Cannabis and Alcohol Co-Use and Condomless Anal Sex Among Men Who Have Sex with Men Living with HIV: An Event-Level Analysis

Abstract

Men who have sex with men (MSM) account for more than two thirds of new HIV diagnoses annually. Sexual behavior that increases risk for onward transmission of HIV is associated with both alcohol and cannabis use. However, little is known about the influence of cannabis and alcohol co-use on engagement in condomless anal sex (CAS). The current study explored daily associations between substance use and CAS among HIV-positive MSM using a 42-day Timeline Followback interview (N=101). Generalized estimating equation (GEE) logistic regression models were used to examine the association between cannabis and alcohol co-use and CAS at the sexual event while controlling for study site, condition, adherence to antiretroviral therapy, sex-related alcohol expectancies, and partner type. Participants provided data for 1,052 sexual activity days, 60.7% of which involved CAS. Of 638 CAS days, 9.1% involved no substances, 72.0% involved either cannabis or alcohol use, and 18.9% involved cannabis and alcohol co-use. Results indicated that the odds of engaging in CAS were higher for sexual events in which cannabis and alcohol co-use occurred (aOR = 2.98; 95% CI: 1.27, 6.97) compared to events in which no substance use occurred (p = .012), but this relationship was no longer significant when cannabis and alcohol co-use was compared to single substance use (aOR = 1.57; 95% CI: 0.85, 2.90; p = .15). Future research should identify specific substance use (e.g., quantity) and partner characteristics (e.g., level of intoxication) that may uniquely influence the relationship between cannabis and alcohol co-use and condomless sex.

Men who have sex with men (MSM) are disproportionately affected by HIV in the United States (U.S.), accounting for more than two thirds (70%) of new HIV diagnoses annually (1). If current infection rates persist, one in six MSM may be diagnosed with HIV in their lifetime (1). The prevalence of HIV in this population underscores the continued need to better understand the risk factors associated with onward transmission. The association between alcohol and sexual behaviors that increase the risk for HIV transmission and acquisition, specifically condomless anal sex (CAS), has been a focal point of HIV research over the past two decades (2–4). There is compelling support in the literature that alcohol consumption is a contributing factor to risk of HIV infection among MSM (5–9), however less is known about the influence of the second most commonly used substance among MSM – cannabis (10).

In a nationally representative sample of MSM, approximately 40% reported current cannabis use compared to 7% of the general population (11). Moreover, among young MSM (ages 18–25), cannabis use is more frequent than alcohol consumption, with 23% endorsing daily cannabis use compared to 2% reporting daily alcohol use (12). Emerging research on cannabis use in conjunction with sexual activity suggests that MSM utilize cannabis prior to sex as frequently as alcohol (63.5% compared to 61.5%, respectively; (13). Although the prevalence of cannabis use among MSM is high, and there is evidence that cannabis may increase engagement in condomless sex in heterosexual samples (14–19), there is limited research on the association between cannabis use and condomless sex among MSM. Theories that consider the pharmacological and expectancy effects of cannabis can be used to generate hypotheses about the potential role of cannabis in condomless sex.

Theoretical Framework of Cannabis Use and Sexual Behavior

The association between cannabis use and condomless sex is theorized to be a result of pharmacological and expectancy effects, both of which are influenced by the co-occurrence of alcohol use. The delta-9-tetrahydrocannabinol (THC) found in cannabis may produce similar pharmacological effects on sexual-risk decision-making as alcohol (20–23). For example, THC has been found to impair information processing (23) and the capacity to inhibit already initiated responses (24,25), increase relaxation, euphoric mood (21), impulsivity (24), risk-taking (26), and subjective intoxication (25), and produce self-reported aphrodisiac effects (27). The pharmacological influence of THC may thus reduce behavioral control and subsequently increase the risk that condomless sex may occur.

Coupled with the pharmacological effect of acute intoxication, individual expectations or beliefs that cannabis will have a disinhibiting impact on one’s sexual behavior may also influence engagement in condomless sex (24,25). An event in which substances are consumed involves several substance expectancies including expectations about the administration of a substance (i.e., stimulus expectancy) and expectancies about the effect or consequence associated with the substance (i.e., outcome expectancies; (28). For example, one cannabis-related sex expectancy may be the belief that cannabis produces strong disinhibiting effects for desired but socially prohibited acts (e.g., having condomless sex with a casual partner; (24). Taken together, acute intoxication from cannabis combined with cannabis-related sex expectancies may lead to greater engagement in condomless sex (24,25,29,30).

