Abstract
Exclusive breastfeeding (EBF) prevalence at the population level is typically assessed through maternal or caregiver reports, which are prone to recall and social desirability biases. The deuterium oxide dose-to-mother (DTM) technique offers an objective alternative by administering a small amount of deuterium-labeled water to lactating mothers. This study compared the prevalence of EBF from maternal recall versus the DTM method in Eswatini and evaluated the strengths and limitations of both approaches in assessing infant feeding practices linked to malnutrition. A total of 102 mother-infant pairs were recruited from three regional hospitals and three health centers in Eswatini. EBF was assessed by maternal recall using the WHO 24-hour infant feeding recall method and objectively measured using the DTM stable isotope technique at three infant age intervals: 6 weeks to 2.4 months (n = 26), 2.5 to 4.0 months (n = 43), and 4.1 to 5.5 months (n = 33). Two cutoff values defined EBF status via the stable isotope method. Agreement between maternal recall and DTM measurements was analyzed using Kappa statistics. Maternal recall consistently overestimated EBF across all time points compared to the DTM method. Using Model 1 cutoffs, prevalences of EBF by maternal recall vs. DTM were 100% vs. 50% (≤2.4 months), 79.5% vs. 18.2% (2.5-4 months), and 71.9% vs. 18.8% (4.1-5.5 months). For Model 2, prevalences were 100% vs. 65.4%, 79.5% vs. 50%, and 71.9% vs. 43.8%, respectively. The difference between the two models is the cutoff that was used to determine EBF using the DTM method. Kappa analysis showed minimal agreement between methods at each time point. The DTM technique provides a more accurate and objective measure of EBF than maternal recall and is the gold standard for smaller sample studies. Maternal recall substantially overestimated exclusive breastfeeding prevalence, while the deuterium oxide technique yielded more reliable, objective assessments of infant feeding practices, highlighting the need to interpret recall data with caution due to overreporting.