Importantly, we examined the association of alcohol use and ART adherence in the context of other important variables such as use of other substances , major depressive disorder , global cognitive function. We hypothesized that PLWHA who are at-risk drinkers will have lower ART adherence than PLWHA who are not at-risk drinkers, and that at-risk alcohol use will still be significantly predictive of ART adherence when controlling for substance use, MDD, and global cognitive function. Participants consisted of 535 HIV-infected adults enrolled in NIH-funded research studies at the University of California, San Diego HIV Neuro behavioral Research Program from 2003 to 2015. The current study is a secondary analysis of existing data from each participant’s baseline visit at the HNRP. All participants were receiving ART at the time of the visit and reported drinking alcohol in the previous 30 days. Measures Non-adherence to ART was determined based on 1) response to part of the self-report AIDS Clinical Trial Group Questionnaire indicating any missed ART doses in the previous four days, or 2) detectable plasma HIV RNA .Alcohol use was measured using the HNRP Substance Use History form, which is a timeline follow-back measure of the amount of daily alcohol used in the previous 30 days. Participants were stratified based on the daily limit defined by the NIAAA guideline for individuals at risk for developing an AUD: >3 drinks per day for women and >4 drinks per day for men, hereafter referred to as “at-risk drinking.” Individuals who consume more than these daily limits are at a near fourfold increase risk of developing alcohol abuse and a seven fold increase risk of developing alcohol dependence compared to those who drink within those limits. Current and lifetime substance use disorders and major depressive disorder were identified via the Composite International Diagnostic Interview .
Global cognitive function was measured via a standardized battery of well-validated neurocognitive tests [see Heaton et al. for descriptions of tests in this battery]. For all neurocognitive measures,vertical grow shelf raw scores were converted to demographically adjusted T scores used to construct one summary Global Mean T score. Unadjusted association between at-risk drinking and adherence was assessed using simple logistic regression. A multi-variable logistic model was used to regress the adherence measure on at-risk drinking adjusted for covariates. Potential covariates included demographic characteristics ; current AUD and current cannabis use; lifetime AUD; lifetime substance use disorder for cannabis, cocaine, hallucinogens, inhalants, methamphetamine, opioids, phencyclidine , sedatives, and other substances combined; current and lifetime major depressive disorder ; global cognitive function; estimated years of HIV infection; and ART regimen type. Current use disorders for substances other than alcohol and cannabis were not considered in this analysis because of low frequencies of participants with such diagnoses. Covariates were considered for inclusion into a multi-variable model if they either showed an association with the adherence measure or differed between drinking level groups at a 0.10 significance level. Group comparisons were done with two-sample t-tests for numeric covariates and chi-square or Fisher’s exact tests for categorical covariates. Backward model selection was applied and the final model was chosen to include only the predictor of primary interest and the covariates with pvalues less than 0.05. Odds ratio was used as the effect size for the strength of these associations, such that OR>1 would indicate a predictor’s association with higher odds of adherence. All analyses were performed using R version 3.2.1 statistical software .Seventy-nine participants endorsed missing an ART dose in the last four days, 46 of whom also had detectable plasma viral load. An additional 192 participants had detectable plasma viral load without reporting a missed ART dose in the last four days. Based on our criteria determining ART non-adherence , 271 participants were identified as ART non-adherent.
See Table 1 for a full list of sample characteristics. Approximately a quarter of the cohort met the criterion for at-risk drinking . Table 2 shows comparisons of demographic, psychiatric, and substance use characteristics between the two drinking level groups. Participants meeting the criterion for at-risk drinking were younger, less educated, and more likely to be diagnosed with lifetime alcohol use and lifetime cocaine use disorders. Table 3 lists results of univariable and multi-variable analyses. In unadjusted analyses, the at-risk drinking group was associated with significantly lower odds of adherence . 57.1% of the participants in the at-risk drinking group were identified as non-adherent compared to 40.3% of the participants in the not at-risk drinking group . Among covariates, older age, greater education, absence of lifetime cocaine substance use disorder and absence of lifetime MDD were associated with greater adherence at a 0.10 significance level. These covariates were included into multi-variable model selection along with three additional covariates that differed by group at a 0.10 level . Following the stepwise procedure, the final model included only at-risk drinking and education. The effects for the remaining covariates did not achieve a 0.05 significance level in the multi-variable analysis. The adjusted effect of at-risk drinking remained significant . Education wasthe only other significant predictor, with OR=1.09, p=0.009 per one year increase. See Figure 2 for odds ratios and corresponding confidence intervals for at-risk drinking and education in the final model. Understanding the factors associated with ART non-adherence is critical for developing strategies to improve patient outcomes for those living with HIV/AIDS. Although alcohol use is broadly associated with worse ART adherence among PLWHA, there is little existing literature on the level of alcohol use that portends ART non-adherence. Results of the current study showed that persons with lower education and persons who meet the NIAAA criteria for at-risk drinking, which indicate consumption of a high level of alcohol per day, are more likely to be ART non-adherent compared to not-at-risk drinkers. This finding is consistent with the studies that have found a positive relationship between level of alcohol use and non-adherence to medications in samples of HIV-positive and HIV-negative participants . The most notable of those found that regardless of HIV status, binge drinking veterans were more likely to be non-adherent to medications on drinking days, post-drinking days, and non-drinking days compared to non-binge drinking veterans .
