Factors Associated with Antiretroviral Adherence among Youth Living with HIV

Abstract

Background:

Youth living with HIV (YLH) have an increased risk for psychosocial stressors that can impact their antiretroviral (ARV) adherence. We examined factors associated with self-reported ARV adherence among YLH ages 12-24 years old.

Setting:

YLH (N=147) were recruited in Los Angeles, California, and New Orleans, Louisiana from 2017-2020.

Methods:

YLH whose self-reported recent (30 days) ARV adherence was “excellent” or “very good” were compared to non-adherent YLH on sociodemographic, clinical, and psychosocial factors using both univariate and multivariate analyses.

Results:

Participants were predominantly male (88%), and 81% identified as gay, bisexual, transgender, queer, or other. The mean duration on ARV was 27 months (range: 0-237 months). Most YLH (71.2%) self-reported being adherent, and 79% of those who self-reported adherence were also virally suppressed (<200 copies/mL). Multivariate analysis indicated being adherent was significantly associated with white race [aOR=8.07, (CI=1.45, 74.0)], Hispanic/Latinx ethnicity [aOR=3.57, CI=(1.16, 12.80)], more social support [aOR=1.11, (CI=1.05, 1.18)], and being on ARV for a shorter duration [aOR=0.99, (CI=0.97, 0.99)]. Mental health symptoms, substance use, age, and history of homelessness or incarceration were unrelated to adherence.

Conclusions:

Enhancing efforts to provide support for adherence to non-white youth, as well as those with limited social support and who have been on ARV treatment longer, may help increase viral suppression among YHL.

Due to the success of antiretroviral (ARV) therapies, human immunodeficiency virus (HIV) infection has become a manageable chronic condition. Persons living with HIV have had an increase in their quality of life and reductions in premature deaths and are also less likely to transmit HIV to others when virally suppressed.^{1, 2} Yet, to achieve viral suppression and garner these benefits, a high level of adherence to ARV is needed (75% or greater depending on the medication).^{2, 3} Suboptimal adherence increases the likelihood of viral rebound, resistance, treatment failure, and transmission.^{4, 5}

Suboptimal adherence is a serious problem among youth living with HIV (YLH). Adolescents now account for 21% of new HIV diagnoses in 2019 in the United States.⁶ They have one of the lowest adherence rates with only 62% of adolescents having an adherence rate of 85%.⁷ With low adherence, YLH are at increased mortality and morbidity.^{1, 4, 5} YLH under 24 years of age have also been shown to have worse clinical outcomes when compared to adults living with HIV at every step of the HIV care cascade, including retention in medical care and ARV adherence.⁸

Medication adherence is generally defined as the extent to which patients follow their medication regimen as prescribed by their health care providers. Several methods can be used in measuring and monitoring medication adherence: pharmacy claims, electronic monitoring, and self-report. Self-reports have been widely used due to their simplicity and validity.^9–11 For example, a multi-center study among adults living with HIV in the United States found relatively high validity between self-reported adherence and those based on biomarkers.¹² The current study examines self-reported adherence among a diverse group of YLH, as well as both risk and protective factors associated with ARV adherence.

Much of the research focused on adolescents and young adults misses at-risk or hidden populations, such as sexual and gender diverse youth (SGDY; i.e., men-who-have-sex-with-men, bisexual, transgender, queer, and other), people who inject drugs, sex workers, and incarcerated youth.¹³ These youth that are under-represented in research studies often face stigma and may be at greater risk for poor health outcomes.¹⁴ Given the difficulty of identifying these young people, it is critical to look beyond clinical care sites to identify SGDY who are not routinely linked to care.

