bs_bs_banner

doi:10.1111/jpc.13109

ORIGINAL ARTICLE

Formula and breast feeding in infant food allergy: A population-

based study
Alice J Goldsmith,1,2 Jennifer J Koplin,2,4 Adrian J Lowe,2,5 Mimi LK Tang,2,3,4 Melanie C Matheson,5
Marnie Robinson,3 Rachel Peters,2,4 Shyamali C Dharmage2,5 and Katrina J Allen2,3,4
1
School of Medicine, Sydney, University of Notre Dame, Sydney, New South Wales, 2 Murdoch Childrens Research Institute, 3Department of Allergy and
Immunology, The Royal Children’s Hospital, 4Department of Paediatrics, The University of Melbourne and 5The Centre for Environmental, Genetic and Analytic
Epidemiology, University of Melbourne, Melbourne, Victoria, Australia

Aim: To determine whether infant-feeding practices, including duration of exclusive breastfeeding and use of partially hydrolysed formula,
modify the risk of developing infant food allergy.
Methods: In an observational population-based study, 1 year olds were recruited from community immunisation clinics in Melbourne,
Australia. Parent-reported data on infant-feeding practices and potential confounders were collected prior to infant skin prick testing for four food
allergens. Sensitised infants attended hospital-based oral food challenges to establish food allergy status. Multiple logistic regression was used to
investigate associations between breastfeeding and formula-feeding and infant food allergy adjusting for possible confounding variables.
Results: A total of 5276 (74% response) infants participated. Of the 4537 for whom food allergy status was determined, 515 (11.3%) were food
allergic (challenge-proven in the context of skin prick testing positive (≥2 mm)). After adjusting for confounding variables, there was no associ-
ation between duration of exclusive breastfeeding and food allergy. Use of partially hydrolysed formula did not reduce the risk of food allergy
compared with cow’s milk formula in the general population (adjusted odds ratios 1.03 (confidence interval 0.67–1.50)).
Conclusion: Duration of exclusive breastfeeding and use of partially hydrolysed formula were not associated with food allergy at 1 year of age
in this large population-based study. These findings have implications for population-based infant-feeding guidelines and do not support the use
of partially hydrolysed formula for food allergy prevention.
Key words: allergy prevention; breastfeeding; food allergy; infant feeding; infant formula.

What is already known on this topic What this paper adds

1 The effect of duration of breastfeeding on the risk of developing 1 Duration of exclusive breastfeeding was not associated with in-
food allergy is controversial. fant food allergy after adjusting for all known important con-
2 Randomised controlled trials of partially hydrolysed formula for founding variables.
the prevention of allergic disease in high-risk infants have pro- 2 Partially hydrolysed formula was not associated with infant food
duced conflicting results. None have examined the relationship allergy in infants with or without a family history of allergic
between partially hydrolysed formula and food allergy in infants disease.
without a family history of allergy. 3 Use of partially hydrolysed formula was common in infants with
3 Infant-feeding guidelines have recommended partially hydro- and without a family history of allergic disease despite guide-
lysed formula for high-risk bottle-fed infants in the first 4– lines recommending partially hydrolysed formula only for a sub-
6 months, despite limited evidence. set of high-risk infants.

Food allergy is an important health problem in Australian chil- significant burden and major health, economic and quality of
dren. Australia has amongst the highest prevalence of allergic life impact for Australia’s children.2–4 Food allergy has risen
disease in the developed world.1 Allergic diseases, including rapidly in recent decades. Australian hospitalisations for food-
asthma, eczema, allergic rhinitis and food allergy, have a related anaphylaxis have dramatically increased since 1990,
particularly in the 0–4 years age group5,6 in line with trends
in the US7 and UK.8 We have previously shown that food al-
lergy affects around 10% of 1-year-old infants in Melbourne,
Correspondence: Professor Katrina Allen, Murdoch Childrens Research
Australia.9
Institute, Royal Children’s Hospital, 50 Flemington Road, Parkville, Vic.
3052, Australia. Fax: +61 39345 4848; email: [email protected] Identifying the modifiable determinants of food allergy is im-
portant to address the economic, social and pathophysiological
Conflict of interest: None declared.
burden in children. Early childhood factors such as the type
Accepted for publication 31 July 2015. and duration of infant feeding have been postulated as likely to

Journal of Paediatrics and Child Health 52 (2016) 377–384 377

© 2016 The Authors
Journal of Paediatrics and Child Health © 2016 Paediatrics and Child Health Division (Royal Australasian College of Physicians)
Formula and breast feeding AJ Goldsmith et al.

