Abstract
Background
Alcohol use has been shown to accelerate disease progression in experimental studies of simian immunodeficiency virus in macaques, but the results in observational studies of HIV have been conflicting.
Methods
We conducted a prospective cohort study of the impact of unhealthy alcohol use on CD4 cell count among HIV-infected persons in southwestern Uganda not yet eligible for antiretroviral treatment (ART). Unhealthy alcohol consumption was 3-month Alcohol Use Disorders Identification Test – Consumption (AUDIT-C) positive (≥3 for women, ≥4 for men) and/or phosphatidylethanol (PEth - an alcohol biomarker) ≥50 ng/ml, modeled as a time-dependent variable in a linear mixed effects model of CD4 count.
Results
At baseline, 43% of the 446 participants were drinking at unhealthy levels and the median CD4 cell count was 550 cells/mm3 (Interquartile Range [IQR] 416-685). The estimated CD4 cell count decline per year was −14.5 cells/mm3 (95% Confidence Interval [CI]: −38.6 to 9.5) for unhealthy drinking vs. −24.0 cells/mm3 (95% CI: −43.6 to −4.5) for refraining from unhealthy drinking, with no significant difference in decline by unhealthy alcohol use (p-value 0.54), adjusting for age, sex, religion, time since HIV diagnosis, and HIV viral load. Additional analyses exploring alternative alcohol measures, participant subgroups, and time-dependent confounding, yielded similar findings.
Conclusion
Unhealthy alcohol use had no apparent impact on the short-term rate of CD4 count decline among HIV-infected ART naïve individuals in Uganda, using biological markers to augment self-report and examining disease progression prior to ART initiation to avoid unmeasured confounding due to misclassification of ART adherence.
Keywords: HIV progression, phosphatidylethanol, sub-Saharan Africa, antiretroviral adherence
INTRODUCTION
A substantial proportion (between 8% and 42%) of persons living with HIV/AIDS (PLWHA) worldwide have been reported to drink at unhealthy levels (i.e., at risk drinking or meeting criteria for an alcohol use disorder).1 Alcohol use has been shown to be a consistent independent risk factor for HIV acquisition,2 and among persons with HIV, alcohol use is associated with decreased retention in care,3 and worse antiretroviral treatment (ART) adherence,4 with a dose-response relationship.5,6
Chronic alcohol use impacts both innate and adaptive immune functioning,7 and chronic alcohol use and HIV independently damage the intestinal mucosa, enabling increased microbial translocation with subsequent increased inflammation.8 Experimental studies in which high doses of alcohol were administered to macaques before and after infection with simian immunodeficiency virus (SIV) found increased levels of SIV viremia and mortality compared to control macaques who were infected with SIV but who received a sucrose control.9-12 Thus alcohol use might be an important factor in HIV disease progression.
Despite the high biologic plausibility of an effect of alcohol use on HIV disease progression, the results of human observational studies have been mixed. No prospective study conducted in the period before the advent of ART found an association between alcohol consumption and the onset of AIDS,13 and a retrospective analysis of persons not yet on ART participating in a large clinical HIV cohort found no association between risky alcohol use and CD4 cell count.14 However, two studies conducted since the advent of ART suggested a detrimental effect of alcohol use prior to ART use, with one study reporting a difference in mean CD4 cell count of 49 cells/mm3 among those reporting heavy drinking compared to those abstaining,15 and another reporting a strong association between frequent alcohol use (≥2 drinks daily) and time to CD4 cell count below 200 cells/mm.3 16 Neither study found an association of alcohol use measures with HIV viral load.
Among longitudinal studies of persons on ART, the findings have been mixed as well. Several studies among persons on ART have found no association between high levels of alcohol use (variously defined as heavy, hazardous, problem, or severe risk alcohol use) and CD4 cell count and/or HIV viral load after controlling for ART adherence.14,15,17-19 Two recent studies conducted mediation analyses to separate out effects of alcohol use on CD4 cell counts due to reduced adherence versus other pathways. One found direct effects of heavy alcohol use on CD4 cell counts,20 while the other found only indirect effects of alcohol use via adherence.21
Several methodological considerations might explain these inconsistent findings. First, inability to accurately measure alcohol use, due to recall bias, especially social desirability bias, may impact the results.2 Biomarkers of alcohol use, such as phosphatidylethanol (PEth), a direct metabolite of alcohol use that is highly specific and reasonably sensitive for measuring prior 2-3 weeks’ alcohol use,22 can provide an objective measure of alcohol intake. Second, in some populations, alcohol use may be associated with illicit drug use, which may be associated with more rapid HIV progression (i.e. illicit stimulant use23,24), and thus a spurious association of alcohol use with HIV progression may occur. A solution to this is to exclude other substance use, or conduct studies in settings with very little substance use, such as in Uganda.25 Third, studies of persons on ART may be susceptible to residual confounding due to imperfect measurement of adherence, such as in the case of exaggerated self-reported adherence.26 Thus, restricting the sample to those who are not yet on ART avoids this potential pitfall. Lastly, the relationship between unhealthy alcohol use and HIV disease progression may be confounded over time, if individuals who engage in unhealthy drinking experience declines in their health, and thus reduce their subsequent drinking. This circumstance may spuriously reduce the apparent relationship between unhealthy drinking and HIV disease progression;27 previous analyses of this issue have not accounted for this possibility.
The main goal of our study was to determine the biological impact of unhealthy alcohol consumption on HIV disease progression, to clarify the previous inconsistent results. We conducted a prospective cohort study among persons not yet on ART in southwestern Uganda, a population with little other substance use, and used PEth, an objective measure of alcohol consumption, to augment self-reported unhealthy alcohol consumption.
METHODS
Study participants
This was a longitudinal prospective cohort study conducted in Mbarara, Uganda. Participants were recruited from the Immune Suppression Syndrome (ISS) Clinic of the Mbarara Regional Referral Hospital of the Mbarara University of Science and Technology (MUST). Study enrollment was conducted from September 2011 to August 2014. Eligibility criteria were: adult (age ≥18) patient of the Mbarara ISS Clinic; living within 60 km (or 120 km for men to increase male enrollment), fluent in either Runyakole (the local language) or English, and not yet meeting eligibility criteria for ART (i.e., CD4 cell count <350 cells/mm3 [cutoff changed to <500 cells/mm3 beginning March 1, 2014], World Health Organization disease stage III or IV, or AIDS defining illnesses). We aimed to include equal numbers of persons drinking at unhealthy levels to those not drinking at such levels. To increase the proportion of unhealthy drinkers recruited, reporting any prior year alcohol consumption became a further eligibility criterion beginning in September 2013; the definition of unhealthy drinking used for analysis (see below) was unchanged.
