Crystal methamphetamine use and initiation among gay, bisexual, and other men who have sex with men living with HIV in a Treatment as Prevention environment

Abstract

Background:

Risk compensation in an HIV Treatment as Prevention (TasP) environment may increase high-risk sexual and substance use behaviours among people living with HIV.

Objective:

To examine recent crystal methamphetamine (CM) use/initiation in a longitudinal cohort of gay, bisexual, and other men who have sex with men (GBMSM) living with HIV in Metro Vancouver, Canada.

Methods:

Eligible participants were GBMSM aged >15 years who reported sex with another man in the past six months. Participants were recruited using respondent-driven sampling and self-completed a computer questionnaire every six months. We used multi-level generalized mixed-effect models to evaluate trends in recent CM use (past six months), multivariable logistic regression to identify covariates of recent CM use, and multivariable survival analysis to identify predictors of CM initiation.

Results:

Of 207 GBMSM living with HIV at enrollment, 44.3% reported recent CM use; there was a statistically non-significant decrease over the study period (41% in first period to 25% in final period, p=0.087). HIV treatment optimism was not associated with CM use/initiation. CM use was positively associated with depressive symptomology, sexual escape motivation, transactional sex, number of anal sex partners, condomless anal sex with seroconcordant partners, STIs, and other substance use. Recent CM use was negatively associated with viral load sorting. CM initiation was predicted by escape motivation, transactional sex, and group sex participation.

Conclusion:

Results suggest that CM use among GBMSM living with HIV is prevalent and increased CM use/initiation is not a consequence of TasP public policy.

Introduction

Crystal methamphetamine (CM) is a fervent psychostimulant that, among other effects, increases sexual desire and stamina (Vearrier, Greenberg, Miller, Okaneku, & Haggerty, 2012). In addition to physical, psychological, and interpersonal harms (Adam Bourne et al., 2015), CM use has been associated with high-risk sexual behaviours and transmission of sexually transmitted infections (STIs; (Hoenigl et al., 2016; Melendez-Torres, Hickson, Reid, Weatherburn, & Bonell, 2016; Pantalone, Huh, Nelson, Pearson, & Simoni, 2014; Rajasingham et al., 2012; Vosburgh, Mansergh, Sullivan, & Purcell, 2012). Commonly associated with social networking applications (Hegazi et al., 2017) and party and play (PnP) or “chemsex” subcultures (Nerlander et al., 2018; Tomkins, George, & Kliner, 2018), prevalence of CM use has been shown to be higher among gay, bisexual, and other men who have sex with men (GBMSM) living with HIV compared with those who are not (Schwarcz et al., 2007). Among GBMSM living with HIV, CM use has been associated with a number of detriments to their HIV care: poor adherence to prescribed highly active antiretroviral therapy (HAART) (A. W. Carrico et al., 2007; Hinkin et al., 2007; Marquez, Mitchell, Hare, John, & Klausner, 2009; Moore et al., 2012), unsuppressed HIV viral load (Fairbairn et al., 2011; Feldman, Thomas, Alexy, & Irvine, 2015; King et al., 2009; Skeer et al., 2012), HIV drug resistance (Colfax et al., 2007; Gorbach et al., 2008), HIV disease progression (Adam W. Carrico, 2011; Shoptaw et al., 2012), and other HIV-related morbidities (Langford et al., 2003; Nath, Maragos, Avison, Schmitt, & Berger, 2001). As the use of CM has been well demonstrated to be pernicious to both individual and population health, it is imperative to establish an understanding of any potential factors contributing to CM use and uptake, especially among GBMSM living with HIV.

Treatment as Prevention (TasP) is a public health strategy that asserts that use of HAART benefits individuals living with HIV through reduced morbidity and mortality and benefits the greater population through decreased secondary HIV transmission via reduced community HIV viral load (Lima et al., 2015; Lima, Hogg, & Montaner, 2010; WHO, 2012). This, together with appreciation of the efficacy of HAART, may contribute to a sense of “HIV treatment optimism”, auspicious attitudes toward HIV treatment because of reduced likelihood and consequence of HIV infection (Van de Ven, Crawford, Kippax, Knox, & Prestage, 2000). Researchers have suggested that HIV treatment optimism may contribute to behavioural risk compensation such as increased sexual risk behaviour and substance use, especially among GBMSM living with HIV (Brennan, Welles, Miner, Ross, & Rosser, 2010; Chen, 2013; Roth et al., 2017; Schwarcz et al., 2007); however, the association between HIV treatment optimism and behavioral risk compensation via substance use has not been thoroughly examined among GBMSM living with HIV.

