Survive and Thrive in Brazil: The Boa Vista Early Childhood Program: Study protocol of a randomized controlled trial

A growing body of evidence suggests that early life health and developmental outcomes can be improved through parental support programs. The objective of this project was to test the feasibility, impact, and relative cost-effectiveness of an adapted “Reach Up and Learn” program delivered through home visiting programs as well as through center-based parenting groups on child health and development in the municipality of Boa Vista, Brazil. A randomized stepped wedge design was used to roll out and evaluate the two parenting platforms in Boa Vista municipality. A total of 39 neighborhoods with high vulnerability index were selected for the study. For the first phase of the program, 9 neighborhoods were randomly selected for home visits, and 2 were randomly selected for the center-based parenting groups. In the second phase of the program, 10 neighborhoods were added to the home visiting program, and 8 were added to the center-based program. In the final phase of the program, the remaining 10 control areas will also be assigned to treatment. Study eligibility will be assessed through a baseline survey completed by all pregnant women in the 39 study areas. Pregnant women will be eligible to participate in the study if they are either classified as poor, were under age 20 when they became pregnant, or if they indicate to have been exposed to domestic or sexual violence. To assess program impact, an endline survey will be conducted when children reach age 2. The primary study outcome is child development at age 2 as measured by the PRIDI instrument. Secondary outcome will be infant mortality, which will be assessed linking municipal vital registration systems to the program rollout.

2 Abstract Background A growing body of evidence suggests that early life health and developmental outcomes can be improved through parental support programs. The objective of this project was to test the feasibility, impact, and relative cost-effectiveness of an adapted "Reach Up and Learn" program delivered through home visiting programs as well as through center-based parenting groups on child health and development in the municipality of Boa Vista, Brazil.

Methods
A randomized stepped wedge design was used to roll out and evaluate the two parenting platforms in Boa Vista municipality. A total of 39 neighborhoods with high vulnerability index were selected for the study. For the first phase of the program, 9 neighborhoods were randomly selected for home visits, and 2 were randomly selected for the centerbased parenting groups. In the second phase of the program, 10 neighborhoods were added to the home visiting program, and 8 were added to the center-based program. In the final phase of the program, the remaining 10 control areas will also be assigned to treatment. Study eligibility will be assessed through a baseline survey completed by all pregnant women in the 39 study areas. Pregnant women will be eligible to participate in the study if they are either classified as poor, were under age 20 when they became pregnant, or if they indicate to have been exposed to domestic or sexual violence. To assess program impact, an endline survey will be conducted when children reach age 2.
The primary study outcome is child development at age 2 as measured by the PRIDI instrument. Secondary outcome will be infant mortality, which will be assessed linking municipal vital registration systems to the program rollout.

Discussion
This trial will assess the feasibility and impact of parenting programs rolled out at medium 3 scale. The results from trial should create evidence urgently needed for guiding Brazil's national Criança Feliz program as well as similar efforts in other countries.

