Study design and participants
The study was an individually-randomized group treatment pilot trial that used a stepped-wedge design across three waves of young people (Supplemental Figure 1). The current report includes data on the outcomes at baseline and 6-months for each wave. One male and one female group from each site are included per wave and groups were filled on a rolling basis through July 2017. Our plan was to have groups including ages 12–14/15 years; 15–19 years; 18/19–24 years, but not enough youth under 15 years enrolled to create the younger groups. The mean age and age range for each group can be seen in Supplemental Figure 1.
YPLWH 12–24 years of age who attended the established adolescent HIV clinics at Kilimanjaro Christian Medical Centre (KCMC) or Mawenzi Regional Referral Hospital (MRRH), both located in Moshi, Tanzania, were approached to participate. Full HIV disclosure was required to attend either clinic, thus all patients attending clinic were fully aware of their HIV status. This was confirmed by each participant as part of the consent process. Inclusion criteria included the ability to understand and participate in the study intervention and receipt of ART. ART is prescribed and obtained by the clinic pharmacy free of charge during adolescent HIV clinic. Young people with cognitive disabilities precluding active participation in the consent process, intervention, or study assessments were excluded. There were no changes to the eligibility criteria or methods during the pilot trial.
Participants 18 years or older provided written informed consent for themselves; those under 18 years provided written assent with written consent provided by their caregivers. Caregivers were not treated as participants in the study and did not provide study data. Ethical approval for the study was granted from Duke University, KCMC, and the National Institute of Medical Review in Tanzania.
Study methods and results reporting followed the 2010 CONSORT checklist for reporting a pilot or feasibility trial [23]. Clinicaltrials.gov outlines the protocol (NCT 02888288) and a full study protocol is available upon request from the corresponding author.
Procedures
To provide sufficient information on the feasibility and acceptability of implementing the intervention and perform preliminary estimates of intervention effects, the targeted enrollment was 108–120 participants in 6 groups of 18–20 young people randomized per group and two groups per wave (see supplemental Figure 1). Due to attrition, enrollment was increased to 130 to meet the target sample size.
From May to July 2016, YPLWH were recruited via a study announcement made by study staff during monthly adolescent HIV clinic at KCMC and MRRH. Interested young people were invited to join a study waitlist that included their name, age, sex, site, and contact information. Young people were then grouped by age, sex, and clinic site in the order in which they signed the waitlist, and the first 60 from each site that made up groups of 18–20 with similar age and same sex were contacted to confirm their commitment to attend all intervention sessions and study visits. Young people were encouraged to invite their caregiver to attend two joint caregiver sessions, but enrollment was not contingent on caregiver participation. Young people who were available to participate and to provide informed written consent/assent were enrolled. Enrollment was not limited to only YPLWH who endorsed mental health challenges on baseline screening in order to establish acceptability and feasibility across a wide spectrum of mental health challenges and the hypothesis that all YPLWH could benefit from the intervention.
At enrollment and subsequent study visits (Supplemental Figure 1), participants responded to a structured interview conducted by one of two research assistants in Kiswahili. The interview included questions about demographics, perception of stigma, ART adherence, risk behaviors, and mental health symptoms. A hair sample was obtained to measure ART drug concentration as a cumulative measure of ART adherence and a five mL blood sample was drawn to measure HIV RNA. The participant file was reviewed to document the date the participant began ART, current ART regimen, most recent CD4 cell count, and (when available) the most recent standard-of-care HIV RNA result.
Randomization and masking
Once a male and female group of 18–20 young people of similar age, same sex, and same site were enrolled, they formed a “wave”. Each wave of individuals in the male and female groups were randomized by coin flip in the presence of the principal investigator and group leaders. Participants randomized to the SYV arm received the intervention; participants randomized to SOC continued with usual attendance to the monthly adolescent HIV clinic, but the SOC arm did not meet together as a study group. SOC, as provided to both study arms, included routine adherence counseling that was uniform across both sites and followed national Tanzanian guidelines [24]. All youth received adherence counseling in the form of health talks provided during the routine clinic; and, per guidelines, those with virologic failure (HIV RNA > 1000 copies) met with an adherence nurse on a monthly basis for three months to document youth adherence, discuss barriers to adherence, and provide support and suggestions to improve adherence. No mental health screening nor mental health specialists were available for referral, thus no specific mental health referrals were made at either clinic. The two research assistants conducting the study visits were blinded to the participant’s study arm assignment.
