Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Feb 6.
Published in final edited form as: AIDS Behav. 2025 Feb 10;29(5):1640–1649. doi: 10.1007/s10461-025-04634-w

Impact of COVID-19-Related Disruptions on Antiretroviral Therapy Adherence Among Young Adults Living with HIV in Southern Uganda

Abel Zemedkun Girma 1, Rachel Brathwaite 1, Yvonne Karamagi 2, Fatumah Nakabuye 3, Noeline Nakasujja 4, William Byansi 5, Proscovia Nabunya 1, Ozge Sensoy Bahar 1, Fred M Ssewamala 1
PMCID: PMC12874465  NIHMSID: NIHMS2136885  PMID: 39928068

Abstract

We investigated how COVID-19-related disruptions influenced antiretroviral therapy (ART) adherence among young adults living with HIV (YALHIV) in Southern Uganda, a region with limited resources and high rates of HIV. Data were analyzed from 499 YALHIV aged 19 to 25, participating in the Suubi+Adherence-R2 COVID-19 Supplement study. The study measured COVID-19 disruptions using an 8-item Coronavirus Impact Scale and evaluated ART adherence through self-reported measures. Our analytical framework was informed by the Health Belief Model and generalized estimating equations were estimated. We find no statistically significant association between COVID-19 disruptions, as quantified by the COVID-19 Impact Score, and sub-optimal ART adherence (OR = 0.99, 95% CI [0.87–1.14]). However, findings revealed that being employed (OR = 1.99, 95% CI [1.07–3.71]) and older age (OR = 1.18, 95% CI [1.02–1.37]) was associated with higher likelihood of poor adherence highlighting the complex interplay between economic activity, working schedules, and health management. Other notable predictors included marital status, with cohabiting individuals showing decreased odds of poor adherence (OR = 0.25, 95% CI [0.08–0.74]) compared to single and separated YALHIV. These insights emphasize the need for multifaceted intervention strategies that consider both individual and systemic factors affecting ART adherence. Tailored interventions must address the socioeconomic challenges intensified by the pandemic and leverage the inherent resilience within this population to enhance ART adherence outcomes for YALHIV in challenging environments.

Trial registration

ClinicalTrials.gov, ID: NCT01790373.

Keywords: HIV, COVID-19, ART adherence, Young adults living with HIV, Health belief model

Introduction

HIV/AIDS continues to pose significant challenges to global public health. With the advent of antiretroviral therapy (ART), HIV has transformed into a chronic inflammatory condition to one that can be managed effectively, marking a significant milestone in the treatment of HIV/AIDS [1, 2]. Given the importance of ART in managing HIV, ART adherence has emerged as a critical concern in the long-term management of the disease [3, 4]. By ensuring high levels of adherence, it is possible to lower morbidity and mortality rates among people living with HIV, thereby substantially improving their quality of life [1]. Consequently, understanding and promoting adherence to ART has become a primary focus in the ongoing effort to manage HIV effectively.

The COVID-19 pandemic significantly impacted people living with HIV (PLWHIV), exacerbating challenges in reducing risks, accessing medical care, and treatment [5, 6]. For example, a study from South Africa showed a 28% decrease in ART initiations in 2020 compared to the previous year [7], with ART initiations declining sharply during the pandemic’s early waves and only partially recovering afterward. Another study also found immediate and significant decline in engagement with HIV services, including testing, treatment, and Pre-Exposure Prophylaxis (PrEP) use, at the onset of COVID-19 restrictions [8]. Such declines in ART initiation underscore the broader challenges the pandemic has posed for HIV treatment programs, as disruptions in access and care continuity may contribute to worse treatment outcomes and increased transmission risks. The necessity for social distancing and the overwhelming of healthcare systems by COVID-19 cases have also often prevented face-to-face clinical follow-ups [9]. Additionally, the economic fallout from the pandemic, including job losses and rising poverty levels, has made it increasingly difficult for vulnerable individuals to access transportation for medication pickup, further complicating adherence to ART [8, 9].

ART adherence among young adults and adolescents living with HIV frequently falls below ideal levels [10]. A multi-cohort study across African and Asian settings found that, among 13,001 adherence assessments, suboptimal adherence was observed in 7.3% of cases in the African cohort [11]. Similarly, short-term adherence rates to ART in South Africa were reported to range between 63% and 88%, underscoring the adherence challenges in the Sub-Saharan context [12]. Given the suboptimal adherence rates among young adults living with HIV (YALHIV), it becomes crucial to focus on this demographic to enhance treatment outcomes and quality of life. This emphasis is particularly pertinent in regions with significant HIV prevalence, such as Uganda. Uganda carries a substantial HIV/AIDS burden, with a national prevalence rate of 6% and a rate of 3.3% among young adults between the ages of 18 and 24 years [11, 12]. YALHIV in Southern Uganda have faced numerous disruptions due to the COVID-19 pandemic [6], including significant changes to their daily routines and impacts on family income, which can potentially complicate their adherence to treatment regimens.

Theoretical Framework: Health Belief Model (HBM)

Our study utilizes the Health Belief Model (HBM) to explore the determinants of suboptimal adherence to ART among YALHIV. Central to the HBM is the assertion that an individual’s health-related behaviors, such as the commitment to an ART regimen, are deeply influenced by their perceptions [13]. These include beliefs about the severity of and their susceptibility to their health condition, the perceived pros and cons of adhering to treatment, cues that trigger action, and their confidence in their ability to execute these health-related actions effectively [14]. This study utilizes the HBM as a framework to dissect how these various factors, compounded by the social disruptions stemming from the COVID-19 pandemic, collectively influence ART adherence behaviors among YALHIV in Southern Uganda. By integrating the pandemic’s impact, which extends from healthcare access disruptions to broader socio-economic challenges, we aim to uncover the intricate series of predictors and barriers that may be associated with poor ART adherence during this period.