Background
Brazil has made remarkable progress with respect to child nutrition and child survival over the past decade, with particularly impressive results in large urban areas. Infant mortality has dropped by roughly 50% since 2000 to 14 deaths per 1000 births across the country [1]. Most of the remaining mortality occurs during the neonatal period. More than 25,000 newborns are estimated to die each year within the first 28 days of their life in Brazil, with most deaths occurring in the first seven days after birth. Previous studies suggest that one of the most effective ways to prevent such deaths are home visiting programs, which support mothers in the first weeks of infant's lives, promote breastfeeding and kangaroo mother care, and ensure appropriate medical care when needed [2][3][4][5][6]. From a child health and development perspective, best outcomes have generally been achieved when continued support was provided to mothers from pregnancy throughout the first years of children's life. A nurturing and stimulating environment sensitive to maternal needs as well as the child's health, nutritional and emotional needs is not only essential for healthy development but can also mitigate harmful effects of risk factors related to poverty and lack of parental resources [7][8][9][10][11][12][13][14][15][16][17].
Parental programs to promote nurturing environments have been successfully implemented in a range of low-and middle-income countries, including UNICEF's Care for Child Development (CCD) framework, Jamaica's Reach Up and Learn program, Early Head Start in the USA and UK's Sure Start [ 9,[18][19][20][21][22].
Brazil has successfully implemented its "family health strategy" -a health focused home visiting program coordinated by primary care units -since 2006 [23][24][25]. In 2016, the government passed the "Criança Feliz" law to further increase the support provided to 4 vulnerable families [26]. Even though the program foresees home visits to all vulnerable families, there is very little guidance on how such programs should be implemented in Brazil, and what kind of content should be promoted.
In this project, we assess the scalability of a locally adapted version of an early childhood program previously developed in Jamaica. Between 2014 and 2016, the Jamaican "Reach up and Learn" curriculum was adapted to the Brazilian context and tested through a small randomized trial in São Paulo (ClinicalTrials.gov NCT0270400). The content, delivered through fortnightly home visits, comprises parenting skills tips as well as child stimulation activities supported with basic toys, addressing motor, social emotional and cognitive development and language for families with children from 9-36 months.
While the reach of this program was not universal in this original trial, the estimated impact on participating mothers was large, and the program was well received at the local level (manuscript with main results under review). While this curriculum was delivered through home visits in the original study in Sao Paulo, recent evidence from Bangladesh [6] suggests that similarly positive impacts can be achieved through center-based parenting groups that are substantially cheaper from a logistical perspective.
Given this and given also that the municipality had previously invested in a center-based model, it was decided to test both delivery platforms within this project. Given the relatively high burden of neonatal mortality in the study area, we also decided to start the intervention earlier. By enrolling pregnant women at the beginning of their third trimester, the revised intervention program aims to improve support for mothers during the antenatal and neonatal periods with the ambition to increase health service utilization and reduce the risk of adverse birth outcomes. Figure 1 presents the theory of change through which the curriculum may improve child survival and development.
Objectives: The objective of this study is to rigorously assess the impact of an extended version of the previously tested and validated parenting curriculum on child survival and development in Boa Vista municipality. Through a three-arm trial, we will assess home visits through child development agents (intervention 1) as well as center-based parenting groups (intervention 2) relative to a randomly selected control group. Through a steppedwedge rollout approximately one third of communities will receive the intervention in phase I of the rollout; in phase II of the program, two thirds of communities will be selected for interventions. All areas will receive interventions in the last phase of the program.
The specific study objectives are: To evaluate the effect of home-and center-based parenting programs on child survival and development; To assess the impact of both programs on parenting behavior and practices; To evaluate the implementation process; To estimate program cost and cost-effectiveness; To assess relative program, reach among extremely poor and immigrant populations; To assess program spillover on older siblings in the household.

Trial Design
The study was designed as a stepped wedge randomized trial with 3 arms. Arm 1 represents intervention delivered through fortnightly home visits; Arm 2 consists of the same content delivered through fortnightly center-based group meetings; arm 3 is the control group.
To avoid contamination, neighborhoods were used as units of randomization. All 53 neighborhoods in Boa Vista were classified according to the Municipality vulnerability criteria. Only neighborhoods classified as B (medium vulnerability) and C (high vulnerability) will be targeted by the program. As of 2017, 42 neighborhoods were in this category.
Three neighborhoods had either existing home visiting programs or other planned interventions, and were excluded from the study, leaving a final study sample of 39 6 neighborhoods.

From a programmatic perspective, the Survive and Thrive in Brazil -The Boa Vista Early
Childhood Program eventually aims at providing support to all children born in the municipality of Boa Vista from the end of the second trimester of pregnancy until children reach age three. This program was designed to closely align with the national "Criança Feliz" program (Legal Framework of Early Childhood, Law 13.257 / 2016) [23], which will also provide partial funding for this initiative. Within all target areas, all pregnant women as well as women with children under the age of one are eligible for the program as long as they are either poor, were under the age of 20 at the time of conception or were ever exposed to domestic or sexual violence.