Intervention
SYV was developed based on our formative research that identified the common mental health needs of YPLWH in Tanzania [9, 21]. We demonstrated that 32% of YPLWH reported symptoms of mental health difficulties (depression, post-traumatic stress, and emotional/behavioral symptoms) as measured quantitatively with the same measures utilized in this study and described in detail below [9, 13, 22]. Components of the SYV intervention were adapted, in part, from a study of a cognitive behavioral therapy for bereaved orphans in Tanzania that reported improved outcomes on indicators of grief, post-traumatic stress, depression, and overall behavioral adjustment that were sustained at 3 and 12 months [25,26,27]. SYV was tailored further to the needs of YPLWH with additional components of interpersonal psychotherapy [28] and motivational interviewing [29] to address the specific challenges identified by those YPLWH in the formative study [9, 21]. The idea was to be responsive to the needs identified by this group of youth while preserving aspects of existing evidence-based treatment models. The group intervention involved ten group sessions, two designed to be held jointly with caregivers, and two individual sessions between the participant and one group leader (Fig. 2). A trauma narrative: “How I learned I have HIV” was conducted first in an individual session with the group leader to discuss their memories and then to prepare for what, if any, aspects of the disclosure event the participant wanted to share with other group members and/or caregivers. Caregivers were invited to participate in the first SYV session to understand the intervention content and share common challenges with other caregivers of YPLWH. In the sixth session, caregivers were asked to attend to hear the young person’s narrative about learning about HIV-infection from the young person’s perspective. A description of the scripted session-by-session intervention protocol has been previously described [22], and the full intervention is available upon request to the corresponding author or at sites https://duke.edu/sautiyavijana.
The study was implemented with Tanzanian group leaders trained to deliver the intervention and who were closely supervised twice weekly. Six young adults (age 23 to 30 years) with a background of either living with HIV and/or having prior experience with mental health research were provided an on-site intensive two-week training with the principal investigator and the U.S.-based clinical psychologist, both of whom provided subsequent weekly supervision for each session. Sessions lasted approximately 90 min and were held on Saturdays at a centrally-located site not affiliated with HIV care, providing a non-clinical space where young people could be comfortable discussing topics freely. Enrolled young people who missed more than two intervention sessions were not allowed to continue as content and group rapport builds from the previous sessions. These young people were offered the option to join a future wave as part of a cross-over group in the stepped-wedge design (Supplemental Figure 1).
Measures
All measures were administered in Kiswahili, the official language of Tanzania. Two different professional translators fluent in both Tanzanian Kiswahili and US English were involved. One translated the original English to Kiswahili, and the other back translated the Kiswahili to English. The back translated and original English were compared and clarifications incorporated. The formative study acted as the pilot of the mental health and stigma measurements [9].
Acceptability and feasibility have been previously reported elsewhere, [22] but did not include 6-month follow -up data as presented here. Measures of acceptability and feasibility included recruitment, attendance and participation during SYV sessions, follow up visits, and the ability of the group leaders to maintain fidelity to the intervention content based on supervisor observation and as documented on fidelity checklists for each session. Adherence to the scripted SYV manual was documented by a group leader observing (but not leading) each session. Fidelity checklists were based on intervention content whereby the observing group leader marked if the content was covered and documented the amount of time utilized in discussion. If content was skipped, the reason was documented and the content was revisited at a later session. The fidelity checklist was designed by the intervention developers (O’Donnell and Dow, Supplemental Table 2). Session notes were also documented by the observer noting each participant’s demeanor and participation in the group discussion. Fidelity checklists (quantitative data) and session notes (qualitative data) were discussed during the weekly post-session debrief meetings between the supervisors (Dow, Mmbaga, and O’Donnell) and group leaders and also ad hoc (as needed).
