Preliminary Impact of Group-Based Interventions on Stigma, Mental Health and Treatment Adherence Among Adolescents Living with HIV in Uganda

Abstract

Objective

To examine the preliminary impact of group cognitive behavioral therapy (G-CBT) and multiple family group-based family strengthening (MFG-FS) to address HIV stigma and improve the mental health functioning of adolescents living with HIV in Uganda.

Study design

We analyzed data from the Suubi4Stigma study, a two-year pilot randomized clinical trial that recruited from 2020 through 2021 adolescents living with HIV (10–14 years) and their caregivers (N=89 dyads), from 9 health clinics. We fitted separate three-level mixed-effects linear regression models to test the effect of the interventions on adolescent outcomes at 3 and 6 months post-intervention initiation.

Results

The average age was 12.2 years, and 56% of participants were female. Participants in the MFG-FS intervention reported lower levels of internalized stigma (mean difference = −0.008, 95% CI= −0.015, −0.001, p=0.025) and depressive symptoms at 3 months (mean difference = − 0.34, 95% CI= −0.53, −0.14, p<0.001), compared with usual care. On the other hand, participants in the G-CBT intervention reported lower levels of anticipated stigma at 3 months (mean difference = −0.039, 95% CI= −0.072, −0.006), p=0.013) and improved self-concept at 6 months follow-up (mean difference = 0.04, 95% CI= 0.01, 0.01, p=0.025).

Conclusion

Outcome trends from this pilot study provide compelling evidence to support testing the efficacy of these group-based interventions on a larger scale.

Globally, an estimated 1.7 million children <15 years are living with HIV, with the majority living in Sub-Saharan Africa (SSA) ¹. Over the last decade, improved approaches optimizing HIV treatment outcomes— including early diagnosis and antiretroviral therapy (ART) initiation, treatment and prevention of infections, monitoring and optimal adherence to ART—have improved health outcomes and reduced child mortality ^{2, 3}. As a result, the number of children growing up and transitioning through adolescence with a highly stigmatized disease has increased.

HIV stigma, characterized by public blame and moral condemnation of contracting the infection ^{4, 5}, is associated with morbidity, mortality, and health disparities among people living with HIV (PLHIV) ^{6, 7}. Stigma is a major barrier to all aspects of the HIV care continuum ^{5, 8, 9, 10} . Similarly, studies have documented the adverse impact of HIV stigma on a range of cognitive, affective, and mental health outcomes among PLHIV, including post-traumatic stress disorder (PTSD), depression, anxiety, and suicidal ideation ¹¹, feelings of rejection, loneliness, social isolation ¹², and low quality of life ¹³. Stigma, once internalized is also associated with negative emotional reactions, including diminished self-efficacy ¹⁴, low self-esteem ¹⁵, and hopelessness ^16,17. In Uganda, the prevalence of depression among adolescents living with HIV (ALHIV) is estimated at more than 40% and 45.6% for anxiety ¹⁸. The incidence and persistence of major depressive disorders have been estimated at 7.6% and 11% respectively ¹⁹ –pointing to the need to identify evidence-based approaches and integrate mental health treatment into health care settings.

While significant efforts have been made in developing and implementing HIV stigma reduction interventions in different settings, including psycho-educational, supportive interventions for treatment adherence, individual empowerment, psychotherapy, narrative, social contact and community-based interventions ^{20, 21}, ALHIV in SSA are significantly underrepresented ²¹. A recent systematic review of health-related needs reported by ALHIV receiving ART in SSA identified stigma reduction as their primary health-related need ²² – pointing to the need for data-driven research to address stigma among ALHIV as they transition into young adulthood.

Group-based treatments have several therapeutic benefits for participating group members, such as fostering cohesiveness, universality, interpersonal learning, and self-understanding ²³. Cognitive behavioral therapy (CBT) – whether individual or group based, has effectively treated various disorders among adults and children ^{24, 25}. The core components of CBT, such as psychoeducation, cognitive restructuring, and skill-building to increase adaptive coping mechanisms ²⁶, may be critical in addressing HIV stigma. Indeed, CBT has been associated with reducing depression, improving medication adherence and viral suppression among individuals with comorbid HIV and depression ²⁷. On the other hand, group-based CBT (G-CBT) has been widely used and found to be more effective than individual CBT on a range of mental disorders among children ^{28, 29}. Moreover, G-CBT is likely to offer more opportunities for normalization, positive peer modeling, reinforcements, social support, and exposure to social situations and feedback sources, given the context of shared experiences ³⁰.