An additional explanation for the association between cannabis use and condomless sex may relate to the use of cannabis in conjunction with alcohol during a single substance use episode (i.e., co-use). Cannabis is the most commonly used substance among those who drink, with over 75% of cannabis users endorsing current alcohol use (31). Indeed, almost three quarters (70.4%) of MSM who use substances fit within an “alcohol and marijuana users” substance-use risk profile (high alcohol and marijuana use but low endorsement of other substances), while only 24.7% and 4.9% are categorized as “low marijuana users” (low to moderate marijuana use and negligible endorsement of all other substances) and “polysubstance users” (moderate to high endorsement of all substances), respectively (32). According to the complement hypothesis, individuals tend to use substances such as cannabis and alcohol together within a fixed, limited time period as part of a common social experience and/or to produce desired interactive effects (e.g., euphoria; (33,34). Further, the pharmacological effects of cannabis and alcohol co-use may be more pronounced among HIV-positive MSM, as it is hypothesized that cannabis, alcohol, and HIV infection synergistically affect cognitive functioning (35,36).

The risk associated with cannabis use in combination with alcohol is greater than that from either drug alone, including increased social and behavioral consequences such as driving while impaired, higher and more frequent consumption levels (37,38), and greater psychological distress (39,40). Furthermore, research with heterosexual college student samples has found that cannabis and alcohol co-use on a given day is associated with higher odds of engaging in condomless sex compared to use of either substance alone (17,41). The deficits in cognition, perception, and social behaviors associated with cannabis and alcohol co-use suggest that the combined use of both substances may influence engagement in condomless sex above the independent effect of either substance alone and may, at least in part, explain the number of contradictory findings across global, situational, and event-level research.

In order to comment on the current status of the cannabis use and sexual behavior literature, it is necessary to characterize the studies in terms of their research design and the information each design can yield. Global association studies focus on aggregated levels of substance use and sexual behaviors of interest over broad recall periods (e.g., cannabis use over the past week and frequency of condomless sex over the past 3 months), which can provide a signal that two variables are related, but cannot supply any information about temporal overlap (42). Situational association studies examine whether a behavior (e.g., cannabis use) has tended to occur, or has ever occurred, in conjunction with another behavior of interest (e.g., “how often do you use cannabis during sex”; (4), but do not yield information about whether substance use is associated with sexual activity at specific events. Alternatively, event-level studies focus on whether a substance was consumed immediately before or during a specific sexual event, ideal for assessing temporal overlap (4). Prospective event-level studies assess the occurrence of specific behaviors on specific occasions at multiple timepoints over a period of weeks or months, whereas retrospective event-level assessments occur at a single timepoint after the event of interest has already occurred (43,44).

Summary of Literature on Cannabis Use and Sexual Behavior Among MSM

A number of studies using global measures have generally linked cannabis use and sexual outcomes in MSM (45–47), with some exceptions (10,12). Similarly, studies that have examined the situation-specific use of cannabis and CAS among MSM (47–50) have mostly found that reporting cannabis use in the context of sex was related to increased participation in CAS. However, the event-level literature on cannabis use and CAS among MSM is less conclusive (5,51–53). While some studies that measure cannabis use during specific sexual events have demonstrated that cannabis use, compared to no substance use, was associated with an increase in the odds of CAS (aOR = 1.55; 95% CI: 1.13, 2.15; aOR = 4.01; 95% CI: 1.35, 11.90; 5,52, respectively), other studies have not demonstrated similar significant findings (aOR = 1.0; 95% CI: 0.80, 1.10; aOR = 0.91; 95% CI: 0.42, 1.95; (51,53, respectively).

One explanation for the discrepant findings across event-level studies may relate to the moderating role of alcohol use in the association between cannabis and CAS. For example, Boone and colleagues (2013) found that almost half (49%) of substance use episodes in their sample of MSM involved more than one substance, and the use of multiple substances during the most recent sexual encounter corresponded to a 62% increase in the odds of having CAS (aOR = 1.62; 95% CI: not reported). Yet, the “alcohol and marijuana users” profile was not explicitly explored. Similarly, in a cross-sectional study of MSM living with HIV in the United Kingdom (UK), it was revealed that polydrug use was associated with a higher prevalence of condomless sex (prevalence of CAS with no drug use was 24% versus prevalence of CAS with use of five or more drugs was 78%), but again, it was unclear which substance use combinations may have contributed to the overall sexual risk (54). Although Kahler et al. (2015) incorporated a daily alcohol use variable in their final model, the influence of cannabis and alcohol co-use on CAS was not examined, limiting our understanding of how these two substances may uniquely influence engagement in CAS when used in conjunction. Thus, although a majority of MSM who use substances can be classified as “alcohol and marijuana users” (32), the evidence linking CAS with cannabis and alcohol co-use among MSM is inconclusive. Moreover, HIV-positive MSM may be uniquely impacted by the synergistic effects of cannabis and alcohol co-use on CAS (35,36), yet no study has sought to examine how the combined use of both substances may influence engagement in CAS among this population.

