Limitations of maternal recall for measuring exclusive breastfeeding rates in South African mothers

Background: Maternal recall is most commonly used to determine exclusive breastfeeding rates. A gold standard stable isotope method is available which can determine intake of breast milk versus water from sources other than breast milk and thus objectively determine exclusive breastfeeding. The objectives of this study were to determine exclusive breastfeeding rates by both maternal recall and the objective stable isotope method and discuss the limitations and usefulness of the two methods. Methods: The study involved 100 mother-infant pairs in a peri-urban area in Durban, South Africa and study visits took place from July 2012 to September 2014. Maternal recall of exclusive breastfeeding was carried out using the World Health Organization’s 24 hour recall of infant feeding and this was compared to the objective measurement of exclusive breastfeeding using the stable isotope technique at three time points: six weeks, three and 5.5 months. The objective measurements were carried out using two different cut off values for exclusive breastfeeding. Kappa analysis was used to quantify the relationship between maternal recall and results from the stable isotope technique for each mother-infant pair. Results: Over reporting of exclusive breastfeeding was common at the three different time points regardless of the cut off value used to assess exclusive breastfeeding by the stable isotope technique. Kappa analysis also revealed only slight or fair agreement (K<0.24) between reported and measured exclusive breastfeeding at all time points. Conclusions: Maternal recall of exclusive breastfeeding is limited in accuracy and should be restricted to large scale epidemiological surveys. The more objective gold standard stable isotope method for measuring intake volumes of breast milk should be used to evaluate interventions with smaller representative samples. The study design was longitudinal and observational and was carried out at Cato Manor Clinic, also known as the Umkhumbane Community Health Centre, in Cato Manor, Durban, South Africa. Mothers could access the study site easily when coming from clinic visits and the clinic was located centrally in the community from which the participants were drawn. The study being reported on in this manuscript is a sub-study of a larger study whose main objective was to determine breast milk intake/output volumes at five different time points over 12 months during the period from July 2012 to September 2014 [16]. The sample size calculated to be adequate to answer this research question was 100 mother-infant pairs. The sub-study only included relevant data that was available for the first three time points (infant age of 6 weeks, 3 and 5.5 months). In order to be eligible for inclusion in the parent study [16], and therefore the sub-study, mothers had to be participating in the Improved Nutrition Program that was taking place concurrently at the Cato Manor Clinic. This was a program that provided approximately 16 breastfeeding and nutrition training sessions to mothers over a period of 12 months [16]. Other eligibility criteria for mothers included the following: well (no clinical symptoms or medical history of cardiac or other chronic conditions); no HIV or other infectious disease; intends to breastfeed her baby for 12 months; intends to live in the neighborhood of the recruitment clinic for 12 months after delivery; and African. However, recruitment was slow, and an amendment was approved by the Biomedical Research Ethics Administration of the University of KwaZulu-Natal (UKZN-BREC) in March 2013 that allowed asymptomatic HIV infected mothers to also participate in the study. This resulted in recruitment of 40 HIV infected and 60 HIV uninfected mothers. Exclusion criteria for the mother were: pregnancy; and BMI < 18.5 kg/m2. Inclusion criteria for the infant were: full term; and birth weight > 2.3 kg. Exclusion criteria for the infant were: twins; any defect that interferes with feeding; and chronic illness e.g. congenital heart disease, cerebral palsy. A WHO 24 h recall questionnaire [17] was used to determine infant feeding by maternal recall at each study visit. Classification of exclusive breastfeeding at the first three time points of the study, viz. infant age of 6 weeks, 3 months and 5.5 months was done according to the WHO definition [18] of breast milk, vitamins and mineral supplements and medicine only. To determine exclusivity of breastfeeding, the DTM method was carried out at infant age of 6 weeks, 3 months and 5.5 months. The method involves the mother drinking an accurately measured 30-g dose of deuterium oxide (99.8% atom % purity, Sercon Ltd. UK, lot no. EB2039). Deuterium is a stable isotope of hydrogen, ie it is not radioactive and is considered safe and ethical for use in human studies and is also found in small quantities naturally in the body. No side effects have been noted at enrichment levels less than 0.2% and in this DTM method the deuterium enrichment level in the mother reaches a maximum of 0.1% in the body and less than half of that amount in the infant [19]. The dose of deuterium oxide given to the mother mixes rapidly with her body water and therefore also appears in her saliva and breast milk. If the infant is breastfed the deuterium oxide will pass through the breast milk to the infant and will then mix with the infant’s body water. The infant will thus have an enriched level of deuterium in its body water, which is sampled in its saliva. Saliva samples are taken from the mother and infant over a period of 14 days and the deuterium enrichment is measured in each sample using a Fourier Transform Infrared Spectrometer (FTIR) compared to their pre-dose saliva sample. Each saliva sample is measured twice and precision is high using this method, Coefficient of Variation (CV) values < 1% are achieved and enables differences in deuterium enrichment over time to be determined. The Solver function of Excel® is then used to fit the deuterium enrichment of the saliva samples from the mother and the infant over the 14-day period to model curves. This function minimizes the sum of the squares of the differences of the deuterium enrichments obtained from the FTIR and model values. This then yields values for breast milk intake (labeled water) and non-milk oral intake (NMOI, ie water from sources other than breast milk, which is unlabeled). An infant was determined to be exclusively breastfed according to the DTM method if the value of NMOI is ≤25 g/day [11]. A recent validation study [20] has proposed a higher cut off value of NMOI ≤82.6 g/day and therefore the results from the study are analyzed using both cut off values. The DTM method has been validated against the test weighing technique for measuring milk intake volumes and good correlations were observed [21]. It is a highly sensitive technique as deuterium enrichments in the infant’s saliva can only originate from the mother’s milk. A Kappa analysis was performed to validate reported infant feeding practice against the infant feeding practice as determined by the DTM method using STATA Version 13, © Statacorp, Texas, USA. Kappa analysis compares the observed level of agreement with the level of agreement expected by chance alone, with a scale of 0, representing an agreement that can be expected by chance alone, to 1.0 which would represent a perfect agreement (or − 1.0 a perfect disagreement) [22]. Further sub-classifications are: 0.01–0.20 slight agreement; 0.21–0.40 fair agreement; 0.41–0.60 moderate agreement; 0.61–0.80 substantial agreement and 0.81–1.00 almost perfect agreement.