Study procedures
Study visits (baseline and follow-up) included a structured interviewer-administered assessment and phlebotomy for laboratory testing. Follow-up visits were conducted every six months, until loss to follow-up, death, ART initiation, study withdrawal, or the end of the study period (December 2015). Those who became eligible to start ART received a final interview and blood draw prior to initiating ART. All procedures were approved by the institutional review boards of the Boston University/Boston Medical Center, MUST, and the University of California, San Francisco, as well as the Uganda National Council for Science and Technology.
Laboratory testing
Whole blood specimens were tested from all study visits to determine the CD4 cell count, and baseline specimens were tested to determine HIV viral RNA level (<40 copies/ml) in batches from frozen (−80C) plasma. For PEth testing, whole blood was pipetted on the day of collection onto Whatman 903 cards and stored at −80C before shipment in batches at room temperature to the United States Drug Testing Laboratories. PEth testing was performed measuring the most common PEth homologue, PEth 16:0/18:1, as previously described.28 The limit of quantification was 8 ng/mL. All baseline DBS were tested for PEth level. No further PEth testing was conducted for participants whose baseline PEth level was <8 ng/mL and who denied current (prior 3 months) alcohol use at all study visits; the PEth level was assumed to be <8 ng/mL for all visits. PEth was tested at all visits for participants who tested PEth positive (≥8 ng/mL) at any visit or who reported any alcohol consumption at any visit.
Measures
The study assessment included demographics, alcohol consumption (using the Alcohol Use Disorders Identification Test – Consumption [AUDIT-C],29 modified to measure alcohol consumption in the prior 3 months). While the interval between study visits was 6 months to maximize available funds and minimize participant fatigue, we chose a 3-month interval for self-report of alcohol use because recall for frequent behaviors may be better for shorter recall periods.30 This period was also chosen to be roughly comparable to the maximum time period PEth can be detected after drinking ceases.31 We also measured physical health functioning (using the Medical Outcomes Study HIV Survey32,33), symptoms of HIV, and symptoms of depression (using the Center for Epidemiologic Studies Depression Scale34).
We defined unhealthy alcohol use, the primary independent variable, as unhealthy drinking detected via self-report (AUDIT-C ≥3 for women and ≥4 for men) and/or PEth, as follows. We used a cutoff of PEth ≥50 ng/mL to indicate unhealthy drinking, a cut-off that was highly sensitive (93%) and reasonably specific (83%) for detecting daily drinking of two or more drinks per day on average (S. Stewart, personal communication). Thus, our primary measure of unhealthy alcohol use was either AUDIT-C positive and/or PEth ≥50 ng/mL.
Statistical Analysis
We calculated descriptive statistics to characterize study participants overall and by unhealthy drinking status. We used chi-square or Fisher’s exact tests for categorical variables and t-tests or the Wilcoxon rank-sum test for continuous variables, as appropriate, to compare baseline characteristics between groups.
We evaluated the association between unhealthy alcohol consumption and CD4 cell count over time as our primary analysis. We used linear mixed effects models with subject-specific random intercepts and slopes (to account for within subject correlation over time), and included a time by unhealthy alcohol use (as a time-dependent variable) interaction term to evaluate the main hypothesis that unhealthy alcohol use is associated with the rate of HIV disease progression. The primary outcome variable was CD4 cell count, assessed every 6 months from baseline to the final study visit (just prior to ART initiation or the end of the study). We adjusted for baseline age, religion, gender, time since HIV diagnosis, and log10 HIV RNA viral load as potential confounders. Prior to regression modeling, we calculated Spearman correlation coefficients for independent variables and covariates (all correlations r<0.40).
We conducted several confirmatory analyses. These included analyses limited to those who were recruited and followed prior to the change in ART eligibility (from CD4 cell count <350 to <500 cells/mm3) to assess the impact of this change. In an analysis, to assess the impact of possible ‘sick quitters’, we included only those participants reporting any prior 3 months alcohol use or PEth ≥8 ng/ml at baseline, and lifetime abstainers (i.e. excluding past drinkers). In another analysis, we restricted the sample to those diagnosed with HIV in the past year, to be more comparable to the macaque models that focused on early infection.9 We also re-ran the primary analysis excluding log10 HIV RNA viral load to assess potential over-fitting by including viral load as a covariate. We also examined using alternative measures of alcohol consumption, such as using PEth alone, as a 3-level variable (PEth <50 ng/mL, PEth ≥50 to <210 ng/mL, PEth ≥210 ng/ml; 210 ng/mL is a suggested PEth cutoff for excessive drinking35), as a continuous variable (log PEth), and modeling self-report, using three AUDIT-C categories (low-level drinking: <3 for women, <4 for men; medium level drinking: ≥3 and <6 for women, ≥4 and <6 for men; high level drinking: ≥6 for men and women). We also applied pattern mixture models to the main analyses to explore departures from the assumption that data are missing at random (MAR).36 For these analyses, we classified participants’ visit patterns as complete (completing all scheduled study visits), monotone (missing one and all following visits), or intermittent (returning at least once after a missed visit) and we assessed interactions between these patterns and the parameters in the mixed effects model.
We additionally conducted analyses using marginal structural models (MSMs) to account for time-dependent covariates that may potentially be both confounders and mediators of the relationship between unhealthy alcohol consumption and HIV disease progression,27,37,38 to determine whether an effect of unhealthy alcohol use on HIV disease progression could be masked if participants reduced their drinking in response to disease progression that was the result of prior unhealthy alcohol use. The parameters of the MSM were estimated using a GEE model of CD4 cell count using inverse probability of treatment weights (IPTW), to balance the joint distribution of all co-variables at each time point, thus eliminating both time independent and time-dependent confounding. Weights were estimated using a logistic GEE model and accounted for time independent and time dependent variables (age, gender, marital status, education, literacy, overall health status, nausea, physical functioning, number of HIV symptoms, depression, and months since HIV diagnosis).
