Abstract
Background.
Young, black men who have sex with men are disproportionately impacted by the U.S. HIV epidemic and HIV-positive, young, black men who have sex with men face stark disparities in HIV clinical outcomes.
Methods.
We performed an observational analysis of the 199 HIV-positive black men aged 18–30 followed for 12 months in healthMpowerment, a randomized controlled trial of an Internet-based HIV prevention intervention, to identify time-varying correlates of self-reported viral suppression using relative risk regression.
Results.
Retention at the 12-month visit was 84%. One hundred five of 162 (65%) participants reported being undetectable at baseline. At 3, 6, and 12 months, 83/115 (72%), 84/103 (82%), and 101/117 (86%) reported an undetectable viral load, respectively. In a multivariable model, participants who reported homelessness (relative risk [RR]=0.85, 95% confidence interval [CI] 0.72, 0.99), who had clinically significant depressive symptoms (RR=0.88, 95%CI 0.79, 0.98), and who used methamphetamine or crack (RR=0.61, 95%CI 0.38, 0.96) were less likely to report an undetectable viral load. Young men who engaged in condomless insertive anal intercourse were more likely to report viral suppression (RR=1.14, 95%CI 1.04, 1.24).
Conclusion.
HIV care for young, black men who have sex with men must be multidimensional to address medical needs in the context of mental health, substance use, and housing insecurity.
Keywords: HIV; young, black, men who have sex with men; homelessness; depression; substance use
Summary
An observational analysis of young, HIV-positive black MSM participating in a RCT of an HIV prevention intervention showed that homelessness, depression, and stimulant use were independent predictors of a detectable viral load.
Introduction
HIV-positive black men who have sex with men (MSM) face significant inequities in HIV care, morbidity, and mortality.1 HIV viral load suppression is an important indicator not only of HIV care, but also of the risk of secondary HIV transmission. A modeling study estimated that only 16% of U.S. HIV-positive black MSM achieved an undetectable viral load.2 There are no estimates of viral suppression specific to HIV-positive, young, black MSM (YBMSM); however, data from clinics serving HIV-positive youth aged 13–24 (81% male, 72% black, 70% gay/bisexual) estimated that only 7% achieved viral suppression.3
Development of programs to create equity in HIV outcomes requires examination of the individual- and structural-level barriers that YBMSM face in navigating HIV care. Therefore, we sought to determine the correlates of viral suppression among HIV-positive black MSM participating in a randomized controlled trial of healthMpowerment (HMP), an Internet-based HIV prevention intervention designed for YBMSM.4,5
Methods
Study population
We conducted an observational analysis of HIV-positive men participating in HMP, a randomized controlled trial of an Internet-based intervention for HIV-positive and HIV-negative black MSM aged 18–30 years.4,5 HMP provided HIV and sexually transmitted infections (STI) prevention and care information, resources, personalized feedback, game-based elements, and a social networking platform to interact with other participants. The control was a website that provided basic information about HIV and STI. While participants were told that the trial would last 3 months, they could access their respective websites for the entire 12-month follow-up period.
Participants were recruited from local community-based organizations; HIV and STI clinics; local college campuses; venues frequented by young, black MSM; and, through online ads posted on Craigslist, Facebook, Grindr, Jack’d, Adam4Adam, and black Gay Chat. Study participants were also offered incentives to invite peers to participate in the study.
Eligibility criteria were (1) age 18–30; (2) born biologically male; (3) self-identification as black or African American; (4) current residence in North Carolina; (5) current access to a mobile device with Internet access and texting capabilities; and (6) any one of the following in the past 6 months: (a) condomless anal sex with a male partner; (b) anal sex with >3 cisgender male or transgender female partners; (c) exchange of sex for money, gifts, shelter, or drugs; or (d) anal sex while under the influence of drugs or alcohol (i.e., 2 hours before or during sex). Participants were eligible if they reported either insertive or receptive anal sex.
All participants completed online surveys at baseline, 3, 6, and 12 months. The study period spanned November 2013 to October 2016.
The study was approved by the University of North Carolina at Chapel Hill Institutional Review Board (IRB #10–2179).
Outcome of interest
The outcome of interest was an undetectable viral load (UVL). Baseline and follow-up surveys queried HIV-positive participants about whether their viral load was checked in the prior three months. If yes, a follow-up question asked if that viral load was undetectable. In a study assessing the accuracy of self-reported UVL, agreement between self-report and clinic-recorded viral load <50 copies/mL was 88%.6
Predictors of interest
Study-related characteristics.
We examined the relationship between self-reported viral suppression and group assignment, study follow-up time, and the total time spent on the intervention website over the 12-month study period calculated from website usage statistics.
Socio-demographic characteristics.
Survey items captured demographic information, including age, identification with other races or ethnicities, sexual identity, relationship status, educational attainment, employment status, homelessness, and involvement with the criminal justice system.
Psychosocial variables.
We used the Center for Epidemiologic Studies Depression Scale (CES-D), a validated 20-item survey, to measure depressive symptoms in the prior 2 weeks (Cronbach’s alpha = 0.90).7 The range of the scale is 0–60. A score of ≥16 indicates clinically significant depressive symptoms. The Generalized Anxiety Disorder 7-item scale (GAD-7) assessed participants’ experience of anxiety symptoms in the prior 2 weeks (Cronbach’s alpha = 0.93).8 The range of the scale is 0–21 with cut-offs of 5, 10, and 15 indicating mild, moderate, and severe anxiety, respectively. We used the Medical Outcomes Study Social Support Survey (MOS-SSS) to assess perceived social support.9 We calculated subscales of emotional support (Cronbach’s alpha = 0.97), tangible support (Cronbach’s alpha = 0.92), affectionate (Cronbach’s alpha = 0.94), and positive social interaction (Cronbach’s alpha = 0.96) as well as the overall support index. Scores ranged from 0–100 with higher scores indicating greater perceived social support. The Lubben Social Network Scale captured participants’ ease and frequency of contact with friend and familial networks (Cronbach’s alpha = 0.83).10 The range of possible scores is 1–30 with lower scores indicating greater social isolation. A score of <12 indicates social isolation. Finally, we queried participants about their experiences with discrimination due to their HIV status, sexual identity, and race. Each scale has a range of 0–10. We used the experiences with HIV discrimination (enacted stigma) subscale of an HIV stigma assessment developed by Steward and colleagues (Cronbach’s alpha = 0.84).11 We used the multiple discrimination scale to assess experiences of discrimination due to sexual orientation (MDS-Gay) and race (MDS-Race; Cronbach’s alpha = 0.88 for both).12
Sexual risk.
