Reporting and methodology for the proposed study follow the Standard Protocol Items Recommendations for Interventional Trials (37), as well as Consolidated Standards of Reporting Trials (CONSORT) - extension to randomised pilot and feasibility trials (38). The intervention has been reported using the Template for Intervention Description and Replication (TIDieR) (39).
Design
This will be a two-group parallel, open label pilot RCT. The study will have two arms, an intervention group and a control group.
Setting
This study will be conducted in a peri-urban community within the south of the eThekwini municipality, KwaZulu-Natal, South Africa. Recruitment, baseline quantitative assessments and qualitative interviews will be conducted face-to-face. Whereas intervention delivery and end of trial assessments will be conducted telephonically.
Participant recruitment
Participants will be recruited via:
-
An existing study database of caregivers of ALHIV who were enrolled in an adolescent study
-
Information sessions within local HIV clinics
Participant eligibility criteria
Participants will be eligible to participate in the trial if they meet the following criteria:
Participants will be excluded from the study should they have any of the following:
-
Caregiver who is unable to comprehend the nature of the study during the information session in either English or isiZulu
-
A caregiver who is experiencing distress, suicidal ideation, or requires urgent medical attention
Screening
Initial contact will be face-to-face at the study clinic. Participant eligibility will be assessed, followed by explanation of the study, assessment of study understanding, and completion of consent procedures (Appendix 1).
Figure 1: Standard protocol items: recommendation for intervention trials (SPIRIT) flow diagram
Baseline and endline assessment
Post-screening, a baseline assessment will be conducted (Appendix 2). After the intervention period, the endline assessment will be conducted. For the assessments, the following data will be collected on an electronic interviewer-administered REDCAP questionnaire.
-
Socio-economic status (e.g. demographics, dwelling type, household occupants, household dwellers’ ages), labour and income (e.g. employment status, income range), and household and social outcomes (e.g. number of children on support grants, source(s) of income).
-
Food security will be assessed using the Food Insecurity Experience Scale (FIES). The FIES has demonstrated good internal reliability in this setting (Rasch reliability infit statistic > 0.7) (40).
-
Depressive symptoms will be measured using the 10-item Centre for Epidemiological Studies Depression Scale (CES-D-10). The CES-D-10 has been psychometrically evaluated in a South African sample and showed acceptable internal consistency across different language groups (Cronbach’s α = 0.69–0.89), and concurrent validity when compared to other depressive symptom measures commonly used (e.g. Patient Health Questionnaire, WHO Disability Assessment Schedule) (41). The scale has also shown good convergent validity (regression co-efficients between known psychosocial measures and EDS ranged from 0.17–0.19, p < 0.001)(42) and internal reliability (Cronbach's α = 0.86)(43). A cut-off point of 12 is deemed optimal to correctly classify individuals with a diagnosis of depression for a South African sample (44).
-
Stigma: The Everyday Discrimination Scale is a recommended scale for measuring intersectional stigma, particularly among people living with HIV (45), and has demonstrated good psychometric properties in samples from the United States of America (46).
Primary clinical outcome measure
-
Wellbeing will be measured using the Care-related Quality of Life (CarerQol) instrument. The CarerQol instrument is a preference-based measure that was developed for use in economic evaluations (47), and has exhibited moderate construct validity (correlation co-efficient with CarerQol and self-rated burden dimensions ranging from 0.2–0.4) among older carers in the Netherlands (48). Clinical and convergent validity have also been supported (49).
Randomisation and allocation concealment
A statistician on the research team who is not involved in the intervention or outcome measure assessments will randomly allocate participants to the intervention or control group. The randomisation list will be generated in STATA (Stata Corp, College Station, TX, V17) using block randomisation techniques. This method is used to ensure a sample size balance across groups over time. Blocks will be small and balanced with predetermined group assignments, which keeps the numbers of subjects in each group similar at all times. A seed of 123 will be used for reproducibility of the same randomisation output obtained. Four blocks will be generated in STATA (i.e. 2 blocks each will be assigned to each study arm) and the list exported to Excel. This list will be emailed to a second researcher who will assign the intervention or control programme to each participant.
