What Is Known
- Breast-feeding from birth may protect against infections, but optimal duration of breast-feeding is unclear.
- Delaying introduction of complementary foods may reduce the risk of infections, but may be more important for formula than for solid foods.
- Breast-feeding is associated with a healthy lifestyle in high-income countries, making interpretation of breast-feeding studies difficult.
What Is New
- Breast-feeding for ≤6 compared to ≥12 months was associated with higher risk of infections, but with similar risks for 6 to 11 and ≥12 months’ duration.
- Breast-fed children who received complementary foods at 4 to 6 months of age had similar risk for infection as those receiving complementary foods after 6 months.
There is an ongoing controversy regarding optimal breast-feeding duration and time of weaning (1). The recommendation from the World Health Organization (WHO) (2) to continue full breast-feeding until 6 months of age has been adopted by several industrialized countries (1,3–5). Still, only 1% to 13% of mothers in the UK, the US, and Norway comply with this recommendation (4,6,7). The WHO furthermore recommends to continue partial breast-feeding up to 2 years or beyond (3). There is a need for more studies providing stronger evidence to underpin a revision of current guidelines, as stated by the WHO expert consultation (8).
Protection against infections is one of several causes considered in the breast-feeding recommendations (9,10). These benefits appear particularly in developing countries with a high burden of infectious diseases in infancy (11–13), where the risk of unsafe formula feeding adds to a higher pressure of infections (8). These observations have partly been reproduced in industrialized countries (6,14).
Breast-feeding in wealthy societies is associated with health conscious behavior (15), and consequently, confounding effects of other healthy choices needs to be thoroughly investigated. Furthermore, previous studies have included few infants fully breast-fed for 6 months. Underlying characteristics of this group could possibly confound the results, in addition to uncertainty in the estimates due to limited sample sizes (6,14,16,17). Moreover, few studies have addressed the optimal duration of breast-feeding beyond 6 months regarding infection prevention (6,14).
The main objective of the present study was to assess whether the risk of hospitalization for infections before the age of 18 months was different in infants introduced to complementary foods from 4 to 6 months compared to the recommended 6 months. Secondary objectives were to study whether duration of any breast-feeding was associated with hospitalization and with infections in general. In addition to a traditional cohort analysis, we assessed whether breast-feeding predicted risk of infections in a matched sibling design.
The Norwegian Mother and Child Cohort Study (MoBa) is a population-based pregnancy cohort study conducted by the Norwegian Institute of Public Health (18). Participants were recruited from all over Norway (rural/urban) from 1999 to 2008, and 95,200 mothers (40.6% of eligible) participated with 1 or more pregnancies. Written informed consent was obtained. The MoBa study was approved by The Regional Committee for Medical Research Ethics in South-Eastern Norway. The present study is based on version VII of the quality-assured data files released for research in June 2012.
Follow-up is conducted by mailed questionnaires. We use information from the baseline questionnaire completed around week 17 of pregnancy and 2 follow-up questionnaires at infant age 6 and 18 months (available at www.fhi.no/moba).
Only singletons whose parents returned questionnaires at 6 and 18 months of age (n = 70,511, flow chart Fig. 1) were included (Supplemental Digital Content 1, Table, http://links.lww.com/MPG/A903, for characteristics of included and nonincluded participants). Children with missing data for any of the study variables were excluded from the analysis.
In order to reduce bias that may occur due to differences in self-reporting of infections and confounding maternal effects, siblings in the cohort were studied in a secondary analysis.
Main Exposure: Infant Feeding
In the 6 months’ questionnaire, ongoing breast-feeding or formula feeding from the first week in monthly intervals and age at introduction of solids was specified up until completion (median 27 weeks). From 6 to 18 months of age, the mothers reported whether they were breast-feeding in 4 intervals (6–8, 9–11, 12–14, and 15–18 months). The median value of each interval was used as a fixed value in calculations of breast-feeding duration. For the regression analyses, duration of any breast-feeding was categorized into 6 intervals (no breast-feeding, 0.1–3, 4–6, 6.1–8, 9–11, and ≥12 months). Full breast-feeding was defined as breast-feeding from birth without any formula or solids. This allows for water and vitamins, different from the strict definition of exclusive breast-feeding.