We asked participants to enumerate their male anal sex partners by HIV status and sexual role over the past 3 months. Survey items captured whether participants exchanged condomless anal sex for money, drugs, shelter, or gifts.
Substance use.
Survey items assessed smoking status, use of alcohol, cocaine, inhalants (e.g., poppers, nitrous oxide, glue), “club” drugs (e.g., ketamine, gamma-hydroxybutyrate [GHB], and methylenedioxy-methamphetamine [MDMA]), erectile dysfunction medications (e.g., sildenafil [Viagra], vardenafil [Levitra], tadalafil [Cialis]), crack, methamphetamine, and opiates in the past 3 months. A current smoker was defined as having smoked cigarettes in the last week. Alcohol use was classified as heavy (≥5 days per week with ≥5 drinks per day) or less than heavy (including no use). We did not include a “no use” category, as only 7 participants reported not using alcohol throughout the entire study period. All other substance use was classified as any or none.
Statistical analyses
Using generalized linear regression with a log link and binomial distribution, we estimated the relative risk (RR) as a measure of association between self-reported viral suppression and study and demographic characteristics, scores on psychosocial health scales, sexual risk, and substance use.13 All predictors were modeled as time-varying covariates except for study characteristics and immutable demographic characteristics. A robust variance estimator accounted for repeated measures over time.
First, we examined univariable models for each predictor of interest within three domains: (1) study and demographic variables, (2) psychosocial variables, and (3) sexual risk and substance use variables. Scores on psychosocial indices were evaluated as continuous variables in the univariable models as most of the scales do not have standardized cut-offs. Thus, the interpretation of the RR is the change in the probability of reporting a UVL associated with a one-unit change in the score on the psychosocial variable. Second, those predictors that were significant at the P<0.2 level were included in a multivariable model examining predictors within each of the three domains. Finally, those predictors that were significant at the P<0.05 level in the three multivariable, domain-specific models were included in a final, full multivariable model. The three domain-specific models and the final model were adjusted for study group. For the final multivariable model, we defined statistical significance at the P<0.05 level. All analyses were conducted using STATA 14.2 (College Station, TX).
Results
We enrolled 199 HIV-positive participants of whom 174 (87%), 167 (84%), and 168 (84%) completed 3, 6, and 12-month online surveys, respectively (Table 1). Ninety-two (46%) were randomized to the control website and 107 (54%) were randomized to the HMP intervention. The median age was 24.5 (interquartile range [IQR] 23–27). Thirty-four percent identified as multiracial or multiethnic. Sixty-eight percent identified as gay, 14% as bisexual, 15% as queer, transgender, or other, and 2% as straight or questioning. Sixty-one percent were single, 67% completed more than a high school education, and 60% were currently employed. Eighty percent had health insurance. One third of participants reported being homeless in the past 3 months and 8% reported being arrested or incarcerated in the past 3 months.
Table 1.
Baseline study and demographic characteristics, psychosocial health, and sexual and substance use behaviors of HIV-positive, young, black men who have sex with men participating in healthMpowerment, 2013–2016.
| Proportion of participants, n (%); and/or median (IQR) N = 199 |
|
|---|---|
| Study-related characteristics | |
|
| |
| Group | |
| Control | 92 (46) |
| Intervention | 107 (54) |
| Retention | |
| 3 months | 174 (87) |
| 6 months | 167 (84) |
| 12 months | 168 (84) |
|
| |
| Socio-demographic characteristics | |
|
| |
| Age | 24.5 (23–27) |
| Multiracial/multiethnic | 65 (34) |
| Sexual identity | |
| Gay | 136 (68) |
| Bisexual | 28 (14) |
| Queer/Transgender/Other | 30 (15) |
| Straight/Questioning | 5 (2) |
| Single | 121 (61) |
| Completed more than high school education | 134 (67) |
| Currently employed | 119 (60) |
| Health Insurance | 160 (80) |
| Homeless, last 3 months | 66 (33) |
| Arrested or incarcerated, last 3 months | 16 (8) |
|
| |
| Psychosocial indices | |
|
| |
| CES-D, score 16 or greater; median (IQR) | 114 (58); 17 (10–26) |
| GAD-7, score 10 or greater; median (IQR) | 29 (15); 6 (2–11) |
| Lubben social network score, score less than 12; median (IQR) |
80 (40); 13 (9–17) |
| Overall social support score | 74 (50–95) |
| Emotional support subscore | 75 (50–100) |
| Tangible support subscore | 75 (44–94) |
| Positive social interaction support subscore | 75 (50–100) |
| Affectionate support subscore | 75 (50–100) |
| Experience with HIV discrimination score | 1 (0–3) |
| Experience with sexual minority discrimination score |
2 (0–4) |
| Experience with racism score | 1 (0–4) |
|
| |
| Sexual behavior | |
|
| |
| Any condomless receptive anal intercourse | 90 (45) |
| Non-concordant condomless receptive anal intercoursea |
31 (16) |
| Any condomless insertive anal intercourse | 71 (36) |
| Non-concordant condomless insertive anal intercourse |
27 (14) |
| Use of alcohol or drugs prior to non- concordant condomless anal intercourse |
57 (29) |
|
| |
| Substance use | |
|
| |
| Cigarettes, past week | 109 (55) |
| Heavy alcohol use, past 3 monthsb | 29 (15) |
| Cocaine use, past 3 months | 26 (13) |
| Inhalants (poppers, glue, nitrous oxide), past 3 months |
14 (7) |
| Erectile dysfunction medications, past 3 months |
7 (4) |
| Ketamine, GHB, MDMA, past 3 months | 14 (7) |
| Methamphetamine or crack use, past 3 months |
18 (9) |
| Opiates, including heroin, past 3 months | 11 (6) |
CES-D, Center for Epidemiologic Studies Depression Scale; GAD-7, Generalized Anxiety Disorder 7-item scale; IQR, interquartile range.