The Patient Health Questionnaire (PHQ-9) was used to measure symptoms of depression. The PHQ-9 has been validated relative to other diagnostic tools in several studies in Africa, [30, 31] including adolescent populations in Kenya and Ghana [32, 33] demonstrating high internal consistency with Chronbach alphas of > 0.70 [32, 34]. This measure includes nine questions with a response range of 0–27 with a score of 10 or greater used as a screening threshold suggestive of moderate to severe depressive symptoms [30, 31] and was used in our prior research [9]. The Strengths and Difficulties Questionnaire (SDQ) was used to measure young people’s emotional and behavioral symptoms. The scale includes 25 questions (5 prosocial behavior questions are not included in the composite score). A cutoff of 17 or higher (score range of 0–40) signals symptomatology of mental health difficulties with mean Cronbach α: 0.73 as assessed by the tool developers [35]. Trauma related symptoms were assessed using the UCLA Post Traumatic Stress Symptoms Exposure Screener and Reaction Index survey [36, 37] modified to a four-question Likert scale (None, Some, Much, and Most) based on Swahili translation as previously described with a cutoff of 18 or greater [9]. Prior use of the tool internationally demonstrated inter-rater reliability and criterion-related validity with children in Zambia [38] and demonstrated internal consistency among adolescent Somalian refugees (Cronbach α = 0.85) [39].
Stigma was measured using a truncated (10 question) Berger HIV Stigma Scale. Included questions related to both internal or ‘negative self-image’ (four questions, 2,3,7,12) and external stigma or ‘public attitudes’ (six questions, 9, 10, 14, 16, 20, 32) and were analyzed in these two categories as well as the total score for interpreting results (scale range 10–40) [40] and were used in our prior research [9].
Oral interviews were documented by group leader notes as part of the two individual sessions. Similarly, group discussion notes were documented by the group leader observer during intervention sessions.
ART adherence was measured by a validated three-question survey that asked participants to rate their adherence in the past 30 days transformed to a 0–100 scale [41]. The mean of these three items was the value used to measure self-reported adherence with a higher number indicating better adherence. A fourth open-ended question asked, “Many people have trouble remembering to take their medication. Please think about the last time you missed your medication. What was the reason?” in order better understand adherence challenges from the participant’s perspective. An objective metric of adherence was assessed by analyzing ART concentrations in hair. Hair samples (20–30 strands) were collected and concentrations of efavirenz (EFV), nevirapine (NVP), lopinavir (LPV), atazanavir (ATV) and ritonavir (RTV) were measured at the University of California, San Francisco (UCSF) Hair Analytical Laboratory (HAL) using validated, reproducible, liquid chromatography–tandem mass spectrometry (LC-MS/MS)-based methods [42]. For levels below the limit of quantitation (BLQ), a value of 0.05 (the BLQ equivalent) was used [43, 44]. Only participants contributing hair samples and receiving the same ART regimen at both baseline and the six-month time points are included in the analysis of hair data.
HIV RNA measurement was performed at the Kilimanjaro Clinical Research Institute Biotechnology Laboratory, which participates in international external quality assurance programs, using the Abbott m2000 platform (Des Plaines, Illinois, USA). Virologic failure was defined as plasma HIV RNA level > 400 copies/ml. HIV RNA lower than the minimal detectable amount of 40 copies/ml were imputed as half the minimum detectable amount (20 copies/ml).
Statistical analysis
Continuous variables were summarized using means and standard deviations (SD), and categorical variables were summarized using counts and percentages. Data were analyzed as intention-to-treat according to the participant’s randomization assignment, regardless of whether the participant attended or participated in the intervention if assigned to the intervention arm. Item nonresponse resulting in missing data was < 3%. If a question was skipped within a quantitative measure (ex. PHQ-9 or Stigma) the entire measure was treated as missing. Analyses were also run using imputed data of missing items with the mean of answered items and it did not appreciably change any of the results.
In the exploratory analyses for each measure, we fitted mixed effects models with a random intercept for group in the treatment arm to take into account the group treatment effects. Residual errors were allowed to vary by arm [45]. For each continuous outcome, we modeled the change in the outcome from baseline to six-month follow-up, adjusting for the baseline level of the outcome, wave, and site as covariates. To create a standardized hair measure collapsing the four anchor antiretrovirals (EFV, NVP, LPV, ATV) into a single measure, we defined a variable that was the log base 2 of the hair level minus the median log base 2 of the hair level for each drug and combined these “standardized” hair values into a single variable [43]. All analyses were conducted using Stata version 15.1 (StataCorp, College Station, TX).