Similarly, implementing multiple family group-based family strengthening (MFG-FS), which encourages shared experiences, foster social support, and enable social networking, is an integral component to families experiencing difficulties ³¹, such as those caring for PLHIV. MFG-FS consist of a combination of family and group work in which a collection of families (multiple caregivers and their children) meet simultaneously, at agreed-upon intervals, with a facilitator present ³². MFGs may include psychoeducation, emotional support, communication, parenting skills, problem-solving strategies, and other therapeutic components ³³. By bringing multiple families together, MFGs are associated with unique benefits beyond usual group work, such as enabling participants to feel comfortable sharing with and supporting one another, helping to foster peer support, decreasing social isolation, facilitating optimism and morale, and enhancing interpersonal and coping skills ³³.

The objective of this study was to test the feasibility, acceptability, and preliminary impact of two evidence and theory-informed interventions: a group-based cognitive behavioral therapy (G-CBT) and a family strengthening intervention delivered via multiple family group (MFG-FS) to address HIV stigma and improve the mental health functioning of ALHIV (10–14 years) in Uganda.

METHODS

Participants

We analyzed data from the Suubi (hope)4Stigma study. A total of 89 adolescent-caregiver dyads (N=178) were recruited to participate in the study. Inclusion criteria for adolescents included: 1) living with HIV and aware of their status; 2) between 10–14 years; 3) enrolled on ART in participating clinics; and 4) living within a family, including the extended family. Adults aged 18+ who identified as the primary caregiver of the child were recruited.

Recruitment

Details on participant recruitment are provided in the study protocol [Author Citation]. In brief, between November 2020 and May 2021, participants were recruited from 9 health care clinics providing HIV-related services across four political districts of Masaka, Kyotera, Kalungu, and Lwengo – a region with the highest HIV prevalence in Uganda (11.7%) compared with the national average of 5.4% ³⁵. Clinics were comparable in terms of the number of adolescents served, facility level and having adolescent clinic days. A clinic staff presented the project idea to adult caregivers of eligible adolescents during appointment days. Verbal consent to be contacted by a research staff member was obtained from interested caregivers. During a one-on-one meeting with the research staff member, caregivers were taken through informed consent process, after which they provided written consent for themselves and their children to participate. Adolescents provided assent separately from their caregivers to avoid coercion. A total of 147 adolescents together with their caregivers from 9 health care clinics were screened; 89 dyads met the inclusion criteria and were recruited into the study (see detailed consort flow diagram below).

Randomization

The study utilized a three-arm cluster randomized design. Prior to randomization, stratified random sampling was used to divide the clinics into two, based on the clinic level of care services provided and number of ALHIV served at the clinic. Participants (children and caregivers) were randomly assigned (at the clinic level) to one of three study conditions resulting in three clinics per condition: 1) Usual care (n=29); 2) G-CBT (n=26), or 3) MFG-FS (n=34). Participants in each clinic received the same intervention corresponding to the group to which the clinic was assigned to minimize contamination. The randomization was done using STATA software and was conducted by an independent research associate based at Washington University in St. Louis.

Sample Size and Power Calculation

The study was planned to recruit 90 adolescent-caregiver dyads. Assuming α=.05 and power=.80, 81 participants would be retained at the final time point following 10% estimated attrition, and a clinic-based conservative unconditional ICC of 9.3%. The sample size of 90 dyads provided adequate statistical power to test feasibility. Additionally, for the target enrollment proportion of 70% to assess feasibility, the width to the limit of the confidence interval was 27.9% (standardized distance: .32). For continuous standard normal variables to assess acceptability, the distance from the mean to the confidence limit was .30. These distances to confidence limits were between small (.20) and medium (.50) effect sizes. For preliminary efficacy exploratory analyses with two time points and paired comparisons of two out of the three groups at 81/3=27 participants per group (N=54 per comparison), minimum detectable standardized mean differences for continuous outcomes ranged from .79 to .97 for within-subject correlations (r) ranging from .20 to .80. The study was powered to detect small-medium distances to confidence limits for descriptive statistics and large longitudinal analysis effects.

Description of the Suubi4Stigma Interventions

Usual care condition.

All study participants received the traditional clinic-based interventions focused on testing, medical and treatment services, and psychosocial support for ALHIV. All HIV patients in Uganda receive testing, ART treatment, and information about disease management. We supplemented usual care with literature focused on living positively with HIV, based on the stories of the lives of ALHIV (11–13 years) in Uganda ³⁶.

G-CBT intervention condition.