Non-concordant is defined as anal intercourse with a man of negative or unknown HIV status.
Heavy alcohol use is defined as alcohol use on 5 or more days per week with 5 or more drinks per day.
Over the course of the 12-month study period, HIV-positive participants used their assigned website for a median of 11 minutes (interquartile range [IQR] 5–38.5 minutes; range 1–1250 minutes) and a mean of 73 minutes (standard deviation [SD] 190 minutes).
At baseline, 162/199 (81%) subjects had a viral load measured in the prior 3 months with 115/174 (66%), 103/167 (62%), and 118/168 (70%) reporting viral load measurements in the prior 3 months on the 3, 6, and 12-month follow-up surveys. One hundred five of 162 (65%) participants reported being undetectable at baseline. At 3, 6, and 12 months, 83/115 (72%), 84/103 (82%), and 101/117 (86%) reported a UVL, respectively. Overall, participants reported being undetectable on 373/497 (75%) online surveys.
Ten participants completed only the baseline survey and 5 (50%) reported a UVL. The 21 participants who completed one follow-up survey after baseline contributed 30 assessments and reported a UVL on 21 (70%) of those surveys. Of the 66 surveys contributed by the 35 participants who completed the two surveys after baseline, a UVL was reported on 45 (68%) of those assessments. Finally, 133 participants who completed all of the online surveys contributed 391 assessments on which a UVL was reported at 302 (77%).
Univariable models
Table 2 presents the proportion of online surveys (including baseline and follow-up surveys) in which participants report being undetectable or not by study features and socio-demographic characteristics. The proportion of surveys in which participants reported a UVL increased over time. In addition, completing more than a high school education and current employment were associated with a greater probability of self-reported viral suppression. Identification as multiracial or multiethnic; identification as queer, transgender, or other; being single; and, homelessness in the prior 3 months were associated with a greater risk of a detectable viral load.
Table 2.
Self-reported undetectable viral load and study and demographic characteristics of HIV-positive, young, black men who have sex with men participating in healthMpowerment, 2013–2016.
| Proportion of surveys on which participants do not report a UVL, n/N (%) | Proportion of surveys on which participants report a UVL, n/N (%) | Univariable RR (95% CI) | P-value | |
|---|---|---|---|---|
| Study-related characteristics | ||||
|
| ||||
| Group | ||||
| Control | 64/243 (26) | 179/243 (74) | REF | |
| Intervention | 60/254 (24) | 194/254 (76) | 1.04 (0.91, 1.18) | 0.582 |
| Website usage | ||||
| 120 minutes or less | 111/432 (26) | 321/432 (74) | REF | |
| More than 120 minutes | 13/65 (20) | 52/65 (80) | 1.08 (0.93, 1.24) | 0.890 |
| Survey time | ||||
| Baseline | 57/162 (35) | 105/162 (65) | REF | |
| 3 months | 32/115 (28) | 83/115 (72) | 1.11 (0.97, 1.23) | 0.129 |
| 6 months | 19/103 (18) | 84/115 (82) | 1.26 (1.10, 1.43) | 0.001 |
| 12 months | 16/117 (14) | 101/117 (86) | 1.33 (1.18, 1.51) | <0.001 |
|
| ||||
| Demographic characteristics | ||||
|
| ||||
| Age | ||||
| 18–24 | 52/206 (25) | 154/206 (75) | REF | |
| 25–35 | 72/291 (25) | 221/291 (75) | 1.00 (0.89, 1.14) | 0.915 |
| Multiracial/multiethnic | ||||
| No | 69/324 (21) | 255/324 (79) | REF | |
| Yes | 36/42 (44) | 46/82 (56) | 0.84 (0.72, 0.98) | 0.023 |
| Sexual identity | ||||
| Gay | 83/366 (23) | 283/366 (77) | REF | |
| Bisexual | 13/63 (21) | 50/63 (79) | 1.03 (0.89, 1.18) | 0.718 |
| Queer/Transgender/Other | 24/57 (42) | 33/57 (58) | 0.75 (0.58, 0.97) | 0.029 |
| Straight/Questioning | 4/11 (36) | 7/11 (64) | 0.82 (0.54, 1.25) | 0.359 |
| Relationship status | ||||
| In a relationship | 34/198 (17) | 164/198 (83) | REF | |
| Single | 90/299 (30) | 209/299 (70) | 0.84 (0.75, 0.95) | 0.003 |
| Highest level of education | ||||
| High school or less | 48/151 (32) | 103/151 (68) | REF | |
| More than high school | 27/345 (22) | 269/345 (78) | 1.14 (0.97, 1.34) | 0.100 |
| Currently employed | ||||
| No | 57/175 (33) | 118/175 (67) | REF | |
| Yes | 67/322 (21) | 255/322 (79) | 1.17 (1.03, 1.34) | 0.019 |
| Health Insurance | ||||
| No | 20/73 (27) | 53/73 (73) | REF | |
| Yes | 104/424 (24) | 320/424 (76) | 1.03 (0.89, 1.21) | 0.614 |
| Homeless, last 3 months | ||||
| No | 72/371 (19) | 299/371 (81) | REF | |
| Yes | 52/124 (42) | 72/124 (58) | 0.72 (0.60, 0.86) | <0.001 |
| Arrested or incarcerated, last 3 months | ||||
| No | 111/456 (25) | 341/456 (75) | REF | |
| Yes | 13/45 (29) | 32/45 (71) | 0.94 (0.76, 1.16) | 0.585 |
CI, confidence interval; REF, reference group; RR, relative risk; UVL, undetectable viral load. Bold indicates P < 0.2 and inclusion into multivariable model.