Purpose of the Present Study

The present study attempts to address this gap in the literature with a secondary data analysis of retrospective event-level data from a sample of 101 MSM living with HIV. The overall objective of this analysis was to examine the event-level association between cannabis and alcohol co-use and CAS with a partner of any HIV status. Serodiscordant condomless sex (i.e., when the participant reported that his sexual partner was HIV-negative or of unknown HIV status) was a secondary outcome of interest. Consistent with research on the additive deficits in cognition, perception, and social behaviors that result from cannabis and alcohol co-use (38–40), we predicted that cannabis and alcohol co-use during a sexual event would be associated with significantly higher odds of engaging in CAS with a partner of any HIV status compared to single substance use (alcohol-only or cannabis-only) and no substance use. We hypothesized a similar pattern of findings for CAS with a partner of negative or unknown HIV status as an outcome.

Methods

Study Procedure

This study is a secondary analysis of data collected from a larger daily diary study designed to examine event-level substance use and sexual behavior among HIV-positive MSM (N = 101). Participants were recruited from Syracuse, NY and San Francisco, CA using social media (e.g., Jack’d, Facebook), visits to infectious disease clinics, study participant referrals, and self-referrals (e.g., flyers). At the baseline study appointment, participants completed several interviewer-administered self-report questionnaires, including a 6-week Timeline Followback (TLFB) interview of recent substance use and sexual activity, and were randomized to six weeks of daily telephone-based surveys completed via phone call using the touchtone keypad to record responses or a no-daily assessment control condition. Inclusion criteria included: (1) 18–65 years of age, (2) self-identified as male or a transgender man, (3) sexually active (i.e., reported having >1 instances of receptive or insertive anal sex with a man in the last 6 weeks), (4) inconsistent condom user (i.e., using condoms more than 0% of the time but less than 100% of the time during sexual intercourse in the previous 6 weeks), and (5) consumption of more than one drink containing alcohol in a typical month. Although participants were not required to be connected with outpatient medical care to be eligible to participate in this study, all participants did report being connected with some form of medical care. Participants returned for a 6-week follow-up appointment in which they were re-administered the same set of self-report questionnaires, including the 6-week TLFB interview. All procedures were approved by the Institutional Review Boards at Syracuse University and University of California, San Francisco.

Measures

Sociodemographic profile.

Demographic variables included in this study were age, race, time since HIV diagnosis, personal income, and highest level of education received. Self-reported adherence to antiretroviral therapy (ART) was also measured by asking participants to report the number of days they missed at least one dose of their HIV medications in the last 30 days, which was then converted to an adherence percentage.

Timeline Followback.

The TLFB is a calendar-assisted structured interview that assesses discrete episodes of alcohol use (44) cannabis use (55), and sexual behavior (56). For this study, the recall period was modified to 6-weeks to match the length of the follow-up period and administered at two time points, resulting in 84 total days of TLFB data for each participant. For alcohol use, the TLFB assessed the number of standard drinks of alcohol consumed on each day (defined as 12 oz. of beer, 5 oz. of wine, or 1.5 oz. 80-proof distilled spirits). Cannabis was coded dichotomously as use or no use for each day. The substance use variable was coded for each day as no substance use, single substance use (alcohol-only or cannabis-only), or cannabis and alcohol co-use. For each sexual episode, participants were asked about condom use versus nonuse for oral, vaginal, and anal sex events, as well as the HIV status of each partner. Sexual partner type was coded as “regular” if they identified the partner as someone they had known for a while and had some commitment to, or as “casual” if they identified the partner as someone they have not known for very long and had little commitment to.

Sex-Related Alcohol Expectancies Questionnaire.

Given that individual expectations or beliefs that alcohol will have a disinhibiting impact on one’s sexual behavior may influence intent to drink, drinking, and post-drinking behavior (e.g., sexual activity), the Sex-Related Alcohol Expectancies Questionnaire (SRAEQ) was utilized as a covariate in the analyses for this study (57). The SRAEQ is a 13-item assessment that measures three domains of sex-related expectancies linked to drinking alcohol: sexual disinhibition, sexual enhancement, and sexual risk-taking. Responses range from 1 (strongly disagree) to 6 (strongly agree), with higher mean scores indicating stronger endorsement of expectancies. The SRAEQ demonstrated adequate reliability in this sample (α = 0.89).

