The purpose of this study is to evaluate the impact of a community-based early childhood development (ECD) program on children's physical and cognitive development. Under the program, targeted communities will be assigned a trained Child Development Agent (CDA) who will have four main tasks and responsibilities: 1) biweekly screening and management (including referral) of acute malnutrition in children; 2) encouragement of caregivers to utilize routine care services for children; 3) screening for symptoms of acute diseases including malaria, diarrhea, and pneumonia and referral for diagnosis and treatment; and 4) organization and mentoring of biweekly caregiver meetings to discuss parenting and promote early childhood cognitive stimulation. The investigators will enroll at baseline around 600 children ages 6 - 12 months and their caregivers, and randomize them at the community-level to receive the ECD program or to remain in the control group. The study period will be one year. At end line, the investigators will collect important indicators of child physical and cognitive development to assess program impact. If the program shows both feasibility and impact, there is the potential to integrate program interventions into existing national community-based health initiatives.

Amendment: the study period has been extended for a second year. After a five month gap when no intervention was provided, biweekly (i.e., fortnightly) community-based parenting groups were restarted in intervention clusters. In the second year of the intervention, CDAs no longer visit households.

The purpose of this study is to evaluate the impact of a community-based early childhood development (ECD) program on child physical and cognitive development. Central to the ECD program is a cadre of community-based health workers called Child Development Agents (CDAs). CDAs have four main tasks and responsibilities: 1) Biweekly screening and management (including referral) of acute malnutrition in children; 2) Encouragement of caregivers to utilize routine care services for children (immunizations, growth monitoring, and vitamin A supplementation); 3) Screening for symptoms of acute diseases including malaria, diarrhea, and pneumonia and referral for diagnosis and treatment; and 4) Organization and mentoring of weekly meetings for caregivers to discuss parenting issues and promote early childhood cognitive stimulation.

Protocols for the community-based evaluation and management of children with acute malnutrition are based on the draft Zambian Community Integrated Management of Acute Malnutrition (CIMAM) guidelines. Children with complicated severe acute malnutrition (SAM) are referred to the nearest health center for evaluation and, if SAM is confirmed based on mid-upper arm circumference (MUAC), then referred to an inpatient feeding center for management. All children with moderate acute malnutrition (MAM) and uncomplicated SAM are administered an appetite test at the health center. Children who pass the appetite test are managed locally as outpatients. Children with acute infectious diseases are treated following the Ministry of Health's adaptation of the Integrated Management of Childhood Illness (IMCI) guidelines and those with complicated SAM or feeding difficulties are referred to higher level facilities (feeding centers). As part of the ECD program, clinical staff members at study health centers receive a short training to improve their adherence to IMCI and CIMAM guidelines.

One aspect of the IECDZ program that is of particular interest from a policy and evaluation perspective is the explicit focus on improving communication between CDAs and clinical staff at facilities handling undernourished children. As part of the ECD program, CDAs meet with staff at referral health centers on a regular basis, and discuss strategies to improve nutritional outcomes of children falling behind.

In addition to these regular meetings, the ECD program strengthens the linkages between CDAs and health centers with a novel mHealth intervention. When a CDA refers a child to a health center for evaluation of acute malnutrition, danger signs, or signs/symptoms suggestive of malaria, pneumonia or diarrhea, the officer-in-charge at the health center receives an short message service (SMS) and a paper referral form to make them aware of the child and their provisional diagnosis. Once the referred child has been attended to at the facility, health center staff members will then send an SMS back to the CDA to make them aware of the final diagnosis and management plan. If the health center refers the child to a feeding center, an SMS is sent to the feeding center along with a paper referral form. All text messages within the proposed system are free of charge to the end users. The mobile health (mHealth) system is meant to improve communication between health workers operating at various levels of the health system. Caregivers themselves do not receive SMS updates. Rather, CDAs will communicate all updates on the health and nutritional status of children directly to caregivers.

The primary objectives of the ECD program are to:

1. Improve community-based screening for uncomplicated acute malnutrition and to strengthen referral to and the quality of management of malnourished children at local health centers;
2. Conduct community-based screening and referral for management of common childhood diseases (malaria, diarrhea, and pneumonia);
3. Strengthen the continuum of care for children with acute malnutrition and disease using an innovative mHealth intervention to link CDAs, the children that they are screening, and health centers and feeding centers; and
4. Improve early cognitive development through community-based weekly meetings with caregivers.