Table 3 presents the median and IQR of scores on each of the psychosocial indices by study visit and self-report of a UVL. Participants who reported greater depressive and anxiety symptoms and scored higher on scales measuring experiences of discrimination based on HIV status and sexual orientation were less likely to report a UVL. Conversely, higher levels of emotional support and lower levels of social isolation were associated with a greater probability of a UVL.
Table 3.
Self-reported undetectable viral load and psychosocial health indices of HIV-positive, young, Black men who have sex with men participating in healthMpowerment, 2013–2016.
| Median (IQR) on surveys in which participants do not report a UVL | Median (IQR) on surveys in which participants report a UVL | Univariable RRa (95% CI) | P-value | |
|---|---|---|---|---|
| CES-D score | 0.99 (0.98, 1.0) | 0.001 | ||
| Baseline | 17.5 (11.5–25.5) | 16 (9–23) | ||
| 3 months | 22 (11–29) | 12 (8–21) | ||
| 6 months | 25 (12–36) | 13 (8–23) | ||
| 12 months | 26.5 (12.5–30) | 15 (7–24) | ||
| GAD-7 score | 0.99 (0.98, 1.0) | 0.058 | ||
| Baseline | 6 (1–10) | 6 (2–11) | ||
| 3 months | 7.5 (1–14) | 3 (0–9) | ||
| 6 months | 6 (0–11) | 4 (0–9) | ||
| 12 months | 10 (2–13) | 5 (0–9) | ||
| Lubben social isolation score | 1.01 (1.00, 1.02) | 0.069 | ||
| Baseline | 13 (10–16) | 14 (9–18) | ||
| 3 months | 10.5 (7–15) | 14 (10–18) | ||
| 6 months | 12 (7–17) | 13 (10–18) | ||
| 12 months | 9.5 (6–12) | 15 (10–18) | ||
| Overall social support scale | 1.00 (0.99, 1.00) | 0.201 | ||
| Baseline | 75 (51–95) | 74 (50–99) | ||
| 3 months | 64 (48–82) | 70 (50–96) | ||
| 6 months | 50 (50–74) | 71 (50–95) | ||
| 12 months | 59 (47–75) | 75 (50–100) | ||
| Emotional support | 1.00 (1.00, 1.01) | 0.120 | ||
| Baseline | 75 (50–97) | 75 (50–100) | ||
| 3 months | 59 (46–94) | 75 (50–100) | ||
| 6 months | 50 (47–75) | 75 (50–100) | ||
| 12 months | 61 (50–90) | 75 (50–100) | ||
| Tangible support | 1.00 (0.99, 1.00) | 0.236 | ||
| Baseline | 76 (44–94) | 75 (44–100) | ||
| 3 months | 62.5 (41–78) | 56 (44–100) | ||
| 6 months | 50 (44–75) | 69 (50–94) | ||
| 12 months | 62.5 (50–81) | 75 (50–100) | ||
| Positive social interaction support | 1.00 (0.99, 1.00) | 0.704 | ||
| Baseline | 71 (50–100) | 75 (50–100) | ||
| 3 months | 67 (42–100) | 75 (50–100) | ||
| 6 months | 50 (42–75) | 71 (50–100) | ||
| 12 months | 67 (50–92) | 67 (50–100) | ||
| Affectionate support | 1.00 (0.99, 1.00) | 0.490 | ||
| Baseline | 75 (50–100) | 75 (46–100) | ||
| 3 months | 58 (42–100) | 75 (42–100) | ||
| 6 months | 50 (50–75) | 67 (46–100) | ||
| 12 months | 54 (33–87.5) | 67 (50–100) | ||
| Experienced HIV discrimination score | 0.97 (0.95, 1.00) | 0.052 | ||
| Baseline | 1 (0–3) | 1 (0–2) | ||
| 3 months | 1 (0–3.5) | 0 (0–3) | ||
| 6 months | 2 (1–5) | 0 (0–2) | ||
| 12 months | 1 (0–4.5) | 0 (0–2) | ||
| Experienced sexual minority discrimination score | 0.97 (0.95, 0.99) | 0.026 | ||
| Baseline | 2 (0–4.5) | 1 (0–4) | ||
| 3 months | 2 (0–5) | 0 (0–4) | ||
| 6 months | 4 (0–7) | 0 (0–3) | ||
| 12 months | 0 (0–2) | 0 (0–2) | ||
| Experienced racial discrimination score | 0.99 (0.96, 1.01) | 0.221 | ||
| Baseline | 0 (0–4) | 1 (0–3) | ||
| 3 months | 2 (0–4) | 0 (0–3) | ||
| 6 months | 0 (0–5) | 0 (0–2) | ||
| 12 months | 0 (0–2.5) | 0 (0–3) |
CES-D, Center for Epidemiologic Studies Depression Scale; CI, confidence interval; GAD-7, Generalized Anxiety Disorder 7-item scale; IQR, interquartile range; RR, relative risk; UVL, undetectable viral load. Bold indicates P < 0.2 and inclusion into multivariable model.
