Mental health, social influences, and HIV pre-exposure prophylaxis (PrEP) utilization among men and transgender individuals screening for HIV prevention trials

Abstract

The effects of mental health comorbidities and social support on the HIV pre-exposure prophylaxis (PrEP) care continuum are unknown. We conducted a cross-sectional study of men and transgender individuals, ≥18 years-old, with ≥2 male or transgender partners, or recent condomless anal intercourse. Surveys assessed demographics, mental health treatment, depressive symptomatology, social support, and PrEP-related social contacts. Logistic regression assessed associations between these factors and PrEP uptake and persistence. Participants (n=247) were 89% cis-male and 46% African-American. Median age was 27 (IQR:23–33). Thirty-seven percent had ever used PrEP, of whom 18% discontinued use. High depressive symptomology was identified in 11% and 9% were receiving mental health treatment. There were no significant associations between depressive symptoms or mental health treatment on the odds of PrEP uptake or discontinuation. Each additional PrEP contact conferred a greater odds of uptake (aOR:1.24, 95% CI: 1.09–1.42). Network-level targets may produce fruitful interventions to increase PrEP uptake.

Introduction

Despite increasing rates of awareness and willingness to use HIV pre-exposure prophylaxis (PrEP) among men and transgender individuals who have sex with men, uptake and sustained utilization of PrEP have remained a challenge in these key populations. The PrEP care continuum in the United States,^1–3 encompassing the stages of PrEP awareness, willingness to take PrEP, uptake, adherence, and persistence, has been characterized by substantial drop off at each stage, and striking racial and age-based disparities. Despite disproportionally high rates of HIV acquisition, African American and Latinx individuals, and adolescents and young adults have had substantial lower rates of PrEP utilization.^4,5 PrEP discontinuation has also been common, as demonstrated in a recent analysis by Hojilla et al documenting a 38% rate of discontinuation at 13 months among men who have sex with men (MSM) at a San Francisco sexual health clinic.⁶ Thus, while the individual-level efficacy of PrEP in preventing HIV has been clearly demonstrated in clinical trials, the population effectiveness of PrEP in reducing the incidence of HIV appears to be hindered by barriers to engagement and retention in PrEP care.

While addressing race- and age-based disparities in PrEP utilization is an essential task toward ameliorating drop off at the stages of the continuum, it is critically important to also identify modifiable barriers to engagement in PrEP care among race and age-diverse populations at high risk of HIV infection.⁷ Mental health, particularly depression, is a key factor to explore in its relationship to PrEP. Among people living with HIV, depression is a well-established risk factor for non-engagement in care and antiretroviral therapy (ART) non-adherence.⁸ Depression has also been associated with increased sexual risk behavior, including condomless anal intercourse, suggesting it may perpetuate HIV transmission risk.^9–13 If depression impacts PrEP utilization, interventions to synergistically improve mental health and access to PrEP clinical services may increase PrEP utilization.¹⁴ From a social ecologic perspective, these interventions could occur at the health systems (e.g. integrated mental health and HIV prevention clinics), individual (e.g. life skills training to address mental health and HIV risk) or social network levels. From a social network perspective, intervening to enhance social support and foster PrEP-related social norms may attenuate the potential impact of depression and improve PrEP uptake. In a recent study of ART-adherence among people living with HIV, social network influences significantly decreased the magnitude of association between depression and ART adherence.¹⁵

Our primary objective was to assess the proportion of individuals at sequential stages of the PrEP care continuum in an at-risk cohort of men who have sex with men and transgender individuals (MSM-TG) screening for HIV prevention trial participation. Secondarily, we aimed to assess associations between depressive symptoms, mental health treatment, and social network influences on PrEP uptake and subsequent discontinuation.

Methods

Study Design, Participants, and Setting:

This cross-sectional study of MSM and TG individuals pre-screening for HIV Vaccine Trials Network (HVTN) and HIV Prevention Trials Network (HPTN) clinical trial participation was conducted in Philadelphia from July 2016 to May 2018. The HVTN 704/HPTN 085, and HPTN 083 trials aim to evaluate the safety and efficacy of the VRC01 broadly neutralizing monoclonal antibody and compare oral emtricitabine/tenofovir disoproxil fumarate versus injectable cabotegravir, respectively, in preventing acquisition of HIV infection. Prior to screening for the clinical trials, potential participants completed a pre-screening instrument measuring sexual risk behaviors, depression, social support, network PrEP awareness, and additional domains described below. Eligible participants were aged ≥18 years, born male or identifying as transgender, with a history in the preceding six months of anal intercourse with ≥2 male or transgender partners, or any condomless anal receptive intercourse with male or transgender partners. Participant recruitment occurred via a variety of methods including face-to-face at venues such as bars, clubs, social service organizations, and community events and social media-based recruitment through advertisements on websites and geosocial applications used to share information, inquire about services, or meet sexual partners.