To assess whether partial breast-feeding was associated with different risk of infections than full breast-feeding, age at introduction of complementary foods (formula or solid foods) in infants breast-fed >6 months was analyzed in categories <1, 1 to 3, and 4 to 5 months with full breast-feeding for 6 months as the reference category. We performed subanalyses for the age at introduction of formula or solids, all adjusted for the duration of any breast-feeding.
Main Outcome Definitions: Infections
The primary outcome was hospitalization due to infection by maternal report from 0 to 18 months. In additional analyses hospitalization was separated in age 0 to 6 and 6 to 18 months. We assessed the validity of this outcome from the questionnaires by comparing it to hospital statistics in a subsample from one major hospital, in which 197 of 212 admissions were correctly reported in questionnaires (sensitivity 93%).
Secondary outcomes were infections regardless of hospitalization from 0 to 18 months indicated from a specified list. Lower respiratory tract infections (LRTIs = pneumonia/bronchiolitis), gastroenteritis (GE), and acute otitis media were dichotomized (none vs any). Upper respiratory tract infections (URTI) and frequent infections (total number of the 4 types of infections) were dichotomized corresponding to the upper quintile of the distribution in the whole sample (URTI >8 or “frequent infections” >10, respectively).
Information regarding maternal age, birth weight, gestational age, parity, mode of delivery, and sex was obtained from the Medical Birth Registry. The first questionnaire provided information on smoking during pregnancy, paternal smoking, and duration of maternal education in 4 categories. Duration of child day care outside home was categorized into 4 groups (Supplemental Digital Content 1, Table, http://links.lww.com/MPG/A903, for covariates and categorization).
Analyses were performed using the SPSS 20.0 statistical software package (IBM SPSS Inc, Chicago, IL) and Stata version 13 (StataCorp LP, College Station, TX). The main analysis was conducted using binary regression models to obtain relative risks (RRs), with robust variance to correct for potential clustering within families. Confounders reported from previous studies were preselected. Other covariates associated with any of the outcomes with 95% confidence intervals for RR not including 1.00 in unadjusted analyses were assessed for inclusion in the multiple regression models, but removed from the final model if the estimates changed <10%.
As sensitivity analyses to assess robustness of findings, we explored the potential effect of using different cut-offs for the total number of URTI and total infections, and fitted Poisson regression models for number of infectious episodes. Sensitivity analyses were consistent with our main analyses (data not shown).
Finally, we estimated the odds ratio for infections in siblings (matched sets conditioned on the maternal origin), using conditional logistic regression models. In this model, covariates that are identical within sibling sets will not contribute to the final estimates. Thus, only sibling sets with different duration of breast-feeding can potentially change the odds ratio from 1. We included as covariates potential confounders that could differ among siblings, comparing an unmatched design to the sibling analysis.
The mean duration of any breast-feeding was 10.0 months (SD 4.5), 80.8% were breast-fed >6 months. Fourteen percent were fully breast-fed for 6 months; 0.9% was never breast-fed. Longer duration of any breast-feeding was strongly associated with higher maternal age and parity, education, nonsmoking status, vaginal delivery, and day care at home (all P < 0.001; Supplemental Digital Content 2, Table, http://links.lww.com/MPG/A903).
Before the age of 18 months, 8.0% were admitted to hospital for infections. GE or LRTI accounted for 68% (Supplemental Digital Content 3, Fig., http://links.lww.com/MPG/A903, which demonstrates that the age at admission was evenly distributed during the period from 0 to 18 months). The frequency of all infections increased gradually with age (Supplemental Digital Content 4, Table, http://links.lww.com/MPG/A903).
Age at Introduction of Complementary Foods and Infections
Age at introduction of complementary foods was a significant predictor for hospitalization for infections among infants who were breast-fed for a minimum of 6 months (adjusted RR 0.98 per month delay, 95% confidence interval 0.97–1.00, P for trend 0.01). This association was driven by a higher risk for hospitalization in those with introduction of complementary foods at <4 months of age (Fig. 2). In categorical analyses, hospitalization for infections was reported in 7.7% of infants fully breast-fed for 6 months compared to 7.3% of those introduced to complementary foods at 4 to 6 months (aRR 0.95, 0.88–1.03, Fig. 2, Supplemental Digital Content 5, Table, http://links.lww.com/MPG/A903).
The risk was significantly lower when formula introduction was delayed (aRR 0.96 per month, 0.93–0.98), but formula introduction from 4 months was not associated with increased risk of hospitalization. For solids introduction, we found a nonsignificant trend in the opposite direction (aRR 1.03 per month, 1.00–1.07, Fig. 2).
