Abstract
Alcohol use is associated with poor outcomes among people living with HIV (PLWH), but it remains unclear which alcohol use measures best predict future HIV viral non-suppression over time. This study aimed to compare the ability of five alcohol use measures to predict risk of suboptimal HIV viral load trajectories over 36 months. We analyzed data from a cohort of PLWH in Florida including survey data linked to the state HIV surveillance system on prospective HIV viral loads over 36 months (n = 783; 66% male; 55% Black; Mage=46, SD = 11). Four trajectory patterns for HIV viral load were identified: consistently low (65.1%), decreasing (15.9%), increasing (10.6%), and consistently high (8.4%). Past year alcohol use frequency (OR = 2.1, CI:1.0-4.4), drinks consumed on a typical drinking day (OR = 2.2, CI: 1.2–4.1), frequency of binge drinking (OR = 2.6, CI:1.3–5.2), and alcohol-related problems score (OR = 1.7, CI:1.1–2.7) were the measures predictive of the risk of future viral non-suppression above specific thresholds.
Keywords: Alcohol measurement, Binge drinking, HIV, Viral suppression, Trajectory analysis
Resumen
El consumo de alcohol está asociado con malos resultados entre las personas que viven con el VIH (PLWH), pero aún no está claro qué medidas de consumo de alcohol predicen mejor la falta de supresión viral del VIH en el futuro con el tiempo. Este estudio tuvo como objetivo comparar la capacidad de cinco medidas de consumo de alcohol para predecir el riesgo de trayectorias subóptimas de la carga viral del VIH durante 36 meses. Analizamos datos de una cohorte de PLWH en Florida, incluidos datos de encuestas vinculadas al sistema estatal de vigilancia del VIH sobre posibles cargas virales del VIH durante 36 meses (n = 783; 66% hombres; 55% afroamericanos; Maños=46, SD = 11). Se identificaron cuatro patrones de trayectoria para la carga viral del VIH: consistentemente baja (65,1%), decreciente (15,9%), creciente (10,6%) y consistentemente alta (8,4%). Frecuencia de consumo de alcohol en el último año (OR = 2,1, IC: 1,0–4,4), bebidas consumidas en un día típico de consumo de alcohol (OR = 2,2, IC: 1,2–4,1), frecuencia de consumo excesivo de alcohol (OR = 2,6, IC: 1,3–5,2), y la puntuación de problemas relacionados con el alcohol (OR = 1,7, IC: 1,1–2,7) fueron las medidas predictivas del riesgo de no supresión viral futura por encima de umbrales específicos.
Introduction
Ending the HIV Epidemic (EHE) is a multi-pronged plan to eliminate new HIV infections in the U.S. by following four key pillars, one of which focuses on treating HIV rapidly and effectively to achieve sustained viral suppression [1]. With the effectiveness of current antiretroviral therapy (ART) and pre-exposure prophylaxis (PrEP), along with the proven concept of undetectable = untransmissible (U = U), the HIV epidemic in the U.S. should theoretically be ended swiftly [1, 2]. This, however, has not been the case. In fact, only 56% of persons living with HIV (PLWH) in the U.S. were considered virally suppressed in 2018 [3]. Biomedical intervention alone is not sufficient to end the HIV epidemic. PLWH may have other challenges that need to be addressed in conjunction with these biomedical interventions. To reach the goal of EHE, it is critical to intervene on such needs, including behaviors like alcohol use, that serve as barriers to achieving viral suppression.
Alcohol use, common among people living with HIV (PLWH), is associated with adverse HIV health outcomes, including reduced engagement and retention in care, poor adherence to treatment (i.e., ART), and viral non-suppression [4-7]. Adherence to treatment is the most important patient-related factor for achieving and sustaining viral suppression [8]. One meta-analysis showed that PLWH who drank alcohol were approximately 55% as likely to adhere to their treatment as persons who did not drink, and those who drank heavily (using the National Institute on Alcohol Abuse and Alcoholism [NIAAA] criteria for men; >14 drinks per week or > 4 drinks in a day) [9] or met criteria for an alcohol use disorder were 47% as likely to be adherent [10]. There are several mechanisms in which alcohol may lead to treatment nonadherence, including alcohol-related adherence avoidance behavior (e.g., due to interactive toxicity beliefs) and prospective memory impairment (e.g., forgetting to take medication) [8]. Depending on how alcohol use is measured and subsequently categorized, the magnitude of association of the relationship between HIV health outcomes and alcohol use varies [11]. Additionally, the period (e.g., 30 days vs. 12 months) in which alcohol use is assessed could alter this association. Further, whether acute alcohol use, patterns of alcohol use, or problems related to alcohol use are being studied could impact the conclusions reached.
The AUDIT-C is one of the most commonly used scales in the alcohol–HIV field [12, 13]. The measure assesses average quantity and frequency of alcohol use, and average frequency of “risky” drinking (> 6 drinks on one occasion), often over 12 months, and can be used to identify persons with “hazardous” alcohol use [14]. The three questions from the AUDIT-C can be considered separately; in clinics, the NIAAA suggests screening persons who drink with just one question similar to the AUDIT-C first (frequency of “binge drinking”; [≥ 4/5 drinks per occasion for women/men]), followed by average quantity and frequency questions if necessary [15]. Williams and colleagues found that persons with medium-to-very-high-levels of alcohol use (according to the AUDIT-C) had significantly lower levels of viral suppression compared to those with no alcohol use [16]. Similarly, using questions adapted from the AUDIT-C, Cook and colleagues found that “heavy drinking” (> 7/14 drinks per week for women/men) was associated with lower odds of achieving viral suppression [5]. Among this same sample, however, “binge drinking” was not associated with viral suppression [17]. When taking frequency of binge drinking into account, though, Myers and colleagues demonstrated that frequent binge drinking (≥ 3 times per week) was significantly associated with viral non-suppression [18].
Acute high volume alcohol use may also be associated with poor HIV health outcomes. Assessing the greatest number of drinks consumed in one day may identify the most extreme binge drinking, which may be associated with greatest HIV-related risk. Parsons and colleagues showed that with each additional drink consumed, odds of medication nonadherence increased by 20% [19]. Other aspects of alcohol use, such as inter- and intra-personal alcohol-related problems, as measured by the Short-Inventory of Problems (SIP) [20], may be related to HIV health outcomes. Though this relationship has not yet been studied, it would be plausible that individuals with higher SIP scores could have worse HIV health outcomes, as they are dealing with increased alcohol-related problems that may be affecting other aspects of their life.
Further, while evidence among most persons not living with HIV indicates that alcohol use becomes harmful at a certain cut-point (above a threshold of use) [21], this is less clear for PLWH. Evidence is mixed regarding whether any amount of alcohol use among PLWH is associated with worse health outcomes. Some studies support the presence of a dose-response relationship between alcohol use and poor outcomes [16, 22] whereas others support a clear cut-point when alcohol use becomes harmful [17, 18].
