Although breast-feeding has long-term benefits for children’s health such as prevention of overweight/obesity and type 2 diabetes and improvement of cognitive function( 1 ), the protective effect of breast-feeding on asthma in childhood still remains controversial( Reference Quigley, Kelly and Sacker 2 – Reference Miyake, Yura and Iki 11 ). Two meta-analyses( Reference Gdalevich, Mimouni and Mimouni 4 , Reference Ip, Chung and Raman 5 ) and a recent study in New Zealand( Reference Silvers, Frampton and Wickens 6 ) reported that breast-feeding continued to be associated with decreased risk of asthma in children after cessation of breast-feeding. However, a large-scale randomized trial( Reference Kramer, Matush and Vanilovich 7 ) and several observational studies( Reference Sears, Greene and Willan 8 – Reference Miyake, Yura and Iki 11 ) found no effect or even an increased risk of asthma. The Japanese Pediatric Guideline for Food Allergy concluded that evidence was insufficient that exclusive breast-feeding had more preventive effects on allergy than partial breast-feeding( 12 ).
In the present study, we evaluated the associations between breast-feeding and hospitalization for asthma in early childhood (i.e. between the ages of 6 and 42 months) by using data from a nationwide longitudinal survey in Japan. We also examined the dose–response relationship between the duration of breast-feeding and children’s hospitalization for asthma.
Materials and methods
Study participants
We extracted the data from the Longitudinal Survey of Babies in the 21st Century conducted by the Japanese Ministry of Health, Labour and Welfare. The survey was targeted at all families whose babies were born between 10–17 January and 10–17 July 2001, all over Japan (n 53 575)( 13 ). Baseline questionnaires were mailed to the targeted families when children were 6 months of age and were returned by 47 015 families (87·8 % response rate). Follow-up questionnaires were annually sent to the families who responded at baseline. The data were also linked individually to birth records from vital statistics of Japan. A detailed description of the survey is available elsewhere( Reference Yamakawa, Yorifuji and Inoue 14 ).
In the present study, we targeted children who were followed until the fourth survey (i.e. between the ages of 6 and 42 months). We restricted study participants to children born after 37 gestational weeks and those who were singleton births, because of the potential influence of prematurity and multiple births on children’s health( Reference Liu, Covar and Spahn 15 ). This resulted in excluding 2418 and 976 children. Children with missing information on infant feeding practices (n 742) were also excluded. Among eligible children (n 43 367), 2761, 2038 and 1677 children were lost to follow-up at the second, third and fourth surveys, respectively. Finally, 1686 children were excluded due to missing information on hospitalization, leaving 35 215 children for analyses.
Infant feeding practices
The first survey included questions on infant feeding practices at age 6 months to determine whether children had been breast-fed, fed colostrum only or never breast-fed. The duration of breast-feeding was then queried. In the present study, we combined the categories of ‘never breast-fed’ and ‘colostrum only’ into a category of ‘formula feeding only’ because the number was small in the category of ‘never breast-fed’ and the duration of feeding colostrum was short. We made the categories of ‘partial breast-feeding’ and ‘exclusive breast-feeding’ by combining queries on breast-feeding and formula feeding. We further divided the category of ‘partial breast-feeding’ into three categories based on the duration of breast-feeding. We do not know when children started being partially breast-fed. Although the first survey was conducted at age 6 months, children breast-fed up to 7 months were included due to the timing of the response. Finally, we had five categories for infant feeding practices as follows: formula feeding only; partial breast-feeding (breast-feeding for 1–2 months, 3–5 months and 6–7 months)( Reference Yamakawa, Yorifuji and Inoue 14 ); and exclusive breast-feeding at 6–7 months of age.
Children’s hospitalization for asthma
We used hospitalization as our health outcome for three reasons. First, it is likely to be reported correctly by families( Reference Oddy, Sly and de Klerk 16 ). Second, children’s hospitalization is used as a marker of the disease severity( Reference Quigley, Kelly and Sacker 2 ). Third, hospitalization may be a more valid outcome for children than hospital visit, because Japan has good access to hospitals for everyone but a medical doctor’s judgement is necessary for hospitalization. A history of hospitalization was queried in the second, third and fourth surveys (i.e. parental report of doctor’s diagnosis). The question was whether or not children had been hospitalized during the past 12 months. Subsequent questions ascertained hospitalization for specific causes including asthma. We thus considered as cases children who were hospitalized for asthma at least once between the ages of 6 and 42 months.