We also performed subanalyses separating the main outcome in time periods of 0 to 6 months and 6 to 18 months. The point estimates did not change substantially, but the confidence intervals became wider due to the lower proportion with events (Supplemental Digital Content 5, Table, http://links.lww.com/MPG/A903).
When we analyzed all infections regardless of hospitalization, introduction of complementary foods <4 months was associated with an increased risk for GE, URTI, and frequent infections but not for LRTI (Supplemental Digital Content 5, Table, http://links.lww.com/MPG/A903).
Duration of Any Breast-feeding and Infections
Compared with infants breast-fed ≥12 months, infants breast-fed for <6 months had an increased risk for hospitalization. The association was primarily driven by GE (Fig. 3A). Breast-feeding <6 months was in general associated with an increased risk for infectious outcomes regardless of hospitalization (Fig. 3B).
Infants breast-fed for 6 to 11 months did not differ significantly from those breast-fed ≥12 months regarding hospitalization (Fig. 3A), but with a slightly reduced risk of GE, otitis media, and frequent infections if breast-fed up to 9 months (Fig. 3B, Supplemental Digital Content 6, Table, http://links.lww.com/MPG/A903).
The median difference in duration of full and any breast-feeding within sibling sets discordant for duration of breast-feeding was 1.5 and 3.0 months, respectively, and 41% had identical duration of any breast-feeding. The first-born sibling had a mean shorter duration of both full and any breast-feeding of 0.4 months. Infants with a matched sibling control with difference in duration of full breast-feeding (n = 12,495) or any breast-feeding (n = 11,149) contributed in the sibling analysis.
In the conditional logistic regression analysis of siblings we found no significant associations for age for introduction of complementary foods with hospital admission for infections, and with risk for infections regardless of hospitalization (Table 1). Similarly, duration of any breast-feeding was not a significant predictor (Table 2). The odds ratios per month were generally closer to 1 in the conditional analysis compared to the unconditional full cohort analysis (Tables 1 and 2).
Significant predictors for infections in the sibling model were birth order, birth weight <2500 g, male sex, and day care outside home (data not shown).
In this large birth cohort, we found a small but significant trend for lower risk for hospitalizations for infections for each month of delaying complementary foods in breast-fed infants. Breast-fed children who, however, received complementary foods at 4 to 6 months of age had a similar risk as those delaying complementary foods to after 6 months, suggesting a threshold effect for duration of full breast-feeding. Breast-feeding for ≤6 months compared to ≥12 months was associated with an increased risk, whereas infants breast-fed for 6 to 11 months had risks of hospitalization similar to those breast-fed ≥12 months. The estimated associations were weaker and nonsignificant in the sibling analysis compared with the full cohort analysis. The present study is the first in this field to include a discordant sibling analysis, an analysis that studies differences within sibling sets and thereby reduces the potential confounding from shared maternal factors.
The similar hospitalization risk for infants introduced to complementary foods from 4 to 6 compared to ≥6 months parallels 2 previous cohort studies (6,17) and the recent randomized controlled EAT trial (19). In contrast, others have found a reduced risk of lower respiratory tract infections with 6 months of full breast-feeding (16,20). It should be noted that the sample sizes were smaller in these studies, and additionally, fewer confounding factors could be assessed.
In our study, infants introduced to formula in addition to breast milk before 4 months had increased risk for infections, which was not found for later formula introduction or for solid foods. The difference between formula and solids as regards infection risk is in accordance with observations from the UK Millennium Cohort Study (6). In the EAT trial, the main difference between the 2 intervention arms was early introduction of solids, with no difference found for infections except for URTI (19). We speculate that the observed difference in our study could be due to a relatively higher effect on the volume of breast milk given when formula was introduced at an early age, but our data do not provide the granularity to study this in more detail.
The strongest association observed in the full cohort analysis was the increased risk of hospitalization for GE in infants not breast-fed or breast-fed for ≤6 months compared to ≥12 months. Observational studies elsewhere have found lower risk for GE early in life among breast-fed infants (17,21), also supported by the randomized trial from the Belarus (22).