HIV treatment optimism may contribute to risk compensation which could include increased CM use and uptake; however, research is limited. Among 197 Black GBMSM of mixed HIV serostatuses, Mimiaga et al. (2010) found that increased HIV treatment optimism was associated with reduced odds of reported CM use, and among a sample of 497 HIV-negative GBMSM living in Vancouver, Canada, treatment optimism was not associated with CM use or initiation (Colyer et al., 2018); no studies have examined this association among GBMSM living with HIV. Research exploring GBMSM’s motivations for CM use is also limited, and no studies have explored HIV treatment optimism as a covariate. Weatherburn et al. (2017) distinguished two broad motivations for combining sex and drugs among gay men: enablement to have the sex they desire (libido, confidence, disinhibition, and stamina), and enhancement of sexual qualities (attraction, physical sensation, intimacy, and adventure). Other research has discussed coping strategies and cultural norms as explanations for substance use among GBMSM (A. Bourne & Weatherburn, 2017; Nakamura, Semple, Strathdee, & Patterson, 2009; Semple, Patterson, & Grant, 2002).

Since 2010, the Canadian province of British Columbia (BC) has adopted TasP as the primary policy framework for HIV prevention and control, and has provided dedicated funding to expand access to HIV testing and treatment and to improve linkages to care for those newly diagnosed with HIV or who have been lost to care (BCCfE, 2017a, 2017b; Heath et al., 2014; Ministry of Health, 2012). We designed this study to investigate the prevalence of CM use over time and to examine the association between the use and initiation of CM and HIV treatment optimism among GBMSM living with HIV in an environment where TasP has been heavily promoted.

Materials and methods

Study design and participants

Data are drawn from the [redacted], a bio-behavioural, longitudinal study of GBMSM in Vancouver, Canada. Participants were recruited through respondent-driven sampling (RDS) from February 2012 to February 2015, with follow-up until February 2016. The sampling process was started by purposely selecting specific participants from the community as “seeds”, who then initiated chains of peer referrals. Participants were given up to 6 coupons and encouraged to recruit from members of their social and sexual networks (Lachowsky et al., 2016). Eligibility criteria included: being 16 years or older, gender identifying as a man, reporting sex with another man in the past 6 months, living in Metro Vancouver, and being able to complete the questionnaire in English. Participants provided written informed consent and completed a computer-assisted self-interview (CASI) on sociodemographic, psychosocial, and behavioural factors. Subsequently, a nurse-administered questionnaire included clinical STI/HIV screening or HIV serology including CD4 count and viral load, and history of STI and mental health diagnoses. Participation in the longitudinal cohort, with visits every 6 months to a maximum of 4 years, was optional. Participants received a CAD $50 honorarium for each visit and an additional CAD $10 for each subsequent participant recruited. This study was approved by the Research Ethics Boards of the University of British Columbia, the University of Victoria, and Simon Fraser University.

Dependent variables

The primary CM prevalence outcome was any CM use in the 6 months prior to a study visit. During the CASI, participants were asked, “In the PAST 6 MONTHS have you used Crystal Methamphetamine (‘Crystal’, ‘meth’)”, with dichotomous “No” or “Yes” response options. Those who selected “Yes” were asked to specify the number of days they used CM in the past 6 months. The secondary CM incidence outcome was first reported use of CM among participants who had not previously reported CM use in the study.

Independent variables

Demographic information included age, sexual orientation, race/ethnicity, and relationship status. Psychosocial factors were measured with several scales:

• HIV Treatment Optimism-Skepticism Scale (HOSS): 12-items with higher scores indicating greater optimism (Van de Ven et al., 2000). Using a 4-point Likert scale from “Strongly Disagree” to “Strongly Agree”, participants provided their level of agreement with items assessing their attitudes toward HIV treatment in terms of reduced likelihood and consequence of HIV (e.g., “If every HIV-positive person took the new treatments, the AIDS epidemic would be over.”) (range: 0–36; α=0.84)

• Escape Motivation Scale: 12-items with greater scores indicating more sexual escape motivations (McKirnan, Vanable, Ostrow, & Hope, 2001). Using a 4-point Likert scale from “Strongly Disagree” to “Strongly Agree”, participants provided their level of agreement with items assessing how much using substances might be related to sexual risk-taking (e.g., “When I am high or drunk, I am more likely to do sexual things I usually wouldn’t do.”) (range: 12–48; α=0.90)