Data Analysis Plan

Data were analyzed using the Statistical Package for Social Sciences (SPSS) versions 23 (SPSS, 2012). The criterion for statistical significance was set to an alpha level of 0.05. Descriptive statistics were used to summarize age, race, time since HIV diagnosis, personal income, highest level of education received, and sex-related alcohol expectancies. For continuous variables, means, medians, standard deviations, percentiles, and ranges were generated; frequencies and proportions were used for categorical and ordinal variables.

Generalized estimating equation (GEE; (58) logistic regression models were used to examine participant substance use and our primary (CAS with a partner of any HIV status) and secondary (CAS with a partner of negative or unknown HIV status) outcomes of interest. The GEE approach was used to account for the correlation from using repeated observations from the same subject over time. Models were fit using an independent working correlation matrix shown by Liang and Zeger (1986) to maintain a relative high efficiency across a range of true correlation structures. To test our hypothesis that cannabis and alcohol co-use during a sexual event would be associated with significantly higher odds of engaging in CAS, compared to single substance use and no substance use, the substance use variable was dummy coded into three categories: no substance use, single substance use, and cannabis and alcohol co-use. The following variables were adjusted for in the multivariate analyses: (1) recruitment site, (2) study condition, (3) adherence to antiretroviral therapy (ART), (4) sex-related alcohol expectancies (i.e., total score at baseline and follow-up), and (5) partner type (casual vs. regular).

Results

Descriptive results

Table 1 displays the demographic characteristics of the sample. Participants were 101 HIV positive MSM. The average age of participants was 41.51 years (SD = 10.44) and the average number of years since HIV diagnosis was 15.64 (SD = 8.97). Many participants had completed a college degree (45.6%), and 50.5% were Black, 29.1% were White, and 20.4% were mixed racial and/or ethnic origin or identified as another race/ethnicity. The average score for the SRAEQ was 34.5 (SD = 8.32; possible range of 13–52), with higher scores indicating stronger endorsement of expectancies.

Table 1.

Characteristics of HIV-positive men who have sex with men (N = 101)

	Total (N = 101)	San Francisco, CA (N = 64)	Syracuse, NY (N = 37)
	M (SD)	M (SD)	M (SD)	t	p
Age	41.51 (10.44)	44.90 (9.13)	35.97 (10.18)	−4.54	<.001**
SRAEQ	34.50 (8.32)	35.82 (8.26)	33.50 (8.98)	−1.80	.073
Years Since HIV Diagnosis	15.64 (8.97)	19.30 (8.63)	9.32 (5.34)	−6.35	<.001**
	n (%)	n (%)	n (%)	χ2	p
Race				10.42	.108
Black	52 (50.5)	39 (60.9)	13 (33.3)
White	30 (29.1)	14 (21.9)	16 (41.0)
Mixed ethnic origin/other	19 (20.4)	11 (17.2)	8 (21.6)
Hispanic ethnicity	13 (12.6)	9 (14.1)	4 (10.3)
ART Adherence				15.33	.168
< 85% adherence	11 (10.9)	7 (10.9)	4 (10.8)
> 85% adherence	90 (89.1)	57 (89.1)	33 (89.1)
Annual Income				25.58	.001**
< $50,000	89 (88.1)	62 (96.9)	27 (72.9
< $50,000	12 (11.9)	2 (3.1)	10 (27.1)
Education				7.51	.185
Did not complete high school	11 (10.9)	7 (10.9)	4 (10.3)
High school graduate or GED	35 (34.7)	28 (43.8)	7 (17.9)
College graduate (AA/BA/BS)	46 (45.6)	25 (39.1)	21 (56.8)
Professional degree (MA/MD/JD/PhD)	9 (8.9)	4 (6.3)	5 (13.5)

Note. M = Mean, SD = Standard Deviation, SRAEQ = Sex-Related Alcohol Expectancies Questionnaire, ART = Antiretroviral Therapy, Adherence was averaged across baseline and follow-up, Percentages may not add up to 100% due to missing data (i.e., participants declining to respond to certain measures).