be important in the development of childhood allergies.10 Infant Cow’s milk allergic: A parent-reported immediate (<1 hour)
feeding is a modifiable factor that can be targeted through popu- reaction (urticaria, angioedema, vomiting or anaphylaxis) to
lation guidelines. Current guidelines lack strong evidence to in- cow’s milk in the context of a positive SPT (2mm).
form infant-feeding recommendations with regard to allergy High risk of allergy: Parent-reported allergic disease (asthma,
prevention, specifically prevention of food allergy.11–13 eczema, allergic rhinitis or food allergy) in at least one first-
Whilst breastfeeding is widely accepted as the food of choice for degree family member (parent or sibling).
infants, its relationship to the development of allergic disease is con- Low risk of allergy: Parent-report of no allergic disease in first-
troversial.11 Various studies have shown either a protective benefit, degree family members.
no effect or increased risk.14,15 There is also limited evidence to sub- Exclusive breastfeeding: Breastfeeding without concurrent
stantiate current guidelines, both nationally and internationally, infant formula, solids or semi-solids.
that advocate the use of partially hydrolysed formula for allergy
prevention in high-risk bottle-fed infants in the first 4– Formula type
6 months.11,13,16,17 Randomised controlled trials examining partially
Formula use was categorised as cow’s milk, partially hydrolysed
hydrolysed formula for the prevention of allergic disease in high-risk
or soy. Other formula types were excluded because of heteroge-
infants have produced conflicting results. In addition, none have ex-
neity of types, relevance to population guidelines and small
amined the relationship between partially hydrolysed formula and
numbers (n < 50). Extensively hydrolysed and amino acid for-
food allergy in infants without a family history of allergy.
mulas are available by specialist prescription for allergic infants,
This article focuses on the impact of breastfeeding and for-
and thus are of limited population relevance.
mula-feeding food allergy at 1 year of age. Using data from a
large population-based study, we examine whether food allergy
is associated with (i) duration of exclusive breastfeeding and (ii)
Statistical analysis
the type of infant formula used in those with and without a fam- We used a separate multivariable logistic regression model for
ily history of allergy. each of the two primary exposures (duration of exclusive
breastfeeding and type of formula) to quantify the association
with food allergy reported as odds ratios (OR) and 95% confi-
Methods dence intervals. Each model was adjusted for confounders identi-
fied from published literature: family history of allergy, infant
HealthNuts study methods have been detailed previously.18
eczema status during breastfeeding, parent reported reactions to
HealthNuts is a population-based observational study investigat-
cow’s milk in the infant and age of introduction of egg.20–22
ing modifiable risk factors for paediatric food allergy. Twelve-
Additional potential confounders were retained if they changed
month-old infants were recruited from immunisation clinics in
the magnitude of the association between either of the exposures
Melbourne. Written informed consent was obtained from par-
and food allergy by more than 10% or were associated with both
ents. Parents completed a questionnaire ascertaining infant-
the exposure and the outcome with a P value <0.05. These
feeding practices and potential confounding variables. Infants
included maternal smoking during pregnancy, household
underwent skin prick testing (SPT) using single-tine lancets to
smoking, preterm delivery (≤36 weeks), number of siblings, maternal
four foods (hen’s egg, peanut, sesame and either cow’s milk
country of birth, socioeconomic status via socio-economic indexes for
(n = 2725) or shrimp (n = 2551)) with a positive histamine and
area (SEIFA (a summary measure of socio-economic conditions
negative saline control. Infants with any detectable wheal on
developed by the Australian Bureau of Statistics))23 and infant ec-
SPT were invited to attend the Royal Children’s Hospital allergy
zema status during formula feeding. Only data that were complete
clinic for repeat SPT with an extended panel of foods (egg, pea-
for all exposures, important confounders and outcomes were
nut, sesame, milk, shrimp, hazelnut, almond, cashew, soy and
included.
wheat) and oral food challenge (OFC). OFCs were performed
A sample size of 5000 infants was calculated to provide suffi-
to egg, peanut and sesame using pre-determined stopping
cient power to detect risk factors present in at least 10% of the
criteria of objective allergic signs occurring within 2 h of the last
population given a prevalence of food allergy of 5–10%. This
dose of allergen. Challenges were called positive if an infant de-
sample size provides 98% power to detect an OR of 1.75 assum-
veloped one or more of the following: ≥3 concurrent non-con-
ing food allergy prevalence of 10%. In our study, 10.5% of in-
tact urticaria lasting 5 min or more; perioral/periorbital
fants used partially hydrolysed formula.
angioedema; vomiting; or circulatory or respiratory compromise,
Analyses investigated the relationship between infant feeding
within 2 h of ingestion of a challenge dose. Details of reactions
and food allergy separately among low and high-risk infants
occurring during challenge have been published previously.19
based on an a priori decision. We also tested for interactions by
comparing models with and without interaction terms using
Definitions the likelihood ratio test.
Stata software (release 12.0; StataCorp, College Station,
Food allergic: A positive OFC in the context of a positive SPT Texas) was used for all analyses.
(wheal 2mm) to that food or 8mm to unchallenged foods (tree
nuts, shrimp, soy or wheat). Ethics approval
Not food allergic: A negative SPT in the community to all four
foods in the context of a positive histamine control, or a negative Ethics approval was obtained from the Victorian State Govern-
OFC in sensitised infants. ment Office for Children (reference no. CDF/07/492), Victorian

378 Journal of Paediatrics and Child Health 52 (2016) 377–384

© 2016 The Authors
Journal of Paediatrics and Child Health © 2016 Paediatrics and Child Health Division (Royal Australasian College of Physicians)
AJ Goldsmith et al. Formula and breast feeding

