Antiretroviral therapy concealment behaviors and their association with antiretroviral therapy adherence among people with HIV: Findings from the Florida Cohort Study

Abstract

Little is known about HIV medication concealment behaviors and the effect of medication concealment on antiretroviral therapy (ART) adherence among people with HIV (PWH). This study aims to 1) describe medication concealment behaviors and factors associated with these behaviors, and 2) assess the association between medication concealment and suboptimal ART adherence. The Florida Cohort Study enrolled adult PWH from community-based clinics around the state from October 2020 to September 2022 (n=416, 62% aged 50+, 56% male, 44% non-Hispanic Black, 18% Hispanic). Participants responded to questions about sociodemographics, ART adherence (≥85%), symptoms of depression, social networks and disclosure to their network, and actions to conceal ART to avoid inadvertent disclosure of their HIV status. Analyses were conducted using multivariable logistic regression models. The most common concealment behavior was hiding ART while having guests over (32%), followed by removing ART labels (26%), and putting ART into a different bottle (16%). Overall, 43% reported ≥1 behavior. In multivariable models, depressive symptoms, incomplete disclosure of HIV to close social networks, and not having a close social network were associated with ART concealment. After adjusting for risk factors for suboptimal ART adherence, endorsing hiding medication while having guests was associated with suboptimal ART adherence (aOR 2.87, 95% CI 1.15–7.55). Taking any action and other individual behaviors were not associated. ART concealment behaviors were common but did not consistently negatively influence adherence when accounting for other factors. PWH may want to receive ART medications in ways that ensure privacy and reduce the risk of inadvertent disclosure.

Introduction

In 2019, the United States (US) Department of Health and Human Health Services (HHS) proposed the Ending HIV Epidemic (EHE) Initiative with the goal of reducing the numbers of incident infections in the U.S. by 75% within 5 years and then by 90% within 10 years [1]. Florida is a high HIV incidence and prevalence setting and, in light of this, the EHE Initiative identified seven Florida counties as priority areas [1,2]. A central pillar of the EHE is achieving rapid and sustained viral suppression for people with HIV (PWH), which requires improved access and adherence to antiretroviral therapy (ART) [1]. To ensure effective treatment, ART needs to be taken with minimal deviation from prescribed directions, requiring consistency in medication-taking behavior [3]. Reaching high levels of ART adherence is vital in PWH to help them reach sustained viral suppression, minimize resistance, and improve individual survival [4]. Research has found that while traditionally 90–95% adherence has been recommended, viral suppression can potentially be maintained at levels of 80% [5,6]. Achieving viral suppression has wide public health implications, since those with an undetectable viral load cannot transmit HIV sexually (referred to as U=U or undetectable equals untransmittable) [7,8]. Suboptimal adherence can lead to viral resistance, having a detectable viral load, increasing the risk of further transmission, and decreasing the effectiveness of key ART medications [9–11]. Resistance can also make it more difficult to achieve the EHE goals if the resistant strains are transmitted to others [12,13].

Despite widespread knowledge of the importance of adherence, poor ART adherence is still common with 42% of PWH not achieving 80% adherence [14]. Research has found that the Southern U.S. has one of the lowest rates of optimal (>90%) ART adherence nationally at 73% [14,15] as well as one of the lowest rates of viral suppression at 63% [16]. Many factors negatively influence ART adherence including a lack of social support, high travel times to the clinic or pharmacy, mental health conditions, and low levels of trust in HIV care providers and the medical establishment [17]. Additionally, race/ethnicity [18,19], age [20], sex [21], substance use [18], and stigma [17,22,23] have all been found to be significantly associated with ART adherence and viral suppression.