^**p ≤.01,

^*p ≤.05,

^αp <.10

Substance use and sexual behavior variables are displayed in Table 2. Participants provided data for a total of 5,816 person-days. Across the sample, participants reported a total of 1,052 sexual activity days (18% of days). Of sexual activity days, almost two thirds (60.7%) involved CAS and 30.9% involved condomless sex with a partner of negative or unknown HIV status. Sixty-one percent of sexual activity days involved sex with a regular partner and 38.4% of sexual activity days involved sex with a casual partner. Substance use was reported before or during sexual activity on 87.5% (N = 921) of days collected. A total of 64 participants (63.4%) endorsed cannabis use and, as part of the eligibility criteria for this study, all participants endorsed alcohol use across the two 42-day recall periods. Cannabis use occurred on almost half (47.4%) of sexual activity days, alcohol use occurred on three quarters (76.5%) of sexual activity days, and cannabis and alcohol co-use occurred on 36.4% of sexual activity days. Over the 84 days, CAS occurred on 9.1% of days in which no substances were used, 25.5% of days in which cannabis was used, 44.9% of days in which alcohol was consumed, and 18.9% of days in which cannabis and alcohol were both used. Of the 325 days in which condomless sex with a partner of negative or unknown HIV status occurred, 20.0% involved no substance use, 40.0% involved cannabis use, 72.9% involved alcohol use, and 32.9% involved cannabis and alcohol co-use.

Table 2.

Characteristics of sexual activity days (N = 1,052) among HIV-positive men who have sex with men

	Overall	Condomless anal sex	Protected sex
	N = 1,052 events	n = 639 events	n = 413 events
	N (%)	N (%)	N (%)
Participant used alcohol only
Yes	805 (76.5)	473 (74.0)	332 (80.4)
No	247 (23.5)	166 (26.0)	81 (19.6)
Participant used cannabis only
Yes	499 (47.4)	269 (42.1)	230 (55.7)
No	553 (52.6)	370 (57.9)	183 (44.3)
Participant used alcohol and cannabis
Yes	383 (36.4)	199 (31.1)	184 (44.6)
No	669 (63.6)	440 (68.9)	229 (55.4)
Partner type
Regular	648 (61.6)	431 (67.4)	217 (52.5)
Casual	404 (38.4)	208 (32.6)	196 (47.5)
HIV status of partner
HIV-positive	423 (40.3)	315 (49.2)	109 (26.4)
HIV-negative or unknown	629 (59.7)	325 (50.8)	304 (73.6)

Note. Percentages may not add up to 100% due to missing data (i.e., participants declining to respond to certain measures).

Primary analyses

Condomless Anal Sex with a Partner of Any HIV Status.

Unadjusted and adjusted analyses for CAS with a partner of any HIV status can be found in Table 3. In the unadjusted analyses, the odds of engaging in CAS were significantly higher for sexual events in which cannabis and alcohol co-use occurred (OR = 2.51; 95% CI: 1.18, 5.37; p = .02) compared to events in which neither cannabis nor alcohol were consumed. However, the unadjusted odds of engaging in CAS were not significantly higher for sexual events in which the participant reported cannabis and alcohol co-use (OR = 0.64; 95% CI: 0.33, 1.22; p = .14) compared to events in which either substance was used independently. Results of the multivariate analysis revealed a similar pattern such that the odds of engaging in CAS were significantly higher for sexual events in which the participant reported cannabis and alcohol co-use (aOR = 2.98; 95% CI: 1.27, 6.97; global p = .01) compared to events in which neither substance was used. However, the odds of engaging in CAS were not significantly higher for events in which both cannabis and alcohol were consumed (aOR = 0.52; 95% CI: 0.26, 1.04; p = .15) compared to events in which either substance was used independently. For the covariates in the multivariate model, the odds of CAS were significantly higher for sexual events that involved a regular, compared to a casual, partner (OR = 2.06; 95% CI: 1.14, 3.72; p = .02). Recruitment site, study condition, adherence to ART, and sex-related alcohol expectancies were not significantly related to CAS with a partner of any HIV status.

Table 3.

Unadjusted and adjusted odds ratios (ORs) and 95% confidence intervals (CI) for seroconcordant and serodiscordant condomless anal sex at baseline and follow-up Timeline Followback (TLFB) interview (CAS) among HIV-positive men who have sex with men