Research shows that gender is often associated with differing health beliefs and access to healthcare resources, which can influence perceived severity and susceptibility [15]. Also, the presence of Generalized Anxiety Disorder (GAD) can amplify perceived obstacles to adherence, such as fears regarding side effects or the social stigma attached to illness, which might undermine an individual’s confidence in their ability to adhere to treatment [16]. The construct of our main exposure, the COVID-19 impact score, provides an opportunity to examine the myriad environmental and psychological hurdles introduced by the pandemic, likely reshaping perceived barriers and benefits for supporting ART adherence. Additionally, in the context of this study, Suubi+Adherence R2, participants’ allocation to different groups—where some received a financial empowerment intervention—introduces variability in exposure to critical support mechanisms [17]. This distinction acts as a significant cue, potentially influencing adherence behaviors by either bolstering motivation through enhanced financial security and empowerment or, conversely, by highlighting the disparities in access to such resources, affecting the consistency of ART adherence. Moreover, socio-demographic factors like age, school enrollment, marital and employment status, along with family size, may influence health beliefs, perceived barriers, and the practicality of maintaining consistent adherence to ART [1, 13]. These factors, collectively examined through the lens of the HBM in this study, provide an understanding of influences on ART adherence behaviors.

Therefore, we propose a comprehensive analysis to not only to identify the core factors affecting adherence but also to understand how the unprecedented disruptions caused by COVID-19 may have exacerbated poor adherence outcomes. Ultimately, our goal is to provide evidence that can inform targeted interventions to address unprecedented social obstacles and demographic barriers to improving ART adherence among YALHIV. Our alternate hypothesis for this study posits that heightened disruptions due to the COVID-19 pandemic are associated with lower levels of ART adherence among YALHIV in Southern Uganda.

Methods

Study Design & Setting

This research utilizes data from the Suubi+Adherence-R2 COVID-19 Supplement. The original Suubi+Adherence study is a longitudinal randomized-control trial conducted from 2012 to 2018. The study enrolled 702 adolescents aged 10 to 16 years to assess the effects of a family-centered economic empowerment intervention on ART adherence among YALHIV in Uganda [17]. Currently, these participants are transitioning into young adulthood and are part of an ongoing follow-up study extending five more years (2020–2025) entitled Suubi+Adherence-R2 [17]. The recruitment methodology, refined in earlier studies(Suubi Maka (R34MH081763) and Bridges to the Future (R01HD070727), involved leveraging clinics/health centers tied to local research partners for participant identification and recruitment. Medical personnel compiled a list of all qualifying youth patients from their clinic records, and during visits, research staff introduced the study to the caregivers of eligible youths younger than 18 years. Caregivers who showed interest signed written consent, while the youths provided assent. In instances where more than one child per family qualified, all children were included to prevent feelings of resentment within families. The study was carried out in six districts of Southwestern Uganda—Masaka, Kalungu, Lwengo, Rakai, Kyotera, and Bukomansimbi. These districts are significantly affected by HIV, showing a prevalence rate of 10.6%, notably surpassing the national average by over 4% [18].

The eligibility criteria for youth in the initial Suubi+Adherence investigation included: (1) being between the ages of 10 to 16 at the time of enrollment; (2) having a HIV-positive diagnosis as verified by medical documents and being aware of their status; (3) receiving ART and being under care at one of the 39 clinics involved in the study area; (4) residing with their family. All individuals who took part in the original Suubi+Adherence study was considered eligible for the Suubi+Adherence-R2 study. Data for this round of the study was gathered from 500 participants in the Suubi+Adherence-R2 study, as an administrative supplement aimed at exploring the effects of the COVID-19 pandemic on YALHIV [17]. By May 2021, all participants were over 18 years old, thus caregiver consent was no longer required. Participants completed interviewer-administered surveys retrospectively providing data on their experiences related to the changes and disruptions brought by the COVID-19 pandemic and their adherence ART treatment [17].

Measures

Outcome: Poor ART Adherence

The primary outcome of our study is defined as poor ART adherence. This measure was derived from participants’ self-reported responses to the question: “In the last 30 days, on how many days did you miss at least one dose of any of your ART medications?” To quantify adherence, we converted the self-reported number of days on which medication was missed into a percentage of days missed out of the last 30 to facilitate a standardized assessment of adherence levels across participants. In line with the findings from previous research [17, 18], we employed the Youden index to determine an optimal cut-point for categorizing adherence. The index identified an 89% threshold as the point that maximizes both the sensitivity and specificity of self-reported ART adherence in relation to achieving viral suppression, defined as ≤ 200 copies/mL. Achieving a viral load of less than 200 copies/mL is considered viral suppression, since the virus is undetectable and the person with HIV will not be able to transmit the virus to others [19]. Therefore, based on the established 89% cut-point, we dichotomized the linearized adherence data into two categories: those who reported missing doses on 11% or more of the past 30 days (indicative of poorer adherence) and those missing doses on less than 11% of days (indicative of better adherence).

Exposure: COVID-19 Related Resource Changes & Disruptions

This primary exposure for this study is defined as COVID-19 related changes, evaluated through a modified 8-item Coronavirus Impact Scale. This scale measured the pandemic’s impact across various life aspects, with participants responding on a four-level ordinal scale to reflect the severity of the disruption [20]. The response options, ranged from 0 to 3, denoted ‘no change’, ‘mild change’, ‘moderate change’, and ‘severe change’. Participants evaluated the pandemic’s influence on eight life areas including (1) routines, (2) family income/employment, (3) food access, (4) medical health care access, (5) mental health treatment access, (6) access to extended family and non-family social supports, (7) experiences of stress related to COVID-19 pandemic, and (8) stress and discord in the family [20]. The scale’s reliability was assessed, revealing a Cronbach’s alpha between 0.64 and 0.75, suggesting a moderate to strong internal consistency of the instrument [20]. This affirms the scale’s utility as a holistic indicator of changes caused by the COVID-19 pandemic. Therefore, we aggregated the scores from the eight domains of the COVID-19 Impact Scale for each participant to derive a comprehensive score ranging from 0 to 24, with higher scores indicating greater disruptions experienced during the COVID-19 pandemic.