State Government Department of Human Services (reference significantly interact with the relationship between food allergy
no. 10/07), Royal Children’s Hospital Melbourne (reference no. and formula type (P = 0.72).
27047A) and University of Notre Dame Australia (reference
no. 012043S) ethics committees. Infant feeding and cow’s milk allergy
Of the 2725 infants who underwent SPT to cow’s milk, 2.2% had
a SPT ≥ 2 mm. Of these, 45% had a parent-reported history of
Results reaction to cow’s milk consistent with an Immunoglobulin
Study population E-mediated reaction; thus, 1.0% (n = 27/2725) were considered
cow’s milk allergic. There was no evidence of an association
Of 7134 infants approached at community immunisation clinics, between duration of breastfeeding or type of infant formula used,
5276 (74%) participated. Of these, 1089 infants were invited for and cow’s milk allergy at 1 year of age after adjusting for
hospital-based allergy testing, and 928 (85%) attended and com- confounding variables (Tables 4 and 5).
pleted further SPT and OFC. Overall, food allergy status was de-
termined for 4537 (86%) infants. Of these, 515 (11.3%) were
food allergic (456 egg allergic, 153 peanut allergic and 35 sesame Discussion
allergic) and 27 classified as cow’s milk allergic. Table 1 details
In a large population-based study of 1 year old infants, neither
characteristics of the study population.
duration of exclusive breastfeeding nor use of partially hydro-
Rates of any breastfeeding were high with 94.3% breastfed in
lysed formula was associated with food allergy. Contrary to in-
the first month and 64.9% at 6 months (Fig. 1). Formula feeding
fant feeding recommendations, the use of partially hydrolysed
was also common with 67.3% (3066) infants being fed formula,
formula was common, even for infants without a family history
either alone or concurrently with breastfeeding, at some stage
of allergic disease. For soy formula, the apparent increased risk of
during the first 12 months. The majority (82%) of formula-fed
food allergy was significantly reduced after adjusting for con-
infants used cow’s milk formula; however, 16% of high-risk
founders. This is the first large, population-based study to inves-
and 13% of low-risk formula-fed infants used partially hydro-
tigate the relationship between use of partially hydrolysed
lysed formula. Use of formula alone rose from 6.4% in the first
formula and the gold standard measure of challenge-proven in-
month, to 37.1% at 6 months and 37.5% at 12 months.
fant food allergy in the general population.
Failure to adjust for confounders of the relationship between
Duration of exclusive breastfeeding and infant food breastfeeding and food allergy can lead to the reporting of poten-
allergy tially spurious associations, in particular, an increased risk of
food allergy or sensitisation in breastfed infants.24–27 An analysis
There was no evidence of a relationship between the duration of of breastfeeding and nut allergy, which did not adjust for any
exclusive breastfeeding and infant food allergy after adjusting for recognised confounders, found an increase in parent-reported
confounding variables. Stratification by family history of allergy nut allergy in breastfed infants.27 Our unadjusted findings also
showed that food allergy was more common in the high-risk co- found a spurious, although weak, association between longer
hort, as would be expected. However, duration of exclusive breastfeeding and an increased risk of food allergy. Adjusting
breastfeeding was not associated with food allergy in either the for confounders eliminated the association. Our data are consis-
high or low-risk group either before or after adjustment (Ta- tent with a meta-analysis of breastfeeding and eczema, which
ble 2). There was no evidence of an interaction between allergy found that any protective benefit of longer duration of
risk and the exclusive breastfeeding and food allergy relationship breastfeeding for childhood allergic disease was lost after
(P = 0.459). adjusting for other confounding factors.28
The evidence that hydrolysed formula reduces the risk of al-
Type of formula and infant food allergy lergic disease, including food allergy, remains weak. However,
infant-feeding guidelines internationally have recommended
Use of partially hydrolysed formula did not alter the risk of infant hydrolysed formula for bottle-fed high-risk infants. Two
food allergy compared with cow’s milk formula. Unadjusted ORs randomised controlled trials have addressed the impact of hy-
suggested soy formula increased the risk of food allergy relative drolysed formulas on allergic manifestations yielding contradic-
to cow’s milk formula; however, no association was evident after tory results. A German randomised controlled trial (German
adjusting for confounding variables. The association between soy Infant Nutritional Intervention study (GINI)) comparing three
formula and food allergy in the unadjusted analysis was ex- hydrolysed formulas with cow’s milk formula in 2252 high-risk
plained by confounding because of infant history of allergic dis- infants found both extensively hydrolysed casein formula and
ease. Infants with eczema and those with parent-reported partially hydrolysed whey formula were protective for atopic
reactions to cow’s milk formula were both more likely to use dermatitis and a diagnosis of any allergic manifestation in the
soy formulas and to have a food allergy (data not shown). Ad- first year of life.29 In contrast, a recently published Australian
justment for these factors reduced the magnitude of the associa- randomised controlled trial (Melbourne Atopic Cohort Study
tion between soy formula and food allergy by >20%. (MACS)) in 620 high-risk infants comparing partially hydrolysed
Stratification according to low versus high-risk infants yielded formula and cow’s milk formula found no difference in allergic
similar results with ORs close to 1.00 (not significant) in both risk outcomes in the first 2 years.30 Neither study measured food
categories (Table 3). Family history of allergy did not allergy using the gold standard OFC, and both were limited to

Journal of Paediatrics and Child Health 52 (2016) 377–384 379

© 2016 The Authors
Journal of Paediatrics and Child Health © 2016 Paediatrics and Child Health Division (Royal Australasian College of Physicians)
Formula and breast feeding AJ Goldsmith et al.