They also found a temporal association within each drinking level group, meaning that participants were most likely to be non-adherent to medications on drinking days, followed by post-drinking days and non-drinking days. For non-binge drinkers, this trend was significantly stronger in HIV+ participants compared to matched HIV- participants. For binge drinkers, the strength of the trend was comparable for HIV+ and HIV- participants,cannabis grow indoor suggesting that PLWHA may be more susceptible to medication non-adherence at lower levels of alcohol consumption compared to HIV- counterparts. The current study supports and augments these previous findings by demonstrating a similar alcohol-adherence association using the widely accepted NIAAA guideline that defines a daily level of alcohol consumption that puts individuals at risk for developing an AUD. Research has shown that individuals who drink within the NIAAA limits are at a much lower risk of having an AUD compared to those who drink beyond those limits . These standard NIAAA criteria are easy to assess and are a consistent method of stratifying alcohol use severity. The criteria for binge drinking, on the other hand, are defined differently by different governmental organizations , welcoming inconsistencies across studies that use binge drinking to stratify drinking level groups. The current study demonstrates that the daily limit defined by the NIAAA guideline for at-risk drinking provides a meaningful demarcation of drinking level in the context of ART adherence among PLWHA. Notably, after consideration of all potential covariates, education was the only other significant predictor in the final logistic regression model besides drinking level. The results showed that at-risk drinking was a better predictor of ART adherence than lifetime cocaine use and methamphetamine use. Although cocaine and methamphetamine use have been shown to have associations with ART adherence , there are several potential reasons why the results of this study found these relationships non-significant. First, alcohol is more widely used in our sample, as the studies conducted at the HNRP generally exclude participants who are current users of drugs other than alcohol and marijuana. Second, it is more socially and societally acceptable to report high levels of alcohol use than it is to report use of cocaine and methamphetamine. Last, we used a more informative variable for alcohol use compared to our variables for cocaine use and methamphetamine use .
That is, the results presented here do not necessarily point to alcohol use as more detrimental to ART adherence than use of other substances. In fact, poly-substance use was very high in our sample and level of alcohol use was found to be significantly associated with lifetime cocaine SUD. A potential future direction may be to more closely examine poly-substance use in the context of level of use for each substance. Despite the lack of significant associations between ART adherence and use of other substances, education showed a strong association to ART adherence. Previous research suggests that low education may be correlated with decreased ART adherence because each is associated with health literacy . Health literacy is defined as the ability to process and understand information about health and health services in order to make the best health-related decisions . Studies of health literacy among PLWHA have shown that higher education is associated with increased health literacy and that greater health literacy is associated with health status and ART adherence . The results of the current study may suggest that PLWHA with higher education may be more likely to understand and appreciate the importance of staying adherent to their ART regimen, even in the context of at-risk drinking. Education and health literacy among PLWHA who drink alcohol, however, has yet to be studied on its own. This is an important topic for future research and subsequent development of better interventions for improving health status among PLWHA, as level of education is immutable but health literacy can be easily enhanced. There are several strengths of the current study including the large sample size and a relatively large group of at-risk drinkers; however, this study also has some limitations. First, alcohol use data are self-reported. The HNRP Substance Use History form is a retrospective self-report assessment that relies on participants’ memory of past drinking patterns. Although most studies rely on self-report to measure alcohol use, requiring participants to give day-by-day or real-time reports may be more reliable. Additionally, we were not able to calculate weekly alcohol quantity from this measure. This might have aided our classification of at-risk drinkers, as the NIAAA guideline also specifies a weekly drinking limit for at-risk drinking ; however, previous research has shown that exceeding the daily limit is associated with a much greater risk of developing an AUD compared to that of exceeding the weekly limit . We also did not administer other alcohol screening tools such as the Alcohol Use Disorders Identification Test , which may have aided our characterization of severity of participants’ drinking patterns. Our adherence measure, using either self-reported non-adherence in the last four days or plasma HIV viral load detectability, also has limitations. We attempted to overcome the social desirability bias that is expected of self-reported adherence by using plasma viral load detectability as a proxy for adherence. Although viral load detectability does not equal nonadherence, non-adherence is likely the greatest indicator of viral detectability. Other reasons for viral load detectability, including virologic failure, drug resistance, and transient viral blips, are not very common and are often associated with ART non-adherence . Further, we did not collect data on other characteristics related to viral load and ART adherence, including drug-resistant genotype, homelessness, and criminal justice history. We did, however, assess for ART regimen type, which may represent a particularly important variable in the use of viral load detectability as a proxy for adherence, as some forms of ART have longer half-lives and are thus more forgiving with missed doses. We found no differences in ART regimen type across drinking groups or adherence groups, providing further evidence in support of our use of viral load detectability as a proxy for ART adherence. Lastly, however, using viral load detectability as a proxy for ART nonadherence does not give as detailed information about medication adherence as would a real-time measure.