A wider disclosure of one’s HIV status to their social network is associated with higher ART adherence [24–27], but a fear of disclosure of one’s HIV status to their social circle is highly associated with anticipated stigma where those who expect to be discriminated against for their condition are less likely to share their status [28,29]. Since ART is most commonly used to treat HIV, the presence of ART medications can inadvertently disclose one’s HIV status. Given this, having to take ART remains a concern for many PWH who wish to keep their HIV status private and PWH who are afraid of experiencing HIV-related stigma [30,31]. As such, steps to hide one’s HIV status can be taken, some of which are disruptive to care and might reduce ART adherence like traveling far away to receive care or actively hiding one’s medications [32]. Medication hiding to conceal one’s disease status has been shown to be associated with reduced adherence to medication for other disease conditions including diabetes and epilepsy [33,34]. Previous studies on the effects of medication concealment behaviors and fear of disclosure on adherence among those with HIV have yielded divergent findings [35–38]. Additionally, many studies have focused on young populations and those outside the US. Little is known about specific HIV medication concealment behaviors and the effect of medication concealment on HIV adherence and viral suppression among adult PWH in the US.

The aims of the present study are 1) to describe medication concealment behaviors and factors associated with these behaviors, and 2) assess if medication concealment is associated with optimal ART adherence among PWH enrolled in the Florida Cohort Study.

Methods

Study Sample

The Florida Cohort study examines healthcare utilization and outcomes among PWH around the state with the goal of informing interventions to improve healthcare services and overall health [39]. Participants were recruited for the Florida Cohort Study from private clinics, health departments, and community settings in six counties (Alachua, Brevard, Hillsborough, Marion, Miami-Dade, and Palm Beach), three of which (Hillsborough, Miami-Dade, and Palm Beach) are EHE priority areas. To be eligible for the parent study, individuals had to be at least 18 years of age, living with HIV, and receive HIV care within Florida. Questionnaires were available in English, Spanish, and Haitian Creole. Participants were eligible for the present analyses if they completed the baseline questionnaire between October 2020 and September 2022, reported being prescribed ART, and answered the question about medication concealment (n=416). All data in this study come from the baseline questionnaires.

Demographics

Participants responded to questions about their demographic information (age, race and ethnicity, rurality, education, sex, etc.). Age was broken into three categories, those under the age of 35, those aged 35–49, and those aged 50 and older. Race and ethnicity were categorized as non-Hispanic White, non-Hispanic Black, Hispanic, and Multiracial or Other. For education, participants were categorized into those who had a high school-level education (or GED) or less and those who had at least some post-secondary education. Those who reported living on the street, in a car, or in a shelter in the preceding 12 months were categorized as unhoused. For sexual orientation, those who identified as gay, lesbian, or bisexual were compared to those who identified as heterosexual or straight. Rurality was classified by the Rural-Urban Commuting Area (RUCA2) codes and the classification scheme developed by the University of Washington [40,41].

ART Concealment

Participants were asked if they had taken any of five actions to conceal their ART in order “to try to prevent others from finding out about your HIV status” in the preceding 12 months. The actions were 1) actively hiding ART when people came to their home, 2) removing labels from the ART pill bottles, 3) putting ART into a different container, 4) changing the pharmacy where they got their ART, and 5) traveling at least 30 miles to pick up their ART. These responses were based on the study team’s past observations and conversations with the community and conversations with providers and PWH. Participants could endorse multiple actions from this list or select a “none of the above” option. An all-inclusive variable was created so that those who reported taking any of the five actions were compared to those who reported taking none of the actions.

Adherence

Participant adherence to their ART regimen in the preceding 30 days was assessed via a single item: “In the past 30 days, on how many days did you miss at least one dose of any of your HIV medication?”. This reported number was used to calculate the percentage of days during which the participant took their medication as prescribed. These percentages were then categorized using a cutoff of 85%. Past research from the Florida Cohort team found that optimal levels of ART adherence that maximized the likelihood of viral suppression were between 80%–90% [42].