• Gay/Bisexual Self-Esteem: 7-items, reverse-coded, with higher scores indicating lower self-esteem (Herek & Glunt, 1995). Using a 4-point Likert scale from “Strongly Agree” to “Strongly Disagree”, participants provided their level of agreement with items assessing their self-esteem as gay/bisexual men (e.g., “I feel that I am a person of worth, at least on an equal basis with others.”) (range: 0–21; α=0.90)

• Hospital Anxiety and Depression Scale: two 7-item subscales with greater scores indicating more anxiety and depression symptomology, respectively (Snaith, 2003; Whelan-Goodinson, Ponsford, & Schˆnberger, 2009; Zigmond & Snaith, 1983). Using various 4-point scales, participants responded to items assessing symptoms of anxiety and depression (e.g., “I get a sort of frightened feeling as if something awful is about to happen”, and “I have lost interest in my appearance”, respectively.) (range: 0–21 for each; α_anxiety=0.86; α_depression=0.81)

HIV prevention and risk reduction strategy variables included recent (i.e., past 6 months [P6M]) consistent condom use, sero-sorting (i.e., selectively choosing sex partners based on HIV-status), sero-positioning (i.e., selective choosing receptive or insertive sexual position based on partners’ HIV statuses), viral load sorting (i.e., “having sex without condoms if my viral load is low or I’m on HIV treatment”), and self-perceived current risk of transmitting HIV. Sexual history and behavioural variables measured over the P6M included number of male sex partners, use of the internet or mobile apps to seek sex, any escort work, anal sex role preference, group sex participation, any recent STI diagnosis, and recent anal sex practice (i.e., no anal sex, only condom protected, condomless anal sex [CAS] with an HIV-positive [seroconcordant] partner, and any CAS with an HIV-negative [serodiscordant] or unknown status partner). Lastly, recent substance use variables included having received or given drugs for sex, binge drinking (defined as 5+ drinks on one occasion), and use of erectile drugs (EDs), amyl nitrite (poppers), gamma hydroxybutyrate (GHB), and ecstasy.

Statistical analysis

We conducted baseline descriptive statistics for cohort participants living with HIV. RDS-weighted population parameters were not applied as recruitment chains were “broken” since not all cross-sectional RDS study participants enrolled in the cohort. To compensate for this, we accounted for non-independence of data introduced via RDS recruitment chains as well as respondents’ multiple visits by using multi-level generalized mixed effect models (i.e., a 3-level model where study visits were clustered within participants clustered within RDS recruitment chains) (Mosley et al., 2018). We used univariable and multivariable logistic regression to examine trends in CM use with 6-month time periods as the independent variable; this was also used to assess factors associated with recent CM use. In addition, given our prospective cohort design, we performed Cox proportional hazards analysis to identify predictors of CM initiation. Variables in the univariable models with a p-value <0.2 were included for consideration in the multivariable models. Multivariable model selections were conducted using a backward elimination technique based on two criteria [Akaike Information Criterion (AIC) and type III p-values] until the final model reached the optimum (minimum) AIC. All statistical tests were two-sided and considered significant at α <0.05. All analyses were conducted using SAS® version 9.4 (SAS, North Carolina, United States).

Results

Descriptive characteristics of study population at enrollment

774 GBMSM participants were recruited, and 698 consented into the prospective cohort (90.2%). Of cohort participants, 201 (28.8%) were living with HIV at the initial study visit and 6 (0.9%) seroconverted during the follow-up period; thus, 207 participants comprised our analytic sample for this longitudinal analysis of CM use. Descriptive statistics of the study sample at enrollment are shown in Table 1.

Table 1.

Descriptive statistics of study sample at enrollment (N=201).

	n	%
Demographics
Age (median, Q1, Q3)	47	39, 51
Sexual Orientation
Gay	173	86.1
Bisexual/Other	28	13.9
Race/Ethnicity
White	157	78.1
Asian	13	6.5
Indigenous	17	8.5
Other	14	7.0
Relationship Status
Single	123	61.2
Partnered/Married	78	38.8
HIV Prevention Practices and Factors
Self-Perceived Current Risk of Transmitting HIV
Low	182	90.6
High	19	9.5
Sexual History and Practices
Number of Anal Sex Partners in P6M (median, Q1, Q3)	5	2, 20
Escort Work
No	139	69.2
Yes, in P6M	16	8.0
Yes, not in P6M	46	22.9
Anal Sex over P6M
No anal sex	19	9.7
Only condom-protected	26	13.3
Any condomless, but only seroconcordant	53	27.0
Any condomless, including an unknown status or serodiscordant partner	98	50.0
Participated in Group Sex in P6M	70	34.8
Diagnosis of any STI in P6M	33	17.9
Alcohol and Substance Use
Received Drugs for Sex
No	147	73.1
Yes, in P6M	34	16.9
Yes, not in P6M	20	10.0
Gave Drugs for Sex
No	166	82.6
Yes, in P6M	19	9.5
Yes, not in P6M	16	8.0
Binge Drinking Frequency
Monthly or less	179	90.9
Weekly/Daily or almost daily	18	9.1
Used EDDs in P6M	96	47.8
Used Poppers in P6M	110	54.7
Used GHB in P6M	70	34.8
Used Ecstasy in P6M	51	25.4