Moreover, youth’s developmental phase may place them at risk for mental health consequences of stigma and sensitivity to social rejection.¹⁵ Those mental health problems not only affect their quality of life but also their ARV adherence and viral suppression.^14–16 Given that risk, it is important to monitor mental health symptoms among YLH over time to assess the potential impact of these challenges on adherence. Additionally, because youth who identify as SGDY often experience stigma and discrimination, they are less likely to have the developmental and personal resources to address social stigma.^{15, 17} The U.S. Centers for Disease Control and Prevention has documented that about 84% of youth and young adults new HIV cases in 2019 involved male-to-male sexual contact.¹⁸ Many of these youth, also face discrimination because they are disproportionately of a minority race or ethnicity (e.g., African-American, Hispanic/Latinx, or Native American). Across studies of race and adherence, being African American has been associated with lower ARV adherence among adults.^19–21

Yet, beyond poor mental health and stigma due to sexual and gender identity or racial/ethnic minority status, there is a cluster of problem behaviors and stressors that often characterize the lives of adolescents,²² which are lower in adulthood: drug abuse, homelessness, incarceration, bartering sex, and having sexually transmitted infections. These challenges are only likely to change slowly over time and to lead to reduced adherence compared to peers without such challenges.^{23, 24} Similar findings emerge among adults.^25–27 Additionally, the longer one is living with HIV, adherence decreases.²⁷ For example, youth who are perinatally infected have the lowest adherence among adolescents.²⁸ While adolescents typically have low adherence to health regimens, the longer young people are infected, the lower their adherence tends to be. Therefore, the relationship between histories of risk and adherence are examined.

While sexual risk behaviors, substance use, and psychosocial life stressors decrease adherence, there are also protective factors that are associated with adherence. Social support is protective against mental health symptoms and is an important factor that can impact adherence among YLH^{24, 29} and adults living with HIV.³⁰ In particular, social support from caregivers and family members is exceptionally important.³⁰ Health care providers are also an important source of support to help with coping, particularly for those who are newly diagnosed.³¹

The current study fills a gap in the existing literature by examining self-reported adherence among a diverse group of YLH, as well as both risk and protective factors associated with ARV adherence. Based on the literature reviewed above, we hypothesized that mental health symptoms and being from an ethnic or racial background that is a minority in the United States would be associated with lower self-reported ARV adherence. We also hypothesized that having more social support and less time on ARVs would be associated with higher self-reported ARV adherence.

In addition, we included a broad range of youth, not only those engaged in clinical care. This study focused on youth in two American cities who received services at a broad range of community agencies and clinical care sites or who engaged actively in social media dating sites.³² Moving beyond adolescent medicine clinics allowed us to identify youth both receiving health care, as well as those only connected to social service sites that serve youth experiencing multiple challenges.

Methods

Study Design, Setting, Recruitment, Inclusion Criteria, and Measurement

This study was part of the Adolescent Medicine Trials Network (ATN) study (Trial Registration: ClinicalTrials.gov NCT03109431) that aimed to increase adherence among YLH.³² YLH were recruited from 13 homeless shelters, clinics, and community-based organizations working with SGD youth in Los Angeles, California, and New Orleans, Louisiana, from May 2017 to May 2020.^32–34 To be included in the study, youth had to be 12-24 years of age, test positive for HIV with established HIV infection (not acutely infected), and be able to provide informed consent. Youth were excluded who did not meet the age or HIV status criteria or appeared to have cognitive difficulties or to be under the influence of alcohol or other substances. Assessments were done at baseline and four-month intervals for the duration of the study.

In addition, YLH were identified on social media dating apps in these areas. Data were obtained from the baseline assessment from this clinical trial that involved follow-up over 24 months. Trained staff collected the following specimens and administered a one-hour interview that included the following:

HIV Testing.

All youth were tested for HIV antigen and antibody screening with a rapid-HIV test using the Alere (Waltham, MA) Determine HIV-1/2Ag/Ab Combo finger stick and the Cepheid (Sunnyvale, CA) Xpert HIV-1 Qual Assay which is Clinical Laboratory Improvement Amendments (CLIA) waived and FDA approved.³³ The test is a point-of-care lateral flow strip that detects both HIV-1/2 antibodies and the HIV-1 p24 antigen using 50 microliters of fingerstick whole blood. The window period is 12-26 days, and results are ready in 20-40 minutes. At the same assessment, a sample of whole blood was collected in order to evaluate viral suppression, which was defined as <200 copies/ml, and CD4 level was also assessed.

