Impact of a teen club model on HIV outcomes among adolescents in rural Neno district, Malawi: a retrospective cohort study

Abstract

Objective

To compare the impact of a teen club model to the standard care model on HIV treatment outcomes among adolescents (10–19 years of age).

Design

Retrospective cohort study.

Setting

HIV clinics in Neno district, Malawi.

Participants

Adolescents living with HIV enrolled in teen clubs (n=235) and matched participants in standard HIV care (n=297).

Outcome measures

Attrition from HIV care, defined as a combination of treatment outcomes ‘died’, ‘defaulted’ and ‘transferred out’.

Results

Over a 4-year follow-up period, adolescents who participated in the teen club had a significantly higher likelihood of remaining in care than those who did not (HR=2.80; 95% CI: 1.46 to 5.34). Teen clubs also increased the probability of having a recent measured viral load (VL) and BMI, but did not change the probability of VL suppression. The age at antiretroviral treatment initiation below 15 years (aHR=0.37; 95% CI: 0.17 to 0.82) reduced the risk of attrition from HIV care, while underweight status (aHR=3.18; 95% CI: 1.71 to 5.92) increased the risk of attrition, after controlling for sex, WHO HIV staging and teen club participation.

Conclusions

The teen club model has the potential to improve treatment outcomes among adolescents in rural Neno district. However, in addition to retaining adolescents in HIV care, greater attention is needed to treatment adherence and viral suppression in this special population. Further understanding of the contextual factors and barriers that adolescents in rural areas face could further improve the teen club model to ensure high-quality HIV care and quality of life.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• The analysis used routine programmatic data from electronic medical records to enable a rigorous evaluation of the teen club intervention.

• Data were available over a follow-up period of 5 years.

• Voluntary enrolment on full disclosure of HIV status may mean that teens who enrol in teen clubs are inherently different from those who decline to participate. Rigorous matching may have potentially accounted for this limitation.

• The duration of exposure to the teen club could not be quantified because of the unavailability of registration date into the teen club as part of the routine records.

Introduction

Malawi has a significant burden of the HIV, with approximately one million people currently living with the disease.^{1 2} Despite substantial gains towards attaining the 95–95–95 UNAIDS targets among adults, progress has been slower among adolescents and young people living with HIV.^{1 2} Children and adolescents have lower rates of retention in care and viral load (VL) suppression than adults.³ The incidence of new HIV infections and deaths among adolescents is also higher than that of the other age groups. In fact, HIV is one of the leading causes of morbidity and mortality among adolescents.⁴

Several challenges exist for adolescents in Malawi, including early sexual debut, loss of parents, substance abuse and poverty.^{5 6} Additionally, poor educational attainment and early pregnancies among girls sometimes result in transactional relationships with older men.^{6 7} These factors make adolescents more likely to engage in risky sexual behaviours, thereby increasing their chances of acquiring and transmitting HIV. Further, adolescents often have limited contact with the formal health system, making it a challenge to be tested for HIV, and if they are already living with HIV, to ensure consistent follow-up.⁸

Adolescents living with HIV who receive services in standard antiretroviral treatment (ART) clinics together with adults have poor medication adherence, low rates of viral suppression and high rates of loss to follow-up.⁹ They struggle with disclosure of their status due to discrimination and stigma. In turn, failure to disclose their status may lead to a lack of emotional support and connection with others with similar challenges.¹⁰ As adolescents comprise approximately 26% of Malawi’s population,¹¹ with many years of life ahead, focusing on this subpopulation is essential in order to control the HIV epidemic.^{8 12} To address challenges faced by adolescents living with HIV, differentiated ART service delivery models of care, known as teen clubs, have been introduced. A teen club is a psychosocial peer-support group intervention that empowers HIV-infected adolescents to build positive relationships, improve self-esteem and ultimately improve clinical and mental health outcomes.¹³ Teen clubs have demonstrated improved retention in care in South Africa, Namibia, Malawi, Eswatini and Kenya, mostly in urban settings.^13–17

