Skip to main content
NCBI home page
Journal of Studies on Alcohol and Drugs logoLink to Journal of Studies on Alcohol and Drugs
. 2023 Feb 19;84(1):79–88. doi: 10.15288/jsad.21-00440

Polysubstance Use Patterns Associated With HIV Disease Severity Among Those With Substance Use Disorders: A Latent Class Analysis

Nicolas Bertholet a,,*, Michael R Winter b, Timothy Heeren c, Alexander Y Walley d, Richard Saitz d,,e,†
PMCID: PMC9948140  PMID: 36799677

Abstract

Objective:

Polysubstance use is common among people with HIV infection (PWH) and with substance use disorder (SUD), but its effects are understudied. We aimed to identify polysubstance use patterns over time and assess their associations with HIV disease severity.

Method:

In 233 PWH with current or past SUD, latent class analysis identified polysubstance use patterns based on the Alcohol Use Disorders Identification Test–Consumption and past-30-day use of cannabis, cocaine, opioids, and tranquilizers at baseline. We categorized changes in use patterns and tested associations between those changes and CD4 count and HIV viral suppression at 12 months in linear and logistic regressions.

Results:

Three patterns were identified at baseline: 18% did not use any substance (NONE—a priori defined); 63% used mostly cannabis and alcohol (CA); and 19% used opioids along with other drugs, including cocaine, tranquilizers, cannabis, and alcohol (MULTI). At 12 months, 40% moved from a high to a lower substance use class (MULTI to CA, either to NONE) or remained as NONE, 43% were in CA both times and 17% increased (NONE to CA, either to MULTI) or remained as MULTI. The adjusted mean CD4 count (for baseline covariates and baseline CD4 count) was significantly lower among participants increasing or remaining in MULTI (523, 95% CI [448, 598], cells/mm3) compared with those who decreased/abstained throughout (607, 95% CI [552, 663], p = .02). No significant difference was observed for HIV viral suppression.

Conclusions:

We identified distinct polysubstance use patterns among PWH with SUD: cannabis/alcohol and opioids with alcohol and other drugs. Changes over time toward fewer substances/ no use were associated with lower HIV disease severity based on CD4 count but not based on HIV viral suppression

Substance use is common among people with HIV infection (PWH), and the relationship between HIV and substance use is complex. Substance use may lead to HIV transmission through risk-taking behaviors (El-Bassel et al., 2014). PWH are more likely than people without HIV to report the use of alcohol, cannabis, opioids, cocaine, and amphetamines (Cofrancesco et al., 2008; Korthuis et al., 2012; Shiau et al., 2017).

Substance use is associated with HIV disease progression: in a Canadian study of 564 PWH who had initiated antiretroviral treatment (ART), viral rebound was associated with heavy alcohol use and high-intensity illicit drug use (notably more than daily heroin use and more than daily crack use) (Ladak et al., 2019). Unhealthy alcohol use or opioid use may reduce adherence to ART (Moore et al., 2004; Williams et al., 2016), risking more rapid HIV disease progression. PoorART adherence linked to alcohol and substance use is associated with lower CD4 count, more rapid HIV disease progression, and worse health outcomes in PWH (Kader et al., 2015).

The use of one substance is often associated with the use of another, making polysubstance use (use of more than one substance) common (Hasin & Grant, 2015; McCabe et al., 2008). Concurrent polysubstance use is reported by a significant proportion of PWH: in a survey of PWH receiving HIV care, 15% reported using two or more substances concurrently (Korthuis et al., 2008). In more than 3,000 PWH receiving care in four U.S. cities (Boston, San Diego, Seattle, Birmingham), Mimiaga et al. (2013) reported a 24% prevalence of recent use of cannabis, 9% for crack cocaine, 9% for amphetamine use, and 2% for opioid use. In that study, the prevalence of polysubstance use was 10.3% (Mimiaga et al., 2013). Among HIV-infected men who have sex with men, Skeer et al. (2012) showed a prevalence of use of three or more substances of 19.9%. In PWH, substance use and polysubstance use are associated with decreased life expectancy and worse health outcomes (Arnsten et al., 2002; Korthuis et al., 2008; Palepu et al., 2003; Samet et al., 2004; Samji et al., 2013; Walley et al., 2008).

Although PWH commonly use multiple substances, most studies tend to analyze the impact of one substance independently (i.e., opioid use vs. no opioid use, while adjusting for other substance use). Many commonly used and validated assessment tools ask separately about each substance and have limited ability to identify detailed patterns of use. Addiction treatment studies often focus on substance-specific treatments (e.g., medications for opioid use disorder for people using opioids, cocaine, and benzodiazepines). Therefore, evidence is lacking about the impact of polysubstance use, and especially groups of substances commonly used among PWH. From a clinical point of view, looking at individual substances provides only limited information on how to approach patients, especially when a substantial portion of them are using multiple substances. Also, individuals may have distinct substance use patterns, and different patterns of use may have different impacts on health outcomes. In addition, because substance use is dynamic over time, patterns of use are likely to evolve, which could have an impact on health outcomes. Assessing whether substance use patterns change over time and whether potential changes are associated with non–substance use clinical outcomes will be helpful to clinicians.

Because substance use patterns and changes are seldom assessed, the stability over time, the types of patterns, and their potential association with HIV outcomes remain unclear among PWH. Latent class analysis (LCA) can be used to study substance use patterns by assigning individuals to classes on a probabilistic basis. Comparisons of health outcomes can be made across classes of multiple substance patterns instead of across individual substances. Compared with modeling involving adjustment for the use of various substances to assess the independent association with a given substance, the identification of classes allows for the study of health outcomes in people using multiple substances representing potential subgroups seen in clinical practice. LCA has been used previously to identify patterns of substance use in diverse populations (hospitalized PWH, adolescents, people using opioids, nightlife attendees; Hannemann et al., 2017; Lynskey et al., 2006; Monga et al., 2007; Shiu-Yee et al., 2018; Tomczyk et al., 2015).

The association between different patterns of substance use and HIV disease progression has not been studied sufficiently. In this study, we aimed to (a) identify polysubstance use patterns in a cohort of PWH with alcohol and/or other SUDs, (b) assess changes in substance use patterns at 12 months, and (c) assess their associations with HIV disease severity.