Participants randomized to this condition received 10 sessions of G-CBT delivered by two trained health para-counselors (5 female and 1 male, aged 24+ years) , with experience in mental health support and experience working with ALHIV in the study region. All paracounselors had been practicing for 4 or more years. Para-counselors received additional training to deliver the Suubi4Stigma G-CBT intervention. The number of adolescents in each group ranged between 8–9 based on the number of adolescents recruited at the clinic. Caregivers did not participate in these sessions. Sessions utilized core components of CBT, such as psychoeducation, cognitive restructuring, and skill-building, to increase adaptive coping ²⁵. Sessions topics included: 1) Introduction to Suubi4Stigma G-CBT Program, 2) HIV/AIDS-associated stigma and depression, 3) Relationship between thoughts and emotions, 4) Identifying thought patterns, 5) Challenging negative thoughts, 6) Identifying and increasing helpful thoughts, 7) Setting goals and shaping your reality, 8) Visualization and guided imagery techniques for mood management, 9) Change talk to improve mood and reduce depressive symptoms, and 10) Group review and ending celebrations. G-CBT content was adapted from existing CBT manuals ^37,38, in consultation with mental health experts in the study region. Sessions were delivered biweekly, and each session lasted approximately one hour. The rate of attendance for G-CBT sessions was between 85–92%. Prior to each session, adolescents who had missed the previous session were taken through what was covered before moving on to the next session.

MFG-FS intervention condition.

Participants randomized to this condition received ten sessions of MFG (also known as “Amaka Amasanyufu” in the local Luganda language), delivered by two trained parents peers – trained in the delivery of MFG in our other studies in the region ³⁹. The number of families (children and caregivers) per group ranged between 5 to 10. Clinics with more than 10 families were split into two groups. The grouping size was intended to promote communication and support within and among families. Sessions focused on the core components of MFG, also known as the 4Rs and 2S (rules, responsibility, relationships, respectful communication, stress, and social support)³⁹. The original sessions 2 and 3 were adapted to focus on HIV, stigma and discrimination, and associated risks. Session topics included: 1) Introduction to “Amaka Amasanyufu,” 2) HIV/AIDS Knowledge and Adherence to Medication, 3) Stigma, Discrimination and Associated Risks, 4) Building on Family Supports, 5) Rules for Home and Problem Solving for Broken Rules, 6) Respectful Communication, 7) Responsibility at Home, 8) Dealing with Stress at Home, 9) Family Relationships and Building Families Up, and 10) Group Review and Ending Celebrations. These biweekly sessions lasted approximately 1 hour. The rate of attendance for MFG-FS sessions was between 88–91%.

Ethical Considerations

All study procedures were approved by Washington University in St. Louis Institutional Review Board (IRB # 202009185), the Uganda Virus Research Institute (GC/127/20/10/792), and the Uganda National Council for Science and Technology (SS632ES). Prior to study participation, voluntary written consent and assent were obtained from caregivers and adolescents, respectively.

Data Collection

Data were collected at baseline, 3- and 6-months post-intervention initiation using a 90-minute interviewer-administered questionnaire. All interviewers received training in human subjects protection and completed Good Clinical Practice (GCP) training before engaging with study participants. All study-related materials, including recruitment materials, consent/assent forms, intervention manuals and data collection tools were translated into Luganda – the widely spoken language in the study region and back translated into English to ensure consistency. A certificate of translation was obtained from Makerere University. Recruitment, intervention delivery and data collection processes were conducted in Luganda.

Measures

Primary outcome

HIV stigma was measured using 9-items from the Berger Stigma Scale ⁴⁰ measuring both internalized and anticipated stigma (Cronbach’s alpha = 0.74), with high scores indicating high levels of stigma.

Secondary outcomes

Depressive symptoms were assessed using the 14-item Child Depression Inventory (CDI), which measures both emotional and functional problems (Cronbach’s alpha 0.61). Hopelessness was measured using the Beck Hopelessness Scale (BHS) ⁴², a 20-item scale that measures children’s hopelessness and pessimistic attitudes toward the future (Cronbach’s alpha 0.68). PTSD symptoms were measured using 31 items from the abbreviated Childhood Post-Traumatic Stress Reaction Index (CPTS-RI) ⁴³, used to screen for exposure to traumatic events and assess PTSD symptoms in school-age children and adolescents (Cronbach’s alpha = 0.92). Self-concept was measured using the Tennessee Self-Concept Scale (TSCS) ⁴⁴, a 20-item scale that measures children’s perception of identity, self-satisfaction, and other behaviors (Cronbach’s alpha = 0.81). Higher scores on the CDI, BHS, CPTS-RI and the TSCS indicated higher levels of depressive symptoms, hopelessness, PTSD, and self-concept, respectively.

To assess self-reported adherence to HIV treatment, participants were asked three questions from Wilson's three-item self-reported adherence measure ⁴⁵: 1) “In the last 30 days, on how many days did you miss at least one dose of your HIV medications (range: 0–30 days)?” 2) “In the last 30 days, how often did you take your HIV medicine in the way you were supposed to (never to always)?” and 3) “In the last 30 days, how good a job did you do at taking your HIV medicine in the way you were supposed to (very poor - excellent)”? Responses from each item were linearized into a continuous scale ranging from 0–100, with low scores representing poor adherence and 100 representing perfect adherence ⁴⁶. Adherence was calculated as the mean of the three individual items and was treated as a continuous variable.

