Introduction

Efforts to end the HIV epidemic will be stymied without efforts to address the needs of adolescents and young adults (AYA). Globally, there are 1.7 million children and adolescents < 15 years-old [1] and over 5 million AYA ages 15–24 years living with HIV [2]. Several biomedical advances have significantly altered the landscape for HIV prevention and treatment: (1) pre-exposure prophylaxis (PrEP), the use of antiretroviral drugs to prevent HIV acquisition [3] and (2) evidence that people living with HIV with a sustained undetectable HIV RNA viral load (VL < 200 copies/mL) do not transmit the virus to a sex partner, leading to the Undetectable = Untransmittable (U = U) public health campaign [4]. Promoting adherence to antiretroviral treatment (ART) among those living with HIV to achieve viral suppression, and to PrEP to reduce the risk of acquisition among those vulnerable to acquiring HIV, particularly in this age group, has been challenging. AYA often have the most difficulty with adherence across chronic health conditions including HIV, which can be related to patient, ART regimen, healthcare, and cultural/societal-related barriers [5,6,7]. Adolescence and young adulthood is also a stage with dramatic cognitive, physical, sexual, social, and emotional changes that can leave AYA prone to sensation-seeking or risk-taking behaviors that may capture their attention more than pill-taking or other HIV prevention activities [8,9,10].

AYA with perinatally-acquired HIV (AYAPHIV) are a population found to be vulnerable to a host of biomedical, psychosocial, neurocognitive, and psychiatric challenges, as well as suboptimal adherence to ART and sexual disease prevention behaviors [11,12,13,14,15,16]. Though improved access to ART and the introduction of safer, more tolerable regimens have allowed AYA living with HIV to survive into adulthood, this population continues to experience worse health outcomes relative to peers not living with HIV [8].

Also of concern are AYA who were perinatally HIV-exposed but uninfected (AYAPHEU), whose experiences with familial HIV and growing up in similarly vulnerable families and communities as AYAPHIV expose them to biogenetic and psychosocial risks (e.g., prenatal exposures, neurological deficits, low self-esteem, stressful life events, and caregiver HIV, substance use, and mental health challenges). Sociodemographic characteristics of AYAPHIV and AYAPHEU in the US are also similar, as both groups are predominantly Black and Latinx and exposed to poverty, housing insecurity, violence, racism, and health care disparities [11, 17]. Globally, AYAPHEU live in communities with high HIV seroprevalence rates, potentially placing them at risk for acquisition of HIV [18]. Although many AYAPHIV and AYAPHEU have demonstrated resilience in the face of numerous stressors, with positive psychosocial outcomes [11–[12, 19]–20], one-third of a longitudinal cohort of AYAPHIV and AYAPHEU in New York City was identified as having consistent anxiety or escalating trajectories of multiple psychiatric disorders, indicating that both groups are vulnerable to psychiatric disorder [16]. Given the growing population of AYAPHIV and AYAPHEU worldwide, understanding the relationship between psychiatric disorder and behaviors associated with HIV viremia (defined in this study as the level of HIV viral load that is detected by a threshold of > 400 copies/mL of blood) or HIV-transmission is critical.

Similar to other AYA, AYAPHIV and AYAPHEU are engaging in sexual risk behaviors, including condomless sex [21,22,23], with one longitudinal study finding that as both groups age, their likelihood of engaging in unprotected sex across follow-up visits more than doubled [24]. In the US specifically, multiple studies suggest that AYAPHIV are at high risk for HIV viremia [12, 14, 25]. Additionally, it is not clear in the literature to what extent US youth in families affected by maternal HIV, including AYAPHEU, are accessing HIV-testing and PrEP. There is evidence in other AYA populations that adherence to PrEP is a challenge, suggesting that condom use remains an important prevention tool [26]. Thus, even with PrEP and U = U campaigns, interventions may need to continue to focus on sexual risk reduction, including condom use, to play a crucial role in HIV prevention efforts.

