Article Text

other Versions

Protective effect of exclusive breastfeeding against infections during infancy: a prospective study
  1. Fani Ladomenou1,
  2. Joanna Moschandreas2,
  3. Anthony Kafatos2,
  4. Yiannis Tselentis3,
  5. Emmanouil Galanakis1
  1. 1Department of Paediatrics, University of Crete, Heraklion, Greece
  2. 2Department of Social Medicine, University of Crete, Heraklion, Greece
  3. 3Department of Clinical Bacteriology, Parasitology, Zoonoses and Geographical Medicine, University of Crete, Heraklion, Greece
  1. Correspondence to Professor Emmanouil Galanakis, Department of Paediatrics, University of Crete, POB 2208, Heraklion 710 03, Greece; emmgalan{at}


Objective To prospectively investigate the effects of breastfeeding on the frequency and severity of infections in a well-defined infant population with adequate vaccination coverage and healthcare standards.

Study design In a representative sample of 926 infants, successfully followed up for 12 months, feeding mode and all infectious episodes, including acute otitis media (AOM), acute respiratory infection (ARI), gastroenteritis, urinary tract infection, conjunctivitis and thrush, were recorded at 1, 3, 6, 9 and 12 months of life.

Results Infants exclusively breastfed for 6 months, as per WHO recommendations, presented with fewer infectious episodes than their partially breastfed or non-breastfed peers and this protective effect persisted after adjustment for potential confounders for ARI (OR 0.58, 95% CI 0.36 to 0.92), AOM (OR 0.37, 95% CI 0.13 to 1.05) and thrush (OR 0.14, 95% CI 0.02 to 1.02). Prolonged exclusive breastfeeding was associated with fewer infectious episodes (rs=−0.07, p=0.019) and fewer admissions to hospital for infection (rs=−0.06, p=0.037) in the first year of life. Partial breastfeeding was not related to protective effect. Several confounding factors, including parental age and education, ethnicity, presence of other siblings, environmental tobacco smoke exposure and season of birth were demonstrated to have an effect on frequency of infections during infancy.

Conclusions Findings from this large-scale prospective study in a well-defined infant population with adequate healthcare standards suggest that exclusive breastfeeding contributes to protection against common infections during infancy regarding and lessens the frequency and severity of infectious episodes. Partial breastfeeding did not seem to provide this protective effect.

Statistics from

Human milk is the most suitable food for newborn and young infants and exclusive breastfeeding for at least the first 6 months of life is recommended.1 The many advantages of breastfeeding have been documented, and include financial, psychosocial and developmental benefits and the prevention of illnesses, including infections.2,,6 Given the immaturity of the infantile immune system, this protection is of paramount importance. Although it has long been shown that breastfed infants are less prone to a variety of infections, some have asserted that differences in breastfeeding and formula-feeding mothers may alter infants' risk of infection and that the protective effect may be attributable to confounding factors and may be less marked in societies with high health standards.7 8 In addition, recent studies have often focussed on particular infections or on the effect of breastfeeding during the first months, with only a few investigating the whole spectrum of infectious episodes and the protective effect of breastfeeding throughout infancy.2 9,,12 In addition, more attention has been paid to the effects of breastfeeding on the frequency rather than the severity of infectious episodes.

What is already known on this topic

Breastfed infants are considered to be less prone to a variety of infections, including acute otitis media, and gastrointestinal and lower respiratory tract infections.

What this study adds

  • ▶. Exclusive breastfeeding helps protect infants against common infections and lessens the frequency and severity of infectious episodes not only in developing countries but also in communities with adequate vaccination coverage and healthcare standards.

  • ▶. Partial breastfeeding does not seem to provide this protective effect against infections in infancy

In this study we prospectively investigated the effect of breastfeeding on both the frequency, expressed as number of infectious episodes, and the severity, expressed as need for doctor visits and hospitalisation for common infections throughout the first year of life, in a well-defined, well-vaccinated infant population with adequate health standards. Based on existing experience in the field, the study was designed first to verify the hypothesis that breastfeeding protects against common infantile infections, second to explore the impact of exclusive breastfeeding and its duration, and third to investigate this impact during both the first and the second 6 months of infancy (months 1–6 and months 6–12).