We examined a secondary outcome, time from enrollment to CD4 cell count below the threshold for ART initiation, using the Cox proportional hazards model using the exact method for handling tied event times.39 Participants were censored at the earliest of the following: ART start, loss to follow-up, study withdrawal, or end of study. For this model, we conducted the additional analyses described for the mixed models above, and also conducted a sensitivity analysis by including a time-varying covariate for date of the guideline change for the CD4 threshold to initiate ART (i.e., before vs. after March 1, 2014). In addition, we fit a separate model with time to CD4 < 500 as the outcome, including only those with CD4 ≥ 500 at enrollment. Lastly, we used an MSM approach to examine the association of unhealthy alcohol use with time to CD4 cell count below the threshold of ART initiation, accounting for time-dependent confounding. All analyses were conducted using two-sided tests and a significance level of 0.05.
Sample size
A priori, we estimated the sample size needed to detect the expected differences in CD4 cell count decline between unhealthy vs. not-unhealthy drinking, with 80% power and a 2-sided test with a significance level of 0.05. We considered the change in CD4 cell count from baseline to the 12-month time point (corresponding to testing an alcohol by time interaction), which is a conservative approach given our analyses based on repeated measures. The standard deviation of change in CD4 cell count over one year was previously 166 cells/ml3,15 with an expected retention rate of 90%, a sample size of 450 would detect a difference in the 1-year decline in CD4 count between the groups of 50 cells/mm3 or greater, similar to the difference in CD4 count previously found.15
RESULTS
Of 1096 persons approached for enrollment, 484 persons enrolled, 445 persons were initially deemed ineligible, and 167 persons declined enrollment. Reasons for declining enrollment included not having time (n=70), not wanting to have blood drawn (n=30), not wanting to participate (n=20), worries about stigma or disclosure (n=15), feeling too weak to participate (n=13), needing permission from someone to participate (n=11), and other, unspecified reasons (n=8). Declining participation did not differ by gender. After enrollment and baseline testing, we found that several participants were not eligible for this study, based on testing of stored specimens of participants whose HIV viral load was found to be low or undetectable (<500 copies/ml). We thus excluded 32 participants who were not HIV antibody positive, 5 who tested positive for the presence of nevirapine or efavirenz (the two most commonly used HIV drugs in Uganda), and one participant who was missing alcohol use data, leaving 446 participants for analysis.
At baseline, thirty percent (30%) of the participants were AUDIT-C positive, and 35% had PEth level ≥50 ng/ml (Table 1). The majority were concordant on AUDIT-C and PEth (57% concordant negative, 21% concordant positive), 13% were AUDIT-C negative but PEth ≥50 ng/mL, and 9% were AUDIT-C positive but PEth <50 ng/mL. Thus 43% of the cohort were defined as drinking at unhealthy levels (AUDIT-C positive and/or PEth ≥50 ng/mL). Two-thirds (68%) of participants were women; all reported lifetime abstention from heroin, methamphetamine, and cocaine and 7 (1.6%) persons reported any lifetime marijuana use and 7 reported any lifetime khat use (1.6%). The median CD4 cell count at baseline was 550 cells/mm3 (IQR 416-685) and the median time since HIV diagnosis was 18.5 months (IQR 1.9-64.6).
Table 1.
ART naïve HIV-infected persons in southwestern Uganda: participant characteristics at baseline (N=446).
| Variable | Response | Overall | Unhealthy drinking* = Yes | Unhealthy drinking* = No | p-value |
|---|---|---|---|---|---|
| All | 446 (100.0%) | 193 (43.3%) | 253 (56.7%) | ||
| Age | <30 | 164 (36.8%) | 66 (40.2%) | 98 (59.8%) | 0.59 |
| 30–40 | 173 (38.8%) | 79 (45.7%) | 94 (54.3%) | ||
| >40 | 109 (24.4%) | 48 (44.0%) | 61 (56.0%) | ||
| Religion | Catholic | 157 (35.2%) | 80 (51.0%) | 77 (49.0%) | <0.01 |
| Moslem | 41 (9.2%) | 8 (19.5%) | 33 (80.5%) | ||
| Saved/Other | 29 (6.5%) | 3 (10.3%) | 26 (89.7%) | ||
| Protestant/Anglican | 219 (49.1%) | 102 (46.6%) | 117 (53.4%) | ||
| Sex | Male | 144 (32.3%) | 85 (59.0%) | 59 (41.0%) | <0.01 |
| Female | 302 (67.7%) | 108 (35.8%) | 194 (64.2%) | ||
| Months since HIV diagnosis | N Mean (Std Dev) Median (25th, 75th) |
446 37.2 (44.0) 18.5 (1.9, 64.6) |
193 33.9 (39.1) 18.6 (1.4, 56.7) |
253 39.7 (47.3) 18.4 (2.5, 72.9) |
0.17 |
| Viral Load (log10) | N Mean (Std Dev) Median (25th, 75th) |
442 3.7(1.0) 3.7 (3.0, 4.3) |
191 3.8 (1.1) 3.8 (3.0, 4.5) |
251 3.6 (1.0) 3.7 (3.0, 4.3) |
0.12 |
| CD4 Count | N Mean (Std Dev) Median (25th, 75th) |
446 570.0 (206.3) 550.0 (416.0, 685.0) |
193 556.5 (197.6) 541.0 (415.0, 666.0) |
253 580.2 (212.5) 553.0 (421.0, 705.0) |
0.23 |
| Alcohol use | |||||
| When last consumed alcohol (self-report) | In the past 3 days 3 days – 3 weeks ago 3 weeks – 3 months ago 3 months – 5 years ago Never or >5 years ago |
114 (26.0%) 68 (15.5%) 57 (13.0%) 77 (17.5%) 123 (28.0%) |
96 (50.8%) 48 (25.4%) 32 (16.9%) 7 (3.7%) 6 (3.2%) |
18 (17.2%) 20 (8.0%) 25 (10.0%) 70 (28.0%) 117 (46.8%) |
<0.01 |
| AUDIT-C | Positive (≥3 for women, ≥4 for men) | 133 (30.0%) | 133 (100.0%) | 0 (0.0%) | <0.01 |
| Negative | 310 (70.0%) | 59 (19.0%) | 251 (81.0%) | ||
| AUDIT-C score | N Mean (Std. Dev) Median (25th, 75th) |
443 2.1 (2.8) 1.0 (0.0, 3.0) |
192 4.4 (2.9) 4.0 (2.0, 6.0) |
251 0.4 (0.7) 0.0 (0.0, 1.0) |
<0.01 |
| PEth level | ≥50 ng/mL | 153 (34.5%) | 153 (100.0%) | 0 (0.0%) | <0.01 |
| <50 ng/mL | 290 (65.5%) | 39 (13.4%) | 251 (86.6%) | ||
| PEth level (ng/ml) | N Mean (Std. Dev) Median (25th, 75th) |
446 160.7 (393.0) 8.5 (BLQ#, 109.0) |
193 365.4 (532.6) 148.0 (60.4, 403.0) |
253 4.5 (9.0) (BLQ#, BLQ#) |
<0.01 |
| AUDIT-C by PEth level | AUDIT-C positive and PEth ≥50 ng/mL | 94 (21.2%) | |||
| AUDIT-C positive and PEth <50 ng/mL | 39 (8.8%) | ||||
| AUDIT-C negative and PEth ≥50 ng/mL | 59 (13.3%) | ||||
| AUDIT-C negative and PEth <50 ng/mL | 251 (56.7%) |
Unhealthy drinking=Yes defined as AUDIT-C+ and/or PEth ≥50 ng/mL; Unhealthy drinking=No defined as AUDIT-C- and PEth<50 ng/mL.