Blinding
This is an open label trial as field staff will be involved in the intervention delivery and study assessments. The statistician involved in the randomisation will be blinded to study arm. Participants will be encouraged not to disclose details regarding their wellbeing programme to other participants during the trial.
Intervention
Participants in the intervention group will receive an economic incentive package comprising a monthly motivational SMS and cash to the value of ZAR 350 (23 USD) for three months. The SMS will promote key behavioural economic principles (aspiration framing, loss aversion, altruism), aligned to social wellbeing dimensions that matter to people in this context (Table 1) (50). The R350 incentive amount was chosen because it is within the margin of the COVID-19 Social Relief of Distress (SRD) grants in South Africa (51). Participants will have 30 days to claim the receipt of their cash. It will be delivered via electronic local banking services. In line with the human centred design (HCD) approach, themed messages will be co-developed with a caregiver advisory board (CAB) were revised and used by the research team to design SMS messages to be sent to participants.
Table 1
Intervention messages and underpinning behavioural economic principle
Message
|
Behavioural Economic principle
|
Message content
|
1
|
Aspiration framing
|
“Taking care of your own needs is important so you can be there to watch your child grow”
|
2
|
Loss aversion
|
“If you need to speak to someone you can call the South African Depression and Anxiety Group at no cost to you (Number:)”
|
3
|
Altruism
|
“Taking care of your child and family’s needs are important”
|
Control
The control group will receive a once-off SMS encouraging access to public health clinic services as per the standard message sent by the Department Health in South Africa to patients accessing clinic services. Participants in this group will not receive cash or a motivational SMS.
Qualitative study
A subset of participants in the intervention (n = 8) and control (n = 8) arm will be invited to participate in in-depth interviews after their baseline and endline quantitative assessments. They will be purposively sampled to ensure the sample is reflective of age, socio-economic status and HIV status. The aim of these IDIs is to provide an authentic account of wellbeing experiences among caregivers in this setting (52), to understand how experiences differ by arm and what could explain these differences. Topic guides (Appendix 3) will be mapped on dimensions from our wellbeing scales (e.g. financial problems, relational problems, household and caregiving chores, mental and physical health, belonging) and preliminary findings from the trial. For those in the intervention arm, specific elements regarding the intervention will be probed as part of endline in-depth interview (i.e. timing, cash incentive amount, delivery method, safety concerns, SMS content, utilisation of the cash incentive, access of mental health support, recommendations to improve the intervention).
Sample size
Our primary outcome is pre-post between difference in caregiver wellbeing scores. Assuming a mean caregiver wellbeing VAS score of 6.5 in M1 versus 8.5 in M5 (SD1.9) in the intervention arm, we would need a minimum sample size of n = 15 per arm to have 80% power to detect a treatment effect of between\({\delta }\) = 0.6 − 0.5 (49, 53).
Statistical analysis
Descriptive statistics will be used to present feasibility indicators and baseline sample characteristics at M1, including means (standard deviations) for normal data and, median (interquartile range) for non-normal data Differences in baseline characteristics between the intervention versus control arm, stratified by age-range of ALHIV (10–14 yrs. vs. 15–19 years) will be assessed using a Chi-square test for categorical variables and Kruskal-Wallis H test for continuous variables. Baseline variables that differ between groups will be included as covariates in the regression analysis. To compare primary outcomes, we will analyse wellbeing score change at M1 and M5 between arms using linear regression. Furthermore, we will explore changes in scores between M2 and M5. To include all participants in the primary analysis (intention to treat analysis), we will perform multiple imputation methods if the loss of those missing from follow-up is significant, and assumptions are met. Participants who complete the questionnaire at baseline (M1) and endline (M5) will be included as per protocol. All statistical analyses will be performed using Stata (Stata Corp, College Station, TX, V17) (see Table 2 for participant timeline).
Table 2: Participant timeline
Progression criteria
The following criteria must be met in order to consider progression to a main RCT: