Tina’s ParTy Line: Polysubstance Use Patterns in Sexual Minority Men Living with HIV Who Use Methamphetamine

Abstract

Background:

This cross-sectional study examined the associations of demographic, structural, and psychological factors with distinct typologies of polysubstance use in sexual minority men (SMM) living with HIV who use methamphetamine.

Methods:

In total, 161 SMM living with HIV who reported methamphetamine use in the past three months were recruited in San Francisco from 2013-2017 for a randomized controlled trial. A latent class analysis was conducted by leveraging baseline measures of self-reported use of 15 substances in the past three months as well as validated screening measures of hazardous alcohol and cannabis use. Correlates of latent class membership were examined using a 3-step categorical latent variable logistic regression.

Results:

Four typologies of substance use were identified: 1) methamphetamine use only (43%), 2) methamphetamine and crack-cocaine use (22%), 3) party and play (PnP) – methamphetamine, gamma-hydroxybutyrate, and amyl nitrites (i.e., poppers) with erectile dysfunction drugs (31%) and 4) high polysubstance use (4%). SMM of color and those with a history of incarceration were more commonly classified as engaging in methamphetamine and crack-cocaine use compared to PnP use. Men with higher sexual compulsivity scores were more commonly classified as engaging in PnP use and polysubstance use.

Conclusions:

There is substantial heterogeneity in polysubstance use patterns among SMM living with HIV who use methamphetamine This will inform the development of tailored substance use interventions addressing the unique needs of SMM of color and targeting sexual compulsivity as a prominent comorbidity for some men.

1. Introduction

The prevalence of methamphetamine (meth) use is more than 10 times higher among gay, bisexual, and other men who have sex with men (referred to here collectively as sexual minority men, or SMM) compared to the general population in the United States (10-27% vs. 0.6% respectively) (1–5). Among SMM who use meth, polysubstance use is common—whereby men report using multiple drugs simultaneously or sequentially over a brief period of time (6–8). For other SMM the use of at least two “club drugs” such as ecstasy, cocaine, meth, gamma-hydroxybutyrate (GHB), or ketamine is particularly common (9–11). At the same time, estimates of the prevalence for polysubstance use among SMM who use meth have varied from approximately one-in-four men to as high as 95% (6, 7, 12, 13). An important gap, addressed in the present study, is that relatively little is known about correlates of distinct, empirically derived typologies of polysubstance use to inform tailored substance use interventions for potentially distinct subgroups of SMM who use meth.

Meth and polysubstance use are linked to increased drug and sexual risk behaviors that heighten risk of HIV or STI infections (6, 13, 14). For example, one cohort study of SMM found that those who use stimulants, erectile dysfunction drugs (EDD), and amyl nitrites (i.e., poppers) had an 8-fold greater risk of HIV seroconversion compared to non-users (15). Additionally, the use of stimulants such as meth has important implications for faster clinical HIV progression (16–18). People living with HIV who use stimulants such as meth experience profound difficulties with antiretroviral therapy (ART) adherence and persistence that may explain slower rates of viral suppression even in the era of HIV treatment as prevention (19–21). Not only do these difficulties that stimulant users experience with HIV disease management increase risk for clinical HIV progression but they also amplify risk for onward HIV transmission in SMM who use meth (22).

Difficulties managing sexual thoughts and behaviors, referred to here as sexual compulsivity, is a prevalent mental health comorbidity that fuels substance use among SMM (23). United States (US) general population estimates of sexual compulsivity are between 3 and 6%, but range from 19-30% of SMM to as high as 51% in samples of highly sexually active SMM (23–25). In the US, sexual minorities experience a higher prevalence of mental health and substance use disorders that are linked to social stigma related to their sexual orientation as well as race/ethnicity for SMM of color (26–28). Sexual compulsivity (referred to as Hypersexual Disorder in the Diagnostic and Statistical Manual of Mental Disorders, 5^th Edition) is one mental health comorbidity among SMM that has been linked to stimulant use, condomless anal intercourse, and HIV-positive serostatus (29–31). In fact, specific patterns of substance use may be due in large part to the desire to enhance sexual experiences, increase libido, and alleviate negative affect due to sexual minority stress among SMM who use meth (1, 31, 32).