The studies above relied on cross-sectional HIV viral load data, from just one time-point. The longitudinal relationship between alcohol use and HIV viral load may follow a similar trend, though there are few longitudinal findings in the literature. Satre and colleagues report on the association between changes in number of days of hazardous alcohol use and HIV viral suppression, using data from a clinical trial testing an alcohol-reduction intervention [23]. Examining data from a 1-year period (baseline, 6-months, and 12-months), the authors found that a reduction in hazardous drinking days was significantly associated with increased likelihood of HIV viral suppression [23]. Cook and colleagues reported similar findings from a randomized clinical trial of Naltrexone among women living with HIV [24]. Using data from a 7-month period (baseline, 2-months, 4-months, and 7-months), the authors found that women who quit heavy drinking had higher rates of HIV viral suppression [24]. However, data regarding the longitudinal nature of this relationship from observational studies are lacking. It is also unclear which alcohol use measures most strongly predict poor HIV outcomes, such as higher HIV viral loads (e.g., viral non-suppression). Group-based trajectory analyses are appropriate to answer this question because they allow one to test the longitudinal relationship between exposure variables (e.g., alcohol use) and an outcome assessed over time (e.g., HIV viral load) [25]. Group-based trajectory analyses are especially useful in describing different types of patterns that cannot be captured using point estimates. If different types of HIV viral load patterns exist, knowing how common the pattern is and whether specific groups of individuals are experiencing each pattern could influence testing frequency. Further, knowing which measure of alcohol use is most strongly associated with future HIV viral suppression could influence the specific questions that are asked of people in clinical settings, research studies, or public health programs such as Ryan White, and individuals who may need additional resources to achieve sustained viral suppression could be identified.
To accomplish this goal, we examined data from a prospective cohort of PLWH. The aims of this study were to (1) identify HIV viral load group-based trajectories over 36 months; and (2) to compare the ability of five alcohol use measures to predict the risk of suboptimal HIV viral load trajectories by assessing whether each measure has a dose-response relationship or a clear cut-point (i.e., threshold) that is more strongly associated with suboptimal trajectories by examining trends in odds ratios, statistical significance, and area under the curve (AUC).
Methods
Study Design
Data for this secondary analysis came from the Florida Cohort, a statewide prospective survey study that aimed to identify factors associated with HIV health outcomes. The current analysis includes data collected at the baseline survey that were linked to longitudinal HIV surveillance data. Detailed methods of the Florida Cohort are published elsewhere [27], but briefly, PLWH were recruited using targeted, convenience sampling at clinics and community settings throughout north, central, and south Florida. A total of 932 PLWH were recruited from eight Florida cities between October 2014 and June 2018. The demographic characteristics of persons in the total sample were similar to those in Florida, overall [27]. To be eligible for participation in the Florida Cohort, participants needed to be living with HIV, 18 years or older, plan to remain in Florida for at least six months, and communicate in English (all sites) or Spanish (two sites). Participants completed a self-administered survey that took approximately 30–45 min to complete. The survey could be completed using either paper-and-pencil or online. The survey included questions about their demographics, substance use, mental health, and HIV-related health behavior. All study procedures were approved by participating Institutional Review Boards and all participants provided informed consent.
HIV viral loads were obtained via the Enhanced HIV/AIDS Reporting System (eHARS) database, an HIV surveillance system that is maintained by the Centers for Disease Control and Prevention (CDC). We established a Data Use Agreement with the Florida Department of Health HIV Surveillance Division, which maintains the state’s eHARS database, to receive updated data every six months, allowing us to follow participants’ HIV health outcomes for 36 months. Florida eHARS is a CDC designed web-based/browser-based HIV surveillance system/application used for the reporting and data management of HIV diagnoses in Florida. Florida has dual reporting of HIV diagnoses from both providers who diagnose HIV and from laboratories performing the testing of specimens by Florida statute and are required to report positive test results to the Florida Department of Health. eHARS encompasses core HIV/AIDS surveillance data activities and projects and assists with the investigation of potential HIV/AIDS cases, management of current data, importing and exporting of data and the transfer of data to CDC. All positive HIV test results including detectable and undetectable viral loads are required to be reported by statute. All these results are imported into the eHARS system which goes through a series of validation checks to identify errors, and the standardization processes ensures that any imported data match existing eHARS data formats. The result is data imported from various sources, such as state or private labs, can be verified as quality data before being written to the eHARS database. Since Florida mandates the reporting of all HIV viral load results [28], we were able to match survey data with HIV viral load for 98% of the study sample.
Measures
HIV viral load
In this study, we analyzed HIV viral load data up to 36 months after the baseline survey. We chose 36 months as the maximum cut-off due to data availability since study recruitment ended in 2018. Viral load tests were grouped in periods of six-months for analysis, based on testing frequency recommendations by HIV.gov [29]. The baseline timepoint (time = 0) included HIV viral load tests from six months prior to study enrollment up until enrollment. Each additional time point was coded as follows: Month 6 (1–6 months); Month 12 (7–12 months); Month 18 (13–18 months); Month 24 (19–24 months); Month 30 (25–30 months); and Month 36 (31–36 months). For participants with more than one HIV viral load test within a defined sixmonth period, we used the highest viral load value.
Alcohol use variables
We examined five measures of alcohol use: (1) past year alcohol use frequency; (2) drinks consumed on a typical drinking day; (3) frequency of binge drinking; (4) most drinks consumed in one day; and (5) alcohol-related problems score. For the current analysis, persons who reported that they had never drank alcohol or had not had any drinks in the past year were grouped together and recoded as not having alcohol in the past year for all alcohol use measures.
Past Year Alcohol Use frequency
Participants were asked, “How often did you have a drink containing alcohol?” with the answer options “Less than once a month,” “1–3 times a month (less than weekly),” “1–3 times a week,” “4–6 times a week,” and “Every day.” This and the following two questions were adapted from the AUDIT-C [14].
Drinks consumed on a typical drinking day
Next, participants were asked, “In the past 12 months, how many standard drinks would you have on a typical day when you were drinking?” The first 657 participants had answer options of “1,” “2,” “3,” “4,” “5,” and “6+”, and subsequent participants had answer options of “1–2,” “3–4,” “5–6,” “7–9,” and “10 or more,” and these were recoded to 3 categories: “1–2,” “3–4,” and “5 or more” to include all results.
Frequency of binge drinking
We next asked about frequency of binge drinking over the past 12 months, “When you were drinking regularly, how often did you have 4 + standard drinks (for women) or 5 + standard drinks (for men) on one occasion?” with the answer options “Never,” “Less than once a month,” “Monthly,” “Weekly,” and “Daily or almost daily.”