Interpretation of relative risk is the change in probability of reporting an undetectable viral load associated with a one-unit change in the score of the psychosocial index.
Study participants were more likely to report a UVL on online surveys in which they also reported condomless insertive anal intercourse (Table 4). Substance use before or during condomless anal sex with a partner of unknown or negative (non-concordant) HIV status was associated with lower probability of a UVL. Study participants were less likely to report a UVL on online assessments in which they reported smoking cigarettes, using methamphetamine or crack, or using opiates.
Table 4.
Self-reported undetectable viral load and sexual risk and substance use of young, HIV-positive, black men who have sex with men participating in healthMpowerment, 2013–2016.
| Proportion of surveys on which participants do not report a UVL, n/N (%) | Proportion of surveys on which participants report a UVL, n/N (%) | Univariable RR (95% CI) | P-value | |
|---|---|---|---|---|
| Sexual risk | ||||
|
| ||||
| Condomless receptive anal intercourse with a partner of any HIV status | ||||
| No | 78/314 (25) | 236/314 (75) | REF | |
| Yes | 46/183 (25) | 137/183 (75) | 1.0 (0.89, 1.11) | 0.942 |
| Non-concordant condomless receptive anal intercoursea | ||||
| No | 106/436 (24) | 330/436 (76) | REF | |
| Yes | 18/58 (31) | 40/58 (69) | 0.91 (0.75, 1.11) | 0.362 |
| Condomless insertive anal intercourse with a partner of any HIV status | ||||
| No | 97/361 (27) | 264/361 (73) | REF | |
| Yes | 27/137 (20) | 109/136 (80) | 1.10 (0.99, 1.23) | 0.086 |
| Non-concordant condomless insertive anal intercourse | ||||
| No | 116/452 (26) | 336/452 (74) | REF | |
| Yes | 8/43 (19) | 35/43 (81) | 1.09 (0.93, 1.28) | 0.261 |
| Use of alcohol or drugs prior to non-concordant condomless anal intercourse | ||||
| No | 94/403 (23) | 309/403 (77) | REF | |
| Yes | 30/94 (32) | 64/94 (68) | 0.89 (0.75, 1.04) | 0.152 |
|
| ||||
| Substance use | ||||
|
| ||||
| Cigarette smoking, past week | ||||
| No | 48/262 (18) | 214/262 (82) | REF | |
| Yes | 76/235 (32) | 159/235 (68) | 0.83 (0.73, 0.93) | 0.002 |
| Alcohol use, past 3 months | ||||
| None, or less than heavy | 107/435 (25) | 328/435 (75) | REF | |
| Heavyb | 17/62 (27) | 45/62 (73) | 0.96 (0.80, 1.15) | 0.679 |
| Cocaine use, past 3 months | ||||
| No | 24/91 (26) | 67/91 (74) | REF | |
| Yes | 15/48 (31) | 33/48 (69) | 0.93 (0.74, 1.18) | 0.568 |
| Inhalants (poppers, glue, nitrous oxide), past 3 months | ||||
| No | 24/92 (26) | 68/92 (74) | REF | |
| Yes | 9/21 (43) | 12/21 (57) | 0.77 (0.51, 1.17) | 0.227 |
| Erectile dysfunction medications, past 3 months | ||||
| No | 25/97 (26) | 72/97 (74) | REF | |
| Yes | 4/11 (36) | 7/11 (64) | 0.86 (0.54, 1.35) | 0.508 |
| Ketamine, GHB, MDMA, past 3 months | ||||
| No | 26/95 (27) | 69/95 (73) | REF | |
| Yes | 8/20 (40) | 12/20 (60) | 0.83 (0.57, 1.20) | 0.322 |
| Methamphetamine or crack use, past 3 months | ||||
| No | 105/465 (23) | 360/465 (77) | REF | |
| Yes | 19/32 (59) | 13/32 (41) | 0.52 (0.32, 0.85) | 0.008 |
| Opiates, including heroin, past 3 months | ||||
| No | 115/480 (24) | 365/480 (76) | REF | |
| Yes | 9/17 (53) | 8/17 (47) | 0.62 (0.36, 1.06) | 0.082 |
CI, confidence interval; REF, reference group; RR, relative risk; UVL, undetectable viral load. Bold indicates P < 0.2 and inclusion into multivariable model.
Non-concordant is defined as anal intercourse with a man of negative or unknown HIV status.
Heavy alcohol use is defined as alcohol use on 5 or more days per week with 5 or more drinks per day.
Multivariable models: domain-specific models
In the multivariable model of study and socio-demographic characteristics (Table 5), participants were more likely to report a UVL over study time. Compared to those participants who reported being housed, those who were homeless in the past 3 months were 36% less likely to report a UVL.
Table 5.
Domain-specific and final multivariable models assessing correlates of an undetectable viral load among young, HIV-positive, black men who have sex with men participating in healthMpowerment, 2013–2015.