Note: Q1,Q3 = first quartile, third quartile values; P6M = past 6 months; STI = sexually transmitted infection; EDDs = erectile dysfunction drugs; GHB = gamma-hydroxybutyrate

At enrollment, 44.3% of GBMSM living with HIV reported any CM use in the previous 6 months. Among these, routes of administration were: 84.3% smoking, 51.7% snorting, 29.2% injecting, and 31.5% hooping (i.e., anal administration). Of those reporting any use, 34.8% used less than monthly, 25.8% used monthly, 15.7% used weekly, 9.0% used more than weekly, and 14.6% used daily or almost daily. Of the 207 baseline participants, 172 (83.1%) returned for at least one follow-up visit, of whom 80 (46.5%) reported recent CM use at enrollment.

During the first 6-month interval of our study (February 2012 – August 2012), 41% reported recent CM use. 25% reported recent use during our last study interval (September 2015 – February 2016); however, there was no statistically significant temporal trend (p=0.087).

CM use differed by retention in the cohort, but not by follow-up time. 11% of individuals reporting CM use at least once did not complete any further visits (were lost to follow-up) compared with 22% of those who reported no CM use during the study period (p=0.038). Nevertheless, there was no difference in median follow-up time between those who ever reported CM use (2.5 years) compared with those who never reported any CM use (2.1 years, p=0.071).

Factors associated with recent crystal methamphetamine use

Descriptive statistics and univariable associations of correlates of recent CM use over all cohort study visits are shown in Table 2, while Table 3 provides the results of our final multivariable regression analysis. The median HIV treatment optimism score was 29 (Quartile 1, Quartile 3 [Q1,Q3]: 26,33) for those not reporting CM use, and 31 (Q1,Q3: 28,34) for those reporting CM use (p=0.123). Treatment optimism was not significantly associated with CM use (OR=1.02, 95%CI:0.97–1.08) in univariable analysis and was not retained in our multivariable model. In the multivariable model, recent CM use was positively associated with higher escape motive scale scores (aOR=1.08, 95%CI:1.02–1.14), higher HADS depression subscale scores (aOR=1.16, 95%CI:1.06–1.27), increased number of recent male anal sex partners (aOR=1.02, 95%CI:1.01–1.04), having CAS with a seroconcordant partner vs. only condom-protected anal sex (aOR=2.84, 95%CI:1.10–7.34), recent STI diagnosis (aOR=2.84, 95%CI:1.09–7.37), having received drugs for sex (P6M: aOR=18.53, 95%CI:3.97–86.55; >P6M: aOR=4.59, 95% CI:1.83–11.50), and recent use of poppers (aOR=2.03, 95%CI:1.06–3.88), GHB (aOR=11.60, 95%CI:4.80–28.03), and ecstasy (aOR=2.85, 95%CI:1.08–7.47). Recent CM use was inversely associated with use of viral load sorting as an HIV prevention strategy (aOR=0.41, 95%CI:0.21–0.81).

Table 2.

Descriptive statistics and univariable associations with any recent (past 6 months) crystal methamphetamine use the course of the study versus no use. All cohort participant visits (N = 915).

