The results of this study suggest the need for further research to better understand the effects of different types of sexual orientation victimization experiences on HIV sexual risk behavior among Black South African MSM.In the United States , alcohol related mortality represents ~10% of all deaths among working-age adults. Recent reports have highlighted the increase in alcohol related mortality due to alcohol poisonings and alcohol-related chronic liver disease. In addition, alcohol misuse can exacerbate chronic health conditions including hypertension and increase the risk for a variety of cancers and other liver diseases unrelated to alcohol. Indeed, with the advent of new effective treatments for hepatiThis C viral infection and the potential decrease in the burden of HCV related liver cirrhosis, alcohol use may become the leading cause of liver disease in the US. Alcohol misuse, which includes hazardous/heavy use, binge drinking, and alcohol use disorder , is prevalent and particularly risky among people with HIV . Blood alcohol levels for a given quantity of alcohol appear to be higher in PWH compared to those without HIV infection, planting racks especially among individuals who are not virologically suppressed. In addition, alcohol misuse decreases optimal engagement in the HIV care continuum, has been associated with lower use of antiretroviral therapy , decreased rates of adherence and viral suppression, and increased mortality.
Alcohol misuse has also been associated with an increase in HIV transmission risk behaviors, and potentially increased hepatotoxicity in combination with protease inhibitors. In a recent multi-site cohort study, alcohol misuse was associated with worse retention in HIV care, which has been associated with poorer treatment outcomes and increased mortality. Furthermore, as PWH age, alcohol misuse may complicate other comorbid conditions such as hypertension and diabetes, and increase the risk for stroke and cognitive decline. Understanding trajectories of alcohol use over time, including predictors of persistence and change in use may assist providers in identifying individuals at particular risk for development and/or maintenance of alcohol misuse, and therefore worsening HIV and other treatment outcomes. Thus, we examined longitudinal trajectories of alcohol use in a large multi-site clinical cohort of PWH and determined clinical and potentially modifiable predictors of each specific trajectory.CNICS collects longitudinal clinical, demographic, laboratory, and medication data from electronic health records and other sources at each site. In addition, the CNICS data repository integrates patient reported behaviors obtained at 4–6 month intervals at clinical care visits through touch-screen based computer-assisted assessments that are self-administered. In some of the clinics, the results are subsequently fed back to the providers . Patients report using validated instruments on alcohol , substance use , depressive and anxiety symptoms , medication adherence, and other health domains.
All analysis were stratified by sex, given the previously reported differences in alcohol consumption trajectories and potential differences in alcohol metabolism by sex. Descriptive staThistics were calculated for alcohol consumption trajectory and the exposures. The main analysis was conducted in 3 stages. First, we used a multinomial logistic model to assess the probability of non-drinking, moderate alcohol use, or alcohol misuse at the first CASI using the covariates listed above, adjusting for age, race, and site. Second, we used a finite mixture model with multinomial distributions in order to classify individuals according to latent alcohol consumption trajectories . Optimal number of classes were determined by fitting models with 3 to 9 classes and assessing stability using 1000 bootstrap resamples and by minimizing the Integrated Classification Likelihood Criterion . Appendix Table 1 shows the goodness of fit and goodness of classification staThistics, along with the combined ICL-BIC. Finally, we examined baseline predictors of latent alcohol consumption trajectories using Vermunt’s 3-step approach. Each predictor was included in separate models adjusted for age, race and site. We also ran a model that included all variables to assess whether associations varied due to correlations with other variables. To account for missing data we used multiple imputation by chained equations where the imputations were based on other available covariates including the outcome of drinking behaviors, generating 21 imputed data sets and pooling the coefficients using Rubin’s Formula. In reporting our results, we do not make any mention of staThistical significance since we are not explicitly testing any hypothesis.
We report 95% confidence intervals to provide an estimate for the precision of our estimates. All analyses were conducted using R v3.2.2 and Mplus version 7.4.Alcohol consumption categories at first CASI differed by sex. Thirty-three percent of men did not drink at their first CASI, while 40% drank moderately and 27% had alcohol misuse. In women, 56% did not drink at their first CASI while 26% drank moderately and 18% had alcohol misuse. Table 2 shows the predictors of alcohol consumption at first CASI for men. Men aged 38 and above were more likely to not drink and less likely to be report alcohol misuse, as compared to men aged 19 to 37. Men with depressive or anxiety symptoms were more likely to report non-drinking or alcohol misuse. Illicit drug and marijuana use were also associated with alcohol misuse, while HCV infection and injection drug use were associated with non-drinking. Men with undetectable viral loads were also more likely to report not drinking and less likely to report alcohol misuse. Finally, men who have sex with men were less likely to report not drinking. After adjusting for all other variables in the model, there was a decrease in the probability of alcohol misuse for men that injected drugs . Table 2 shows the predictors of alcohol consumption at first CASI in women. Older women were more likely to report not drinking. Depressive and anxiety symptoms, illicit drug use and marijuana use were associated with higher odds of alcohol misuse. Women that were HCV infected were more likely to report non-use or misuse. After adjusting for all other covariates, there was no longer an increased probability of alcohol misuse in HCV infected women .Figure 1 shows the predicted trajectories of alcohol consumption, along with the prevalence of each trajectory. Men that did not drink at first CASI were either in the stable non-drinking category or increased their drinking to moderate use or alcohol misuse . Men that drank moderately at first CASI were classified into stable moderate use , reduction or increase in alcohol use. Finally, men starting with alcohol misuse were classified into stable misuse with a long-term trend towards decreased drinking and those that decreased alcohol misuse more quickly , either by alcohol cessation or reduction to moderate use. For women, the predicted categories and distribution of women across categories were similar to those observed for men, with the exception that we did not have enough data to support a third class of women who drank moderately at first CASI and increased their drinking to levels of misuse. These women were classified in the trajectory of stable moderate drinking, where there is a small trend towards increased alcohol misuse . Appendix Figures 1 and 2 show the observed alcohol consumption trajectories weighted by the probability of membership into each class, and Appendix Figures 3 and 4 show observed proportions of individuals in each class and alcohol consumption category every 6 months. All show good agreement with the predicted categories and a large amount of variability within each trajectory, racks industries as some patients change categories frequently during the study.In this longitudinal study of PWH across 7 HIV clinics in the US, we found that alcohol drinking trajectories varied widely. Many individuals had stable drinking trajectories— remaining persistently in one of the three categories of none, moderate, or alcohol misuse.