Measures:

All participants completed survey measures via paper-and-pencil or on a secure tablet computer. Demographic measures included self-reported race, ethnicity, sex, gender, and age. Depressive symptomatology was assessed via the Patient Health Questionnaire-8 (PHQ-8) survey. The parent instrument, the PHQ-9, is a short-form self-administered screening tool for major depressive disorder with excellent internal reliability and validity in diagnosing major depression.¹⁶ For this study, we used the abbreviated PHQ-8 measure which omits the final question regarding active suicidality.¹⁷ A PHQ-8 score ≥10 has 88% sensitivity and 88% specificity for major depression, which was the score threshold for our high depressive symptoms definition.¹⁶ Mental health treatment was assessed by the question “Are you under the care of a health care provider for any mental health issues?” The PrEP-related social contacts measure was defined as the number of individuals with whom participants discussed PrEP in the seven days prior to survey administration. Social support was measured with the Multidimensional Scale of Perceived Social Support (MSPSS), a 12-item questionnaire which measures perceived social support from significant others, friends, and family.¹⁸

The survey contained a statement that PrEP was a daily pill to prevent HIV. Awareness of PrEP was assessed by the question: “Before considering to participate in this trial, have you ever heard of PrEP?” Willingness was assessed by the question: “Would you consider taking a pill every day (PrEP) to prevent becoming infected with HIV?” Uptake was assessed by the question: “Have you ever taken PrEP?” Discontinuation of PrEP was defined as answering “yes” to the PrEP uptake question, and “no” to questions regarding PrEP use during any sexual relationships with male or transgender partners in the past six months.

Statistical analysis:

The proportion of individuals at the various stages of the continuum were reported as proportions of the total study sample. Depressive symptomatology was defined by the continuous score on the PHQ-8 in order to capture the effect of increasing levels of depressive symptomatology on PrEP-related outcomes. As there are no established population norms for the MSPSS,¹⁹ we examined the distribution of scores in the sample to set a threshold value for low versus not low social support. As scores were highly left skewed toward moderate to high support, we dichotomized scores with a cut-off value for low social support at the 25th percentile value (3.4) consistent with responses of very strongly disagree, strongly disagree, mildly disagree or neutral to the survey’s social support statements. We performed bivariate analyses with chi-square and Fisher’s exact testing for categorical variables and Wilcoxon rank-sum testing for continuous variables to assess for associations between demographic and psychosocial variables between those who had initiated PrEP versus not, and those who had discontinued PrEP versus not.

To assess associations between mental health and social network factors and the outcomes of PrEP uptake and, separately, discontinuation we performed multivariable logistic regression adjusting for age, race, and gender as potential confounders based on prior literature demonstrating associations between these demographic covariates and both depression and PrEP utilization.^5,20 In order to first determine whether perceived social support and PrEP-related social contacts modified the hypothesized effect of depression on PrEP utilization, we tested separate interaction terms for these variables. As neither point estimates nor p values of the interaction terms were suggestive of outcome difference association between depression and PrEP utilization by perceived social support or PrEP-related social contacts, MSPSS score and the number of PrEP social contacts were included as covariates in the final multivariate models. All multivariable models with missing data were performed via complete case analyses.

Power and Sample Size:

Our sample size was driven by the goal of detecting a 20% difference in PrEP uptake between participants with high depressive symptoms (PHQ-8 score ≥10) vs. without high depressive symptoms with 80% power and an alpha of 0.05. Given prior literature demonstrating a depression rate in MSM of approximately 30%²¹ and a PrEP utilization rate early in the present sample of approximately 30%, we targeted a sample size of n=244 to reach this goal.

The research was approved by the University of Pennsylvania Institutional Review Board.

