Building on the success of community-based peer education and counselling in breastfeeding, it is possible to address young child feeding and health practices in the surveyed district. Opportunities that can optimise success and encourage behaviour change in the district include mothers willingness to modify some feeding options; support of family members; seasonal availability and accessibility of foods; established set-up of village peers and existence of health centres and health staff in some villages. This study has developed a feasible, context-specific nutrition education package for use in rural Tanzania. The package is composed of 4 components, namely 1) education and counselling of mothers, 2) training of community-based nutrition counsellors, 3) sensitisation meetings with health staff and family members, and 4) supervision of community-based nutrition counsellors. The intervention will use a parallel cluster-randomised controlled trial design where infants will be recruited when aged 6 months and followed up for 9 months. The intervention expects to provide information and, where appropriate, recommendations to strengthen the nutrition component in the health education programme of the Tanzania child health services.

The study hypothesises that the nutrition education package will be more effective than the routine health education in improving feeding practices, dietary adequacy and growth of infants and young children. Specific objectives of the study include:

• To evaluate the effectiveness of a nutrition education package on feeding practices and dietary adequacy
• To determine the effectiveness of a nutrition education package in improving growth and health of infants and young children
• To document the process of implementation of the nutrition education package to promote optimal feeding and health practice

Design:

A parallel cluster randomised controlled trial will be conducted in rural central Tanzania in 9 intervention and 9 control villages. A simple randomisation with a 1:1 allocation using a list of random numbers generated in Microsoft Excel 2007 will be used to randomise villages to either control or intervention group. Generation of allocation sequence and randomisation of clusters will be done by a statistician blinded to study groups and not participating in the research.

Recruitment and collection of baseline information:

After cluster randomisation, sub-village leaders will identify all infants aged 6 months and their parents in a systematic door-to-door survey. Parents will be invited to a meeting where the nature and purpose of the trial and eligibility criteria will be explained. Thereafter, their infants will be screened for eligibility.

Parents and infants who meet the inclusion criteria, agree to participate in the trial and give a written informed consent will have their infants recruited. Baseline information of recruited infants and their parents (i.e. household's, parents' and infant's characteristics) will be collected using a structured questionnaire. Structured interviews with village and sub-village leaders will also be conducted to collect village information (i.e. population, health facility, water supply, schools, livelihoods, resources, economic opportunities).

Intervention:

The control group will receive routine health education offered monthly by health staff at health facilities. Routine health education is a standard government health service for children below the age of five years. It is offered to mothers during growth monitoring and immunisation contacts. The sessions usually take 10-15 minutes and focus on general health issues including child feeding, prevention of diseases such as malaria and importance of immunisations. The intervention group will receive a nutrition education package in addition to the routine health education. With the nutrition education package, mothers and caregivers will receive 3 sessions of education and counselling on optimal infant and young child feeding and health, when a child is aged 6, 9 and 12 months. Nutrition counsellors will receive training at baseline and the session will be repeated after 6 months. After training, the nutrition counsellors will conduct monthly home visits to counsel and support mothers and their families. Supervisory visits will be conducted by the research team after every 2 months to assess counsellors' work. Separate sensitisation meetings with families and health staff responsible for child health will be conducted at baseline and meeting will be repeated after 6 months.

Data collection:

Data collection forms (structured questionnaires, observation checklists, etc.) have been pre-tested in a neighbouring district and standardised. Interviewers will administer the surveys in national language, Kiswahili. After data collection, all filled forms will be manually checked for completeness and consistency. To enhance blinding, precise objectives of the study and village allocation to trial will not be disclosed to data collectors; nutrition counsellors will not be responsible for data collection; and data collection schedule will be randomised.

Impact evaluation data: In the intervention group, data on primary and secondary outcomes will be collected at baseline and ages 9, 12 and 15 months. In the control group, the same variables will be collected at baseline, and ages 12 and 15 months.

Process evaluation data: A process evaluation will document the intervention implementation process so as to 1) assess whether the intervention activities are implemented as planned (i.e. fidelity), 2) evaluate the extent to which the intervention reaches the intended mothers and their families (i.e. reach), 3) determine the degree to which targeted mothers are exposed to intervention components and extent to which they use intervention resources (i.e. exposure or dose received), and 4) describe the setting (i.e. contextual factors, facilitators, barriers, contamination) into which the intervention is being implemented that may have an influence on intervention effectiveness. In control villages, process evaluation will identify and describe actions, events and context which may reveal new interventions, evidence of contamination or other factors external to the intervention. Process evaluation data will be collected two months after baseline (when infants are 8 months-old) and at the end of intervention (when infants are 15 months-old). Structured and semi-structured interviews, review of records (e.g. training and education attendance sheet, activity logs, training test scores, etc) and structured observations will be carried out with mothers, nutrition counsellors, health workers and village leaders.

Monitoring:

The research team will visit intervention villages every 2 weeks in the first 2 months of intervention, and thereafter visits will be made at 2-months interval. Researchers will log and report all activities related to the delivery and quality of training, education, supervision and sensitisation meetings using monitoring tools (e.g. forms, logs, observation checklists, supervisory reports). Nutrition counsellors will document home visits in workbooks. Actions or events external to the intervention which happen in the intervention villages will be documented in event forms and semi-structured interviews.

Data analysis:

Data will be entered in Epi-data version 3.1 and consistency analysed by range checks of data values. Data will be analysed using STATA release 12.0 (STATA Corporation, Texas, 2007). Frequency distributions will be run to identify outliers. Impact evaluation data will be compared between intervention and control villages using hierarchical or multi-level models and presented at the cluster and individual levels. The methods will provide adjustment for potential covariates and confounders at the cluster and individual levels. Data will be analysed by intention to treat. The process evaluation data will use both inductive and deductive approaches to analyse qualitative data using NVivo 10 for Windows (QSR International, Australia).

Inclusion Criteria:

• Infants aged 6-7 months at the time of recruitment into the study
• Infants currently being breastfed
• Parents (or caregivers) anticipated local residence for the study duration

Exclusion Criteria:

• Congenital or chronic abnormalities impairing feeding or physical growth measurements
• Oedema
• Severely ill or clinical complications warranting hospitalisation