monte Carlo simula tion model was built considering administration of HAART, dynamic s of hazardous drinking, and alcohol treatment utilization. In the rest of this section, I will summarize the major findings of above 3 studies, the overall strengths and weakness, and implications for future studies and public health policy . Study 1 Utili zation of Alcohol T reatment We hypothesized that utilization of alcohol treatment was low among women with HIV infection. Our finding is suppo rting this hypothesis: a mong 474 HIV positive women identified in WIHS who reported hazardous drinking in the past 6 months, one in five recently engaged in any alcohol treatment. The rate of alcohol treatment use found in our study is generally consistent with many community studies. The 2002 National
89 Survey on Drug Use and Health (DSDUH) found that among individuals with combined AUD or SUD, the prevalence of past year treatment utilization for substance use disorders (SUD) was 22.1% (Mojtabai, 2005) . In another study on low socioeconomic class women with substance use problems, the rates of past year substance abuse treatment could be as high as 44.0% (Hansen et al., 2004; Rosen et al., 2004) . Mojtabai suggest that high levels of utilization of alcohol or substance abuse treatments may suggest more severe issues of AUD or SUD (Mojtabai, 2005) . The se cond hypothesis of this study was that ena bling and need factors may play important roles determining the utilization of alcohol treatment. This hypothesis is strongly supported by our results. First, we found no association of predisposing factors with alcohol treatment utilization. This finding is expected because results of previous studies are mixed on the association of age with alcohol treatment seeking (Cohen et al., 2007; Grella et al., 1999; Ilgen et al., 2011; S and Bowe, 2008; Tighe and Saxe, 2006; Weisner et al., 2002) . In terms of race/ethnicity, our finding is consistent with prior work in which no racial difference is generally found in terms of receiving any alcohol treatment (Hatzenbuehler et al., 2008) . Second, we found that a mixture of enabling (income and social support) and need factors (drinking levels and drug use) were associated with the utilization of alcohol treatment. The finding of the negative association of income with alcohol treatment utilization was generally consistent with prior observations among non HIV infected individuals with AUD (Cohen et al., 2007; Ilgen et al., 2011; Mojtabai, 2005) . Individuals with low socio economic classes are over represented in the criminal justice which has traditionally been closely associated with alcohol and substance abuse
90 treatment (Weisner and Schmidt, 1995) . In addition, the availability of public entitlements h as resulted in many Individuals of low socio economic status seeki ng substance abuse treatments. For example, the most common form of treatment reported in the Chapter 3 was participation in Alcoholic Anonymous, which is available at no cost. We found gre ater odds of receiving alcohol treatment among HIV infected women with social support. Emotional support obtained from families and friends may help HIV infected women overcome significant social stigma associated with the alcohol abuse and HIV infection a nd motivate care seeking to address both conditions. Also, transportation, or important family services, may alleviate life stressors, enhancing access to alcohol treatment fo r many HIV infected women with AUD. The lack of adequate supporting system has contributed substantially to the low rates of alcohol treatment uptake (Ilgen et al., 2011) . Future alcohol treatment programs focusing on HIV infected women may consider the integ ration of specific social support strategies to facilitate care access and quality of care. This positive association between recent illicit drug use and utilization of alcohol treatment is consistent with prior research (Cohen et al., 2007; Dawson et al., 2012; Hasin, 1994; Kessler RC et al., 2001 ; Mojtabai et al., 2002) . Cohen and colleagues (2007) suggested that drug use disorders were the comorbid psychiatric disorders that had the strongest influence on alcohol treatment utilization. In the current study, concurrent drug use was reported by more t han two thirds of the sample. A dditive distress experienced by HIV positive women may increase their motivation to seek care
91 for drinking problems. It is also possible that HIV positive women who acknowledge recent drug use may be mandated by crimin al justice systems to receive treatment for substance use in general, including their alcohol abuse; they may also be more likely to come to the attention of care providers and receive referrals to alcohol treatment. Study 2 Alcohol Treatment a nd Drinking O utcomes In this study, we hypothesize d the positive association between receipt of alcohol treatment and achieving better drinking outcomes, such as no unhealthy drinking, and abstinence . In a large cohort of HIV positive women with hazardous drinking, w e found that receipt of alcohol treatment was associated with increased odds of achieving no unhealthy drinking and abstinence, respectively. These findings support our hypothesis that engaging in alco hol treatment may improve drinking outcomes among women with HIV infection . Our findings of alcohol treatment effect in HIV positive women appear to be consistent with prior research conducted among individuals with HIV infection (Hasin et al., 2013; Velasquez et al., 2009) . R eduction in drinking may have important health implications to HIV patients. However, research data are lacking for direct effect of a lcohol treatment on any measurable improvement for HIV outcomes among HIV infected individuals with drinking problems. Our findings of alcohol treatment effect on drinking outcome warrant future study to assess whether receipt of any alcohol treatment has any effect on HIV outcomes, such as medication adherence , viral suppression or survival . Study 3 Survival Impact of Hazardous Drinking and Alcohol Treatment We hypothesized that hazardous drinking was associated with substantial reduction in the survival of HIV infected individuals. We found that hazardous drinking
92 among HIV positive patients resulted in an average of 1.65 years reduction in survival year s, supporting this hypothesis. Our estimate of the decrement in survival due to hazardous drinking was lower than estimates from previous studies. In a study projecting alcohol use and survival in HIV patients, researchers found that daily hazardous drinkers was 4 6.4 years shorter in overall life expectancy compared with abstainers (Braithwaite et al., 2005) . Our estimates favoring a small survival impact may reflect our modeling of the dynamic of drinking patterns. Rather than assuming consiste nt drinking behavior such as haz ardous drinking every day , we considered the reality that many hazardous drinkers could resolve by themselves from heavy drinking (Sobell et al., 2000, 1996, 1993) and had certain chance to relapse to hazardous drinking in the future (Booth et al., 2001; Dawson et al., 2006; Moos and Moos, 2006) . If life long hazardous drinking was considered in our mod el, the estimated survival reduction was 5.0 years which was close to previous study assuming daily hazardous drinking . The second hypothesis for this study was that utilization of alcohol treatment would result in improvement in survival among HIV infected individuals with hazardous drinking. The results of our simulation model provide certain evidence supporting this hypothesis, but the effect size is modest and appear s to vary substantially according to treatment characteristics such as the chance of receiving treatment and treatment effect . For example, if the chance of getting any alcohol treatment per quarter increased to 20%, death was delayed by 0.31 years, suggesting a 72% increase compared with tha t from the 10% chance. Additionally, if treatment effect doubles (doubling the chance of ceasing hazardous drinking in the next quarter post a hazardous drinking episode,