| Model 1: Study and socio-demographic characteristics | |||
|---|---|---|---|
| Multivariable RR | 95% CI | P-value | |
|
|
|||
| Survey time, continuous | 1.06 | 1.01, 1.10 | 0.028 |
| Multiracial/multiethnic | 0.90 | 0.79, 1.01 | 0.144 |
| Queer/transgender/other | 0.67 | 0.44, 1.02 | 0.064 |
| Currently single | 0.90 | 0.80, 1.02 | 0.143 |
| Completed high school or more | 1.15 | 0.92, 1.45 | 0.308 |
| Currently employed | 1.04 | 0.88, 1.22 | 0.772 |
| Homeless, past 3 months | 0.64 | 0.48, 0.84 | 0.002 |
| Model 2: Psychosocial health indices | |||
| Multivariable RR | 95% CI | P-value | |
|
|
|||
| CES-D score (≥16 v <16) | 0.85 | 0.76, 0.96 | 0.020 |
| GAD-7 score (≥10 v <10) | 0.91 | 0.73, 1.13 | 0.410 |
| Lubben social network scale (<12 v ≥12) | 0.99 | 0.89, 1.11 | 0.924 |
| Emotional support, continuous a | 1.0 | 0.99, 1.01 | 0.733 |
| Experienced HIV discrimination, continuous a | 1.0 | 0.97, 1.03 | 0.991 |
| Experienced sexual minority discrimination, continuous a | 0.99 | 0.96, 1.01 | 0.274 |
| Model 3: Sexual and substance use behaviors | |||
| Multivariable RR | 95% CI | P-value | |
|
|
|||
| Condomless insertive anal intercourse, past 3 months | 1.10 | 1.00, 1.22 | 0.032 |
| Sex or drugs during condomless sex with a discordant partner, past 3 months | 0.99 | 0.87, 1.12 | 0.845 |
| Smoking, past week | 0.87 | 0.78, 0.97 | 0.014 |
| Methamphetamine or crack use, past 3 months | 0.56 | 0.33, 0.92 | 0.023 |
| Opiate use, including heroin, past 3 months | 0.89 | 0.56, 1.40 | 0.603 |
| Model 4: Full multivariable model | |||
| Multivariable RR | 95% CI | P-value | |
|
|
|||
| Survey time, continuous | 1.02 | 1.01, 1.03 | < 0.001 |
| Homelessness, past 3 months | 0.85 | 0.72, 0.99 | 0.041 |
| CES-D score (≥16 v <16) | 0.88 | 0.79, 0.98 | 0.025 |
| Cigarette smoking, past week | 0.92 | 0.83, 1.03 | 0.143 |
| Condomless insertive anal intercourse, past 3 months | 1.14 | 1.04, 1.24 | 0.005 |
| Methamphetamine or crack use, past 3 months | 0.61 | 0.38, 0.96 | 0.034 |
CES-D, Center for Epidemiologic Studies Depression Scale; CI, confidence interval; GAD-7, Generalized Anxiety Disorder 7 Item Scale; RR, relative risk. All models are adjusted for intervention group. Bold indicates P < 0.05.
Interpretation of relative risk is the change in probability of reporting an undetectable viral load associated with a one-unit change in the score of the psychosocial index.
Compared to participants who had a score of <16 on CES-D, those who had a score of ≥16 were 13% less likely to report a UVL in the model of psychosocial indices. On the 242 surveys in which participants scored ≥16 on CES-D, 78 (32%) reported a detectable viral load. In contrast, participants reported a detectable viral load on 45 of 251 (18%) surveys in which they scored <16 on CES-D.
In the model of sexual and substance use behaviors, participants who reported condomless insertive anal intercourse were 10% more likely to have a UVL compared to those who did not report this sexual practice. Those who smoked were 13% less likely to a UVL while those who used methamphetamine or crack were 44% less likely a UVL.
Multivariable models: full model
In the final, full multivariable model, each successive study visit was associated with a 2% increase in report of viral suppression (Table 5). Participants who reported being homeless in the prior 3 months were 15% less likely to report a UVL than those who were housed. Compared with a score of <16, a score of ≥16 on CES-D was associated with a 12% lower probability of a self-reported UVL. Compared to those who did not report condomless insertive anal intercourse, those who did were 14% more likely to report a UVL. Those who used methamphetamine or crack in the prior 3 months were 39% less likely to report a UVL than those who did not use these drugs. Smoking was not statistically significant in the final model.
Discussion
In contrast to prior reports,2,3 rates of viral suppression in the HMP sample were robust. Fifty percent (102/199) of HIV-positive participants were recruited from HIV clinics and community organizations serving HIV-positive individuals which may explain the high rates of viral suppression in our population. Our findings are not inconsistent, however, with trends in viral suppression among black MSM sampled in the Medical Monitoring Project, where the prevalence of viral suppression increased from 65% to 73% from 2009 to 2013.14
We observed an increase in viral suppression over time. Of the total 497 online assessments, there were 257 (52%) at which we could assess a change in viral suppression, meaning that a participant reported two or more viral load measurements over the 12-month study period. Participants reported a detectable viral load after being undetectable at 13/257 (5%) assessments, no change in viral suppression at 209/257 (81%) assessments (29/209 [14%] remained detectable while 180/209 [86%] remained suppressed), and a change from a detectable viral load to an undetectable one at 35/257 (14%) assessments. The latter change from a detectable viral load to an undetectable one was stable over time, reported by 17/97 (17%), 10/82 (12%), and 8/78 (11%) participants at 3, 6, and 12 months, respectively (P-value for trend=0.101). In addition, participants’ report of health insurance (75–80%), receipt of HIV care (87–95%), and anti-retroviral medication use (82–86%) were high and stable over the study period. In the setting of this stable care measures and transition to a UVL over time, retention of participants who were more likely to report a UVL may provide an alternative explanation to the finding of an increase in viral suppression over time.
In addition to limitations to external generalizability of our observed rates of viral suppression due to our recruitment sources, self-report of viral suppression may have led to misclassification of our outcome of interest. While the accuracy of self-report of a UVL has been estimated to be 88% among MSM, this estimate may be affected by age, housing status, education, employment status, and depressive symptoms.6
At baseline, one-third of participants reported being homeless in the prior 3 months. Participants who reported homelessness were significantly less likely to report a UVL. Prior studies have shown that lack of stable housing is a significant barrier to HIV care, anti-retroviral uptake and adherence, and sustained viral suppression; moreover, providing housing to homeless or unstably housed individuals with HIV improves HIV clinical outcomes.15 In comparison to white youth, black youth are less likely to identify with the stigmatizing term “homeless” and, thus, not utilize, or be reached by, services relevant to homeless youth.16 HMP participants who reported homelessness were significantly more likely to score ≥16 on CES-D (RR=1.71, 95%CI: 1.44, 2.01) and to use methamphetamine or crack (RR=3.07, 95%CI: 1.64, 5.74). Thus, structural interventions to ameliorate housing challenges must also address concurrent mental illness and substance use.