The district of Neno is in Malawi’s southern region and is one of the most difficult to reach areas because of its topography and poor road network, which is impassable in the wet seasons. Partners In Health, a non-governmental organisation, has accompanied the Ministry of Health (MoH) in Malawi since 2007, employing a close partnership with national and local governments to build strong equitable health systems for vulnerable communities in Neno. The remote nature of Neno called for an integrated model of healthcare provision. According to this model, all 14 of the district’s health facilities provide decentralised treatment for a variety of chronic conditions—not only just HIV, but also hypertension, diabetes, chronic lung disease and mental illness, among others—so that patients can maximise the benefits of time-consuming and expensive journeys to the health facility. Compared with most districts in the country, which provide HIV services separately, HIV outcomes have improved in Neno throughout the years,¹⁸ and stigma has considerably reduced.

Following the recommendation of the MoH, teen clubs were first introduced in Neno in 2015 as a differentiated model of HIV care targeting adolescents. By May 2021 when this analysis was conducted, teen clubs were operating in 6 out of 14 health facilities. While some adolescents living in Neno joined the teen clubs, others continued receiving standard HIV care in the integrated chronic care clinics, often based on geography. This cohort study compares HIV outcomes and assesses factors associated with attrition from care between adolescents managed in teen clubs as compared with adolescents receiving standard HIV care in rural Neno district.

Methods

Two models of HIV care

In May 2021, approximately 8313 patients with HIV were actively receiving care in Neno, including 516 adolescents between 10 and 19 years of age (online supplemental table S1). All clinical management of HIV followed the Malawi Clinical HIV Guidelines (fourth edition).¹⁹ All data of enrolled patients with HIV were recorded in standardised paper-based charts called ‘mastercards’ with oversight by the MoH department of HIV and AIDS. Data from the mastercards for each HIV client were retrospectively entered into an electronic medical records system (OpenMRS) by medical informatics technicians. In the standard care model, adolescents are managed together with paediatric and adult patients. As an alternative, from 2015 onward, adolescents had the option of joining a teen club in their usual (or nearby) health facility. The number of teens in each club varied depending on the size of the HIV cohort at the site (online supplemental table S1).

Enrolment in teen clubs was voluntary on full disclosure to fellow members, healthcare providers and facilitators of the clubs. The teen clubs provided all of the regular HIV care, including clinical encounters with medication refills, VL testing, CD4 count testing, enhanced adherence counselling and screening for malnutrition, tuberculosis and sexually transmitted infections (online supplemental table S2). Additionally, teen clubs offered targeted education focused on disclosure, VL suppression, stigma, sexual and reproductive health and rights, and general aspects of positive living. These small group sessions intended to facilitate peer-to-peer sharing, learning and support. Adolescents were also involved in sports and games to develop a sense of community among the participants and leaders of the club. To facilitate attendance, adolescents received refreshments, meal allowances and transport reimbursement for each session. Monthly sessions were staffed and facilitated by clinicians, nurses, clerks and expert clients. Expert clients are persons living with HIV who adhere to ART and are virologically suppressed. They undergo a 6-week specialised training focused on providing emotional and social peer support. In their roles, expert clients assist teen club members with issues related to disclosure and medication adherence, while acting as role models and encouraging resilience and independence.

Study design and population

The analysis was conducted using a retrospective matched cohort study design with routinely collected patient data obtained from the OpenMRS electronic medical records. Eligible patients for the analysis included adolescents 10–19 years of age who were enrolled in the HIV programme in Neno for any part of the study period: between January 2015 (when the teen club model was initiated in Neno district) and May 2021 (when the data were extracted). The sampling frame of eligible adolescents contained 829 patients from all 14 health facilities in the district, comprised of 235 teen club participants and 594 non-participants (figure 1). All teen club participants (n=235) were selected for the cohort analysis. A corresponding sample of 297 (of 594) adolescents receiving standard care was selected, matched by sex, age at ART initiation and the year started on ART (figure 1) using the ‘joinby’ function in STATA software (V.14.1) which keeps all matching controls for each case. Individuals who could not be matched on the three criteria were excluded from the analysis.

