Our study provides with valuable data relating to pooled analysis, extracted from 50 studies of Se concentrations in EHM. It could be used for estimates of Se intake in human milk-fed infants during, and shortly after the first year of life.

We could not find in the literature similar studies which reviewed in a systematic manner Se contents of human milk over various periods of lactation.

Our study has, however, several inherent limitations. The first one is that our review, although systematic, is a cross-sectional analysis of the literature, and in no manner can be considered as longitudinal. Indeed, all 50 papers retained for analysis used different groups (in terms of stages of lactation). Some articles only measured Se concentrations in one specific period of lactation, while others used several periods, and mostly cross-sectionally. Consequently, at various periods of lactation, the pooled sample size varied from 27 combined measurements (in the > 6–12 months of lactation period) to 5708 measurements (in the ≥ 3 days–3 months of lactation). Obviously, the confidence interval of the measurement was affected by the sample size. Moreover, while Se values were apparently the highest in Colostrum and the first trimester of lactation, and subsequently increased, we believe that the temporal variations we describe here should be confirmed only in a longitudinal manner. Nevertheless, studies that were retained for analysis, whenever they had longitudinal measurements of colostrum and subsequent measurements, found universally that the colostrum concentrations were the highest ones (for instance references [13,14,15]). The reason for such high Se concentrations in colostrum might be the fact that Se is secreted together with proteins [16, 17], and we know that colostrum is characterized by a very high protein concentration [18].

Another limitation is that in we could not in this study correlate maternal Se status with EHM Se concentrations. A few individual studies that attempted such analyses had inconsistent results [19,20,21]. However, it is generally believed that human milk Se concentrations might be related to maternal intake, and therefore to geographic location of the population studied [17]. Unfortunately, it was not possible in this study, to precisely determine from the country of origin of each article what was the maternal Se intake.

From our results, we may calculate that an exclusively breastfed, 3 months old infant weighing an average of 6.4 kg, and taking an average of 700 cc of term human milk per day [22] is expected to have an average Selenium intake of 27.5 × 0.7 = 19.25 µg, or 3 µg /kg body weight. This number is actually higher than previous estimates of Se intake in breastfed infants of 2.3 mcg/kg per day [8]. It is however lower than current recommendations of Se intake, but these recommendations are not likely to lead to toxicity.