Interactions: Gender and GAD Status

In this research, we included two interaction terms in the regression models: (1) gender-by-COVID-19 impact score and (2) GAD status-by-COVID-19 impact score. This allowed us to explore whether the effect of COVID-19 disruptions on ART adherence varies depending on participants’ gender or anxiety status. The inclusion of the gender-by-COVID-19 impact score interaction term is informed by existing literature that underscores gender differences in the experience and impact of health-related events [15]. These differences are not merely biological but are also shaped by societal norms, roles, and expectations that influence behaviors, access to resources, and vulnerability to stressors, including those related to a pandemic [13, 15]. By examining gender interactions, we aim to uncover nuanced insights into how the COVID-19 related disruptions has differentially influenced ART adherence among males and females. The rationale for including the GAD status-by-COVID-19 impact score, measured by the Generalized Anxiety Disorder-7 (GAD-7) questionnaire, as an interaction term lies in the profound impact of mental health on an individual’s ability to maintain treatment adherence, particularly under the stress of a global health crisis [16]. We dichotomized GAD-7 responses at a cutoff point of 5 to allows us to distinguish between minimal and non-minimal levels of anxiety [6, 19], acknowledging that even mild levels of anxiety can significantly affect health behaviors. This distinction is crucial in the context of lower-income countries, where mental health often receives limited attention, and the cultural perception and tolerance of anxiety may differ from those in higher-income settings [21].

By examining these interaction effects, this study seeks to explore the multifaceted relationships between COVID-19 disruptions, gender, anxiety levels, and their collective impact on poor ART adherence. This approach not only enhances our understanding of the barriers to adherence but also facilitates the identification of specific groups who may be at greater risk of poor adherence due to these intersecting factors.

Controls: Demographic Information

Essential demographic variables were recorded using a socio-demographic survey. The variables included age (years), educational attendance (currently enrolled or unenrolled in school), marital status (single/ separated, married, or cohabiting), employment situation (employed or unemployed), and the number of individuals per household.

Analysis

The HBM framework guided the inclusion of factors critical to effective ART management. For the analysis, binary logistic regression was employed, utilizing the binary categorization of ART adherence. In addition, our analysis incorporated the use of generalized estimating equations (GEE), which helps overcome certain constraints associated with logistic regression, especially in handling clustered binary data. By applying cluster-based robust standard errors, GEE effectively acknowledges and adjusts for the interdependencies of data points that are grouped within clinics [22]. This approach models the correlations of residuals within clusters of clinics to ensure a more nuanced analysis [22]. Furthermore, to control for potential confounders and remove biases, demographic factors were adjusted for in the models. Additionally, the allocation of participants to either the Suubi+Adherence-R2 intervention or control group was adjusted for, aiming to remove any bias when evaluating the effect of the pandemic on poor ART adherence.

To assess the significance of the gender-by-COVID-19 and GAD status-by-COVID-19 interaction effects, the relevant interaction terms were included in the regression model to evaluate their impact based on a p-value threshold of less than 0.05. If any interaction term ends up being statistically significant, a stratified analysis will be performed based on that variable. This stratification will allow us to dissect and understand the differences in ART adherence behaviors, offering a clearer view of how combined influences play out across different segments of the YALHIV population. Such insights are instrumental in customizing interventions to the needs of those most affected by the compounded effects of these factors.

Finally, to verify the stability and reliability of our regression analyses, we conducted a series of diagnostic evaluations, including DFFit, Cook’s distance, and variance inflation factors. These metrics provided insight on the reliability and stability of our findings and model. We established a threshold for statistical significance at a p-value of 0.05 for all performed analyses. Data management and analysis were carried out using STATA 17.

Results

Description of Participants’ Baseline Characteristics and Bivariate Association with ART Adherence

The final analysis sample consisted of 499 YALHIV, and a high majority (93.8%) had good ART adherence. The median age was 21 years, interquartile range (IQR) 19–22 years (Table 1). A small majority of participants (55.9%) were female and the chi-square test revealed a statistically significant relationship between gender and ART adherence (χ2(1) = 5.60, p = 0.016) (Table 1). Approximately 81.6% of the participants were not currently enrolled in school (Table 1). Approximately, 65.7% were single or separated, 10.2% were married and 24.1% were cohabiting (Table 1). In examining the association between current school enrollment and ART adherence, we found that there was a statistically significant relationship (χ2(2) = 6.72, p = 0.040) (Table 1). Relatively equal proportions of participants were employed, and a small majority (54.1%) resided with </= 4 family members in their household (Table 1).

Table 1.