Most drinks consumed in one day
Then, we asked, “During the last 30 days (month), what is the largest number of drinks containing alcohol that you drank within a 24-hour period?” with the answer options “Less than 1 drink,” “1 drink,” “2 drinks,” “3 drinks,” “4 drinks,” “5 to 7 drinks,” “8 to 11 drinks,” “12 to 17 drinks,” “18 to 23 drinks,” “24 to 35 drinks,” and “36 drinks or more.” Due to low frequency of larger quantities of drinks, we combined the last five options to be “8 or more drinks.”
Alcohol-related problems score
Alcohol-related problems over the past 12 months were measured by the Short Inventory of Problems (SIP-2R) [20]. The SIP-2R is a validated, 15-item scale that measures alcohol-related problems on a total scale from 0 to 15. For this analysis, scores were grouped into categories “0”, “1–5”, “6–10”, or “11–15”.
Socio-demographics and other baseline characteristics
Participants selected their sex at birth, race, and ethnicity by selecting the answer option that they felt best described them. Participants also self-reported their age, education, insurance status, housing status, and past year drug use at the time of the questionnaire.
Data Analysis
Of the 932 participants who completed the baseline survey, 783 were included in our analyses. We excluded 149 participants because: (1) We could not link surveillance data to the survey data (n = 14); (2) There were too few (less than three) HIV viral loads available (n = 62); (3) Data were missing for any one of the five alcohol measures (n = 73).
All statistical analyses were conducted using SAS version 9.4. To correct for the right-skewed and highly variable distribution of HIV viral load, we conducted a log10 transformation, creating the variable log10-HIV viral load. We used a semi-parametric, group-based method to fit log10-HIV viral load trajectories over 36 months [30].
Group-based trajectory analyses require subjective decisions to be made on the number of trajectories to include and the shape of each trajectory (e.g., quadratic, cubic); we followed published recommendations and literature to make these decisions [25, 26, 31]. Using PROC TRAJ with the DROPOUT function, a four trajectory model was determined to have the best fit, based on Bayesian information criterion (BIC), significant polynomial terms, and high average posterior group membership probabilities (greater than 70% is considered adequate) [25, 32, 33]. Parsimony in number of trajectory groups and group membership probability greater than 5% was prioritized [33]. Rstudio was used to plot the final trajectories, displaying the observed mean log10-HIV viral loads for each group.
Next, we conducted Pearson’s Chi-square tests (sex at birth, race/ethnicity, alcohol use measures) and ANOVA tests (age) to determine which demographic and alcohol use measures were significantly associated with membership in the four trajectory groups at an alpha level < 0.05. We then dichotomized the outcome variable (trajectory group membership) as optimal (e.g., consistent viral suppression) vs. suboptimal (e.g., non-consistent viral suppression). Consistent viral suppression was defined as having all HIV viral load values < 200 copies/ml (equivalent to a log10-HIV viral load < 2.3 copies/ml) [34]. For each alcohol use variable, we conducted proportional odds models with the alcohol use variable as the independent variable and the viral load outcome as the dependent variable (reference: optimal). Alcohol use variables were treated ordinally. We controlled for sex at birth, race/ethnicity, and age. Race/ethnicity was dichotomized (for the proportional odds models) as either “Non-Hispanic, White” or “Other” which included persons who identified as Hispanic, Black, Native American, Asian, Multi-Racial, or another race not otherwise listed. We also created receiver operating characteristic (ROC) curves for each alcohol variable (controlling for sex at birth, race/ethnicity, and age) to calculate the AUC, using the PLOTS(only) = ROC option [35]. An AUC equal to 1 means the measure has perfect discriminatory ability to classify an individual to the correct group; whereas an AUC equal to 0.5 is considered a chance level [36].
Sensitivity analysis
We conducted an a priori sensitivity analysis with a different dichotomization of the trajectory groups by combining the group with consistent viral suppression with a group that started off virally unsuppressed but became consistently suppressed over time. This was done because it was unclear whether the group that switched from virally unsuppressed to suppressed was more similar to the consistent viral suppression group or those without consistent viral suppression.
Results
Sample characteristics
Of the 783 participants, the majority were male (66%) and non-Hispanic, Black (55%), with a mean age of 46 years (SD = 11). Additional sample characteristics are reported in Table 1.
Table 1.
Baseline characteristics among sample of persons living with HIV in Florida (n = 783)
Frequency (%) |
|
---|---|
Recruitment Region | |
North Florida | 239 (31%) |
Central Florida | 424 (54%) |
South Florida | 120 (15%) |
Sex at Birth | |
Male | 513 (66%) |
Female | 270 (34%) |
Race/Ethnicity | |
Non-Hispanic, White | 170 (22%) |
Non-Hispanic, Black | 429 (55%) |
Non-Hispanic, Other | 29 (4%) |
Hispanic | 155 (20%) |
Age | |
Mean (SD) | 46.3 (11.4) |
Educationa | |
Less than High School | 263 (34%) |
High School or Equivalent | 230 (29%) |
More than High School | 287 (37%) |
Insurancea | |
Uninsured | 46 (6%) |
Insured | 712 (94%) |
Homelessab | |
No | 655 (85%) |
Yes | 116 (15%) |
Any Past Year Drug Usea | |
No | 290 (42%) |
Yes | 398 (58%) |
Total frequency does not equal n = 783 due to missing data
Participants were considered homeless if they reported living in a homeless shelter, emergency shelter, car, street, or in an abandoned building in the past 12 months
More than two-thirds (68%) of the sample reported past 12-month alcohol use. In terms of past year alcohol use frequency, approximately one-quarter (24%) of the sample reported at least weekly alcohol use. For drinks consumed on a typical drinking day, approximately 32% of the sample reported consuming at least 3 drinks. Regarding frequency of binge drinking, 16% reported binge drinking at least weekly. Further, for most drinks consumed in one day, 15% reported consuming at least 5 drinks on one day in the past 30 days. Lastly, in terms of alcohol-related problems score, about one-third of persons who reported alcohol use in the past 12-months also reported experiencing at least one alcohol-related problem (not shown in tables).
Log10-HIV viral load trajectories
The observed mean log10-HIV viral loads at each time period are shown in Fig. 1. Four trajectory groups were identified: (1) consistently low viral load (n = 513; 65.1%); (2) decreasing viral load (n = 123; 15.9%); (3) increasing viral load (n = 83; 10.6%); and (4) consistently high viral load (n = 64; 8.4%). The consistently low viral load group remained below a log10-HIV load of 2.3 the entire 36 months, indicating that these persons were, on average, virally suppressed throughout the entirety of the study, as opposed to the other three trajectory groups. The decreasing viral load group did, however, achieve HIV viral suppression at the 12-month period and remained virally suppressed for the duration of the study (Fig. 1).