Design, setting and study population

The study was conducted as a prospective observational study in a fixed cohort in Crete, an island of 601 159 inhabitants (2001 census data). From a total of 6878 births in 2004, a representative cohort of 1049 (15.2%) mother–infant pairs was recruited, as described elsewhere.13 14 Sampling was based on consecutive births on random days from October to December 2004 and from April to July 2005. This sampling during two periods facilitated assessment of the effects of breastfeeding on infections in relation to season of birth. During the study period, infants were routinely immunised with diphtheria-tetanus-acellular pertussis (three doses), poliomyelitis (inactivated poliovirus vaccine, two or three doses), Haemophilus influenzae type b (three doses), hepatitis B (three doses), Neisseria meningitidis group C conjugate (three doses) and 7-valent Streptococcus pneumoniae conjugate (three doses) vaccines, which made them a well-vaccinated population. Infants were mainly cared for by family paediatricians, and for out-of-hour services, general hospitals with emergency paediatric facilities were easily available. Vaccinations were financed by state social security funds, resulting in high immunisation rates during infancy, although lower rates have been observed in older children and adolescents.15 Day care attendance was not common during early infancy.


All mothers were interviewed by the same investigator (FL) while in the maternity ward. Special emphasis was given to asking simple questions and to facilitating the participation of mothers with low literacy levels.13 14 Mothers received detailed information about the study and their right to abstain at any time. All agreed to participate except for one immigrant mother, who responded to the interview but had no phone access for further contact. The questionnaire at the maternity visit consisted of 19 multiple choice and 14 open ended questions regarding information on family composition, sources of medical care, ethnicity and years of education completed by mother and father, mother and infant health during pregnancy and delivery and breastfeeding experience and intention. Subsequently, mothers were contacted by phone by the same investigator at months 1, 3, 6, 9 and 12 postpartum. The follow-up questionnaires consisted mostly of multiple choice items regarding information on breastfeeding, immunisation coverage, doctor visits and hospitalisations and all episodes of infant illness since the previous interview. The Research Committee, Medical Faculty of the University of Crete approved the study protocol.


Following previously published reports, exclusive breastfeeding was defined as an infant receiving only breast milk and no other liquids or solids, and partial breastfeeding as an infant receiving breast milk in combination with infant formula or other liquids or solids.1 13 14 16 Our definitions focused on the breast milk intake of infants and did not specify types of other food or liquids. However, in the study cohort, infants as a rule received milk as their sole source of nutrition and were not given water based liquids such as sugar waters or diluted fruit juices; hence, exclusive breastfeeding was generally identical to full breastfeeding. The majority of infants were weaned at the age of 6 months. Some immigrant mothers might introduce solids before the age of 4 months in their country of origin, but we assume that they followed their physician's advice in the study community. Environmental tobacco smoke exposure was defined as parental smoking during the first 12 months of life, and parents who reported smoking in all questionnaires were defined as regular smokers.14 The diagnostic criteria for infections are given in table 18 17 18. In children with multiple episodes of the same illness, a symptom-free week was required before a new episode was considered to have occurred. Hospital admissions were classified according to discharge diagnosis, as per parental report. The structure of the questionnaires did not allow for reliable additional information on severity of infectious episodes. Episodes of acute otitis media (AOM), acute respiratory infections (ARI), gastroenteritis, urinary tract infections (UTI), conjunctivitis and thrush were collectively classified as common infantile infections.

Table 1

Definitions of infectious episodes in infancy

Statistical analysis

Duration of breastfeeding was expressed in weeks. Episodes of infections were defined and calculated as described above and an index of all episodes was created. Frequency of infections was estimated by the total number of episodes. Severity of infections was estimated by the number of doctor visits and admissions to hospital. For quantitative variables, such as weeks of breastfeeding and number of infectious episodes, the Mann–Whitney U test and Spearman's rank correlation coefficient rs were used. For categorical variables, such as exclusive breastfeeding for 6 months or lack of breastfeeding, Fisher's exact test, OR and 95% CI were calculated. 95% CIs for differences in proportions were obtained using the normal approximation to the binomial distribution. Multiple generalised linear models were fit in order to evaluate the effects of exclusive breastfeeding on frequency of infections while adjusting for potential confounders. Initially a Poisson regression approach was used with robust standard errors. Standardised residual plots indicated that the Poisson model was appropriate for assessing the effects of breastfeeding on the total number of infections. The significance of the model was assessed using the χ2 likelihood ratio statistic. Multiple logistic regression analysis was subsequently performed using AOM, gastroenteritis, ARI, thrush and hospital admissions as binary (yes/no) response variables. Potential confounders evaluated were ethnic origin, average parental age (years), average years of parental education, birth weight, sex, season of birth (autumn/spring), gestation duration, delivery mode, parental smoking (cigarettes per day) and number of siblings. The significance level was set at 5% (α=0.05). The statistical package SPSS V.15 was used.


