Our research group recently published an estimation of a dietary intake of total and individual polyphenols in a representative sample of the Brazilian general population aged 10 years or older^{(Reference Carnauba, Hassimotto and Lajolo1)}. In comparison with available data, the present estimates of total polyphenols intake were lower than that previously reported for other populations, including Latin American countries (Mexico)^{(Reference Zamora-Ros, Biessy and Rothwell2)} and small Brazilian samples^{(Reference Miranda, Steluti and Fisberg3,Reference Nascimento-Souza, de Paiva and Pérez-Jiménez4)} . As we pointed out in the ‘Discussion’ section, the intake of fruit and vegetable was very low, as well as the food variety in the Brazilian diet, which illustrates the low quality of their diet and the low polyphenol intake^{(Reference Carnauba, Hassimotto and Lajolo1)}.

We would like to thank Tomoyuki Kawada for his comment, and we agree that large heterogeneity in intakes of polyphenols could be justified by ethnic differences. As we highlighted in the ‘Discussion’ section, the different dietary preferences and habits according to different populations, which are related to the food availability and local cultures, could be an explanation to the differences between polyphenol intake across populations^{(Reference Carnauba, Hassimotto and Lajolo1)}. In relation to the study developed with adults and elderly adults from São Paulo^{(Reference Miranda, Steluti and Fisberg3)}, the median polyphenol intake was 292·2 mg/d for adults (v. 205 mg/1000 kcal/d (4184 kJ/d) as aglycone equivalents for the entire Brazilian population) and 348·6 mg/d for elderly adults (v. 225 mg/1000 kcal/d as aglycone equivalents for the entire Brazilian population). Although the city of São Paulo is one of the most important of Brazil, its dietary habits do not represent the entire population of the state of São Paulo, as well as the country. We consider our analysis more accurate because, in addition to using a larger sample, estimates were performed using sample weights to allow population representativeness.

We appreciate the opportunity to respond the letter from Kawada’s and his interest in our work.