Infectious diseases are a leading cause of morbidity and hospitalization in infants and children.1 Breast-feeding offers substantial protection against infectious diseases in infancy,2–4 including respiratory infections,3,5–8 gastrointestinal infections (GI),9–12 and otitis media.6,11,13 Breast-feeding reduces outpatient treatment for respiratory, GI, and febrile illnesses up to 18 months of age and reduces hospitalization for respiratory illnesses up to 9 months of age.8 However, most of these studies from western developed countries are subject to possible bias by uncontrolled or uncontrollable confounding by socioeconomic position.
Furthermore, few studies have examined the impact of breast-feeding on infectious disease beyond the first few years of life. A longer duration of breast-feeding has been found to protect against pneumonia in children of 1 to 2 years of age6 and otitis media in children until 3 years of age.14–17 Breast-feeding has also been shown to reduce the risk of respiratory illness in children up to 7 years of age18 and to protect against Haemophilus influenzae type b (Hib) infection up to 10 years of age.19 The majority of the studies on the benefits of breast-feeding in older children have methodological limitations including retrospective design,9,15,19 failure to adjust adequately for known confounders,20 small sample size,18 and failure to stratify children by age group to differentiate the effects of breast-feeding on infants and older children.21–23 The distinction between the effect of breast-feeding in infancy and in childhood is important in understanding the biologic mechanism of breast-feeding—whether it confers passive immunity in infancy alone or if it promotes maturation of infants' active immunity and protection beyond infancy.24
In contrast to long-term developed countries, where breast-feeding mothers tend to be better educated,25 educated Hong-Kong-born mothers tend to cease breast-feeding relatively early whereas less educated or migrant mothers tend to continue breast-feeding.26 Moreover, some common infectious diseases, such as influenza, have different patterns in Hong Kong than in other countries.27 Hence, economically developed Hong Kong offers a useful nonwestern setting in which to examine the long-term effects of breast-feeding.
We prospectively examined whether breast-feeding was associated with hospitalizations for infectious diseases until 8 years of age in a large, population-based birth cohort of Hong Kong children. Our primary hypothesis was that breast-feeding would protect children against hospital admission for respiratory and gastrointestinal infections up to 8 years of age. We also specifically examined whether breast-feeding was protective beyond the immediate period of breast-feeding.
Sources of Data
The basis of this study is a large, prospective population-based birth cohort “Children of 1997,” which had initially been recruited through all 49 publicly funded Maternal and Child Health Centers of the Hong Kong Department of Health in 1997.28 The sampling frame consisted of all infants born in April and May 1997 and brought to one of the health centers for their first postnatal visit. For the index year, 92% of infants born in Hong Kong went at least once to one of these clinics, which provide free-of-charge preventive care and immunizations. The “Children of 1997” birth cohort study recruited 8327 mother-infant pairs, covering 88% of all births during the recruitment period. At their first (baseline) visit, mothers provided information on infant feeding, household smoking patterns, socioeconomic position (parental education, employment status, and type of housing), and birth characteristics (mode of delivery, gestational age, maternal age, and parity) through a standardized self-administered questionnaire in Chinese.
Follow-up questionnaires were administered at subsequent well-child visits at 3, 9, and 18 months after birth to update information on infant feeding status, health-care utilization, and household smoking patterns. Infants participating in the birth cohort study were similar to the general population on several sociodemographic characteristics, such as parental education and type of housing, with relatively small Cohen effect sizes (<0.17).
In 2005–2006, more detailed information on socioeconomic position was abstracted manually from the original hardcopy clinic records, and hospital discharge data were obtained by record linkage from the Hospital Authority. The Hospital Authority manages all public hospitals in Hong Kong, and these hospitals account for 89% of acute bed-days and 87% of all hospital admissions for children under 8 years old in Hong Kong.29 Information contained in the Hospital Authority discharge records includes the date of each hospital admission and discharge, name of hospital, principal and secondary reasons for admission, and principal medical procedures.