Method

Study design and participants

The present study was conducted with data from the Boston Alcohol Research Collaboration on HIV/AIDS (ARCH), a prospective cohort study of 250 adults. Inclusion criteria in the Boston ARCH were (a) documentation of HIV infection in medical records or HIV viral load >10,000 copies/ml, (b) past-12-month drug or alcohol dependence according to criteria from the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV; American Psychiatric Association, 1994), or ever injection drug use, (c) the ability to speak English, (d) being at least 18 years old, and (e) willingness to provide contact information for one or more persons. Pregnancy and plans to leave Boston in the next year were exclusion criteria. More details of the cohort have been reported previously (Saitz et al., 2018; Ventura et al., 2017). Enrollment took place in an urban academic hospital–based infectious disease clinic and a community health center primary care clinic serving homeless patients over a 2-year period (2012–2014). At study enrollment, participants met the substance inclusion criteria as follows: 42.8% of participants reported current alcohol and/or drug dependence without a history of injection drug use, 36.8% reported current alcohol and/or drug dependence and injection drug use, and 20.4% reported a history of injection drug use (with no current alcohol or drug dependence). Participants were assessed at baseline and 12 months later. Participants provided written informed consent for the study. The Boston University Medical Campus Institutional Review Board approved the study.

The Mini International Neuropsychiatric Interview was used to assess the presence of past 12-month DSM-IV drug or alcohol dependence (Sheehan et al., 2010). Past-30-day substance use was assessed using the Addiction Severity Index (McLellan et al., 1992). Participants also completed the Alcohol Use Disorders Identification Test–Consumption (AUDIT-C; Bush et al., 1998). To describe the impact of alcohol and other drug use among the cohort, we reported results from the Alcohol Short Index of Problems (SIP) and the Drug Short Index of Problems (SIP-D; Alterman et al., 2009). These were used to assess the face validity of the identified classes.

Latent class analysis

The following baseline variables were used in the LCA: any past-30-day use of cannabis, cocaine, opioids, and tranquilizers and the presence or absence of unhealthy alcohol use. Past-30-day use of cannabis, cocaine, opioids, and tranquilizers was assessed using the Addiction Severity Index (McLellan et al., 1992). Tranquilizers included benzodiazepines, benzodiazepine derivatives or other drugs such as barbiturates, gamma hydroxybutyrate (GHB), or Quaalude (participants were presented a card with examples). Because of a low prevalence of use in the sample, use of other stimulants (distinct from cocaine) (n = 4), inhalants (n = 4), and hallucinogens (n = 1) was not included in the analysis. Presence or absence of unhealthy alcohol use was assessed with the AUDIT-C (Bush et al., 1998). The AUDIT-C is a validated and effective tool to identify unhealthy alcohol use. It takes into account the volume and the pattern of drinking. Presence of unhealthy alcohol use was defined as an AUDIT-C score greater than 2 for women and 3 for men.

From a conceptual point of view, we considered individuals not using any substance as different. Thus, in a first step, individuals reporting no past-30-day use of cannabis, cocaine, opioids, or tranquilizers and an absence of unhealthy alcohol use (AUDIT-C score ≤ 2 for women and ≤ 3 for men) were placed into an a priori–defined separate class (hereafter referred to as “NONE”). LCA was conducted on the remaining sample of 206 reporting any substance use.

We determined the appropriate number of classes by beginning with a one-class model and then increasing the number of classes by one. Models were compared using the Bayesian information criterion (BIC) value and entropy value, the Lo–Mendell–Rubin adjusted likelihood ratio test (LMR-A), and the parametric bootstrap likelihood ratio test (BLRT). We retained the model with the lowest BIC, entropy greater than 0.8, and significant LMR-A and BLRT.

Based on the results of the LCA, we classified each participant's substance use at baseline and applied the classifications at the 12-month follow-up (i.e., the LCA was only carried out on baseline data). We manually assigned participants to a substance use class (NONE, CA, MULTI) at 12 months based on their substance use profile at 12 months. The LCA performed at baseline assigned each substance use profile (combination of yes/no responses to the five substance use questions included in the LCA) to an LCA class, and we used these same assignments to categorize participants at 12 months. We then assessed whether they changed their pattern of use at 12 months. Groupings of changes in patterns of use over time were determined after identification of classes. Groupings were determined by consensus among authors based on clinical considerations before conducting the analyses on the associations between changes in patterns and HIV disease progression. We considered three patterns of change in substance use over time, based on clinical considerations: a favorable course (i.e., maintaining an absence of use or decreasing the number of substances used), a nonfavorable course (i.e., maintaining the use of opioids and multiple substances or increasing the number of substances used), and maintenance of unhealthy use of alcohol with cannabis use. Then we assessed the association between changes in patterns of use from baseline to 12 months and HIV disease progression using linear and logistic regression models. Mean differences (betas) and 95% confidence intervals (CIs), along with adjusted least-squares means, were reported for linear models; for logistic regression models, odds ratios and 95% confidence intervals were reported. Baseline characteristics were compared across classes through descriptive statistics (mean and standard deviation for continuous variables, proportions for categorical variables) and differences across classes were assessed using Kruskal–Wallis tests or chi-square tests as appropriate. Analyses were conducted with SAS Version 9.4 (SAS Institute Inc., Cary, NC), except for the LCA, which was conducted in MPlus Version 7.31 (Muthén & Muthén, 1998–2012).

Outcomes

The outcomes (at 12 months) were (a) CD4 cell count (cells/mm3) and (b) HIV viral suppression (HIV viral load: <200 copies/ml). Participants’ medical records were reviewed for the most recent HIV viral load (copies/ml) and the most recent CD4 cell count (cells/mm3). If unavailable within 3 months before assessment, CD4 and HIV viral load were tested. Three different models were used for each outcome: first, an unadjusted model; second, a model adjusted for baseline covariates of age, sex, race/ethnicity (Black/ African American, Hispanic, White, multiracial/other), smoking (current, former, never), homelessness, Charlson comorbidity index (0–1, 2–3, ≥4; Charlson et al., 1987), and duration of HIV infection (in years); and third, a model with additional adjustment for the baseline value of CD4 cell count/HIV viral suppression. The Charlson comorbidity index is a weighted index used to characterize comorbidity in clinical research (Quan et al., 2005).