Polysubstance use can amplify the hedonic experiences of using by achieving interactions between substances or counteracting the negative side effects (13). For example, meth may also be combined with EDD to counteract meth-induced erectile dysfunction and extend sexual encounters (33). ChemSex, a term popularized in Europe to describe the co-use of substances among SMM to increase sexual pleasure, involves the combination mephedrone, GHB, ecstasy, and meth (34). In the United States, some SMM also engage in polysubstance use to enhance sexual experiences—often termed party and play (PnP). PnP is more commonly linked to co-use of meth with other substances such as GHB, poppers, and EDD (35, 36).

Latent class analysis (LCA) is an advanced approach to structural equation modeling where latent sub-groups are constructed based on multivariate, categorical responses that holds promise for objectively classifying polysubstance use patterns among SMM. McCarty-Caplan and colleagues (2014) performed a LCA that identified three classes of substance use: low or moderate use, sex-drug use (e.g., meth, club drugs, EDD, and poppers), and polysubstance use. Those the sex-drug category were more likely to be HIV-positive and engage in condomless insertive anal sex with partner of unknown HIV serostatus (7). This is supported by findings from other LCAs where those classified as polysubstance users were more likely to be SMM and report greater sexual sensation seeking or were greater than three times more likely to report condomless anal intercourse or having an STI in the last year (37, 38).

This cross-sectional study had two primary objectives. First, we sought to characterize distinct typologies of polysubstance use in SMM living with HIV who use meth with LCA. Second, we examined demographic, structural, and psychological correlates of latent class membership for distinct typologies of substance use to inform the development of tailored substance use intervention approaches for this high priority population.

2. Methods

SMM living with HIV who use meth were recruited from substance use disorder treatment programs, HIV medical clinics, AIDS service organizations, the community, and referrals from active participants for a randomized controlled trial of a positive affect intervention delivered during contingency management for stimulant abstinence (39, 40). At an in-person screening visit, participants completed a signed informed consent. All participants who completed the screening met the following inclusion criteria: 1) 18 years of age or older; 2) man who has sex with men; 3) documentation of HIV-positive serostatus (i.e., letter of diagnosis or ART medications matched to photo identification); and 4) self-reported meth use in the past three months. The screening visit included administration of self-report measures, on-site urine toxicology testing, and collection of hair samples for toxicology testing of men who provided a urine sample that was non-reactive for meth. Those with documented evidence of recent meth use in urine or hair toxicology testing completed a separate baseline assessment approximately one week later that included self-report measures, a second urine sample for on-site toxicology testing, and a peripheral venous blood sample to measure HIV disease markers. Data were collected in San Francisco, California from 2013 through 2017. All relevant procedures were approved by the Institutional Review Board for the University of California, San Francisco with reliance agreements from the University of Miami and Northwestern University. All participants completed a signed informed consent and a certificate of confidentiality was provided by the National Institute on Drug Abuse.

2.1. Measures

2.1.1. Demographics and structural factors.

Participants reported their age, race/ethnicity, education level, and income in a demographic questionnaire. Participants also reported whether they had been homeless in the past year as well as whether they had been incarcerated (i.e., jail or prison) in their lifetime.

2.1.2. Substance use.

Participants reported how often they used meth, powder cocaine, crack-cocaine, ketamine, methyl enedioxy methamphetamine (MDMA), GHB, poppers, EDD medications while feeling the effects of substances, benzodiazepines, poppers, barbiturates, methadone and prescription opioids in the past three months. Consistent with validated measures, participants were instructed to report use of substances that were either prescribed, used more than prescribed by the doctor, or used for reasons other than prescribed (41). Frequency of use during the past three months for each substance was rated separately on a Likert-type scale from zero (not at all) to seven (daily). Frequency ratings for each substance were dichotomized as no use or less than once a month (coded as 0) versus monthly or greater (coded as 1) in the past three months. This was based on previous research defining problematic substance use as one or more times per month (42). By extension polysubstance use was defined as the monthly use of at least one drug in addition to meth.