Data Analysis Procedures

Analyses were conducted using Stata version 17.0 using an intention-to-treat approach. We summarized the continuous variables using means and standard deviations and counts with percentages for the categorical variables. For this analysis, we had seven continuous outcomes: internalized stigma and anticipated stigma (primary outcomes), as well as depressive symptoms, self-concept, hopelessness, PTSD and ART adherence (secondary outcomes). We reported summary statistics for all the outcomes, stratified by study groups, for each time point. For each outcome, we checked whether the assumptions of normality and constant variance of residuals across levels of the predicted values were not violated, by plotting histograms of standardized residuals and scatter plots of standard residuals against predicted values. All models met the assumption of normality of residuals. We then fitted separate three-level mixed-effects linear regression models to test the effect of the interventions on the outcomes at 3 and 6 months. In each model, we included fixed categorical effects for the study group (usual care, MFG-FS, and G-CBT), time (baseline, 3 and 6 months), the group-by-time interaction term, and the random participant and clinic intercepts. We assessed for the main effects of the interventions, time, and their interaction. Regardless of significance of group-by-time interaction term, we also examined the pairwise comparisons for the group means at post-baseline time points, using Sidak’s method to adjust the p-values (see supplementary materials). We reported robust Huber-White cluster-adjusted confidence intervals. Significance levels were set at p = 0.05/2 for the primary outcomes and 0.05/5 for the secondary outcomes, after applying Bonferroni’s adjustment for repeated measures on the primary outcomes.

RESULTS

Sample Characteristics

At baseline, majority of participants were female (62.9%), the mean age was 12.2 years, and 44.9% identified as orphans i.e., had lost a biological father or mother (Table I). The mean household size was 6 people. The mean baseline self-reported adherence across all study groups was 91.77%. The distribution of HIV stigma, and all mental health measures by study group and time points are presented in Table II.

Table I.

Baseline sample characteristics

Variables	Total sample	Usual care	MFG-FS	G-CBT
	n = 89 (%)	n = 29 (%)	n = 34 (%)	n = 26 (%)
	n (%)/mean (SD)	n (%)/mean (SD)	n (%)/mean (SD)	n (%)/mean (SD)
Gender
Male	33 (37.08)	12 (41.38)	12 (35.29)	9 (34.62)
Female	56 (62.92)	17 (58.62)	22 (64.71)	17 (65.38)
Age in years (min/max: 10–14)	12.21 (1.41)	12.45 (1.35)	11.65 (1.52)	12.69(1.09)
Orphanhood status
Nonorphan	49 (55.06)	14 (48.28)	20 (58.82)	15 (57.69)
Single orphan	33 (37.08)	13 (44.83)	11 (32.35)	9 (34.62)
Double orphan	7 (7.87)	2 (6.90)	3 (8.82)	2 (7.69)
Household size (min/max: 2–14)	6.42 (2.66)	6.14 (2.60)	6.79 (3.00)	6.23 (2.25)

Table II.