Fig. 1.
Each line represents the observed mean log10-HIV viral load at the given time period. Four trajectory patterns for HIV viral load were identified: consistently low (65.1%), decreasing (15.9%), increasing (10.6%), and consistently high (8.4%). (Note: A log10-HIV viral load < 2.3 indicates HIV viral suppression).
Distributions of demographic and alcohol use variables by trajectory group are displayed in Table 2. Of note, Non-Hispanic, White and Hispanic persons were represented in higher proportions in the consistently low viral group compared to Non-Hispanic, Black and Other persons. Additionally, persons in the consistently low viral load group were older, on average, than those in the other groups.
Table 2.
Baseline characteristics among sample of persons living with HIV in Florida by group-based log10-HIV viral load trajectories (n = 783)
Low (n = 513) |
Decreasing (n = 123) |
Increasing (n = 83) | High (n = 64) |
||
---|---|---|---|---|---|
Sex at Birth | x2(3) = 8.719, p = 0.003 | ||||
Male | 327 (64%) | 92 (18%) | 48 (9%) | 46 (9%) | |
Female | 186 (69%) | 31 (11%) | 35 (13%) | 18 (7%) | |
Race/Ethnicity | x2(9) = 28.204, p = 0.001 | ||||
Non-Hispanic, White | 126 (74%) | 26 (15%) | 7 (4%) | 11 (6%) | |
Non-Hispanic, Black | 259 (60%) | 68 (16%) | 57 (13%) | 45 (10%) | |
Non-Hispanic, Other | 14 (48%) | 7 (24%) | 6 (21%) | 2 (7%) | |
Hispanic | 114 (74%) | 22 (14%) | 13 (8%) | 6 (4%) | |
Age | F(3) = 60.233, p < 0.001 | ||||
Mean (SD) | 48.6 (10.8) | 41.3 (11.5) | 44.0 (11.2) | 40.8 (10.2) | |
Past year alcohol use frequency | x2(15) = 19.850, p = 0.221 | ||||
No alcohol in past year | 178 (72%) | 31 (13%) | 24 (10%) | 14 (6%) | |
Less than once a month | 114 (60%) | 35 (18%) | 27 (14%) | 15 (8%) | |
Less than weekly | 100 (63%) | 25 (16%) | 16 (10%) | 17 (11%) | |
1–3 times a week | 82 (71%) | 16 (14%) | 8 (7%) | 10 (9%) | |
4–6 times a week | 19 (53%) | 7 (19%) | 5 (14%) | 5 (14%) | |
Everyday | 20 (57%) | 9 (26%) | 3 (9%) | 3 (9%) | |
Drinks consumed on a typical drinking dayb | x2(9) = 15.946, p = 0.068 | ||||
No alcohol in past year | 178 (72%) | 31 (13%) | 24 (10%) | 14 (6%) | |
1–2 drinks | 179 (62%) | 46 (16%) | 36 (13%) | 26 (9%) | |
3–4 drinks | 124 (67%) | 33 (18%) | 14 (8%) | 15 (8%) | |
5 or more drinks | 32 (51%) | 13 (21%) | 9 (14%) | 9 (14%) | |
Frequency of binge drinkingab | x2(15) = 26.291, p = 0.035 | ||||
No alcohol in past year | 178 (72%) | 31 (13%) | 24 (10%) | 14 (6%) | |
Never | 117 (61%) | 34 (18%) | 25 (13%) | 15 (8%) | |
Less than once a month | 92 (65%) | 24 (17%) | 11 (8%) | 14 (10%) | |
Monthly | 51 (67%) | 16 (21%) | 7 (9%) | 2 (3%) | |
Weekly | 54 (64%) | 10 (12%) | 8 (9%) | 13 (15%) | |
Daily or almost daily | 21 (49%) | 8 (19%) | 8 (19%) | 6 (14%) | |
Most drinks consumed in one dayc | x2(21) = 27.652, p = 0.150 | ||||
No alcohol in past year | 178 (72%) | 31 (13%) | 24 (10%) | 14 (6%) | |
Less than 1 drink | 79 (59%) | 22 (16%) | 16 (12%) | 17 (13%) | |
1 drink | 27 (56%) | 10 (21%) | 7 (15%) | 4 (8%) | |
2 drinks | 58 (72%) | 10 (12%) | 8 (10%) | 5 (6%) | |
3 drinks | 51 (69%) | 8 (11%) | 8 (11%) | 7 (9%) | |
4 drinks | 49 (60%) | 13 (16%) | 13 (16%) | 7 (9%) | |
5–7 drinks | 40 (59%) | 18 (26%) | 3 (4%) | 7 (10%) | |
8 or more drinks | 31 (63%) | 11 (22%) | 4 (8%) | 3 (6%) | |
Alcohol-related problems scoread | x2(12) = 18.350, p = 0.106 | ||||
No alcohol in past year | 178 (72%) | 31 (13%) | 24 (10%) | 14 (6%) | |
0 | 185 (68%) | 42 (15%) | 26 (9%) | 21 (8%) | |
1–5 | 84 (54%) | 35 (23%) | 20 (13%) | 16 (10%) | |
6–10 | 31 (60%) | 8 (15%) | 7 (13%) | 6 (12%) | |
11–15 | 35 (64%) | 7 (13%) | 6 (11%) | 7 (13%) |
SD, Standard Deviation
Past year
Binge drinking is defined as consuming 4 + standard drinks (women) or 5 + standard drinks (men) on one occasion
Past 30 days
Measured by the Short Inventory of Problems-Revised
Alcohol Use Measures
The following section presents results from proportional odds models adjusted for sex at birth, race/ethnicity, and age. No dose-response relationship was identified for any of the five alcohol use measures. Past year alcohol use frequency was identified to have a cut-point that was significantly associated with future suboptimal viral load trajectories; persons who reported drinking 4–6 times a week were more than twice as likely to be in a suboptimal trajectory (OR: 2.1, 95% CI: 1.0–4.4, p = 0.047). Every day drinking was not significantly associated with suboptimal trajectories but followed a similar trend (OR: 1.8, 95% CI: 0.9–3.9, p = 0.116). Drinks consumed on a typical drinking day was also identified to have a cut-point significantly associated with suboptimal viral load trajectories; persons who reported consuming 5 or more drinks on a typical drinking day were more than twice as likely to be in a suboptimal trajectory group (OR: 2.2, 95% CI: 1.2–4.1, p = 0.008). Frequency of binge drinking also had a clear cut-point that was significantly associated with future suboptimal viral load trajectories; persons who reported daily or almost daily binge drinking were almost three times as likely to be in a suboptimal trajectory group (OR: 2.6, 95% CI: 1.3–5.2, p = 0.006). Most drinks consumed in one day was not significantly associated with HIV viral load trajectory (Table 3). Lastly, alcohol-related problems score was identified to have a cut-point significantly associated with future suboptimal viral load trajectories; persons who scored a 1–5 were almost twice as likely to be in a suboptimal HIV viral load trajectory (OR: 1.7, 95% CI: 1.1–2.7, p = 0.015). A similar, but non-significant trend was observed for persons who scored a 6–10 (OR: 1.9, 95% CI: 1.0–3.6, p = 0.050). ROC analyses showed similar AUCs for all five of the measures, ranging from 0.68 to 0.69.