93 and reducing the future chance of relapse by half), our estimate of survival years ga in ed at the 30% utilization chance was almost one year. Strengths and Limitations study is the first study examining the utilization of alcohol treatment among HIV infected wom en with hazardous drinking . We also provided the profile of those who received treatment and those who did not. Second , we are the first study examining the effect of alcohol treatment on drinking outcomes using a large, multicenter cohort of HIV positive wom en, adding to the liter ature about alcohol effect among HIV infected individuals . Finally, the computer model we built is the first assessing the potential survival impact of alcohol treatment among HIV infected individuals with hazardous drinking, providing important informatio n for understanding long term alcohol treatment effect. At the same time, several limitations are worth noting. First, our data on alcohol treatment use from WIHS are relative old, although information on alcohol treatment utilization among such a large c ohort of HIV patients is extremely rare . Despite increasing understanding of AUD as a chronic condition, the landsca pe of alcohol treatments has been relatively stable in the past two decades except that pharmacotherapies have begun to play a more importan t role in the management of alcoholics (Friedmann, 2013; Room et al., 2005) . Future research on alcohol treatment utilization among HIV infected individuals should include specific questions assessing the access and utilization of pharmacotherapies (e.g., naltrexone, acamprosate, disulfiram, gabapentin) . Second, the formal diagnosis o f AUD is lacking for our sample. However, quantity and frequency questions of alcohol consumption have been widely used to identify people with drinking problems (Canagasaby and Vinson, 2005; Chung
94 et al., 2012; Dawson et al., 2005; Rubinsky et al., 2013; Willenbring et al., 2009) . In addition, we cho se a higher cut off point to define hazardous drinking, resulting in higher specificity in identifying subjects in need of treatment. Third, we did not distinguish specific treatment modality when assessing determinants of treatment utilization and when examining treatment effect. Use of multiple treatment types was prevalent in our study sample and thus, the limited sample size precluded our ability to examine individual treatment. Fourth, our simulation model, like any model, is at best an approximation of the reality (Briggs et al., 2006; Clarke et al., 2010) . Due to the limited data on alcohol treatment among men with HIV infection, we used the data obtained from a large cohort of HI V infected women to populate our model. Men are g enerally more likely to drink heavily and to engage in alcohol specific treatment then women with drinking problems (Cohen et al., 2007) , and therefore, our estimates on survival impact of both hazardous drinking as well as alcohol treatment tend to be underestimated. Implications for Public H ealth and F uture research Despite above limitations, results from these three studies have important implic ations. Study 1 found low rate of alcohol treatment utilization among HIV infected women with hazardous drinking, suggesting more efforts to enhance utilization. Findings of study 1 a lso provide a unique profile of HIV infected women with hazardous drinking who received alcohol treatments and those who did not. For those who engage in care for AUD, continuing monitor is needed to ensure that drinking outcomes have been improved and sus efforts for education, information, and encouragement of care seeking are necessary. Because this study is exclusive to women, future study can be expanded to men to
95 investigate if levels of uti lization and characteristics associated with utilization differ by gender. Our second study provides promising evidence supporting alcohol treatment effect on improving drinking outcomes among women with HIV infection . Hazardous drinking is common among women with HIV infection. Improved screening and referral to treatment program may have profound implications on their overall health and well being. Future research can focus on men or consider using mixed gender population to investigate potential gende r difference in treatment effect at different settings . Also, the large sample data are needed for comparative effectiveness studies focusing on marginal effect of different treatment modalities. Our third study has at least two important public health im plications. First, our estimated survival impact of current alcohol treatment is modest among HIV infected individuals, suggesting much room needs to be done to improve the current treatment delivery and effectiveness. Second, our results suggest increasin g the utilization rate to 20% or 30% would bring a relatively large marginal gain in survival, suggesting a possible target point for improving alcohol treatment delivery to HIV infected population. Future research based on this model can be easily adapted to examine survival impact of specific type of alcohol treatment, such as medications , and to perform comparative effectiveness analysis. More importantly, when costs data for treatment are available, a formal cost effectiveness analysis can be conducted to generate data relevant to the public health decision makers regarding the care planning to HIV infected individuals with drinking problems .
96 APPENDIX WIHS QUESTIONS ABOUT ALCOHOL USE AND ALCOHOL TREATMENT
97 LIST OF REFERENCES Aberg, J.A., Gallant, J.E., Ghanem, K.G., Emmanuel, P., Zingman, B.S., Horberg, M.A., 2014. Primary Care Guidelines for the Management of Persons Infected With HIV: 2013 Update by the HIV Medicine Association of the Infectious Diseases Society of America. Clin. Infect. Dis. 58, e1 e34. Alter, M.J., 2006. Epidemiology of viral hepatitis and HIV co infection. J. Hepatol., Proceedings of the 1st European Consensus Conference on the Treatment of Chronic Hepatitis B and C in HIV Co infected Patients Proceedings of the 1st European Consensus Conference on the Treatment of Chronic Hepatitis B and C in HIV Co infected Patients 44, Supplement 1, S6 S9. American Psychiatric Association, 2013. Diagnos tic and Statistical Manual of Mental Disorders, DSM Library. American Psychiatric Association. Andersen, R.M., 1995. Revisiting the behavioral model and access to medical care: does it matter? J. Health Soc. Behav. 36, 1 10. Andersen, R.M., 2008. National Health Surveys and the Behavioral Model of Health Services Use. [Miscellaneous Article]. Med. Care July 2008 46, 647 653. Antiretroviral Therapy Cohort Collaboration, 2008. Life expectancy of individuals on combination antiretroviral therapy in high income countries: a collaborative analysis of 14 cohort studies. Lancet 372, 293 299. Anton, R.F., Moak, D.H., Waid, L.R., Latham, P.K., Malcolm, R.J., Dias, J.K., 2003. Naltrexone and Cognitive Behavioral Therapy for the Treatment of Outpatient Alcoholics: Resu lts of a Placebo Controlled Trial. FOCUS 1, 183 189. Azar, M.M., Springer, S.A., Meyer, J.P., Altice, F.L., 2010. A systematic review of the impact of alcohol use disorders on HIV treatment outcomes, adherence to antiretroviral therapy and health care utilization. Drug Alcohol Depend. 112, 178 193. Babitsch, B., Gohl, D., von Lengerke, T., 2012. Re Model of Health Services Use: a systematic review of studies from 1998 2011. Psycho Soc. Med. 9, Doc11. Bacon, M.C., Wyl, V. von, Alden, C., Sharp, G., Robison, E., Hessol, N., Gange, S., Interagency HIV Study: an Observational Cohort Brings Clinical Sciences to the Bench. Clin. Diagn. Lab. Immunol. 12, 1013 1 019.