Results

Results of the LCA and characteristics by class membership

After a priori classification of participants with no use of cannabis, opioids, cocaine, or tranquilizers and an absence of unhealthy alcohol use in a separate class (NONE, n = 44), a two-class solution had the best fit for the remaining 206 participants in the sample. Of note, none of the participants reporting use of other stimulants (distinct from cocaine), inhalants, and hallucinogens (i.e., the substance use variables not used in the LCA because of a low prevalence of use in the sample) were classified in the NONE class. Fit statistics and entropy for the LCA are available in a supplemental table. Class 1 included 157 participants, and Class 2 included 49. Estimates of any past-30-day drug use and AUDIT-C score by class are presented in Table 1. Class 1 comprised participants with a high probability of cannabis use (52%) and unhealthy alcohol use (M AUDIT-C score = 6.04). The probability of cocaine use was 28%. Class 2 comprised participants all with opioid use (100%) and a high probability of cocaine use (68%), cannabis use (60%), tranquilizer use (36%), and unhealthy alcohol use (M AUDIT-C score = 6.09). Therefore, in addition to the a priori–defined class NONE, we identified one class of participants with mostly cannabis and unhealthy alcohol use (hereafter referred to as “CA”; n = 157, 63% of the sample) and one class of participants who used opioids and had a high prevalence of cocaine, tranquilizers, cannabis, and unhealthy alcohol use (hereafter referred to as “MULTI”; n = 49, 19% of the sample). Baseline characteristics of the study sample, by class, are presented in Table 2. Substance use was assessed over the past 30 days. Thus, one could report no use but still fulfill criteria for an SUD. The prevalence of SUDs is reported in Table 2 to assess the face validity of the identified classes.

Table 1.

Estimates of any past-30-day drug use and AUDIT-C score

graphic file with name jsad.21-00440.tbl1.jpg

Probability of Class 1 (CA), (n = 157) Estimate (SE) Class 2 (MULTI), (n = 49) Estimate (SE)
Used cocaine past 30 days 27.8% (3.7) 68.3% (13.0)
Used opioidsa past 30 days 7.3% (8.5) 100.0% (0)
Used cannabis past 30 days 52.1% (4.2) 60.0% (10.3)
Used tranquilizers past 30 days 3.4% (1.5) 35.8% (12.2)
Mean AUDIT-C score 6.04 (0.31) 6.09 (0.57)
Open in a new tab

Notes: AUDIT-C = Alcohol Use Disorders Identification Test–Consumption questions; CA = cannabis and alcohol; MULTI = opioids and other drugs.

a

Includes heroin, nonprescribed use of buprenorphine, methadone, and misused prescription opioids.

Table 2.

Baseline characteristics by class

graphic file with name jsad.21-00440.tbl2.jpg

Variable NONE (n = 44) n (%) or M (SD) CA (n = 157) n (%) or M (SD) MULTI (n = 49)n (%) or M (SD) p
Age, in years, M (SD) 53.9 (7.1) 48.0 (9.6) 46.4 (9.5) .0001
Sex
 Male 27.(61.4%) 101.(64.3%) 29.(59.2%) .7903
 Female 17.(38.6%) 56.(35.7%) 20.(40.8%)
Race/ethnicity
 Black or African American 20.(45.5%) 82.(52.2%) 23.(46.9%) .1090
 White 10.(22.7%) 28.(17.8%) 13.(26.5%)
 Hispanic 14.(31.8%) 35.(22.3%) 13.(26.5%)
 Multiracial/other 0 12.(7.6%) 0
Current employment status
 Not employed 34.(77.3%) 131.(83.4%) 45.(91.8%) .1528
Homelessnessa
 Yes 6.(13.6%) 40.(25.5%) 17.(34.7%) .0649
Any MINI DSM-IV drug dependence
 Yes 13.(29.6%) 123.(78.3%) 44.(89.8%) <.0001
Cocaine dependence
 Yes 11.(25%) 87.(55.4%) 36.(73.5%) <.0001
Opioid dependence
 Yes 8.(18.2%) 31.(19.8%) 34.(69.4%) <.0001
Cannabis dependence
 Yes 1.(2.3%) 51.(32.7%) 14.(28.6%) .0003
Tranquilizer dependence
 Yes 3.(6.8%) 13.(8.3%) 11.(22.5%) .0132
Alcohol dependence
 Yes 8.(18.2%) 110.(70.1%) 32.(65.3%) <.0001
Total grams of alcohol consumed,  past 30 days, M (SD) 40.1 (112.3) 857.8 (1483.2) 1,380.9 (2491.8) <.0001
Number of heavy drinking days,  past 30 days, M (SD) 0.2 (0.8) 5.4 (8.8) 5.7 (6.6) <.0001
Total SIP Score, M (SD) 1.9 (4.5) 10.8 (12.3) 12.9 (13.2) <.0001
Total SIP-D Score, M (SD) 6.1 (11.9) 13.6 (13.9) 21.(12.9) <.0001
Currently Taking ART
 Yes 42.(95.5%) 137.(87.8%) 41.(83.7%) .1977
Duration of HIV infection, in years,
 M (SD) 19.1 (8.2) 16.2 (8.4) 14.6 (8.3) .0390
Ever took ART
 Yes 43.(97.7%) 149.(95.5%) 47.(95.9%) .9002
Duration of ART, in years,b M (SD) Smoking status 16.0 (8.7) 11.7 (8.5) 10.7 (7.6) .0061
 Current smoker 28.(63.6%) 123.(78.9%) 44.(89.8%) .0397
 Former smoker 10.(22.7%) 18.(11.5%) 2.(4.1%)
 Never smoker 6.(13.6%) 15.(9.6%) 3.(6.1%)
Charlson Comorbidity Index, M (SD) 3.7 (2.6) 2.9 (2.5) 2.3 (2.1) .0110
CD4 Cell Count, M (SD) 569.5 (320.1) 548.5 (294.3) 633.5 (314) .2810
HIV viral load >200 8.(18.2%) 47.(30.1%) 16.(32.7%) .2327
Open in a new tab

Notes: NONE = a priori defined class with no use of opioids, cannabis, cocaine, or tranquilizers and no unhealthy alcohol use; CA= class with mostly cannabis and unhealthy alcohol use; MULTI = class with mostly opioids, cocaine, tranquilizers, cannabis, and unhealthy alcohol use. It may be understood as opioid and other drugs (including unhealthy alcohol use); MINI = Mini International Neuropsychiatric Interview; DSM-IV = Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition; SIP = Alcohol Short Index of Problems; SIP-D = Drug Short Index of Problems; ART = antiretroviral treatment.

a

Has spent one or more nights on the street or in a shelter in the past 6 months;

b

only among those with “yes” to “ever took ART.”

There were significant differences between classes in terms of the presence/absence of drug and alcohol dependence that were consistent with the class estimates of use. In addition, participants in MULTI reported fewer problems related to alcohol (lower SIP score) and more problems related to drug use (SIP-D score), whereas participants in NONE had the lowest SIP-D score. The prevalence of smoking increased from NONE to CA to MULTI. The Charlson comorbidity index was the highest in NONE, indicating poorer health. There were no differences in CD4 cell count or HIV viral suppression between groups. A limited number of participants reported injection drug use (n = 22).