Follow-up and breastfeeding rates

Of the initial sample of 1049 mother–infant pairs, 1027 (97.9%), 996 (94.9%), 984 (93.8%), 915 (87.2%) and 926 (88.2%) were available for follow-up at months 1, 3, 6, 9 and 12 postpartum, respectively. The principal reason for mothers being lost to follow-up was inability to contact them again by phone; as these mothers were lost at different time points, reliable information on breastfeeding rates and infectious episodes was not available in this group. Breastfeeding rates for the total cohort were 60.7%, 35.3%, 16.7% and 4.48% (including 24.6%, 17.2%, 10.2% and 3.49% of exclusive breastfeeding) at months 1, 3, 6 and 12, respectively. Complementary foods were given to all infants after 6 months. There were no infant deaths during the 12-month follow-up period. Among the families of the final cohort, 14 had twins, and one had triplets. In only one family was the mother single. Immigrant mothers, comprising 13.1% of the study population, tended to breastfeed more than Greek-born mothers with 21.5% vs 8.0% exclusively breastfeeding at 6 months, respectively, a difference of 13.5% (95% CI 6.0% to 21.1%, p<0.0001).

Common infections during infancy

Episodes of common infections and hospital admissions for these infections in the 926 infants successfully followed up for 12 months are given in table 2. ARI, AOM, gastroenteritis, UTI, thrush and conjunctivitis accounted for 3533 (88.7%) of the 3984 infectious episodes reported during infancy. The remaining infections consisted mainly of presumed febrile viral infection without a focus, exanthema subitum and skin infection (171, 38 and 11 episodes, respectively). During the 12 first months of life, 162 of the 926 infants (17.5%) were hospitalised for various reasons (a total of 212 admissions); among these infants, 134 (14.5%) were hospitalised for infection (a total of 175 admissions).

Table 2

Infectious episodes and hospital admissions due to infection during the first year of life in 926 infants

Any breastfeeding and infections in the first year of life

Duration of exclusive breastfeeding (expressed in weeks) was statistically significant but was only very weakly negatively correlated with fewer infectious episodes (Spearman's rs=−0.07, p=0.019) and fewer admissions to hospital for infection (Spearman's rs=−0.059, p=0.037) in the first year of life. Partial breastfeeding was not found to be related to the number of infectious episodes or hospitalisations due to infection.

Exclusive breastfeeding and infections

Infants exclusively breastfed for at least 6 months (26 weeks, n=91) were less prone than their peers with partial or no breastfeeding (n=835) to present with infectious episodes and to be hospitalised for infection. For ARI, AOM and thrush, this protective effect persisted after adjustment for potential confounders (table 3). Furthermore, when compared to their non-breastfed peers (n=146), infants exclusively breastfed for at least 6 months presented with fewer total infectious episodes during the first 12 months of life (3.4 vs 4.4, p=0.001) and required less doctor visits for AOM (0.3 vs 0.6, p=0.011) and ARI (1.1 vs 1.6, p=0.002) during the first 12 months of life, and for gastroenteritis (0.04 vs 0.09, p=0.065) during the first 6 months. Exclusive breastfeeding seemed to protect against hospital admissions for infection (OR 0.980, 95% CI 0.961 to 0.999; p=0.036) and thrush (OR 0.973, 95% CI 0.951 to 0.996; p=0.022). For hospitalisations for any infection, statistically significant confounding factors were birth period (OR 0.53, 95% CI 0.33 to 0.85, with a lower risk for babies born in the autumn and winter) and mean years of parental education (OR 0.91, 95% CI 0.84 to 0.98). These factors were also statistically significant in relation to hospitalisation for any reason, with singletons additionally having a lower risk of hospitalisation than twins or triplets (OR 0.16, 95% CI 0.03 to 0.76).

Table 3

Infections during the first 6 months in infants exclusively breastfed throughout the 6 months (EBF, n=91) compared to those with partial or no breastfeeding (non-EBF, n=835)

The Poisson regression model assessing the effect of breastfeeding and other factors on the total number of infections was found to be statistically significant (likelihood ratio χ2 statistic 159.2 on 11 df, p<0.0001). Factors found with Poisson regression analysis to have a protective effect on the total number of infections in the first year of life were exclusive breastfeeding (p<0.005), non-Greek ethnicity (p<0.001), advanced maternal age (p<0.0001) and no other siblings in the family (p<0.0001). Breastfed infants were found to have, on average, 0.7 fewer infections in their first year of life than their non-exclusively breastfed counterparts (p=0.005), after adjusting for potential confounders.