Medication Adherence.

Self-reported medication adherence was a single item measure based on studies by Wilson and colleagues³⁵ and Lu and colleagues³⁶ that has been used in clinical trials conducted by the Adult AIDS Clinical Trials Group (AACTG).³⁷ Participants were asked to rate their ability to take all their HIV medications as prescribed, answering on a six-point scale from “very poor” to “excellent.”

Antiretroviral duration.

We calculated the duration of ARV medication usage based on the difference between self-reported first date of ARV and the date of the data collection. In some cases, participants reported starting ARV later than the data of the survey. The ARV duration for these individuals was coded as 0.

Sociodemographic characteristics.

We collected data on a variety of background and demographic characteristics, including age, sex, gender identity, sexual orientation, ethnicity, education, income, and employment.

Psychosocial stressors.

YLH reported on histories of incarceration, homelessness, engagement in sex work (exchanged sex for drugs or money in their lifetime), psychiatric hospitalizations, and inpatient substance abuse treatment.

Sexual transmitted infections (STIs).

STIs were measured by self-report over the past 12 months.

Substance use.

Heavy drinking was defined as drinking 7 or more drinks on a typical day when drinking. Drug abuse was measured by asking participants if they have ever used any drugs over the past 4 months other than marijuana, and if so, how many of the following substances: cocaine or crack, heroin, ecstasy, methamphetamines, prescription stimulants or amphetamines, gamma hydroxybutyric acid, ketamine, poppers, inhalants, hallucinogens, prescription painkillers not used as prescribed, and other prescription medications not used as prescribed.

Social support.

The three-item self-rated social support subscale³⁸ from the Coping Self-Efficacy Scale was used to measure social support. Items were rated on a scale of 0 to 30 (higher scores indicating more support) and were as follows: “I can get friends to help me with things I need,” “I can get emotional support from friends and family,” and “I can make new friends.”

Mental health symptoms.

The Generalized Anxiety Disorder 7-item (GAD-7) scale was used to assess anxiety, and the Patient Health Questionnaire (PHQ-9) was used to measure depression.^{39, 40} Both measures are valid and reliable self-report questionnaires that have been previously used to measure depression and anxiety among people living with HIV.^{41, 42}

Data Analysis

To determine factors associated with medication adherence at baseline, we first ran logistic regression models with adherence as the outcome and each of the following as a univariate predictor: ARV duration, CD4 count, homelessness, incarceration, heaving drinking, mental health hospitalization, substance use treatment, number of drugs used, a variety of specific drugs, engagement in sex work, a STI diagnosis, depression, anxiety, and social support. Adherent was defined as having “excellent” or “very good” self-rated adherence in the past 30 days. We also investigated the association between viral load and adherence to validate our adherence measure.

The next step was a multivariate analysis, looking at how different factors jointly explain adherence. A-priori, we decided to include ARV duration, age, sex at birth, gender, sexual orientation, race, and ethnicity in the logistic regression model. Because they were strongly associated (p <0.10) with adherence in the univariate analysis, we also included drug use (none, one, two or more), social support, and depression (PHQ-9 score) in the model. We chose this variable selection approach over an automated variable selection process like stepwise regression because of statistical inference issues with stepwise,⁴³ because we already had a large set of pre-determined variables and there was a relatively small number of possible additions. To aid interpretability, we reduced the gender variable down to only cis-gender and trans/gender-diverse identity. The main outcomes were adjusted odds ratios (AOR) with 95% confidence intervals (CI). Statistical analysis was performed using R version 4.2.0 (RStudio Inc., Boston, MA, USA).