Characteristics of YALHIV in Southern Uganda by levels of ART Adherence

Total n (%) Good ART Adherence n (%) Poor ART Adherence n (%) χ2 (df), p-value/z score, p-value
N (%) 499 (100) 468 (93.8) 31 (6.2)
Intervention Group Assignment χ2(1) = 0.58, 0.4471
Treatment 273 (54.7) 254 (54.3) 19 (61.3)
Control 226 (45.3) 214 (45.7) 12 (38.7)
Age (years), median (IQR) 21 (19–22) 21 (19–22) 22 (20–23) z=−1.16, 0.2452
Gender χ 2 (1) = 5.60, 0.016 1
Male 220 (44.1) 200 (42.7) 20 (64.5)
Female 279 (55.9) 268 (57.3) 11 (35.5)
Currently enrolled in school χ2(1) = 0.38, 0.5391
Yes 92 (18.4) 85 (18.2) 7 (22.6)
No 407 (81.6) 383 (81.4) 24 (77.4)
Marital Status χ2(2) = 6.72, 0.040 3
Single/Separated 328 (65.7) 301 (64.3) 27 (87.1)
Married 51 (10.2) 50 (10.7) 1 (3.23)
Cohabiting 120 (24.1) 117 (25.0) 3 (9.7)
Employment Status χ2(1) = 2.52, 0.1121
Unemployed 246 (49.3) 235 (50.2) 11 (35.5)
Employed 253 (50.7) 233 (49.8) 20 (64.5)
Household size χ2(2) = 0.10, 0.9531
</= 4 members 270 (54.1) 254 (54.3) 16 (51.6)
5–6 members 110 (22.0) 103 (22.0) 7 (22.6)
>/=7 members 119 (23.9) 111 (23.7) 8 (25.8)
GAD Category χ2(2) = 0.81, 0.3671
Minimal 296 (59.3) 280 (59.8) 16 (51.6)
Non-minimal 203 (40.7) 188 (40.2) 15 (48.4)
COVID-19 Impact Score, median (IQR) 9 (6–13) 9 (6–13) 8 (5–14) z = 0.03, 0.9752
Open in a new tab
1

p-value derived from Chi-square Test;

2

p-value derived from Kruskal-Wallis Test;

3

p-value derived from Fisher’s Exact Test;

Bolded values are significant;

IQR-Interquartile range

Approximately 59.3% of the participants had minimal levels of GAD, though there was not a statistically significant relationship between GAD status and poor ART adherence (Table 1). Finally, the participants had a median COVID-19 Impact Score of 9 (IQR 6–13), and the chi-squared test revealed no statistically significant relationship between COVID-19 disruptions and poor ART adherence (Table 1).

Analysis of Effect Modifiers (Interaction Effects): Gender-by-COVID-19 and GAD Status-by-COVID-19

Both gender and GAD status were evaluated for their roles in modifying the relationship between COVID-19-related disruptions and poor ART adherence among YALHIV in Southern Uganda. Despite initial considerations of these factors as potential effect modifiers in the context of ART adherence, the statistical models indicated that neither gender nor GAD status significantly altered the influence of COVID-19 disruptions on ART adherence outcomes. Specifically, the interaction term for gender and COVID-19 impact score was not statistically significant (OR = 0.97, 95% CI [0.86–1.09]). Similarly, the interaction term for non-minimal GAD status and COVID-19 impact score also did not reach statistical significance (OR = 1.01, 95% CI [0.87–1.18]). Consequently, in the final model construction, gender and non-minimal GAD status were incorporated as control variables alongside other demographic and psychosocial factors to ensure a comprehensive adjustment for potential confounders in assessing the primary exposure’s effect.

Logistic Regression Findings

This study examined the impact of COVID-19 disruptions and factors associated with poor ART adherence among YALHIV in Southern Uganda. The Odds ratios and 95% CI derived using GEE, as detailed in Table 2, highlights several significant predictors of poor ART adherence among YALHIV.

Table 2.

Binary logistic regression model showing factors associated with poor ART adherence among YALHIV in Southern Uganda

OR (95% CI) p-value
COVID-19 Impact Score 0.99 (0.87–1.14) 0.931
Intervention Group Assignment (ref: Treatment)
Control 0.81 (0.34–1.92) 0.633
Age 1.18 (1.02–1.37) 0.027
Gender (ref: female)
Male 2.12 (1.17–3.84) 0.013
Currently enrolled in school (ref: No)
Yes 1.63 (0.66–4.03) 0.286
Marital Status (ref: single/separated)
Married 0.22 (0.03–1.60) 0.134
Cohabiting 0.25 (0.08–0.74) 0.012
Employment Status (ref: unemployed)
Employed 1.99 (1.07–3.71) 0.029
Household size (ref: </=4 members)
5–6 members 0.94 (0.37–2.38) 0.892
>/=7 members 1.12 (0.53–2.38) 0.759
GAD Category (ref: minimal GAD)
Non-minimal GAD 1.78 (0.72–4.42) 0.211
Pseudo R 2 0.083
Wald X 2 53.27
Sig of the model < 0.001
Model performance, AUC 71.9
Open in a new tab

AUC: Area Under the ROC Curve; OR: Odds Ratio; CI: Confidence Interval;

Bolded values are significant

Age emerged as a significant predictor, with older participants more likely to exhibit poor ART adherence (OR = 1.18, 95% CI [1.02–1.37]), suggesting that as YALHIV age, they may face increasing challenges in adhering to their medication regimens. Gender also significantly influenced ART adherence; males demonstrated over twice the odds of poor adherence compared to females (OR = 2.12, 95% CI [1.17–3.84]), indicating a gender disparity in ART adherence.

Marital status and living arrangements further influenced adherence. Participants cohabiting with a partner were significantly less likely to have poor ART adherence compared to those who were single or separated (OR = 0.25, 95% CI [0.08–0.74]), suggesting the potential supportive role of partners in managing HIV treatment. Employment status was another significant factor; being employed was associated with almost twice the odds of poor adherence (OR = 1.99, 95% CI [1.07–3.71]), possibly reflecting the challenges of balancing work commitments with consistent medication intake.

The analysis did not find significant associations between school enrollment, household size, and GAD status with ART adherence, indicating that these factors might not directly impact ART adherence behaviors in this cohort. Notably, the COVID-19 Impact Score did not significantly affect poor ART adherence (OR = 0.99, 95% CI [0.87–1.14]), suggesting that the overall perceived impact of the pandemic’s disruptions may not directly be associated with adherence behaviors among YALHIV in Southern Uganda.