2.1.3. Alcohol Use Disorders Identification Test (AUDIT).

This 10-item screening measure for hazardous alcohol use was developed by the World Health Organization to assess alcohol consumption, drinking behaviors, and alcohol-related problems (43). AUDIT scores of eight or more are indicative of hazardous alcohol use.

2.1.4. Cannabis Use Identification Test-Revised (CUDIT-R).

This 8-item screening measure is validated to identify problematic patterns of cannabis use (44). CUDIT-R scores of eight or more are indicative of hazardous cannabis use.

2.1.5. Psychological factors.

The Center for Epidemiological Studies- Depression (CES-D) is a 20-item measure asking participants to rate how often over the past week they have experienced symptoms associated with depression. Those with a CES-D total score of 16 or greater (coded as 1) were classified as being at greater risk of depression compared to those with a score below 16 (coded as 0) who were classified as not screening positive for depression (45). The Differential Emotions Scales was used to assess continuous positive affect and negative affect sub-scales separately (46). The sample had an overall mean for positive affect of 25.48 (standard deviation = 6.78; Cronbach’s alpha =0.87) and a mean for negative affect of 13.31 (standard deviation = 5.83; Cronbach’s alpha = 0.85). The Post-Traumatic Stress Disorder Checklist – Civilian (PCL-C) is a standardized self-report rating scale for PTSD comprising 17 items that correspond to the key symptoms of PTSD (47), the total PCL-C score was examined. The overall sample had a mean PCL-C score of 46.49 (standard deviation = 14.30; Cronbach’s alpha = 0.92). Sexual compulsivity was measured using the Sexual Compulsivity Scale, a 10-item continuous composite score indexing difficulties with controlling sexual thoughts and behaviors (48). The overall sample had a sexual compulsivity mean of 18.74 (standard deviation = 7.54; Cronbach’s alpha = 0.92).

2.1.6. Detectable HIV viral load.

HIV viral load testing was performed to detect plasma HIV RNA using the Abbott Real Time HIV-1 assay (Abbott Molecular, Inc.; Des Plaines, IL). The lower limit of detection was 40 copies/mL. For this study viral load was dichotomized as detectable (≥ 40 copies/mL coded as 1) versus undetectable (< 40 copies/mL coded as 0).

2.2. Statistical Analysis

LCA was used to identify patterns of polysubstance use in SMM living with HIV who use methamphetamine. LCA is a person-centered analytic approach that aims to identify underlying patterns of covariance in the data structure to identify ‘classes’ or sub-groups of participants. Fifteen binary substance use indicator variables for meth, powder cocaine, crack-cocaine, ketamine, MDMA, GHB, poppers, EDD medications while feeling the effects of substances, benzodiazepines, barbiturates, prescription opioids, heroin, hazardous alcohol use, and hazardous cannabis use were entered into the LCA model. As shown in Figure 1, the LCA model using 15 binary indicators began with one class and added classes incrementally to 5 classes.

Figure 1:

Latent Class Analysis of Substance Use Typologies Among Sexual Minority Men Living with HIV Who Use Methamphetamine

Each model was tested for fit using 4 separate measures: the Bayesian Information Criteria (BIC), adjusted Bayesian Information Criteria (ABIC), the Lo-Mendell-Rubin adjusted likelihood ratio test (LMR), and the parametric bootstrap likelihood ratio test (BLRT). The BIC and ABIC are interpreted such that the lowest value is considered the best fit (49). The LMR and the BLRT both provide comparisons between models, such that significant values indicate the model is an improved fit over the model with less classes. Entropy is not used as a measure of fit, however lower entropy values associated with a given model may indicate that the model is not illustrating groups with uniquely separate characteristics. Meaningfulness of the patterns of drug use was also considered in the selection of the final class structure. Classes were assigned using most likely class membership (50), and correlates of class membership were first examined using chi-square and analysis of variance (ANOVA) tests.

