The effect of family support on self-reported adherence to ART among adolescents perinatally infected with HIV in Uganda: A mediation analysis

Abstract

Introduction:

This study examined the mechanisms through which family support affects self-reported adherence to antiretroviral therapy among adolescents perinatally infected with HIV in Uganda.

Methods:

Longitudinal data from 702 adolescent boys and girls (10–16 years) were analyzed. Structural equation models were conducted to assess the direct, indirect, and total effects of family support on adherence.

Results:

Results showed a significant indirect effect of family support on adherence (β = .112, 95% confidence interval [CI]: 0.052–0.173, p < .001). Specific indirect effects of family support through saving attitudes (β = .058, 95% CI: 0.008–0.108, p = .024), and communication with the guardian (β = .056, 95% CI: 0.012–0.100), p = .013), as well as the total effect of family support on adherence (β = .146 (95% CI: 0.032–0.259, p = .012), were statistically significant. Mediation contributed 76.7% of the total effects.

Conclusion:

Findings support strategies to help promote family support and strengthen open communication between adolescents living with HIV and their caregivers.

1 |. INTRODUCTION

An estimated 1.7 million children under < 15 years are living with HIV globally, the majority of these live in sub-Saharan Africa (SSA) (UNAIDS, 2021). Advances in antiretroviral therapy (ART) over the last decade have greatly improved health outcomes, disease progression and life expectancy for people living with HIV (Burger et al., 2022; Forsythe et al., 2019). However, these benefits depend greatly on an individual’s ability to adhere to prescribed medication regimens (Peltzer & Pengpid, 2013). While the World Health Organization established guidelines recommending ART to be initiated for anyone living with HIV regardless of CD4 count (World Health Organization, 2015), ART coverage for children is still lower than that of adults in developing countries, especially in SSA (Ahmed & Lemma, 2019).

HIV medication adherence is complicated for adolescents in low resource settings (Enane et al., 2018; Mark et al., 2017). Globally, only 40% of adolescents on ART had suppressed viral load compared to 67% of adults (UNAIDS, 2021). Adolescents living with HIV (ALHIV) have to confront the psychosocial issues, maintain adherence to drugs and learn to negotiate sexual relationships while undergoing rapid physical and psychological development (Enane et al., 2018; Hazra et al., 2010). Factors associated with suboptimal adherence to ART among ALHIV in low-resource settings include developmental characteristics, social and economic factors such as poverty, stigma and discrimination, burdensome ART regimens, lack of social support, serostatus nondisclosure, limited adherence clinical knowledge, clinic environments, including the lack of adolescent-friendly services to facilitate access to the limited available health care services (Ammon et al., 2018; Hudelson & Cluver, 2015). Yet, suboptimal adherence is associated with increased risk of viral progression, HIV transmission and viral resistance (Ajose et al., 2012; Bangsberg et al., 2001; San-Andres et al., 2003). As such, promoting adherence remains an important care element for children and ALHIV.

1.1 |. Family support and medication adherence

Social support, especially from family members, has been shown to influence adherence to treatment protocols (DiMatteo, 2004a; Shahin et al., 2021; Williams et al., 2006). Social support can be categorized into emotional (e.g., love, trust, care, and empathy), instrumental (e.g., provision of tangible or practical support), informational (e.g., provision of advice) and appraisal support (e.g., self-esteem, confidence building) (Cohen & Wills, 1985; House, 1981; Langford et al., 1997). Each type of support provides unique resources to mitigates stressors (Cohen & McKay, 1984; House, 1981). For example, while emotional support may alleviate stress through the provision of psychological resources, including communication, listening and empathy, instrumental support provides tangible resources and assistance that directly alleviate stressors.

In HIV-related studies, perceived emotional support from family members (Poudel et al., 2015), family cohesion and social support from caregivers (Damulira et al., 2019), as well as caregiver characteristics such as having an adult other than the biological parent as the primary caregiver, caregiver education level (Williams et al., 2006), and caregiver’s knowledge about the importance of treatment and adherence (Arage et al., 2014; Biadgilign et al., 2009), were all associated with improved adherence. Family cohesion and positive child-caregiver communication have also been associated with adherence self-efficacy (Nabunya et al., 2020). Nonadherence was related to worse parent–child communication, higher levels of caregiver stress, lower caregiver quality of life and worse caregiver cognitive functioning (Mellins et al., 2004). Instrumental support—the focus of this study, involves tangible, or practical forms of support, such as task assistance, funds, and direct interventions (Cohen & McKay, 1984; French et al., 2018; Friedman & King, 1994). Instrumental support has been documented as having the strongest effect on adherence compared to other forms of support (DiMatteo, 2004b). Family members promote treatment adherence by facilitating access to HIV care through the provision of financial support, having accurate knowledge about medication and adherence, providing motivation to adhere to treatment, directly reminding or helping the patient to take medication, meal preparation and facilitating medical transportation to health clinics (Fredriksen-Goldsen et al., 2011).