	No CM use P6M		Any CM Use P6M		Univariable
	n	%	n	%	OR	95% CI
Demographics
Age (median, Q1, Q3)	49	43, 55	47	39.5, 51	0.95	0.92	0.99
Sexual Orientation
Gay	539	66.4	273	33.6	Ref
Bisexual/Other	56	54.4	47	45.6	1.49	0.64	3.45
Race/Ethnicity
White	461	63.7	263	36.3	Ref
Asian	49	81.7	11	18.3	0.23	0.04	1.27
Indigenous	53	65.4	28	34.6	1.11	0.29	4.19
Other	32	64.0	18	36.0	0.95	0.20	4.38
Relationship Status
Single	338	60.4	222	39.6	Ref
Partnered/Married	257	72.4	98	27.6	0.53	0.30	0.95
Self-Perceived Current Health
Excellent/Very good	310	70.0	133	30.0	Ref
Good	193	62.1	118	37.9	1.23	0.71	2.13
Fair/Poor	92	58.2	66	41.8	1.65	0.82	3.32
Psychosocial Factors and Scales
HIV Treatment Optimism Scale (α=0.84) (median, Q1, Q3)	29	26, 33	31	28, 34	1.02	0.97	1.08
Escape Motivation Scale (α=0.90) (median, Q1, Q3)	28	24, 32	31	28.5, 36	1.14	1.09	1.20
Gay/Bisexual Self-Esteem Scale (α=0.90) (median, Q1, Q3)	6	2, 9	7	4, 10	1.12	1.04	1.21
HADS Anxiety Subscale (α=0.86) (median, Q1, Q3)	6	3, 9	8	5, 11	1.09	1.02	1.16
HADS Depression Subscale (α=0.81) (median, Q1, Q3)	3	1, 6	6	2, 8	1.16	1.07	1.25
HIV Prevention Practices and Factors
Consistent Condom Use (Ref: No)	210	79.8	53	20.2	0.37	0.20	0.67
Sero-position for Anal Sex (Ref: No)	155	56.0	122	44.0	1.79	1.06	3.03
Sero-sort for Anal Sex (Ref: No)	230	56.1	180	43.9	2.11	1.29	3.44
Condomless Anal Sex if own Viral Load is low or on HIV Treatment (Ref: No)	229	63.3	133	36.7	1.00	0.61	1.65
Self-Perceived Current Risk of Transmitting HIV
Low	565	65.2	302	34.8	Ref
High	29	61.7	18	38.3	1.54	0.51	4.66
Sexual History and Practices
Number of Anal Sex Partners in P6M (median, Q1, Q3)	2	0, 5	5	2, 25	1.07	1.04	1.10
Used Internet to Seek Sex
Not in the past 6 months	289	80.1	72	19.9	Ref
Less than once per month	103	62.4	62	37.6	3.21	1.59	6.47
About once per month	55	55.6	44	44.4	4.62	1.99	10.72
More than once per month	148	51.0	142	49.0	8.97	4.49	17.94
Escort Work
No	445	69.7	193	30.3	Ref
Yes, in P6M	11	21.6	40	78.4	13.80	3.66	52.12
Yes, not in P6M	139	61.5	87	38.5	1.66	0.73	3.79
Anal Sex over P6M
Only condom-protected	94	76.4	29	23.6	Ref
No anal sex	178	87.7	25	12.3	0.49	0.20	1.23
Any condomless, but only seroconcordant	104	53.3	91	46.7	3.98	1.69	9.35
Any condomless, including unknown status or serodiscordant partner	211	55.5	169	44.5	2.83	1.30	6.16
Participated in Group Sex in P6M (Ref: No)	123	50.8	119	49.2	3.74	2.10	6.66
Diagnosis of any STI in P6M (Ref: No)	40	47.1	45	52.9	2.71	1.30	5.65
Alcohol and Substance Use
Received Drugs for Sex
No	485	77.0	145	23.0	Ref
Yes, in P6M	11	12.8	75	87.2	70.12	19.36	253.99
Yes, not in P6M	99	49.7	100	50.3	4.94	2.19	11.13
Gave Drugs for Sex
No	506	70.6	211	29.4	Ref
Yes, in P6M	15	24.6	46	75.4	12.76	3.42	47.63
Yes, not in P6M	74	54.0	63	46.0	2.71	1.06	6.92
Binge Drinking Frequency
Monthly or less	534	65.7	279	34.3	Ref
Weekly/Daily or almost daily	47	58.8	33	41.3	1.90	0.78	4.63
Used EDDs in P6M (Ref: No)	194	50.7	189	49.3	3.65	2.11	6.31
Used Poppers in P6M (Ref: No)	185	48.2	199	51.8	4.25	2.54	7.10
Used GHB in P6M (Ref: No)	29	14.0	178	86.0	29.13	14.32	59.27
Used Ecstasy in P6M (Ref: No)	33	23.6	107	76.4	9.95	4.59	21.54

Note: Q1,Q3 = first quartile, third quartile values; 95% CI = 95% confidence interval; HADS = Hospital Anxiety and Depression Scale; P6M = past 6 months; STI = sexually transmitted infection; EDDs = erectile dysfunction drugs; GHB = gamma hydroxybutyrate; Data with bold emphasis indicates statistical significance at p<0.05.

Table 3.

Final multivariable model of factors associated with any recent (past 6 months) crystal methamphetamine use over the course of the study versus no use. All cohort study visits (N = 915).

