The evaluation employed a quasi-experimental design to test the hypothesis that health and nutrition education and promotion of use of antenatal care (ANC) services in the first 1,000 days of life will reduce stunting in children under five years of age. Repeated data collection surveys were done at baseline in June - September 2016, and endline in July- September 2019 in both control and interventions sites.
The study was conducted in five regions of Tanzania mainland. Simiyu and Ruvuma regions were purposively selected as interventions sites by the implementers as they had a high prevalence of stunting, 33 and 44% respectively(6). Simiyu is located in the Northern part of Tanzania and the South East of Lake Victoria while the Ruvuma is located in the Southern part of Tanzania. Four districts matched on health services availability, population, nutritional and mortality indicators were purposeful selected as controls. Uyui and Nzega districts located in Tabora region were paired with Simiyu region while Rufiji and Ruangwa districts in Coast and Lindi region respectively were paired with Ruvuma region. Districts with ongoing or planned similar nutrition program at the beginning of program activities were excluded from the controls sampling frame.
Sample size and sampling
The sample size was calculated to detect a 10% percent absolute baseline-to-end-line change in stunting as a key indicator, using a baseline rate of stunting in the intervention and control districts of 40% and 26 % respectively. Assuming a 15% and 5% absolute drop in stunting in intervention and control districts respectively over the 4-year period, a 5% type I error and 80% power, and an intervention to control ratio of 1:2, a total of 840 households in intervention districts and 1,680 in control districts was dimmed sufficiently after accounting for the 10% non-response. A two-stage cluster sampling process was used, with the first stage involving sampling of 56 villages, 28 from each site proportional to the district population size. This was followed by random sampling of 30 and 60 households from each village located in the intervention and control site respectively. All households with children under two years in the selected villages form part of a sampling frame for the second stage sampling.
Study Participants and Eligibility
Mothers who delivered within the 24 months preceding the survey and all children under-five years residing in the selected households were eligible. The 24 months were selected because child nutrition deteriorates in the period 6-24 months and after that age, the growth stagnates(11). Infant feeding information was collected for all children under 24 months, as longer duration will introduce feeding pattern misclassification due to recall bias (12). In household with two or more women with children less than 24 months of age, the one with the youngest child was interviewed to reduce recall bias.
Anthropometric measurements (weight and height) were taken from all under-five children who slept in the selected household a night preceding the survey and also to the blood parents of the children under-fives. Weight was assessed using the calibrated United Nations Children Funds (UNICEF) Electronic Scales(13). The scale was placed on a flat surface, and children aged 2 years and above weighed while standing on a scale while those aged less than 2 years had their weight taken as a difference of weight of mother or caretakers with the child combined to that of the mothers alone. Weight was recorded to the nearest 0.1kg. Height was measured using a height board. Children less than two years were measured lying on the length board while those aged two years and above were measured standing upright on the board. Height was recorded to the nearest 0.1cm. Measurement procedures adhered to standardized protocols for anthropometry used in the construction of the international growth reference (13).
A modified version of the validated Demographic and Health Survey (DHS) questionnaire was used during the interviews(6). The tools captured i) mother’s demographic and reproductive history; ii) alcohol consumption and cigarette smoking history iii) history of health service utilization during ANC and the postnatal period, including Iron and Folic Acid supplementation, presumptive treatment for malaria and use of insecticide-treated bed nets during pregnancy and the postnatal period; iv) sources of breastfeeding information and a series of questions testing maternal knowledge on breastfeeding, complementary feeding and healthy eating during pregnancy and lactation; and v) infant feeding practices.
Data Management and Analysis:
Data was analyzed using STATA version 13 software. Descriptive statistics were done to establish the markers of household-level characteristics, health and nutrition indicators at each round of data collection. The anthropometry Z-scores were calculated using the WHO 2006 growth references(14). The impact of a program was calculated using the difference in difference estimates with a p-value less than 0.05 at 95% level considered as significant. Principal components analysis was used to create a wealth index for each household based on the asset’s availability at household level.
Quality Assessment and Control
All study tools were reviewed by relevant experts prior to data collection. Questionnaires were translated to Kiswahili and back-translated to English to ensure meaning was maintained. Field workers were trained for four days prior to each round of data collection to reduce within and between inter and intra-data variability. The training was followed by a day of field-testing of data tools. Weight and height/length measurements were standardized using FANTA guidelines (15). All weighing scales were calibrated using an object of known weight prior to use. Data collection was done using electronic devices and uploaded to a central data server on daily basis allowing instant review and feedback to the field team.
Ethical clearance from institutional (IHI/IRB/No:09-2016) and national (NIMR/HQ/R.8a/Vol.IX /2208) Research Ethics Committee were obtained. The study team pay courtesy to regional and district authority prior to beginning data collection activities. Permission letters were obtained and presented to village leaders during data collection activities. Before interviews, RA’s informed all study participants about the study objectives, risk and benefit of participating and study procedures. Only respondent who provided written consent were interviewed.