Results

A total of 147 YLH were included in our analyses out of 170 youth who were initially recruited for this study as we excluded those who had missing data on the variables of interest. Table 1 shows descriptive statistics of participants’ baseline characteristics. Most participants (80%) identified as cisgender male and homosexual (59%). The average duration of taking ARV was 27 months (range 0-237 months). Most (71%) reported being adherent, and 79% of those who self-reported being adherent were virally suppressed (<200 copies/mL). Of those who self-reported non-adherence, 61% were virally suppressed. Overall, viral suppression was 74%. CD4 counts were not significantly different based on self-reported adherence.

Table 1.

Potential factors associated with self-reported adherence (N=147)

Participant Characteristics	Adherent N=104 (71%)	Non-Adherent N=43 (29%)	Total N=147 N (%)
Virally suppressed (<200 copies/mL)	82 (79%)	26 (61%)	108 (74%)
Viral load (Mean/Median/SD)	9530 [20] (33000)	25800 [32] (64800)	14300 [20] (45000)
Time on ARVs (Months; Mean/Median/SD)	21.9 [12] 33.5	40.9 [27] 38.1	27.4 [15] 39.2
Time since diagnosis (Months; Mean/Median/SD)	28.6 [15] 42.8	56.9 [35.5] 68.5	36.9 [23] 53.0
CD4 Count (cells/mm3; Mean/Median/SD)	955 [600] (1782)	588 [668] (342)	883 [607] (1607)
Age in years (Mean/Median/SD)	22.1 [22] (2.0)	21.4 [22] (2.3)	22.9 [22] (2.1)
Sex assigned at birth
Male	94 (90%)	36 (84%)	130 (88%)
Female	10 (10%)	7 (16%)	17 (12%)
Gender
Cisgender male	85 (82%)	33 (77%)	118 (80%)
Cisgender female	9 (9%)	7 (16%)	16 (11%)
Transgender woman	7 (7%)	3 (7%)	10 (7%)
Transgender man	1 (1%)	0 (0%)	1 (1%)
Gender non-binary/gender non-conforming	2 (2%)	0 (0%)	2 (1%)
Sexual orientation
Homosexual	62 (60%)	25 (58%)	87 (59%)
Heterosexual	16 (15%)	10 (23%)	26 (18%)
Bisexual	19 (18%)	5 (12%)	24 (16%)
Other	7 (7%)	2 (5%)	9 (6%)
Ethnicity
Hispanic/Latinx	30 (29%)	7 (16%)	37 (25%)
Not Hispanic/Latinx	74 (71%)	36 (84%)	110 (75%)
Race
White	16 (15%)	2 (5%)	18 (12%)
Non-white	88 (85%)	41 (95%)	129 (88%)
Homelessness	32 (31%)	18 (42%)	50 (34%)
Previously incarcerated	20 (19%)	8 (19%)	28 (19%)
Heavy drinking	9 (9%)	7 (16%)	16 (11%)
Previous mental health hospitalization	16 (15%)	10 (23%)	26 (18%)
Previous substance use treatment	13 (13%)	8 (19%)	21 (14%)
Self-reported drug use
No drugs	18 (17%)	13 (30%)	31 (21%)
One drug	49 (47%)	16 (37%)	65 (44%)
Two or more drugs	37 (36%)	14 (33%)	51 (35%)
Cannabis	82 (79%)	30 (70%)	112 (76%)
Cocaine	14 (14%)	6 (14%)	20 (14%)
Methamphetamine	15 (14%)	4 (9%)	19 (12.9%)
Other amphetamines	13 (13%)	4 (9%)	20 (14%)
Inhalants	23 (22%)	8 (19%)	31 (21%)
Sedatives	18 (17%)	5 (12%)	23 (16%)
Hallucinogens	11 (11%)	4 (9%)	15 (10%)
Opioids	9 (9%)	2 (5%)	11 (8%)
Prior sex work	22 (21%)	14 (33%)	36 (25%)
Self-reported sexually transmitted infections for chlamydia, gonorrhea, or syphilis	54 (52%)	16 (37%)	70 (48%)
Depression	4.8 [4] (4.4)	6.4 [5] (5.6)	5.3 [4] (4.8)
Anxiety	4.9 [4] (4.9)	5.8 [3.5] (6.3)	5.2 [4] (5.3)
Social Support	24.7 [27] (6.1)	19.7 [20.5] (8.4)	23.3 [26] (7.2)