Using baseline data and data from multiple follow-up interviews with a longitudinal cohort of AYAPHIV and AYAPHEU, this study examined: (a) the prevalence of condomless sex over time from adolescence through young adulthood; (b) the association between condomless sex and PHIV-status (AYPHIV vs. AYAPHEU) and, among AYAPHIV only, viral load; and (c) the association of condomless sex with sociodemographic factors (gender, race, ethnicity) and trajectories of psychiatric functioning across time.

Methods

The Child and Adolescent Self-Awareness and Health Study (CASAH) is an ongoing longitudinal study examining behavioral health, psychosocial functioning, and health outcomes in a cohort (N = 340) of youth with (a) perinatally-acquired HIV (N = 206) or (b) perinatal HIV-exposure, but who are not living with HIV (N = 134). Participants were recruited from four urban medical centers in New York City between 2003 and 2008. Inclusion criteria at baseline were: (1) youths ages 9–16 years with perinatal HIV-exposure; (2) cognitive capacity to complete an extended psychosocial interview; (3) English- or Spanish-speaking; and (4) caregiver with the legal ability to sign consent for youth participation. Providers identified eligible patients in their clinics and referred interested caregivers and youth to the study team. Caregivers and youth provided consent/assent (caregivers for youth < 18 years; youth assent < 18 years and consent ≥ 18 years).

CASAH data are collected approximately every 12–18 months through (1) structured, research assistant-administered quantitative interviews; (2) audio computer-assisted self-interviews (ACASI); and (3) medical chart reviews. Although on average participants completed interviews every 12–18 months, this did not always happen due to multiple factors. CASAH was not originally conceived as a longitudinal study and instead began with five years of data collection for two interviews—baseline and follow-up 1. We received additional funding at several different time points for additional interviews (CASAH 2: 2008–2013, follow-ups 2–4; CASAH 3: 2013–2018, follow-ups 5–7). Moreover, participants needed to be 13-years-old at the start of CASAH 2 and 18-years-old at the start of CASAH 3, given the study’s aims. As a result, there are larger gaps in time between some interviews for participants. The data for these analyses were taken from participants’ baseline visit (ages 9–16) through their sixth follow-up (ages 18–29). This study was approved by the Columbia University-New York State Psychiatric Institute Institutional Review Board.

Measures

Sociodemographic and PHIV-status factors

Youth reported on the following sociodemographic variables: gender (Male vs. Female), race (African-American/Black vs. non-Black), ethnicity (Latinx vs. Non-Latinx), and PHIV-status (PHIV vs. PHEU).

Condomless sex

The ACASI was used to measure recent (past three months) condomless sex at each time point. ACASI is a data collection software that allows participants to listen to pre-recorded questions through headphones and to enter their responses on a touch screen or keypad. Using the ACASI, participants were asked how many times in the past three months they engaged in either anal or vaginal sex without a condom. For every participant who answered this question at one or more time points from baseline-follow-up 6 (N = 335), a variable was calculated by looking at the proportion of total visits with recent condomless sex. In this proportion variable, the numerator represents a given participant’s total number of visits with recent condomless sex, and the denominator represents the total number of visits for which the participant answered the recent condomless sex question.

Questions concerning PrEP use and awareness

Following its authorization for use among adults by the US Food and Drug Administration in 2012, we asked AYAPHEU about PrEP use beginning at follow-up 6 (2015–2018). During this visit, the youth completed a questionnaire that measured their use and knowledge of PrEP. All the participants were asked if they had ever heard of PrEP, and PHEU participants were asked if they had ever taken any type of medication to protect themselves from acquiring HIV infection. Both questions are dichotomous yes/no variables.

HIV Biologic Measures (AYAPHIV only)

Data on CD4 + T-lymphocyte counts (cells/uL) and plasma HIV RNA viral load (copies/mL), as close to the date of the interview as possible, were obtained from medical records. Due to changing values for “undetectable” viral load throughout the history of CASAH and to the unavailability of ultrasensitive HIV-1 RNA assay during the first few study interviews, we dichotomized viral load as ≤ 400 and > 400 copies/mL.