At baseline, 58% of the HIV-positive participants scored ≥16 on CES-D. Scientists have demonstrated that depression is highly prevalent among black MSM17 and that depression is associated with a longer time to viral suppression and a shorter time to virologic failure among HIV-positive individuals initiating anti-retroviral therapy.18 black Americans experience significant disparities in access to mental health care,19 disparities that are likely more acute among HIV-positive YBMSM.
Participants reporting condomless insertive anal intercourse were more likely to have a UVL. In a study of young MSM attending adolescent HIV clinics, black MSM with detectable viral loads were less likely to report condomless anal intercourse than MSM of other races/ethnicities with detectable viral loads.20 HIV-positive HMP participants who reported non-concordant condomless insertive anal intercourse were also more likely to be undetectable, but this association was not statistically significant. Thus, HIV-positive YBMSM may use their knowledge of HIV transmission risk to guide their sexual practices, engaging in condomless insertive anal sex only when undetectable and, thus, not infectious.
Less than 10% of the HIV-positive HMP participants reported methamphetamine or crack use at baseline, but use of psychostimulants was the strongest predictor of a detectable viral load. Stimulant use among HIV-positive people has been associated with poorer adherence to anti-retroviral therapy, virologic failure, more rapid HIV progression, as well as sexual behaviors that risk secondary HIV transmission.21 In the HMP population, >50% reported smoking at baseline. Smokers were more likely to be homeless (RR=2.04, 95%CI: 1.47, 2.81), score ≥16 on CES-D (RR=1.33, 95%CI: 1.10, 1.60), and use methamphetamine or crack (RR=2.94, 95%CI: 1.37, 6.32) than non-smokers. Thus, smoking appears to be a marker of psychosocial instability that interferes with successful viral suppression.22
Attention to intersectional identities in HIV care may improve clinical outcomes among YBMSM. Identification with multiple races and ethnicities and identification as queer, transgender, or other were significantly associated with a detectable viral load in univariable models. In our sample, participants identifying as queer, transgender, or other were more likely to meet CES-D criteria for depression (RR=1.36, 95%CI: 1.17, 1.60) and report homelessness (RR=1.43, 95%CI: 1.05, 1.94). Participants identifying as multiracial or multiethnic were more likely to be depressed (RR=1.43, 95%CI: 1.16, 1.76), use stimulants (RR=3.33, 95%CI: 1.72, 6.44), and report homelessness (RR=1.53; 95%CI: 1.11, 2.13).
Additional limitations deserve discussion. Sixty-two to 81% of participants reported having a viral load measurement in the prior 3 months over the study period. Missing viral load data could be explained by the more recent every-6-month schedule of routine HIV care visits; however, there are likely other factors in our data that may also explain this missingness. Employed participants were more likely to report a viral load in the prior 3 months (RR=1.12, 95%CI: 1.02, 1.23) while homelessness participants (RR=0.86, 95% 0.77, 0.96) and those experiencing greater social isolation (RR=0.89, 95% 0.81, 0.97) were less likely to report a viral load in the prior 3 months. Those with a score of ≥16 on CES-D (RR=0.93, 95% 0.81, 1.01) and those who reported methamphetamine or crack use (RR=0.87, 95%CI: 0.87, 1.04) were somewhat less likely to have had a viral load in the prior 3 months. The factors associated with missing viral load are also associated with our outcome of interest, report of a UVL. If those missing viral load data were actually detectable, the missing viral load information in our data would likely bias our relative risks toward the null. We should also note that the strength of the associations we observed in our data may be particular to HIV-positive, young, black MSM with high levels of care engagement and may not be generalizable to less engaged populations. It is possible that the strength of association between the predictors of having a detectable viral load that we observed (e.g., homelessness, depression, stimulant use) may be larger in a more vulnerable population. Additionally, a larger sample size may have allowed us to observe associations between report of a UVL and other variables in our data that did not reach statistical significance.
In a sample of HIV-positive YBMSM with a high rate of viral suppression, homelessness, depression, and stimulant use were independent predictors of a detectable viral load. Future investigation should be aimed at further elucidating pathways to homelessness for YBMSM and developing services that are accessible and acceptable to homeless black youth. The significant impact of crack and methamphetamine use on viral suppression, the high prevalence of depression among YBMSM, and the co-occurrence of these factors with homelessness call for universal integration of HIV care, mental health services, and substance use treatment. Additionally, HIV services should continue to foster, and capitalize on, young black men’s integration of HIV knowledge and sexual practices to reduce HIV transmission risk. Thus, the health behaviors of HIV-positive YBMSM must be addressed in the context of their intersecting social, economic, political, and sexual landscapes.23–25 Only interventions that address poverty, racism, HIV stigma, and homophobia at the governmental, community, and familial levels will ameliorate challenges to individual-level physical and emotional health in order to produce a durable state of health equity for YBMSM living with HIV.
Sources of support:
NIMH R01MH093275-01 to LHW.