Table 3.
Odds of not achieving consistently low log10-HIV viral loads by different alcohol use domains (n = 783)
Adjusted Modela | AUC | ||
---|---|---|---|
Past year alcohol use frequency | 0.68 | ||
No alcohol in past year (n = 247) | Reference | Reference | |
Less than once a month (n = 191) | 1.4 | 0.9–2.2 | |
Less than weekly (n = 158) | 1.2 | 0.7–1.8 | |
1–3 times a week (n = 116) | 1.0 | 0.6–1.6 | |
4–6 times a week (n = 36) | 2.1 | 1.0–4.4** | |
Every day (n = 35) | 1.8 | 0.9–3.9 | |
Drinks consumed on a typical drinking dayb | 0.69 | ||
No alcohol in past year (n = 247) | Reference | Reference | |
1–2 drinks (n = 287) | 1.3 | 0.9–1.9 | |
3–4 drinks (n = 186) | 1.1 | 0.7–1.6 | |
5 or more drinks (n = 63) | 2.2 | 1.2–4.1** | |
Frequency of binge drinkingbc | 0.69 | ||
No alcohol in past year (n = 247) | Reference | Reference | |
Never (n = 191) | 1.3 | 0.9–2.0 | |
Less than once a month (n = 141) | 1.1 | 0.7–1.8 | |
Monthly (n = 76) | 1.1 | 0.6–1.9 | |
Weekly (n = 85) | 1.3 | 0.8–2.3 | |
Daily or almost daily (n = 43) | 2.6 | 1.3–5.2** | |
Most drinks consumed in one dayd | 0.68 | ||
No alcohol in past year (n = 247) | Reference | Reference | |
Less than 1 drink (n = 134) | 1.5 | 0.9–2.4 | |
1 drink (n = 48) | 1.6 | 0.8–3.1 | |
2 drinks (n = 81) | 0.9 | 0.5–1.6 | |
3 drinks (n = 74) | 1.0 | 0.5–1.7 | |
4 drinks (n = 82) | 1.6 | 0.9–2.7 | |
5–7 drinks (n = 68) | 1.4 | 0.8–2.5 | |
8 or more drinks (n = 49) | 1.4 | 0.7–2.7 | |
Alcohol-related problems scorebe | 0.69 | ||
No alcohol in past year (n = 247) | Reference | Reference | |
0 (n = 274) | 1.0 | 0.7–1.5 | |
1–5 (n = 155) | 1.7 | 1.1–2.7** | |
6–10 (n = 52) | 1.9 | 1.0–3.6 | |
11–15 (n = 55) | 1.3 | 0.7–2.5 |
AUC, Area Under Curve; CI, Confidence Interval
Adjusted for sex at birth, race/ethnicity, and age
Past year
Binge drinking is defined as consuming 4 + standard drinks (women) or 5 + standard drinks (men) on one occasion
Past 30 days
Measured by the Short Inventory of Problems-Revised
p < 0.05
Sensitivity analysis
Results from the sensitivity analysis (comparing the consistently low and decreasing trajectories to the increasing and consistently high trajectories) were similar to that of the overall analysis but shifted towards the null. The only significant difference noted was among the most drinks consumed in one day measure. Persons who reported consuming 5–7 drinks in one day in the past 30 days were two times more likely to be in a suboptimal viral load trajectory compared to those who had no alcohol in the past year (OR: 2.0, 95% CI: 1.1–2.3, p = 0.036), whereas this relationship was not observed in the original analysis. In this case, 5–7 drinks in one day was identified as being associated with a suboptimal viral suppression trajectory, but 8 or more drinks was not (OR: 1.5, 95% CI: 0.7–3.1, p = 0.274).
Discussion
In this statewide sample of PLWH, we identified four HIV viral load trajectory groups. One trajectory group consisted of individuals who, on average, were virally suppressed for the entirety of the study period whereas the other three consisted of individuals who, on average, had at least 2 time-points of viral non-suppression. Overall, no dose-response relationships were identified; specific cut-points (i.e., high thresholds) of alcohol use in terms of both quantity (drinks consumed on a typical drinking day) and frequency (past year alcohol use and binge drinking) were associated with increased risk of future viral non-suppression. Further, experiencing the lowest threshold of alcohol-related problems (score of 1–5) was also associated with an increased risk of future viral non-suppression.
By initially examining the four distinct trajectories, HIV viral load patterns other than virally suppressed or non-suppressed were identified. Almost 20% experienced consistently high viral loads or increasing viral load over three years. Persons in these two groups may benefit from more frequent testing. Though these two groups are of particular concern, we chose to group all three virally non-suppressed trajectories together, as members of each could contribute to HIV transmission and may require enhanced intervention. Persons with viral non-suppression may be ideal candidates for the new, long-lasting injectable ART. However, the efficacy, safety, and durability of injectable ART is still being investigated for persons with a history of suboptimal ART adherence (ClinicalTrials.gov number, NCT03635788).
Given the significant associations between individual alcohol use questions and risk of future viral non-suppression, our findings indicate that brief screeners within clinics may be sufficient to assess alcohol use and predict a possibility of future issues with viral load within a population of PLWH. While we chose to analyze each alcohol use measure separately (e.g., not as a scale), we did find that all three questions adapted from the AUDIT-C were significantly associated with suboptimal trajectories after adjusting for socio-demographics. Use of the AUDIT-C is highly convenient in research studies, as it is commonly used and allows for cross-study comparisons [12, 13], but it may not be necessary to ask all three questions, clinically. Of the three questions (and all five measures studied), frequency of binge drinking in the past 12 months was most strongly associated with future viral non-suppression. This finding, consistent with Myers and colleagues’ findings [18], also helps validate the NIAAA recommendation to assess frequency of binge drinking as a single measure for “unhealthy alcohol use” [15]. Identifying single items that were strongly associated with viral non-suppression may have special significance in clinical settings where the patient-provider time is already limited [37].