3. Results

Among the 161 participants who completed a screening visit, the majority were racial/ethnic minorities (54%): 16% Black, 27% Hispanic/Latino, and 11% Other, with the remaining 46% identifying as non-Hispanic white. The mean age of participants was 43.8 years (SD = 8.98). Two-thirds (67%) of participants made less than $16,000 a year. One-in-four (25%) had a high school education or less, 47% had some college or trade school education, and 29% had completed college or graduate degrees. Nearly one-third (32%) had a detectable viral load. Approximately one-third (32%) had experienced homelessness in past year and 38% had been incarcerated in jail or prison at some point in their life.

With regard to self-reported substance use, participants were most likely to report past using GHB (43%), poppers (37%), and EDD while feeling the effects of other substances (39%) at least once a month in the past three months. They were least likely to report using barbiturates (5%), ketamine (8%), heroin (11%), methadone (11%), and MDMA (15%) at least once a month in the past three months. Participants reported similar frequencies of crack-cocaine, benzodiazepine, prescription opioids, and powder cocaine use (24%, 24%, 22%, and 21% respectively). One-in-five participants (21%) screened positive for hazardous alcohol use or hazardous cannabis use (21%).

3.1. Substance use classes.

The LCA model with the four-class solution was identified as the best-fitting model for identifying patterns related to self-reported substance use. This model had a high entropy, the lowest AIC and ABIC values, and a statistically significant bootstrap likelihood ratio test (BLRT). Specifically, the estimates for AIC, ABIC, BLRT, and entropy for the four-class model were 2045.2, 2040.3, 47.9 (p-value < 0.05), and 0.871 respectively; compared with those obtained from the three-class model (ABIC = 2059.8; LMR = 84.3, p-value = 0.12; BLRT = 85.4, p-value <0.01 ; entropy = 0.875) and the five-class model (ABIC = 2045.3; LMR = 29.4, p-value = 0.07; BLRT = 29.7, p-value = 0.21; entropy = 0.87). The four-class model was selected based on the model fit indices, its consistency with prior studies, and its ability to provide class meaningful heterogeneity without diminishing cell size to examine correlates of class membership. Table 1 depicts the results for the model fit criteria. Figure 2 depicts the visual representation of the results for the four-class model.

Table 1.

Fit Indices for Latent Classes of Substance Use in Sexual Minority Men Living with HIV Who Use Methamphetamine (N = 161)

	AIC	BIC	ABIC	LMR LRT	BLRT	Entropy
1 Group	224.637	2287.858	2240.372	-	-	-
2 Group	2115.599	2211.314	2113.175	157.738; p-value = 0.0374	159.675; p-value <0.0001	0.822
3 Group	2063.491	2208.608	2059.817	84.338; p-value = 0.1201	85.375; p-value <0.0001	0.875
4 Group	2045.223	2239.741	2040.297	47.371; p-value = 0.69	47.953; p-value = 0.012	0.871
5 Group	2051.519	2295.439	2045.342	29.380; p-value = 0.0641	29.741; p-value 0.2083	0.87

^*p <0.05;

^**p < 0.01

AIC =Akaike Information Criterion ; BIC = Bayesian Information Criterion; ABIC = Adjusted Bayesian Information Criterion ; LMR LRT = Lo-Mendell-Rubin adjusted likelihood ratio test ; BLRT = Bootstrap Likelihood Ratio Test ;

Figure 2.

Probability of Substance Use by Class

As shown in Figure 2a, Class 1 (n = 69) exhibited high usage of meth only (94%). This class was 44.6% non-Hispanic white, and one-third (33.9%) had experienced homelessness in the last year. Class 1 had the highest number of individuals with undetectable viral load (56.7%) compared to the other classes.

As shown in Figure 2b, Class 2 (n = 35) was characterized by high meth and crack-cocaine co-usage, all participants reported using meth in the past three months and 83.5% reported using crack-cocaine in the past three months (Figure 2b). Stimulants accounted for most of the drug usage in this group, including a higher probability of using powder cocaine (41.2%) than any other substances. Class 2 was the most racially diverse: 28.6% Black, 28.6% Non-Hispanic White, 28.6% Hispanic/Latino, and 14.2% reported as other. Class 2 also had the highest number of individuals who had experienced incarceration in their lifetime (51.4%).