While several studies have investigated the role of family support, the mechanisms through which family support impacts ART adherence is rarely investigated. This study examines the mediators of the effect of family support, specifically, instrumental support, on self-reported ART adherence among ALHIV in Uganda. Comprehensive psychosocial services for individuals living with HIV in Uganda, including access to comprehensive counseling services and home-based care services are very low and some have been phased out completely, leaving families with the primary responsibility of providing care and support with no or minimal government support. As such, understanding the mechanisms through which specific types of support from family members affect adherence, is essential in promoting and facilitating treatment adherence, especially among children and adolescents, whose adherence needs depend largely on their caregivers.

2 |. METHODS

2.1 |. Sample and setting

Data from the Suubi+Adherence study (2012–2018), a randomized experimental study were analyzed. The study examined a family-based economic empowerment intervention aimed at improving adherence to medication for ALHIV in rural Uganda (Ssewamala et al., 2019). A total of 702 ALHIV between ages 10–16 years were enrolled in the study. Adolescents were eligible to participate if they were HIV-positive as confirmed by medical records, and disclosed to, prescribed antiretroviral therapy, lived within a family, not an institution, and enrolled in one of the 39 health clinics in the study region. Participants were recruited from health clinics located in Rakai, Masaka, Lwengo, Lyantonde, Bukomasimbi, and Kalungu Districts—a region with high prevalence for HIV, compared to the national average (Uganda AIDS Commission, 2021). Details on participant recruitment and the intervention are provided in the study protocol (Ssewamala et al., 2019) and elsewhere (Ssewamala et al., 2020; Tozan et al., 2021).

2.2 |. Data collection and measures

Data was collected using a 90-min interviewer-administered assessment battery by trained research assistants. Assessment measures and all study related materials were translated into Luganda—the local language and back translated into English to ensure consistency. All measures have been tested in our previous studies in the study region (Nyoni et al., 2019; Osuji et al., 2018; Ssewamala & Ismayilova, 2009). A description of measures utilized in this analysis is provide below.

2.2.1 |. Outcome variable

Self-reported adherence was measured using six items: (1) When was the last time you missed any of your medication? (1 = “Never missed medication” to 6 = “More than 3 months ago”); (2) In the last 30 days, on how many days did you miss at least one dose of any of your HIV medicine? (3) In the last 30 days, how often did you take your HIV medicine in the way you were supposed to? (1 = “Never” to 5 = “Always”); (4) In general, over the past 6 months, how often did you miss taking your medication? (1 = “I have not taken any medication over the past 6 months” to 6 = “I never miss taking any of my medicines/I take my medicines all the time”); 5) How hard is it for you to take your HIV medicine in the way you are supposed to? (1 = “Extremely hard” to 5 = “Not hard at all”); and (6) How often do you argue with the person helping you to take your medicine? (1 = “Never” to 5 = “Always”). A latent construct for adherence was generated using these six questions.

2.2.2 |. Independent variable

Family support was measured by three items assessing instrumental support: (1) Who helps you take your HIV medicines or helps you remember to take them (such as mother, father, friend, nurse/medical professional, etc.)? (2) How often does this person help you when it comes to taking your medication? (5 = always to 1 = never); and (3) How likely is it that this person would know when you miss your medication? (4 = very likely to 1 = very unlikely). Responses from items 2 and 3 were summed and the total score was dichotomized into 0 = total score of 5 or less (less responsible), and 1 = total score of more than 5 (more responsible). A latent construct for family support was generated using items 2 and 3.