Summary of the outcomes by study arm and time points

Outcomes	Baseline		3 months		6 months
	n = 29		n = 34		n = 26
		Mean (SD)	n	Mean (SD)	n	Mean (SD)
Internalized stigma (min/max: 0–15)
Usual care	29	7.00 (2.60)	29	5.76 (1.68)	29	5.59 (2.15)
MFG-FS	34	6.53 (3.77)	34	6.59 (3.01)	34	6.21 (2.28)
G-CBT	26	6.19 (2.77)	23	6.43 (2.11)	21	5.38 (1.75)
All	89	6.58 (3.13)	86	6.27 (2.39)	84	5.79 (2.12)
Anticipated stigma (min/max 0–9)
Usual care	29	2.83 (2.75)	29	2.79 (2.13)	29	1.72(1.81)
MFG-FS	34	2.44 (2.56)	34	3.53 (2.78)	34	2.41 (2.30)
G-CBT	26	2.46 (2.37)	23	3.17 (2.31)	21	3.00 (2.88)
All	89	2.57 (2.55)	86	3.19 (2.44)	84	2.32 (2.33)
Depressive Symptoms (min/max: 0–16)
Usual care	29	4.90 (3.48)	28	3.43 (3.02)	29	3.28 (2.07)
MFG-FS	34	4.85 (3.82)	33	3.79 (3.52)	34	4.18 (3.33)
G-CBT	26	5.38 (3.49)	25	3.16 (2.58)	21	3.10 (2.57)
All	89	5.02 (3.58)	86	3.49 (3.08)	84	3.60 (2.77)
Self-concept (min/max: 0–60)
Usual care	29	38.03 (5.80)	29	38.83 (7.54)	29	38.28 (4.96)
MFG-FS	31	36.39 (7.41)	32	36.00 (7.81)	31	37.39 (6.18)
G-CBT	26	39.92 (7.53)	23	35.35 (8.13)	18	34.61 (7.04)
All	89	38.01 (7.01)	86	36.80 (7.86)	84	37.08 (6.06)
Hopelessness (min/max: 0–16)
Usual care	29	6.00 (3.22)	28	4.57 (3.27)	29	4.14 (2.90)
MFG-FS	34	5.50 (3.03)	33	4.70 (3.38)	34	5.00 (3.87)
G-CBT	26	6.27 (3.42)	25	4.80 (2.99)	21	4.55 (3.30)
All	89	5.89 (3.19)	86	4.69 (3.20)	84	4.55 (3.30)
Post-Traumatic Stress Disorder (min/max: 0–73)
Usual care	29	19.76 (3.25)	28	14.71 (12.55)	29	12.21 (12.47)
MFG-FS	34	20.56 (4.94)	33	13.67 (12.52)	34	13.06 (12.84)
G-CBT	26	22.88 (15.69)	25	10.08 (15.49)	21	9.24 (10.91)
All	89	20.98 (14.53)	86	12.97 (13.43)	84	11.81 (12.21)
ART adherence (min/max: 56.5% – 100%)
Usual care	29	90.15 (12.17)	28	93.97 (9.63)	29	93.14 (9.24)
MFG-FS	34	92.55 (9.31)	33	91.99 (9.90)	34	91.37 (9.59)
G-CBT	26	91.77 (9.82)	25	91.67 (10.68)	21	88.41 (15.30)
All	89	91.77 (9.82)	86	92.54 (9.98)	84	91.24 (11.18)

Mixed-effects Model Results for Stigma and Adherence

There were significant intervention main effects for internalized stigma (chi square = 14.85, p<0.01), and anticipated stigma (chi square = 8.80, p=0.012) (Table III). These results show that the interventions had a significant impact on both forms of stigma when compared with usual care. Further examination of the intervention effects involved performing pairwise comparisons across groups at different time points to determine which intervention or time point was responsible for the significant observed group main differences (Table IV). Adolescents who participated in the MFG-FS intervention reported significantly lower levels of internalized stigma at 3-months (mean difference = −0.008, 95% CI= −0.015, −0.001, p=0.025) compared with usual care. On the other hand, participants in the G-CBT intervention reported lower less anticipated stigma (mean difference = −0.039, 95% CI= −0.072, −0.006), p=0.013) compared with usual care at 6-moths follow-up.

Table III.

Mixed-effects model showing effects of the interventions on stigma and adherence

Outcomes	Internalized stigma*	Anticipated stigma	Adherence
	β (95% CI)	β (95% CI)	β (95% CI)
Group: χ2(df), P-value	14.85 (2), P < .001	8.80 (2), P = .012	3.29 (2), P = .193
Usual care	Ref	Ref	Ref
MFG-FS	−0.01 (−0.02 – −0.001)	0.01 (−0.02 – 0.05)	2.40 (−1.99 – 6.78)
G-CBT	0.01 (−0.01 – 0.01)	0.01 (−0.01, 0.03)	2.41 (−1.95 – 6.77)
Time: χ2(df), P value	4.05 (2), P = .132	7.05 (2), P = .029	3.03 (2), P = .219
Baseline	Ref	Ref	Ref
3 months	0.01 (−0.01 – 0.01)*	−0.01 (−0.04 – 0.03)	3.88 (−2.51 – 10.28)
6 months	0.08 (0.02 – 0.13) *	0.03 (−0.01 – 0.06)	2.99 (−4.26 – 10.24)
Group-by-time interaction: χ2(df), P value	4.38 (4), P = .357	8.77 (4), P = .067	19.82 (4), P < .001
Usual care x baseline	Ref	Ref	Ref
MFG-FS x 3 months	0.05 (−0.11 – 0.20)*	−0.03 (−0.09 – 0.03)	−4.51 (−11.21 –2.19)
MFG-FS x 6 months	0.01 (−0.17 – 0.18)*	−0.04 (−0.08 – 0.01)	−4.16 (−11.58 – 3.25)
G-CBT x 3 months	−0.06 (−0.20 – 0.07)*	−0.02 (−0.06 – 0.02)	−4.70 (−12.06 – 2.66)
G-CBT x 6 months	−0.07 (−0.18 – 0.04)*	−0.05 (−0.08 – −0.01)	−7.30 (−14.98 – 0.38)
Constant	0.09 (0.07 – 0.11)	0.21 (0.19 – 0.23)	90.15(85.81 – 94.49)
Random effects
Clinic level variance	<0.001 (<0.001, 0.001)	<0.001 (<0.001, 0.001)	<0.001 (<0.001, 0.001)
Child-level variance	0.006 (0.002 – 0.023)	0.023 (0.008 – 0.065)	4.54 (3.08 – 6.68)
Variance of residuals	0.026 (0.022 – 0.030)	0.084 (0.079 – 0.090)	9.11 (8.49 – 9.78)
Number of participants	89	89	89
Number of observations	259	259	259

The model for internalized stigma controlled for depressive symptoms and self-concept.