Model Performance and Significance

The model exhibited moderate explanatory power with a Pseudo R2 of 0.083 and was statistically significant (Wald X2 = 53.27, p < 0.001). This suggests that 8.3% of the variance in ART adherence can be explained by the model. The model’s performance, assessed by the Area Under the ROC Curve (AUC), was 71.9%, indicating a good ability to discriminate between those with poor versus good ART adherence.

Diagnostic and Assumption Testing Results

Diagnostic evaluations were performed on the logistic regression model to establish its robustness and validity, with a particular focus on multicollinearity, influence, and residuals. The assessment of multicollinearity revealed that all predictors had VIF below 5, indicating no significant multicollinearity within the model; the mean VIF was low at 1.2. Regarding influential observations, Cook’s distances for all data points were well below the threshold of concern, with the highest recorded value being only 0.045, significantly less than the benchmark value of 1 that might indicate an influential case. DFFits, another measure of influence, also did not exceed the critical value of 1, further supporting the absence of overly influential data points in our analysis.

However, an examination of standardized residuals identified 31 cases with values exceeding 3, with the highest among these reaching 4.1. While these values surpass the conventional cutoff of 3, suggesting potential outliers or influential observations, a closer inspection revealed that all these cases belonged to the group with poor ART adherence. This led us to conduct a thorough review for possible data entry and coding inaccuracies, yet no errors were detected, confirming the authenticity of these observations. Given the value of Cook’s distance and DFFits met the criteria of acceptability, and the specific grouping of the observations with high residuals, it was determined appropriate to retain these cases in the model. This examination not only reinforces the model’s integrity but also underscores its capacity to reflect the nuanced dynamics of ART adherence among the study population.

Discussion

This study aimed to understand the dynamics of ART adherence among YALHIV in Southern Uganda, with a particular focus on the role of COVID-19-related disruptions. Contrary to our hypothesis, our analysis revealed that the overall impact of COVID-19 disruptions, as quantified by the COVID-19 Impact Score, did not significantly influence ART adherence among the participants. This finding diverges from the anticipated hypothesis that the pandemic’s social disruptions would directly be associated with adherence behaviors. This may be due to first, the resilience and adaptability demonstrated by YALHIV which could have mitigated the potential negative impacts of COVID-19 disruptions on ART adherence. Previous research has highlighted the capacity of individuals living with chronic conditions to employ coping strategies that help maintain treatment adherence amidst adversities [21, 22]. Secondly, the extensive public health messaging and community support mobilized in response to the pandemic might have played a protective role, especially in contexts where HIV awareness is high and support systems are active. A recent study highlights several approaches implemented in Uganda to support treatment continuity amid lockdowns [23]. These strategies included extending multi-month dispensing, home-based ART delivery, and leveraging Community Drug Distribution Points to facilitate ART refills at outreach sites closer to patients [24]. Additionally, for many young adults, spending more time at home due to lockdowns may have created a supportive environment for ART adherence. Being at home may provide a more private and less stigmatizing environment for managing daily medication routines, as individuals experiencing high levels of HIV stigma have been shown to exhibit lower ART adherence compared to those facing less stigma [25].

An interesting aspect of our analysis was the absence of significant interaction effects between COVID-19 disruptions and both gender and GAD status on ART adherence. This null finding challenges some of the anticipated dynamics posited by the HBM and raises important questions about the interaction between individual characteristics and external stressors. One possibility is that the effects of such disruptions are more universal and do not disproportionately affect adherence based on gender or anxiety levels within this context. It could also suggest that YALHIV have developed means to counter the specific challenges posed by the pandemic, which are equally effective across different genders and levels of GAD symptoms. Alternatively, the interplay of these factors may be more subtle than can be detected by the binary interaction terms used in this study. The absence of significant findings for these interactions prompts further investigation into the complex and potentially non-linear relationships that might exist among these variables.

However, our analysis identified several significant predictors of poor ART adherence, including age, gender, marital status, and employment status, which align with existing literature on factors associated with ART adherence [13, 26]. For instance, being male was associated with lower adherence, which could reflect gender-specific barriers to accessing care or societal norms impacting health-seeking behaviors [15]. Notably, employment status was identified as a considerable predictor, with our findings revealing that individuals who were employed were nearly twice as likely to demonstrate poor ART adherence. This finding is counterintuitive as employment is typically associated with better health outcomes and behaviors due to increased economic stability [27]. However, for YALHIV in the Suubi+Adherence R2 study, being employed might introduce unique challenges, such as time constraints or workplace stigma, that complicate adherence to ART regimens. The demands of maintaining employment could potentially limit the time and energy available for healthcare management, leading to missed doses. Additionally, the fear of disclosing one’s HIV status in the workplace might deter individuals from adhering to their medication schedules [28].

The HBM, which served as the theoretical backbone of our investigation, provides a valuable lens through which to interpret the findings related to the determinants of poor ART adherence among YALHIV. Our analysis revealed that employment status significantly influenced ART adherence, underscoring the HBM’s assertion regarding the impact of perceived barriers. For individuals engaged in employment, the challenges of managing work responsibilities alongside ART regimens potentially heightened perceived barriers, influencing their adherence behaviors negatively. Perhaps obtaining time-off from work to collect ART medication at clinics was challenging to obtain and or justify since it may lead to missed income. This finding exemplifies how socioeconomic factors intertwine with health beliefs to shape health behaviors, suggesting that interventions aimed at improving ART adherence need to address these perceived barriers directly. Further research on the specific employment-related barriers to adherence will be useful for understanding how to overcome these challenges.

Despite the anticipated role of COVID-19 disruptions as a significant factor affecting ART adherence, our results did not identify the pandemic’s impact as a substantial predictor. This outcome might reflect a complex interplay of heightened perceived severity and susceptibility due to the pandemic, counterbalanced by effective coping mechanisms or support systems that reduced its negative influence. The HBM underscores the importance of cues to action, which in the context of the pandemic, may have taken various forms, from increased health communication to community support initiatives, potentially sustaining adherence levels despite the challenges [1, 13].