Below the limit of quantification
The median duration of follow-up was 12.4 months (Interquartile range [IQR] 6.5-22.5), and median the number of study visits per participant was 3 (IQR 2-5). Over the course of the study, two-thirds (67%) graduated from the cohort due to starting ART, 25% were followed until the end of the study, while 8% were lost to follow-up or withdrew from the study.
Primary outcome: CD4 cell count
The unadjusted and adjusted mixed models showed declines in CD4 cell count over time, but no statistically significant difference in the rate of the decline by unhealthy drinking (Table 2). The estimated decline in CD4 cell count from baseline over 1 year was −14.5 cells/mm3 (95% Confidence Interval [CI]: −38.6 to 9.5) for unhealthy drinking vs. −24.0 cells/mm3 (95% CI: −43.6 to −4.5) for not drinking at unhealthy levels in the adjusted model (Table 3), and the p-value for interaction was not statistically significant (p=0.54). We did not find a significant relationship between level of drinking and CD4 cell count decline over time in any of the additional analyses (limiting visits to those that occurred prior to ART eligibility changes, excluding past drinkers, excluding those diagnosed more than one year prior to enrollment, not including HIV viral load in the model, examining alternative measures for unhealthy alcohol use, and using MSM techniques to account for potential time-dependent confounding) (Table 3). We did not detect significant differences across patterns of missing data in the pattern mixture models (p-value for interaction: 0.68) nor did we detect a relationship between unhealthy alcohol use and HIV disease progression within any missing data pattern group (all p>0.30).
Table 2.
Unadjusted and adjusted models of CD4 cell count in ART naïve HIV-infected persons in southwestern Uganda.
| Model 1: Unadjusted n=447 |
Model 2: Adjusted for age, religion, sex, time since HIV diagnosis, HIV viral load n=443 |
|||
|---|---|---|---|---|
|
| ||||
| β (95% CI) | p-value | β (95% CI) | p-value | |
|
| ||||
| Unhealthy Drinking* (main effect) | 0.68 | |||
| Yes | −16.25 (−50.07, 17.58) | −7.26 (−42.22, 27.70) | ||
| No (ref) | — | — | ||
|
| ||||
| Months since Baseline (main effect) | −2.27 (−3.92, −0.63) | 0.01 | −2.00 (−3.63, −0.37) | 0.02 |
|
| ||||
| Interaction Term | ||||
| Unhealthy drinking* (Yes vs. No) × Months since Baseline | 0.82 (−1.75, 3.39) | 0.53 | 0.79 (−1.76, 3.34) | 0.54 |
|
| ||||
| Age | 0.65 | |||
| <30 | −19.10 (−65.03, 26.83) | |||
| 30–40 | −17.79 (−59.21, 23.63) | |||
| >40 (ref) | — | |||
|
| ||||
| Religion | 0.53 | |||
| Catholic | 2.63 (−32.36, 37.62) | |||
| Moslem | −22.64 (−79.65, 34.36) | |||
| Saved/Other | −42.43 (−109.37, 24.50) | |||
| Protestant (ref) | — | |||
|
| ||||
| Sex | 0.20 | |||
| Male | −24.07 (−60.50, 12.36) | |||
| Female (ref) | — | |||
|
| ||||
| Months since HIV diagnosis | −0.17 (−0.57, 0.24) | 0.41 | ||
|
| ||||
| HIV Viral Load (log10) | −49.09 (−65.14, −33.04) | <0.01 | ||
Unhealthy drinking=Yes defined as AUDIT-C+ and/or PEth ≥50 ng/mL; Unhealthy drinking=No defined as AUDIT-C- and PEth <50 ng/mL.
Table 3.
Estimated 12-month change in CD4 cell count by drinking status for primary and additional analyses of CD4 cell count in ART naïve HIV-infected persons in southwestern Uganda.