	Multivariable Model
	aOR	95% CI
Psychosocial Factors and Scales
Escape Motivation Scale	1.08	1.02	1.14
HADS Depression Subscale	1.16	1.06	1.27
HIV Prevention Practices and Factors
Condomless Anal Sex if own Viral Load is low or on HIV Treatment (Ref: No)	0.41	0.21	0.81
Sexual History and Practices
Number of Anal Sex Partners in P6M	1.02	1.01	1.04
Anal Sex over P6M (Ref: Only condom-protected)
No anal sex	0.54	0.19	1.50
Any condomless, but only seroconcordant	2.84	1.10	7.34
Any condomless, including an unknown status or serodiscordant partner	0.99	0.40	2.46
Diagnosis of any STI in P6M (Ref: No)	2.84	1.09	7.37
Alcohol and Substance Use
Received Drugs for Sex (Ref: No)
Yes, in P6M	18.53	3.97	86.55
Yes, not in P6M	4.59	1.83	11.50
Used Poppers in P6M (Ref: No)	2.03	1.06	3.88
Used GHB in P6M (Ref: No)	11.60	4.80	28.03
Used Ecstasy in P6M (Ref: No)	2.85	1.08	7.47

95% CI = 95% confidence interval; HADS = Hospital Anxiety and Depression Scale; P6M = past 6 months; STI = sexually transmitted infection; GHB = gamma hydroxybutyrate; Data with bold emphasis indicates statistical significance at p<0.05.

Predictors of crystal methamphetamine initiation

Of the 172 participants who completed at least one follow-up visit, 92 (53.5%) had no reported use of CM at enrollment, comprising the study sample for our analysis of CM initiation. Of these, 14 (15.2%) initiated CM use over the course of the study. The incidence rate (IR) was 6.53 per 100 person-years (95%CI: 3.86–11.04). Descriptive statistics and univariable associations comparing variables of interest between study visits indicating first use of CM and study visits with no previous CM use are reported in Table 4.

Table 4.

Descriptive statistics and univariable associations with initiation of crystal methamphetamine use over the past 6 months (P6M) versus no previously reported use. Last visit included for descriptive statistics (N=92). All cohort participant visits until first reported CM use for survival analysis (N = 389).