98 Bagby, G.J., Zhang, P., Purcell, J.E., Didier, P.J., Nelson, S., 2006. Chronic binge ethanol consumption accelerates progression of simian immunodeficiency virus disease. Alcohol. Clin. Exp. Res. 30, 1781 1790. Ballinger, G.A., 2004. Using Generalized Estimating Equations for Longitudinal Data Analysis. Organ. Res. Methods 7, 127 150. Barkan, S.E., Melnick, S.L., Preston Martin, S., Weber, K., Kalish, L.A., Miotti, P., Interagency HIV Study. WIHS Collaborative Study Group. Epidemiol. Camb. Mass 9, 117 125. Baum, M.K., Rafie, C., Lai, S., Sales, S., Page, J.B., Campa, A., 2010. Alcohol Use Accelerates HIV Disease Progression. AIDS Res. Hum. Retroviruses 26, 511 518. Benhamou, Y., Bochet , M., Di Martino, V., Charlotte, F., Azria, F., Coutellier, A., Vidaud, M., Bricaire, F., Opolon, P., Katlama, C., Poynard, T., 1999. Liver fibrosis progression in human immunodeficiency virus and hepatitis C virus coinfected patients. The Multivirc Group. Hepatol. Baltim. Md 30, 1054 1058. Booth, B.M., Fortney, S.M., Fortney, J.C., Curran, G.M., Kirchner, J.E., 2001. Short term course of drinking in an untreated sample of at risk drinkers. J. Stud. Alcohol 62, 580 588. Braithwaite, Justice, A.C., Chang, C. C.H., Fusco, J.S., Raffanti, S.R., Wong, J.B., Roberts, M.S., 2005. Estimating the proportion of patients infected with HIV who will die of comorbid diseases. Am. J. Med. 118, 890 898. Braithwaite, S., McGinnis, K.A., Conigliaro, J., Maisto, S.A., Crystal , S., Day, N., Cook, R.L., Gordon, A., Bridges, M.W., Seiler, J.F.S., Justice, A.C., 2005. A temporal and dose response association between alcohol consumption and medication adherence among veterans in care. Alcohol. Clin. Exp. Res. 29, 1190 1197. Briggs, A.H., Claxton, K., Sculpher, M.J., 2006. Decision Modelling for Health Economic Evaluation. Oxford University Press. Cainelli, F., Concia, E., Vento, S., 2001. Mortality due to hepatitis C virus related cirrhosis in patients infected with HIV type 1: a ro le for alcohol. Clin. Infect. Dis. Off. Publ. Infect. Dis. Soc. Am. 33, 1795 1797. Canagasaby, A., Vinson, D.C., 2005. Screening for hazardous or harmful drinking using one or two quantity frequency questions. Alcohol Alcohol. Oxf. Oxfs. 40, 208 213. Cente r for Disease Control and Prevention, 2014. HIV Among Women Fact Sheet.
99 Centers for Disease Control and Prevention, 2013. HIV Surveillance Report, 2011 23. Chander, G., 2011. Addressing alcohol use in HIV infected persons. Top. Antivir. Med. 19, 143 147. C hander, G., Lau, B., Moore, R.D., 2006. Hazardous Alcohol Use: A Risk Factor for Non Adherence and Lack of Suppression in HIV Infection. J. Acquir. Immune Defic. Syndr. 1999 43, 411 417. Chung, T., Smith, G.T., Donovan, J.E., Windle, M., Faden, V.B., Chen, C.M., Martin, C.S., 2012. Drinking Frequency as a Brief Screen for Adolescent Alcohol Problems. Pediatrics 129, 205 212. Clarke, P.M., Wolstenholme, J.L., Wordsworth, S., 2010. Applied Methods of Cost effectiveness Analysis in Healthcare. Oxford Universit y Press. Cohen, E., Feinn, R., Arias, A., Kranzler, H.R., 2007. Alcohol treatment utilization: findings from the National Epidemiologic Survey on Alcohol and Related Conditions. Drug Alcohol Depend. 86, 214 221. Cohen, M.H., French, A.L., Benning, L., Kova cs, A., Anastos, K., Young, M., Minkoff, H., Hessol, N.A., 2002. Causes of death among women with human immunodeficiency virus infection in the era of combination antiretroviral therapy. Am. J. Med. 113, 91 98. Cook, J.A., Burke Miller, J.K., Cohen, M.H., Cook, R.L., Vlahov, D., Wilson, T.E., Golub, E.T., Schwartz, R.M., Howard, A.A., Ponath, C., Plankey, M.W., Levine, A.M., Levine, A., Grey, D.D., 2008. Crack cocaine, disease progression, and mortality in a multicenter cohort of HIV 1 positive women. AIDS Lond. Engl. 22, 1355 1363. Cook, R.L., Sereika, S.M., Hunt, S.C., Woodward, W.C., Erlen, J.A., Conigliaro, J., 2001. Problem drinking and medication adherence among persons with HIV infection. J. Gen. Intern. Med. 16, 83 88. Cook, R.L., Zhu, F., Belnap, B. H., Weber, K., Cook, J.A., Vlahov, D., Wilson, T.E., Hessol, N.A., Plankey, M., Howard, A.A., Cole, S.R., Sharp, G.B., Richardson, J.L., Cohen, M.H., 2009. Longitudinal trends in hazardous alcohol consumption among women with human immunodeficiency virus i nfection, 1995 2006. Am. J. Epidemiol. 169, 1025 1032. Cunningham, C.O., Sohler, N.L., Wong, M.D., Relf, M., Cunningham, W.E., Drainoni, M. L., Bradford, J., Pounds, M.B., Cabral, H.D., 2007. Utilization of health care services in hard to reach marginalize d HIV infected individuals. AIDS Patient Care STDs 21, 177 186.