However, among others, alcohol use changed over time. Among men, approximately 25% who did not drink or drank moderately at first CASI increased their alcohol use, while 22% with moderate or alcohol misuse reduced their use. Among women, 13% who did not drink or drank moderately at first CASI increased their use, while 29% with alcohol misuse or moderate use decreased their use. Given the persistence of alcohol misuse in approximately 20% of all men and 13% of all women, and the escalation of alcohol use among those with initial non- or moderate use, these results reinforce the need to routinely screen for alcohol use in HIV clinical encounters, even among those who previously reported no alcohol use. Despite the potential negative consequences of alcohol misuse among PWH, there have been only a few studies examining their trajectories of alcohol use and how they change their alcohol consumption patterns over time. These studies have focused on specific groups of PWH, including women, veterans, and veteran men who have sex with men. Two of the studies described drinking trajectories as defined by drinking scores for the entire duration of the study and then in separate analyses examined associations with alcohol misuse at each visit, while a third study modeled trajectories and their predictors but only in women. There was, therefore, a need for analyses that jointly model alcohol use trajectories and their predictors, allowing for an understanding of correlates of complex trajectories. Our finite mixture modeling method allowed for this characterization and description of drinking trajectories and their determinants. Overall, in both men and women, increasing age was associated with decreasing alcohol use. This is consistent with other cohort studies in the general population that demonstrate that alcohol use decreases with age. Among men, alcohol misuse was associated with increased odds of depressive and anxiety symptoms, former and current illicit drug use, marijuana use and detectable viral loads. These findings are consistent with numerous other studies that demonstrate that alcohol misuse frequently co-occurs with other substance use and mental health disorders and that those with alcohol misuse are less likely to be virologically suppressed. Of interest, in our trajectory analysis, both depressive and anxiety symptoms were associated with reducing alcohol consumption to a lower category of use. Alcohol reduction may have led to a decrease in these mental health symptoms, or a decrease in depressive and anxiety symptoms may have resulted in decreased alcohol use if an individual was self-medicating mental health symptoms with alcohol use. It is also possible that changes in mental health symptoms or drug use associated with changes in alcohol trajectories may have been mediated by entry into psychiatric care or alcohol and drug treatments, which were not systematically captured in this clinical cohort. This may be secondary to messaging from providers that alcohol use should be reduced among individuals with HCV infection or secondary to a general worsening in health status associated with the infection that may lead to a reduction in alcohol consumption. Marijuana use was associated with maintaining alcohol misuse in men and women. This finding is consistent with a recent trajectory analysis of marijuana consumption where alcohol misuse was associated with increasing marijuana use over time. An analysis conducted in young adults found that marijuana and alcohol consumption trajectories in the transition to adulthood mirrored each other, which may reflect the patterns we are observing in a somewhat older population. Factors associated with alcohol misuse among women in this sample are similar to findings from the Women’s Interagency HIV Study examining hazardous alcohol use over an 11-year period. Consistent with their study, we found that depressive symptoms , current and former illicit drug use and HCV infection were associated with alcohol misuse. However, in contrast to a later study in WIHS examining alcohol consumption trajectory patterns, we did not find that any of these clinical comorbidities was associated with our trajectory of persistent misuse whereas their study found that depressive symptoms, HCV infection and cocaine use were all associated with this trajectory. This difference may be due issues of staThistical power, given that our sample size of women was less than half of the WIHS sample. This study has several strengths. Our sample included a large cohort of patients in routine ambulatory HIV clinical care throughout the US, which increases the generalizability of our results. Second, patients were followed at approximately 4–6 month intervals, allowing for a detailed measurement of medium term alcohol consumption trajectories. Third, use of a finite mixture modeling approach where trajectories and their determinants are modeled in the same framework allows for a better and more integrated description of the alcohol burden in PWLH. Finite mixture models are flexible enough to accommodate outcomes of continuous and categorical nature.