Covariates
We selected the following characteristics as potential confounding factors. Maternal educational attainment was obtained in the second survey. We classified the original eight categories into four as follows: junior high school or others; high school; junior college (2 years) or vocational school; and university (4 years) or higher. Maternal smoking habit was ascertained at the first survey. Children’s sex and parity were listed in the birth record. The information on parity (i.e. the number of deliveries) was used as proxy for the presence of older siblings. Persons who usually took care of children were asked about in the first survey, and we assumed children being taken care of by nursery teachers as attending day care.
Statistical analysis
To evaluate the impact of loss to follow-up, we first compared baseline characteristics between eligible children, children included in the analyses and those who were lost to follow-up between the first and fourth surveys.
We used logistic regression models to estimate crude odds ratios for the associations of infant feeding practices with children’s hospitalization for asthma using the category of ‘formula feeding only’ as a reference (model 1). Next, we adjusted for maternal factors (i.e. educational attainment and smoking habit; model 2). We then additionally adjusted for children’s factors (i.e. sex, day-care attendance and presence of older siblings) in the final model (model 3). Furthermore, instead of using the five infant feeding categories, we examined a dose–response relationship between breast-feeding duration (i.e. 0 months, 1 month, 2 months, 3 months, 4 months, 5 months and 6–7 months) and children’s hospitalization for asthma, adjusting for the same set of covariates.
In sensitivity analyses, we further adjusted for paternal annual income (as a continuous variable) in addition to the same set of covariates in model 3 to examine the potential impacts from familial socio-economic status as suggested in a previous study( Reference Yngve and Sjöström 17 ). Furthermore, we performed the same analyses excluding asthma cases at the second survey (i.e. between the ages of 6 and 18 months), because diagnosing asthma could be difficult at an earlier age.
All confidence intervals were calculated at the 95 % level. The statistical software package STATA SE version 12 was used for all analyses.
The study was approved by the Institutional Review Board of Okayama University Graduate School of Medicine, Dentistry and Pharmaceutical Sciences (number 881).
Results
Table 1 shows the baseline characteristics of eligible children at the first survey, children in the analyses and those who were lost to follow-up between the first and fourth surveys. Children who were lost to follow-up were more likely than the eligible children to have been formula-fed and to have mothers who smoked.
Table 1 Baseline characteristics of eligible children, children in the analyses and children lost to follow-up between the first and fourth surveys, Longitudinal Survey of Babies in the 21st Century, Japan
Table 2 shows the demographic characteristics of children included in the analyses according to infant feeding practices. Children who had been exclusively breast-fed at 6–7 months of age were more likely to have mothers with a higher academic attainment and who were non-smokers than those who had been formula-fed. In addition, children who had been exclusively breast-fed at 6–7 months of age were less likely to attend day care and were more likely to have older siblings.
Table 2 Demographic characteristics of children included in the analyses by infant feeding practices (n 35 215), Longitudinal Survey of Babies in the 21st Century, Japan
Table 3 shows the results for the associations between infant feeding practices and children’s hospitalization for asthma between the ages of 6 and 42 months. After adjusting for the covariates, the protective associations were marginally significant in models 2 and 3. The adjusted OR was 0·77 (95 % CI 0·56, 1·06) for exclusive breast-feeding at 6–7 months of age in the final model (model 3). One-month longer duration of breast-feeding was associated with a 4 % decreased risk of hospitalization for asthma (OR=0·96; 95 % CI 0·92, 0·99).
Table 3 Associations of infant feeding practices with hospitalization for asthma in children between the ages of 6 and 42 months (n 35 215), Longitudinal Survey of Babies in the 21st Century, Japan
Model 1: no adjustment was performed.
Model 2: maternal factors (educational attainment and smoking habit) were adjusted for.
Model 3: children’s factors (sex, day-care attendance and presence of older siblings) in addition to maternal factors were adjusted for.
In the sensitivity analyses, even after we further adjusted for paternal income in addition to the covariates in model 3, the main findings for the associations with hospitalization due to asthma did not change substantially (data not shown). When we conducted the analysis with the same covariates in model 3 excluding the outcome at the second survey, the estimates were still protective but less precise. The OR for exclusive breast-feeding at 6–7 months of age was 0·88 (95 % CI 0·61, 1·26). The dose–response relationship turned to be marginally significant (OR=0·97; 95 % CI 0·93, 1·01).
Discussion
We examined the effects of breast-feeding on children’s hospitalization for asthma between the ages of 6 and 42 months using a large Japanese nationwide data set. We found that exclusive breast-feeding at 6–7 months of age was associated with decreased risk of hospitalization for asthma. Moreover, we observed a dose–response relationship between breast-feeding duration and hospitalization for asthma. These protective associations did not change substantially in the sensitivity analyses.