2.2.3 |. Mediator variables

Mediator variables included caregiver support, caregiver communication, food security, having personal savings, savings attitudes, and school enrollment. Caregiver support was measured using 18-items related to emotional support (e.g., acceptance and warmth). Responses were rated on a 5-point scale (with 1 = Never and 5 = Always). Summary scores were generated, with higher scores indicating higher levels of perceived caregiver support. Communication with the parent/guardian was assessed by asking participants how often they discuss specific topics (such as HIV/AIDS, puberty, education and future planning, among others) with their caregivers, on a 5-point scale (with 1 = Never and 5 = Always). Food security was assessed by a composite score generated using three items: (1) whether the participant had breakfast on the interview day (1 = yes, 0 = no), (2) how many times the participant ate fish or meat in the week preceding the interview (0 = once or none, 1 = twice or more), and the number of meals that the participant has per day (0 = one or none, 1 = two or more meals). Personal savings were assessed by asking participants whether they have any money saved anywhere (yes/no). Savings attitudes were assessed by asking participants to state the importance of saving for six goals related to family assistance, family business, education, vocational training, buying an animal for income generation or moving into one’s own house. Responses were rated on a 5-point scale (with 1 = Not important at all and 5 = Extremely important). For school enrollment, participants were asked whether they were currently enrolled in school at the time of baseline assessments (yes/no). Participants’ household size, gender and pill burden were included in the model as control variables.

2.3 |. Analysis procedures

Descriptive analyses were conducted on the sample and household characteristics, self-reported adherence and family support factors using STATA 17. Mplus version 8 was used to fit structural equation models that assessed the direct, indirect, and total effects of family support on self-reported ART adherence. The mediators included in the models were saving attitudes, having savings, being enrolled in school, food security, caregiver support, and communication with the caregiver. We controlled for participants’ gender, household size, and pill burden. For all the analyses, cluster adjusted robust standard errors were reported. In the results tables, nonstandardized coefficients, (B), were reported, in addition to standardized coefficients, (β), and the corresponding 95% confidence intervals (CI). p values that were .05 or less were considered significant. A χ² goodness of fit value was calculated; however, it was significant. Therefore, model fitness was estimated by considering a Root Mean Square Error of Approximation (RMSEA) of less than 0.06, Comparative Fit Index (CFI) of more than 0.92, and the Standardized Root Mean Square Residual (SRMR) of less than 0.08.

3 |. RESULTS

3.1 |. Sample characteristics

Baseline sample characteristics are presented in Table 1. Fifty-six percent (56%) of participants were females. The average age of participants was 12.4 years. More than one-third of participants (37.8%) were single orphans—meaning they had one surviving biological parent and 35% had both their biological parents living (non-orphans). Majority of participants (87.3%) were currently enrolled in school (at baseline assessment). In terms of household characteristics, almost half of the participants (47%) reported a surviving biological parent as their primary caregiver and 29.4% reported their grandparent(s). Participants lived in households with an average of 5.7 individuals, with 2.4 children under 18 years. Only 29% of participants reported having personal savings. Most participants (77.2%) were taking more than one type of HIV medication daily. In addition, 91% of adolescent reported never to rarely argue with the person helping them to take their medication.

TABLE 1.

Sample characteristics (N =702).

Variable	n (%)
Gender
Female	396 (56.4)
Male	306 (43.6)
Age (Mean ± SD)	12.42 ± 1.98
Orphanhood status
Double orphan	192 (27.4)
Single orphan	264 (37.6)
Non-orphan	246 (35)
School enrollment
Currently enrolled	613 (87.3)
Not enrolled	89 (12.7)
Household characteristics
Primary caregiver
Biological parent	331 (47.1)
Grandparents	206 (29.4)
Other relative (siblings, aunt, uncle, other)	165 (23.5)
Mean number of people in HH ± SD (min/max: 2–18)	5.7 ± 2.6
Mean number of children in HH ± SD (min/max: 1–14)	2.4 ± 1.9
Mean food security score ± SD (min/max: 0–7)	3.9 ± 1.6
Availability of savings (yes)	205 (29.2)
Mean saving attitudes score ± SD (min/max: 13–30)	26.5 ± 3.2
How many different HIV medicines do you take?
One	160 (22.8)
More than one	542 (77.2)

3.2 |. Family support and self-reported adherence

Adolescents’ self-reported adherence to medication and family support is presented in Table 2. Adolescents reported high levels of adherence to medication, with 70% reporting no missed medicines at all, 72% had not missed any of their medication in the last 6 months, and about 90% reported usually or always taking their medication as prescribed. The majority (85.6%) reported no difficulty (not hard to not hard at all) taking their medication.

TABLE 2.

Self-reported adherence and family-level factors.