Supplemental Table shows the results of the single predictor logistic regression models with relevant odds ratios and p-values. Table 2 shows the details of the results of the final logistic regression model. Based on the results from the final multivariate logistic regression model, youth of white race as compared to non-white race [aOR=8.07, CI=(1.45, 74.0), p=0.033], of Hispanic/Latinx ethnicity [aOR=3.57, CI=(1.16, 12.80), p=0.035], with higher social support [aOR=1.11, CI=(1.04, 1.19), p=0.001] and with shorter ARV treatment duration [aOR=0.98, CI=(0.97, 1.00), p=0.010] were more likely to report being adherent. Those who were assigned female sex at birth [aOR=8.08, CI=(0.79, 99.8), p=0.086] or were bisexual [as compared to heterosexuals; aOR=7.08, CI=(0.94, 57.1), p=0.058] had increased odds of reporting being adherent but these findings were not statistically significant (p >.05). Similarly, those who reported lower depression had increased odds of self-reporting ARV adherence [aOR=0.33, CI=(0.09, 1.08), p=0.067], but this result was also not statistically significant (p >.05).

Table 2.

Results of the Multivariate Logistic Regression Analysis for Youth Living with HIV* (N=147), eliminating risk factors unassociated with adherence in the univariate analyses.

Variable	aOR	95% CI	P-value
ARV treatment duration (months)	0.98	(0.97, 1.00)	0.010
Social support	1.11	(1.04, 1.19)	0.001
Age (years)	1.03	(0.83, 1.29)	0.700
Sex at birth (Ref: male): Female	8.08	(0.79, 99.8)	0.086
Gender (Ref: cis-gender): Trans/gender diverse	4.43	(0.83, 34.3)	0.110
Sexual orientation (Ref: Heterosexual):
Bisexual	7.08	(0.94, 57.1)	0.058
Gay	4.29	(0.68, 27.0)	0.110
Other	3.98	(0.35, 51.9)	0.300
Race (Ref: non-white):White	8.07	(1.45, 74.0)	0.033
Ethnicity (Ref: Not Hispanic/Latinx): Hispanic/Latinx	3.57	(1.16, 12.80)	0.035
Number of Drugs Used (Ref: None):
One	2.73	(0.81, 9.36)	0.100
Two or More	1.60	(0.39, 6.38)	0.500
Depression	0.33	(0.09, 1.08)	0.067

Discussion

This study makes a unique contribution to the literature on adherence among YLH. Compared to prior studies,^44–46 self-reported medication adherence was highly related to viral load and was about 10% higher than previous reports. About 79% of participants who self-reported adherence were suppressed compared to 61% of those who reported non-adherence were virally suppressed. Although a majority of youth who perceive themselves as non-adherent were virally suppressed, this may lead to concern that these youth may not be maintaining long-term habits for achieving prolonged viral suppression. However, as drug regimens have changed, viral suppression may be achieved with lower levels of adherence. Studies have shown that adhering about four days a week to medication may result in viral suppression with current ARV regimens.^{47, 48} It may be key to tailor adherence messages to the new medication regimens and set realistic expectations for YLH about how many days per week they must be adherent for optimal health outcomes. These issues also may become unimportant as the field moves to long-term injectable ARV.^{49, 50} Our analyses showed that having social support is significantly associated with optimal ARV adherence levels, similar to other studies with adolescents living with HIV.^51–53 Relationships may both motivate higher adherence, to protect friends and family from the consequences of non-adherence, or as a trigger to remind young people to adhere to their medications. This finding is expected and points to the importance of designing interventions for young people that build on their social networks.