Table 1 Demographic, infant-feeding and allergy risk variables of the study cohort defined by food allergy status (column percentages)

Characteristic Not food allergic, n (%) Food allergic, n (%)

Total 4044 (88.7) 515 (11.3)

Infant feeding variables
Completed months of exclusive breastfeeding (n = 3841)
0 1007 (29.6) 116 (26.7)
1 219 (6.4) 20 (4.6)
2 138 (4.1) 19 (4.4)
3 200 (5.9) 23 (5.3)
4 464 (13.6) 62 (14.3)
5 606 (17.8) 96 (22.1)
6 710 (20.9) 91 (21.0)
>6 63 (1.9) 7 (1.6)
Use of formula (n = 2958)
Cow’s milk formula 2225 (82.5) 255 (76.8)
Partially hydrolysed 421 (15.6) 57 (17.2)
Soy 51 (1.9) 20 (6.0)
Allergy risk variables
Family history (n = 4559)
Immediate family history of allergy (asthma, food allergy, allergic rhinitis or eczema in a 2754 (68.1) 397 (77.1)
first degree relative)
No immediate family history of allergy 1290 (31.9) 118 (22.9)
Eczema status and breastfeeding (n = 4267)
No eczema 3082 (81.3) 192 (40.2)
Eczema while breastfeeding 452 (11.9) 234 (49.1)
Eczema after breastfeeding 256 (6.8) 51 (10.7)
Eczema and formula use (n = 4274)
No eczema 3082 (81.1) 192 (40.3)
First eczema before formula feeding started 360 (9.48) 171 (35.9)
First eczema after formula feeding started 356 (9.37) 113 (23.7)
Other infant variables
Gender (n = 4515)
Male 2017 (50.2) 297 (57.8)
Female 2003 (49.8) 217 (42.2)
Preterm delivery (n = 4330)
Preterm delivery (≤36 weeks) 249 (6.5) 17 (3.5)
>36 weeks gestation 3595 (93.5) 469 (96.5)
Number of siblings (n = 4515)
No siblings 1937 (48.4) 290 (56.7)
1 sibling 1349 (33.7) 159 (31.1)
2 siblings 532 (13.3) 51 (10.0)
3 or more siblings 185 (4.6) 11 (2.2)
Mother’s country of birth (n = 4451)
Australia 2942 (74.5) 320 (63.5)
Europe 367 (9.3) 47 (9.3)
East Asia 303 (7.7) 94 (18.7)
Other 335 (8.5) 43 (8.5)
SES via SEIFA quintiles (n = 4545)
1 (least disadvantaged) 828 (20.5) 76 (14.8)
2 792 (19.6) 117 (22.8)
3 855 (21.2) 101 (19.7)
4 787 (19.5) 110 (21.4)
5 (most disadvantaged) 769 (19.1) 110 (21.4)
Maternal smoking during pregnancy (n = 4429)
Yes 197 (5.0) 12 (2.4)
No 3734 (95.0) 486 (97.6)
Parent-reported reaction to cow’s milk formula (n = 4453)
Yes 186 (4.7) 89 (17.8)
No 3766 (95.3) 412 (82.2)

(Continues)

380 Journal of Paediatrics and Child Health 52 (2016) 377–384

© 2016 The Authors
Journal of Paediatrics and Child Health © 2016 Paediatrics and Child Health Division (Royal Australasian College of Physicians)
AJ Goldsmith et al. Formula and breast feeding

Table 1. (Continued)

Characteristic Not food allergic, n (%) Food allergic, n (%)

Age at first introduction of egg products (n = 4353)
<4 months 35 (0.9) 6 (1.2)
4–6 months 989 (25.4) 92 (18.9)
7–9 months 1575 (40.5) 178 (36.5)
10–12 months 1170 (30.1) 154 (31.6)
>12 months 119 (3.1) 57 (11.7)

SES, socio-economic status.