Characterization of changes in substance use patterns over time

Of the 250 participants, data were available for 233 to assess for substance use pattern changes at 12 months (3 participants withdrew from the study, 3 died, and 11 others did not complete the 12-month assessment). Of the 233, 40% (n = 92) had a favorable course (i.e., change from baseline to 12 months: MULTI to CA: n = 23, either to NONE: n = 41, or remained as NONE: n = 28; 43% (n = 101) were in CA both times; 17% (n = 40) had a nonfavorable course (i.e., change from baseline to 12 months: NONE to CA: n = 11; or either to MULTI: n = 17, including the 12 who were in MULTI at both assessment time points). Details of the changes from baseline to 12 months are presented in Figure 1. Figure 1 is a Sankey diagram, a flow diagram in which the width of each band is proportional to the flow rate (the change over time).

Figure 1.

Figure 1.

Open in a new tab

Sankey diagram—changes from baseline to 12 months in substance use patterns. Notes: Class membership at baseline is presented on the left (blue). Class membership at 12 months is presented on the right (green). Bandwidths are proportional to the flow rate. NONE = a priori defined class with no use of opioids, cannabis, cocaine, or tranquilizers, and no unhealthy alcohol use; CA = class with mostly cannabis and unhealthy alcohol use; MULTI = class with mostly opioids, cocaine, tranquilizers, cannabis, and unhealthy alcohol use. It may be understood as opioid and other drugs (including unhealthy alcohol use). NO DATA = 3 participants withdrew from the study, 3 died, and 11 others also did not complete the 12-month assessment.

Association between 12-month changes in substance use patterns and HIV outcomes

Three groups were compared: participants who had a favorable course (n = 92), participants who had a nonfavorable course (n = 40), and participants who remained in CA (n = 101). Tables 3 and 4 present unadjusted models, models adjusted for baseline covariates, and models adjusting for baseline covariates and baseline value of CD4 cell count and HIV viral suppression. In the unadjusted model, in the model adjusted for baseline covariates, and in the model adjusting for baseline covariates and baseline CD4 count, participants who had a nonfavorable course had a significantly lower CD4 cell count compared with those who had a favorable course (-84, 95% CI [-156, -12.3]). Participants remaining in CA had a significantly lower CD4 count compared with those with a favorable course in the unadjusted model and in the model adjusted for baseline covariates, but not in the model adjusted for baseline covariates and baseline CD4 count. For models looking at the absence of HIV viral suppression, none of the associations were statistically significant.

Table 3.

Associations between 12-month changes in alcohol and drug use patterns and 12-month CD4 cell count

graphic file with name jsad.21-00440.tbl3.jpg

CD4 N Unadjusted Adjusted (1)a Adjusted (2)b
M CD4 [95% CI] Difference vs. NONE categoryc[95% CI] Adjusted M CD4 [95% CI] Difference vs. NONE category[95% CI] Adjusted M CD4 [95% CI] Difference vs. NONE category[95% CI]
Moved from a high substance use class to a lower class or remained in NONE 89 668.7 [598.9, 738.4] ref. 604.7 [509.2, 700.3] ref. 607.2[551.7, 662.6] ref.
Remained in CA 100 523.2 [464.6, 581.8] -145.5[-235.4, -55.6] 467.8[376.2, 559.4] -136.9 [-232.1, -41.7] 570.4 [516.4, 624.5] -36.7 [-92.8, 19.4]
Moved from a low substance use class to a higher class 39 519.1 [417.0, 621.1] -149.6[-268.1, -31.1] 445.7 [316.7, 574.6] -159.1[-281.9, -36.2] 523.2 [448.0, 598.4] -83.9[-155.6, -12.3]
Open in a new tab

Notes: n = 228, due to missing CD4 cell count; NONE = a priori defined class with no use of opioids, cannabis, cocaine, or tranquilizers and no unhealthy alcohol use; CA = class with mostly cannabis and unhealthy alcohol use; MULTI = class with mostly opioids, cocaine, tranquilizers, cannabis, and unhealthy alcohol use. It may be understood as opioid and other drugs (including unhealthy alcohol use); ref. = reference; CI = confidence interval.

a

Adjusted for baseline variables: age, sex, race (Black/African American, Hispanic, White, multiracial/other), smoking (current, former, never), homelessness, Charlson comorbidities (0–1, 2–3, ≥4), duration of HIV infection (years);

b

additionally adjusted for the baseline value of the outcome measure;

c

difference is the regression coefficient from multiple regression model.

Table 4.

Associations between 12-month changes in alcohol and drug use patterns and 12-month absence of HIV viral suppression

graphic file with name jsad.21-00440.tbl4.jpg

ABSENCE of HIV viral suppression (>200 copies/ml) N Unadjusted Adjusted (1)a OR [95% CI] Adjusted (2)b OR [95% CI]
N (%) OR [95% CI]
Moved from a high substance use class to a lower class or remained in NONE 89 13 (14.6%) ref. ref. ref.
Remained in CA 100 22 (22.0%) 1.65[0.78, 3.51] 1.68[0.74, 3.83] 1.55[0.64, 3.75]
Moved from a low substance use class to a higher class or remained in MULTI 39 11 (28.2%) 2.30[0.92, 5.72] 2.20[0.81, 6.02] 1.64[0.55, 4.89]
Open in a new tab

Notes: n = 228, due to missing HIV viral load. NONE = a priori defined class with no use of opioids, cannabis, cocaine, or tranquilizers and no unhealthy alcohol use; CA = class with mostly cannabis and unhealthy alcohol use; MULTI = class with mostly opioids, cocaine, tranquilizers, cannabis, and unhealthy alcohol use. It may be understood as opioid and other drugs (including unhealthy alcohol use); OR = odds ratio; CI = confidence interval; ref. = reference.

a

Adjusted for baseline variables: age, sex, race (Black/African American, Hispanic, White, multiracial/other), smoking (current, former, ever), homelessness, Charlson comorbidities (0–1, 2–3, 4+), duration of HIV infection (years);

b

additionally adjusted for the baseline value of the outcome measure.