Figure 1.

Data flow diagram describing how the analysis dataset of 532 individuals was derived from the full dataset of 829 adolescents enrolled in HIV care during any part of the study period (January 2015 to May 2021). The numbers between this figure and online supplemental table S1 do not correspond because this figure includes all patients who had been in the cohort over the study period.

Data analysis

All data were retrieved from the OpenMRS database and imported into Microsoft Access for data verification and cleaning. The main exposure variable was participation in a teen club during any part of the patient’s course of treatment. The main outcome variable was attrition from care, obtained by combining treatment outcomes of ‘died’, ‘defaulted’ and ‘transferred out’ (table 1). Additional analysis variables are listed in table 1, including a secondary outcome variable of VL suppression. Summary statistics included frequencies and percentages for categorical variables. For bivariate analyses, the χ² test was used to assess differences between adolescents enrolled in teen clubs and those receiving standard care. Time to attrition from care was calculated using the Kaplan-Meier curve.

Table 1.

Summary of analysis variables

Variable	Variable type	Definition
Year of ART initiation	Categorical	Categorised as 2000–2007, 2008–2015 or 2016–2021.
Age at ART initiation (years)	Ordinal	Calculated as the difference between ART initiation date and date of birth and categorised as <10, 10–14 and 15–19 years.
Sex	Categorical	Recorded at enrolment (male or female).
WHO clinical stage at ART initiation	Ordinal	WHO clinical stage at initiation, ranging from mild (stage 1) to severe (stage 4) HIV infection.
Enrolled in teen club	Binary	Binary variable: 1 if the patient has been exposed to the teen club model and 0 if not exposed.
Body mass index (kg/m2) at recent visit	Ordinal	Calculated at last visit from recorded weight and height and categorised as underweight (<18.5), normal (18.5–24.9), overweight (25–29.9) and obese (30+). Patients who had no recent (within the last 6 months) weight and height measurement were classified as ‘not available’.
VL (copies/mL) at recent visit	Binary	Recorded at last measurement and categorised as 1 for suppressed (<1000 copies/mL or <40 copies/mL, according to the recommended cut-off values in use at the time*) and 0 for unsuppressed. Those who had no recent (within 12–24 months, depending on recommended testing interval at the time*) VL measurements since starting ART were classified as ‘not available’.
Age at treatment outcome (years)	Ordinal	Calculated as the difference between outcome date and date of birth and categorised as <10 years, 10–14 years and 15–19 years. For patients who were still on ART at the time of analysis, the outcome date is equivalent to the analysis date.
Treatment outcome	Categorical	Currently on ART: next appointment date is in the future or less than 8 weeks since last missed appointment; Defaulted: 8 or more weeks since last missed appointment; Transferred out: Transferred out of the district; Died: Died

*Prior to June 2019, recent VL measurement meant within the last 24 months and suppression status was defined as <1000 copies/mL; from June 2019 onward, recent measurement meant within the last 12 months and suppression was defined as <40 copies/mL.

ART, antiretroviral treatment; VL, viral load.

Kaplan-Meier survival analysis with the log-rank test was used to compare the survival of adolescents in the teen clubs to those in standard care. Risk ratios were used to evaluate differences in outcome rates between the two groups. Finally, the factors associated with attrition from care among adolescents were determined using univariable and multivariable Cox proportional hazards regression models, using p<0.05 as a cut-off for statistical significance. In the final multivariable model, covariates with a p value<0.2 in univariable analysis were included. Analyses were performed using STATA (V.14.1) and R (V.3.2.3) software.