While Cook and colleagues found a null association between binge drinking and viral suppression [17], we found that both binge drinking (4/5 or more drinks consumed on a typical drinking day for women/men) and frequency of binge drinking were associated with viral suppression. Both studies used similar sample populations (i.e., both used the Florida Cohort study), but Cook and colleagues only examined the relationship between alcohol use and viral suppression at one time-point. This could explain the difference in findings, highlighting the importance of studying longitudinal trends of HIV viral load. The current study adds to the literature by indicating which measures of alcohol use are associated with viral load 36 months in the future. The longitudinal nature of this study can help researchers and clinicians contribute to the EHE pillar of PLWH achieving sustained viral suppression by identifying factors (e.g., thresholds of alcohol use) on which to intervene.
This was the first study to examine the relationship between alcohol-related problems score and prospective viral suppression. There was no dose-response relationship regarding alcohol-related problems score and viral suppression. We identified a low cut-point (score of 1–5) of alcohol-related problems that was significantly associated with an increased risk of future viral non-suppression, indicating that any number of alcohol-related problems may indicate the need for additional intervention. However, given the complexity of this measure compared to measures of quantity and frequency, it may be less appropriate to implement in the clinic setting. If an individual does indicate that they are experiencing alcohol-related problems similar to those measured by the SIP (e.g., impulse control, interpersonal, intrapersonal, physical, social), however, their provider may consider them at higher risk for future viral non-suppression.
We did not identify any dose-response relationships between level of alcohol use within each measure and risk of being in a suboptimal trajectory. Rather, specific cut-points within the measures were associated with increased risk. This is in contrast with previous research that identified a dose-dependent association between alcohol use and HIV viral load [16, 22]. We only used alcohol use data from one point in time, however, which may make it more difficult to detect a dose-response relationship in relation to viral suppression. Additionally, the AUCs of each measure were nearly identical, suggesting that no one measure is “better” than another, judging by AUC alone. Our finding that most of the single measures were associated with risk of future viral non-suppression at higher thresholds indicates the importance of assessing beyond a simple “yes/no”. Aside from HIV health outcomes, healthcare providers also need to consider the effect of alcohol on other aspects of the person’s health [15].
Strengths and Limitations
This study had several strengths. First, we leveraged a large, longitudinal statewide dataset to examine 36 months of prospective HIV viral loads with little missing data (> 75% of viral loads available at 36 months). Our use of group-based trajectory analyses allowed us to identify distinct subgroups that exist longitudinally within our sample population [38]. This is the first study to use continuous HIV viral load to inform group-based trajectory analysis; the trajectory groups that resulted were similar to trajectories produced for HIV care engagement [39], CD4 count [40], and likelihood of viral suppression [41, 42]. The inclusion of multiple alcohol use measures within the survey allowed us to compare these measures in the same sample. By analyzing the data ordinally, we were able to identify clear cut-points associated with poor outcomes, identifying patterns that may have been missed or less clear with another analytic approach.
Our study also had several limitations. We restricted our analysis to participants with at least 3 HIV viral loads to detect distinct longitudinal patterns. As a result, persons with the worst outcomes (e.g., lost to care, death) may have been excluded. Demographically, the sample population was similar to the population of PLWH within the state of Florida [27], but may not be representative of all PLWH. Given that Florida ranks third highest in terms of new HIV cases (following the District of Columbia and Georgia) [43], understanding state-specific factors that impact PLWH’s health outcomes is critical to ending the HIV epidemic. Further, Florida’s HIV epidemic is emblematic of the South, which is the region with the highest incidence and prevalence of HIV in the U.S. [44]. Additionally, we only used alcohol questionnaire data from baseline, so it is unclear how changes in behaviors such as alcohol use may affect changes in HIV viral load over time. Recent studies that included multiple waves of data suggest that changes in alcohol use are associated with changes in ART adherence [45] and likelihood of viral suppression [23]. It is also worth noting that a dose-response relationship is not a metric that fits perfectly for all alcohol measures included (e.g., frequency of consumption). For this study, however, we considered possible dose-response relationships to determine whether any significant increases in risk for viral non-suppression were associated with greater alcohol involvement. Alcohol use was self-reported and thus may be subject to social desirability bias. The mid-study changes made to the question assessing drinks consumed on a typical drinking day and subsequent need to combine higher number of drinks limited the sensitivity of this measure. The relatively short, 30-day timeframe of the question assessing most drinks consumed in one day is also a limitation. In this regard, it is also difficult to compare with the other measures which assessed past 12-month alcohol use. Further, we chose not to control for other factors that may be related to alcohol use and HIV viral suppression such as housing status, mental health conditions, and drug use [46-53] since these are not necessarily factors that can be quickly assessed in a clinical setting while conducting an alcohol screen.
Conclusions
After adjusting for sociodemographic factors, we found that past year alcohol use frequency, drinks consumed on a typical drinking day, frequency of binge drinking, and alcohol-related problems score were associated with increased odds of being in a suboptimal viral load trajectory. No measure displayed a dose-response relationship; rather clear cut-points that were associated with suboptimal trajectories were detected. Frequency of binge drinking was most strongly associated with suboptimal trajectories and could be especially useful for assessing risk for future non-suppression in a clinic, especially where there is limited patient-provider time. Persons who indicate that they binge drink daily or almost daily may require additional resources to reach sustained viral suppression. Future prospective studies are needed to investigate the role of additional factors such as housing, mental health, and other substance use on achieving HIV viral suppression.
Acknowledgements
This work was supported by the National Institute on Alcohol Abuse and Alcoholism under grant U24AA022002.
Footnotes
Availability of data and material: Data can be accessed by request via sharc-research.org.
Code Availability Code can be requested by contacting the corresponding author.
Conflict of interest The authors declare that they have no conflicts of interest.
Ethics approval All procedures performed in studies were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Consent to participate Informed consent was obtained from all individual participants in the study.
Consent for publication The authors affirm that human research participants provided informed consent for publication of their deidentified data.