Information on breast-feeding was obtained from self-administered questionnaires in Chinese completed by the primary caregiver (most commonly mothers), with help as necessary, at the first postnatal (baseline) visit to the Maternal and Child Health Centers and at subsequent routine visits when the infants were about 3, 9, and 18 months of age. At the baseline visit the primary caregiver was asked “How is the infant currently fed?” specified as “Exclusively breast-fed,” “Partially breast-fed,” and “Only formula-fed.” The terms exclusively breast-fed and partially breast-fed were not defined but were intended to refer to the type of milk—exclusively breast-fed meaning only breast milk and partially breast-fed meaning breast milk plus formula. At follow-up visits, the primary caregiver was asked “How has the infant been fed from birth until now?”—specified as 1 of the 3 original options or an additional option applying to the initially exclusively breast-fed of “Initially breast-fed, but now formula-fed,” with the caregiver asked to also recall the age in months when breast-feeding terminated. As always with studies with repeated measurements, follow-up information was sometimes missing and answers were sometimes contradictory. In such cases, we gave precedence to the most contemporaneous and hence most likely accurate response (eg, for early feeding patterns, information that was reported at 3 months rather than 18 months).
Breast-feeding is usually of limited duration in Hong Kong. Maternity leave in Hong Kong is 10 weeks, which the mother usually starts about 2 to 4 weeks before the due date. In this cohort, few mothers continued breast-feeding beyond 3 months: at 3 months, 6% of infants were still breast-fed; at 6 months, 3% of infants were still breast-fed; at 9 months, 2% were still breast-fed; and at 18 months, 1% were still breast-fed. Therefore, for this analysis we categorized infants by the type of milk they received during the first 3 months: breast milk only, breast milk plus formula, and formula only. Infants were classified as receiving “breast milk only” if they were reported at 3 months or later as receiving only breast milk, without reported prior use of infant formula or other breast milk substitutes. Infants were classified as “partially breast-fed” if, prior to or at 3 months, they were reported as receiving breast milk, but were also reported as receiving formula at baseline, at 3 months, or, if that information was missing, at the next available follow-up. (This includes infants who were initially given breast milk only and then later given formula, with or without breast milk; it also includes infants who were given both breast milk and formula from birth onwards.) Infants were classified as “never breast-fed” if they were formula-fed at the baseline interview and were receiving no breast milk either at 3 months, or, if that information was missing, at the next available follow-up. We do not have the age of starting complementary feeding, and so infants in all 3 of the groups may also have received nonmilk liquids or solid foods.
Outcome: Hospital Use
Hospital utilization was assessed from birth up to 8 years of age. We also considered hospital utilization in various age groups so as to clarify the association beyond infancy. These age groups were chosen to reflect diet transitions, motor development, and increasing social contact (ie, 0–6 months, 6–24 months, 2–5 years, and 5–8 years), as well as to distinguish the primary period of breast-feeding in our cohort (0 up to 6 months) from the rest of the follow-up period. Admissions are coded at discharge according to the International Classification of Diseases, Ninth Version Clinical Modification (ICD-9CM). Admissions with a principal diagnosis of ICD-9CM 33, 34.0, 381–2, 460–6, 477, 480–7, 477, 490, or 493 were classified as respiratory (and related) infections; ICD-9CM 001–009, 535.00, 535.50, 558.9, 538, 535.40, or 787.91 were classified as gastrointestinal infections; ICD-9CM 10–32, 34.1–139, 320–1, 370, 372.0–372.3, 390–2, 540–2, 590, 595, 599.0, 680–6, 771, 780.3, or 780.6 were classified as other infections; and ICD-9CM 800–999 or E800-E999 were classified as accidents. Admissions for accidents (ICD-9CM 800–999 or E800-E999) were used as a control to examine possible residual confounding by socioeconomic position. (With adequate control for socioeconomic position, there should be no relation between breast-feeding and admissions for accidents.) We included both day-patient and in-patient admissions. Children with no record of hospital admission were assumed to have none.
The average hospital stay for infants delivered by cesarean section was 7.9 days, whereas that by natural birth was 3.0 days.30 Therefore, the inclusion of hospitalizations in the immediate neonatal period could have introduced bias. For this reason, admissions occurring within the first 8 days of life were excluded.