Risk Factors for Non-Adherence

The questionnaire also asked about alcohol and other substance use. Alcohol use was assessed using the AUDIT-C score [43–45]. Those who were assigned female at birth with a score of 3 or more or those who were assigned male at birth with a score of 4 or more were categorized as having at-risk drinking, and those with some use but who fell below these thresholds were categorized as having not at-risk drinking [45]. Participants who reported no alcohol use in the past 12 months were categorized as abstinent. For both injection and non-injection drug use, those who reported any use in the past 12 months were compared to those who did not report past-year use. Symptoms of anxiety and depression were assessed using the GAD-7 [46] and PHQ-8 [47], respectively. Those with a score of 10 or more on these measures were classified as having present symptoms.

Disclosure and Stigmatizing Experiences

Participants were asked to estimate the number of people in their close social network, defined as “people that you feel close to, such as family and best friends, and other people important to you or with whom you communicate frequently.” Those who indicated that their close social network included at least one person were then asked if they had disclosed their HIV status to all, some, or none of their network. Disclosure was categorized into three groups: those without a social network, those who had disclosed to none or some of their close social network (incomplete disclosure), and those who reported disclosure to their entire network (complete disclosure). Three individual items related to recent stigmatizing experiences were also assessed. These items were developed by the research team based on unpublished past work within the community and modern communication platforms that are not accounted for in other stigma measures. Three potentially stigmatizing experiences were included and assessed individually: hearing friends or coworkers say negative things about PWH, seeing negative posts about PWH on social media, and being rejected on dating applications or receiving “mean” messages on these applications due to their HIV status. For each of the three variables participants were categorized so that those who reported having that experience in the past 12 months were compared to those who had never had that experience or who those who had experienced the stigmatizing event more than 12 months ago.

Statistical Analyses

The research team assessed the data for missing values and conducted single imputation using the ‘missRanger’ package in R v4.2.2 [48,49]. MissRanger uses a random forest method to replace missing values. These analyses were run so that five non-missing values were sampled for predictive mean matching and the default of 500 chained trees. A table with the proportion of missing values is presented in the supplementary information along with all analyses using a complete case analysis. All available data in the dataset were used for imputation.

Basic descriptive statistics were calculated for optimal ART adherence, overall ART concealment, each concealment behavior individually, demographics, and all other covariates. In bivariate and multivariable analyses where ART concealment was the outcome, only the overall ART concealment variable was used. For models where ART adherence was the outcome separate models were run for overall ART concealment and each behavior individually. Simple logistic regression models were used for bivariate analyses to examine which demographic, risk factors for ART nonadherence, disclosure and stigma variables were associated with medication concealment. Variable selection for the multivariable models was done empirically due to the lack of evidence in the literature, variables with significant associations to medication concealment were included. To assess whether medication concealment behaviors were associated with suboptimal ART adherence, multivariable logistic regression models were used which included risk factors for nonadherence to ART identified in the literature and variables associated with adherence in our dataset, as well as the overall ART concealment variable [18,23,38,50–53]. The models including changing pharmacies and traveling 30 or more miles behaviors used a Firth logit model (‘logistf’ package in R) to account for the small cell sizes [54,55]. Exploratory mediation analyses were conducted using the ‘mediation’ package in R to see if reporting one or more ART concealment behavior mediated the relationship between risk factors for suboptimal ART adherence and ART adherence [56,57]. Exposures in these analyses included the variables that were associated with both ART adherence and overall ART concealment and were adjusted for other factors related to both.

Ethics Statement

The Florida Cohort was approved by the local Institutional Review Board and all participants gave written informed consent prior to their participation in the study.

Results

Overall, 416 participants were included in the present analyses. The majority were over the age of 50 (62%), identified as male (56%), had a high school education or less (53%), and the vast majority had adhered to their ART regimen at least 85% of the time (93%). A plurality of participants were non-Hispanic Black (44%), and around one-eighth were Hispanic (18%) See Table 1 for participant demographics and behaviors by ART concealment endorsement.

Table 1.