100 Cunningham, W.E., Sohler, N.L., Tobias, C., Drainoni, M., Bradford, J., Davis, C., Cabral, H.J., Cunningham, C.O., Eldred, L., Wong, M.D., 2006. Health services utilization for people with HIV infection: comparison of a population targeted for outreach with the U.S. population in care. Med. Care 44, 1038 1047. Dawson, D.A., Goldstein, R.B., Grant, B.F., 2012. Factors associated with first utilization of different types of care for alcohol probl ems. J. Stud. Alcohol Drugs 73, 647 656. Dawson, D.A., Grant, B.F., Stinson, F.S., Chou, P.S., 2006. Estimating the effect of help seeking on achieving recovery from alcohol dependence. Addict. Abingdon Engl. 101, 824 834. Dawson, D.A., Grant, B.F., Stinson, F.S., Zhou, Y., 2005. Effectiveness of the derived Alcohol Use Disorders Identification Test (AUDIT C) in screening for alcohol use disorders and risk drinking in the US general population. Alcohol. Clin. Exp. Res. 29, 844 854. DHHS, 2013. Panel o n Antiretroviral Guidelines for Adults and Adolescents. Guidelines for the use of antiretroviral agents in HIV 1 infected adults and adolescents. Drummond, M.F., 2005. Methods for the Economic Evaluation of Health Care Programmes. Oxford University Press. Farnham, P.G., Sansom, S.L., Hutchinson, A.B., 2012. How Much Should We Pay for a New HIV Diagnosis? A Mathematical Model of HIV Screening in US Clinical Settings. Med. Decis. Making 32, 459 469. Fitzmaurice, G.M., Laird, N.M., Ware, J.H., 2012. Applied Lo ngitudinal Analysis. John Wiley & Sons. Freedberg, K.A., Losina, E., Weinstein, M.C., Paltiel, A.D., Cohen, C.J., Seage, G.R., Craven, D.E., Zhang, H., Kimmel, A.D., Goldie, S.J., 2001. The Cost Effectiveness of Combination Antiretroviral Therapy for HIV D isease. N. Engl. J. Med. 344, 824 831. Friedmann, P.D., 2013. Alcohol Use in Adults. N. Engl. J. Med. 368, 365 373. Friedmann, P.D., Saitz, R., Gogineni, A., Zhang, J.X., Stein, M.D., 2001. Validation of de to Helping Patients with 238.
101 Fuller, R.K., Branchey, L., Brightwell, D.R., Derman, R.M., Emrick, C.D., Iber, F.L., James, K.E., Lacoursiere, R.B., Lee, K.K., Lowenstam, I., 1986. Disulfiram treatment of alc oholism. A Veterans Administration cooperative study. JAMA 256, 1449 1455. Galvan, F.H., Bing, E.G., Fleishman, J.A., London, A.S., Caetano, R., Burnam, M.A., Longshore, D., Morton, S.C., Orlando, M., Shapiro, M., 2002. The prevalence of alcohol consumptio n and heavy drinking among people with HIV in the United States: results from the HIV Cost and Services Utilization Study. J. Stud. Alcohol 63, 179 186. Goldstein, R.B., Dawson, D.A., Grant, B.F., 2010. Antisocial Behavioral Syndromes in Adulthood and Alco hol Use Disorder Treatment Over Three Year Follow Up: Results From Wave 2 of the National Epidemiologic Survey on Alcohol and Related Conditions. J. Am. Psychiatr. Nurses Assoc. 16, 212 226. Gordon, A.J., Maisto, S.A., McNeil, M., Kraemer, K.L., Conigliaro , R.L., Kelley, M.E., Conigliaro, J., 2001. Three questions can detect hazardous drinkers. J. Fam. Pract. 50, 313 320. Govender, S., Otwombe, K., Essien, T., Panchia, R., de Bruyn, G., Mohapi, L., Gray, G., Martinson, N., 2014. CD4 Counts and Viral Loads o f Newly Diagnosed HIV Infected Individuals: Implications for Treatment as Prevention. PLoS ONE 9, e90754. Grant, B.F., 1997. Barriers to alcoholism treatment: reasons for not seeking treatment in a general population sample. J. Stud. Alcohol 58, 365 371. G ras, L., Kesselring, A.M., Griffin, J.T., van Sighem, A.I., Fraser, C., Ghani, A.C., Miedema, F., Reiss, P., Lange, J.M.A., de Wolf, F., ATHENA, Netherlands National Observational Cohort Study, 2007. CD4 cell counts of 800 cells/mm3 or greater after 7 year s of highly active antiretroviral therapy are feasible in most patients starting with 350 cells/mm3 or greater. J. Acquir. Immune Defic. Syndr. 1999 45, 183 192. Grella, C.E., Polinsky, M.L., Hser, Y.I., Perry, S.M., 1999. Characteristics of women only and mixed gender drug abuse treatment programs. J. Subst. Abuse Treat. 17, 37 44. Greub, G., Ledergerber, B., Battegay, M., Grob, P., Perrin, L., Furrer, H., Burgisser, P., Erb, P., Boggian, K., Piffaretti, J.C., Hirschel, B., Janin, P., Francioli, P., Flepp, M., Telenti, A., 2000. Clinical progression, survival, and immune recovery during antiretroviral therapy in patients with HIV 1 and hepatitis C virus coinfection: the Swiss HIV Cohort Study. Lancet 356, 1800 1805.