Psychiatric Disorders

Psychiatric disorder diagnoses were assessed using adolescent and young adult versions (depending on a participant’s age at follow-up) of The Diagnostic Interview Schedule for Children, 4th Edition (DISC-IV). The DISC-IV [27] is a well-validated, structured instrument that asks participants about symptoms experienced in the past year and throughout their lifetime for common psychiatric disorders as defined by the American Psychiatric Association’s Diagnostic and Statistical Manual of Mental Disorders (DSM) [28]. The adolescent and young adult versions of the DISC-IV are similar, and our study’s variables were the same for both versions. Psychiatric trajectories over time were adopted from a prior study, in which Nguyen et al. [16] used multivariate latent trajectory analysis to identify three longitudinal patterns of psychiatric disorders based on the co-occurrence of mood, behavioral, anxiety, and substance use disorders from enrollment through follow-up 5, spanning a median of nine years. The final 3-class model of trajectories was supported by model fit indices and interpretability (BIC = 3606, 3579, 3569, 3596 for 2-class, 3-class, 4-class, and 5-class models, respectively). The three trajectories included: (1) relatively low probability of psychiatric disorder, i.e., mood or behavioral disorders were low, decreasing, or not reported across all visits, with very few participants meeting criteria for more than one disorder at any time point; (2) consistent probability of anxiety disorder, i.e., a near-50% probability of participants endorsing anxiety disorders throughout all visits; and (3) persistent and escalating probability of psychiatric comorbidity, i.e., substantial probability of comorbidity between mood, behavioral, substance use, and anxiety-related disorders across all visits. Each participant was placed into one of these categories, and the resulting psychiatric trajectory variable was used in analyses for this paper.

Statistical Analysis

First, the number of follow-up visits with condomless sex was examined in the overall population by looking at the proportion of AYA reporting condomless sex at one or more visits, two or more visits, or half or more visits. The majority of participants (86.1%) contributed data for at least two interviews, with 54% completing all study visits. Only two participants in the CASAH cohort completed just a baseline interview; 13.9% of the sample completed only a baseline and follow-up 1 visit. Using linear regression we examined differences in the proportion of visits with any recent condomless sex by PHIV-status, gender, race, ethnicity, and psychiatric disorder trajectory. Each of the demographic variables or key covariates of interest (i.e., PHIV status, gender, race, ethnicity, and psychiatric disorder trajectory) was entered into the model separately (i.e., one in each model) rather than all in one model simultaneously. Among AYAPHIV, we used generalized estimating equations (GEE) with independent working correlation and robust estimate of standard error to assess the relationship between any condomless sex and unsuppressed viral load from follow-up 2 to follow-up 6. All models controlled for age. For descriptive purposes, we also included in analyses the proportion of AYAPHIV with an unsuppressed viral load, as transmission of HIV to others is more likely to happen when participants are not suppressed. For AYAPHEU, we included a proportion variable for participants who had ever taken PrEP, as PrEP use would help protect this group from HIV-infection even if condoms are not used.

Results

Sociodemographic and HIV-related Characteristics

At enrollment, participants included 340 youth (ages 9–16 years) and their caregivers. Half of participants were female (51%), 60% Black, 50% Latinx, and 61% with PHIV, with no significant sociodemographic differences by PHIV-status. By follow-up 6, participants included 223 young adults (ages 19–28 years). Half of participants were female (53%), 61% Black, 50% Hispanic/Latinx, and 60% with PHIV. At follow-up 6, PHIV participants were on average one year older than PHEU participants (23.8 vs. 22.8; t=-2.9(221), p = 0.005), but the groups did not differ significantly by gender, race, or ethnicity.

Among AYAPHIV, median CD4 cell count at follow-up 6 was 475 cells/uL and median viral load was 28 copies/mL, with 67% having a viral load ≤ 400 copies/mL. Almost all AYAPHIV reported taking HIV medications (98%). At more than half of visits, 75% of AYAPHIV had an unsuppressed viral load (> 400 copies/mL).

Psychiatric Trajectories

As presented in Nguyen et al. [16], using data from CASAH, 63% of the overall sample was categorized as having “low probability of psychiatric disorder” over time, 26% as having “consistent probability of anxiety disorder” over time, and 11% as having “persistent or escalating probability of psychiatric comorbidity.” Participants who demonstrated escalating psychiatric comorbidity were significantly older, and participants with consistent anxiety were more likely female. No significant differences in rates of psychiatric trajectories were found between PHIV-status groups.