Fig. 1 Formula use and breastfeeding in the first 12 months. Percentages do not add up to 100% because some infants were no longer breastfeeding or
formula feeding.

high allergy risk cohorts. The Cochrane review of hydrolysed allergy amongst Israeli infants who were exposed to cow’s milk
formulas, conducted prior to publication of the MACS findings, formula within the first 14 days of life suggesting early exposure
concluded that there is some evidence, although limited, that to cow’s milk protein might promote tolerance.35 It is possible
use of a hydrolysed formula compared with a cow’s milk for- that very early exposure may modify the development of cow’s
mula reduces infant cow’s milk allergy.31 milk allergy and that the timing of the window of opportunity
It has been hypothesised that hydrolyzing cow’s milk formula for promoting tolerance is earlier for milk compared with
through enzymatic breakdown of proteins reduces exposure to egg.22,35 This is supported by the findings that infants introduced
intact proteins and may reduce sensitisation and thus develop- to milk at 4–6 months were more likely to be milk allergic,35
ment of allergic disease.32 Our findings are consistent with re- while in a separate study, introduction of egg at 4–6 months
cent evidence suggesting that allergen avoidance may not be was found to be protective.22
effective in preventing allergy. The new concept of a window A possible explanation for our findings is that most infants
of opportunity32 proposes that early exposure to potentially al- were predominantly breastfed for the first 4–6 months of life,
lergenic foods may promote persistent oral tolerance. Evidence and exposure to cow’s milk protein was not at sufficient amounts
supporting this concept has been reported for egg,22 peanut33 at the appropriate developmental opportunity to develop toler-
and cow’s milk.34 Katz et al. found lower rates of cow’s milk ance. That is to say, the protective effect of formula might relate

Journal of Paediatrics and Child Health 52 (2016) 377–384 381

© 2016 The Authors
Journal of Paediatrics and Child Health © 2016 Paediatrics and Child Health Division (Royal Australasian College of Physicians)
Formula and breast feeding AJ Goldsmith et al.

Table 2 Relationship between duration of exclusive breastfeeding and infant food allergy stratified by low-risk versus high-risk

Low risk infants (n = 977) High risk infants (n = 2236)

Completed
months of
exclusive Food Food
breastfeeding allergic (%) OR (95% CI) P value OR (95% CI)† P value allergic (%) OR (95% CI) P value OR (95% CI)† P value

0 7.19 1.00 — 1.00 — 10.74 1.00 — 1.00 —

1 6.25 0.86 (0.31–2.36) 0.770 1.09 (0.36–3.32) 0.884 8.53 0.77 (0.40–1.51) 0.454 0.74 (0.37–1.49) 0.399
2 8.33 1.17 (0.38–3.58) 0.779 1.17 (0.34–3.99) 0.804 13.04 1.24 (0.65–2.41) 0.511 1.09 (0.54–2.22) 0.804
3 7.69 1.08 (0.39–2.97) 0.888 1.10 (0.35–3.47) 0.868 11.48 1.08 (0.58–1.99) 0.811 1.11 (0.57–2.15) 0.756
4 13.07 1.94 (1.02–3.70) 0.044 1.52 (0.71–3.27) 0.285 9.77 0.90 (0.57–1.41) 0.651 0.77 (0.47–1.27) 0.303
5 8.18 1.15 (0.56–2.36) 0.705 1.10 (0.48–2.55) 0.815 14.73 1.44 (1.00–2.06) 0.051 1.38 (0.92–2.08) 0.126
6 6.67 0.92 (0.44–1.92) 0.828 0.59 (0.25–1.43) 0.246 13.21 1.27 (0.88–1.81) 0.200 1.07 (0.72–1.61) 0.728

†Adjusted for infant eczema that occurred during breastfeeding, family history of allergic disease, maternal smoking during pregnancy, preterm delivery
(<38 weeks), number of siblings, maternal country of birth, SES via SEIFA and age of introduction of egg. CI, confidence interval; RO, odds ratio.

Table 3 Relationship between type of formula and infant food allergy stratified by low-risk versus high-risk infants

Low risk infants (n = 764) High risk infants (n = 1625)

Type of formula % Food % Food

allergic (n) OR (95% CI) P value OR (95% CI)† P value allergic (n) OR (95% CI) P value OR (95% CI) † P value

Cow’s milk 8.29 (54) 1.00 — 1.00 10.84 (142) 1.00 — 1.00 —
Partially hydrolysed 6.86 (7) 0.81 (0.36–1.84) 0.623 0.81 (0.33–1.94) 0.633 12.73 (35) 1.20 (0.81–1.78) 0.367 1.09 (0.72–1.67) 0.684
Soy 27.27 (3) 4.15 (1.07–16.09) 0.040 1.06 (0.17–6.65) 0.946 22.50 (9) 2.39 (1.11–5.12) 0.025 1.32 (0.56–3.13) 0.524

†Adjusted for infant eczema that occurred during breastfeeding, infant eczema that occurred before formula feeding, family history of allergic disease, ma-
ternal smoking during pregnancy, preterm delivery (<38 weeks), parent-reported reactions to cow’s milk formula and age of introduction of egg. CI, con-
fidence interval; RO, odds ratio.