For lower respiratory tract infections, observational studies report conflicting results (14,17,21). The only randomized controlled trial from an industrialized country did not find any difference in risk for LRTI or hospital admission between the group promoted to breast-feeding compared with controls (22). The full cohort analysis in our study suggests a higher risk of hospitalization for LRTI if breast-feeding lasts ≤4 compared to ≥12 months.
The novel sibling analysis suggested that the associations observed in the full cohort could be biased by residual or unobserved confounding from shared maternal factors. In principle, this matched design eliminates confounding from factors fully shared by the siblings, such as maternal genotype, and reduces confounding from factors likely to be highly correlated among siblings such as rearing practices and parental reporting bias of the infections. The reduced sample size in the cohort belonging to siblings, however, implies wider confidence intervals in the sibling analyses. A matched design has inherent limitations, with the possibility of increasing bias by nonshared confounders (23). Therefore, we cannot automatically conclude that the sibling design produces a less biased result. Nevertheless, our interpretation of the findings among siblings is that the mother may be more important for the infection risk than the feeding mode, consistent with another recent sibling study on breast-feeding and a multitude of other long-term outcomes (24). This study did not assess the risk of infections.
Few studies have assessed the potential benefit of the WHO recommended breast-feeding after 12 months with regard to infections (3), the observation period has been limited to 6 to 12 months (14,17,20–22,25). We found similar risk for hospitalization in infants breast-fed for 6 to 11 months compared to ≥12 months. This finding suggests that breast-feeding plays a minor role in infection prevention after 12 months.
Our study is the largest prospective birth cohort of infant feeding and infections to date, with the potential of more precise estimates as compared to previous studies. Data collected in the present study allowed for the assessment of multiple potential confounders, which reduce the risk of overestimating the potential effects of breast-feeding due to the associated healthy behaviors.
The data collection of exposures and outcomes in a recall time period limited to 6 to 12 months reduces the risk of bias, and other studies have shown good agreement with retrospective breast-feeding data and 24-hour recall (26,27). Infectious symptoms occurring close to changes in diet could be more likely to be notified by the parents. Because infections occurred and infant feeding were given in the same time intervals, reverse causation with infections influencing on breast-feeding cannot be ruled out. A serious infection could lead to interruption of breast-feeding and thus overestimate the apparent benefits of breast-feeding. Alternatively, motivation for breast-feeding could be increased with frequent infections, potentially underestimating the benefits.
A potential limitation of our study is that the difference in adjacent groups for duration of breast-feeding may be too small to induce detectable associations. Comparing full breast-feeding for 3 to 4 months against 6 months (omitting 5 months), however, led to essentially the same conclusions (data not shown). The time of infection within each interval was not specified in the questionnaires, leading to an inability to determine this more accurately.
Our results are likely generalizable to industrialized countries with a similar burden of infectious disease and breast-feeding coverage. A higher background risk of infections in low- and middle-income countries could explain differing results in less resourced populations (11–13,28,29). Participants in our cohort study tend to have a higher level of education and age than the general population, and the participants lost during the study similarly were younger and less educated. In general, the associations between exposures and outcomes have been shown to be robust to selection bias (30). A more heterogeneous cohort will also be more prone to confounding effects of socioeconomic differences.
The recommendations for infant feeding need to take all aspects of the health of the child and the mother into account. Health outcomes later in life as allergies, asthma, autoimmune diseases, neurodevelopment, and obesity are important considerations and may be associated to the duration of breast-feeding. Our findings, supported by the sibling analysis, suggest that effects of prolonged breast-feeding may be overestimated by the healthy behavior among mothers who follow the official recommendations for infant feeding.
To conclude, introduction of complementary foods from 4 to 6 months as compared to introduction ≥6 months of age in breast-fed infants was not a significant risk factor for hospitalization due to infections. Infections thus do not seem to be of major importance for these 2 alternative recommendations for introduction of complementary foods in a high-income society. Our results, however, do support the recommendation to fully breast-feed for 4 months and to continue breast-feeding beyond 6 months, and suggest that protection against infections is limited to the first 12 months.
The authors are grateful to all the participating families in Norway who take part in this ongoing cohort study.
1. Fewtrell M, Wilson DC, Booth I, et al. Six months of exclusive breast feeding: how good is the evidence? BMJ
2. Van’t Hof MA, Haschke F. The Euro-Growth Study: why, who, and how. J Pediatr Gastroenterol Nutr
2000; 31 (suppl 1):S3–S13.