Sample Demographics of 416 PLWH in the Florida Cohort Study and their Relationship to ART Concealment Behaviors

Characteristic	No Did Not Endorse ART Concealment Behaviors (N=238)	Endorsed 1+ ART Concealment Behaviors (N=178)	OR (95% CI)
Age
<35	23 (39.7)	35 (60.3)	Ref
35–49	48 (48.0)	52 (52.0)	0.71 (0.37–1.37)
50+	167 (64.7)	91 (35.3)	0.36 (0.20–0.64)
Symptoms of Anxiety
Absent	205 (62.3)	124 (37.7)	Ref
Present	33 (37.9)	54 (62.1)	2.71 (1.67–4.44)
Drinking Status
No alcohol use	82 (70.7)	34 (29.3)	Ref
Not at risk	84 (56.8)	64 (43.2)	1.84 (1.10–3.10)
At risk	72 (47.4)	80 (52.6)	2.68 (1.62–4.50)
Symptoms of Depression
Absent	199 (64.0)	112 (36.0)	Ref
Present	39 (37.1)	66 (62.9)	3.01 (1.91–4.79)
Disclosure to Close Social Network
Complete Disclosure	106 (73.6)	38 (26.4)	Ref
Incomplete Disclosure	116 (49.2)	120 (50.8)	2.89 (1.85–4.56)
No Close Network	16 (44.4)	20 (55.6)	3.49 (1.65–7.51)
Other Drug Use in Past 12 months
No	111 (60.3)	73 (39.7)	Ref
Yes	127 (54.7)	105 (45.3)	1.26 (0.85–1.86)
Highest Level of Education
High School or less	127 (57.5)	94 (42.5)	Ref
At least some college	111 (56.9)	84 (43.1)	1.02 (0.69–1.51)
Gender
Male	138 (59.7)	93 (40.3)	Ref
Female	91 (54.8)	75 (45.2)	1.22 (0.82–1.83)
Transgender or Non-Binary	9 (47.4)	10 (52.6)	1.65 (0.64–4.30)
Housing Status
Unhoused	22 (45.8)	26 (54.2)	Ref
Housed	216 (58.7)	152 (41.3)	0.60 (0.32–1.09)
Injection Drug Use in Past 12 months
Yes	9 (37.5)	15 (62.5)	Ref
No	229 (58.4)	163 (41.6)	0.43 (0.18–0.98)
Marital Status
Single	131 (57.0)	99 (43.0)	Ref
Divorced/Widowed	63 (57.3)	47 (42.7)	0.99 (0.62–1.56)
Long-Term Partner	44 (57.9)	32 (42.1)	0.96 (0.57–1.62)
ART Adherence
85%+	230 (59.3)	158 (40.7)	Ref
<85%	8 (28.6)	20 (71.4)	3.64 (1.62–8.97)
Sexual Orientation
Non-Sexual Minority	124 (59.6)	84 (40.4)	Ref
Sexual Minority	114 (54.8)	94 (45.2)	1.22 (0.83–1.80)
Race/Ethnicity
Non-Hispanic White	75 (55.6)	60 (44.4)	Ref
Non-Hispanic Black	102 (56.4)	79 (43.6)	0.97 (0.62–1.52)
Hispanic	43 (58.9)	30 (41.1)	0.87 (0.49–1.55)
Other/Multiracial	18 (66.7)	9 (33.3)	0.63 (0.25–1.46)
Received Stigmatizing Messages on Dating Apps
Did not experience	216 (58.5)	153 (41.5)	Ref
Experienced	22 (46.8)	25 (53.2)	1.60 (0.87–2.97)
Heard Stigmatizing Comments
Did not experience	178 (64.0)	100 (36.0)	Ref
Experienced	60 (43.5)	78 (56.5)	2.31 (1.53–3.52)
Saw Stigmatizing Comments Online
Did not experience	174 (61.7)	108 (38.3)	Ref
Experienced	64 (47.8)	70 (52.2)	1.76 (1.16–2.67)
Rurality
Rural	31 (52.5)	28 (47.5)	Ref
Urban	207 (58.0)	150 (42.0)	0.80 (0.46–1.40)
Study Site
North Central Florida	69 (52.3)	63 (47.7)	Ref
Melbourne	57 (53.3)	50 (46.7)	0.96 (0.58–1.60)
Miami	44 (60.3)	29 (39.7)	0.72 (0.40–1.29)
Tampa	68 (65.4)	36 (34.6)	0.58 (0.34–0.98)