Discussion

Among PWH with SUD, polysubstance use is highly prevalent and different patterns of substance use can be identified. In this sample of PWH with SUD (current alcohol and/or drug dependence or reporting a history of injection drug use) enrolled at an urban academic hospital–based infectious disease clinic and a community health center primary care clinic serving homeless patients, we were able to identify three distinct substance use patterns: abstinence from drugs and absence of unhealthy alcohol use, use of cannabis with unhealthy alcohol use, and use of opioids with unhealthy alcohol use and other drugs. These patterns correspond to clinical presentations and reflect the complexity of substance use in this population. Of note, in this cohort of PWH with SUD, opioid use does seem to occur along with other substance use, and when illicit and prescription drugs (including cannabis) are used, unhealthy alcohol use was typically present.

The patterns changed over time, with about two in five people's patterns changing to a likely lower risk pattern. Participants with a favorable course (i.e., with changes toward fewer substances or abstinence from drugs and absence of unhealthy alcohol use) had a significantly higher CD4 count compared with those with a nonfavorable course. Because CD4 count is a key marker of HIV progression and mortality risk, the observed changes related to a decrease in the number of substances used are relevant (Egger et al., 2002). Associations were not significant for HIV viral suppression.

The patterns found in this sample appear to differ from what has been identified in hospitalized PWH. Shiu-Yee et al. (2018) identified five substance use patterns: minimal drug use, cocaine use, substantial cocaine use/unhealthy alcohol use, polysubstance use, and substantial cocaine/ heroin use. Differences might be explained by the study sample (hospitalized PWH vs. recruitment in primary care infectious disease clinic and community health center primary care clinic; substance use vs. SUD) and data used (Shiu-Yee et al. used past-12-month substance use to conduct the LCA). Nevertheless, some commonalities can be identified, notably the fact that opioid use is happening alongside other substance use (largely cocaine, with a 66% probability estimate in Shiu-Yee et al. and 68% in our study). Also, the proportion of participants classified in an opioid use and polysubstance use class appear similar (opioid use class = 15% and polysubstance use class = 6% in Shiu-Yee et al., and 19% in opioid and polysubstance use in our study).

In a sample of PWH age 50 and older, Parsons et al. (2014) identified four distinct patterns of substance use: exclusive alcohol use, alcohol and marijuana, alcohol and cocaine/crack, and multiple substance use (weekly use of opiates in addition to other substances). The proportion of PWH not currently using substances was 16% (compared with 18% in our sample), even though the study did not restrict inclusion to participants with SUD. Similar to our study, the patterns identified indicate that substances are almost exclusively used in combinations and that opioid use is part of a pattern involving the use of other substances, including cocaine and alcohol (Parsons et al.).

A high prevalence of cannabis and unhealthy alcohol use among people using substances has also been observed in samples of people without HIV infection who report using drugs. Monga et al. (2007) found that among different patterns of drug use in a population using opioids, the prevalence of cannabis and alcohol use was more than 50% in all identified substance use patterns. What the current available research suggests is that, across different samples of PWH, polysubstance use patterns are prevalent and that opioid use is happening alongside other substance use, including cocaine and alcohol.

Evidence indicates that substance use patterns are associated with clinical outcomes. Substance use patterns are associated with substance use–related problem severity: Shiu-Yee et al. (2018) showed that participants with heroin/cocaine use had more severe substance use health-related problems. Substance use patterns may also be associated with other health-related outcomes. Reporting on cross-sectional associations, Parsons et al. (2014) indicate that participants reporting use of alcohol and crack cocaine and those with polysubstance use (weekly use of opiates in addition to other substances) had lower medication adherence and were less likely to have an undetectable HIV viral load. In a cluster analysis of substance use conducted among PWH in Spain, Fuster-RuizdeApodaca et al. (2019) showed that PWH with heroin and cocaine use had the poorest biological markers of HIV infection. Thus, the LCA approach allows characterization of groups of individuals who use substances on the basis of their substance use patterns, and this information can be used to tailor interventions adapted to the various substance use patterns and associated health consequences.

Given the available evidence and the ability to identify patterns of use, polysubstance use and its association with HIV outcomes should be studied specifically by substance use pattern. Given our results and the current evidence, it is especially important to take into account that opioid use among PWH is unlikely to happen with no associated use of other substances and unhealthy alcohol use. Of note, being able to support reductions in polysubstance use patterns may bring benefits for ART adherence and immune response, as alcohol and opioids have been linked to poor ART adherence and immunosuppression (Roy et al., 2011; Walley et al., 2008; Wang et al., 2005; Williams et al., 2016), but is not known whether changes in polysubstance use may bring additive or synergistic effects. The contribution of polysubstance use on adherence and whether the use of multiple substances multiplies the risk of poor ART adherence and immunosuppression remains to be studied in detail. Although not investigated in the current study, there is a clear benefit of opioid agonist therapy and addiction care for PWH with opioid use disorder, and access to care should be encouraged. Among people with opioid use disorder, opioid agonist therapy is the standard treatment, and it may have an indirect impact on other substance use (Dong et al., 2020).

The observed associations between substance use pattern changes and CD4 count are in line with what has been observed for PWH reaching abstinence or decreasing drug use without abstinence. Indeed, switching from use to no use of heroin, cocaine, and heavy alcohol use has been demonstrated to be associated with better HIV treatment outcomes (Lucas et al., 2002). In a study of 695 PWH, those who remained free of heroin, cocaine, or heavy alcohol use had better adherence to ART, more frequent viral suppression, and increased CD4 cell counts. Those who stopped using substances had an improvement in ART adherence and HIV outcomes, notably an increase in CD4 cell count (Lucas et al., 2002). Drug use may also play a role in preventing immune reconstitution in virologically suppressed patients with HIV (Jiang et al., 2018). This may explain why we observed an effect of polysubstance use on CD4 count but not on HIV viral suppression. In a 2020 study, Nance et al. found that abstinence or a reduction in use was associated with HIV viral suppression. Abstinence was associated with the highest odds of HIV viral suppression, but reducing opioid use or methamphetamine/crystal use, even without abstinence, was also associated with HIV viral suppression (Nance et al., 2020). The smaller differences observed in our study between those remaining in a pattern with cannabis and alcohol use and those decreasing substance use are also in line with results by Nance et al., who reported that a reduction in cannabis use had the smallest impact on HIV viral load. This study and our results suggest that treatments and interventions aimed at reducing substance use may have a positive impact on HIV disease progression. Observed improvements even when abstinence is not achieved should encourage interventions aimed at reduction of use when abstinence is not currently achievable or not acceptable for the patient. A study from the same cohort as this study found that virologic control was associated with fewer SUD criteria and did not find an association with abstinence (Nolan et al., 2017).

