Relationship Between Breast Milk Feeding and Atopic Dermatitis in Children

Objective: To determine whether or not the breast milk feeding has a role in the prevalence of atopic dermatitis among children. Methods: The target population of the study was all children participating in health check-up program for 3-year-old children in 60 municipalities locating 10 selected prefectures during designated 2 months between October and December 1997. Using a questionnaire, information on nutrition in infants (breast milk only, bottled milk only, or mixed), parity, mothers' age at birth, and a history of atopic dermatitis was obtained. Besides, data on potential confounding factors were obtained. Results: Questionnaires from 3856 children (81.6% of those who were to participate in the programs, and 96.4% of children who participated them) were analyzed. After the adjustment for all potential confounding factors using unconditional logistic models, the risk of atopic dermatitis was slightly higher among children with breast milk (odds ratio [OR]= 1.16 with 95% confidence interval [CI] 0.96-1.40). Mothers' age at birth (OR for those who were more than 30 years or older in comparison with those who were younger than 30 years = 1.15; 95% CI, 0.96-1.37) and those with second or later parity orders (OR=1.14, 95% CI;0.95-1.35) showed odds ratios that were higher than unity without statistical significance. Conclusion: Breast milk elevates the risk of atopic dermatitis slightly without statistical significance; the risk may be, however, higher in children in second or later parity orders. J Epidemiol, 2000 ; 10 : 74-78

. On the other hand, Japanese data revealed that atopic dermatitis was more prevalent among children with breast feeding than among those with bottle feeding 8); from the Japanese results, some researchers guess that dioxins and furans in breast milk cause the dermatitis.
To discuss the relationship between breast milk and atopic dermatitis in children, we observed the prevalence of the disease and infant feeding in Tochigi prefecture using data for cider pollinosis 9). According to the study, the following facts were observed. The breast feeding elevated the risk of atopic dermatitis slightly. However, the dermatitis was more prevalent among those with second or later parity orders.
To certify and get consistency of the results obtained by the aforementioned observation, we observed another data set obtained from 10 prefectures all over Japan using the identical questionnaire.

METHODS
The target population was all participants of health check-up programs conducted by Maternal and Child Health Act of 1965. All the 3-year-old children that participated in the program in 60 municipalities of 10 prefectures during designated 2 months between October and December 1997, were asked to join the study. The parents or guardians were sent a questionnaire by mail before the health check-up, and asked to bring the fulfilled questionnaire at the check-up.
The questionnaire was originally constructed to observe the prevalence of cider pollinosis among young women who have 3-year-old children, but included items about feeding of the child when he/she was an infant, whether or not for the child to have a history of atopic dermatitis and current eczema, and other potential confounding factors, such as mother's age, panty order, and mother's medical history (see Tables 3 and 4 in detail). The main dependent variables were whether a child had a history of atopic dermatitis, with a physicians diagnosis. In addition whether the child had current eczema, which was not a physician's finding at the health check-up but a self-administrated answer by parents or guardians, was observed as a dependent variable supplementally because the eczema is a skin disorder as well. The main independent variable was the feeding methods when he/she was under 1 year of age; breast feeding, bottle feeding, or their mixture. Breast feeding effect and mixed feeding (both breast and bottle) effect in comparison with bottle feeding were observed as dummy variables in logistic models (see below). Sex, mother's age, parity order, mother's history associated with allergy (pollinosis, asthma, atopic dermatitis, allergic rhinitis, and urticaria), whether the mother was engaged in an occupation, whether to keep animals as a pet, whether to live with (a) smoker(s) were used as independent variables as well to control potential confounding factors.
Using unconditional logistic models, we calculated odds ratios and their 95 percent confidence intervals. If a 95 percent confidence interval did not include 1.0, we consider the finding with statistical significance. The SPSS 8.0 was used for the calculation.

RESULTS
A total of 4726 3-year-old children were to take the exam during the study period, and 3999 (84.6 percent) took it indeed.
Of the 3999 participants of the exam, 3856 parents or guardians (81.6 percent of the all target children, and 96.4 per- cent of the examinees) responded to the survey. Table 1 shows the numbers of municipalities and children (guardians), and prevalence of the history of atopic dermatitis and current eczema by prefecture. The prevalence of history of atopic dermatitis was slightly high in Akita prefecture and low in Okayama prefecture, but was between 10 and 30 percent in all the prefectures. Table 2 shows the proportion of those with a history of atopic dermatitis and present eczema, by nutrition during infants. The prevalence of the all items was highest among those with breast feeding, followed by those with mixed feeding, and bottle feeding.
Odds ratios and their 95 percent confidence intervals for the history of atopic dermatitis are shown in Table 3. Mothers' histories of atopic dermatitis, allergic rhinitis, and urticaria were significant risk factors of children's atopic dermatitis, whereas mothers' histories of pollinosis and asthma, which were significant risk factors in the crude analyses, were not significant after adjustment of other factors. Breast feeding elevated the risk of atopic dermatitis in comparison with bottle feeding, but was not statistically significant. The second or later parity order was a significant risk factor in the crude analysis, but the significance diminished by the adjustment. However, the same trend still existed after the adjustment. Table 4 shows the odds ratios and their 95 percent confidence intervals for present eczema. Mothers' histories of allergic disorders also played as risk factors for eczema, in spite of the lack of significance after the adjustment in some diseases. Breast feeding elevated the risk of eczema, but mixed feeding lowered in comparison with bottle feeding only. In these analyses, odds ratios that were higher than unity were observed for second parity order or later without statistical significance.