| Model description | Unadjusted estimate (95% CI) |
p-value for interaction | Adjusted estimate (95% CI) (Adjusted for age, religion, sex, time since HIV diagnosis, HIV viral load) |
p-value for interaction |
|---|---|---|---|---|
| Primary model | 0.53 | 0.54 | ||
| Unhealthy drinking* Yes | −17.43 (−41.73, 6.87) | −14.53 (−38.58, 9.52) | ||
| No | −27.30 (−47.02, −7.57) | −24.02 (−43.56, −4.49) | ||
| Model limited to observations prior to change in ART eligibility in 2014 (n=379) | 0.20 | 0.13 | ||
| Unhealthy drinking* Yes | −66.31 (−117.48, −15.13) | −61.07 (−112.86, −9.29) | ||
| No | −107.27 (−143.29, −71.25) | −109.63 (−145.97, −73.28) | ||
| Model excluding past drinkers (n=338) | 0.32 | 0.38 | ||
| Unhealthy drinking* Yes | −18.35 (−43.46, 6.76) | −16.23 (−41.14, 8.68) | ||
| No | −36.49 (−62.93, −10.05) | −32.27 (−58.54, −6.00) | ||
| Model limited to persons diagnosed in the past year (n=191) | 0.45 | 0.49 | ||
| Unhealthy drinking* Yes | −44.44 (−84.14, −4.74) | −40.23 (−79.94, −0.52) | ||
| No | −24.91 (−58.96, 9.15) | −22.50 (−56.69, 11.70) | ||
| Viral load excluded from model | 0.54 | |||
| Unhealthy drinking* Yes | — | −16.86 (−41.16, 7.44) | ||
| No | — | −26.56 (−46.29, −6.83) | ||
| Model using PEth categories to represent drinking level | 0.69 | 0.71 | ||
| PEth ≥210 ng/mL | −9.06 (−49.10, 30.98) | −5.77 (−45.49, 33.95) | ||
| PEth ≥50 ng/mL and <210 ng/mL | −28.56 (−71.96, 14.83) | −23.11 (−66.38, 20.15) | ||
| PEth <50 ng/mL or confirmed abstainer | −27.79 (−47.78, −7.81) | −24.05 (−43.82, −4.28) | ||
| Model using continuous PEth to represent drinking level | ||||
| Log PEth (per 1 unit log10 PEth change, per 12 months) | −31.57 (−54.53, −8.62) | 0.47 | −23.14 (−46.57, 0.29) | 0.49 |
| Model using AUDIT-C categories to represent self-reported drinking level | 0.26 | 0.29 | ||
| High: AUDIT-C ≥6 | 5.98 (−34.60, 46.56) | 7.74 (−32.65, 48.13) | ||
| Medium: AUDIT-C positive** and AUDIT-C<6 | −20.18 (−57.79, 17.43) | −15.91 (−53.45, 21.63) | ||
| Low: AUDIT-C negative | −30.52 (−48.47, −12.56) | −27.04 (−44.82, −9.25) | ||
| Marginal structural model& | 0.22 | |||
| Unhealthy drinking* Yes | −2.85 (−69.17, 63.47) | |||
| No | — | 46.32 (−11.09, 103.72) |
Unhealthy drinking=Yes defined as AUDIT-C+ and/or PEth >50 ng/mL; Unhealthy drinking=No defined as AUDIT-C- and PEth <50 ng/mL.
AUDIT-C- positive defined as ≥3 for women, ≥4 for men
Estimate from weighted model
Secondary outcome: Time to ART eligibility
The unadjusted and adjusted hazard ratios for unhealthy drinking in examining time to ART eligibility (i.e. CD4 cell count <350 cells/mm3 prior to March 1, 2014, CD4 cell count <500 cells/mm3 thereafter) were 1.17 (0.91-1.49) and 1.10 (0.84-1.43), respectively (Table 4). When we included a time-dependent indicator to reflect when the threshold for starting ART changed, when we examined time to CD4 cell count to 500 cells/mm3, and when we limited the data to visits completed before the change in ART eligibility, the hazard ratios for unhealthy alcohol use were not substantially different from above. The additional analyses of time to ART eligibility (e.g., excluding past drinkers, as above), yielded similar results (Table 5).
Table 4.
Cox proportional hazard models of time to CD4 count below level for ART eligibility in ART naïve HIV infected persons in southwestern Uganda.
| Model 1: Unadjusted n=446 |
Model 2: Adjusted for age, religion, sex, time since HIV diagnosis, HIV viral load n=442 |
|||
|---|---|---|---|---|
|
| ||||
| Hazard Ratio (95% CI) | p-value | Hazard Ratio (95% CI) | p-value | |
|
| ||||
| Unhealthy Drinking* | 0.49 | |||
| Yes | 1.17 (0.91, 1.49) | 0.22 | 1.10 (0.84, 1.43) | |
| No (ref) | — | — | ||
|
| ||||
| Age | 0.52 | |||
| <30 | 0.84 (0.59, 1.19) | |||
| 30-40 | 0.97 (0.71, 1.33) | |||
| >40 (ref) | — | |||
|
| ||||
| Religion | 0.12 | |||
| Catholic | 1.02 (0.78, 1.34) | |||
| Moslem | 1.60 (1.08, 2.39) | |||
| Saved/Other | 1.02 (0.61, 1.72) | |||
| Protestant (ref) | — | |||
|
| ||||
| Sex | 0.60 | |||
| Male | 1.08 (0.82, 1.42) | |||
| Female (ref) | — | |||
|
| ||||
| Months since HIV | ||||
| Diagnosis | 1.00 (1.00, 1.00) | 0.52 | ||
|
| ||||
| HIV Viral Load (log10) | 1.33 (1.18, 1.51) | <0.01 | ||
Unhealthy drinking=Yes defined as AUDIT-C+ and/or PEth ≥50 ng/mL; Unhealthy drinking=No defined as AUDIT-C− and PEth <50 ng/mL.
Table 5.
Cox proportional hazards models for time to CD4 count below ART threshold, for secondary variable and additional analyses in ART naïve HIV infected persons in southwestern Uganda.
| Unadjusted hazard ratio (95% CI) | p-value | Adjusted hazard ratio (95% CI) (Adjusted for age, religion, sex, time since HIV diagnosis, HIV viral load) | p-value | |
|---|---|---|---|---|
| Primary model | 0.22 | 0.49 | ||
| Unhealthy drinking*Yes | 1.17 (0.91, 1.49) | 1.10 (0.84, 1.43) | ||
| No | ref | ref | ||
| Model limited to person-time prior to change in ART eligibility in 2014 (n=379) | 0.28 | 0.86 | ||
| Unhealthy drinking*Yes | 1.20 (0.86, 1.67) | 1.03 (0.72, 1.49) | ||
| No | ref | ref | ||
| Model excluding past drinkers (n=337) | 0.51 | 1.00 | ||
| Unhealthy drinking* Yes | 1.10 (0.83, 1.45) | 1.00 (0.74, 1.35) | ||
| No | Ref | ref | ||
| Model limited to persons diagnosed in the past year (n=190) | 0.33 | 0.92 | ||
| Unhealthy drinking* Yes | 1.20 (0.83, 1.73) | 0.98 (0.64, 1.49) | ||
| No | Ref | Ref | ||
| Viral load excluded from model | 0.32 | |||
| Unhealthy drinking* Yes | — | 1.14 (0.88, 1.48) | ||
| No | — | ref | ||
| Model of time to CD4<500 (n=263) | 0.50 | 0.52 | ||
| Unhealthy drinking* Yes | 0.90 (0.65, 1.24) | 0.89 (0.62, 1.27) | ||
| No | Ref | ref | ||
| Model including indicator for change in threshold for starting ART | 0.80 | 0.63 | ||
| Unhealthy drinking* Yes | 1.03 (0.80, 1.33) | 0.94 (0.72, 1.22) | ||
| No | Ref | Ref | ||
| Model using continuous log10 PEth to represent drinking level | 1.09 (0.97, 1.21) | 0.13 | 1.06 (0.94, 1.20) | 0.36 |
| Model using PEth categories to represent drinking level | 0.44 | 0.76 | ||
| PEth ≥210 ng/mL | 1.22 (0.90, 1.67) | 1.07 (0.76, 1.50) | ||
| PEth ≥50 ng/mL and <210 ng/mL | 1.01 (0.71, 1.44) | 0.94 (0.65, 1.36) | ||
| PEth <50 ng/mL | ref | ref | ||
| Model using AUDIT-C categories to represent self-reported drinking level | 0.56 | 0.73 | ||
| AUDIT-C ≥6 | 1.01 (0.70, 1.45) | 0.98 (0.66, 1.44) | ||
| AUDIT-C positive and AUDIT-C<6 | 0.82 (0.56, 1.19) | 0.86 (0.58, 1.26) | ||
| AUDIT-C negative# | ref | ref | ||
| Marginal structural model& | 0.69 | |||
| Unhealthy drinking* Yes | 1.10 (0.79, 1.52) | |||
| No | ref |
Unhealthy drinking=Yes defined as AUDIT-C+ and/or PEth ≥50 ng/mL; Unhealthy drinking=No defined as AUDIT-C- and PEth <50 ng/mL.