Patient and public involvement

Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.

Results

Patient characteristics in teen clubs and standard care

There were no differences between teen club participants and non-participants in terms of time of ART initiation, age at ART initiation and sex, confirming adequate matching (table 2). Most adolescents in the cohort were female (59%), initiated ART when they were <10 years of age (69%) and started ART between 2008 and 2015 (66%). There was also no difference between groups in terms of WHO clinical stage at ART initiation, with most participants (62%) having mild disease (stage 1). Teen club participants were more likely to have a recent body mass index (BMI) measurement (97% vs 93%) and VL measurement (87% vs 76%), but not necessarily to have viral suppression (table 2).

Table 2.

Comparison of patient characteristics between teen club and standard care

	Total n=532	Standard care n=297	Teen club n=235	P value
Year of ART initiation*
2000–2007	34 (7%)	15 (5%)	19 (8%)	0.17
2008–2015	352 (66%)	193 (65%)	159 (68%)
2016–2021	146 (27%)	89 (30%)	57 (24%)
Age at ART initiation*
<10 years	365 (69%)	198 (67%)	167 (71%)	0.21
10–14 years	142 (27%)	81 (27%)	61 (26%)
15–19 years	25 (5%)	18 (6%)	7 (3%)
Sex*
Female	316 (59%)	181 (61%)	135 (57%)	0.44
Male	216 (41%)	116 (39%)	100 (43%)
WHO clinical stage at ART initiation
Stage 1	330 (62%)	183 (62%)	147 (62%)	0.96
Stage 2	36 (7%)	20 (7%)	16 (7%)
Stage 3	126 (24%)	70 (23%)	56 (24%)
Stage 4	40 (7%)	24 (8%)	16 (7%)
Recent body mass index (within 6 months of last visit date)
Underweight	158 (29%)	97 (33%)	61 (26%)	0.06
Normal	227 (43%)	112 (38%)	115 (49%)
Overweight	117 (22%)	65 (22%)	52 (22%)
Not available	30 (6%)	23 (7%)	7 (3%)
Recent viral load status (within 12–24 months of last visit date)†
Suppressed	275 (52%)	148 (50%)	127 (54%)	0.003
Non-suppressed	155 (29%)	77 (26%)	78 (33%)
Not available	102 (19%)	72 (24%)	30 (13%)
Age at treatment outcome
10–14 years	264 (50%)	164 (55%)	100 (43%)	0.04
15–19 years	268 (50%)	133 (45%)	135 (57%)
Treatment outcome
Currently on ART	422 (79%)	210 (71%)	212 (90%)	<0.001
Transferred out	78 (15%)	62 (21%)	16 (7%)
Defaulted	21 (4%)	15 (5%)	6 (3%)
Died	11 (2%)	10 (3%)	1 (<1%)

P value<0.05 indicates statistically significant difference in variable distribution between standard care and teen club models.

*These variables were used in matching.

†Prior to June 2019, recent viral load measurement meant within the last 24 months and suppression status was defined as <1000 copies/mL; from June 2019 onward, recent measurement meant within the last 12 months and suppression was defined as <40 copies/mL.

ART, antiretroviral treatment.

Attrition and associated factors

The Kaplan-Meier curve shows a higher probability of being retained on ART among adolescents enrolled in teen clubs versus those in standard care (HR=2.80; 95% CI: 1.47 to 5.34) over the course of 50 months from treatment initiation (figure 2). The result is confirmed by the univariable Cox proportional hazards regression model (table 3). The curves remain close, but the log-rank test indicates statistical significance over the follow-up period (p=0.002). For those enrolled in teen clubs, retention on ART was at 96%, 91% and 83% after 10, 30 and 50 months in care, respectively. In standard ART care, retention over the same time interval was at 91%, 85% and 76%.

Figure 2.