	CAS with a partner of any HIV status				CAS with a serodiscordant partner
	Unadjusted (n = 638 events)		Adjusted (n = 638 events)		Unadjusted (n = 233 events)		Adjusted (n = 233 events)
	OR (95% CI)	p value	aOR (95% CI)	p value	OR (95% CI)	p value	aOR (95% CI)	p value
		.017		.012		.011		.066
No substance use (referent)	1.00		1.00		1.00		1.00
Single substance use	1.57 (0.82, 3.00)		1.90 (0.96, 3.80)		2.47 (1.39, 4.38)		2.15 (1.15, 4.00)
Cannabis and alcohol co-use	2.51 (1.18, 5.37)		2.98 (1.27, 6.97)		2.61 (1.39, 4.38)		2.18 (0.95, 5.02)
Single substance use (referent)	1.00	.136	1.00	.154	1.00	.860	1.00	.961
No substance use	1.60 (0.86, 3.00)		1.57 (0.85, 2.90)		0.40 (0.23, 0.72)		1.02 (0.54, 1.92)
Cannabis and alcohol co-use	0.64 (0.33, 1.22)		0.52 (0.26, 1.04)		1.06 (0.57, 1.97)		0.47 (0.25, 0.87)
Recruitment site				.283				.222
San Francisco (referent)	-		1.00		-		1.00
Syracuse	-		1.65 (0.66, 4.10)		-		0.58 (0.24, 1.39)
Study condition				.619				.948
Control (referent)	-		1.00		-		1.00
IVR	-		0.82 (0.38, 1.78)		-		1.03 (0.48, 2.17)
Partner type				.016
Regular (referent)	-		1.00		-		1.00	.617
Casual	-		2.06 (1.14, 3.72)		-		0.86 (0.48, 1.54)
Adherence to ART	-		0.99 (0.98, 1.01)	.284	-		0.99 (0.98, 1.01)	.527
Sex-related alcohol expectancies	-		0.99 (0.93, 1.04)	.630	-		0.99 (0.97, 1.05)	.472

Note. OR = Odds Ratio, aOR = Adjusted Odds Ratio, CI = Confidence Interval; Generalized estimating equation (GEE) models with independence working correlations

Condomless Anal Sex with a Partner of Negative or Unknown HIV Status.

In the unadjusted models, cannabis and alcohol co-use, compared to no substance use, was significantly associated with a higher odds of engaging in CAS with a partner of negative or unknown HIV status (OR = 2.61, 95% CI: 1.24, 5.50; p = .01; Table 3). The odds of engaging in CAS with a serodiscordant partner were not significantly higher during sexual events in which participants reported cannabis and alcohol co-use (OR = 1.06; 95% CI: 0.57, 1.97) compared to single substance use (p = .86). In the multivariate analyses, a nonsignificant trend suggested that the odds of engaging in CAS with a serodiscordant partner were higher for sexual events in which both cannabis and alcohol were consumed (aOR = 2.18, 95% CI: 0.95, 5.02; p = .07) compared to no substance use. Although not a statistically significant difference, cannabis and alcohol co-use was associated with lower odds of engaging in CAS with a serodiscordant partner (aOR = 0.47; 95% CI: 0.25, 0.87; p = 0.96) compared to single substance use. Recruitment site, study condition, adherence to ART, sex-related alcohol expectancies, and partner type were not significantly related to CAS with a serodiscordant partner.

Discussion

This study is among the first to examine event-level associationns between cannabis and alcohol co-use and CAS in a sample of MSM living with HIV. Data from two 6-week TLFB interviews were used to examine daily associations between cannabis and alcohol co-use and condomless sex. Results confirmed our hypothesis that CAS with a partner of any HIV status was almost three times more likely to occur when cannabis and alcohol were both consumed on a given sexual activity day compared to when no substances were consumed. Similar effect sizes were observed for CAS with a partner of unknown or negative HIV status, in which CAS was 2.18 times more likely to occur when both cannabis and alcohol, as compared to no substances, were consumed. Consistent with previous research (e.g., (9), these findings suggest that the occurrence of substance use on a sexual activity day incurs greater risk for the transmission and acquisition of HIV relative to days in which no substance use occurs. Notably, the effect sizes garnered in this study were larger than those obtained in samples of uninfected individuals, indicating that cannabis and alcohol co-use may have unique synergistic effects on impairment among individuals living with HIV (17,35,36). Conversely, condomless sex was less likely to occur when both cannabis and alcohol were consumed during a sexual event compared to consumption of either substance alone regardless of partner HIV status.

One potential explanation for this pattern of findings is that combining cannabis and alcohol into a single co-use variable may have masked the main effect of the substance most responsible for driving the increase in likelihood of condomless sex occurring. Indeed, alcohol use occurred on 44.9% of condomless sexual activity days whereas cannabis use occurred on only 25.5% of CAS days, suggesting that alcohol may have been the driving force behind the main effect of co-use on CAS. It is also possible that the TLFB data did not accurately reflect the occurrence of intoxication from use of both substances during a sexual event compared to the intoxicating effect of a single substance. For example, a participant may have used cannabis in the morning of a given sexual activity day but consumed alcohol in the evening prior to the sex event. As such, simultaneous intoxication from both cannabis and alcohol may not have occurred. Future research should refine data collection techniques to more thoroughly assess the presence of intoxication from substance co-use during a given sex event. Additionally, differences in situational factors surrounding substance co-use versus single use (e.g., location, time of day), as well as how such factors may interact with individual characteristics (e.g., HIV knowledge), was not explicitly assessed in this study and may partially account for these results (41). Previous research has hypothesized that a constellation of dispositional and personality traits, such as impulsivity, and attitudes and beliefs may function as “third variables” to influence engagement in condomless sex rather than a unique relationship between substance use and sexual behavior (4). Research that tests for an event-level relationship between cannabis and alcohol co-use and condomless sex while also examining the moderating effect of relevant individual difference characteristics and situational factors would provide a better understanding of the nature of this relationship.