AUDIT-C-negative defined as AUDIT-C<3 for women, <4 for men
Estimate from weighted model, estimates at 12 months
DISCUSSION
We conducted a large prospective study of CD4 cell count over time in persons with HIV who were not yet on ART and found no significant difference in the rate of CD4 decline for unhealthy vs. no unhealthy drinking in the primary analysis, nor in any of the extensive confirmatory analyses. The clinical significance of this is that unhealthy alcohol use does not appear to have a direct short-term impact on CD4 cell counts among persons with HIV who are not on ART. This is consistent with several prospective studies conducted in East Africa18,19,21 and elsewhere,14,17 but differs from some others.15,16,20 The gender distribution (68% female) was representative of that of HIV in eastern and southern Africa.40 The PEth levels revealed some discordance with self-report, consistent with previous studies of persons with HIV.41-45 The methodological strengths of this study are notable. We recruited a large number of unhealthy drinkers, incorporated biological measures of drinking, restricted the sample to persons not on ART, and did not include persons using other substances. In addition, we conducted multiple confirmatory analyses (i.e., examining alternative measures of alcohol use, including using higher AUDIT-C and PEth cutoffs to examine very heavy drinking, key subgroups of participants, and accounting for potential time-dependent confounding) that were consistent with the primary results, arguing for the robustness of these findings.
Despite finding no impact on CD4 cell count decline, there is evidence that alcohol use negatively impacts the HIV epidemic and individuals with HIV in ways other than CD4 decline. First, alcohol use has been a consistent risk factor for HIV acquisition,46-48 thus drinkers make up a disproportionate number of those infected with HIV. Alcohol use can impact the risk of onward transmission of HIV as well, presumably as a consequence of increased sexual risk behavior. The literature strongly suggests that persons with HIV who consume alcohol have lower ART adherence,4 putting viral suppression and HIV outcomes at risk. Drinkers also may be more likely to transmit HIV due to increased vaginal shedding.49-51 In addition to acquisition and transmission of HIV, some evidence suggests that alcohol use adversely impacts gut factors related to microbial translocation, despite the absence of consequences of the latter on CD4 cell count decline.52
This study was limited by the changes in the ART initiation criterion that occurred during the course of the study. These changes restricted our ability to examine CD4 cell count decline to a narrow range of starting values (i.e., above 500 cells/mm3) and decreased the amount of follow up time, because, by design, we were following participants only until they started ART. The relatively short follow-up duration (median 12.5 months) may have limited our ability to find associations of alcohol use with longer-term outcomes, such as a multi-factor measure of HIV morbidity.53 Because this study was observational, unmeasured confounding may have obscured our results. For example, we did not measure exercise or nutritional status.
We conclude from these results, despite some past literature to the contrary, that there is no clinically meaningful biological impact of unhealthy alcohol use on the CD4 cell count decline among persons with HIV not yet on ART. However, unhealthy alcohol use has been previously shown to decrease ART adherence and increase HIV transmission, thus adversely impacting both individual and population-level outcomes. In addition, there is suggestive evidence that unhealthy alcohol use may negatively impact inflammation; for these reasons, unhealthy alcohol use should be strongly discouraged for persons with HIV. However, unhealthy alcohol use does not appear to have a short-term direct biological impact on CD4 cell count.
Acknowledgments
Sources of support: National Institutes of Health U01AA020776, K24AA022586, U24AA020778, and U24AA020779
Footnotes
These data were presented (in part) at the 38th Annual meeting of the Research Society on Alcoholism Scientific Meeting, June 2015, San Antonio, TX.