We used multivariable regression analyses to assess the association between breast-feeding and hospitalization for various causes during the first 8 years of life. For the primary analysis, we used Cox regression to assess the adjusted association between breast-feeding and time to first hospitalization, from which we reported the hazard ratio (HRs) of admission for all the causes considered (respiratory infections, gastrointestinal infections, other infections, all infections, and accidents) with associated 95% confidence intervals (CIs). Children who died were censored at the date of death. The proportional-hazards assumption was checked by visual inspection of plots of log (−log (S)) against time, where S was the estimated survival function. Children with an admission in an earlier age-group might be more susceptible to admission when older; however, in age-stratified analysis results were similar with and without adjustment for admission at an earlier age.
Finally, because the data were overdispersed for a Poisson model, we used negative binomial regression to assess the adjusted association between breast-feeding and the number of hospitalizations and days of hospitalization. A zero-inflated negative binomial model produced a similar pattern of results, and so for parsimony was not used. For the multivariable analyses, we added potential confounders in turn (categorized as presented later), and we retained confounders based on subject-matter relevance and a change-in-estimate criteria for the association of breast-feeding with earlier admission for any infection.31 On this basis we included sex, type of hospital at birth, and household per capita income (quintiles) as confounders. We did not include highest parental education or maternal and infant characteristics, as these had minimal additional impact on the estimates.
Potential effect modification of the effect of breast-feeding on hospitalization by sex and small-for-gestational age (SGA) status was assessed by statistical significance of interaction terms and heterogeneity of effect across strata. SGA status was defined as birth weight below the tenth percentile by sex and gestational age using the Williams curve,32 for which the cut-offs were derived from 1.3 million non-Hispanic white middle class infants from 1970–1976 birth cohorts in California. In addition, as there is increasing evidence that premature birth is associated with poorer health across many domains,33,34 we also examined possible differential effects of breast-feeding in premature (<37 weeks gestation) and low-birth-weight babies (<2500 g). We assessed the significance of interaction terms in models with and without the interaction term, examining the statistical significance of the likelihood ratio test of the difference between the 2 models on the relevant χ2 distribution.
Ethical approval was obtained from the Institutional Review Board of The Hong Kong West Cluster and the Li Ka-Shing Faculty of Medicine, University of Hong Kong. All analyses were conducted using Stata version 9.2 statistical software (Stata Corp, College Station, TX).
There are 8327 mother-infant pairs in the birth cohort. We excluded 546 with missing breast-feeding status or type of hospital at birth, leaving 7781 (93%) birth-cohort members in this analysis.
Among these 7781 children, half (n = 3881; 50%) were admitted to a public hospital (day-patient or in-patient) for any illness at least once before the end of December 2005 (eTable 1, http://links.lww.com/EDE/A427). Of these, 1966 (51%) were admitted for respiratory infections, 894 (23%) for gastrointestinal infections, and 863 (22%) for other infections.
Within the first 3 months, 5.9% (n = 460) were admitted for any infections, of which 0.8% had admissions at least twice. Of these 460 admitted children, half (53%) were admitted for respiratory infections, one-fifth (19%) were admitted for gastrointestinal infections and 28% were admitted for other infections.
Table 1 shows the baseline characteristics of the mothers and children by breast-feeding status. Fewer than half (43%; n = 3342) of birth-cohort mothers initiated breast-feeding. Just over one-third (37%; n = 2851) gave both breast milk and formula during the first 3 months, and just 6% (n = 491) gave only breast milk during that period. Higher education and higher income were positively associated with breast-feeding initiation, although less educated and lower income women were more likely to give only breast milk for at least 3 months. Women with cesarean delivery and secondhand smoke exposure during pregnancy or postpartum were less likely to breast-feed.
Table 2 shows the hazard ratios for time to first hospital admission by breast-feeding status for each outcome within each age range, adjusted for sex, type of birth hospital, and household per capita income. Breast-feeding was associated with a lower risk of hospitalization in the first 6 months of life for respiratory infections, gastrointestinal infections and any infections, with a clear dose-response relationship. Breast-feeding was not associated with a lower risk of hospitalization for infections beyond 6 months of age.
The associations of breast-feeding with the number of hospital admissions and with the number of hospital admission days for infectious diseases were similar (eTables 2 and 3, http://links.lww.com/EDE/A427). Breast-feeding was associated with a lower number of admissions, and fewer bed-days in the first 6 months of life, with clear-dose response relationships. However, there were no similar associations in the older age groups or for the overall period from birth to 8 years. There was no association between breast-feeding and hospital admission for accidents in any analysis. Finally, there was little evidence that the association of breast-feeding with time to first admission varied with sex, SGA status, low-birth-weight status, or prematurity status (data not shown).