A past study in Japan suggested that exclusive breast-feeding was associated with increased risk of asthma among children( Reference Takemura, Sakurai and Honjo 10 ). By contrast, a recent study in New Zealand showed that exclusive breast-feeding protected against asthma up to 6 years of age( Reference Silvers, Frampton and Wickens 6 ). These inconsistencies may be derived from variations in study designs and settings (e.g. children’s age, the duration of exclusive breast-feeding, different environments surrounding children, residual confounding or reverse causation). From a perspective of mechanisms to prevent asthma, children may receive benefits from excluding infant formula, foods or other drinks containing potential allergenic factors as well as from consuming anti-inflammatory, nutritional or other elements in breast milk( Reference Silvers, Frampton and Wickens 6 , Reference Oddy 18 – Reference Oddy, Holt and Sly 20 ).
A major strength of the present study is that we used a large nationally representative sample of children( Reference Silvers, Frampton and Wickens 6 , Reference Kramer, Matush and Vanilovich 7 , Reference Takemura, Sakurai and Honjo 10 , Reference Miyake, Yura and Iki 11 ). In addition, the response rate at baseline was very high (i.e. 87·8 %), which strengthens the validity of our findings. The data on duration of breast-feeding should be accurate because the information on feeding practices was collected at when the children’s age was 6–7 months.
An important limitation of the present study is that we do not examine the maternal or children’s history of allergy. Although influence by maternal history of allergy is conflicting( Reference Gdalevich, Mimouni and Mimouni 4 , Reference Friedman and Zeiger 21 , Reference Kull, Almqvist and Lilja 22 ), a meta-analysis suggested that breast-feeding was protective for asthma in early childhood irrespective of maternal allergy adjustment( Reference Gdalevich, Mimouni and Mimouni 4 ). Moreover, maternal or children’s history of allergy was found to affect decisions regarding breast-feeding( Reference Lowe, Carlin and Bennett 23 ), which may attenuate the protective associations of breast-feeding with asthma among children if those at higher risk of asthma because of familial history of allergy or children’s eczema are more likely to be breast-fed.
We have the following weakness on the breast-feeding information. First, we could not evaluate the effect of breast-feeding beyond 6–7 months of age. Second, we could not ascertain through the questionnaire whether children were given any liquids or solid foods until age 6 months, although weaning is recommended to be started at 5–6 months of age in Japan( 24 ). Thus, predominantly breast-fed children may be included in the category of ‘exclusive breast-feeding’, resulting in the underestimation of the present result.
Medical doctors may have difficulty in diagnosing asthma for young children( Reference Liu, Covar and Spahn 15 ). Although children who had severe asthma are unlikely to be diagnosed as respiratory tract infections, children with asthma may have been misdiagnosed as respiratory tract infections but not as asthma. In addition, children who had wheezing due to respiratory tract infections could be misdiagnosed as asthma. These misclassifications, however, would be non-differential, moving the effect estimates toward the null. In the present study, we conducted sensitivity analysis excluding asthma cases between the ages of 6 and 18 months, and the estimates were still protective.
Furthermore, the loss of participant children from the first survey to the fourth survey should be considered. As shown in Table 1, children who were lost to follow-up were less likely to have been breast-fed and more likely to have mothers who smoked, compared with the eligible children. Such a differential loss would underestimate the protective effects of breast-feeding on hospitalization for asthma.
Conclusion
In conclusion, we found beneficial effects of breast-feeding on hospitalization for asthma in early childhood.
Acknowledgements
Acknowledgements: The authors would like to thank Dr Kazuko Sugai, NHO Fukuyama Medical Center, for valuable comments. Financial support: This work was funded by Health and Labour Sciences Research Grants on Health Research on Children, Youth and Families (number H24-Jisedai-Ippan-004). The funder had no role in the design, analysis or writing of this article. Conflict of interest: None. Authorship: M.Y. designed the study and its analytical strategy, performed the data analyses, drafted the initial manuscript, and approved the final manuscript as submitted. T.Y. obtained the funding for the study, designed the study and its analytical strategy, obtained the data, reviewed and revised the manuscript, and approved the final manuscript as submitted. T.K. carried out the initial analyses, reviewed and revised the manuscript, and approved the final manuscript as submitted. Y.Y. contributed to the interpretation of results, reviewed and revised the manuscript, and approved the final manuscript as submitted. H.D. coordinated and supervised the study, critically reviewed and revised the manuscript, and approved the final manuscript as submitted. Ethics of human subject participation: The study was approved by the Institutional Review Board of Okayama University Graduate School of Medicine, Dentistry and Pharmaceutical Sciences (number 881).