102 Group, P.M.R., 1998. Matching patients wi th alcohol disorders to treatments: Clinical implications from Project MATCH. J Ment Health 7, 589 602. Hahn, J.A., Samet, J.H., 2010. Alcohol and HIV Disease Progression: Weighing the Evidence. Curr. HIV/AIDS Rep. 7, 226 233. Hall, H.I., Song, R., Rhodes, P., Prejean, J., An, Q., Lee, L.M., Karon, J., Brookmeyer, R., Kaplan, E.H., McKenna, M.T., Janssen, R.S., HIV Incidence Surveillance Group, 2008. Estimation of HIV incidence in the United States. JAMA J. Am. Med. Assoc. 300, 520 529. Hansen, H., Alegria, M., Caban, C.A.M., Pena, M.M., Lai, S., Shrout, P., 2004. Drug Treatment, Health, and Social Service Utilization by Substance Abusing Women From a Community Based Sample. Med. Care Novemb. 2004 42, 1117 1124. Harris, K.M., Edlund, M.J., 2005. Use of Menta l Health Care and Substance Abuse Treatment Among Adults With Co occurring Disorders. Psychiatr. Serv. 56, 954 959. Hasin, D.S., 1994. Treatment/self help for alcohol related problems: relationship to social pressure and alcohol dependence. J. Stud. Alcoho l 55, 660 666. Waxman, R., Wainberg, M., Helzer, J., Johnston, B., 2013. Reducing heavy drinking in HIV primary care: a randomized trial of brief intervention, with a nd without technological enhancement. Addict. Abingdon Engl. 108, 1230 1240. Hatzenbuehler, M.L., Keyes, K.M., Narrow, W.E., Grant, B.F., Hasin, D.S., 2008. Racial/ethnic disparities in service utilization for individuals with co occurring mental health an d substance use disorders in the general population. J. Clin. Psychiatry 69, 1112 1121. Heather, N., Peters, T.J., Stockwell, T., 2001. International Handbook of Alcohol Dependence and Problems. Wiley. Hendershot, C.S., Stoner, S.A., Pantalone, D.W., Simon i, J.M., 2009. Alcohol use and antiretroviral adherence: Review and meta analysis. J. Acquir. Immune Defic. Syndr. 1999 52, 180. Hutton, H.E., McCaul, M.E., Chander, G., Jenckes, M.W., Nollen, C., Sharp, V.L., Erbelding, E.J., 2012. Alcohol Use, Anal Sex, and Other Risky Sexual Behaviors Among HIV Infected Women and Men. AIDS Behav. 1 11. Ilgen, M.A., Price, A.M., Burnett Zeigler, I., Perron, B., Islam, K., Bohnert, A.S.B., Zivin, K., 2011. Longitudinal predictors of addictions treatment utilization in trea tment naÃ¯ve adults with alcohol use disorders. Drug Alcohol Depend. 113, 215 221.
103 of liver disease in patients with HIV infection. The Lancet 377, 1198 1209. Kalichman , S.C., Grebler, T., Amaral, C.M., McNerey, M., White, D., Kalichman, M.O., Cherry, C., Eaton, L., 2013. Intentional non adherence to medications among HIV positive alcohol drinkers: prospective study of interactive toxicity beliefs. J. Gen. Intern. Med. 2 8, 399 405. Kaner, E.F.S., Dickinson, H.O., Beyer, F., Pienaar, E., Schlesinger, C., Campbell, F., Saunders, J.B., Burnand, B., Heather, N., 2009. The effectiveness of brief alcohol interventions in primary care settings: A systematic review. Drug Alcohol Rev. 28, 301 323. Kaslow, R.A., Blackwelder, W.C., Ostrow, D.G., Yerg, D., Palenicek, J., Coulson, A.H., Valdiserri, R.O., 1989. No evidence for a role of alcohol or other psychoactive drugs in accelerating immunodeficiency in HIV 1 positive individuals. A report from the Multicenter AIDS Cohort Study. JAMA 261, 3424 3429. Kessler RC, Aguilar Gaxiola S, Berglund PA, et al, 2001. PAtterns and predictors of treatment seeking after onset of a substance use disorder. Arch. Gen. Psychiatry 58, 1065 1071. Kieffer , T.L., Finucane, M.M., Nettles, R.E., Quinn, T.C., Broman, K.W., Ray, S.C., Persaud, D., Siliciano, R.F., 2004. Genotypic analysis of HIV 1 drug resistance at the limit of detection: virus production without evolution in treated adults with undetectable H IV loads. J. Infect. Dis. 189, 1452 1465. King, G., Zeng, L., 2001. Logistic Regression in Rare Events Data. Polit. Anal. 9, 137 163. KrÃ¤mer, A., Kretzschmar, M., Krickeberg, K., 2009. Modern Infectious Disease Epidemiology: Concepts, Methods, Mathematical Models, and Public Health. Springer New York. Kushi, L.H., Byers, T., Doyle, C., Bandera, E.V., McCullough, M., Gansler, T., Andrews, K.S., Thun, M.J., 2006. American Cancer Society Guidelines on Nutrition and Physical Activity for Cancer Prevention: Redu cing the Risk of Cancer With Healthy Food Choices and Physical Activity. CA. Cancer J. Clin. 56, 254 281. Litten, R.Z., Castle, I. J.P., Falk, D., Ryan, M., Fertig, J., Chen, C.M., Yi, H., 2013. The Placebo Effect in Clinical Trials for Alcohol Dependence: An Exploratory Analysis of 51 Naltrexone and Acamprosate Studies. Alcohol. Clin. Exp. Res. 37, 2128 2137.