As shown in Figure 2c, Class 3 (n = 51) had the highest EDD use while partying along with use of GHB and poppers in the past three months (84.5%, 96.2%, and 65.1% respectively) in combination with high meth use (100%). This combination of substances suggests sexual enhancement motives consistent with PnP settings. Class 3 was predominately Non-Hispanic White (64.7%) and had the lowest percentage of individuals who had experienced lifetime incarceration or homelessness in the past year (25.5% and 25.5%, respectively). Class 3 also had the second highest mean score for sexual compulsivity and second highest mean score for PTSD (see Table 2).

Table 2.

Demographic, Structural, and Psychological Correlates of Substance Use Class Membership (N = 161)

	Meth Use Only (n = 69)	Meth and Crack Use (n = 35)	PnP (n = 51)	Polysubstance Use (n = 6)	p-value
	n (%)	n (%)	n (%)	n (%)
Race/Ethnicity ┼					P < 0.0001; p-value < 0.01**
Black	12 (17.4)	10 (28.6)	1 (2.0)	2 (33.3)
Non-Hispanic White	29 (42.0)	10 (28.6)	33 (64.7)	3 (50.0)
Hispanic/Latino	19 (27.5)	10 (28.6)	13 (25.5)	1 (16.7)
Other	9 (13.1)	5 (14.2)	4 (7.8)	0 (0)
Homeless (Past Year)					P = 0.0018; p-value =0.35
Yes	25 (36.2)	11 (31.4)	12 (25.5)	3 (50)
No	44 (63.8)	24 (68.6)	39 (76.5)	3 (50)
Incarceration (Lifetime)					P = 0.0002; p-value =0.03
Yes	28 (40.6)	18 (51.4)	12 (25.5)	3 (50)
No	41 (59.4)	17 (48.6)	39 (76.5)	3 (50)
Detectable Viral Load					P = 0.0053; p-value =0.64
Yes	15 (21.7)	11 (31.5)	12 (23.6)	2 (33.3)
No	39 (56.6)	16 (45.7)	28 (54.9)	3 (50.0)
Missing	15 (21.7)	8 (22.8)	11 (21.5)	1 (16.7)
Depression					P = 0.0056; p-value =0.71
Yes	48 (69.6)	26 (74.3)	40 (78.4)	5 (83.3)
No	21 (30.4)	9 (25.7)	11 (21.6)	1 (16.7)
	M (SD)	M (SD)	M (SD)	M (SD)
Negative Affect	12.6 (5.5)	13.3 (5.7)	13.63 (6.0)	15.3 (8.3)	F = 0.7; p-value 0.55
Positive Affect	26.0 (6.5)	26.1 (7.5)	25.12 (6.7)	23.6 (6.7)	F= 0.81; p-value 0.49
Sexual Compulsivity	17.2 (7.5)	17.9 (7.1)	19.46 (7.5)	23.1 (8.6)	F = 3.35; p-value 0.02
PTSD	45.5 (14.0)	42.4 (12.3)	49.5 (15.3)	55.3 (13.7)	F = 2.67; p-value 0.04

^┼p-value reflects non-Hispanic white versus persons of color

As shown in Figure 2d, Class 4 (n = 6) exhibited high use of all substances in the past three months (prescription opioids, benzodiazepines, powder cocaine, crack-cocaine, meth, methadone, heroin, benzodiazepine, barbiturates, poppers, ketamine, MDMA, GHB, and EDD). Although Class 4 was the smallest class, a large percentage of participants experienced homelessness in last year and incarceration in their lifetime (50% and 50% respectively). Half of participants (50%) had a detectable viral load. As shown in Table 2, Class 4 also had the highest mean PTSD, negative affect, and sexual compulsivity scores (55.3, 15.3, and 23.1 respectively).

3.2. Correlates of class membership.

As shown in Table 2, bivariate were conducted to examine correlates of class membership. There were significant overall differences in race/ethnicity by class membership such that persons of color were over-represented in Class 2, meth and crack-cocaine use. There were also significant overall differences in lifetime incarceration by class membership as was seen in class 2. Sexual compulsivity and PTSD scores differed significantly as a function of class membership such that higher mean scores were reported by participants in Class 3, PnP. Tests of categorical latent variable multinomial logistic regression using the 3-step procedure were performed to examine differences as a function of class membership. Those in Class 3 (PnP) were less likely to experience incarcerations, meaning those who had never been to jail were 2.2 times more likely to be in Class 3 than the reference group (Class 1) (p-value = 0.05). When examining race, those that identified as Persons of Color were 2.53 times more likely to be in Class 2 (Meth and Crack) than in Class 1 (meth only reference group) (p-value = 0.02).