Kaplan-Meier survival curve for adolescents enrolled in teen clubs (top teal line with shaded 95% CI) and standard care (bottom red line with shaded 95% CI) over 50-month follow-up. Nine patients were censored at month 0 due to a very short-time interval (<30 days) between dates of antiretroviral treatment initiation and outcome.

Table 3.

Summary of univariable and multivariable Cox proportional hazards regression analyses of predictors of attrition* from HIV care

	Univariable analysis		Multivariable analysis
	HR (95% CI)	P value	aHR (95% CI)	P value
Enrolment in teen club
Yes (ref)	1.00		1.00
No	2.80 (1.47 to 5.34)	0.002	2.39 (1.24 to 4.63)	0.009
Age at antiretroviral treatment initiation
<10 years	0.02 (0.01 to 0.05)	<0.001	0.04 (0.01 to 0.14)	<0.001
10–14 years	0.25 (0.12 to 0.54)	<0.001	0.37 (0.17 to 0.82)	0.015
15–19 years (ref)	1.00		1.00
Sex
Female (ref)	1.00		1.00
Male	0.63 (0.36 to 1.12)	0.14	0.98 (0.52 to 1.84)	0.955
WHO clinical stage at initiation
Stage 1 (ref)	1.00		1.00
Stage 2	0.17 (0.02 to 1.26)	0.083	0.59 (0.07 to 4.71)	0.614
Stage 3	0.14 (0.04 to 0.46)	0.001	0.74 (0.20 to 2.80)	0.660
Stage 4	0.15 (0.02 to 1.10)	0.062	0.67 (0.08 to 5.70)	0.714
Recent body mass index (within 6 months of last visit date)
Underweight	1.64 (0.91 to 2.97)	0.100	3.18 (1.71 to 5.92)	<0.001
Normal (ref)	1.00		1.00
Overweight	0.35 (0.12 to 1.03)	0.057	0.34 (0.11 to 0.99)	0.050

Attrition from HIV care is defined as a combination of treatment outcomes ‘died’, ‘defaulted’ and ‘transferred out’.

The univariable Cox proportional hazards regression analysis showed that younger age at ART initiation (both <10 and 10–14 year age groups as compared with the 15–19 reference group) was associated with a lower risk of attrition (p<0.001). Higher WHO clinical stage seemed to decrease the likelihood of attrition, although only stage 3 was statistically significant. BMI was borderline significant in the univariable model, and sex was not associated with attrition (table 3). VL also was not statistically significant (not shown in table 3 because it was not significant in either model). Adjusting for all covariates in the multivariable model slightly reduced the effect of remaining in the standard care model on attrition to aHR=2.39 (95% CI: 1.24 to 4.63). The effect of lower age at ART initiation remained significant in the multivariable model. Being underweight became a significant risk factor in the multivariable model (aHR=3.18; 95% CI: 1.71 to 5.92). The WHO clinical stage lost significance and sex remained not significant (table 3).

Discussion

Malawi has made significant progress toward meeting the UNAIDS 95–95–95 targets among the general population aged 15–49 years. In this population, 88.3% of those living with HIV knew their status, 97.9% of those aware of their status were on ART and 96.9% of those on ART were virally suppressed in 2020.² However, progress is further behind among young people (15–24 years of age), with the corresponding values of 76.2%, 95.9% and 90.7%, and no data for children aged<15 years. Within this context, there is a need to investigate differentiated programmes that can improve HIV care in young people.