Results

Participants (n=247) were predominately cisgender males and racial and ethnic minorities (Table 1). The median PHQ-8 score was 3 (Interquartile range [IQR]: 0–6) and 11% of the study population had a score ≥10, suggesting clinically significant depressive symptoms. Nine percent were currently under the care of a mental health provider, of whom 37% were being treated for depression. Two participants were excluded from the analysis as they were missing data for all PrEP care continuum measures.

Table 1:

Characteristics of the Study Sample by PrEP Utilization History (n=247)

Participant Characteristic	Overall Sample	PrEP Use n=92 (37%)	No PrEP Use n=154 (62%)1	p-Value
Age Category
18–24	83 (34%)	29 (32%)	53 (34%)	0.90
25–34	115 (47%)	43 (47%)	72 (47%)
35+	49 (20%)	20 (22%)	29 (19%)
Race 2
White	96 (39%)	32 (35%)	64 (42%)	0.29
African American	114 (46%)	48 (52%)	65 (42%)	0.13
Asian/Pacific Islander	10 (4%)	5 (5%)	5 (3%)	0.40
Native American	7 (3%)	5 (5%)	2 (1%)	0.06
Other	36 (15%)	12 (13%)	24 (16%)	0.59
Latinx Ethnicity 3	46 (19%)	15 (16%)	31 (20%)	0.44
Gender
Cisgender male	217 (89%)	78 (85%)	139 (90%)	0.09
Transgender female	19 (8%)	10 (11%)	9 (6%)
Transgender male	4 (2%)	0 (0%)	4 (3%)
Other gender	6 (2%)	4 (4%)	2 (1%)
High depressive symptoms 42	28 (11%)	13 (14%)	15 (10%)	0.27
Receiving mental health care 5	23 (9%)	12 (13%)	11 (7%)	0.11
PrEP Social Contacts (Median/IQR) 6	1 (IQR 0–2)	2 (0.5–3)	1 (0–2)	<0.001
Low Social Support 7	58 (23%)	24 (26%)	33 (21%)	0.45

¹n=1 (0.4%) missing data

²Race does not equal 100% because participants could choose from multiple race categories.

³n=4 (1.6%) missing data

⁴PHQ-8 ≥10; n=5 (2%) missing data

⁵n=22 missing data (9%)

⁶n=11 (4.5%) missing data

⁷n=3 (1.2%) missing data

The proportions of participants along the stages of the PrEP care continuum at their HIV prevention clinical trial prescreening visit are displayed in Figure 1. The vast majority of the cohort was aware of PrEP (93%) and willing to take it (92%). However, despite a high willingness to take PrEP, only 37% of participants had ever taken PrEP, and of these, 18% had not used PrEP in the past six months despite ongoing sexual activity. For those unwilling to take PrEP (n=19), the most common reasons were not wanting to take daily medication (6%), concern for side effects (2%), and concern about the expense of PrEP (2%).

Figure 1:

HIV Pre Exposure Prophylaxis Continuum in the Study Population (n=247)

In the bivariate analyses comparing participants who had and had not initiated PrEP (Table 1), we found no associations between age, race, depressive symptoms, mental health treatment, social support, and PrEP uptake. Having a higher number of social contacts with whom the individual had discussed PrEP was associated with PrEP uptake. In the adjusted multivariable model (Table 2) there was no significant association between depressive symptoms and PrEP use. However, each additional social contact with whom participants discussed PrEP conferred a 24% greater odds of having ever used PrEP. Transgender women had a statistically significant nearly four-fold higher odds of PrEP utilization compared with other genders.

Table 2:

Multivariable Regression Assessing Mental Health, Social Influences, and PrEP Uptake (n=202)

Covariate	Adjusted Odds Ratio	95% Confidence Interval	p-Value
Depressive Symptom Score (PHQ-8)1	1.04	0.97–1.12	0.27
Receiving mental health care	2.05	0.75–5.58	0.16
PrEP social contacts in the last 7 days 2,3	1.24	1.09–1.42	<0.01
Social Support
Not low social support	Reference
Low social support4	1.35	0.64–2.84	0.43
Gender
Cisgender males, transgender males, and others	Reference
Transgender female	4.01	1.20–13.40	0.02
Age
18–24	Reference
25–34	1.26	0.62–2.55	0.52
35+	1.01	0.39–2.61	0.98
African-American race	1.63	0.87–3.06	0.12
Latinx ethnicity	0.54	0.23–1.30	0.18

¹Per each 1 unit change in score

²Per each additional social contact

³Bolded values indicate statistical significance.