Consistent with previous studies,3,8,35 we found that, even in a highly developed region, breast-feeding provides substantial protection from hospitalization for infectious disease in infancy. Even a short duration of breast-feeding reduced hospitalization for respiratory and possibly nongastrointestinal infections. Nonetheless, we were unable to show that breast-feeding provided any protection from hospitalization for infections beyond infancy up until 8 years of age. Our findings concerning infancy are similar to other studies from developing and developed locations, but our findings at older ages differ from most previous studies, which have found protective effects of breast-feeding against infectious disease in older children. Most previous studies have come from low- and middle-income countries,9,21–23 or from marginalized populations,36 where patterns of infectious disease are different from developed countries.4 Of the studies from developed countries that found breast-feeding protective in older children, one was an ecological study,20 another used parent-reported respiratory symptoms (including wheeze) as the outcome,18 and others were limited to specific diseases, such as otitis media15 or H. influenzae.19 Most studies that considered overall infections did not identify age at illness, ie, whether during the typical period of breast-feeding or later.15,18,37 Therefore, it is not clear whether the protection against serious infection illnesses occurred during breast-feeding or after the cessation of breast-feeding.
In our prospective birth cohort, breast-feeding data were collected prospectively, generally before the onset of health outcomes of interest, thus minimizing reporting bias. Unlike previous studies examining illnesses and hospitalizations,8,35 case ascertainment was based on doctor diagnosis and public-hospital discharge records rather than self-report, thereby eliminating parental recall bias and misclassification. Furthermore, we used a number of regression models to make full use of the hospitalization data, and we considered hospital admissions in terms of 4 different outcomes. These different models all produced similar results. Additionally, we considered the effect on the estimates of a range of confounders, including several measures of socioeconomic position. We considered whether our findings could be due to residual confounding by examining whether breast-feeding status was associated with admission for accidents. In the age range from 0 to 6 months, where breast-feeding for 3 months or more was associated with reduced risk of hospitalization for infections, there was no association between breast-feeding and hospital admission for accidents, suggesting adequate control for social position.
Our study also had inevitable limitations. We were not able to account for admissions to private hospitals in this analysis. Admissions to private hospitals are most likely to occur among children from more socioeconomically advantaged families, possibly resulting in an overestimation of the protective effect of breast-feeding on infectious disease hospitalization. However, we adjusted for indicators of private hospital use (birth in a private hospital and household income), as well as other markers of socioeconomic position.
In our data, protection against infectious disease from breast-feeding was dose-dependent, with longer durations of exclusive breast-feeding conferring greater benefits.38 The low rate of breast-feeding beyond 6 months may not be suitable for studying the effect of long-term breast-feeding; however, it is ideal for studying the long-term effects of breast-feeding during early infancy, a critical period of development.39–41
As with all observational research, we are unable to rule out a reverse causality bias in that early hospitalization may have resulted in a shorter duration of breast-feeding. However, we did exclude admissions during the first 8 days of life. Furthermore, given that we categorized infant feeding based on the first 3 months, reverse causality would be a concern only for admission during these early months; among children admitted, only 12% were admitted during the first 3 months.
In addition, we do not have data on when weaning foods and nonmilk liquids were introduced. Therefore, we are not able to simultaneously examine the association of complementary feeding and hospitalization, nor can we consider the infants categorized as “breast milk only” to be “exclusively breast-fed” according to the WHO definitions.42
Finally, hospital admission data may not be sensitive enough to capture a protective effect of breast-feeding against infectious disease in older children. Health services utilization data (ie, outpatient and doctor visits) may be more representative of the overall burden of infectious disease in older children. Hospitalizations, however, reflect more severe diseases, with substantially greater costs than outpatient care. Even in this population with a high hospital admission rate, there was no evidence of protective effect of breast-feeding after infancy up to 8 years of age. This conclusion is consistent with a cohort study conducted in the United Kingdom, which found that the protective effect of breast-feeding in infants wears off soon after the discontinuation of breast-feeding.35 Whether there might be beneficial effects of breast-feeding on children's health in later years is a question that remains unanswered.