	No reported use of CM		First use of CM after baseline		Univariable
	n or (median)	% or (Q1, Q3)	n or (median)	% or (Q1, Q3)	HR	95% CI
Demographics
Age (median, Q1, Q3)	50.5	44, 57	49.5	34, 58	0.98	0.93	1.03
Sexual Orientation
Gay	75	86.2	12	13.8	Ref
Bisexual/Other	3	60.0	2	40.0	2.23	0.49	10.10
Race/Ethnicity
White	63	84.0	12	16.0	Ref
Asian	8	100.0	0	0.0	NA
Indigenous	4	80.0	1	20.0	1.18	0.15	9.06
Other	3	75.0	1	25.0	1.69	0.22	13.02
Relationship Status
Single	46	79.3	12	20.7	Ref
Partnered/Married	32	94.1	2	5.9	0.26	0.06	1.15
Self-Perceived Current Health
Excellent/Very good	42	84.0	8	16.0	Ref
Good	23	85.2	4	14.8	0.86	0.26	2.86
Fair/Poor	13	86.7	2	13.3	0.96	0.20	4.51
Psychosocial Factors and Scales
HIV Treatment Optimism Scale (α=0.84) (median, Q1, Q3)	31	28, 34	30	25, 32	0.98	0.89	1.09
Escape Motivation Scale (α=0.90) (median, Q1, Q3)	26	23, 31	32	28, 35	1.10	1.03	1.18
Gay/Bisexual Self-Esteem Scale (α=0.90) (median, Q1, Q3)	6	1, 8	8	6, 10	1.10	0.99	1.22
HADS Anxiety Subscale (α=0.86) (median, Q1, Q3)	7	3, 9	5	3, 7	0.98	0.87	1.11
HADS Depression Subscale (α=0.81) (median, Q1, Q3)	4	1, 7	5	1, 7	1.05	0.92	1.19
HIV Prevention Practices and Factors
Consistent Condom Use (Ref: No)	29	87.9	4	12.1	0.86	0.27	2.78
Sero-position for Anal Sex (Ref: No)	19	67.9	9	32.1	5.56	1.86	16.61
Sero-sort for Anal Sex (Ref: No)	32	78.0	9	22.0	2.71	0.90	8.12
Condomless Anal Sex if own Viral Load is low or on HIV Treatment (Ref: No)	37	90.2	4	9.8	0.64	0.20	2.04
Self-Perceived Current Risk of Transmitting HIV
Low	75	84.3	14	15.7	Ref
High	3	100.0	0	0.0	NA
Sexual History and Practices
Number of Anal Sex Partners in P6M	1	0, 3	3	2, 30	1.02	1.00	1.05
Used Internet to Seek Sex
Not in the past 6 months	42	93.3	3	6.7	Ref
Less than once per month	12	70.6	5	29.4	4.16	0.99	17.43
About once per month	8	88.9	1	11.1	2.18	0.22	21.22
More than once per month	16	76.2	5	23.8	2.91	0.69	12.25
Escort Work
No	58	84.1	11	15.9	Ref
Yes, in P6M	3	75.0	1	25.0	2.97	0.38	23.62
Yes, not in P6M	17	89.5	2	10.5	0.61	0.14	2.76
Anal Sex over P6M
Only condom-protected	10	83.3	2	16.7	Ref
No anal sex	30	96.8	1	3.2	0.20	0.02	2.16
Any condomless, but only seroconcordant	8	57.1	6	42.9	2.01	0.40	10.00
Any condomless, including unknown status or serodiscordant partner	29	87.9	4	12.1	0.71	0.13	1.26
Participated in Group Sex in P6M (Ref: No)	17	70.8	7	29.2	3.51	1.23	10.03
Diagnosis of any STI in P6M (Ref: No)	5	83.3	1	16.7	1.63	0.21	12.67
Alcohol and Substance Use
Received Drugs for Sex
No	68	89.5	8	10.5	Ref
Yes, in P6M	0	0.0	2	100.0	10.03	2.02	49.65
Yes, not in P6M	10	71.4	4	28.6	2.95	0.88	9.84
Gave Drugs for Sex
No	68	88.3	9	11.7	Ref
Yes, in P6M		0.0	1	100.0	4.95	0.61	40.31
Yes, not in P6M	10	71.4	4	28.6	2.87	0.88	9.35
Binge Drinking Frequency
Monthly or less	68	86.1	11	13.9	Ref
Weekly/Daily or almost daily	6	75.0	2	25.0	2.63	0.58	12.04
Used EDDs in P6M (Ref: No)	26	78.8	7	21.2	1.95	0.68	5.57
Used Poppers in P6M (Ref: No)	21	72.4	8	27.6	3.57	1.23	10.35
Used GHB in P6M (Ref: No)	3	33.3	6	66.7	10.57	3.58	31.23
Used Ecstasy in P6M (Ref: No)	2	50.0	2	50.0	4.11	0.90	18.85

Note: Q1,Q3 = first quartile, third quartile values; 95% CI = 95% confidence interval; HADS = Hospital Anxiety and Depression Scale; P6M = past 6 months; STI = sexually transmitted infection; EDDs = erectile dysfunction drugs; GHB = gamma hydroxybutyrate; Data with bold emphasis indicates statistical significance at p<0.05.

HIV treatment optimism was not associated with CM initiation (HR=0.98, 95%CI:0.89–1.09) and was not included in the final multivariable model (Table 5). CM initiation was associated with higher scores on the escape motive scale (adjusted hazards ratio [aHR]=1.09, 95%CI:1.02–1.16), recent group sex participation (aHR=4.82, 95%CI:1.50–15.49), and having received drugs for sex in the P6M (aHR=8.44, 95%CI:1.61–44.25).

Table 5.

Final multivariable model of factors associated with first reported use of crystal methamphetamine use over the past 6 months (P6M) versus no reported use throughout the study period. All cohort participant visits until first reported CM use for survival analysis (N = 389).

	Multivariable Model
	aHR	95% CI
Psychosocial Factors and Scales
Escape Motivation Scale (α=0.90)	1.09	1.02	1.16
Sexual History and Practices
Participated in Group Sex in P6M (Ref: No)
Yes	4.82	1.50	15.49
Alcohol and Substance Use
Received Drugs for Sex (Ref: No)
Yes	8.44	1.61	44.25

Note: 95% CI = 95% confidence interval; P6M = past 6 months; Data with bold emphasis indicates statistical significance at p<0.05.

Discussion

Among GBMSM living with HIV in Vancouver, we found that the prevalence of CM use decreased over the 4-year study period (41% in 2012 to 25% in 2016), but that this trend was not statistically significant. Additionally, we found no association between HIV treatment optimism and recent CM use or CM initiation. These observations occurred during a time when TasP was actively promoted in BC as it was adopted as policy in 2010 (BCCfE, 2017a, 2017b; Heath et al., 2014). These results demonstrate an absence of evidence that would reject the null hypothesis that CM use among GBMSM living with HIV in Vancouver is not associated with the expansion of HIV treatment and any consequent HIV treatment optimism. It is important to note however that CM use was highly prevalent throughout our study.