⁴Upper tertile score on Multidimensional Scale of Perceived Social Support

The bivariate analysis comparing those participants who had discontinued PrEP compared to those were still using PrEP at the time of survey completion is displayed in Table 3. In the multivariable model examining PrEP discontinuation (Table 4), there was no significant association between any characteristic assessed and PrEP discontinuation.

Table 3:

Characteristics of Participants with History of PrEP Utilization by Current PrEP Status (n=92)

Participant Characteristic	Overall Sample (n=92)	Discontinued PrEP Use (n=17)	Currently using PrEP (n=75)	p-Value
Age Category
18–24	29 (32%)	6 (35%)	23 (31%)	0.88
25–34	43 (47%)	8 (47%)	35 (47%)
35+	20 (22%)	3 (18%)	17 (23%)
Race 1
White	32 (35%)	8 (47%)	24 (32%)	0.24
African American	48 (52%)	8 (47%)	40 (53%)	0.64
Asian/Pacific Islander	5 (5%)	2 (12%)	3 (4%)	0.20
Native American	5 (5%)	0 (0%)	5 (7%)	0.27
Other	12 (13%)	1 (6%)	11 (15%)	0.33
Latinx Ethnicity	15 (16%)	2 (12%)	13 (17%)	0.56
Gender
Cisgender male	78 (85%)	14 (82%)	64 (85%)	0.93
Transgender female	10 (11%)	2 (12%)	8 (11%)
Transgender male	0 (0%)	0 (0%)	0 (0%)
Other gender	4 (4%)	1 (6%)	3 (4%)
High depressive symptoms Depression (PHQ-8≥10) 2	13 (15%)	3 (18%)	10 (13%)	0.69
Receiving mental health care	12 (13%)	2 (12%)	10 (13%)	0.80
PrEP Social Contacts (Median/IQR) 3	1 (IQR 0–2)	1 (0–3)	2 (1–4)	0.37
Low Social Support 4	24 (26%)	6 (35%)	18 (24%)	0.34

¹Race does not equal 100% because participants could choose from multiple race categories.

²n=5 (0.02%) missing data

³n=19 (0.07%) missing data

⁴n=3 (0.01%) missing data

Table 4:

Multivariable Regression Assessing Mental Health, Social Influences, and PrEP Uptake (n=74)

Covariate	Adjusted Odds Ratio	95% Confidence Interval	p-value
Depressive Symptom Score (PHQ-8)1	1.02	0.88–1.18	0.81
Receiving mental health care	0.62	0.10–3.93	0.61
Number of PrEP social contacts in the last 7 days2	0.99	0.84–1.16	0.89
Social Support
Not low social support	Reference
Low Social Support3	1.64	0.34–7.90	0.54
Gender
Cisgender males, transgender males, and other	Reference
Transgender female	0.96	0.19–5.87	0.96
Age
18–24	Reference
25–34	0.97	0.22–4.17	0.96
35+	1.00	0.16–6.17	1.00
African-American race	0.50	0.13–1.90	0.31
Latinx ethnicity	0.67	0.11–3.94	0.65

¹Per each 1 unit change in score

²Per each additional social contact

³Score dichotomized at ≥10 vs. not on Multidimensional Scale of Perceived Social Support

Discussion

In this racially and ethnically-diverse urban sample of MSM-TG, screening for clinical trials related to HIV prevention, only 34% had previously initiated PrEP and 30% had continued PrEP use over the past six months in the context of ongoing self-reported sexual activity that placed them at risk for HIV. While we identified a substantial drop off in the PrEP continuum between the states of willingness to take PrEP and initiating PrEP, the 34% rate of utilization is higher than has been noted in many recent studies of these key populations.^22–27 In the “One Thousand Strong” national cohort study of gay and bisexual men, only 4% of PrEP-eligible individuals had been prescribed PrEP.² Moreover, our data underscore the fact that once PrEP is prescribed, additional barriers and facilitators for progression along the continuum need to be addressed to improve the gap between PrEP eligibility and, ultimately, sustained PrEP utilization during periods of HIV risk.