4. Discussion

Findings of this LCA demonstrate that even among SMM living with HIV who use meth there is substantial heterogeneity in substance use patterns. Consistent with previous studies (6, 12, 51), polysubstance use was observed among nearly two-thirds of participants (65%) such that individuals were classified as engaging in meth and crack-cocaine use (Class 2), PnP use that likely reflects a desire for sexual enhancement (Class 3), and use of multiple substances (Class 4). We also noted prominent differences in class membership as a function of race/ethnicity such that meth and crack-cocaine use was the most diverse class where participants of color were overrepresented and more likely to have a history of incarceration relative to other classes.

Those in Class 1 (meth only) and Class 2 (meth and crack) consisted of 58% and 71% racial/ethnic minorities respectively, compared to Class 3 (PnP) which was predominantly Non-Hispanic White (65%). Notably, Classes 1 and 2 also had the highest number of participants that had been homeless in the last year and/or with a history of incarceration, especially compared to the mostly White PnP Class. It is important to note the potential role of racial disparities in incarceration. Although Non-Hispanic Whites in the US are three times more likely to have used crack-cocaine than Blacks, the proportion of Black men in prison for drug charges is 13 times greater than White men (52). Our findings build upon other studies documenting the intersectional relationship between homelessness, incarceration, substance use, and HIV treatment outcomes (53–55). Finally, recent evidence suggests that meth use is increasing among SMM, particularly Black and Latinx SMM (56). Overall, members in Classes 1 (meth only) and 2 (meth and crack) might benefit from substance use disorder treatment programs that address discrimination, housing insecurity, and lack of access to services due to the intersectionality of substance use, race/ethnicity, and incarceration history (31).

Class 3 (PnP) had the highest EDD use while partying along with use of GHB and poppers in the past three months (85%, 96%, and 65% respectively) in combination with high meth use (100%). This combination of substances suggests sexual enhancement motives, which consistent with higher mean scores for sexual compulsivity. There may also be negative reinforcement motives for PnP use as evidence by higher scores PTSD symptoms. Other studies have noted associations among sexual compulsivity, traumatic stress, and stimulant use, whereby traumatic stress and stimulant use were associated with increased frequency of sexual urges and thoughts (57). Our findings suggest those engaging in PnP use could benefit from trauma-informed interventions that address perceived difficulties with controlling sexual behaviors as a potential driver of substance use (58).

Our findings should be understood in light of their limitations. This is a modest sample size for an LCA, leading to small group membership and diminished statistical power. Multivariate logistic regression also overestimates odds ratios in studies with small to moderate samples size and given the small sample size there is induced bias away from null, which can be seen in the overinflated odds ratio and large confidence intervals. Future research with larger sample sizes should be performed to elucidate the relationship between distinct substance use classes and condomless anal sex. The present study also included aggregate substance use measures that asked participants to recall the frequency with which they used each substance over the past three months. Further research employing timeline follow-back methods is needed to examine dynamic patterns of co-use for various substances. Finally, the information gained from this sample of SMM living with HIV who use meth was from San Francisco, California, and may only be applicable to similar urban areas with large communities of SMM. Further research is needed with more representative samples to examine generalizability of our findings.

Despite these limitations, the present study further elucidates the substantial heterogeneity in substance use patterns among SMM living with HIV who use meth. Treatments that are often focused myopically on meth is not realistic for many SMM with different contexts and patterns of substance use requiring tailored approaches (59). Findings clearly identify the need for tailored intervention approaches addressing the unique needs of ethnic minority men, particularly those facing profound structural barriers like a history of incarceration. This study also underscores the potential benefits of trauma-informed interventions targeting the co-occurrence of difficulties controlling sexual behaviors and PnP use.