The results from this prospective study in a well-defined infant population suggest a potential protective effect of breastfeeding against common infections during the first year of life and further indicate that this protection is due to exclusive breastfeeding and its duration: the longer the exclusive breastfeeding, the fewer the number of infectious episodes, physician consultations and hospitalisations during infancy. Thus, compared to their non-breastfed peers, infants exclusively breastfed for at least 6 months, in accordance with WHO recommendations,1 were shown to have fewer common infections and less severe episodes resulting in fewer doctor visits and less hospitalisations.

Some studies have already suggested that breastfeeding has a protective effect against ARI,2 5,,10 AOM3 11 19 20 and gastroenteritis,2,,5 9 11 and to a lesser degree against UTI6 21 22 and thrush.23 Several explanations have been proposed for this protection, including maternal antibodies, nutritional, immunoregulatory and immunomodulatory factors in maternal milk, and the effect of maternal milk on infantile indigenous flora.24,,26 Bottle feeding may result in increased exposure to pathogens and, in addition, bottle-fed infants are held in a different position, which might affect Eustachian tube function and hence susceptibility to AOM.27 Our findings support the hypothesis of diverse mechanisms, as protection against different specific infections was shown during the first and second halves of infancy. Regarding thrush, our findings suggest that factors in maternal milk provide protection against Candida colonisation in the oral cavities of breastfed infants,28 despite the existence of C albicans on the mother's breasts. Moreover, our findings support previous observations29 30 suggesting different oral carriage of Candida species in exclusively and partially breastfed infants, with the latter being more prone to Candida oral carriage.

Our findings suggest that maximum protection is provided by exclusive breastfeeding throughout the first 6 months of life. A shorter duration of breastfeeding was less protective and partial breastfeeding had no substantial protective effect. Therefore, in studies which do not analyse exclusive and partial breastfeeding separately, any protection observed for overall breastfeeding might be due to the subpopulation of infants exclusively breastfed. Our findings on exclusive versus partial breastfeeding support the hypothesis of either an immunomodulatory effect hampered by the introduction of formula feeding, or a threshold level for the passive immunity conferred to the infant by secretory immunoglobulin A and other protective complexes in breast milk.31

Based on the experience of previous studies, we tried to avoid pitfalls and followed the methodology recommended by Bauchner et al,7 focusing on avoidance of detection bias, clear definitions of the outcome event and of breastfeeding, and adjustment for potential confounding variables. By using systematic sampling and carefully designed questionnaires, we sought to minimise potential reporting or observer bias. Nevertheless, our study shared the limitations of interview-based investigations. The study relied on maternal recall alone, which precluded precise estimation of the severity of infectious episodes; furthermore, breastfeeding mothers might be more prone to report illnesses because of the closer link with their infant. However, this detection bias would result in underestimation rather than overestimation of the protective effects of breastfeeding. Our definition of diarrhoea might lack specificity, as breastfed infants commonly have loose stools and as, on the other hand, liquids or food increase the risk of diarrhoea.32 Confounding factors have already been investigated for susceptibility to ARI, AOM and thrush,19,,22 33 34 although the effect of such factors is inconsistent in different studies. In our study, analysis of confounding factors often concluded that the protection initially attributed to breastfeeding was actually the result of intervening variables. As we have already reported, exclusive breastfeeding seemed to significantly protect against total infectious episodes in infants with environmental tobacco smoke exposure.14 These findings point to the complexity of nutritional, environmental and host and pathogen factors that do or do not result in infection, but do not eliminate the protective value of breastfeeding regarding the frequency and severity of infection, as documented by reduced infectious episodes, doctor visits and hospitalisations.

Despite limitations, our results suggest breastfeeding has beneficial effects against common infections in infancy. This protection seems to be related to the exclusive breastfeeding and its duration. Therefore, mothers should be advised by health professionals that, in addition to all the other benefits, exclusive breastfeeding helps prevent infections in babies and lessens the frequency and severity of infectious episodes.

View Abstract


  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the Research Committee, Medical Faculty, University of Crete, Greece.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.