Engaging in at least one ART concealment behavior was reported by 180 participants, just under half of the sample (43%; see Figure 1). Hiding ART while having guests over was the most frequently endorsed behavior (32%), followed by removing ART labels (26%), and putting ART into a different bottle (16%). Changing pharmacies and traveling for at least 30 miles were both endorsed less frequently (4% and 3% of participants, respectively).

Figure 1.

ART Concealment Behaviors Endorsed by 416 PLWH in the Florida Cohort

In bivariate analyses, having symptoms of depression or anxiety, drinking alcohol (including both participants who did and did not meet the criteria for at-risk drinking), having incomplete disclosure to their close social network or not having a close social network, hearing stigmatizing comments from friends or co-workers, and seeing stigmatizing messages online, were associated with endorsing any ART concealment behavior (see Table 1). Conversely, those aged 50 and older and those with optimal ART adherence were less likely to endorse any ART concealment behaviors. The multivariable model was adjusted for significant variables from the bivariate analysis (age, drinking status, depression, anxiety, disclosure, and stigmatizing experiences). In the multivariable model, symptoms of depression, at-risk drinking, incomplete disclosure, and reporting not having a close social network were still significantly associated with ART concealment. Age, stigmatizing experiences, symptoms of anxiety, and drinking alcohol below at-risk levels were no longer associated with ART concealment (see Table 2).

Table 2.

Multivariable Models of Factors Associated with ART Concealment Behaviors among 416 persons with HIV in Florida

Characteristic	aOR (85% CI)a
Age
<35	Ref
35–49	0.96 (0.47–1.95)
50+	0.52 (0.27–1.01)
Symptoms of Anxiety
Absent	Ref
Present	1.43 (0.77–2.67)
Drinking Status
Non-drinker	Ref
Not at risk	1.59 (0.91–2.80)
At risk	2.38 (1.37–4.18)
Symptoms of Depression
Absent	Ref
Present	2.04 (1.15–3.66)
Disclosure to Close Social Network
Complete Disclosure	Ref
Incomplete Disclosure	2.94 (1.82–4.84)
No Close Network	3.65 (1.63–8.31)
Heard Stigmatizing Comments
Did not experience	Ref
Experienced	1.36 (0.82–2.25)
Saw Stigmatizing Comments Online
Did not experience	Ref
Experienced	1.34 (0.81–2.22)

^aModel included age, symptoms of anxiety and depression, drinking, disclosure to close social network, heard stigmatizing comments from friends or coworkers, and saw stigmatizing comments online

Endorsing at least one ART concealment behavior, hiding ART while having guests, and removing labels from ART bottles were significantly associated with suboptimal ART adherence in the bivariate analyses. Endorsing at least one ART concealment behavior and removing ART labels were no longer significantly associated when other variables associated with ART adherence were added to the model (see Table 3). Only hiding ART while having guests was associated with ART adherence in the adjusted model. In the mediation models, the indirect path was not significant for any of the four identified models where the exposures were drinking status, symptoms of depression, disclosure to close social network, and seeing stigmatizing messages online (data not shown).

Table 3.