The present study has several limitations. First, although a cohort of 250 PWH with SUD is a substantial size, the number of participants and the low prevalence of use of some drugs did not allow for inclusion of a large number of substances in the LCA. Also, sample size may have affected our ability to detect the large differences observed in viral load suppression. Second, data used relied on self-reports, which can be subject to social desirability bias, especially for the use of illicit substances; on the other hand, the assessment was confidential, nonjudgmental, had no implications for clinical care or services, and was able to detect substances used less frequently that might not have been detected with biological tests. Third, we did assess changes from baseline to 12 months but relied on measures for drugs that captured use over the past 30 days; such use likely correlates with but may not be the same as use during intervening periods. Any inaccuracies would likely have biased analyses toward the null. Substance use is also not equivalent to an SUD, as use may happen in the absence of an SUD (and absence of use over the past 30 days could be reported by participants with a current SUD). Fourth, study participants recruited from clinical settings were, in general, reasonably well connected to clinical HIV care. Fifth, as our analyses were adjusted for baseline covariates, changes in substance use patterns over time may reflect changes in social stressors occurring over time (changes in housing stability, employment status, etc.) that were not taken into account in the baseline adjustment. The generalizability of our results should be made with caution to other populations of PWH with SUDs. Of note, there may be geographical and historical factors influencing substance use availability and use practices. It is also important to take into account that the current study sample was recruited among people accessing care.

Conclusions

LCA allows for identification of subgroups in populations. This technique has been used with note to identify subgroups of people with higher HIV-risk behaviors (Harrell et al., 2012; Meacham et al., 2015; Trenz et al., 2013). Our findings suggest that polysubstance use is common and that there are identifiable patterns. Patterns change over time, and changes toward fewer substances or no use were associated with lower HIV disease severity (based on CD4 count but not based on HIV viral suppression) compared with the use of multiple substances. These observations suggest that future studies of PWH with SUD should account for polysubstance use. The findings could motivate patients to change their substance use to improve their HIV disease severity, and they also support the need in clinical practice to address multiple substances in PWH with SUD to improve health outcomes. Intervention studies providing patients with this information and that address polysubstance use in patients and assess outcomes would further confirm the clinical utility of these findings. These results also have public health implications, notably because they point to the existence of groups with different substance use patterns that may need specific interventions. Substance use patterns allow for identification of populations more at risk, and thus in need of more attention. When opioid use is targeted in public health interventions, it should be taken into account that changes in other substance use, including alcohol, should be targeted as well.

Footnotes

This study was supported by Grants U01AA020784, U24AA020778, U24AA020779, and UL1TR001430 from the National Institute on Alcohol Abuse and Alcoholism and National Center for Advancing Translational Science.