Table 4 Relationship between duration of exclusive breastfeeding and cow’s milk allergy among 2725 infants who underwent skin prick testing to cow’s
milk

Completed months of
exclusive breastfeeding Number Cow’s milk allergy OR (95% CI) P value OR (95% CI)† P value

0 499 0.6 (3) 1.00 — 1.00 —

1 121 0 (0) 1.00 — 1.00 —
2 81 0 (0) 1.00 — 1.00 —
3 114 0.88 (1) 1.46 (0.15–14.20) 0.743 1.33 (0.12–14.42) 0.815
4 296 0.68 (2) 1.12 (0.19–6.77) 0.898 0.43 (0.06–3.08) 0.400
5 357 1.68 (6) 2.83 (0.70–11.38) 0.144 1.54 (0.33–7.11) 0.582
6 351 2.85 (10) 4.84 (1.32–17.75) 0.017 3.19 (0.76–13.34) 0.113

†Adjusted for infant eczema that occurred during breastfeeding, family history of allergic disease, maternal smoking during pregnancy, preterm delivery
(<38 weeks), number of siblings, maternal country of birth, SES via SEIFA and age of introduction of egg. CI, confidence interval; OR, odds ratio.

382 Journal of Paediatrics and Child Health 52 (2016) 377–384

© 2016 The Authors
Journal of Paediatrics and Child Health © 2016 Paediatrics and Child Health Division (Royal Australasian College of Physicians)
AJ Goldsmith et al. Formula and breast feeding

Table 5 Relationship between type of formula and cow’s milk allergy among 2725 infants who underwent skin prick testing to cow’s milk

Type of formula n % Cow’s milk allergy (n) OR (95% CI) P value OR (95% CI)† P value

Cow’s milk 1084 0.74 (8) 1.00 — 1.00 —

Partially hydrolysed 242 1.65 (4) 2.26 (0.68–7.57) 0.186 3.05 (0.76–12.22) 0.116
Soy 26 3.85 (1) 5.38 (0.65–44.66) 0.119 1.48 (0.11–19.11) 0.766

†Adjusted for infant eczema that occurred during breastfeeding, infant eczema that occurred before formula feeding, family history of allergic disease, ma-
ternal smoking during pregnancy, preterm delivery (<38 weeks), parent-reported reaction to cow’s milk formula and age of introduction of egg. CI, confi-
dence interval; RO, odds ratio.

to the timing, rather than type, of formula exposure. Similarly, Early signs of atopic disease (i.e. visible infant eczema) are
both the GINI and MACS study cohorts had a high rate of known to have an important influence on maternal infant-feeding
breastfeeding in the first 6 months, which may also explain the decisions.20 This study was able to control for this, thus taking into
null effect of formula for the protection of allergic disease.29,30 account the potential for reverse causation. This study also adjusted
Key strengths of this study include a high community recruit- for other important known potential confounders including family
ment rate, large sample size and standardised, well-defined food history of allergy, parental smoking history, timing of introduction
allergy criteria including OFC. The study included a representa- of solids and parent report of infant food reactions. As in all obser-
tive population sample,18 and infants across the spectrum of al- vational studies, we are unable to completely exclude the possibil-
lergy risk (i.e. infants with and without a family history of ity of confounding by unknown factors.
allergic disease), thus enabling translation of findings to the gen- In conclusion, although partially hydrolysed formula is cur-
eral population. rently recommended for high-risk bottle-fed infants, uptake of
Observational studies offer the only practical and ethically fea- guidelines appears to be variable. Although, use of partially hy-
sible approach to studying breastfeeding associations because drolysed formula is common, it is not restricted to the recom-
breastfeeding is the preferred infant food. Compared with a mended high-risk population. This supports the likelihood that
randomised trial, our observational study is ethically feasible infant-feeding choices are informed by numerous influences, in-
and likely to be more representative of population infant-feeding cluding community beliefs, understandings of allergy risk, for-
practices. Randomised controlled trials comparing different for- mula types and guidelines, health professional advice and
mula types for the prevention of food allergy among those who advertising influences. Thus, our data on the use of infant for-
choose not to exclusively breastfeed are feasible; however, to mula, and effect of partially hydrolysed formula on allergic dis-
date, these have only been conducted in high risk infants. A ease at a population level, have important implications for
large population-based randomised controlled trial of partially population-based infant-feeding guidelines.
hydrolysed formula compared with cow’s milk formula could
theoretically be undertaken; however, this would require a very
large sample size to investigate the effect among low-risk infants Acknowledgements
and infants without a family history of allergy are likely to be dif-
ficult to recruit to an allergy study. We thank the children and parents who participated in the
The use of parent-reported questionnaire (completed at re- HealthNuts Study as well as the staff of Melbourne’s Local
cruitment when the child was 11–15 months) is subject to recall Government Areas for access to community Immunisation
error. Questionnaire data were collected prior to food allergy Clinics. We thank ALK Abello, S.A. Madrid, Spain for supplying
testing; therefore, recall is likely to be non-differential between the SPT reagents and the HealthNuts safety committee: Associate
the allergic and non-allergic groups. Family history of food al- Professor Noel Cranswick (Australian Paediatric Pharmacology
lergy or infant eczema might also impact on recall; however, Research Unit, Murdoch Children’s Research Institute), Dr Jo
we have adjusted for these factors in our analysis. Smart (Department of Allergy and Immunology, Royal
Our analysis of cow’s milk allergy was likely underpowered to Children’s Hospital, Melbourne, Australia) and Associate
detect associations as IgE mediated cow’s milk allergy was rare at Professor Jo Douglass (Director, Department of Allergy and
12 months; however, this was not the primary outcome and was Immunology, Royal Melbourne Hospital, Melbourne, Australia).
included as a secondary analysis. The HealthNuts study was supported by funding from the
Our study examined food allergy at 1 year. How breastfeeding National Health and Medical Research Council (NHMRC) of Aus-
practices affect the development of allergic disease, particularly tralia, Ilhan Food Allergy Foundation, AnaphylaxiStop, the Charles
asthma, in the longer term remains unclear. Further follow-up of and Sylvia Viertel Medical Research Foundation and the Victorian
the HealthNuts cohort, which is currently underway, may yield Government’s Operational Infrastructure Support Program.
further insights. Previous studies have found that associations K. A. is a Viertel Senior Medical Research Fellow and has
change during childhood. In a population-based longitudinal study received speaker fees from Abbott, Nutricia and Pfizer. J. K. is partly
of 8280 Australian infants, exclusive breastfeeding for the first supported by a postdoctoral fellowship from an NHMRC Capacity
3 months was associated with a reduced risk of parent-reported Building Grant in Population Health. A. L., M. M. and S. D. are sup-
food allergy at 7 years, but an increased risk in adult life.36 ported by the NHMRC. A. L. was previously supported by Dairy