4. McGuire SUS. Dept. of Health and Human Services. The Surgeon General's Call to Action to Support Breastfeeding. U.S. Dept. of Health and Human Services, Office of the Surgeon General. Adv Nutr
5. Eidelman AI, Schanler RJ. AAP Policy statement. Breastfeeding and the use of human milk
6. Quigley MA, Kelly YJ, Sacker A. Infant feeding, solid foods and hospitalisation in the first 8 months after birth. Arch Dis Child
9. Kramer MS, Kakuma R. Optimal duration of exclusive breastfeeding. Cochrane Database Syst Rev
10. Duijts L, Ramadhani MK, Moll HA. Breastfeeding protects against infectious diseases during infancy
in industrialized countries. A systematic review. Matern Child Nutr
11. Khan MU. Breastfeeding, growth and diarrhoea in rural Bangladesh children. Hum Nutr Clin Nutr
12. Khadivzadeh T, Parsai S. Effect of exclusive breastfeeding and complementary feeding on infant growth and morbidity. East Mediterr Health J
13. Cohen RJ, Brown KH, Canahuati J, et al. Effects of age of introduction of complementary foods on infant breast milk intake, total energy intake, and growth: a randomised intervention study in Honduras. Lancet
14. Rebhan B, Kohlhuber M, Schwegler U, et al. Breastfeeding duration and exclusivity associated with infants’ health and growth: data from a prospective cohort study in Bavaria, Germany. Acta Paediatr
15. Haggkvist AP, Brantsaeter AL, Grjibovski AM, et al. Prevalence of breast-feeding in the Norwegian Mother and Child Cohort Study and health service-related correlates of cessation of full breast-feeding. Public Health Nutr
16. Chantry CJ, Howard CR, Auinger P. Full breastfeeding duration and associated decrease in respiratory tract infection in US children. Pediatrics
17. Duijts L, Jaddoe VW, Hofman A, et al. Prolonged and exclusive breastfeeding reduces the risk of infectious diseases in infancy
18. Magnus P, Irgens LM, Haug K, et al. Cohort profile: the Norwegian Mother and Child Cohort Study (MoBa). Int J Epidemiol
19. Perkin MR, Logan K, Tseng A, et al. Randomized trial of introduction of allergenic foods in breast-fed infants. N Engl J Med
20. Oddy WH, Sly PD, De Klerk NH, et al. Breast feeding and respiratory morbidity in infancy
: a birth cohort study. Arch Dis Child
21. Quigley MA, Kelly YJ, Sacker A. Breastfeeding and hospitalization for diarrheal and respiratory infection in the United Kingdom Millennium Cohort Study. Pediatrics
22. Kramer MS, Chalmers B, Hodnett ED, et al. Promotion of Breastfeeding Intervention Trial (PROBIT): a randomized trial in the Republic of Belarus. JAMA
23. Frisell T, Oberg S, Kuja-Halkola R, et al. Sibling comparison designs: bias from non-shared confounders and measurement error. Epidemiology
24. Colen CG, Ramey DM. Is breast truly best? Estimating the effects of breastfeeding on long-term child health and wellbeing in the United States using sibling comparisons. Soc Sci Med
25. Paricio Talayero JM, Lizan-Garcia M, Otero PA, et al. Full breastfeeding and hospitalization as a result of infections in the first year of life. Pediatrics
26. Burnham L, Buczek M, Braun N, et al. Determining length of breastfeeding exclusivity: validity of maternal report 2 years after birth. J Hum Lact
27. Natland ST, Andersen LF, Nilsen TI, et al. Maternal recall of breastfeeding duration twenty years after delivery. BMC Med Res Methodol
28. Dewey KG, Cohen RJ, Brown KH, et al. Age of introduction of complementary foods and growth of term, low-birth-weight, breast-fed infants: a randomized intervention study in Honduras. Am J Clin Nutr
29. Onayade AA, Abiona TC, Abayomi IO, et al. The first six month growth and illness of exclusively and non-exclusively breast-fed infants in Nigeria. East Afr Med J
30. Nilsen RM, Vollset SE, Gjessing HK, et al. Self-selection and bias in a large prospective pregnancy cohort in Norway. Paediatr Perinat Epidemiol
human milk; infancy; infectious disease; weaning
Supplemental Digital Content
© 2017 by European Society for Pediatric Gastroenterology, Hepatology, and Nutrition and North American Society for Pediatric Gastroenterology,