This study demonstrated improved HIV treatment outcomes among adolescents enrolled in teen clubs between 2015 and 2021 in comparison to those in standard care. For those in standard care, we found the overall ART retention rate at 24 months to be approximately 86% compared with those in teen clubs at 93% in the same time frame. The standard care retention rate is comparable to that of a Namibian study with 89% retention at 24 months in the standard care group.¹⁴ Similarly, elsewhere in Malawi, higher retention rates were observed in teen clubs (92%–96%).^{15 20} Retention in our teen club was also higher than that of a specialised adolescent programme in Uganda at 24 months (83%).²¹

We found no statistically significant difference in viral suppression between those in the teen club (62%) and those in standard care (66%). On average, 64% of the adolescents in both standard care and teen clubs were classified as virally suppressed (<1000 or <40 copies/mL, depending on the VL testing guidelines at the time), falling below the UNAIDS 95% target. This value was also far below the 90.7% national estimate of viral suppression in the 15–24 age group in 2020.² Similar studies with adolescents conducted in South Africa, Malawi and Eswatini showed higher viral suppression rates of 83%–91%.^{13 15 16} Further work is needed in hard-to-reach areas such as Neno district to identify and explore factors behind low viral suppression in this cohort of adolescents, as well as closing gaps in missing VL data in patient records, which may have impacted the results.

Sex did not impact the likelihood of attrition. This finding is consistent with a Namibian study that found no difference in attrition at 24 months between males and females.¹⁴ However, other studies have suggested higher attrition among females compared with males.^{22 23} Older age, particularly the 15–19 age group, and underweight status were risk factors for attrition after controlling for teen club participation, suggesting that greater attention is needed to these high risk groups of adolescents in both teen club and standard care models. The finding of higher risk of attrition in the 15–19 age group than among younger age groups is consistent with other studies with adolescents.²³ The reasons for why older adolescents experience adherence challenges are multifactorial, including potential effects of stigma, substance abuse and economic struggles that are associated with transitioning from adolescence to adulthood.^{24 25} Further study of different implementation models to improve support for this population is required.

Strengths and limitations

The study had several strengths and limitations. One of the key strengths is the utilisation of routine programmatic data from electronic medical records to enable a disaggregated evaluation of the differentiated care of teen clubs; the depth of analysis would not be possible with aggregated data from registers. Additionally, a relatively long follow-up period of adolescents in HIV care was available for analysis (up to 5 years for individual patients). The study also had several limitations. First, enrolment in teen clubs was voluntary, with full disclosure as a prerequisite—this may have introduced selection bias, with adolescents who chose to enrol in teen clubs being inherently different from those who declined to participate. To account for this potential bias, the characteristics of non-participants were matched as closely as possible to those of teen club participants. Second, the last recorded outcome for each patient was used, which is deficient for some outcomes. For example, looking at patients who currently have a ‘defaulted’ status does not account for those who have a history of frequent defaulting but currently happen to be active in care. The same method was applied to teen club participants and non-participants; therefore, this limitation is not likely to have introduced any bias into the analysis. Lastly, the duration of exposure to the teen club could not be quantified because of the unavailability of registration date into the teen club as part of the routine records. Additional fields to document teen club exposure (start and end dates, as applicable) will be incorporated into the electronic medical record in the future, enabling further studies.

Conclusion

Teen clubs provide a tailored patient-centred approach to HIV care, and in this adaptation, social support, peer support and a dedicated clinical team managing the teen clubs play a role in reducing attrition. The findings of this study suggest that expanding teen clubs in the district has a potential to retain adolescents in HIV care, which is an important first step. Greater attention in this population is needed to treatment adherence and viral suppression, critical aspects of HIV care for adolescents. Further qualitative research into specific barriers that adolescents face is warranted, with a special emphasis on older and underweight high-risk groups. Lastly, incorporating additional socioeconomic support modalities into the teen club model could help adolescents overcome treatment barriers and enable a smooth transition to adulthood.

Data availability statement

Data are available upon reasonable request. Data may be obtained from a third party and are not publicly available. Data may be obtained from a third party and are not publicly available. The authors have the data until the publication of the article. The authors can facilitate data access if requested with proper permission from the Ministry of Health.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

The study was approved by the Malawi National Health Sciences Research Committee (protocol # 1216). Informed consent was waived because all data were collected through routine clinical care and deidentified before being retrospectively analysed as per the study protocol.