Condomless Sex

The average age of onset of vaginal or anal sex in the CASAH cohort was 15 years, with 95% of participants reporting the onset of penetrative sex before age 20. By follow-up 6, all participants had initiated sexual activity, with 70% of AYAPHEU and 62% of AYAPHIV reporting having had recent vaginal or anal sex in the past three months. Longitudinally, 56% reported recent condomless sex at one or more follow-up visits, 33% at two or more follow-up visits, and 16% at half or more follow-up visits. The mean proportion of total visits with condomless sex did not differ by gender or race, however, it did differ by PHIV-status, ethnicity, and psychiatric disorder trajectory. Adjusting for age, AYAPHEU reported a higher mean proportion of follow-up visits with recent condomless sex compared to AYAPHIV (24% vs. 19%, adjusted regression coefficient = -0.063, t = -2.395, p = 0.017). Additionally, Latinx AYA reported a higher mean proportion of visits with recent condomless sex compared to their non-Latinx counterparts (25% vs. 17%, adjusted regression coefficient = 0.064, t = 2.476, p = 0.014).

Participants with escalating psychiatric comorbidities had the highest mean proportion of visits with recent condomless sex (44%), followed by those with persistent anxiety (23%) and low probability of psychiatric disorder (17%). Both the escalating psychiatric comorbidities and persistent anxiety trajectories groups were significantly higher in their mean proportion of visits with condomless sex in comparison to the low probability of psychiatric disorder trajectory group (44% vs. 17%, adjusted regression coefficient = 0.223, t = 5.066, p < 0.001; 23% vs. 17%, adjusted regression coefficient = 0.062, t = 2.019, p = 0.043).

When analyzing all the data from baseline to follow-up 6 for AYAPHIV, recent condomless sex was not associated with viral load (adjusted regression coefficient = 0.213, Wald Chi-Squared = 1.051, p = 0.305) while adjusting for age; however, 75% of AYAPHIV had an unsuppressed viral load at more than half of their visits. The mean gap between interview date and CD4/viral load abstraction was 1.5 months (range = 0-31.3) in follow-up 2 and 2.4 months (range = 0-10.2) in follow-up 6.

Table 1 Summary of Recent Sexual Activity & Condomless Sex, Baseline-Follow-up (FU) 6
Full size table
Table 2 Regression Analyses Examining Factors Related to the Proportion of Visits with Any Condomless Sex
Full size table

PrEP Use

Although all AYAPHEU were adults (≥ 18 years) and therefore eligible to receive PrEP by follow-up 6, only two (2.2%) participants reported ever taking PrEP, despite many living in communities with a relatively high HIV-seroprevalence rate. When asked to rate on a 10-point Likert scale how likely they would be to use PrEP in the future (1 = very unlikely; 10 = very likely), 63.3% of AYAPHEU provided a response ≥ 6 (mean score = 6.41).

Discussion

Condomless sex in the context of low rates of viral suppression and limited or no PrEP use endangers all vulnerable populations, including AYAPHIV and AYAPHEU, who are at high risk of HIV transmission and acquisition. Unquestionably, biomedical advances such as PrEP and U = U have made a substantive difference in reducing HIV transmission [3, 29]. However, biomedical advances by themselves have unfortunately not been enough to end the epidemic, with HIV diagnoses continuing to increase in many contexts, including among youth ages 13–24 [30]. In our longitudinal cohort study of adolescents and young adults, over half of AYAPHIV and AYAPHEU participants reported condomless sex at one or more follow-up visits, and one-third at multiple visits. AYAPHEU engaged more frequently in condomless sex than AYAPHIV, which may be due to multiple factors, including the latter group growing up with HIV and more frequent access to health care and education around HIV prevention messaging by providers [31]. Additionally, our analyses suggest that Latinx AYA and AYA with increased psychiatric burden may be particularly likely to engage in condomless sex. Finally, a majority of AYAPHIV were not virally suppressed at one or more study visits, and very few AYAPHEU endorsed using PrEP. The ramifications of these data are important, as there are millions of AYA living with either perinatal HIV-infection or -exposure globally [32]. The data suggest that it remains crucial for evidence-based interventions to promote condom use, medication adherence, and other behavioral changes related to HIV treatment and prevention.