Variables	n (%)
Self-reported adherence
When was the last time you missed any of your medication?
Never missed	494 (70.4)
Within the past week to More than 3 months ago	208 (29.6)
How often did you take your medication in a prescribed manner in the last 30 days?
Usually to always	630 (89.7)
Sometimes to Never	72 (10.3)
How often did you miss taking your medication in the past 6 months?
Never missed	506 (72.1)
A little bit to have not taken any medication	196 (27.9)
How hard is it for you to take your HIV medicine as prescribed?
Extremely hard to hard	101 (14.4)
Not hard to not hard at all	601 (85.6)
How often do you argue with the person helping you take medication?
Never to rarely	640 (91.2)
Sometimes to always	62 (8.8)
Family support
Who helps you take or remember taking you medication?
Biological parent	292 (41.6)
Grandparent	181 (25.8)
Other person (extended family member, teachers, nurse)	130 (18.5)
No one	99 (14.1)
How often does this person help you to take your medication?
Never to half the time	140 (19.9)
Most of the time to always	562 (80.1)
How likely would this person know when you miss taking your medication?
Somewhat likely to very likely	585 (83.3)
Somewhat unlikely to very unlikely	117 (16.7)
Caregiver support (Mean, SD)
Frequency of communication with caregiver (min/max: 10–55)	24.8 ± 8.0

Regarding family support, majority of adolescents (86%) reported receiving instrumental support. Specifically, 41.6% of adolescents reported that their biological parent helped them take or remember to take their medication and 14% reported no support at all. For those who received support, 80% reported that support was received always or most of the time, and 83.4% reported that the individual supporting them would likely or very likely know if the adolescent missed taking their medication.

3.3 |. Mediation analysis

Table 3 presents results from the structural equation modeling analysis for the mediators of the effect of family support on self-reported adherence. The SEM results showed that there was a significant indirect effect of family support on ART adherence (B = 0.008, β = .112 [95% CI: 0.052–0.173], p < .001). Particularly, the specific indirect effects of family support through saving attitudes (B = 0.004, β = .058 [95% CI: 0.008–0.108], p = .024), and communication with the guardian (B = 0.004, β = .056 [95% CI: 0.012–0.100], p = .013) were statistically significant. Also, the total effect of family support on ART adherence was statistically significant (B = 0.011, β = .146 [95% CI: 0.032–0.259], p = .012). However, the direct effect was not significant (Figure 1).

TABLE 3.

Mediation analysis.

Variable	B	β (95% CI)	p Value
Family support on mediators
Saving attitudes	0.189	.391 (0.321–0.461)	<.001
Caregiver support	0.035	.154 (0.069–0.239)	<.001
Having savings	−0.046	−.109 (−0.180–−0.039)	.002
Enrolled in school	−0.002	−.012 (−0.087–0.063)	.751
Food security	−0.001	−.016 (−0.077–0.045)	.608
Communication with the guardian	0.197	.351 (0.271–0.431)	<.001
Mediators on ART adherence
Saving attitudes	0.023	.148 (0.022–0.275)	.021
Caregiver support	−0.002	−.006 (−0.140–0.128)	.925
Having savings	−0.002	−.009 (−0.119–0.101)	.872
Enrolled in school	0.022	.053 (−0.010–0.117)	.099
Food security	0.095	.071 (−0.013–0.156)	.098
Communication with the guardian	0.021	.159 (0.047–0.271)	.005
Specific indirect effects
Saving attitudes	0.004	.058 (0.008–0.108)	.024
Caregiver support	<0.001	−.001 (−0.022–0.020)	.925
Having savings	<0.001	.001 (−0.011–0.013)	.873
Enrolled in school	<0.001	−.001 (−0.005–0.003)	.755
Food security	<0.001	−.001 (−0.006–0.004)	.653
Communication with the guardian	0.004	.056 (0.012–0.100)	.013
Effects
Direct effect	0.003	.034 (−0.093–0.160)	.602
Total indirect effect	0.008	.112 (0.052–0.173)	<.001
Total effect	0.011	.146 (0.032–0.259)	.012
Effect mediated		76.7%
Model fitness
R2		β = .072, p = .016
χ2		χ2 (81) = 131.6, p <.001
CFI		0.939
RMSEA		0.032
SRMR		0.037

Note: Bold values represent statistically significant associations.

Abbreviations: CFI, Comparative Fit Index; RMSEA, Root Mean Square Error of Approximation; SRMR, Standardized Root Mean Square Residual.

FIGURE 1.

Structural equation model showing the effect of family support on self-reported ART adherence among ALHIV in Southern Uganda. We reported standardized coefficients. Significant associations are illustrated with bold arrows. ALHIV, adolescents living with HIV; ART, antiretroviral therapy.