Journal of Paediatrics and Child Health 52 (2016) 377–384 383

© 2016 The Authors
Journal of Paediatrics and Child Health © 2016 Paediatrics and Child Health Division (Royal Australasian College of Physicians)
Formula and breast feeding AJ Goldsmith et al.

Australia, C. R. C. for Asthma and Airways and VicHealth. M. T. is a 17 National Health and Medical Research Council. Dietary Guidelines for
member of Medical Advisory Boards for Nestlé Nutrition Institute Children and Adolescents in Australia: Incorporating the Infant Feeding
and Nutricia, has sat on advisory committees for Pfizer and has re- Guidelines for Health Workers. Canberra: National Health and Medical
Research Council, 2003. Available from: http://www.nhmrc.gov.au/
ceived payment for speaking engagements with Danone.
guidelines/publications/n29-n30-n31-n32-n33-n34 [accessed 5 October
2012].
18 Osborne NJ, Koplin JJ, Martin PE et al. The HealthNuts population-based
study of paediatric food allergy: validity, safety and acceptability. Clin.
Exp. Allergy 2010; 40: 1516–22.
References 19 Koplin JJ, Tang MK, Martin PE et al. Predetermined challenge eligibility and
cessation criteria for oral food challenges in the HealthNuts population-
1 Asher MI, Montefort S, Björkstén B et al. Worldwide time trends in the
based study of infants. J. Allergy Clin. Immunol. 2012; 129: 1145–7.
prevalence of symptoms of asthma, allergic rhinoconjunctivitis, and
20 Lowe AJ, Carlin JB, Bennett CM et al. Atopic disease and breast-feeding-
eczema in childhood: ISAAC Phases One and Three repeat
cause or consequence? J. Allergy Clin. Immunol. 2006; 117: 682–7.
multicountry cross-sectional surveys. Lancet 2006; 368: 733–43.
21 Koplin J, Dharmage SC, Gurrin L et al. Soy consumption is not a risk
2 Australasian Society of Clinical Immunology and Allergy (ASCIA). The
factor for peanut sensitization. J. Allergy Clin. Immunol. 2008; 121:
Economic Impact of Allergic Disease in Australia: Not to be Sneezed at.
1455–9.
Report. Access Economic Pty Limited for ASCIA, 2007.
22 Koplin JJ, Osborne NJ, Wake M et al. Can early introduction of egg
3 Australian Institute of Health and Welfare. A Picture of Australia’s
prevent egg allergy in infants? A population-based study. J. Allergy Clin.
Children 2009. Canberra: Australian Institute of Health and Welfare,
Immunol. 2010; 126: 807–13.
2009. Cat. No.: PHE 112.
23 Pink B. An Introduction to Socio-Economic Indexes for Areas (SEIFA).
4 Marklund B, Ahlstedt S, Nordström G. Health-related quality of life in
Canberra: Australian Bureau of Statistics; 2006. Report No.:
food hypersensitive schoolchildren and their families: parents’
2039.0.55.001.
perceptions. Health Qual. Life Outcomes 2006; 12: 1–12.
24 Businco L, Marchetti F, Pellegrini G, Cantani A, Perlini R. Prevention of
5 Liew WK, Williamson E, Tang MK. Anaphylaxis fatalities and admissions
atopic disease in “at risk newborns” by prolonged breast-feeding. Ann.
in Australia. J. Allergy Clin. Immunol. 2009; 123: 434–42.
Allergy 1983; 51: 296–9.
6 Poulos LM, Waters A-M, Correll PK, Loblay RH, Marks GB. Trends in
25 Saarinen U, Kajosaari M. Breastfeeding as prophylaxis against atopic
hospitalizations for anaphylaxis, angioedema, and urticaria in
disease: prospective follow-up study until 17 years old. Lancet 1995;
Australia, 1993–1994 to 2004–2005. J. Allergy Clin. Immunol. 2007;
346: 1065–9.
120: 878–84.
26 Wetzig H, Schulz R, Diez U, Herbarth O, Viehweg B, Borte M.
7 Lin R, Anderson A. Increasing anaphylaxis hospitalizations in the first 2
Associations between duration of breast-feeding, sensitization to hens’
decades of life: New York State, 199–2006. Ann. Allergy Asthma
eggs and eczema infantum in one and two year old children at high risk
Immunol. 2008; 101: 387–93.
of atopy. Int. J. Hyg. Environ. Health 2000; 203: 17–21.
8 Gupta R, Sheikh A, Strachan DP, Anderson HR. Time trends in allergic 27 Paton J, Kljakovic M, Ciszek K, Ding P. Infant feeding practices and nut