Previous studies point to a range of personal, social, and structural factors—e.g., disparities in access to HIV testing and treatment, poor access to and retention in care, HIV-related stigma and discrimination, and poor adherence to PrEP—as potential barriers to ending the HIV epidemic [33, 34]. With PrEP and U = U not yet actualizing their full potential worldwide, there is concern that overemphasis on these innovations may cause HIV-affected AYA to ignore other concerns related to condomless sex, including other sexually transmitted infections and unplanned pregnancy [35,36,37]. Because AYAPHIV and AYAPHEU have been shown to be at increased risk for psychiatric disorder, suboptimal adherence, and sexual risk-taking [24, 25, 38], our findings emphasize the need for attention to mental healthcare to be used in combination with PrEP and U = U campaigns for maximizing HIV risk reduction. Continued development of products such as long-acting agents for HIV prevention and treatments that decrease the burden of adherence is also necessary.

Mental health has consistently been identified as a critical barrier to HIV prevention and treatment [23, 25, 31]. Our longitudinal dataset showed an association between psychiatric trajectories across adolescence and into young adulthood and condomless sex over time, particularly for youth with multiple co-occurring psychiatric disorders, including substance abuse. These data continue to support the urgent need to evaluate and provide care to AYA living with HIV, both to reduce mental health symptoms as well as identify risks for onward HIV transmission. In line with other studies [39], our findings indicate that AYAPHIV and AYAPHEU are at equal risk for psychiatric problems across this developmental period. However, results from this analysis also indicate that AYAPHEU are at greater risk for condomless sex compared to AYAPHIV, which can result in several poor health outcomes, including HIV, sexually transmitted infections, and unplanned pregnancy. Reaching AYA has been a challenge in the mental health field, not just in HIV care [40]. Importantly, calls for integrating mental health care into primary health care have been consistently made throughout the HIV epidemic [11, 38, 39] and may be important not just for AYAPHIV, but also for reaching AYAPHEU and other vulnerable uninfected populations to maximize prevention efforts [41].

This study also found that youth of Latinx ethnicity had greater odds of engaging in condomless sex. This aligns with previous research that has reported low condom use in many Latinx populations in the US [42,43,44]. Barriers to condom use in Latinx populations include social and personal factors: gender inequality and aversion to condom use by Latino men [45]; lack of skills in negotiating condom use [46]; ineffective condom use [44]; adverse partner reactions to condom use [47]; and negative perception toward condoms [48]. The HIV epidemic in the US continues to disproportionately affect Latinx populations, who represent about 19% of the country’s population but 30% of new infections. AYA are particularly vulnerable, with about 60% of yearly infections among Latinx individuals ages 13–34 [49]. We did not have sufficient numbers of gender and sexual minorities to examine group differences. In line with other studies, future research examining the evolving gender and sexuality constructs among Latinx AYA affected by HIV in the US may be important to prevention efforts [50].

In addition to limited data on gender and sexual minorities, another limitation of this study is that our findings may not be generalizable to participants living outside of New York City. Although demographics from CASAH participants are similar to national studies, with AYAPHIV and AYAPHEU largely from inner-city, low-income, ethnic minority families, it is not clear how reflective our data are of similar populations across the US or in low-to-middle income countries, where the majority of this population reside. Although our cohort is predominantly Black and Latinx, preventing comparisons to other racial and ethnic groups, these demographics align with US and NYC epidemiological data. Other limitations include generalizability to young people not followed in a research study and lack of knowledge about participants’ sexual partners and their HIV-status. For instance, condomless sex among AYAPHIV whose partners are also HIV + would be less risky. We also do not know if AYAPHIV participants had a detectable viral load while engaging in condomless sex. Because viral load can fluctuate, it is possible that participants monitored changes with their sexual partners before having condomless sex. Participants could have also used other methods to protect against sexually transmitted infections and HIV transmission, such as sero-sorting and monogamy. Unfortunately, we do not have the data to examine these variables. Future studies on these relationships between behaviors and prevention efforts across development will likely be important.