References
- 1.Fauci AS, Redfield RR, Sigounas G, Weahkee MD, Giroir BP. Ending the HIV Epidemic: A Plan for the United States. JAMA. 2019;321:844. [DOI] [PubMed] [Google Scholar]
- 2.Eisinger RW, Dieffenbach CW, Fauci AS. HIV Viral Load and Transmissibility of HIV Infection: Undetectable Equals Untransmittable. JAMA. 2019;321:451. [DOI] [PubMed] [Google Scholar]
- 3.Kaiser Family Foundation. HIV Viral Suppression Rate in U.S. Lowest Among Comparable High-Income Countries, 2018 [Internet]. 2021. Available from: https://www.kff.org/hivaids/slide/hiv-viral-suppression-rate-in-u-s-lowest-among-comparable-high-income-countries/. [Google Scholar]
- 4.Williams EC, Hahn JA, Saitz R, Bryant K, Lira MC, Samet JH. Alcohol Use and Human Immunodeficiency Virus (HIV) Infection: Current Knowledge, Implications, and Future Directions. Alcohol Clin Exp Res. 2016;40:2056–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Rehm J, Probst C, Shield KD, Shuper PA. Does alcohol use have a causal effect on HIV incidence and disease progression? A review of the literature and a modeling strategy for quantifying the effect. Popul Health Metr. 2017;15:4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Azar MM, Springer SA, Meyer JP, Altice FL. 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. 2010;112:178–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Pandrea I, Happel KI, Amedee AM, Bagby GJ, Nelson S. Alcohol’s role in HIV transmission and disease progression. Alcohol Res Health J Natl Inst Alcohol Abuse Alcohol. 2010;33:203–18. [PMC free article] [PubMed] [Google Scholar]
- 8.Woolf-King SE, Sheinfil AZ, Ramos J, Foley JD, Moskal D, Firkey M, et al. A conceptual model of alcohol use and adherence to antiretroviral therapy: systematic review and theoretical implications for mechanisms of action. Health Psychol Rev. 2020;1–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.National Institute on Alcohol Abuse and Alcoholism. Drinking Levels Defined [Internet]. Available from: https://www.niaaa.nih.gov/alcohol-health/overview-alcohol-consumption/moderate-binge-drinking. [Google Scholar]
- 10.Hendershot CS, Stoner SA, Pantalone DW, Simoni JM. Alcohol Use and Antiretroviral Adherence: Review and Meta-Analysis. JAIDS J Acquir Immune Defic Syndr. 2009;52:180–202. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Cook RL, Clark DB. Is There an Association Between Alcohol Consumption and Sexually Transmitted Diseases? A Systematic Review. Sex Transm Dis. 2005;32:156–64. [DOI] [PubMed] [Google Scholar]
- 12.Velloza J, Kemp CG, Aunon FM, Ramaiya MK, Creegan E, Simoni JM. Alcohol Use and Antiretroviral Therapy Non-Adherence Among Adults Living with HIV/AIDS in Sub-Saharan Africa: A Systematic Review and Meta-Analysis. AIDS Behav. 2020;24:1727–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Vagenas P, Azar MM, Copenhaver MM, Springer SA, Molina PE, Altice FL. The Impact of Alcohol Use and Related Disorders on the HIV Continuum of Care: a Systematic Review: Alcohol and the HIV Continuum of Care. Curr HIV/AIDS Rep. 2015;12:421–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Bradley KA, DeBenedetti AF, Volk RJ, Williams EC, Frank D, Kivlahan DR. AUDIT-C as a Brief Screen for Alcohol Misuse in Primary Care. Alcohol Clin Exp Res. 2007;31:1208–17. [DOI] [PubMed] [Google Scholar]
- 15.National Institute on Alcohol Abuse and Alcoholism. Helping patients who drink too much: a clinician’s guide [Internet]. 2005. Available from: https://pubs.niaaa.nih.gov/publications/practitioner/cliniciansguide2005/guide.pdf. [Google Scholar]
- 16.Williams EC, McGinnis KA, Edelman EJ, Matson TE, Gordon AJ, Marshall BDL, et al. Level of Alcohol Use Associated with HIV Care Continuum Targets in a National U.S. Sample of Persons Living with HIV Receiving Healthcare. AIDS Behav. 2019;23:140–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Cook RL, Zhou Z, Kelso-Chichetto NE, Janelle J, Morano JP, Somboonwit C, et al. Alcohol consumption patterns and HIV viral suppression among persons receiving HIV care in Florida: an observational study. Addict Sci Clin Pract [Internet]. 2017;12. Available from: http://ascpjournal.biomedcentral.com/articles/ 10.1186/s13722-017-0090-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Myers B, Lombard C, Joska JA, Abdullah F, Naledi T, Lund C, et al. Associations Between Patterns of Alcohol Use and Viral Load Suppression Amongst Women Living with HIV in South Africa. AIDS Behav [Internet]. 2021. [cited 2021 Sep 15]; Available from: https://link.springer.com/ 10.1007/s10461-021-03263-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Parsons JT, Rosof E, Mustanski B. The temporal relationship between alcohol consumption and HIV-medication adherence: A multilevel model of direct and moderating effects. Health Psychol. 2008;27:628–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Miller WR, Tonigan JS, Longabaugh R. The Drinker Inventory of Consequences (DrInC): (563232012-001) [Internet]. American Psychological Association; 1995. [cited 2018 Oct 8]. Available from: http://doi.apa.org/get-pe-doi.cfm?doi=10.1037/e563232012-001. [Google Scholar]
- 21.National Collaborating Centre for Mental Health Staff. Alcohol Use Disorders: The NICE Guideline on the Diagnosis, Assessment and Management of Harmful Drinking and Alcohol Dependence. [Internet]. 2011. [cited 2021 Nov 6]. Available from: https://www.ncbi.nlm.nih.gov/books/NBK65487. [Google Scholar]
- 22.Braithwaite RS, McGinnis KA, Conigliaro J, Maisto SA, Crystal S, Day N, et al. A Temporal and Dose-Response Association Between Alcohol Consumption and Medication Adherence Among Veterans in Care. Alcohol Clin Exp Res. 2005;29:1190–7. [DOI] [PubMed] [Google Scholar]
- 23.Satre DD, Sarovar V, Leyden W, Hare CB, Catz SL, Bryant KJ, et al. Changes in Days of Unhealthy Alcohol Use and Antiretroviral Therapy Adherence, HIV RNA Levels, and Condomless Sex: A Secondary Analysis of Clinical Trial Data. AIDS Behav [Internet]. 2019. [cited 2020 Jan 7]; Available from: http://link.springer.com/ 10.1007/s10461-019-02742-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Cook RL, Zhou Z, Miguez MJ, Quiros C, Espinoza L, Lewis JE, et al. Reduction in Drinking was Associated With Improved Clinical Outcomes in Women With HIV Infection and Unhealthy Alcohol Use: Results From a Randomized Clinical Trial of Oral Naltrexone Versus Placebo. Alcohol Clin Exp Res. 2019;43:1790–800. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Cook RL, Kelso NE, Brumback BA, Chen X. Analytic strategies to evaluate the association of time-varying exposures to HIV-related outcomes: Alcohol consumption as an example. Curr HIV Res. 2016;14:85–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Nagin DS, Odgers CL. Group-Based Trajectory Modeling (Nearly) Two Decades Later. J Quant Criminol. 2010;26:445–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Ibañez GE, Zhou Z, Cook CL, Slade TA, Somboonwit C, Morano J, et al. The Florida Cohort study: methodology, initial findings and lessons learned from a multisite cohort of people living with HIV in Florida. AIDS Care. 2020;1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Florida Department of Health. Laboratory Reporting Guidelines for Reportable Diseases and Conditions in Florida [Internet]. 2016. Available from: https://www.floridahealth.gov/diseases-and-conditions/disease-reporting-and-management/_documents/guidelines-laboratories.pdf. [Google Scholar]