104 Liu, C., Weber, K., Robison, E., Hu, Z., Jacobson, L.P., Gange, S.J., 2006. Assessing the effect of HAART on change in quality of life among HIV infe cted women. AIDS Res. Ther. 3, 6. Maartens, G., Celum, C., Lewin, S.R., 2014. HIV infection: epidemiology, pathogenesis, treatment, and prevention. The Lancet 384, 258 271. Maisel, N.C., Blodgett, J.C., Wilbourne, P.L., Humphreys, K., Finney, J.W., 2013. M eta analysis of naltrexone and acamprosate for treating alcohol use disorders: when are these medications most helpful? Addiction 108, 275 293. Mason, B.J., Goodman, A.M., Dixon, R.M., Hameed, M.H., Hulot, T., Wesnes, K., Hunter, J.A., Boyeson, M.G., 2002. A pharmacokinetic and pharmacodynamic drug interaction study of acamprosate and naltrexone. Neuropsychopharmacol. Off. Publ. Am. Coll. Neuropsychopharmacol. 27, 596 606. Mason, B.J., Quello, S., Goodell, V., Shadan, F., Kyle, M., Begovic, A., 2014. Gabape ntin treatment for alcohol dependence: a randomized clinical trial. JAMA Intern. Med. 174, 70 77. McGinnis, K.A., Fultz, S.L., Skanderson, M., Conigliaro, J., Bryant, K., Justice, A.C., 2006. Hepatocellular carcinoma and non f HIV, hepatitis C infection, and alcohol abuse. J. Clin. Oncol. Off. J. Am. Soc. Clin. Oncol. 24, 5005 5009. Mellors, J.W., Munoz, A., Giorgi, J.V., Margolick, J.B., Tassoni, C.J., Gupta, P., Kingsley, L.A., Todd, J.A., Saah, A.J., Detels, R., Phair, J.P. , Rinaldo, J., Charles R., 1997. Plasma Viral Load and CD4+ Lymphocytes as Prognostic Markers of HIV 1 Infection. Ann. Intern. Med. 126, 946 954. Miller, M.F., Haley, C., Koziel, M.J., Rowley, C.F., 2005. Impact of hepatitis C virus on immune restoration i n HIV infected patients who start highly active antiretroviral therapy: a meta analysis. Clin. Infect. Dis. Off. Publ. Infect. Dis. Soc. Am. 41, 713 720. Mojtabai, R., 2005. Use of specialty substance abuse and mental health services in adults with substan ce use disorders in the community. Drug Alcohol Depend. 78, 345 354. Mojtabai, R., Olfson, M., Mechanic, D., 2002. Perceived need and help seeking in adults with mood, anxiety, or substance use disorders. Arch. Gen. Psychiatry 59, 77 84. Moos, R.H., Moos, B.S., 2004. Long Term Influence of Duration and Frequency of Participation in Alcoholics Anonymous on Individuals With Alcohol Use Disorders. J. Consult. Clin. Psychol. 72, 81 90.
105 Moos, R.H., Moos, B.S., 2006. Rates and predictors of relapse after natural and treated remission from alcohol use disorders. Addiction 101, 212 222. Mowbray, O., 2014. The moderating role of social networks in the relationship between alcohol consumption and treatment utilization for alcohol related problems. J. Subst. Abuse Trea t. 46, 597 601. Murphy, S., Xu, J., Kochanek, K., Division of Vital Statistics, 2013. Deaths: Final Data for 2010. Natl. Vital Stat. Rep. 61. Naranjo, C.A., Sellers, E.M., 1986. Clinical assessment and pharmacotherapy of the alcohol withdrawal syndrome. Re cent Dev. Alcohol. Off. Publ. Am. Med. Soc. Alcohol. Res. Soc. Alcohol. Natl. Counc. Alcohol. 4, 265 281. Neblett, R.C., Hutton, H.E., Lau, B., McCaul, M.E., Moore, R.D., Chander, G., 2011. Alcohol Consumption Among HIV Infected Women: Impact on Time to An tiretroviral Therapy and Survival. J. Womens Health 20, 279 286. Birch, D., Schulte, B., Schmidt, C., Reimer, J., Kaner, E., 2014. The Impact of Brief Alcohol Interventions in Primary Healthcare: A Systematic Review of Reviews. Alcohol Alcohol 49, 66 78. Ouimette, P.C., Finney, J.W., Gima, K., Moos, R.H., 1999. A Comparative Evaluation of Substance Abuse Treatment III. Examining Mechanisms Underlying Patient Treatment Matching Hypotheses for 12 Step and Cognitive Behavioral Treatments for substance Abuse. Alcohol. Clin. Exp. Res. 23, 545 551. Paltiel, A.D., Weinstein, M.C., Kimmel, A.D., Seage, G.R., Losina, E., Zhang, H., Freedberg, K.A., Walensky, R.P., 2005. Expanded Screening for HIV in the United Sta tes An Analysis of Cost Effectiveness. N. Engl. J. Med. 352, 586 595. Parry, C.D., Morojele, N.K., Myers, B.J., Kekwaletswe, C.T., Manda, S.O., Sorsdahl, K., Ramjee, G., Hahn, J.A., Rehm, J., Shuper, P.A., 2014. Efficacy of an alcohol focused interventio n for improving adherence to antiretroviral therapy (ART) and HIV treatment outcomes a randomised controlled trial protocol. BMC Infect. Dis. 14, 500. Parsons, J.T., Golub, S.A., Rosof, E., Holder, C., 2007. Motivational Interviewing and Cognitive Behavi oral Intervention to Improve HIV Medication Adherence Among Hazardous Drinkers: A Randomized Controlled Trial. JAIDS J. Acquir. Immune Defic. Syndr. 46, 443 450.