We found a dose response relationship between having recently discussed PrEP with others and ever using PrEP, which suggests that social network norms may influence utilization. Our findings add to a growing body of literature demonstrating the impact of social networks on engagement along the PrEP care continuum. The “uConnect” online study of young black MSM demonstrated that PrEP awareness over the course of the study was associated with having a higher number of social media (i.e., Facebook) contacts who were PrEP aware.²⁸ In Mehotra et al’s recent study of social relationships and PrEP uptake and adherence, disclosure of sexual orientation to social support network members was associated with both increased PrEP uptake and adherence.²⁹ These data suggest that intervening at the level of the social network and altering social norms regarding PrEP use may translate into improved outcomes along the continuum.

We did not identify a significant association between depressive symptoms or current mental health treatment on PrEP uptake or discontinuation. There are few studies examining the impact of mental health conditions on PrEP utilization, and these have often yielded conflicting results. In the iPrEX trial, half of participants had depressive symptom scores consistent with clinical depression at least once during the trial, and higher scores were associated with increased condomless receptive anal intercourse.³⁰ Yet, depression had only a modest association with non-adherence to PrEP.²¹ In one report conducted in a real-world setting, suboptimal PrEP adherence was associated with anxiety and childhood trauma, there was no association between depression and adherence.³¹ Interestingly, the prevalence of high depressive symptoms and mental health utilization in our sample was lower than in prior studies of depression in men and trans women who have sex with men.^12,13 It is possible that our study sample, all of whom were screening for clinical trial participation, may have minimized symptoms due to a misperception that depression could be a study exclusion criterion. Alternatively, a mentally healthier subpopulation of individuals considered participation. This may have led to misclassification of participants, and our study being underpowered to detect an association between mental health and PrEP utilization. Further longitudinal research is needed in large community-based samples to investigate the time-varying relationship between mental health co-morbidities, mental health care utilization, and PrEP continuum outcomes. These future data may help determine the utility of interventions to synergistically target both mental health and PrEP use, and the potential for co-locating PrEP and mental health services.

An unexpected finding in our study was the higher odds of PrEP uptake in transgender women compared to cisgender and transgender men. Transgender women are a population disproportionately affected by HIV.^32,33 Compared to cisgender women of reproductive age, the odds of HIV acquisition are 49 times higher among transgender women.³⁴ Transgender women face higher rates of homelessness, unemployment, depression and victimization than cisgender men, all of which could act as barriers to receiving HIV prevention care including PrEP.^35–37 Early data in the post-PrEP-licensure period demonstrated low rates of PrEP awareness and uptake in young transgender women.^38,39 Our data provide a signal that these statistics may be changing, and efforts to improve PrEP utilization in transgender women may be slowly making headway.

Our study has both limitations and strengths. The cross-sectional study design precluded establishing a causal relationship between depression, social network factors and PrEP utilization. As the timing of PrEP initiation with respect to onset of depressive symptoms or mental health treatment about PrEP were unknown in our data, we could not determine which factor may have influenced the other. Although our data represent a single city, which may limit generalizability, it provides a population-level snapshot of the current PrEP care continuum in a highly diverse urban cohort. Thus, the PrEP care continuum presented here may be more generalizable to urban populations of MSM and transgender individuals than health system or clinic-level data which include patients who may have a more uniform degree of access to health care. Lastly, as the timing of PrEP start and stop dates were unknown in the sample and PrEP usage was by self-report, we cannot provide estimates for the length of time on PrEP for participants prior to discontinuation. However, our PrEP discontinuation definition required participants to be sexually active and at risk for HIV acquisition in the six-month period prior to enrollment during which discontinuation was assessed. Few previous studies have assessed PrEP persistence in the context of ongoing vulnerability to HIV. As there is little to no benefit to PrEP persistence in the absence of HIV risk,⁴⁰ incorporating time-varying sexual activity into definitions of PrEP adherence and persistence can strengthen the accuracy of PrEP care continuum metrics.

In conclusion, we identified that over one-third of men and transgender individuals in a racially and ethnically diverse urban sample had ever utilized PrEP and that 75% of these individuals had persistent PrEP use while vulnerable to HIV infection in the prior six months. The odds of PrEP utilization in this sample increased with increasing numbers of social contacts with whom the participant had discussed PrEP in the past week. Future research is needed to better delineate how social network structure, function, and characteristics may be harnessed to improve PrEP utilization.