Our final linear regression model showed that higher breakfast frequency was associated with lower B-Pb (beta = -0.10 units of log-transformed B-Pb compared with children who did not eat breakfast regularly, p = 0.02) after controlling for covariates (Table ​(Table2).2). This beta, when calculate back to original scale of B-Pb, means that children eating breakfast regularly had 0.8 μg/dL (or 10%) lower B-Pb than those did not eat breakfast regularly after excluding the impacts of other factors. Other covariates including gender, age, living area, mother's education, and father's occupation were also predictors of B-Pb. While being a boy, living in a rural area, older age, and parents as professional workers were associated with higher B-Pb, children of mothers with higher education tended to have lower B-Pb. In a linear regression model of 270 children with data of calculated daily dietary intake of micronutrients (Ca, Fe, Zn, Cu, Mg, and vitamins), we did not find associations between dietary intakes of minerals and log-transformed B-Pb (data not shown). In the Logistic regression model of lead poisoning risk (Table ​(Table3),3), boys had 1.92 higher odds of being lead poisoning than girls, and that father's university level education reduced the odds to 0.38 times that of father with less than high school education. Compared with 3-year-old children, 4-year or 5-year old children had about 2 time odds of being lead poisoning. Regular breakfast, however, did not predict lead poisoning (B-Pb≥10 μg/dL). Blood micronutrient levels were not associated with B-Pb or the risk of lead toxicity (data not shown).

A number of studies have shown that food in the gastrointestinal tract reduces the absorption of ingested lead in adults [7,9,10,23]. Experimental studies showed that the bioavailability of ingested lead in adults when taken with a meal was about 10 times lower than that when ingested after fasting [7]. The smaller magnitude of the influence in children (15% change) may due to relatively lower lead exposure dose in children, observational study design, and the difference in the kinetics of lead between children and adults.

The mechanisms behind the impact of fasting on gastrointestinal tract absorption of lead are not well understood. Mineral micronutrients, especially the presence of calcium and phosphate in the intestinal lumen, may play a role by competing with lead for absorption [7]. Epidemiologic studies have shown increasing dietary calcium intake is inversely associated with B-Pb in children [15,16]. Higher dietary iron intake is also associated with lower B-Pb in the U.S. children[17,18]. In the present study, however, we did not find significant associations between blood levels of these mineral micronutrients and B-Pb. We did find significantly higher blood levels of zinc in children who ate breakfast regularly, which may suggest that their dietary zinc intake was greater and perhaps competed with lead for absorption. Our analysis of dietary micronutrient intake in a subgroup of 270 children, however, did not suggest higher zinc intake in regular breakfast eaters. In previous studies, while some studies observed associations between dietary intake/supplementation (Ca, Fe, and Zn) and B-Pb [8,15-17]others did not [11,24]. In a study in the US, Gallicchio et al. did not found associations between B-Pb and daily micronutrient intakes (iron, calcium, vitamin C, or vitamin D) after controlling lead exposure and child's age [11]. A clinical trial showed that Ca supplementation did not statistically significantly change B-Pb among children 1 to 6 years of age when B-Pb were 10- 45 μg/dL [24]. There are two possible explanations for the inconsistency with this study. First, B-Pb are lower than those in previous studies. Second, children in this study are younger than the ones in previous studies.

The associations between socio-demographic factors and B-Pb found in previous studies were also observed in the present study [19,25-29]. Some of these factors including age and gender are strongly associated with B-Pb, but not modifiable. Other factors including living area and parental education and occupation might not be possible to improve solely through public health intervention. Targeting children's breakfast consumption habits is relatively more feasible at both the family and community levels.

Several limitations must be considered when interpreting the findings from the present study. We have no data on the key variable of environmental lead exposure, a factor that might explain the higher B-Pb and greater risk of toxicity in older children if cumulative lead exposure increases over time. Clearly, our findings regarding increased B-Pb with increasing age merit repeated measures over time together with measures of environmental lead. We have information of dietary intake of micronutrients on only 270 of the larger sample of 1344 children, and thus cannot be certain that these micronutrient intake levels are representative of the entire group. Parental report of regular breakfast consumption in their children may lead to bias and misclassification. A guide was provided to parents to answer this question. However, for those children primarily living with their grandparents, the information may be less accurate. We did not ask parents the number of days the child ate breakfast during the week but instead used an arbitrary definition of "at least 5 days per week" as an indication of regular breakfast consumption. This was suggested by our local partners who believed that this would reasonably distinguish children. Second, the association observed in this cross-sectional study may not represent a causal relationship. There may be a tendency for higher B-Pb's to be associated with a decreased appetite for eating breakfast. Given the data collected from this study, we cannot determine to what extent appetite affected the reported breakfast consumption. Third, the lack of association in the multivariate logistic regression analysis on lead poisoning is likely due to the limited magnitude of nutritional influence at relative low exposure level [24] and the small number of children with elevated B-Pb (7.9% for children with regular breakfast and 10.4% for children with irregular breakfast). Fourth, other potential confounders (e.g., household income and drinking water source) which were not assessed or included in the multiple regression models might contribute to the lower B-Pb observed among children who ate breakfast regularly. Finally, given the impact of fasting on lead absorption, regular breakfast is expected to modify the associations between external exposure levels and B-Pb which we cannot examine in this study due to lack of external exposure data.