SUNRISE CRT: Radio Intervention to Stimulate Early Child Development

Overview

SUNRISE aims to develop and evaluate a radio campaign to promote nurturing care behaviours in the first 3 years of life, particularly responsive parenting and providing opportunities for early learning. The campaign will be broadcast by local radio stations in Burkina Faso for 3 years and comprise: 60-second radio 'spots' including scripted dramas and modelling of responsive parent-child interactions, broadcast 10 times a day in weekly cycles; plus long format evening programs 2-3 times/week, incorporating longer dramas, real life testimonials and practical 'how to' advice, with space for listeners to call in, express opinions and ask questions. The campaign's impact on early child development (ECD) will be evaluated using a cluster RCT design, alongside a process and health economic evaluation.

STUDY AIMS SUNRISE aims to develop and evaluate, in a cluster randomised trial, a radio campaign in Burkina Faso, to promote nurturing care behaviours in the first 3 years of life, focusing particularly on responsive parenting and providing opportunities for early learning. Children's developmental outcomes will be assessed over the lifetime of the intervention until children are aged 30-32 months. TRIAL CLUSTERS The trial clusters are defined by the broadcast reach of local radio stations. Eligible clusters for the trial were identified as follows. First 23 potential clusters in areas of Burkina Faso where there is no security risk were identified by modelling the reach and signal strength of local radio stations based on tower location, antenna characteristics, and terrain features. A media survey was then carried out to check the radio stations listened to in each of these clusters. This resulted in 15 clusters which satisfy the following conditions: there are at most two dominant radio stations within the cluster; there is no discernible broadcast from stations in neighbouring clusters; they are not based in the large cities of Ouagadougou or Bobo-Dioulassou; and they are not subject to any security risk. MAPPING EVALUATION AREAS (EAs) Trial participants will be recruited from an evaluation area within each cluster that is close to the radio station(s), has a high signal strength confirmed by a motorbike survey and likely to have a high listenership based on the media survey. The following will be excluded from the area: towns, villages within 5 Km of towns, villages already on the national electricity grid and villages with populations of 5000 or more as they are likely to be a priority for the national electrification programme. BASELINE SURVEY DATA COLLECTION A baseline survey was carried out before the start of the campaign. Participants were all eligible mothers with a child/children aged less than 3 years. Data comprised socio-demographic and socio-economic factors; parental education levels and literacy; and radio ownership and listening patterns. In addition, the fieldworkers administered the following to mothers of children aged 2 years: * The short form of the Caregiver Reported Early Development Instruments (CREDI) which is designed to serve as a population-level measure of ECD for children from birth to age three, and which has 20 questions primarily focused on milestones and behaviours that are easy for caregivers to understand, observe, and describe. * Questions from the support for learning/stimulating environment section of the Family Care Indicators (FCI) survey developed by UNICEF to assess key caregiving/family home environment variables considered critical for early child development. This has been used very extensively, including in the global MICS surveys. Data from the baseline was used to: 1. Inform the content of the SUNRISE radio campaign. 2. Provide a sampling frame for the surveillance system of 3-monthly household visits that will be used to recruit trial participants and conduct ECD assessments with GIS maps of households within the EAs. 3. Carry out the restricted randomization (see section below). RANDOMISATION OF CLUSTERS The SUNRISE campaign will be broadcast by the local radio stations in eight of the clusters chosen at random to be in the intervention group; the other 7 clusters will be in the control group and their radio stations will continue to broadcast as normal. This was carried out using restricted randomization to ensure that the intervention and control clusters were balanced at baseline with respect to radio exposure and predictors of ECD outcomes as follows: Mean CREDI z-score; difference between intervention and control groups to be a maximum of 0.1. Mean % key FCI behaviours followed by mothers; difference between intervention and control groups to be a maximum of 10%. * mothers listening to the radio in the last week; difference between intervention and control groups to be a maximum of 10%. * mothers who are literate; difference between intervention and control groups to be a maximum of 10%. The restricted randomization was carried out by the independent trial statistician using the cvcrand command in Stata. This performs covariate-constrained randomisation which is suitable for cluster randomised trials with a small number of clusters. PARTICIPANTS At least 125 eligible children born in each cluster after the campaign has been running for three months will be enrolled together with their families. They will be identified through 3-monthly surveillance visits to all compounds identified at baseline; it is expected that recruitment will be achieved within a 6-month birth period. They will form the trial cohort and be followed 3-monthly until the end of the radio campaign. The intervention's impact on ECD nurturing care behaviours and ECD outcomes in the first 3 years of life will be evaluated. DATA COLLECTION The trial will be supported by a surveillance system with regular 3-monthly visits by the cluster fieldworker (FW) to all households. These visits will be used to recruit the trial participants, maximize their follow-up during the trial and collect a variety of data on ECD behaviours and outcomes as well as process evaluation data on radio listening patterns. Details of all primary and secondary outcome assessments are presented in the next section. SAMPLE SIZE Assessments are anticipated to be carried out on at least 100 children per cluster, after allowing for a maximum 20% loss