106 Pineda, J.A., Romero GÃ³mez, M., DÃaz GarcÃa, F., GirÃ³n GonzÃ¡lez, J.A., Montero, J.L., Torre Ci sneros, J., Andrade, R.J., GonzÃ¡lez Serrano, M., Aguilar, J., Aguilar Guisado, M., Navarro, J.M., SalmerÃ³n, J., Caballero Granado, F.J., GarcÃa GarcÃa, J.A., Grupo Andaluz para el Estudio de las Enfermedades Infecciosas, Grupo Andaluz para el Estudio del H Ãgado, 2005. HIV coinfection shortens the survival of patients with hepatitis C virus related decompensated cirrhosis. Hepatol. Baltim. Md 41, 779 789. Prabhu, V.S., Farnham, P.G., Hutchinson, A.B., Soorapanth, S., Heffelfinger, J.D., Golden, M.R., Brook s, J.T., Rimland, D., Sansom, S.L., 2011. Cost Effectiveness of HIV Screening in STD Clinics, Emergency Departments, and Inpatient Units: A Model Based Analysis. PLoS ONE 6, e19936. Purcell, D.W., Johnson, C.H., Lansky, A., Prejean, J., Stein, R., Denning, P., Gau, Z., Weinstock, H., Su, J., Crepaz, N., 2012. Estimating the population size of men who have sex with men in the United States to obtain HIV and syphilis rates. Open AIDS J. 6, 98 107. Ray, L.A., Chin, P.F., Miotto, K., 2010. Naltrexone for the tr eatment of alcoholism: clinical findings, mechanisms of action, and pharmacogenetics. CNS Neurol. Disord. Drug Targets 9, 13 22. Room, R., Babor, T., Rehm, J., 2005. Alcohol and public health. The Lancet 365, 519 530. Rosen, D., Tolman, R.M., Warner, L.A., 2004. Low abuse and mental health services. J. Health Care Poor Underserved 15, 206 219. RÃ¶sner, S., Hackl Herrwerth, A., Leucht, S., Lehert, P., Vecchi, S., Soyka, M., 2010. Acamprosate for alcohol dependence. Cochrane Dat abase Syst. Rev. CD004332. Rothlind, J.C., Greenfield, T.M., Bruce, A.V., Meyerhoff, D.J., Flenniken, D.L., Lindgren, J.A., Weiner, M.W., 2005. Heavy alcohol consumption in individuals with HIV infection: effects on neuropsychological performance. J. Int. Neuropsychol. Soc. JINS 11, 70 83. Rubin, D.B., 1987. Frontmatter. In: Multiple Imputation for Nonresponse in Surveys. John Wiley & Sons, Inc., pp. i xxix. Rubinsky, A.D., Dawson, D.A., Williams, E.C., Kivlahan, D.R., Bradley, K.A., 2013. AUDIT C scores as a scaled marker of mean daily drinking, alcohol use disorder severity, and probability of alcohol dependence in a U.S. general population sample of drinkers. Alcohol. Clin. Exp. Res. 37, 1380 1390.
107 Salmon Ceron, D., Lewden, C., Morlat, P., BÃ©vilacqua, S., Jougla, E., Bonnet, F., HÃ©ripret, L., Costagliola, D., May, T., ChÃªne, G., 2005. Liver disease as a major cause of death among HIV infected patients: role of hepatitis C and B viruses and alcohol. J. Hepatol. 42, 799 805. Samet, J.H., Horton, N.J., Meli, S., Dukes, K., Tripps, T., Sullivan, L., Freedberg, K.A., 2005. A randomized controlled trial to enhance antiretroviral therapy adherence in patients with a history of alcohol problems. Antivir. Ther. 10, 83 93. Samet, J.H., Phillips, S.J., Horton, N.J., T raphagen, E.T., Freedberg, K.A., 2004. Detecting alcohol problems in HIV infected patients: use of the CAGE questionnaire. AIDS Res. Hum. Retroviruses 20, 151 155. Samji, H., Cescon, A., Hogg, R.S., Modur, S.P., Althoff, K.N., Buchacz, K., Burchell, A.N., Cohen, M., Gebo, K.A., Gill, M.J., Justice, A., Kirk, G., Klein, M.B., Korthuis, P.T., Martin, J., Napravnik, S., Rourke, S.B., Sterling, T.R., Silverberg, M.J., Deeks, S., Jacobson, L.P., Bosch, R.J., Kitahata, M.M., Goedert, J.J., Moore, R., Gange, S.J., for The North American AIDS Cohort Collaboration on Research and Design (NA ACCORD) of IeDEA, 2013. Closing the Gap: Increases in Life Expectancy among Treated HIV Positive Individuals in the United States and Canada. PLoS ONE 8, e81355. Sanders, G.D., Ba youmi, A.M., Sundaram, V., Bilir, S.P., Neukermans, C.P., Rydzak, C.E., Douglass, L.R., Lazzeroni, L.C., Holodniy, M., Owens, D.K., 2005. Cost Effectiveness of Screening for HIV in the Era of Highly Active Antiretroviral Therapy. N. Engl. J. Med. 352, 570 585. Schafer, J.L., 1999. Multiple imputation: a primer. Stat. Methods Med. Res. 8, 3 15. S, H., Bowe, T., 2008. Predictors of initiation and engagement in VA substance use disorder (SUD) treatment. Psychol. Serv. 5, 228 238. Smith, C.J., Ryom, L., Weber, R., Morlat, P., Pradier, C., Reiss, P., Kowalska, J.D., de Wit, S., Law, M., el Sadr, W., Kirk, O., Friis Phillips, A.N., Sabin, C.A., Lundgren, J.D., 2014. Trends in underlying causes of death in people with HIV from 19 99 to 2011 (D:A:D): a multicohort collaboration. The Lancet 384, 241 248. Sobell, L.C., Cunningham, J.A., Sobell, M.B., 1996. Recovery from alcohol problems with and without treatment: prevalence in two population surveys. Am. J. Public Health 86, 966 972. Sobell, L.C., Ellingstad, T.P., Sobell, M.B., 2000. Natural recovery from alcohol and drug problems: methodological review of the research with suggestions for future directions. Addict. Abingdon Engl. 95, 749 764.
108 Sobell, L.C., Sobell, M.B., Toneatto, T. , Leo, G.I., 1993. What Triggers the Resolution of Alcohol Problems Without Treatment?*. Alcohol. Clin. Exp. Res. 17, 217 224. Soto, B., SÃ¡nchez Quijano, A., Rodrigo, L., del Olmo, J.A., GarcÃa Bengoechea, M., HernÃ¡ndez Quero, J., Rey, C., Abad, M.A., Rodr Ãguez, M., Sales Gilabert, M., GonzÃ¡lez, F., MirÃ³n, P., Caruz, A., Relimpio, F., Torronteras, R., Leal, M., Lissen, E., 1997. Human immunodeficiency virus infection modifies the natural history of chronic parenterally acquired hepatitis C with an unusually rapid progression to cirrhosis. J. Hepatol. 26, 1 5. Spanagel, R., Vengeliene, V., Jandeleit, B., Fischer, W. N., Grindstaff, K., Zhang, X., Gallop, M.A., Krstew, E.V., Lawrence, A.J., Kiefer, F., 2014. Acamprosate Produces Its Anti Relapse Effects Via Ca lcium. Neuropsychopharmacology 39, 783 791. Spire, B., Lucas, G.M., Carrieri, M.P., 2007. Adherence to HIV treatment among IDUs and the role of opioid substitution treatment (OST). Int. J. Drug Policy 18, 262 270. 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. Alcohol. Clin. Exp. Res. 29, 837 843. Substance Abuse and Mental Health Services Administration, 2013. Results from t he 2012 National Survey on Drug Use and Health: Summary of National Findings. Tandon, S.D., Cluxton Keller, F., Leis, J., Le, H. N., Perry, D.F., 2012. A comparison of three screening tools to identify perinatal depression among low income African America n women. J. Affect. Disord. 136, 155 162. Thomas, S.L., Newell, M. L., Peckham, C.S., Ades, A.E., Hall, A.J., 1998. A review of hepatitis C virus (HCV) vertical transmission: risks of transmission to infants born to mothers with and without HCV viraemia or human immunodeficiency virus infection. Int. J. Epidemiol. 27, 108 120. Thun, M.J., Peto, R., Lopez, A.D., Monaco, J.H., Henley, S.J., Heath, C.W., Jr, Doll, R., 1997. Alcohol consumption and mortality among middle aged and elderly U.S. adults. N. Engl. J . Med. 337, 1705 1714. Tighe, E., Saxe, L., 2006. Community Based Substance Abuse Reduction and the Gap between Treatment Need and Treatment Utilization: Analysis of Data from the 312. Timko , C., Debenedetti, A., Billow, R., 2006. Intensive referral to 12 Step self help groups and 6 month substance use disorder outcomes. Addict. Abingdon Engl. 101, 678 688.