The reduction in infectious-disease hospitalizations in infants observed in this study is consistent with a biologic effect of breast milk directly stimulating the recipient's immune system via the transfer of anti-idiotypic antibodies and lymphocytes, resulting in a better-functioning immune system.43 In formula-fed infants, infection can also result from contamination of the formula itself or of the bottles and plastic nipples used to deliver the formula.44 In older children, it has been hypothesized that an improvement in immune function with early breast-feeding could produce long-term protection against infections and allergies.45 Results from this study, however, indicate that breast-feeding does not reduce the risk of hospitalization from infectious disease beyond infancy, suggesting that breast-feeding provides no long-term benefits in the development of the immune system.
We are indebted to Vivian Wong at the Hospital Authority for her advice and help. We thank Connie Hui for her assistance with the record linkage. We also thank colleagues at the Hospital Authority, Student Health Service and Family Health Service of the Department of Health for their assistance and collaboration. Finally, we express our heartfelt appreciation to all the infants and parents who have participated and continue to participate in the 1997 Hong Kong “Children of 1997” birth cohort.
1.Hospital Authority. Statistical report, 2000–2001. Hong Kong SAR: Hospital Authority; 2002.
2.American Academy of Pediatrics. Breastfeeding and the use of human milk. Pediatrics.
3.Bachrach VR, Schwarz E, Bachrach LR. Breastfeeding and the risk of hospitalization for respiratory disease in infancy: a meta-analysis. Arch Pediatr Adolesc Med
4.WHO Collaborative Study Team on the Role of Breastfeeding on the Prevention of Infant Mortality. Effect of breastfeeding on infant and child mortality due to infectious diseases in less developed countries: a pooled analysis. Lancet.
5.Cesar JA, Victora CG, Barros FC, Santos IS, Flores JA. Impact of breast feeding on admission for pneumonia during postneonatal period in Brazil: nested case-control study. BMJ
6.Ford K, Labbok M. Breast-feeding and child health in the United States. J Biosoc Sci
7.Levine OS, Farley M, Harrison LH, Lefkowitz L, McGeer A, Schwartz B. Risk factors for invasive pneumococcal disease in children: a population-based case-control study in North America. Pediatrics
8.Leung GM, Lam TH, Ho LM, Lau YL. Health consequences of breast-feeding: doctors' visits and hospitalizations during the first 18 months of life in Hong Kong Chinese infants. Epidemiology
9.Clemens JD, Rao MR, Chakraborty J, et al. Breastfeeding and the risk of life-threatening enterotoxigenic Escherichia coli
diarrhea in Bangladeshi infants and children. Pediatrics
10.Sachdev HP, Kumar S, Singh KK, Puri RK. Does breastfeeding influence mortality in children hospitalized with diarrhoea? J Trop Pediatr
11.Scariati PD, Grummer-Strawn LM, Fein SB. A longitudinal analysis of infant morbidity and the extent of breastfeeding in the United States. Pediatrics
12.Wright AL, Bauer M, Naylor A, Sutcliffe E, Clark L. Increasing breastfeeding rates to reduce infant illness at the community level. Pediatrics
13.Duffy LC, Faden H, Wasielewski R, Wolf J, Krystofik D. Exclusive breastfeeding protects against bacterial colonization and day care exposure to otitis media. Pediatrics
14.Owen MJ, Baldwin CD, Swank PR, Pannu AK, Johnson DL, Howie VM. Relation of infant feeding practices, cigarette smoke exposure, and group child care to the onset and duration of otitis media with effusion in the first two years of life. J Pediatr
15.Pukander J, Luotonen J, Timonen M, Karma P. Risk factors affecting the occurrence of acute otitis media among 2–3-year-old urban children. Acta Otolaryngol
16.Saarinen UM. Prolonged breast feeding as prophylaxis for recurrent otitis media. Acta Paediatr Scand