Factors associated with suboptimal ART Adherence among 416 PWH in Florida

	ART Adherence 85%+	ART Adherence <85%	OR (95%CI)	aOR (95% CI)a
Endorsed 1+ Action to Conceal ART
No	230 (59.3)	8 (28.6)	Ref	Ref
Yes	158 (40.7)	20 (71.4)	3.64 (1.62–8.97)	1.69 (0.65–4.62)
Hid ART while having guests
No	274 (70.6)	10 (35.7)	Ref	Ref
Yes	114 (29.4)	18 (64.3)	4.33 (1.97–10.01)	2.87 (1.15–7.55)
Removed labels from ART bottles
No	296 (76.3)	14 (50.0)	Ref	Ref
Yes	92 (23.7)	14 (50.0)	3.22 (1.47–7.05)	1.80 (0.73–4.39)
Put ART in a different container
No	327 (84.3)	23 (82.1)	Ref	Ref
Yes	61 (15.7)	5 (17.9)	1.17 (0.38–2.96)	0.42 (0.12–1.24)
Changed pharmacies
No	371 (95.6)	27 (96.4)	Ref	Ref
Yes	17 (4.4)	1 (3.6)	1.16 (0.12–4.88)	0.47 (0.04–2.54)
Travelled 30+ miles to get ART
No	376 (96.9)	28 (100.0)	Ref	Ref
Yes	12 (3.1)	0 (0.0)	0.53 (0.004–4.20)	0.33 (0.002–2.86)

^aSeparate models were used for each of the variables in the table. All models adjusted for age, drinking status, injection and non-injection drug use, disclosure to their close social network, symptoms of depression, housing status, and saw stigmatizing messages on social media.

Discussion

Our analysis revealed that 43% of PWH participating in the Florida Cohort Study endorsed at least one ART concealment behavior that was done to prevent disclosure of HIV to others. Of these, the most frequently ssreported ART concealment behavior was hiding ART when having guests over. Furthermore, many participants removed ART labels or put their ART in a different bottle. Virological failure and reduced ART adherence have been linked to patient repacking of ART [58]. In addition, those with incomplete disclosure to their social network were 3 times more likely and those with no close social network were over 4 times more likely to endorse any ART concealment behavior compared with those who had complete disclosure to their social network. ART concealment behaviors were also positively associated with symptoms of depression and at-risk drinking patterns.

Since concealment of HIV status from friends and family members requires people to be more guarded in what they say and do to avoid inadvertent disclosure, it is possible that this increases the stress placed on PWH [35]. This is supported by the fact that several studies have reported that openness with friends and family has been linked to psychological benefits [25,59,60]. Additionally, disclosing HIV status to at least one friend has been found to be associated with improved immunological status [35]. Some studies have found that PWH might avoid disclosure not only because of worries of stigma, but because they have yet to accept their status themselves. A study in Australia found that men who have sex with men with strong social support networks were more accepting of their diagnoses and coped better with their diagnoses than those without strong social support networks and, as a result, could receive better social support and healthcare support [61]. However, it may not be safe for PWH to disclose their status to their social network in some circumstances as consequences of HIV-related stigma could include financial insecurity and violence [62–64]. Therefore, safe and selective disclosure of HIV status is crucial and mediation concealment may be a protective response to a hostile environment, and HIV providers should consider this in their care plans. Providers should work with their patients to ensure that medication is administered in a way that maintains privacy for patient safety and wellbeing, and ensure patients have a form of social support, especially if they do not have support from their social network. Long-acting injectables, which do not require PWH to keep medications at their residence, could be a promising option for many PWH to protect against inadvertent disclosure of their HIV status [65–67]. However, this might not be a feasible option for all PWH as it requires regular visits to their HIV care provider. Additionally, increased trips to a medical care facility can increase the risk for inadvertent disclosure and stigma [68,69].

A minority of participants in our study reported traveling at least 30 miles to obtain ART medication, which was highlighted as a pill-hiding mechanism. A few studies have revealed that participants are usually opposed to receiving care close to their homes due to increased perceived risk of HIV status disclosure [70]. Moreover, individuals that have experienced enacted stigma have been associated with long-distance travel to seek HIV care [71]. Thus, individuals avoiding stigma have developed solutions that result in an increase in the distance of travel for health care delivery despite efforts in the HIV healthcare sphere aimed at improving accessibility. This exemplifies how stigma continues to be a barrier to HIV care-seeking behaviors in a manner that can burden and negatively impact the quality of life and health outcomes of PWH [22].

