Study population and setting
Data for this study were collected in the context of an evaluation of a community-based infant and young child nutrition program, called Akhoni Shomay. The program was implemented starting in May 2011 in Karimganj, a rural sub-district of Kishoreganj with a population of ~ 320,000, approximately 120 km north of Dhaka. To evaluate the program, a prospective cohort of 2400 pregnant women was recruited, consisting of 1200 women from Karimganj and 1200 from a neighboring sub-district (Katiadi). Akhoni Shomay promoted optimal IYCF practices through individual counselling for mothers, as well as group counselling for other key influencers of IYCF practices, such as fathers and grandmothers. The program also encouraged home-based fortification of complementary foods using micronutrient powders.
Women were recruited during their 7th month of gestation, in three waves: January and February 2011, May and June 2011, and September and October 2011. The recruitment waves were designed to coincide with just prior to program launch, immediately after program launch and several months after program launch (to allow for adequate program dissemination), respectively. Follow-up of mother-child dyads occurred at 3, 9, 16 and 24 months of child’s age, using back-translated, pilot-tested questionnaires. Detailed descriptions of the intervention program, study setting, population and data collection tools are also available elsewhere [19,20,21,22,23].
Variable derivation
Complementary feeding knowledge
Complementary feeding knowledge was assessed by using a 19-item instrument administered at the 3-month follow-up (see Additional file 1). Questions were derived from the WHO guiding principles for complementary feeding [24], and assessed maternal knowledge of recommended age of complementary feeding initiation, techniques for responsive feeding, and types of foods and how to prepare them. A scale of complementary feeding knowledge was created from the sum of correct responses. However, if a mother identified force-feeding an infant as acceptable, 1 point was deducted from her knowledge score. The scores were then categorized into tertiles for analysis, with a higher score reflecting better knowledge of optimal IYCF practices.
Attitudes
Attitudes were assessed using a 10-item instrument, also administered at the 3-month follow-up, and addressed nutritional importance and cost of complementary foods and nutritional supplements, and ease of continued breastfeeding for the mother. Respondents answered using a 5-point scale ranging from ‘1 = Strongly Agree’ to ‘5 = Strongly Disagree’. Favorable attitudes were reverse-coded before analysis. Factor analysis using principal-axis factoring method and an orthogonal varimax rotation extracted three factors. Eight of the 10 items had factor loadings of > 0.5. An overall attitude scale was created from the simple sum of scores for these eight items (theoretical range: 8–40). The scores were then categorized into tertiles for analysis, with a higher score reflecting more favorable attitudes towards complementary feeding.
Infant age at complementary feeding initiation
Infant age (in months) at complementary feeding initiation was estimated using maternal recall at 9-month follow-up (see Additional file 2). Complementary feeding was considered early if reported age was ≤4 months, timely at age 5–6 months, and late if reported age was ≥7 months.
Other covariates
Information on household characteristics and ownership of assets, as well as maternal age, literacy and parity was collected at enrollment into the study at 28–32 weeks’ gestation.
Statistical analysis
Data were imported into Statistical Analysis Software (SAS), version 9.3 for analysis (see Additional file 3). For categorical variables, frequencies and percentages were calculated; for continuous variables, median (range) was calculated. Household socioeconomic status was assessed via characteristics of the respondents’ dwelling and ownership of assets. An asset based socioeconomic status score was created using methods described by Filmer and Pritchett [25].
A 3-level categorical variable was created for infant age at complementary feeding initiation, with timely initiation as referent. Tertiles were created for maternal knowledge and attitude scores for analysis at 0–7, 8–9, and 10–15 for the knowledge score and at 18–25, 26, and 27–34 for the attitudes score.
The study population was stratified into four groups for analysis, based on district of residence and wave of enrollment. All participants from Katiadi, the control district, were included in one group, while participants from Karimganj, the intervention district, were divided into three groups, based on their timing of enrollment into the study and hence potential for exposure to program messaging.
Polytomous logistic regression, stratified by district of residence and wave of enrollment, and adjusted for socioeconomic status, infant gender, maternal age, literacy, and parity, was used to determine the association between maternal knowledge and attitudes and timing of complementary feeding initiation. Using a polytomous regression model, instead of ordinal logistic regression, allows for comparisons of ‘early’ or ‘late’ versus ‘timely’ initiation of complementary feeding, separately. The corresponding odds ratios and their 95% confidence intervals are interpreted using ‘timely’ complementary feeding initiation as referent.
Ethics
The Research Review Committee (RRC) and the Ethical Review Committee (ERC) of the International Center for Diarrheal Disease Research, Bangladesh (icddr,b) approved the study. Written informed consent was obtained from each woman at the time of enrollment into the study. At each follow-up visit, a description of the information to be collected at that point was provided and verbal consent was obtained.