References
- 1.Williams EC, Hahn JA, Saitz R, et al. Alcohol Use and Human Immunodeficiency Virus (HIV) Infection: Current Knowledge, Implications, and Future Directions. Alcohol Clin Exp Res. 2016;40(10):2056–2072. doi: 10.1111/acer.13204. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Baliunas D, Rehm J, Irving H, et al. Alcohol consumption and risk of incident human immunodeficiency virus infection: a meta-analysis. Int J Public Health. 2010;55(3):159–166. doi: 10.1007/s00038-009-0095-x. [DOI] [PubMed] [Google Scholar]
- 3.Monroe AK, Lau B, Mugavero MJ, et al. Heavy Alcohol Use is Associated with Worse Retention in HIV Care. J Acquir Immune Defic Syndr. 2016 doi: 10.1097/QAI.0000000000001083. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Hendershot CS, Stoner SA, Pantalone DW, et al. Alcohol use and antiretroviral adherence: review and meta-analysis. J Acquir Immune Defic Syndr. 2009;52(2):180–202. doi: 10.1097/QAI.0b013e3181b18b6e. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Sileo KM, Simbayi LC, Abrams A, et al. The role of alcohol use in antiretroviral adherence among individuals living with HIV in South Africa: Event-level findings from a daily diary study. Drug Alcohol Depend. 2016;167:103–111. doi: 10.1016/j.drugalcdep.2016.07.028. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Braithwaite RS, McGinnis KA, Conigliaro J, et al. A temporal and dose-response association between alcohol consumption and medication adherence among veterans in care. Alcohol Clin Exp Res. 2005;29(7):1190–1197. doi: 10.1097/01.alc.0000171937.87731.28. [DOI] [PubMed] [Google Scholar]
- 7.Szabo G, Saha B. Alcohol’s Effect on Host Defense. Alcohol Res. 2015;37(2):159–170. [PMC free article] [PubMed] [Google Scholar]
- 8.Bagby GJ, Amedee AM, Siggins RW, et al. Alcohol and HIV Effects on the Immune System. Alcohol Research : Current Reviews. 2015;37(2):287–297. [PMC free article] [PubMed] [Google Scholar]
- 9.Bagby GJ, Stoltz DA, Zhang P, et al. The effect of chronic binge ethanol consumption on the primary stage of SIV infection in rhesus macaques. Alcohol Clin Exp Res. 2003;27(3):495–502. doi: 10.1097/01.ALC.0000057947.57330.BE. [DOI] [PubMed] [Google Scholar]
- 10.Bagby GJ, Zhang P, Purcell JE, et al. Chronic binge ethanol consumption accelerates progression of simian immunodeficiency virus disease. Alcohol Clin Exp Res. 2006;30(10):1781–1790. doi: 10.1111/j.1530-0277.2006.00211.x. [DOI] [PubMed] [Google Scholar]
- 11.Kumar R, Perez-Casanova AE, Tirado G, et al. Increased viral replication in simian immunodeficiency virus/simian-HIV-infected macaques with self-administering model of chronic alcohol consumption. J Acquir Immune Defic Syndr. 2005;39(4):386–390. doi: 10.1097/01.qai.0000164517.01293.84. [DOI] [PubMed] [Google Scholar]
- 12.Poonia B, Nelson S, Bagby GJ, et al. Chronic alcohol consumption results in higher simian immunodeficiency virus replication in mucosally inoculated rhesus macaques. AIDS Res Hum Retroviruses. 2006;22(6):589–594. doi: 10.1089/aid.2006.22.589. [DOI] [PubMed] [Google Scholar]
- 13.Hahn JA, Samet JH. Alcohol and HIV disease progression: weighing the evidence. Curr HIV/AIDS Rep. 2010;7(4):226–233. doi: 10.1007/s11904-010-0060-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Conen A, Wang Q, Glass TR, et al. Association of alcohol consumption and HIV surrogate markers in participants of the Swiss HIV Cohort Study. J Acquir Immune Defic Syndr. 2013 doi: 10.1097/QAI.0b013e3182a61ea9. [DOI] [PubMed] [Google Scholar]
- 15.Samet JH, Cheng DM, Libman H, et al. Alcohol consumption and HIV disease progression. J Acquir Immune Defic Syndr. 2007;46(2):194–199. doi: 10.1097/QAI.0b013e318142aabb. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Baum MK, Rafie C, Lai S, et al. Alcohol use accelerates HIV disease progression. AIDS Res Hum Retroviruses. 2010;26(5):511–518. doi: 10.1089/aid.2009.0211. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Kowalski S, Colantuoni E, Lau B, et al. Alcohol consumption and CD4 T-cell count response among persons initiating antiretroviral therapy. J Acquir Immune Defic Syndr. 2012;61(4):455–461. doi: 10.1097/QAI.0b013e3182712d39. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Weiser SD, Palar K, Frongillo EA, et al. Longitudinal assessment of associations between food insecurity, antiretroviral adherence and HIV treatment outcomes in rural Uganda. AIDS. 2014;28(1):115–120. doi: 10.1097/01.aids.0000433238.93986.35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Cagle A, McGrath C, Richardson BA, et al. Alcohol use and immune reconstitution among HIV-infected patients on antiretroviral therapy in Nairobi, Kenya. AIDS Care. 2017:1–6. doi: 10.1080/09540121.2017.1281881. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Kahler CW, Liu T, Cioe PA, et al. Direct and Indirect Effects of Heavy Alcohol Use on Clinical Outcomes in a Longitudinal Study of HIV Patients on ART. AIDS Behav. 2016 doi: 10.1007/s10461-016-1474-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Wandera B, Tumwesigye NM, Nankabirwa JI, et al. Hazardous alcohol consumption is not associated with CD4+ T-cell count decline among PLHIV in Kampala Uganda: A prospective cohort study. PLoS One. 2017;12(6):e0180015. doi: 10.1371/journal.pone.0180015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Wurst FM, Thon N, Yegles M, et al. Ethanol metabolites: their role in the assessment of alcohol intake. Alcohol Clin Exp Res. 2015;39(11):2060–2072. doi: 10.1111/acer.12851. [DOI] [PubMed] [Google Scholar]
- 23.Carrico AW, Shoptaw S, Cox C, et al. Stimulant use and progression to AIDS or mortality after the initiation of highly active antiretroviral therapy. J Acquir Immune Defic Syndr. 2014;67(5):508–513. doi: 10.1097/QAI.0000000000000364. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Cook JA, Burke-Miller JK, Cohen MH, et al. Crack cocaine, disease progression, and mortality in a multicenter cohort of HIV-1 positive women. Aids. 2008;22(11):1355–1363. doi: 10.1097/QAD.0b013e32830507f2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.John-Langba J, Ezeh A, G G, et al. Alcohol, Drug Use, and Sexual-risk Behaviors among Adolescents in Four Sub-Saharan African Countries. http://paa2006.princeton.edu/download.aspx?submissionId=61153. Accessed December 7, 2008.