In the model, mediation contributed 76.7% of the total effects. Controlling for participants’ characteristics, we were able to explain only 7.2% of the variance (R² = .072, p = .016). The model fitness parameters included CFI = 0.939, RMSEA = 0.032, and SRMR = 0.037.

4 |. DISCUSSION

This paper examined the effect of family support on self-reported adherence to medication among ALHIV in Uganda. First, study findings indicate high levels of self-reported adherence to ART. We find that 70% of the sample reported that they had never missed any medication and about 90% reported taking their medication in a prescribed way. This finding is consistent with previous studies in the region that have documented self-reported adherence among ALHIV (Dachew et al., 2014; Vreeman et al., 2018). It is possible that once adolescents are disclosed to and understand the importance of taking their medication, they are more likely to adhere to medication without conflicting with their caregivers. Indeed, studies have documented improved adherence as one of the benefits associated with HIV disclosure among children and adolescents (Cluver et al., 2015; Krauss et al., 2016; Nichols et al., 2017). Alternatively, it could be that given the young age of adolescents in the sample, it is easier for them to take their medication as prescribed. Young children also tend to report higher levels of adherence compared to older adolescents. However, adherence tend to decline as children grow older potentially due to the transfer of responsibility from caregivers to adolescents themselves (Adejumo et al., 2015; Mellins et al., 2004).

Second, majority of participants (86%) reported receiving family support in the form of instrumental support from family members. Given the young age of study participants, it is not surprising that family members are actively involved to make sure that adolescents adhere to their treatment, specifically, by helping them or reminding them to take their medication (Denison et al., 2015; Gichane et al., 2018). However, while the positive role of instrumental support on ART adherence has been documented (Fredriksen-Goldsen et al., 2011), its direct effect on adherence was not significant in our study—signifying that this form of support alone may not be enough to promote adherence. For adolescents who did not report family support (14%), it could be that they are fully responsible for their own treatment adherence needs.

Third, regarding potential pathways, we find that family support impacted self-reported adherence through both communication with the caregiver and positive savings attitudes. Specifically, family support may improve child-caregiver communication, which in turn, was associated with self-reported adherence. Studies have documented that children’s knowledge of the purpose of taking medication, as well as how to take it correctly depends on their caregivers and require open communication with the child (Bikaako-Kajura et al., 2006; Stein et al., 2019; Taddeo et al., 2008). It is possible that caregivers who are actively involved in their children’s treatment are more likely to engage in open communication, regarding HIV illness in general, as well as the importance of adhering to medication. Moreover, given that communication among caregivers of children living with HIV is still low (Kajubi et al., 2014, 2016), this finding support other studies that have documented the importance of effective parent-child communication in fostering pediatric adherence to treatment (DiMatteo, 2004b; Kajubi et al., 2014, 2016; Mellins et al., 2004; Stein et al., 2019).

For savings attitudes, it is possible that family support improves saving attitudes around the importance of saving to take care of one’s family, including their medical expenses. Specifically, positive savings attitudes may act as a proxy for savings behaviors. For example, those with positive savings attitudes may be more likely to save—which may help to meet their medical needs, including affording transportation to get to their clinic appointments and also purchase food to facilitate medication intake. While self-reported savings was not a significant pathway in our analyses, in a separate analysis that utilized actual saving records from financial institution, participants with savings reported better ART adherence (Brathwaite et al., 2022). This finding aligns with previous studies related to financial savings, possession of household economic assets, and their association with greater odds of self-reported adherence (Bermudez et al., 2016) and viral suppression among ALHIV (Bermudez et al., 2018).

A few limitations are worth noting. We report self-reported adherence which may be influenced by social desirability. Also, we analyzed cross-sectional data which does not allow us to examine the impact of family support over time.

Despite these limitations, study findings contribute to the limited literature that has investigated the pathways through which family support affects ART treatment adherence among ALHIV. Our study findings indicate that family support (i.e., provision of instrumental support), specifically through child-caregiver communication and positive attitudes towards financial savings is crucial in facilitating and promoting ART adherence among ALHIV. Programs that work to promote treatment outcomes, especially for children and ALHIV, should incorporate strategies that help improve overall family support and open communication with children on issues regarding HIV, as well as medication and adherence, as they try to navigate the transition through adolescence and into young adulthood.

DATA AVAILABILITY STATEMENT

The datasets analyzed during the current study are available from the corresponding author on reasonable request.