Although half of the condomless sexual events reported by participants in the sample were with known seroconcordant partners, in which the risks associated with CAS are primarily related to acquiring sexually transmitted infections or different strains of HIV (59,60), the results revealed a similar association between cannabis and alcohol co-use and CAS with partners of negative or unknown HIV status. Specifically, the odds of engaging in CAS with a serodiscordant partner were significantly higher for sexual events in which both cannabis and alcohol were consumed compared to no substance use, but lower when compared to single substance use events. Importantly, the decision to use condoms between serodiscordant partners may have been influenced by the presence of other biomedical HIV prevention strategies, such as pre-exposure prophylaxis (PrEP) and “treatment-as-prevention” (TasP), which can substantially decrease the likelihood of sexually-based HIV acquisition and transmission, even when condoms are not used (61,62). Indeed, most participants in this study were adherent to ART and thus likely had undetectable viral loads, which poses minimal risk for HIV transmission (61). Additionally, although rates of PrEP uptake remain low among MSM (63), uninfected partners who were adherent to PrEP may have been at low risk for HIV acquisition (62). Taken together, occurrences of CAS among participants or partners who were “biologically protected” may have conferred little to no risk of onward HIV transmission, and subsequently impacted the decision to forego condom use. However, since this study was a secondary data analysis of a parent study that was not originally designed to assess for these participant and partner characteristics, it is not possible to discern from these data the factors that may have influenced condom use decision-making or the amount of risk associated with condomless sex.

Consistent with previous studies (64,65), sexual encounters with a regular partner were more likely to involve condomless sex compared to sexual encounters with a casual partner. Rates of condomless sex among MSM appear to increase incrementally as familiarity with a partner and seriousness of a relationship increases (65). Consequently, 68% of HIV transmissions have been shown to result from sex with a regular sexual partner (66). Emotional factors, such as greater feelings of closeness and intimacy, have been suggested to underlie the effects of relationship type on condomless sex among MSM (67). Although endorsing sex with a regular partner, as compared to a casual partner, is associated with greater HIV risk, this must be interpreted in the context of the other emotional and health benefits that come with engaging in close, positive romantic relations and potential utilization of other HIV prevention strategies. Prior research has shown that being in an intimate relationship may buffer against stressful life experiences (68), which may be particularly important for PLWH as both acute and chronic stress have been shown to increase inflammation, exacerbate tissue damage, and increase the risk for developing non AIDS-related co-morbidities (69). Additionally, individuals in close relationships may choose to replace consistent condom use with other HIV prevention strategies (e.g., PrEP), which can reduce the risk of contracting HIV by more than 90% with adequate adherence (70). Decisions about the safety of condomless sex in regular sexual partnerships may also be influenced by TasP approaches, including consistently adhering to ART regimens and maintaining an undetectable viral load (71,72). While this study provides further evidence that regular sexual partnerships may result in greater risk of condomless sex, future research is needed to examine how characteristics of substance co-use (e.g., quantity) and attributes of partner type (e.g., serostatus, number of previous sexual encounters, level of intoxication) interact to influence engagement in condomless sex.

Limitations

This study has several limitations. Although the TLFB is a reliable and valid measure of daily substance use and sexual behavior (55,73,74), retrospective reporting carries the risk of recall bias and memory failure (75). To supplement potential recall biases resulting from retrospective self-report measures, biomarkers of alcohol use (e.g., Phosphatidylethanol; (76), cannabis use (e.g., Tetrahydrocannabinol; (77), and condomless sex (e.g., sexually transmitted infections; (78) could be utilized to better quantify substance use and sexual behavior in this population. Indeed, research suggests that self-report measures of substance use and sexual behavior among MSM generally underestimate levels of risk behavior compared with biomarkers, highlighting the need to correlate self-report measures with objective measures (e.g., biomarkers) in future research (79). Recall bias would also be ameliorated with the use of prospective intensive longitudinal methods, such as daily diary studies (80). Daily diary studies, in which participants complete end-of-day reports on their behaviors and experiences that day for a period of days or weeks, allow for reliable and accurate information on precise temporal ordering of events with relatively little recall error (80). Furthermore, this study incorporated some individual difference characteristics, such as sex-related alcohol expectancies, as covariates in each model; however, additional individual difference characteristics (e.g., HIV knowledge), as well as additional situational factors (e.g., location of sex event) should be incorporated in future research to better elucidate how these variables may influence the relationship between substance use and sexual behavior.