- 26.Stirratt MJ, Dunbar-Jacob J, Crane HM, et al. Self-report measures of medication adherence behavior: recommendations on optimal use. Transl Behav Med. 2015;5(4):470–482. doi: 10.1007/s13142-015-0315-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Robins JM, Hernan MA, Brumback B. Marginal structural models and causal inference in epidemiology. Epidemiology. 2000;11(5):550–560. doi: 10.1097/00001648-200009000-00011. [DOI] [PubMed] [Google Scholar]
- 28.Jones J, Jones M, Plate C, et al. The detection of 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanol in human dried blood spots. Analytical Methods. 2011;3(5):1101. [Google Scholar]
- 29.Bradley KA, DeBenedetti AF, Volk RJ, et al. AUDIT-C as a brief screen for alcohol misuse in primary care. Alcohol Clin Exp Res. 2007;31(7):1208–1217. doi: 10.1111/j.1530-0277.2007.00403.x. [DOI] [PubMed] [Google Scholar]
- 30.Napper LE, Fisher DG, Reynolds GL, et al. HIV risk behavior self-report reliability at different recall periods. AIDS Behav. 2010;14(1):152–161. doi: 10.1007/s10461-009-9575-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Hahn JA, Anton RF, Javors MA. The Formation, Elimination, Interpretation, and Future Research Needs of Phosphatidylethanol for Research Studies and Clinical Practice. Alcohol Clin Exp Res. 2016;40(11):2292–2295. doi: 10.1111/acer.13213. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Stangl AL, Bunnell R, Wamai N, et al. Measuring quality of life in rural Uganda: reliability and validity of summary scores from the medical outcomes study HIV health survey (MOS-HIV) Qual Life Res. 2012;21(9):1655–1663. doi: 10.1007/s11136-011-0075-5. [DOI] [PubMed] [Google Scholar]
- 33.Wu AW, Hays RD, Kelly S, et al. Applications of the Medical Outcomes Study health-related quality of life measures in HIV/AIDS. Qual Life Res. 1997;6(6):531–554. doi: 10.1023/a:1018460132567. [DOI] [PubMed] [Google Scholar]
- 34.Radloff LS. The CES-D scale: A self-report depression scale for research in the general population. J Appl Psychol Meas. 1977;1:385–401. [Google Scholar]
- 35.Helander A, Hansson T. National harmonization of the alcohol biomarker PEth. Lakartidningen. 2013;110(39-40):1747–1748. [PubMed] [Google Scholar]
- 36.Hedeker D, Gibbons RD. Application of random-effects pattern-mixture models for missing data in longitudinal studies. Psychological methods. 1997;2(1):64. [Google Scholar]
- 37.Hernan MA, Brumback B, Robins JM. Marginal structural models to estimate the causal effect of zidovudine on the survival of HIV-positive men. Epidemiology. 2000;11(5):561–570. doi: 10.1097/00001648-200009000-00012. [DOI] [PubMed] [Google Scholar]
- 38.Hernan MA, Brumback BA, Robins JM. Estimating the causal effect of zidovudine on CD4 count with a marginal structural model for repeated measures. Stat Med. 2002;21(12):1689–1709. doi: 10.1002/sim.1144. [DOI] [PubMed] [Google Scholar]
- 39.Kalbfleisch JD, Prentice RL. The statistlcal analysis of failure time data. 2nd. Hoboken: Wiley; 2002. [Google Scholar]
- 40.2017; aidsinfo.unaids.org. Accessed 10/6, 2017.
- 41.Muyindike WR, Lloyd-Travaglini C, Fatch R, et al. Phosphatidylethanol confirmed alcohol use among ART-naive HIV-infected persons who denied consumption in rural Uganda. AIDS Care. 2017:1–6. doi: 10.1080/09540121.2017.1290209. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Bajunirwe F, Haberer JE, Boum Y, 2nd, et al. Comparison of self-reported alcohol consumption to phosphatidylethanol measurement among HIV-infected patients initiating antiretroviral treatment in southwestern Uganda. PLoS One. 2014;9(12):e113152. doi: 10.1371/journal.pone.0113152. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Papas RK, Gakinya BN, Mwaniki MM, et al. Associations Between the Phosphatidylethanol Alcohol Biomarker and Self-Reported Alcohol Use in a Sample of HIV-Infected Outpatient Drinkers in Western Kenya. Alcohol Clin Exp Res. 2016;40(8):1779–1787. doi: 10.1111/acer.13132. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Littlefield AK, Brown JL, DiClemente RJ, et al. Phosphatidylethanol (PEth) as a Biomarker of Alcohol Consumption in HIV-Infected Young Russian Women: Comparison to Self-Report Assessments of Alcohol Use. AIDS Behav. 2017;21(7):1938–1949. doi: 10.1007/s10461-017-1769-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Wang Y, Chen X, Hahn JA, et al. Phosphatidylethanol (PEth) in Comparison to Self-Reported Alcohol Consumption among HIV-infected Women in a Randomized Controlled Trial of Naltrexone for Reducing Hazardous Drinking. Alcohol Clin Exp Res. 2017 doi: 10.1111/acer.13540. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Joseph Davey D, Kilembe W, Wall KM, et al. Risky Sex and HIV Acquisition Among HIV Serodiscordant Couples in Zambia, 2002-2012: What Does Alcohol Have To Do With It? AIDS Behav. 2017 doi: 10.1007/s10461-017-1733-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Chersich MF, Bosire W, King’ola N, et al. Effects of hazardous and harmful alcohol use on HIV incidence and sexual behaviour: a cohort study of Kenyan female sex workers. Global Health. 2014;10:22. doi: 10.1186/1744-8603-10-22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Kiwanuka N, Ssetaala A, Ssekandi I, et al. Population attributable fraction of incident HIV infections associated with alcohol consumption in fishing communities around Lake Victoria, Uganda. PLoS One. 2017;12(2):e0171200. doi: 10.1371/journal.pone.0171200. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Theall KP, Amedee A, Clark RA, et al. Alcohol consumption and HIV-1 vaginal RNA shedding among women. J Stud Alcohol Drugs. 2008;69(3):454–458. doi: 10.15288/jsad.2008.69.454. [DOI] [PubMed] [Google Scholar]
- 50.Homans J, Christensen S, Stiller T, et al. Permissive and protective factors associated with presence, level, and longitudinal pattern of cervicovaginal HIV shedding. J Acquir Immune Defic Syndr. 2012;60(1):99–110. doi: 10.1097/QAI.0b013e31824aeaaa. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Loganantharaj N, Nichols WA, Bagby GJ, et al. The effects of chronic binge alcohol on the genital microenvironment of simian immunodeficiency virus-infected female rhesus macaques. AIDS Res Hum Retroviruses. 2014;30(8):783–791. doi: 10.1089/aid.2014.0065. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Carrico AW, Hunt PW, Emenyonu NI, et al. Unhealthy Alcohol Use is Associated with Monocyte Activation Prior to Starting Antiretroviral Therapy. Alcohol Clin Exp Res. 2015;39(12):2422–2426. doi: 10.1111/acer.12908. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Marshall BDL, Tate JP, McGinnis KA, et al. Long-term alcohol use patterns and HIV disease severity. Aids. 2017;31(9):1313–1321. doi: 10.1097/QAD.0000000000001473. [DOI] [PMC free article] [PubMed] [Google Scholar]