Strengths and Limitations

The interpretation of the findings from this study are subject to some limitations due to its design, which may influence the results. A primary limitation lies in the reliance on self-reported data for measuring ART adherence. Despite using a validated question and operationalization method to assess adherence in a manner indicative of viral suppression, the nature of self-reporting can introduce response bias. Additionally, while the COVID Impact Scale used in this study captures a broad range of pandemic-related disruptions relevant to ART adherence, it was not specifically validated in the context of East Africa or among young adults. This limits its contextual specificity, although the scale’s constructs align with known challenges faced by young adults in the region during the pandemic [6, 9]. Participants might under-report or over-report their adherence due to factors such as recall bias, social desirability bias, or misinterpretations of the survey questions, which are all common biases in survey-based research and could impact the accuracy of the reported adherence levels. Additionally, our sample was drawn from a supplement of a main study, where participants were already enrolled in a longitudinal study. This may introduce some selection bias, as they might have been more engaged with adherence protocols due to their ongoing participation, potentially increasing their adherence rates compared to the broader population of YALHIV.

To mitigate the limitations associated with self-reported data and other study design aspects, several methodological strengths were integrated into our analysis. One significant strength is the use of cluster-based robust standard errors, which enhances the reliability of our findings by accounting for the data’s clustered nature. This approach allows for the consideration of potential correlations within clusters of observations, offering a more nuanced understanding of ART adherence among YALHIV in Southern Uganda. Additionally, the study benefits from a comprehensive analytical framework informed by the Health Belief Model, providing a deep understanding of the multifaceted influences on ART adherence. This theoretical grounding helps to contextualize the findings within broader health behavior research, highlighting the interplay between individual perceptions, factors, and adherence behaviors.

Implications

The absence of a significant direct association between COVID-19 disruptions and ART adherence does not diminish the pandemic’s profound effects on the healthcare system and individuals’ lives. This finding prompts a broader consideration of how social disruptions, while not directly altering adherence behaviors, may still influence the well-being and health outcomes of YALHIV through indirect pathways. For example, economic hardships and the psychological toll of the pandemic may exacerbate existing challenges faced by individuals living with HIV, from mental health issues to accessing healthcare services [6, 9]. Additional mediation analyses may help explain the pathways through which COVID-19 disruptions may have impacted ART adherence among YALHIV.

This insight underscores the importance of developing multifaceted intervention strategies that address not only the direct determinants of ART adherence but also the broader socio-economic and psychological contexts in which YALHIV navigate their care. Interventions that enhance economic stability, provide mental health support, and promote accessible, youth-friendly healthcare services could be crucial in supporting ART adherence in the face of future societal disruptions. Moreover, the significance of demographic factors such as gender and employment status in influencing ART adherence highlights the need for targeted approaches that consider the unique challenges faced by different sub-groups within the YALHIV population. Tailoring interventions to address the specific barriers encountered by males, for instance, or providing additional support for employed individuals, could contribute to more effective adherence support strategies. Future research could benefit from using a validated, context-specific COVID impact measure and a mixed-methods design, providing a richer understanding of ART adherence challenges in this demographic.

Conclusion

In conclusion, while the anticipated overall impact of COVID-19-related disruptions on ART adherence among YALHIV in Southern Uganda was not observed, this study brings to light the complex interplay of factors influencing adherence. It reinforces the need for holistic, adaptable intervention strategies that can address both the direct and indirect influences on ART adherence. As we move forward, acknowledging and addressing the broader social determinants of health will be key in supporting the ongoing efforts to improve ART adherence and overall outcomes for YALHIV, especially in the context of global challenges such as the COVID-19 pandemic.

Acknowledgements

Acknowledgments to the field coordinators—Fatuma Nakabuye, Flavia Namuwonge, Vicent Ssentumbwe, Phionah Namatovu, and all the young adults who participated in the study. In addition, special thanks to Dr. Kim Johnson for her invaluable assistance with proposal development and manuscript review.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The COVID-19 Supplement (3D43TW011541-02S1) was funded by Fogarty International Center (FIC) of the National Institutes of Health as a supplement leveraging an R01 study funded by Eunice Kennedy Shriver National Institute of Child Health & Human Development (NICHD) (R01HD074949—FM Ssewamala, P Nabunya & O Sensoy-Bahaar, MPI) and a D43 Training Program titled The Child mental Health in HIV-impacted Low-resource settings in Developing countries: Global Research Fellowship (D43TW011541 — FM Ssewamala, N Nakasujja & M McKay, MPIs). The supplement was to examine the impact of the COVID-19 epidemic response and social distancing measures on adolescent’s mental health disorders (Support for Mentorship for Drs. Yvonne Karamagi and William Byansi through Makerere University).

Footnotes

Ethical Approval This study was approved by the Institutional Review Board at Washington University in St. Louis (IRB # 201704066) and the in-country local IRBs in Uganda: Makerere University School of Public Health Review Committee (Protocol # 210), and Uganda National Council of Science and Technology (UNCST, SS 2969).

Consent All participants provided written consent to prior to enrollment in the study. They were assured that participation is voluntary and medical care will not be affected. Procedures for confidentiality, including handling of data, Certificates of Confidentiality, and HIPAA were explained.

Competing Interests The authors have no competing interests to declare that are relevant to the content of this article.