It is important to note that this study focused on a sample of MSM living with HIV engaged in outpatient medical care and that most participants self-reported having an undetectable viral load. Thus, the actual risk of HIV transmission to partners of negative or unknown HIV status may have been quite modest, and subsequently have influenced participants’ behavioral intentions regarding sex (81). HIV-positive MSM who have not been diagnosed, or connected with care, and/or those who are non-adherent to ART, are a high public health priority as they are more likely to have a detectable viral load that increases risk for onward transmission (82). Additional partner characteristics that may influence the likelihood of HIV transmission and dynamics of condom use negotiation, such as partners’ adherence to PrEP and partners’ substance use, were not collected on the TLFB and should be incorporated into future studies (83,84). This study also did not delineate between varying types of sexual behavior (i.e., insertive vs. receptive anal sex), which can also influence condom use decision-making (85). Last, given that this was a secondary data analysis of a study examining the relationship between alcohol use and condomless sex among MSM living with HIV, various characteristics of cannabis use that may correlate with engagement in condomless sex, such as level of intoxication at the time of the sex event (86), were not assessed. Consequently, our ability to comment on the specific pharmacological effects of various cannabis use behaviors on condomless sex is restricted. Rapidly changing cannabis use laws suggest that the influence of cannabis on sexual behavior must continuously be assessed and the generalizability of these findings are geographically restricted to New York and California (87).

In light of the study’s limitations, there were a number of critical strengths. First, the retrospective event-level design of this study allowed for detailed day-level analyses and was ideal for assessing temporal overlap of cannabis and alcohol co-use and sexual behavior (4). In this context, temporal overlap refers to cannabis and/or alcohol use occurring prior to sexual behavior, such that the acute pharmacological effects of the substances co-occur with sexual behavior during the same event. Information pertaining to the temporal overlap of substance use and sexual behavior allows for examination of the hypothesis that intoxication during sexual activity is associated with decreased condom use and informs the development of interventions that can adequately address substance use-related HIV-risk behavior among MSM. Further, this study recruited a sample of exclusively HIV-positive MSM, a population that continues to contribute to the HIV epidemic through engagement in condomless sex with non-infected counterparts (9). While biomedical HIV prevention strategies can significantly decrease the likelihood of onward transmission in the absence of condom use (62), a sizable number of HIV-negative and HIV-positive MSM remain “biologically unprotected” from acquiring or transmitting HIV (88,89). Condoms, therefore, maintain an important role in HIV prevention and efforts to better understand the factors purported to hinder condom use among HIV-positive MSM, such as substance use, continue to be crucial for prevention. Additionally, this sample of HIV-positive MSM was considerably diverse in terms of race, socioeconomic status, and educational attainment and participants were recruited from two very distinct geographic locations (Syracuse, NY and San Francisco, CA). Both the geographic and demographic diversity of the sample contribute to the generalizability of our findings to the larger HIV-positive MSM population.

Conclusion

These findings add important information to the emerging body of research on event-level associations between cannabis and alcohol co-use and condomless sex among HIV-positive MSM. Results suggest that CAS is almost three times more likely to occur on days in which both cannabis and alcohol are consumed compared to days in which no substances are used, but not in comparison with days in which a single substance is consumed. In sexual encounters with a regular partner, substance use was associated with greater likelihood of condomless sex compared to encounters with a casual partner. Interventions to address condomless sex in HIV-positive MSM may benefit from highlighting the increased risks of engaging in condomless sexual activity in the context of cannabis and alcohol co-use. An important question for future research is whether cannabis and alcohol are commonly used in combination as part of an intentional effort to become intoxicated and have sex, or if the intoxicating effects of co-use catalyze sexual activity and reduce cognitive ability to enact protected behaviors such as condom use. More research is needed regarding characteristics of substance use (e.g., quantity) and their interaction with partner type (e.g., relationship status) that may further increase the risk of engaging in condomless sex. Future research should also continue the use of more sophisticated longitudinal methods to examine causal links between these variables. Condomless sex is more likely to occur among MSM living with HIV when both cannabis and alcohol are consumed during a sexual activity day compared to no substance use, suggesting that interventions targeting CAS incorporate educational and behavioral components that address the additive risk associated with engaging in substance co-use.

Availability of data and material:

Data from this study are available upon request.