For our analyses, our primary focus was on average condomless sex across time, specifically across the developmental stage of adolescence through young adulthood, which is accounted for through use of the GEE modeling. Although we controlled for age, we did not examine specific developmental differences by different stages of adolescence (early, middle, late) and young adulthood. Given our limited sample size, we were concerned about losing statistical power as well as the potential for having too many comparisons across multiple time points, which could have opened up hypothesis testing to erroneous effects. That said, understanding specific differences between time points or developmental stages throughout this transition period may be critical to sexual risk reduction and warrants more investigation in future studies with larger sample sizes. An added limitation in this study’s methodology is our use of ACASI, which may not always be the best way to capture participants’ sexual risk-taking. Other methods of data collection such as diaries and daily calls may capture sexual behavior in an immediate context more accurately.

Continued research is needed to identify other factors associated with condomless sex in AYAPHIV and AYAPHEU. Potential factors include neurocognitive impairment and structural barriers such as poverty, lack of access to and retention in health care, and limited knowledge regarding HIV risk reduction. Although this paper found AYAPHEU vulnerable to condomless sex and largely unengaged in PrEP use at the time of this study, they are a harder group to identify, and without specialized clinics, much less likely to be engaged in health care than AYAPHIV. Given that young adulthood is a time when many individuals are not routinely engaged in care, strategies are needed for identifying and engaging this population in sexual risk reduction interventions. Given that rates of condomless sex in both groups were high, continued research is also needed to explore motivation to use or not use condoms to help inform risk reduction strategies. Additionally, given that our data on PrEP was collected relatively early in its rollout in the US—a couple of years following FDA approval for Truvada in adult populations and at the same time as rollout of Descovy—our examination of PrEP use was exploratory and may explain in part why so few participants reported knowledge of PrEP use. Future research should look at demographic correlates of willingness to use PrEP in this population.

Finally, mental health impairment has been consistently identified as a barrier to HIV prevention and treatment efforts, with previous CASAH data supporting this finding [25]. Therefore, we suggest health care providers continue to explore sexual health, sexual risk reduction, and mental health in AYA living with or at risk for HIV. It is encouraging and should be acknowledged that calls for strategies to integrate behavioral health (mental health and substance use) with HIV treatment and prevention efforts have increased, with funding streams identified (e.g., Ending the HIV Epidemic in the US) [51, 52].

In conclusion, substantive biomedical advances in HIV treatment and prevention suggest that ending the HIV epidemic is in reach [51]. However, efforts to end the epidemic have not yet been actualized, and AYA remain a high-risk population, with increased odds of psychiatric disorder, suboptimal medication adherence, and HIV diagnoses continuing to increase in youth ages 13–24 [30]. Therefore, even in the context of PrEP and U = U, condom use remains an important method of sexual risk reduction, especially in vulnerable populations such as AYAPHIV and AYAPHEU, whose numbers are in the millions globally [32]. Given its continued importance in sexual risk reduction, strategies to promote condom use may need to move beyond a focus on negative health outcomes to include associations of condoms with positive sexual health. As recent research points out, despite billions of dollars and donor funding spent each year by agencies responsible for sexual health, incorporating pleasure into efforts to encourage condom use has been largely absent, as most campaigns continue to define safe sex as a means to avoid negative health outcomes such as sexually transmitted infections and HIV infection [53]. Eroticizing condom use and associating it with pleasure has been associated with improvements in behavioral outcomes and risk-preventative attitudes [53] and may be an important strategy for young people, including AYAPHIV and AYAPHEU. As we move closer to reaching global targets to end the HIV epidemic, it is critical to not homogenize different populations of HIV-affected populations, who may have unique stressors that make reliance on biomedical advances challenging. For these reasons, we stress that condoms continue to be promoted alongside PrEP and U = U and revamped in how they are encouraged for use in different populations of HIV-affected AYA.