17.Sassen ML, Brand R, Grote JJ. Breast-feeding and acute otitis media. Am J Otolaryngol
18.Wilson AC, Forsyth JS, Greene SA, Irvine L, Hau C, Howie PW. Relation of infant diet to childhood health: seven year follow up of cohort of children in Dundee infant feeding study. BMJ
19.Silfverdal SA, Bodin L, Hugosson S, et al. Protective effect of breastfeeding on invasive Haemophilus influenzae
infection: a case-control study in Swedish preschool children. Int J Epidemiol
20.Silfverdal SA, Bodin L, Olcen P. Protective effect of breastfeeding: an ecologic study of Haemophilus influenzae
meningitis and breastfeeding in a Swedish population. Int J Epidemiol
21.Broor S, Pandey RM, Ghosh M, et al. Risk factors for severe acute lower respiratory tract infection in under-five children. Indian Pediatr
22.Mahalanabis D, Alam AN, Rahman N, Hasnat A. Prognostic indicators and risk factors for increased duration of acute diarrhoea and for persistent diarrhoea in children. Int J Epidemiol
23.Teka T, Faruque AS, Fuchs GJ. Risk factors for deaths in under-age-five children attending a diarrhoea treatment centre. Acta Paediatr
24.Hanson LA, Korotkova M, Lundin S, et al. The transfer of immunity from mother to child. Ann N Y Acad Sci
25.Wright CM, Parkinson K, Scott J. Breast-feeding in a UK urban context: who breast-feeds, for how long and does it matter? Public Health Nutr
26.Kwok MK, Schooling CM, Lam TH, Leung GM. Does breastfeeding protect against childhood overweight? Hong Kong's “Children of 1997” birth cohort. Int J Epidemiol
27.Chiu SS, Lau YL, Chan KH, Wong WH, Peiris JS. Influenza-related hospitalizations among children in Hong Kong. N Engl J Med
28.Lam TH, Leung GM, Ho LM. The effects of environmental tobacco smoke on health services utilization in the first eighteen months of life. Pediatrics
29.Hui LL, Schooling CM, Wong MY, Ho LM, Lam TH, Leung GM. Infant growth during the first year of life and subsequent hospitalization to 8 years of age. Epidemiology
30.Leung GM, Ho LM, Tin KY, Schooling CM, Lam TH. Health care consequences of cesarean birth during the first 18 months of life. Epidemiology
31.Rothman KJ, Greenland S, Lash TL. Modern Epidemiology.
3rd ed. Philadelphia: Lippincott Williams & Wilkins; 2008.
32.Williams RL, Creasy RK, Cunningham GC, Hawes WE, Norris FD, Tashiro M. Fetal growth and perinatal viability in California. Obstet Gynecol
33.Moss TJ. Respiratory consequences of preterm birth. Clin Exp Pharmacol Physiol
34.Hofman PL, Regan F, Harris M, Robinson E, Jackson W, Cutfield WS. The metabolic consequences of prematurity. Growth Horm IGF Res
. 2004;14(suppl A):S136–S139.
35.Quigley MA, Kelly YJ, Sacker A. Breastfeeding and hospitalization for diarrheal and respiratory infection in the United Kingdom Millennium Cohort Study. Pediatrics
36.Oddy WH, Kickett-Tucker C, De Maio J, et al. The association of infant feeding with parent-reported infections and hospitalisations in the West Australian Aboriginal Child Health Survey. Aust N Z J Public Health
37.Hetzner NM, Razza RA, Malone LM, Brooks-Gunn J. Associations among feeding behaviors during infancy and child illness at two years. Matern Child Health J
38.Kramer MS, Kakuma R. The optimal duration of exclusive breastfeeding: a systematic review. Adv Exp Med Biol
39.Gluckman PD, Hanson MA. Living with the past: evolution, development, and patterns of disease. Science
40.Wells JC. Historical cohort studies and the early origins of disease hypothesis: making sense of the evidence. Proc Nutr Soc
41.Singhal A, Lucas A. Early origins of cardiovascular disease: is there a unifying hypothesis? Lancet
43.Hanson LA. Human milk and host defence: immediate and long-term effects. Acta Paediatr
44.Riordan J. Breastfeeding and Human Lactation.
3rd ed. Sudbury, MA: Jones and Bartlett; 2005.
45.Hanson LA. Breastfeeding provides passive and likely long-lasting active immunity. Ann Allergy Asthma Immunol
Supplemental Digital Content
© 2010 Lippincott Williams & Wilkins, Inc.