References

  1. Alterman A. I., Cacciola J. S., Ivey M. A., Habing B., Lynch K. G. Reliability and validity of the alcohol Short Index of Problems and a newly constructed drug Short Index of Problems. Journal of Studies on Alcohol and Drugs. 2009;70:304–307. doi: 10.15288/jsad.2009.70.304. doi:10.15288/jsad.2009.70.304. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. American Psychiatric Association. (4th ed.). Washington, DC: Author; 1994. Diagnostic and statistical manual of mental disorders. [Google Scholar]
  3. Arnsten J. H., Demas P. A., Grant R. W., Gourevitch M. N., Farzadegan H., Howard A. A., Schoenbaum E. E. Impact of active drug use on antiretroviral therapy adherence and viral suppression in HIV-infected drug users. Journal of General Internal Medicine. 2002;17:377–381. doi: 10.1046/j.1525-1497.2002.10644.x. doi:10.1007/s11606-002-0044-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bush K., Kivlahan D. R., McDonell M. B., Fihn S. D., Bradley K. A. The AUDIT alcohol consumption questions (AUDIT-C): An effective brief screening test for problem drinking. Archives of Internal Medicine. 1998;158:1789–1795. doi: 10.1001/archinte.158.16.1789. doi:10.1001/archinte.158.16.1789. [DOI] [PubMed] [Google Scholar]
  5. Charlson M. E., Pompei P., Ales K. L., MacKenzie C. R. A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. Journal of Chronic Diseases. 1987;40:373–383. doi: 10.1016/0021-9681(87)90171-8. doi:10.1016/0021-9681(87)90171-8. [DOI] [PubMed] [Google Scholar]
  6. Cofrancesco J., Jr., Scherzer R., Tien P. C., Gibert C. L., Southwell H., Sidney S., Grunfeld C. Illicit drug use and HIV treatment outcomes in a US cohort. AIDS. 2008;22:357–365. doi: 10.1097/QAD.0b013e3282f3cc21. doi:10.1097/QAD.0b013e3282f3cc21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Dong H., Hayashi K., Milloy M. J., DeBeck K., Singer J., Wong H., Kerr T. Changes in substance use in relation to opioid agonist therapy among people who use drugs in a Canadian setting. Drug and Alcohol Dependence. 2020;212:108005. doi: 10.1016/j.drugalcdep.2020.108005. doi:10.1016/j.drugalcdep.2020.108005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Egger M., May M., Chêne G., Phillips A. N., Ledergerber B., Dabis F., Sterne J. A. C., the ART Cohort Collaboration Prognosis of HIV-1-infected patients starting highly active antiretroviral therapy: A collaborative analysis of prospective studies. The Lancet. 2002;360:119–129. doi: 10.1016/s0140-6736(02)09411-4. doi:10.1016/S0140-6736(02)09411-4. [DOI] [PubMed] [Google Scholar]
  9. El-Bassel N., Shaw S. A., Dasgupta A., Strathdee S. A. Drug use as a driver of HIV risks: Re-emerging and emerging issues. Current Opinion in HIV and AIDS. 2014;9:150–155. doi: 10.1097/COH.0000000000000035. doi:10.1097/COH.0000000000000035. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Fuster-RuizdeApodaca M. J., Castro-Granell V., Garin N.,Laguía A., Jaén Á., Iniesta C., Galindo M. J.2019Prevalence and patterns of illicit drug use in people living with HIV in Spain: A cross-sectional study PLoS One 14e0211252.doi:10.1371/journal.pone.0211252 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Hannemann T. V., Kraus L., Piontek D. Consumption patterns of nightlife attendees in Munich: A latent-class analysis. Substance Use & Misuse. 2017;52:1511–1521. doi: 10.1080/10826084.2017.1290115. doi:10.1080/10826084.2017.1290115. [DOI] [PubMed] [Google Scholar]
  12. Harrell P. T., Mancha B. E., Petras H., Trenz R. C., Latimer W. W. Latent classes of heroin and cocaine users predict unique HIV/HCV risk factors. Drug and Alcohol Dependence. 2012;122:220–227. doi: 10.1016/j.drugalcdep.2011.10.001. doi:10.1016/j.drugalcdep.2011.10.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Hasin D. S., Grant B. F. The National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) Waves 1 and 2: Review and summary of findings. Social Psychiatry and Psychiatric Epidemiology. 2015;50:1609–1640. doi: 10.1007/s00127-015-1088-0. doi:10.1007/s00127-015-1088-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Jiang W., Luo Z., Martin L., Wan Z., Fu P., Wagner A., McRae-Clark A. Drug use is associated with anti-CD4 IgG-mediated CD4+ T cell death and poor CD4+ T cell recovery in viral-suppressive HIV-infected individuals under antiretroviral therapy. Current HIV Research. 2018;16:143–150. doi: 10.2174/1570162X16666180703151208. doi:10.2174/1570162X16666180703151208. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Kader R., Govender R., Seedat S., Koch J. R., Parry C. Understanding the impact of hazardous and harmful use of alcohol and/or other drugs on ARV adherence and disease progression. PLoS One. 2015;10:e0125088. doi: 10.1371/journal.pone.0125088. doi:10.1371/journal.pone.0125088. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Korthuis P. T., Fiellin D. A., McGinnis K. A., Skanderson M., Justice A. C., Gordon A. J., Kraemer K. L. Unhealthy alcohol and illicit drug use are associated with decreased quality of HIV care. Journal of Acquired Immune Deficiency Syndromes. 2012;61:171–178. doi: 10.1097/QAI.0b013e31826741aa. doi:10.1097/QAI.0b013e31826741aa. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Korthuis P. T., Zephyrin L. C., Fleishman J. A., Saha S., Josephs J. S., McGrath M. M., Gebo K. A. Health-related quality of life in HIV-infected patients: The role of substance use. AIDS Patient Care and STDs. 2008;22:859–867. doi: 10.1089/apc.2008.0005. doi:10.1089/apc.2008.0005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Ladak F., Socias E., Nolan S., Dong H., Kerr T., Wood E., Milloy M.-J. Substance use patterns and HIV-1 RNA viral load rebound among HIV-positive illicit drug users in a Canadian setting. Antiviral Therapy. 2019;24:19–25. doi: 10.3851/IMP3265. doi:10.3851/IMP3265. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Lucas G. M., Gebo K. A., Chaisson R. E., Moore R. D. Longitudinal assessment of the effects of drug and alcohol abuse on HIV-1 treatment outcomes in an urban clinic. AIDS. 2002;16:767–774. doi: 10.1097/00002030-200203290-00012. doi:10.1097/00002030-200203290-00012. [DOI] [PubMed] [Google Scholar]
  20. Lynskey M. T., Agrawal A., Bucholz K. K., Nelson E. C., Madden P. A. F., Todorov A. A., Heath A. C. Subtypes of illicit drug users: A latent class analysis of data from an Australian twin sample. Twin Research and Human Genetics. 2006;9:523–530. doi: 10.1375/183242706778024964. doi:10.1375/twin.9.4.523. [DOI] [PubMed] [Google Scholar]
  21. McCabe S. E., Cranford J. A., West B. T. Trends in prescription drug abuse and dependence, co-occurrence with other substance use disorders, and treatment utilization: Results from two national surveys. Addictive Behaviors. 2008;33:1297–1305. doi: 10.1016/j.addbeh.2008.06.005. doi:10.1016/j.addbeh.2008.06.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. McLellan A. T., Kushner H., Metzger D., Peters R., Smith I., Gris-som G., Argeriou M. The Fifth Edition of the Addiction Severity Index. Journal of Substance Abuse Treatment. 1992;9:199–213. doi: 10.1016/0740-5472(92)90062-s. doi:10.1016/0740-5472(92)90062-s. [DOI] [PubMed] [Google Scholar]