Interventions:
Home visit arm (Arm 1) The intervention will be delivered during fortnightly home visits with the presence of the child and at least one of the main caregivers (the program focusses on the mother, but ideally with the father participation or any other family member that routinely spends time with the child). The home visiting curriculum is divided in 3 main modules: 1) pregnancy module: this module is designed to make mothers aware of pregnancy danger signs, to encourage adherence to antenatal care (attendance, performing exams and supplementation), to improve bonding and positive parental practices and to prepare women for breastfeeding. 2) Neonatal module: comprises 3 home visits during the baby's first month of life. One visit during the first week, a second visit at 15 days and the third visit at 28 days. The primary focus of these visits is babies' health, breastfeeding and bonding. 3) child module for ages 02-36 months. Each home visit has 3 or 4 play activities to address child development domains (gross motor, fine motor, language, cognitive development and social emotional development). A recycled materials toy kit is used to support these activities. Visits take in average 45 minutes; after demonstrating and practicing each activity with the caregiver, the visitor leaves the toys, and mother/caregiver is encouraged to play and interact with the child in the two-week interval between the visits. Activities are age-appropriate, and toys and learning materials are exchanged for a new set at each visit. Home visits will be conducted by newly hired and trained child development agents, who will be tasked to complete 60 home visits per month (3 visits per workday), supporting 30 families. 8 Group-meeting arm: Group meetings are designed to deliver essentially the same content. Meetings are held fortnightly at the CRAS center (Social Services Centers) for groups of 8 mothers and their children, other members of the family can participate as well. Groups are formed with participants at a similar stage of gestation or with children of similar age. Group composition remains fixed to increase bonding over time. The only difference between the center and the home-based curriculums is the number of activities per session: due to the larger number of participants, the number of activities is reduced to 2 at centers. The neonatal module is also reduced to 1 or 2 sessions, since we do not expect mothers to attend the sessions during the first 2 weeks of babies' life. Group moderators will be hired by the project and will be tasked to schedule 3 age-groups specific 1 hour and 30 min sessions each day.
Program teams will be based at the 7 regional social service units (CRAS) and will deliver the intervention according to the services coverage area. Group meetings will be held at the centers, where a meeting room will be fully dedicated to the project and are more easily accessed by mothers, since they are distributed within the more vulnerable areas.

Outcomes:
The primary study outcome is child development at age 2 as assessed by the PRIDI scale.
The secondary study outcome is child survival, which will be assessed using data from the municipality's vital registration system (SIM and SINASC). Maternal depression will be assessed using the Center for Epidemiologic Studies Depression Scale (CES-D Scale).
Intention to treat analysis will be used to compare outcomes across the 3 study arms.
Additional per-protocol analysis will be conducted using data on program compliance. Our primary measure of compliance in the home visiting arm will be the number of home visits completed. Our primary compliance measure in the center-based groups will be the number of sessions attended by the mother.

Randomization
Randomization of program rollout timing was based on a simple random number draw in Stata. The number of treated areas in phases 1 and 2 were determined based on logistical feasibility and an initial agreement between the municipality and the project team: in phase one, 9 areas were chosen for home visits and 2 for center-based programs. In phases 2 and 3, program reach was gradually increased to reach all communities as outlined in the Study Design section above. In phase I, 11 neighborhoods will receive the intervention; in phase II, 27; in phase III, all B and C neighborhoods will receive interventions. Figure 3 summarizes the neighborhood selection process and shows the random allocation of interventions by phase. Figure 4 shows the spatial location of areas chosen for each intervention phase. In the first few years of the program, a small sample of women with children under age 1 10 will also be enrolled. This sample will be used to assess the relative impact of reduced exposure to the interventions.