DISCUSSION
Same as the previous study in Tochigi prefecture 9), we defined the districts and period of time for the observation in the current study first, and obtained the items with very high response rate. These facts mean that this study includes less selection bias, and the validity did not decrease because of the selection bias. Besides, the study population extended to 10 prefectures all over the country as shown in Table 1; the exter- Table 2. Frequency of the history of atopic dermatitis and present eczema by nutrition in infant.
percentages in parentheses unknown excluded Table 3. Odds ratios and their 95% confidence intervals for the history of atopic dermatitis . nal validity should be much higher than that of the previous study.
Recently, one of the most important environmental issues in maternal and child health is dioxins and related chemicals in breast milk. The breast milk, which is fat-rich, is the greatest opportunity to excrete dioxins, which is fat-soluble, from human bodies 10). If dioxins in the breast milk cause atopic dermatitis, the chemicals may influence other parts of human bodies. If so, we should recommend stopping breast feeding. On the other hand, the breast feeding has many advantages that bottled milk does not have. Therefore, we have to compare the advantages and disadvantages of stopping breast feeding. Thus, the evaluation of health effects of dioxins contained by the breast milk is essential. Unfortunately, we have little data about this, especially in epidemiologic field.
If dioxins and furans in breast milk cause atopic dermatitis of children, first parity children should be more risky because their breast milk are expected to be more concentrated with the chemicals 1612). Even if breast feeding is a risk factor of atopic dermatitis in childhood, thus, we should consider other mechanisms than dioxins/furans in breast milk. From this point of view, we conducted the current study following the previous Tochigi study 9). This is not a study in which the level of exposure to dioxins and related chemicals was observed. Exposure levels of such chemicals just started to be measured recently. We guess three reasons not to observe the human exposure levels in Japan; indifference, invasiveness, and cost. European countries started to decrease dioxins in 1980's 13); Japan gets behind in 10 or 20 years. Exposure levels of human body to dioxins are evaluated through three kinds of human specimen; adipose tissue, blood, and breast milk. To get the adipose tissue from an alive human body is invasive. Because the concentration of dioxins in the blood is very low, a 50 or 100 ml of blood sample is required; this is also invasive. The breast milk is limited to only young females. To measure the concentration of dioxins is so expen-sive that about 200-300 thousand yen per specimen is required. Therefore, even the measurement of dioxins in human bodies started in earnest a few years ago in Japan; few epidemiologic data exist in this country. Concerning such situations in Japan, we stress the importance of our data to discuss the relationship between dioxins and health effects even if we did not observed the concentration of the chemicals in the breast milk.
The prevalence of atopic dermatitis was 20.1 percent, and this is quite similar to the figure obtained in Tochigi prefecture 9). This value, about one fifth, was not so far from the results provided by some other studies in the past in Japan 9).
Breast feeding was a possible risk factor of atopic dermatitis without statistical significance after adjustment of other factors in the current study. This finding has already been revealed in the previous Tochigi study 9). The interpretation is; (1) by chance, (2) breast feeding as a real risk factor of atopic dermatitis, (3) existence of confounding factors which could not be adjusted in the study, and (4) combination of them. In particular, the small odds ratio (1.16) is vulnerable to confounding factors. From this study we cannot say that breast milk is a definite risk factor of atopic dermatitis in children.
In addition, even if breast milk is the risk factor of atopic dermatitis, our results did not support the hypothesis that dioxins and furans in breast milk cause the dermatitis . As well known, dioxins and furans are lipid-soluble and there is few pathways to excrete from human body. Therefore, their halflife time in human body is very long . Breast milk, which is rich in lipids, provides the largest opportunity to reduce the chemicals in the body 11,14). Thus, breast milk for children with second or later parity order is expected to contain less dioxins and furans than that for the first baby . However, the current data as well as Tochigi data 9) show that second or later parity order is a small risk factor, and are inconsistent to the dioxin theory . Similar results in the supplemental observation for present eczema were obtained as well.
In conclusion, the findings in this study did not support the hypothesis that dioxins and furans in breast milk induce atopic dermatitis in children. Other mechanisms than the dioxin theory should be considered even if breast feeding is one of the risk factors of atopic dermatitis.