The proposed evaluation will be structured around the ECD program roll out, and will employ a cluster-randomized trial design. For the purposes of the evaluation, a cluster is demarcated by the boundaries of a health zone, a geographic unit defined and used by the Zambia Ministry of Health to determine catchment populations for public health facilities. Each health zone and the children recruited from within the zone comprise a cluster for the trial. The ECD interventions will be implemented in 15 randomly selected health zones across five facilities starting in August 2014. The remaining 15 health zones in the catchments of the five study facilities will continue using current practices and will serve as controls. Within each health zone, a total of around 20 child/caregiver pairs will be recruited and enrolled in the evaluation. In total, the investigators anticipate around 300 child/caregiver pairs being enrolled in each arm of the study, for a total of 600 child/caregiver pairs.

The study sample will be randomized at the level of the health zone to either the ECD program or to the control group using stratified random sampling methods. Prior to randomization, basic demographic information will be collected on all health zones, including: distance to the study health center, total population, and number of eligible children ages 6 to 12 months. Next, zones will be stratified according to these demographic data. Finally, within strata zones will be assigned to either the ECD intervention or to the control group using a random number generator to achieve balance within strata.

Study data will be collected from four main sources: 1) a baseline interview with study caregivers; 2) an endline interview with caregivers and a child assessment; 3) IECDZ program monitoring systems, including CDA records and SMS software; and 4) semi-structured interviews with CDAs and clinic staff.

Baseline caregiver interview - Study CDAs will conduct a baseline survey of both control and intervention households which will include: a) basic demographic and socioeconomic status, b) availability of books, toys and other reading material at home, time spent by caregivers reading to the child, as well as time spent by caregivers playing with the child, c) anthropometric screening (weight, height, and MUAC), d) review of vaccinations and child health visits using child health card, e) assessment of maternal or caregiver mental well-being, f) assessment of maternal depression with the self-reporting questionnaire-20 (SRQ-20), and g) presence of other early childhood development programs in the community.

End line caregiver interview and child assessment - Twelve months after enrollment all children will be visited for a final interview and assessment. This assessment will include the questions asked at baseline as well as an assessment of child development using a standardized test-the INTERGROWTH 21st century assessment tool.

ECD program monitoring systems - Data on key process indicators will be collected from ECD program records to assess the activities of CDAs and study clinics. Important indicators include: a) volume of CDA household visits; b) volume of children identified and referred for malnutrition and/or disease; c) volume of successful referrals; and d) volume of SMS sent between CDAs and study clinics. Additional ECD program monitoring systems dedicated to the caregiver groups' cognitive stimulation are: a) frequency of caregiver groups, b) caregiver knowledge of topics covered by the curriculum, and c) measures of caregiver-child interaction quality. These data do not exist for control areas, so they will not be used to evaluate the impact of the ECD program. However, these data will provide important information that will be used to better understand any program impacts that are identified.

Interviews with CDAs and clinic staff - A series of semi-structured interviews will be conducted with CDAs and clinic staff members to assess both awareness of child nutrition and disease issues and general perceptions of the ECD program. These interviews will be conducted with all study CDAs and staff from all study facilities at three time points.

For analyses, the primary independent variable of interest is receipt of the ECD program. Assuming proper randomization, we anticipate that all factors that might potentially confound the relationships of interest are uncorrelated with receipt of the program. We will compare baseline characteristics between the two groups to assess balance. We will collect data on potential confounders to increase precision, analyze heterogeneity and, if necessary, control for any potential imbalance. Potential confounders include: maternal and paternal education, assets, income and employment, family composition and child gender. We will also control for baseline height-for-age z-score and other baseline child health measures.

Categorical variables will be compared between groups using a chi-square test, or Fisher's exact test if cell sizes are small; continuous variables using a t-test if normally distributed or non-parametric Wilcoxon rank sum tests, if non-normal. We will model our impact measures as a function of treatment and controls using a standard regression framework.

Amendment: the study period has been extended for a second year.

Inclusion criteria

• Child ages 6 to 12 months at the time of intervention launch residing in the catchment areas of the five study health facilities
• Child's mother or main caregiver must be 15 years or older
• Child's primary caregiver must be a female (because the participants in the caregiver groups may feel uncomfortable discussing certain issues if a man is present)

Exclusion criteria

• Caregivers who are unwilling to provide informed consent
• Families that plan to move from their health center catchment zone during the twelve month period of the study