A shorter duration of ARV treatment was a critical factor related to ARV adherence. Indeed, medication adherence is a complex behavior and many factors that influence it have been identified among people living with HIV including both adults and adolescents. Others’ research on adherence similarly documented that the longer time since diagnosis was significantly associated with nonadherence.^{54, 55} Our findings confirm that treatment duration still is an important factor even during the current era of single-pill fixed-dose once daily regimens. Therefore, adherence interventions need to be longitudinal over time to account for such a challenge.

Race and ethnicity appeared to play a role in adherence for this sample of youth. White youth were about 8 times more likely to report adherence than their non-white peers. However, Hispanic/Latinx youth were 3.5 times more likely to report higher levels of adherence. Most studies on racial and ethnic disparities among YLH have been conducted with youth with perinatally acquired HIV but also suggest racial and ethnic disparities. In their longitudinal study of youth with perinatal HIV, Kacanek and colleagues⁵⁶ found that black and Hispanic youth had lower baseline rates of 3-day recall non-adherence and missed doses in the past month, but differences were not statistically significant. Naar-King and colleagues⁵⁷ found that African American youth had lower levels of non-adherence but noted that race may be proxy for other factors associated with health disparities. It is possible that there are specific factors, such as motivational readiness, that may play a role in adherence for youth of color living with HIV.⁵⁸ Using data from 13 studies not specific to perinatal HIV acquisition or to youth, Simoni and colleagues²⁰ found disparities in adherence related to race and ethnicity, with lowest rates among African Americans. Existing interventions may be insufficient to reduce these disparities for YLH, and the complexity of the intersectionality of adherence with other factors warrants further exploration.

Interestingly, several factors that were reported in previous studies as predictors of non-adherence^59–61 did not have as clear of effects in our study. In particular, there was no evidence anxiety had an effect and while we had some confidence in a depression effect, the evidence was relatively weak. At the time of entry into the study, we found low rates of symptoms of both anxiety and depression. The mean scores were considerably lower than the clinical thresholds set for these scales. This finding is different from earlier reports¹⁶ and may reflect that HIV is now a chronic disease, not likely to result in premature death. However, 17% of this sample had been previously hospitalized for mental health problems, which would have led us to expect higher rates of depression and anxiety.

Alternatively, this is a sample that reported engaging in many behaviors that can be associated with increased risk for poor adherence, such as homelessness, incarceration, and heavy drinking. These kinds of behaviors did not characterize earlier samples of YLH in the United States. It may be that youth with a greater number of life challenges are becoming HIV infected in the settings that are associated with homelessness or heavy drinking (e.g., bars). The findings may also be associated with our recruitment strategy which focused on intentionally obtaining a diverse sample of youth whose risks are not reflected in previous studies of youth in this same age range. However, two of these risks, heavy drinking and sex work, were not associated with increased odd of adherence.

Strengths and Limitations

In addition to the diversity of study sample, the strengths of this study are the richness of the variables collected among YLH that have an impact on their day-to-day medication-taking behavior. We also had measures of both self-reported adherence and viral load, supporting the accuracy of those variables. However, this study has some limitations. ARV adherence was measured through self-report, which is known to over-estimate adherence. Among adolescents, Craker and colleagues⁶² found that when WisePill caps were monitored over time, YLH overestimated their adherence by 20-40%. We also used only one measure of adherence, however, 79% of those reporting being adherent were also virally suppressed, supporting the validity of the measure. We believe that since most of those who reportedly had high adherence were also virally suppressed, that the bias in adherence measurement is limited. In future studies, we hope to conduct the analysis among a larger sample size and to use a more objective measure of adherence such as pharmacy refill records or electronically monitored adherence.

Conclusions

These results highlight the importance of support from others and duration of ARV treatment as critical factors in shaping adherence to ARV medications as well as the possible impact of racial, but not ethnic, disparities. A better understanding of clinical, psychosocial, and socioeconomic factors is needed to develop effective ARV adherence interventions. These behaviors are complex and are likely impacted by multiple aspects of YLH’s lives that may can be challenging to address but are critical to health and well-being.