Recruitment
Blinding: Given the nature of the intervention, blinding of participants is not possible.
Endline interviewers will be blind to group assignment.

Sample size and power calculations
Based on the estimated number of births in the targeted B and C areas, we anticipate completing approximately 12,000 baselines (screening interviews) over the project period.
Out of these mother-child dyads, we will randomly select 3000 for the endline assessment.
We assume an average causal effect of 0.5SD on compliant mothers. With an anticipated average compliance rate of 50%, the study is powered to detect an intent-to-treat effect of 0.25SD with power 0.9 between each of the two intervention arms and the control arm, assuming an average sample of 120 households in each of the 11 neighborhoods treated in Phase 1, and an average sample of 60 households in each of the 28 control neighborhoods, assuming an intra-class correlation of 0.02.
In order to assess spillovers within households, we will also assess all older siblings under age 4 in households selected for endline. Based on the high fertility rates observed at baseline, we anticipate a sample of 750 older siblings. Assuming a uniform distribution of siblings across clusters, the study is powered to detect a 0.3SD change in PRIDI score with power 0.8.
In order to assess domain-specific changes in children's development, we will also invite 80 mothers from each arm for a detailed assessment using the Bayley's Scales for Infant and Toddler development. For these assessments, we will randomly select moms from the control group as well as compliant mothers from the two intervention arms. Endline surveys are scheduled to start in March 2020 and will target children at age 2.
With the targeted sample size of 3000 households we anticipate to complete endline by March 2021. BSID-3 assessments will start in July and should be completed within six months.

Data management:
Baseline and endline data will be collected in electronic format on tablets and storied on a secure server. The project leader will be responsible for the anonymization of the data system. A unique identifier will be used in the data system to preserve participant's privacy and confidentiality.

Statistical methods:
For statistical analysis, standard regression models will be used for the continuous PRIDI outcome. Logistic regression models will be used for the secondary mortality outcome. All analysis will be conducted using the Stata 15 statistical software package. Primary analysis will be intent-to-treat. To assess the impact of partial compliance, we will also estimate average treatment effects on the treated (per-protocol analysis).

Ethical clearance:
The study was approved by the Researcher's institution IRB under protocol number 12 CAAE:73722917.4.0000.0076 in August 2017. Each participant signs the consent form at enrollment, during the baseline interview visit. Should the participant be a minor, the consent form is signed by the responsible adult.
Trial Status: As of November 2019, trial enrollment is open and is scheduled to continue until July 2020. The present protocol is in its first version, since November 2017.

Discussion
A large body of evidence has highlighted the importance of early parenting and nurturing care on child development. However, equitable early childhood programs and public policies are crucial, for ensuring continued progress in Brazil's as well as international efforts [11,[29][30][31]. In 2016, our group adapted and evaluated a home-based early childhood stimulation program with positive impact on child development. The proposed project aims to build on that evidence, creating a program that addresses not only child development, but child survival as well. We will evaluate the feasibility, impact and costeffectiveness of this program at the municipality level. The proposed project will tackle the two most salient problems for children under 5 in Brazil: the continued high rates of neonatal mortality, and the large disparities in early childhood development. Children growing up in poor urban areas of Brazil continue to be exposed to a substantial amount of adversity in early childhood due to exposure to pollutants, external and domestic violence, unstable family environment, maternal depression and inadequate learning opportunities. Early disadvantages appear to be particularly large in illegal slum settlements characterized by poor hygiene, high levels of environmental pollution, high levels of community violence and in many cases also social isolation, often resulting in high incidence and prevalence of maternal depression. Our aim is to use this transition to 13 scale project to provide information to the Brazilian government, contributing to these new national efforts. If successful, the tested interventions can potentially be used at the regional level and also nationwide and can ideally help Brazil to further accelerate its positive trends in neonatal survival and to help create healthy early life environment for all children in the country.