Study design and setting
A prospective study on infant feeding practices was carried out among mother-child pairs in Kinshasa from October 2012 to July 2013. Mothers were recruited during the first week after childbirth and followed for six months.
Sampling procedure
Kinshasa has six health districts, two (Ndjili and Kalamu) were randomly selected for the study. The list of all maternity facilities located in these two selected health districts and the monthly mean number of births occurred in each maternity facility were obtained from the National Reproductive Health Program. From this list, 12 eligible maternity facilities were randomly selected. To be eligible for the study, at least 60 births should occur in the maternity facility per month. In the selected maternity facility, all eligible women who gave birth from October 2012 to December 2012 were recruited. The mother was considered to be eligible when she was at least 18 years old, had antenatal care visits in the same maternity facility and gave birth to a single living full-term child who was free of any serious health conditions that would require transfer to an intensive care unit.
Data collection
To collect data, six trained surveyors led face-to-face interviews at seven times: during the first week after birth, and at 1, 2, 3, 4, 5, and 6 months. Those interviews were conducted at the mother’s home to allow her to feel free while talking about what she had experienced in the health facility and how she was feeding her infant.
Twice a week, surveyors identified eligible mothers in the registers of the selected maternity facilities. Then, mothers were contacted at home during the first week after birth; they were informed about the study and its purpose. Those agreeing to participate gave written consent. During this first visit, the mother’s socioeconomic, psychosocial and medical data were collected. She was also asked about baby-friendly practices experienced during the stay in the maternity and difficulties related to breastfeeding encountered. During interviews led at 1, 2, 3, 4, 5 and 6 months, mothers were asked about how their infants were fed the previous day. If any food or drink other than breast milk was mentioned, the mother was asked about when she started to feed her infant with the food or drink in question. The study instrument was translated in Lingala and back in French, the original language.
Definition of variables
The breastfeeding terms used in this study were those recommended by the WHO. An infant was considered to be exclusively breastfed when he or she had received only breast milk with no other liquids (including water) or solids [22]. The duration of EBF was determined using information about the age at which other types of milks, liquids and/or complementary foods were introduced. This duration was initially measured in days, and then converted in weeks.
The mother’s intention to breastfeed exclusively was defined as the planned length of EBF. The Breastfeeding Self-Efficacy Short Form (BSE-SF) was used to assess the mother’s confidence in her ability to breastfeed [23], while the attitude toward breastfeeding was evaluated with the lowa Infant Feeding Attitude Scale (IFAS) [24]. A cross-cultural adaptation was made from the original scales; items that were not culturally adapted to Kinshasa population were removed. Thus, 10 items from each original scale were used in this study. Each item was measured on a 5-point scale, and the total score ranged from 5 to 50. For the purpose of bivariate and multivariate analysis, confidence in the ability to breastfeed was split into three groups: not confident (less than 35), fairly confident (from 35 to 44) and very confident (at least 45). For the same purpose, the attitude toward breastfeeding was also split into three groups: negative attitude (less than 35), fairly positive attitude (from 35 to 44) and very positive attitude (at least 45). In absence of a validated standardized questionnaire, the mother’s knowledge on breastfeeding was determined with a pretested questionnaire adapted from previous studies [25, 26]. This questionnaire included ten questions related to the following aspects: timing of breastfeeding initiation, importance of colostrum, average number of feeds a child should receive per day, duration of EBF and mother’s understanding of the usefulness of breastfeeding. Each question was given a rating of 0 or 1 depending on whether the answer was wrong or correct. Therefore, the total score ranged from 0 to 10. For the purpose mentioned above, the knowledge on breastfeeding was split into three groups: low level of knowledge (less than 5), fairly good level of knowledge (from 5 to 6) and very good level of knowledge (at least 7).
Ten questions were used to assess baby-friendly practices experienced by mothers during the stay in the maternity (see Additional file 1). Thus, this variable ranged from 0 to 10. These questions covered six of the ten steps to successful breastfeeding namely: informing all pregnant women about the benefits and management of breastfeeding (step 3), helping mothers to initiate breastfeeding within a half-hour of birth (step 4), giving newborn infants no food or drink other than breast milk, unless medically indicated (step 6), allowing mothers and infants to remain together 24 h a day (step 7), encouraging breastfeeding on demand (step 8) and giving no artificial teats or pacifiers to breastfed infants (step 9).
The family income was defined as the average amount spent by the family, per day and per individual, for foods. The family was considered to have a low income when this amount was less than one United States Dollars (1 USD).
The questionnaire used to collect data was assessed for reliability using the test-retest method. Twenty breastfeeding mothers were interviewed, and then re-interviewed after seven days. The correlation coefficient ranged from 0.84 to 0.96.
Data analysis
Information collected was entered in EpiData version 3.1 and analyzed with Stata version 12. Characteristics of study participants were summarized using the median and interquartile range (IQR) for continuous variables and proportions for categorical variables namely: occupation, education, marital status and parity.
Survival analysis was used to examine the duration of EBF. For this analysis, the duration of EBF was right censored at the age of the child at the moment the mother-infant pair had dropped out of the study and at six months for mothers who continued to exclusively breastfeed at the end of the study. The child death was also considered as a censoring event. The Kaplan Meier Method was used to determine the median duration of EBF. The Cox Proportional Model helped to identify predictors of discontinuing EBF before six months. Crude hazard ratios (CHR) and adjusted hazard ratios (AHR) had been calculated in bivariate and multivariate analysis respectively. The following variables were included in the model, which was built using the stepwise approach: age, education level, marital status, occupation, family income, parity, number of antenatal care visits reached, intention to breastfeed exclusively, confidence in the ability to breastfeed, attitude toward breastfeeding, knowledge on breastfeeding, breastfeeding problems during the first week and number of BFHI practices experienced. The final model was built with six variables namely: number of antenatal care visits reached, intention to breastfeed exclusively, confidence in the ability to breastfeed, knowledge on breastfeeding, breastfeeding problems during the first week and number of BFHI practices experienced. The compliance of the model with the assumption of proportionality was assessed using the plot based on Schoenfeld residues. Statistical significance was set at p < 0.05.
Ethical consideration
The study was approved by the Ethical Committee of the Kinshasa School of Public Health. At the time of recruitment, all mothers were informed about the study and provided a signed informed consent. Participation was entirely voluntary and mothers had the liberty of withdrawing from the study at any time.