Data/Code Availability

The data generated during and/or analysed during the current study are not publicly available due to the following reasons but are available from the corresponding author on reasonable request. Ethics approval, participant permissions, and all other relevant approvals were granted for this data sharing. Our study participants are adolescents living with HIV and highly stigmatized. Thus, we stated in the consent form that only de-identified individual-level data may be shared outside of the research team and only upon completion of the following conditions: A formal research question is specified a priori; Names, affiliations, and roles of any other individuals who will access the shared data; The deliverable(s)—e.g., manuscript, conference presentation—are specified a priori; Proper credit and attribution—e.g., authorship, co-authorship, and order—for each deliverable are specified a priori; A statement indicating an understanding that the data cannot be further shared with any additional individual(s) or parties without the PI’s permission; IRB approval for use of the data (or documentation that IRB has determined the research is exempt). The requestors are expected to handle converting electronic formats (though the research team will consider converting to tab-delimited text format if possible). These conditions are in line with Washington University’s Office of Research Administration’s data sharing agreement. Our study was reviewed by the following Ethics Review Boards: Washington University in St. Louis (IRB # 201704066); Makerere University School of Public Health (Protocol 210); Uganda National Council for Science and Technology (Protocol SS 2969). Data Access Requests can be sent to any of the following Associate Deans—at Washington University’s Brown School. Provided the conditions outlined above are met, there should not be concern about data sharing. The team is open to data sharing provided the points outlined above, which were part of the study protocol, data sharing plan, and consenting process, are met. Siomari Collazo-Colón, JD, Associate Dean for Administration, Hillman Hall, Room 254, Brown School, Washington University in St. Louis [o] 314.935.8675 [f] 314.935.8511 [e] scollazo@wustl.edu; OR the William E. Gordon Distinguished Professor Associate Dean for Transdisciplinary Faculty Research Professor of Medicine, Washington University School of Medicine Goldfarb, Room 343 Brown School Washington University in St. Louis [o] 314.935.8521 [e] fms1@wustl.edu (the corresponding author).

References

  • 1.Suryana K, Suharsono H, Indrayani AW, Wisma Ariani LNA, Putra WWS, Yaniswari NMD. Factors associated with anti-retroviral therapy adherence among patients living with HIV during the COVID-19 pandemic: a cross-sectional study. Front Psychiatry. Sep. 2022;13:824062. 10.3389/fpsyt.2022.824062. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Hornschuh S, Dietrich JJ, Tshabalala C, Laher F. Antiretroviral treatment adherence: knowledge and experiences among adolescents and young adults in Soweto, South Africa. AIDS Res Treat. 2017;2017:5192516. 10.1155/2017/5192516. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Hirsch JD, Gonzales M, Rosenquist A, Miller TA, Gilmer TP, Best BM. Antiretroviral therapy adherence, medication use, and health care costs during 3 years of a community pharmacy medication therapy management program for Medi-Cal beneficiaries with HIV/AIDS, J. Manag. Care Pharm. JMCP, vol. 17, no. 3, pp. 213–223, Apr. 2011, 10.18553/jmcp.2011.17.3.213 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Achappa B, Madi D, Bhaskaran U, Ramapuram JT, Rao S, Mahalingam S. Adherence to Antiretroviral Therapy Among People Living with HIV. North Am. J. Med. Sci 5(3):220–223, Mar. 2013, 10.4103/1947-2714.109196 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Jiang H, Zhou Y, Tang W. Maintaining HIV care during the COVID-19 pandemic. Lancet HIV. May 2020;7(5):e308–9. 10.1016/S2352-3018(20)30105-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Girma AZ et al. Feb., The impact of COVID-19 changes and disruptions on generalized anxiety disorder among young adults living with HIV (YLHIV) in Uganda. J. Health Psychol 2024. 10.1177/13591053241228205 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Benade M, Long L, Rosen S, Meyer-Rath G, Tucker J-M, Miot J. Reduction in initiations of HIV treatment in South Africa during the COVID pandemic. BMC Health Serv Res. Mar. 2022;22:428. 10.1186/s12913-022-07714-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Yao D, et al. The impact of COVID-19 restrictions on HIV prevention and treatment services for key populations in South Africa: an interrupted time series analysis. BMC Public Health. Sep. 2024;24(1):2386. 10.1186/s12889-024-19679-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Shiau S, Krause KD, Valera P, Swaminathan S, Halkitis PN. The Burden of COVID-19 in People Living with HIV: A Syndemic Perspective. AIDS Behav. 24(8):2244–2249, Aug. 2020, 10.1007/s10461-020-02871-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Kim S-H, Gerver SM, Fidler S, Ward H. Adherence to antiretroviral therapy in adolescents living with HIV: systematic review and meta-analysis. AIDS Lond Engl. Aug. 2014;28(13): 1945–56. 10.1097/QAD.0000000000000316. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Bijker R, et al. Adherence to antiretroviral therapy for HIV in sub-saharan Africa and Asia: a comparative analysis of two regional cohorts. J Int AIDS Soc. Mar. 2017;20(1):21218. 10.7448/IAS.20.1.21218. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Moosa A, Gengiah TN, Lewis L, Naidoo K. Long-term adherence to antiretroviral therapy in a South African adult patient cohort: a retrospective study. BMC Infect Dis. Sep. 2019;19(1):775. 10.1186/s12879-019-4410-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Addo MK, Aboagye RG, Tarkang EE. Factors influencing adherence to antiretroviral therapy among HIV/AIDS patients in the Ga West Municipality, Ghana. IJID Reg. Apr. 2022;3:218–25. 10.1016/j.ijregi.2022.04.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Jones CL, Jensen JD, Scherr CL, Brown NR, Christy K, Weaver J. The Health Belief Model as an explanatory Framework in Communication Research: exploring parallel, serial, and Moderated Mediation. Health Commun. 2015;30(6):566–76. 10.1080/10410236.2013.873363. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Berg KM, Demas PA, Howard AA, Schoenbaum EE, Gourevitch MN, Arnsten JH. Gender Differences in Factors Associated with adherence to antiretroviral therapy. J Gen Intern Med. Nov. 2004;19(11):1111–7. 10.1111/j.1525-1497.2004.30445.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Wykowski J, Kemp CG, Velloza J, Rao D, Drain PK. Associations between Anxiety and Adherence to Antiretroviral Medications in Low- and Middle-Income Countries: A Systematic Review and Meta-analysis, AIDS Behav., vol. 23, no. 8, pp. 2059–2071, Aug. 2019, 10.1007/s10461-018-02390-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Ssewamala FM et al. Suubi+Adherence-Round 2: a study protocol to examine the longitudinal HIV treatment adherence among youth living with HIV transitioning into young adulthood in Southern Uganda. BMC Public Health. 2021;21:179. 10.1186/s12889-021-10202-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.The Republic of Uganda. The HIV and AIDS Uganda Country Progress Report, 2014. [Online]. Available: http://www.unaids.org/sites/default/files/country/documents/UGA_narrative_report_2015.pdf
  • 19.Center for Disease Control and, Prevention HIV. Treatment as Prevention| HIV Risk and Prevention| HIV/AIDS| CDC. Accessed: Mar. 19, 2024. [Online]. Available: https://www.cdc.gov/hiv/risk/art/index.html
  • 20.Stoddard J, et al. The Coronavirus Impact Scale: construction, validation, and comparisons in diverse clinical samples. Jaacap Open. Jun. 2023;1(1):48–59. 10.1016/j.jaacop.2023.03.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Patel V. Mental health in low- and middle-income countries. Br Med Bull. Jan. 2007;1, pp. 81–96, 81–2. 10.1093/bmb/ldm010. [DOI] [PubMed] [Google Scholar]
  • 22.Glantz SA, Slinker BK, Neilands TB. Regression with a Qualitative Dependent Variable: Logistic Regression. In Primer of Applied Regression and Analysis of Variance, 3e, Book, Section vols., New York, NY: McGraw-Hill Education, 2017. Accessed: Aug. 09, 2023. [Online]. Available: accessbiomedicalscience.mhmedical.com/content.aspx?aid=1141901984. [Google Scholar]
  • 23.Zakumumpa H, Tumwine C, Milliam K, Spicer N. Dispensing antiretrovirals during Covid-19 lockdown: re-discovering community-based ART delivery models in Uganda. BMC Health Serv Res. Jul. 2021;21(1):692. 10.1186/s12913-021-06607-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Bailey LE, Siberry GK, Agaba P, Douglas M, Clinkscales JR, Godfrey C. The impact of COVID-19 on multi-month dispensing (MMD) policies for antiretroviral therapy (ART) and MMD uptake in 21 PEPFAR-supported countries: a multi-country analysis, J. Int. AIDS Soc, vol. 24, no. Suppl 6, p. e25794, Oct. 2021, 10.1002/jia2.25794 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Rintamaki LS, Davis TC, Skripkauskas S, Bennett CL, Wolf MS. Social stigma concerns and HIV medication adherence. AIDS Patient Care STDs. May 2006;20(5):359–68. 10.1089/apc.2006.20.359. [DOI] [PubMed] [Google Scholar]
  • 26.Chatoo A, Lee S. Association of coping strategies and medication adherence: a systematic review. Innov Pharm. Dec. 2022; 13(3). 10.24926/iip.v13i3.4991. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Nachega JB, et al. Association between antiretroviral therapy adherence and employment status: systematic review and meta-analysis. Bull World Health Organ. Jan. 2015;93(1):29–41. 10.2471/BLT.14.138149. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Martinez J et al. Feb., The Impact of Stigma on Medication Adherence Among HIV-Positive Adolescent and Young Adult Females and the Moderating Effects of Coping and Satisfaction with Health Care. AIDS Patient Care STDs. 26(2): 108–115, 2012, 10.1089/apc.2011.0178 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