Our study also revealed that those with suboptimal adherence were more likely to report at least one ART concealment behavior overall or removing labels from ART bottles compared to those with optimal adherence, but these associations were no longer significant in adjusted analyses. This finding indicates that factors associated with medication concealment behaviors, rather than the behaviors themselves, might be the most strongly associated with suboptimal ART adherence. Hiding ART while having guests was significantly associated with suboptimal adherence in both unadjusted and adjusted analyses, reflecting how stigma may prevent patients from disclosing their status to family, friends, and others who might visit their home. In totality, our results support findings in the literature that factors such as depression, substance use, disclosure, and stigma [17,18,23,38,53] are closely associated with ART adherence, and contribute to the literature by linking them to ART concealment behaviors as well. This highlights the need to improve ART adherence through stigma reduction efforts implemented through multilevel psychosocial interventions involving community members, social networks, family members, and healthcare workers [18,72].

Finally, it is possible that the lack of a significant relationship between those with medication hiding behaviors and adherence in adjusted analysis could be because, in some cases, medication hiding might be beneficial to adherence for patients at risk of domestic violence. There is evidence that women living with HIV who are at risk for experiencing intimate partner violence at home hide their ART to prevent their partners from finding out their status while attempting to adhere to their regimen [73]. While it is unclear whether medication hiding improves adherence in these situations, providers should again consider the needs of individual PWH and not consider PWH as a monolith when prescribing and administering care to their patients. Future research should consider domestic and other types of violence PWH who engage in medication hiding behaviors might be dealing with and examine not only the association with adherence but the balance of risks between adherence and individual safety.

Limitations

Data from the Florida Cohort are ideal for understanding the barriers and facilitators to ART adherence among a sample of PWH living in an area that experiences disproportionately negative HIV outcomes compared with the rest of the US. However, the population is largely recruited from clinics and most participants are engaged in care, so these results may not be generalizable to out-of-care PWH. In-care PWH tend to have better HIV-related outcomes (e.g., viral suppression among those retained in care was 89.5% in Florida in 2021 compared to an overall viral suppression rate of 69.3%), which may result from differing health behaviors. We also relied on self-reported ART adherence, which may overestimate adherence [74]. The potential misclassification of participants as being optimally adherent decreases the sample of suboptimally adherent participants and decrease our ability to detect differences. Additionally, our analysis does not account for certain variables that have demonstrated significant associations with adherence in the literature, most notably social support [60,75]. This was because these questions were not asked of participants in the study. We also used cross-sectional data as variables of interest were not available at follow-up visits. Future longitudinal studies could investigate how instances of these behaviors impact subsequent ART adherence. Further, our questions describing ART concealment are not exhaustive; they do not include all the ways in which people can conceal their medications. This question was written in consultation with HIV healthcare providers and community members with awareness of how people commonly hide their ART from others. However, future studies can expand upon this question. Qualitative interviews may be particularly valuable for capturing the diversity of methods employed by PWH to hide their ART and understand their reasons for and attitudes towards ART concealment.

Conclusion

Medication concealment behaviors were commonly endorsed among this sample of PWH and were associated with depression, at-risk drinking, incomplete disclosure to social network, and not having a social network. Medication concealment behaviors were associated with reduced likelihood of optimal ART adherence. However, only hiding medications while having guests was associated with adherence in adjusted multivariable analyses, while other behaviors were no longer associated. These findings suggest that PWH may want to receive ART medications in ways that ensure privacy. Further research is warranted to circumnavigate the challenges presented by ART concealment behaviors or ascertain how alternate forms of ART, such as long-acting injectables, can improve adherence. Future studies could also investigate the impacts of medication concealment on adherence to HIV preexposure prophylaxis.

Data and Code Availability:

The data used in this study can be requested from the Southern HIV and Alcohol Research Consortium (https://sharc-research.org/research/data/sharc-concepts-system/) and the code can be provided by the corresponding author upon request.