Consistent with our results, Lea et al. (2016) reported no trend in regular (at least monthly) CM use from 2010 to 2014 among GBMSM living with HIV in Australia (prevalence: 10.9% in 2010 and 10.8% in 2014); however, a significant increase in any CM use, 22.8% in 2010 to 27.4% in 2014, was reported. The latter is more similar to our measure; however reported prevalence is much lower.

Significant correlates of CM use included depressive symptomology, escape motivation, and other substance use in our analyses, with escape motivation also being a significant predictor of CM initiation. We speculate that GBMSM living with HIV participated in CM and other drug-use behavior in efforts to cope with psychological distress originating from compounding stigma from HIV-positive serostatus and sexual minority identity. This contributes to evidence of a syndemic association between depression and CM use among some GBMSM living with HIV (Bousman et al., 2009; Halkitis et al., 2008; Peck, Reback, Yang, Rotheram-Fuller, & Shoptaw, 2005), and to research concluding that CM is used by some individuals living with HIV to cope with their HIV status (Nakamura et al., 2009; Semple et al., 2002). While CM use is a risk factor for HIV acquisition among HIV-negative GBMSM (Hoenigl et al., 2016; Plankey et al., 2007; Rajasingham et al., 2012; Vosburgh et al., 2012), most (65%) CM-using GBMSM living with HIV in a cross-sectional study reported only initiating CM use after seroconversion (Halkitis, Levy, & Solomon, 2016). These coping motivations correspond with our finding of escape motivation being a significant correlate of recent CM use and a predictor of initiation, where individuals report substance use as a plausible means to facilitate “escape” from fear of HIV transmission risk and stigma. We further postulate that this escape from inhibition contributes to both increased numbers of sexual partners and increased likelihood of STI diagnosis and we hypothesize CM use to be a mediator between escape motivations and these behavioural and biological outcomes.

Our analyses describe an independent association between CM use and seroconcordant condomless anal sex, but not serodiscordant/unknown status condomless anal sex, though the latter was significant in univariable analysis. As GBMSM have significantly higher rates of CM use (Schwarcz et al., 2007), this may be a function of sexual subcultures, or it may suggest that these individuals have not completely “escaped” their anxiety regarding potential HIV transmission. Semple et al. (2006) showed that, consistent with our study, CM-using GBMSM living with HIV engaged in significantly fewer acts of anal sex with serodiscordant partners as compared with seroconcordant partners, but that mean levels of CAS were high. Nevertheless, our analyses provide some reassurance to HIV prevention efforts in that, despite high prevalence of CM use and CAS among GBMSM living with HIV, they appear to engage in CAS mostly with other partners who are also living with HIV. Still, given the significance of serodiscordant/unknown status CAS in our univariable analysis and previous research describing CM use interrupting HAART adherence and contributing to unsuppressed viral load (A. W. Carrico et al., 2007; Fairbairn et al., 2011; Feldman et al., 2015; Hinkin et al., 2007; King et al., 2009; Marquez et al., 2009; Moore et al., 2012; Skeer et al., 2012), we reason that in the context of CM, the risk of serodiscordant/unknown status CAS remains noteworthy.

Readers should be cautious when reviewing our results as our data may be affected by reporting and social desirability biases; although, the use of a CASI was intended to minimize this. Our limited sample size and power may have contributed to potential Type II errors. CM users were less likely to be lost-to-follow-up, thereby contributing to a conservative bias in the trend analysis of CM use. For the survival analysis, our only measure of CM use at baseline pertained to the six months prior to study visit, thus we could not report any historical lifetime use prior to this time frame. Consequently, anyone with historical use was considered a non-user in the survival analysis and all initiation events were considered first-time use events; we also had limited statistical power for this analysis (n=14). Lastly, given the longitudinal nature of our analyses, we were unable to use RDS-weighting to develop population parameter estimates over time, but did account for the violation of non-independence of data by using multi-level modeling.

In conclusion, among GBMSM living with HIV in Vancouver, HIV treatment optimism associated with TasP was not independently associated with CM use nor initiation of CM use. Prevalence of CM use was high and decreased non-significantly; CM use was associated with depressive symptomology, escape motivation, and other substance use, suggesting that GBMSM living with HIV may be using CM as a method of temporarily escaping from distress and stigma due to multiple marginalizations. Further work is needed to expand interventions to address better mental health and stigma, positive coping strategies, and substance use harms among GBMSM living with HIV.