to follow-up from recruitment. This will give 90% power to detect an effect size of 0.25, assuming an intra-class correlation of 0.01 (personal communication, from the Save the Children ECD trial in Rwanda) and 80% power to detect an effect size of 0.21. It will also give 90% power to detect an effect size of 0.4 on other ECD outcomes, the level achieved in several successful ECD interventions, assuming an intra-class correlation of 0.04, which is informed by unpublished results from the SPRING trial, and 80% power to detect an effect size of 0.33. STATISTICAL METHODS Findings will be reported according to the CONSORT guidelines for cluster randomized trials. Analyses will be intention to treat and include all data from trial children and their families, regardless of their exposure to the radio campaign. Random-effects linear regression models using individual-level data will be used to adjust for the clustered design and any imbalances between intervention and control arms. Effect sizes will be presented as standardized mean differences with 95% confidence intervals (CI). The data analysis plan will be published prior to the end of recruitment. PROCESS EVALUATION Tracking and post-broadcast feedback Independent radio trackers will determine if spots were broadcast as planned. Post-broadcast feedback on the campaign will be obtained through periodic focus group discussions (FGDs) and semi-structured key informant interviews. These data will be collected from at least two different communities every quarter, throughout the campaign duration. Surveillance monitoring data Monitoring data on radio coverage will be collected every 3 months from trial mothers including listenership in the past week and questions concerning child development advice heard on the radio. These data will also be collected from a separate monitoring cohort of mothers who participated in the baseline survey together with 6-monthly questions on key behaviours \& knowledge targeted by the campaign. QUALITATIVE RESEARCH Acceptability will be measured through 25-30 qualitative interviews with caregivers, who will be asked about their experience and views on the intervention approach and content. Data from the post-production feedback research will also be utilized. The mechanisms and barriers to behavioural uptake will be explored through qualitative interviews with a further sample of 25-30 families. Approximately five intervention communities will be selected to reflect the diversity within the study areas. Within communities' respondents will be selected to give diversity in key characteristics such as parity and socio-economic status. Qualitative analysis will begin through the identification of themes during daily debriefing sessions. Key analytical categories will then be identified, and the interviews systematically indexed into the categories and interpreted and written up. During analysis, consensus coding methods will be used for initial interviews. ECONOMIC EVALUATION SUNRISE offers the opportunity to fully cost, from the provider perspective, a large-scale intervention to promote ECD. These provider costs will be used to calculate a range of incremental cost effectiveness ratios including cost per beneficiary and cost per unit of cognitive gain. The full provider costing will then form the basis for affordability estimates calculated as a percentage of national gross domestic product. The cost of a fully-scaled programme will also be compared with current health and education spending in context. Simple modelling of the gains from improved early childhood stimulation, using estimates from the Lancet Series in Early Child Development, would then be used to conduct a fiscal space analysis to explore the expansion in national income likely to arise from the effects of the program, and what percentage of that growth in income the program cost constitutes. Finally, the equity impact of this intervention will be measured, as early child development interventions are known to attenuate the effects of poverty. DATA MANAGEMENT Data protection and confidentiality procedures will be specified and followed, in keeping with Good Clinical Practice (GCP) and the GDPR 2018. Data collection will be carried out using REDCap which is a secure web application for building and managing online surveys and databases; REDCap is 21 CFR Part 11, FISMA, HIPAA, and GDPR compliant. All data (including video observations) will be collected on secure, password protected handheld tablets within an encrypted database, with the data transferred daily to a secure password protected cloud server, managed by IPA in Ouagadougou. All electronic data will be backed up locally on a weekly basis on encrypted external hard drives and stored offsite in Ouagadougou. Encrypted transfer of pseudonymized quantitative data to LSHTM will take place monthly via secure internet connection to a GCP-compliant LSHTM database, which allows full auditing of all data input and edits. Qualitative interviews will be audio-recorded on digital audio recorders. As video and audio data cannot be de-identified, transfer of this material to the UK will be done through UCL's Data Safe Haven, which is a dedicated highly secure system for storing and handling sensitive data. The Data Safe Haven has been certified to the ISO27001 information security standard and conforms to the NHS Information Governance Toolkit. Video recordings will be viewed only by members of the SUNRISE team and for the purposes of the research, unless further explicit written consent is obtained. Data concerning cluster allocation will not be stored in the local or LSHTM research databases, but held separately by the trial statistician in a secure password protected file. GOVERNANCE The trial will be overseen by an independent Trial Steering Committee (TSC) and Data Monitoring and Ethics Committtee (DMEC). TSC members are: Linda Richter (Chair), Jose Martines, Therese Stukel, Ali Sie. DMEC members are Rajiv Bahl (Chair)., Thierry Martens, Charlotte Tawiah-Agyemang, Susan Walker, Helen Weiss (independent statistician). ETHICS APPROVAL As a non-CTIMP, no regulatory approvals for medicinal products will be required. The study will be reviewed by the ethics committees of the Ministry of Health, Burkina Faso, UCL and LSHTM. LSHTM will be the study sponsor. The trial will be registered with ClinicalTrials.gov prior to initiation.