^*Coefficients have been multiplied by a factor of 10.

Table IV.

Mixed-effects model showing effects of the interventions on adolescents’ mental health

Variables	Depressive symptoms	Self-concept	Hopelessness	PTSD
	β (95% CI)	β (95% CI)	β (95% CI)	β (95% CI)
Group: χ2(df), P-value	49.88 (2), P < .001	15.40 (2), P < .001	0.84 (2), P = .658	0.65(2), P = .724
Usual care	Ref	Ref	Ref	Ref
MFG-FS	−0.01 (−0.41 – 0.40)	0.01 (−0.01 – 0.01)*	−0.50 (−1.21 – 0.21)	0.80 (−3.36 – 4.96)
G-CBT	0.03 (−0.46 – 0.51)	−0.01 (−0.01 – 0.01)*	0.27 (−0.45 – 0.99)	3.13 (−2.54 – 8.80)
Time: χ2(df), P-value	13.43 (2), P = .001	9.56 (2), P = .008	19.05 (2), P < .001	19.26 (2), P < .001
Baseline	Ref	Ref	Ref	Ref
3 months	−0.29 (−0.58 – −0.01)	0.01 (−0.04 – 0.05)*	−1.42 (−3.82 – 1.00)	−4.91 (−12.04 – 2.23)
6 months	−0.34 (−0.73 – 0.05)	0.03 (−0.01 – 0.13)*	−1.86 (−3.09 – −0.63)	−7.55 (−15.82 – 0.71)
Group-by-time interaction: χ2(df), P-value	10.52 (4), P = .033	23.17 (4), P < .001	16.13 (4), P = .003	1.76 (4), P = .780
Usual care x baseline	Ref	Ref	Ref	Ref
MFG-FS x 3 months	0.19 (−0.10, 0.48)	0.01 (−0.04 – 0.05)*	0.63 (−2.02 – 3.29)	2.05 (−9.62 – 5.51)
MFG-FS x 6 months	0.11 (−0.30 – 0.52)	−0.17 (−0.38 – 0.03)*	1.36 (−0.49 – 3.22)	0.05 (−8.32 – 8.43)
G-CBT x 3 months	−0.18 (−0.66 – 0.29)	0.04 (−0.04 – 0.09)*	−0.04 (−2.51 – 2.44)	−7.79 (−20.64 – 5.06)
G-CBT x 6 months	−0.09 (−0.76 – 0.59)	0.04 (0.02 – 0.07) *	0.01 (−1.30 – 1.32)	−5.75 (−18.34 – 6.84)
Constant	1.47 (1.07 – 1.86)	0.02 (0.01 – 0.03)	6.00 (5.44 – 6.56)	19.76 (16.60 – 22.92)
Random effects
Clinic level variance	<0.001 (<0.001, <0.001)	<0.001 (<0.001, <0.001)	<0.001 (<0.001, <0.001)	<0.001 (<0.001, <0.001)
Child-level variance	0.291 (0.163 – 0.521)	0.010(0.001 – 0.068)	1.61 (1.27 – 2.04)	6.00 (3.99 – 9.03)
Variance of residuals	0.608 (−.552 – 0.669)	0.006 (0.005 – 0.008)	2.76 (2.50 – 3.05)	11.88 (11.15 – 12.65)
Number of participants	89	89	89	89
Number of observations	259	259	259	259

The model for self-concept was adjusted for depressive symptoms and anticipated stigma.

^*Coefficients have been multiplied by a factor of 10.

Regarding ART adherence (Table III), we observed significant group-by-time interaction effects for ART adherence (χ²(4) = 19.82, p<0.001), suggesting the presence of a significant difference between the groups in at least one of the endpoints. Pairwise comparisons (Table IV) revealed a statistically significant mean difference between participants in the MFG-FS and G-CBT groups at 6 months (mean difference = −3.12, 95% CI= −5.97, −0.27, p=0.027), suggesting that participants in the G-CBT intervention had better ART adherence compared with those who received the MFG intervention.

Mixed-effects Model Results for Mental Health Functioning

We found significant group main effects for depressive symptoms (chi square = 49.88, p<0.001) and self-concept (chi square = 15.40, p<0.001) (Table III), indicating that the interventions had a significant impact on depressive symptoms and self-concept. Pairwise comparisons revealed that participating in MFG-FS intervention was associated with a significant impact on reducing depressive symptoms at 3 months compared with G-CBT (mean difference = −0.34, 95% CI= −0.53, −0.14, p<0.001) (Table IV. In addition, adolescents who participated in the G-CBT intervention reported higher levels of self-concept compared with those in usual care at 6momths follow-up (mean difference = 0.04, 95% CI= 0.01, 0.01, p=0.025).

DISCUSSION

This study examined the preliminary impact of G-CBT and MFG-FS interventions to address HIV stigma and improve the mental health functioning and adherence outcomes of ALHIV. Given the benefits of group-based treatments ²³, we hypothesized that relative to the usual care condition, participants in both treatment conditions will have lower mean levels of HIV stigma, improved mental health functioning and treatment adherence.

We observed statistically significant differences between groups on HIV internalized and anticipated stigma at 3-months follow-up. Specifically, participating in MFG-FS was associated with lower levels of internalized stigma, and participating in G-CBT was associated with lower levels of anticipated stigma. These findings suggest the unique contribution of each intervention on stigma. Consistent with our hypotheses, it is possible that by bringing adolescents and families together in groups, they were able to reap the benefits associated with group therapies, by enhancing peer support and reducing social isolation ³³ –ultimately resulting in reduced internalized and anticipated stigma. Indeed, a similar group-based pilot intervention in Tanzania among young people documented reduced internalized stigma at 6-months ⁴⁷. Taken together, these outcome trends support the development of a larger trial to test the efficacy of these interventions.

In addition, participating in the interventions was associated with reduced depressive symptoms and improvement in self-concept. Specifically, the MFG-FS intervention was associated with reduced depressive symptoms at 3-months, and the G-CBT intervention was associated with improved self-efficacy at 6-months. This finding is consistent with previous studies that documented the positive impact of group and family strengthening interventions in reducing depressive symptoms among people living with HIV ^{26, 27, 48}, as well as improved self-esteem among depressed youth ^{11, 50}. Similarly, this finding underscores the importance of bringing children and families together in a group setting to help facilitate peer support, shared experiences and enhance interpersonal and coping skills ³³. While we did not find significant group effects for hopelessness and PTSD, these outcomes significantly improved over time across groups.

Finally, for treatment adherence, we did not observe statistically significant differences across groups. This could be attributed to the higher levels of self-reported adherence at baseline. Children within this age group heavily depend on their caregivers for their treatment needs ⁵¹, and they tend to report higher levels of adherence. Thus, it is possible that the interventions did not do much to improve the already high levels of treatment adherence. However, given that self-reports tend to suffer from social desirability, more objective measures should be considered. Despite not being significant, adherence among ALHIV in the G-CBT group was higher compared with the MFG arm. G-CBT may have led to improved ART adherence due to its structured approach focused on skill-building, such as problem-solving and coping strategies that directly influence self-management behaviors. Additionally, the cognitive aspect of G-CBT could help in shaping beliefs about medication adherence. In contrast, while MFG facilitates support and shared experiences, its less direct focus on individual behavioral change techniques may account for the observed differences in adherence outcomes.

We acknowledge the following limitations. First, we report results from a small pilot study not sufficiently powered to conduct formal hypotheses testing; and delivered over a very short period of time (3 months). The potential for evaluating hypotheses is limited by the small sample size. As a result, certain potentially important associations could have been missed. Likewise, as Freedland et al. pointed out, small sample sizes in pilot studies might result in spurious positive results ⁵². Second, all measures were self-reported, and findings may be affected by social desirability bias. Third, some measures of mental health functioning (such as the CDI and CPTS-RI), are only meant to screen for symptoms and not to establish clinical diagnosis of mental health disorders. Finally, given that data was collected during the COVID-19 pandemic, the associated challenges may have limited the impact of the interventions on adolescents’ outcomes.

In conclusion, our study contributes to the limited literature focused on HIV stigma reduction interventions among ALHIV in SSA. These preliminary findings demonstrate the potential of participating in group-based interventions in reducing internalized and anticipated stigma, depressive symptoms and improving self-concept in the short term. These outcome trends provide compelling evidence to support testing the efficacy of these interventions in a larger trial.

Table V.

Pairwise comparisons by group across time points for stigma and antiretroviral therapy adherence

Pairwise comparisons	Internalized stigma		Anticipated stigma		Adherence
	MD (95% CI)	P value	MD (95% CI)	P value	MD (95% CI)	P value
MFG-FS x usual care at baseline	−0.01 (−0.03, 0.01)	.087	0.01 (−0.02, 0.05)	.766	2.40 (−2.94, 7.33)	.633
MFG-FS x usual care at 3 months	0.01 (−0.01, −0.001)	.025	−0.02 (−0.06, 0.03)	.812	−2.12 (−6.78, 2.54)	.624
MFG-FS x usual care at 6 months	−0.01 (−0.03, 0.01)	.355	−0.02 (−0.05, 0.01)	.182	−1.77 (−6.99, 3.45)	.803
G-CBT x usual care at baseline	0.02 (−0.11, 0.16)*	.969	0.01 (−0.02, 0.03)	.903	2.41 (−2.90, 7.72)	.624
G-CBT x usual care at 3 months	−0.01 (−0.02, 0.01)	.899	−0.01 (−0.06, 0.03)	.842	−2.29 (−8.69, 4.10)	.776
G-CBT x usual care at 6 months	−0.01 (−0.02, 0.01)	.809	−0.04 (−0.07, −0.01)	.013	−4.89 (10.6, 0.81)	.117
MFG-FS x G-CBT at baseline	0.01 (−0.01,0.03)	.051	−0.01 (−0.04, 0.02)	.902	0.02 (−0.88, 0.91)	1.000
MFG-FS x G-CBT at 3 months	0.01 (−0.01,0.02)	.918	0.01 (−0.03, 0.04)	.996	−0.17 (−5.27, 4.92)	1.000
MFG-FS x G-CBT at 6 months	0.01 (−0.01,0.02)	.508	−0.02 (−0.04, 0.01)	.204	−3.12 (−5.97, −0.27)	.027
Number of participants	89		89		89
Number of observations	259		259		259

^*Coefficients have been multiplied by a factor of 10.

Table VI.

Pairwise comparisons by group across time points for depressive symptoms, self-concept, hopelessness, and post-traumatic stress disorder

Pairwise comparisons	Depressive symptoms		Self-concept		Hopelessness		PTSD
	MD (95% CI)	P value	MD (95% CI)	P value	MD (95% CI)	P value	MD (95% CI)	P value
MFG-FS x usual care at baseline	−0.01 (−0.50, 0.48)	1.000	0.02 (−0.06, 0.05)*	.178	−0.500 (−1.37, 0.37)	.429	0.80 (−4.27, 5.87)	.975
MFG-FS x usual care at 3 months	0.18 (−0.12, 0.48)	.389	0.02 (−0.02, 0.05)*	.147	0.13 (−2.46, 2.73)	.999	−1.25 (−9.35, 6.84)	.976
MFG-FS x usual care at 6 months	0.10 (−0.04, 0.24)	.234	0.01 (−0.03, 0.04)*	.968	0.86 (−0.73, 2.45)	.478	0.85 (−9.27, 11.0)	.996
G-CBT x usual care at baseline	0.03 (−0.56, 0.62)	.999	−0.09 (−0.02, 0.03)*	.211	0.27 (−0.61, 1.15)	.846	3.13 (−3.78, 10.0)	.627
G-CBT x usual care at 3 months	−0.16 (−0.51,0.20)	.655	0.03 (−0.02, 0.08)*	.375	0.23 (−2.16, 2.62)	.994	−4.67 (−14.2, 4.90)	.568
G-CBT x usual care at 6 months	−0.06 (−0.42, 0.31)	.974	0.04 (0.01, 0.07) *	.025	0.28 (−0.61, 1.16)	.838	−2.63 (−14.6, 9.31)	.936
MFG-FS x G-CBT at baseline	0.04 (−0.31,0.38)	.993	−0.03 (−0.06, −.001)	.017	0.77 (0.01, 1.53)	.048	2.33 (−4.29, 8.94)	.785
MFG-FS x G-CBT at 3 months	−0.34 (−0.53, −0.14)	<.001	0.01 (0.05, 0.06)*	.983	0.10 (−0.97, 1.17)	.995	−3.42 (−13.0, 6.17)	.778
MFG-FS x G-CBT at 6 months	−0.16 (−0.52, 0.20)	.645	0.03 (−0.02, 0.08)*	.312	−0.58 (−1.61, 0.74)	.644	−3.48 (−10.5, 3.54)	.555
Number of participants	89		89		89		89
Number of observations	259		259		259		259

^*Coefficients have been multiplied by a factor of 10.

List of Abbreviations

ALHIV

Adolescents Living with HIV

ART

Antiretroviral therapy

BHS

Beck Hopelessness Scale

CBT

Cognitive Behavioral Therapy

CDI

Child Depression Inventory

CPTS-RI

Childhood Post-Traumatic Stress Reaction Index

G-CBT

Group-based Cognitive Behavioral Therapy

GCP

Good Clinical Practice

MFG-FS

Multiple Family Group-based Family Strengthening

PLHIV

People Living with HIV

PTSD

Post-Traumatic Stress Disorder

SSA

Sub-Saharan Africa

TSCS

Tennessee Self-Concept Scale

Data availability:

On reasonable request, the data sets included in this analysis are available from the corresponding author.