References
- 1.Millett GA, Peterson JL, Flores SA, et al. Comparisons of disparities and risks of HIV infection in black and other men who have sex with men in Canada, UK, and USA: a meta-analysis. Lancet Lond Engl 2012;380(9839):341–8. [DOI] [PubMed] [Google Scholar]
- 2.Rosenberg ES, Millett GA, Sullivan PS, Del Rio C, Curran JW. Understanding the HIV disparities between black and white men who have sex with men in the USA using the HIV care continuum: a modeling study. Lancet HIV 2014;1(3):e112–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Kapogiannis B, Xu J, Mayer KH, et al. The HIV Continuum of Care for Adolescents and Young Adults (12–24 years) Attending 13 Urban US Centers of the NICHD-ATN-CDC-HRSA SMILE Collaborative. Vancouver, Canada: 2015. [DOI] [PMC free article] [PubMed]
- 4.Hightow-Weidman LB, Muessig KE, Pike EC, et al. HealthMpowerment.org: Building Community Through a Mobile-Optimized, Online Health Promotion Intervention. Health Educ Behav Off Publ Soc Public Health Educ 2015;42(4):493–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Hightow-Weidman LB, Fowler B, Kibe J, et al. HealthMpowerment.org: development of a theory-based HIV/STI website for young black MSM. AIDS Educ Prev Off Publ Int Soc AIDS Educ 2011;23(1):1–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Sewell J, Daskalopoulou M, Nakagawa F, et al. Accuracy of self-report of HIV viral load among people with HIV on antiretroviral treatment. HIV Med 2016; [DOI] [PMC free article] [PubMed]
- 7.Weissman MM, Sholomskas D, Pottenger M, Prusoff BA, Locke BZ. Assessing depressive symptoms in five psychiatric populations: a validation study. Am J Epidemiol 1977;106(3):203–14. [DOI] [PubMed] [Google Scholar]
- 8.Spitzer RL, Kroenke K, Williams JBW, Löwe B. A brief measure for assessing generalized anxiety disorder: the GAD-7. Arch Intern Med 2006;166(10):1092–7. [DOI] [PubMed] [Google Scholar]
- 9.Stewart A, Ware JE. Measuring Functioning and Well-Being [Internet]. 1992. [cited 2017 Jan 31];Available from: http://www.rand.org/pubs/commercial_books/CB361.html
- 10.Lubben J, Blozik E, Gillmann G, et al. Performance of an abbreviated version of the Lubben Social Network Scale among three European community-dwelling older adult populations. The Gerontologist 2006;46(4):503–13. [DOI] [PubMed] [Google Scholar]
- 11.Steward WT, Herek GM, Ramakrishna J, et al. HIV-related stigma: adapting a theoretical framework for use in India. Soc Sci Med 1982 2008;67(8):1225–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Bogart LM, Landrine H, Galvan FH, Wagner GJ, Klein DJ. Perceived discrimination and physical health among HIV-positive Black and Latino men who have sex with men. AIDS Behav 2013;17(4):1431–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Cumming P. Methods for estimating adjusted risk ratios. Stata J 2009;9(2):175–96. [Google Scholar]
- 14.Beer L, Bradley H, Mattson CL, Johnson CH, Hoots B, Shouse RL. Trends in Racial and Ethnic Disparities in Antiretroviral Therapy Prescription and Viral Suppression in the United States, 2009–2013: JAIDS J Acquir Immune Defic Syndr 2016;73(4):446–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Aidala AA, Wilson MG, Shubert V, et al. Housing Status, Medical Care, and Health Outcomes Among People Living With HIV/AIDS: A Systematic Review. Am J Public Health 2016;106(1):e1–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Hickler B, Auerswald CL. The worlds of homeless white and African American youth in San Francisco, California: a cultural epidemiological comparison. Soc Sci Med 1982 2009;68(5):824–31. [DOI] [PubMed] [Google Scholar]
- 17.Williams JK, Wilton L, Magnus M, et al. Relation of Childhood Sexual Abuse, Intimate Partner Violence, and Depression to Risk Factors for HIV Among Black Men Who Have Sex With Men in 6 US Cities. Am J Public Health 2015;105(12):2473–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Pence BW, Miller WC, Gaynes BN, Eron JJ. Psychiatric illness and virologic response in patients initiating highly active antiretroviral therapy. J Acquir Immune Defic Syndr 1999 2007;44(2):159–66. [DOI] [PubMed] [Google Scholar]
- 19.Cook BL, Trinh N-H, Li Z, Hou SS-Y, Progovac AM. Trends in Racial-Ethnic Disparities in Access to Mental Health Care, 2004–2012. Psychiatr Serv Wash DC 2017;68(1):9–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Wilson PA, Kahana SY, Fernandez MI, et al. Sexual Risk Behavior Among Virologically Detectable Human Immunodeficiency Virus-Infected Young Men Who Have Sex With Men. JAMA Pediatr 2016;170(2):125–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Carrico AW. Substance use and HIV disease progression in the HAART era: Implications for the primary prevention of HIV. Life Sci 2011;88(21–22):940–7. [DOI] [PubMed] [Google Scholar]
- 22.Cropsey KL, Willig JH, Mugavero MJ, et al. Cigarette Smokers are Less Likely to Have Undetectable Viral Loads: Results From Four HIV Clinics. J Addict Med 2016;10(1):13–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Baral S, Logie CH, Grosso A, Wirtz AL, Beyrer C. Modified social ecological model: a tool to guide the assessment of the risks and risk contexts of HIV epidemics. BMC Public Health 2013;13:482. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Bowen EA. A Multilevel Ecological Model of HIV Risk for People Who Are Homeless or Unstably Housed and Who Use Drugs in the Urban United States. Soc Work Public Health 2016;31(4):264–75. [DOI] [PubMed] [Google Scholar]
- 25.Stevens R, Bernadini S, Jemmott JB. Social environment and sexual risk-taking among gay and transgender African American youth. Cult Health Sex 2013;15(10):1148–61. [DOI] [PMC free article] [PubMed] [Google Scholar]