Optimal infant and young child feeding practices are largely inadequate in rural and poor households in Ghana. Child welfare clinics (CWCs) focus on immunization, supplementation, and growth monitoring and promotion (GMP) activities among infants and young children (IYC). An essential component of GMP activities is to counsel mothers to practice optimal IYC feeding and health practices, but most of the time this is completely missing or not specific enough to be effective.

This study will test the effect of a mass media nutrition education program on caregiver infant and young child feeding knowledge, attitudes and practices using a community-based cluster randomized design. Formative research will first be undertaken to determine the beliefs, attitudes, and constraints that prevent caregivers from adequately feeding their infants and young children with members of the community (caregivers with children under-five, men and elderly women).The information gathered will then be used to design specific messages that directly address the IYC feeding challenges of the community. Clusters will be randomized into active (radio messages with in-person engagement and follow-up) and passive (only radio messages) arms of the intervention. Farm Radio International (implementers of the mass media programming) will ensure the regular broadcast of twice weekly messages for a period of about 12-18 months.

This study will be a community-based cluster randomized controlled trial. All communities in the Upper Manya Krobo District (UMKD) that are accessible 12 months of the year (with the exception of 19 communities that are participating in an agriculture-education intervention) will be considered eligible. These communities (n =168) will be stratified based on sub-district and the median community population size for the sub-district. Four communities will be randomly selected from each sub-district: two large and two small communities. These 24 communities will then be randomly assigned to active or passive radio engagement.

Each community will be considered a cluster. To achieve a total sample of 1258 caregivers, 52 women per cluster will be enrolled. If researchers are unable to enroll the sample size in one community, the cluster will be expanded to include the closest village and enrolment will continue until the sample size is achieved.

Enrolment strategy. A census of all communities was completed in 2013 and will be used to decide on the selection strategy. Field workers a priori will randomly select the 52 households in each cluster (only one mother per household, randomly chosen if there is > 1). If the selected household does not have a child < 5 y of age, the closest house will be approached. This will continue until enrolment is completed.

Data Collection Procedures

Phase 1: Baseline survey

A baseline survey will be carried out to document the prevailing IYC feeding knowledge, attitudes and practices of caregivers using a semi-structured questionnaire. Questions will cover:

• socio-demographic information
• radio ownership and listening practices
• exposure to other sources of nutrition education in the district
• IYC feeding knowledge and practices
• knowledge and awareness about anaemia
• knowledge and awareness about undernutrition
• knowledge, attitudes and practices about water, sanitation, and hygiene

Phase 2: Formative research

The formative research will begin in only the active engagement communities.

The total sample for the formative research will be no more than 144 women with children under 5 years of age 144 men with children under 5 years of age, and 144 elderly women who take care of children under 5 years of age. Participation will be solicited at a community-wide meeting to explain the project. Interested individuals will be included until the sample size is reached. Written informed consent will be obtained as noted below.

Participation in the formative research will include:

(i) Farm Radio International (implementers of the mass media programming) will carry out the focus groups to determine the beliefs, attitudes, and constraints that prevent caregivers from adequately feeding their infants and young children. In addition, information on radio-listening habits, time preference for programmes, and formats that are of interest to the communities will be collected. Information gathered will be used to design radio nutrition and health messages and jingles that directly address the IYC feeding challenges in the area. Three focus group discussions will be undertaken in each of the 12 active radio engagement communities. The focus groups will include 8-12 individuals each, for the following groups: i) women with children under 5 years of age ii) men with children under 5 years of age iii) elderly women who take care of children under 5 years of age (grandmothers, great aunts)

(ii) These same participants from the focus groups will be followed up and will take part in monthly and quarterly assessments of the radio programming using semi-structured questionnaires. The aims of these assessments are:

i) obtain feedback from community members to improve content the radio messages, ii) determine whether the messages are being received as intended, and iii) assess intermediary adoption of messages

Phase 3: Intervention

The interactive radio education intervention will involve weekly radio programming with local radio stations on key nutrition and health topics for up to 18 months. The topics will be consistent with the Ministry of Health's community nutrition and health education messages. The programme will be repeated once every week. The intensity of the intervention for an individual will differ based on the assignment of the participant's community:

See description of the active and passive arms.

Phase 4: Post-intervention survey

An endline survey will be carried out within a maximum of 3 months after the end of the mass media intervention, to determine the effect of the intervention on IYC feeding and sanitation knowledge, attitudes and practices of caregivers. The questions will cover:

• recollection of key messages (free recall-no probes)
• self-reported level of engagement with radio programming
• exposure to other sources of nutrition education in the district
• IYC feeding knowledge and practices
• knowledge and awareness about anaemia
• knowledge and awareness about undernutrition
• knowledge, attitudes and practices about water, sanitation and hygiene The endline survey will be completed with all 1258 women who participated in the baseline survey.

Data Analysis Focus Group Data All recordings of the focus group discussions will transcribed from the local language (Krobo and Twi) to English and analyzed using N-Vivo version10. The thematic content analysis approach will be used to analyze the data. Common themes emerging from the discussions will be grouped and the results summarized based on these themes. Selected quotes of participants will also be presented.

Change from baseline in self-reported knowledge, attitudes and practices. All analysis will be by intention-to-treat basis. Analyses will be performed on individual level summaries using a difference-in-difference (DiD) approach. For each target knowledge, attitude and behavior, the reported prevalence will be estimated at baseline and endline, and the difference between surveys calculated. ANOVA tests will be used to detect differences in caregiver infant and young child feeding knowledge, attitudes and practices between intervention group and control group while adjusting for important baseline characteristics and the effect of cluster.

Inclusion Criteria:

• lives in cluster randomly assigned to either active or passive arm
• has child < 5 y of age
• is the primary caregiver of the child

Exclusion Criteria:

• has physical condition/illness that prevents from taking care of child
• has physical condition/illness that prevents from responding to surveys

The study design and communities' status as active or passive are not discussed with the field staff.