109 Tsui, J.I., Saitz, R., Cheng, D.M., Nunes, D., Libman, H., Alperen, J.K., Samet, J.H., 2007. Awareness of Hepatitis C Diagnosis is Associated with Less Alcohol Use Among Persons Co Infected with HIV. J. Gen. Intern. Med. 22, 822 825. Tucker, J.S., Burnam, M.A., Sherbourne, C.D., Kung, F. Y., Gifford, A.L., 2003. Substance use and mental hea lth correlates of nonadherence to antiretroviral medications in a sample of patients with human immunodeficiency virus infection. Am. J. Med. 114, 573 580. Ulett, K.B., Willig, J.H., Lin, H. Y., Routman, J.S., Abroms, S., Allison, J., Chatham, A., Raper, J .L., Saag, M.S., Mugavero, M.J., 2009. The Therapeutic Implications of Timely Linkage and Early Retention in HIV Care. AIDS Patient Care STDs 23, 41 49. Velasquez, M.M., von Sternberg, K., Johnson, D.H., Green, C., Carbonari, J.P., Parsons, J.T., 2009. Red ucing sexual risk behaviors and alcohol use among HIV positive men who have sex with men: a randomized clinical trial. J. Consult. Clin. Psychol. 77, 657 667. Volberding, P.A., Deeks, S.G., 2010. Antiretroviral therapy and management of HIV infection. The Lancet 376, 49 62. Vo, T.T.N., Ledergerber, B., Keiser, O., Hirschel, B., Furrer, H., Battegay, M., Cavassini, M., Bernasconi, E., Vernazza, P., Weber, R., 2008. Durability and Outcome of Initial Antiretroviral Treatments Received during 2000 2005 by Patie nts in the Swiss HIV Cohort Study. J. Infect. Dis. 197, 1685 1694. Wada, N., Jacobson, L.P., Cohen, M., French, A., Phair, J., MuÃ±oz, A., 2014. Cause specific mortality among HIV infected individuals, by CD4+ cell count at HAART initiation, compared with H IV uninfected individuals: AIDS 28, 257 265. Walensky, R.P., Ross, E.L., Kumarasamy, N., Wood, R., Noubary, F., Paltiel, A.D., Nakamura, Y.M., Godbole, S.V., Panchia, R., Sanne, I., Weinstein, M.C., Losina, E., Mayer, K.H., Chen, Y.Q., Wang, L., McCauley, M., Gamble, T., Seage, G.R., 3rd, Cohen, M.S., Freedberg, K.A., 2013. Cost effectiveness of HIV treatment as prevention in serodiscordant couples. N. Engl. J. Med. 369, 1715 1725. Weber, R., Ruppik, M., Rickenbach, M., Spoerri, A., Furrer, H., Battegay, M. , Cavassini, M., Calmy, A., Bernasconi, E., Schmid, P., Flepp, M., Kowalska, J., Ledergerber, B., Swiss HIV Cohort Study (SHCS), 2013. Decreasing mortality and changing patterns of causes of death in the Swiss HIV Cohort Study. HIV Med. 14, 195 207. Weisne r, C., 1993. Toward an Alcohol Treatment Entry Model: A Comparison of Problem Drinkers in the General Population and in Treatment. Alcohol. Clin. Exp. Res. 17, 746 752.
11 0 Weisner, C., Matzger, H., Tam, T., Schmidt, L., 2002. Who Goes to Alcohol and Drug Trea tment? Understanding Utilization within the Context of Insurance. J. Stud. Alcohol Drugs 63, 673. Weisner, C., Schmidt, L.A., 1995. Expanding the frame of health services research in the drug abuse field. Health Serv. Res. 30, 707 726. Willenbring, M.L., M assey, S.H., Gardner, M.B., 2009. Helping patients who drink too much: an evidence based guide for primary care clinicians. Am. Fam. Physician 80, 44 50. Wu, E.S., Metzger, D.S., Lynch, K.G., Douglas, S.D., 2011. Association between Alcohol Use and HIV Vir al Load. J. Acquir. Immune Defic. Syndr. 1999 56, 129 130. Wu, L. T., Ringwalt, C.L., Williams, C.E., 2003. Use of Substance Abuse Treatment Services by Persons With Mental Health and Substance Use Problems. Psychiatr. Serv. 54, 363 369. Ye, Y., Kaskutas, L.A., 2009. Using propensity scores to adjust for selection bias when assessing the effectiveness of Alcoholics Anonymous in observational studies. Drug Alcohol Depend. 104, 56 64. Zeger, S.L., Liang, K. Y., Albert, P.S., 1988. Models for Longitudinal Data : A Generalized Estimating Equation Approach. Biometrics 44, 1049 1060.
111 BIOGRAPHICAL SKETCH Xing di hu received his Bachelor of E ngineering in mi croelectronics from the Soochow University in China in 2007. He received Master of Science in e l ectrical e ngineering in December of 2008 and Ph.D. in e pidemiology from the University of Florida in the spring of 2015.