- 29.HIV.gov. Table 3. Laboratory Testing Schedule for Monitoring People with HIV Before and After Initiation of Antiretroviral Therapy [Internet]. Clin. Info Available from: https://clinicalinfo.hiv.gov/en/table/table-3-laboratory-testing-schedule-monitoring-people-hiv-and-after-initiation-antiretroviral. [Google Scholar]
- 30.Jones BL, Nagin DS, Roeder K. A SAS Procedure Based on Mixture Models for Estimating Developmental Trajectories. Sociol Methods Res. 2001;29:374–93. [Google Scholar]
- 31.Marshall BDL, Operario D, Bryant KJ, Cook RL, Edelman EJ, Gaither JR, et al. Drinking trajectories among HIV-infected men who have sex with men: A cohort study of United States veterans. Drug Alcohol Depend. 2015;148:69–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Haviland AM, Jones BL, Nagin DS. Group-based Trajectory Modeling Extended to Account for Nonrandom Participant Attrition. Sociol Methods Res. 2011;40:367–90. [Google Scholar]
- 33.Jones BL, Nagin DS. Advances in Group-Based Trajectory Modeling and an SAS Procedure for Estimating Them. Sociol Methods Res. 2007;35:542–71. [Google Scholar]
- 34.Centers for Disease Control and Prevention. HIV Treatment as Prevention [Internet]. HIV Risk Prev. 2021. Available from: https://www.cdc.gov/hiv/risk/art/index.html. [Google Scholar]
- 35.Support SAS. Usage Note 65611: Modify the ROC plot produced by PROC LOGISTIC [Internet]. Available from: http://support.sas.com/kb/65/611.html. [Google Scholar]
- 36.Hajian-Tilaki K Receiver Operating Characteristic (ROC) Curve Analysis for Medical Diagnostic Test Evaluation. Casp J Intern Med. 2013;4:627–35. [PMC free article] [PubMed] [Google Scholar]
- 37.Tai-Seale M, McGuire TG, Zhang W. Time Allocation in Primary Care Office Visits: Time Allocation in Primary Care. Health Serv Res. 2007;42:1871–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Nagin DS. Group-Based Trajectory Modeling: An Overview. Ann Nutr Metab. 2014;65:205–10. [DOI] [PubMed] [Google Scholar]
- 39.Powers KA, Samoff E, Weaver MA, Sampson LA, Miller WC, Leone PA, et al. Longitudinal HIV Care Trajectories in North Carolina. JAIDS J Acquir Immune Defic Syndr. 2017;74:88–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Platt L, Xu A, Giddy J, Bogart LM, Boulle A, Parker RA, et al. Identifying and predicting longitudinal trajectories of care for people newly diagnosed with HIV in South Africa. PLoS ONE. 2020;15:e0238975. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Ocampo JMF, Plankey M, Zou K, Collmann J, Wang C, Young MA, et al. Trajectory analyses of virologic outcomes reflecting community-based HIV treatment in Washington DC 1994–2012. BMC Public Health. 2015;15:1277. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Kassaye SG, Wang C, Ocampo JMF, Wilson TE, Anastos K, Cohen M, et al. Viremia Trajectories of HIV in HIV-Positive Women in the United States, 1994–2017. JAMA Netw Open. 2019;2:e193822. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Florida Department of Health. State of the Epidemic 2018. State HIV Slide Sets [Internet]. 2020. Dec. Available from: http://www.floridahealth.gov/diseases-and-conditions/aids/surveillance/epislide-sets.html. [Google Scholar]
- 44.Centers for Disease Control and Prevention. HIV in the United States by Region [Internet]. 2020. Oct. Available from: https://www.cdc.gov/hiv/statistics/overview/geographicdistribution.html. [Google Scholar]
- 45.Williams EC, McGinnis KA, Rubinsky AD, Matson TE, Bobb JF, Lapham GT, et al. Alcohol Use and Antiretroviral Adherence Among Patients Living with HIV: Is Change in Alcohol Use Associated with Change in Adherence? AIDS Behav. 2021;25:203–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.King WD, Larkins S, Hucks-Ortiz C, Wang P-C, Gorbach PM, Veniegas R, et al. Factors Associated with HIV Viral Load in a Respondent Driven Sample in Los Angeles. AIDS Behav. 2009;13:145–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Thakarar K, Morgan JR, Gaeta JM, Hohl C, Drainoni M-L. Homelessness. HIV, and Incomplete Viral Suppression. J Health Care Poor Underserved. 2016;27:145–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Riley ED, Vittinghoff E, Koss CA, Christopoulos KA, Clemenzi-Allen A, Dilworth SE, et al. Housing First: Unsuppressed Viral Load Among Women Living with HIV in San Francisco. AIDS Behav. 2019;23:2326–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Gardner LI, Marks G, Shahani L, Giordano TP, Wilson TE, Drainoni M-L. et al. Assessing efficacy of a retention-in-care intervention among HIV patients with depression, anxiety, heavy alcohol consumption and illicit drug use. AIDS. 2016;30:1111–9. [DOI] [PubMed] [Google Scholar]
- 50.DiPrete BL, Pence BW, Bengtson AM, Moore RD, Grelotti DJ, O’Cleirigh C, et al. The Depression Treatment Cascade: Disparities by Alcohol Use, Drug Use, and Panic Symptoms Among Patients in Routine HIV Care in the United States. AIDS Behav. 2019;23:592–601. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Dvorak RD, Simons JS. Daily associations between anxiety and alcohol use: Variation by sustained attention, set shifting, and gender. Psychol Addict Behav. 2014;28:969–79. [DOI] [PubMed] [Google Scholar]
- 52.Grant BF, Chou SP, Saha TD, Pickering RP, Kerridge BT, Ruan WJ, et al. Prevalence of 12-Month Alcohol Use, High-Risk Drinking, and DSM-IV Alcohol Use Disorder in the United States, 2001–2002 to 2012–2013: Results From the National Epidemiologic Survey on Alcohol and Related Conditions. JAMA Psychiatry. 2017;74:911. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Garey L, Bakhshaie J, Sharp C, Neighbors C, Zvolensky MJ, Gonzalez A. Anxiety, depression, and HIV symptoms among persons living with HIV/AIDS: the role of hazardous drinking. AIDS Care. 2015;27:80–5. [DOI] [PubMed] [Google Scholar]