The data generated during and/or analysed during the current study are not publicly available due to the following reasons but are available from the corresponding author on reasonable request. Ethics approval, participant permissions, and all other relevant approvals were granted for this data sharing. Our study participants are adolescents living with HIV and highly stigmatized. Thus, we stated in the consent form that only de-identified individual-level data may be shared outside of the research team and only upon completion of the following conditions: A formal research question is specified a priori; Names, affiliations, and roles of any other individuals who will access the shared data; The deliverable(s)—e.g., manuscript, conference presentation—are specified a priori; Proper credit and attribution—e.g., authorship, co-authorship, and order—for each deliverable are specified a priori; A statement indicating an understanding that the data cannot be further shared with any additional individual(s) or parties without the PI’s permission; IRB approval for use of the data (or documentation that IRB has determined the research is exempt). The requestors are expected to handle converting electronic formats (though the research team will consider converting to tab-delimited text format if possible). These conditions are in line with Washington University’s Office of Research Administration’s data sharing agreement. Our study was reviewed by the following Ethics Review Boards: Washington University in St. Louis (IRB # 201704066); Makerere University School of Public Health (Protocol 210); Uganda National Council for Science and Technology (Protocol SS 2969). Data Access Requests can be sent to any of the following Associate Deans—at Washington University’s Brown School. Provided the conditions outlined above are met, there should not be concern about data sharing. The team is open to data sharing provided the points outlined above, which were part of the study protocol, data sharing plan, and consenting process, are met. Siomari Collazo-Colón, JD, Associate Dean for Administration, Hillman Hall, Room 254, Brown School, Washington University in St. Louis [o] 314.935.8675 [f] 314.935.8511 [e] scollazo@wustl.edu; OR the William E. Gordon Distinguished Professor Associate Dean for Transdisciplinary Faculty Research Professor of Medicine, Washington University School of Medicine Goldfarb, Room 343 Brown School Washington University in St. Louis [o] 314.935.8521 [e] fms1@wustl.edu (the corresponding author).

RESOURCES