disorders in the UK. Thorax 2007; 62: 91–6. allergy over time in Australian school entrant children. Int. J. Pediatr.
9 Osborne NJ, Koplin JJ, Martin PE et al. Prevalence of challenge-proven 2012. DOI: 10.1155/2012/675724.
IgE-mediated food allergy using population-based sampling and 28 Yang YW, Tsai CL, Lu CY. Exclusive breastfeeding and incident
predetermined challenge criteria in infants. J. Allergy Clin. Immunol. atopic dermatitis in childhood: a systematic review and meta-
2011; 127: 668–76.e1–2. analysis of prospective cohort studies. Br. J. Dermatol. 2009; 161:
10 Muraro A, Dreborg S, Halken S et al. Dietary prevention of allergic 373–83.
diseases in infants and small children. Part III: Critical review of 29 von Berg A, Koletzko S, Grübl A et al. The effect of hydrolyzed cow’s
published peer-reviewed observational and interventional studies and milk formula for allergy prevention in the first year of life: the German
final recommendations. Pediatr. Allergy Immunol. 2004; 15: 291–307. infant nutritional intervention study, a randomized double-blind trial.
11 Greer FR, Sicherer SH, Burks AW. Effects of early nutritional J. Allergy Clin. Immunol. 2003; 111: 533–4.
interventions on the development of atopic disease in infants and 30 Lowe AJ, Hosking CS, Bennett CM et al. Effect of a partially hydrolyzed
children: the role of maternal dietary restriction, breastfeeding, timing whey infant formula at weaning on risk of allergic disease in high-risk
of introduction of complementary foods, and hydrolyzed formulas. children: a randomized controlled trial. J. Allergy Clin. Immunol. 2011;
Pediatrics 2008; 121: 183–91. 128: 360–5. e4.
12 Osborn DA, Sinn JKH. Formulas containing hydrolysed protein for 31 Osborne DA, Sinn J. Formulas containing hydrolysed protein for
prevention of allergy and food intolerance in infants. Cochrane prevention of allergy and food intolerance in infants. Cochrane
Database Syst. Rev. 2006; 2: CD003664. Database Syst. Rev. 2006; 4: CD003664.
13 Lowe A, Dharmage SC, Allen KJ, Tang MLK, Hill DJ. The role of partially 32 Høst A, Koletzko B, Dreborg S et al. Dietary products used in infants for
hydrolyzed whey formula for the prevention of allergic disease: treatment and prevention of food allergy. Arch. Dis. Child. 1999; 81: 80–4.
evidence and gaps [Invited Review]. Expert Rev. Clin. Immunol. 2013; 33 Prescott SL, Smith P, Tang M et al. The importance of early
9: 31–41. complementary feeding in the development of oral tolerance: concerns
14 Grimshaw K, Allen K, Edwards C et al. Infant feeding and allergy and controversies. Pediatr. Allergy Immunol. 2008; 19: 375–80.
prevention: a review of current knowledge and recommendations. 34 Du Toit G, Katz Y, Sasieni P et al. Early consumption of peanuts in
Allergy 2009; 64: 1407–16. infancy is associated with a low prevalence of peanut allergy. J. Allergy
15 Matheson MC, Allen KJ, Tang MLK. Understanding the evidence for and Clin. Immunol. 2008; 122: 984–91.
against the role of breastfeeding in allergy prevention. Clin. Exp. Allergy 35 Katz Y, Rajuan N, Goldberg MR et al. Early exposure to cow’s milk
2012; 26: 1–25. protein is protective against IgE-mediated cow’s milk protein allergy.
16 Australasian Society of Clinical Immunology and Allergy. Infant Feeding J. Allergy Clin. Immunol. 2010; 126: 77–82. e1.
Advice. Australasian Society of Clinical Immunology and Allergy, 2010. 36 Matheson MC, Erbas B, Balasuriya A et al. Breast-feeding and atopic
Available from: http://www.allergy.org.au/health-professionals/papers/ disease: a cohort study from childhood to middle age. J. Allergy Clin.
ascia-infant-feeding-advice [accessed 15 October 2012]. Immunol. 2007; 120: 1051–7.

384 Journal of Paediatrics and Child Health 52 (2016) 377–384

© 2016 The Authors
Journal of Paediatrics and Child Health © 2016 Paediatrics and Child Health Division (Royal Australasian College of Physicians)