  23. Meacham M. C., Rudolph A. E., Strathdee S. A., Rusch M. L., Brouwer K. C., Patterson T. L., Roesch S. C. Polydrug use and HIV risk among people who inject heroin in Tijuana, Mexico: A latent class analysis. Substance Use & Misuse. 2015;50:1351–1359. doi: 10.3109/10826084.2015.1013132. doi:10.3109/10826084.2015.1013132. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Mimiaga M. J., Reisner S. L., Grasso C., Crane H. M., Safren S. A., Kitahata M. M., Mayer K. H. Substance use among HIV-infected patients engaged in primary care in the United States: Findings from the Centers for AIDS Research Network of Integrated Clinical Systems cohort. American Journal of Public Health. 2013;103:1457–1467. doi: 10.2105/AJPH.2012.301162. doi:10.2105/AJPH.2012.301162. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Monga N., Rehm J., Fischer B., Brissette S., Bruneau J., El-Guebaly N., Bahl S. Using latent class analysis (LCA) to analyze patterns of drug use in a population of illegal opioid users. Drug and Alcohol Dependence. 2007;88:1–8. doi: 10.1016/j.drugalcdep.2006.08.029. doi:10.1016/j.drugalcdep.2006.08.029. [DOI] [PubMed] [Google Scholar]
  26. Moore R. D., Keruly J. C., Chaisson R. E. Differences in HIV disease progression by injecting drug use in HIV-infected persons in care. Journal of Acquired Immune Deficiency Syndromes. 2004;35:46–51. doi: 10.1097/00126334-200401010-00006. doi:10.1097/00126334-200401010-00006. [DOI] [PubMed] [Google Scholar]
  27. Muthén L.K. and Muthén B.O. Seventh edition. Los Angeles, CA: Muthén & Muthén; 1998-2012. Mplus user's guide. [Google Scholar]
  28. Nance R. M., Trejo M. E. P., Whitney B. M., Delaney J. A. C., Altice F., Beckwith C. G., Crane H. M. Impact of abstinence and of reducing illicit drug use without abstinence on human immunodeficiency virus viral load. Clinical Infectious Diseases. 2020;70:867–874. doi: 10.1093/cid/ciz299. doi:10.1093/cid/ciz299. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Nolan S., Walley A. Y., Heeren T. C., Patts G. J., Ventura A. S., Sullivan M. M., Saitz R. HIV-infected individuals who use alcohol and other drugs, and virologic suppression. AIDS Care. 2017;29:1129–1136. doi: 10.1080/09540121.2017.1327646. doi:10.1080/09540121.2017.1327646. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Palepu A., Tyndall M., Yip B., O’Shaughnessy M. V., Hogg R. S., Montaner J. S. Impaired virologic response to highly active antiretroviral therapy associated with ongoing injection drug use. Journal of Acquired Immune Deficiency Syndromes. 2003;32:522–526. doi: 10.1097/00126334-200304150-00009. doi:10.1097/00126334-200304150-00009. [DOI] [PubMed] [Google Scholar]
  31. Parsons J. T., Starks T. J., Millar B. M., Boonrai K., Marcotte D. Patterns of substance use among HIV-positive adults over 50: Implications for treatment and medication adherence. Drug and Alcohol Dependence. 2014;139:33–40. doi: 10.1016/j.drugalcdep.2014.02.704. doi:10.1016/j.drugalcdep.2014.02.704. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Quan H., Sundararajan V., Halfon P., Fong A., Burnand B., Luthi J.-C., Ghali W. A. Coding algorithms for defining comorbidi-ties in ICD-9-CM and ICD-10 administrative data. Medical Care. 2005;43:1130–1139. doi: 10.1097/01.mlr.0000182534.19832.83. doi:10.1097/01.mlr.0000182534.19832.83. [DOI] [PubMed] [Google Scholar]
  33. Roy S., Ninkovic J., Banerjee S., Charboneau R. G., Das S., Dutta R., Barke R. A. Opioid drug abuse and modulation of immune function: consequences in the susceptibility to opportunistic infections. Journal of Neuroimmune Pharmacology. 2011;6:442–465. doi: 10.1007/s11481-011-9292-5. doi:10.1007/s11481-011-9292-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Saitz R., Mesic A., Ventura A. S., Winter M. R., Heeren T. C., Sullivan M. M., Samet J. H. Alcohol consumption and bone mineral density in people with HIV and substance use disorder: A prospective cohort study. Alcoholism: Clinical and Experimental Research. 2018;42:1518–1529. doi: 10.1111/acer.13801. doi:10.1111/acer.13801. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Samet J. H., Phillips S. J., Horton N. J., Traphagen E. T., Freedberg K. A. Detecting alcohol problems in HIV-infected patients: Use of the CAGE questionnaire. AIDS Research and Human Retroviruses. 2004;20:151–155. doi: 10.1089/088922204773004860. doi:10.1089/088922204773004860. [DOI] [PubMed] [Google Scholar]
  36. Samji H., Cescon A., Hogg R. S., Modur S. P., Althoff K. N., Buchacz K., Gange S. J., & the North American AIDS Cohort Collaboration on Research and Design (NA-ACCORD) of IeDEA Closing the gap: Increases in life expectancy among treated HIV-positive individuals in the United States and Canada. PLoS One. 2013;8:e81355. doi: 10.1371/journal.pone.0081355. doi:10.1371/journal.pone.0081355. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Sheehan D. V., Sheehan K. H., Shytle R. D., Janavs J., Bannon Y., Rogers J. E., Wilkinson B. Reliability and validity of the Mini International Neuropsychiatric Interview for Children and Adolescents (MINI-KID) Journal of Clinical Psychiatry. 2010;71:313–326. doi: 10.4088/JCP.09m05305whi. doi:10.4088/JCP.09m05305whi. [DOI] [PubMed] [Google Scholar]
  38. Shiau S., Arpadi S. M., Yin M. T., Martins S. S. Patterns of drug use and HIV infection among adults in a nationally representative sample. Addictive Behaviors. 2017;68:39–44. doi: 10.1016/j.addbeh.2017.01.015. doi:10.1016/j.addbeh.2017.01.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Shiu-Yee K., Brincks A. M., Feaster D. J., Frimpong J. A., Nijhawan A., Mandler R. N, Metsch L. R. Patterns of substance use and arrest histories among hospitalized HIV drug users: A latent class analysis. AIDS and Behavior. 2018;22:2757–2765. doi: 10.1007/s10461-017-2024-y. doi:10.1007/s10461-017-2024-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Skeer M. R., Mimiaga M. J., Mayer K. H., O’Cleirigh C., Covahey C., Safren S. A. Patterns of substance use among a large urban cohort of HIV-infected men who have sex with men in primary care. AIDS and Behavior. 2012;16:676–689. doi: 10.1007/s10461-011-9880-7. doi:10.1007/s10461-011-9880-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Tomczyk S., Hanewinkel R., Isensee B. Multiple substance use patterns in adolescents—A multilevel latent class analysis. Drug and Alcohol Dependence. 2015;155:208–214. doi: 10.1016/j.drugalcdep.2015.07.016. doi:10.1016/j.drugalcdep.2015.07.016. [DOI] [PubMed] [Google Scholar]
  42. Trenz R. C., Scherer M., Duncan A., Harrell P. T., Moleko A. G., Latimer W. W. Latent class analysis of polysubstance use, sexual risk behaviors, and infectious disease among South African drug users. Drug and Alcohol Dependence. 2013;132:441–448. doi: 10.1016/j.drugalcdep.2013.03.004. doi:10.1016/j.drugalcdep.2013.03.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Ventura A. S., Winter M. R., Heeren T. C., Sullivan M. M., Walley A. Y., Holick M. F., Saitz R. Lifetime and recent alcohol use and bone mineral density in adults with HIV infection and substance dependence. Medicine. 2017;96:e6759. doi: 10.1097/MD.0000000000006759. doi:10.1097/MD.0000000000006759. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Walley A. Y., Cheng D. M., Libman H., Nunes D., Horsburgh C. R., Jr.,Saitz R., Samet J. H.2008Recent drug use, homelessness and increased short-term mortality in HIV-infected persons with alcohol problems AIDS 22415–420.doi:10.1097/QAD.0b013e3282f423f8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Wang X., Tan N., Douglas S. D., Zhang T., Wang Y. J., Ho W. Z. Morphine inhibits CD8+ T cell-mediated, noncytolytic, anti-HIV activity in latently infected immune cells. Journal of Leukocyte Biology. 2005;78:772–776. doi: 10.1189/jlb.0305167. doi:10.1189/jlb.0305167. [DOI] [PubMed] [Google Scholar]
  46. Williams E. C., Hahn J. A., Saitz R., Bryant K., Lira M. C., Samet J. H. Alcohol use and human immunodeficiency virus (HIV) infection: Current knowledge, implications, and future directions. Alcoholism: Clinical and Experimental Research. 2016;40:2056–2072. doi: 10.1111/acer.13204. doi:10.1111/acer.13204. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Journal of Studies on Alcohol and Drugs are provided here courtesy of Rutgers University. Center of Alcohol Studies

RESOURCES