Infant formulas containing hydrolysed protein for prevention of allergic disease

David A Osborn; John KH Sinn; Lisa J Jones

doi:10.1002/14651858.CD003664.pub6

. 2018 Oct 19;2018(10):CD003664. doi: 10.1002/14651858.CD003664.pub6

Infant formulas containing hydrolysed protein for prevention of allergic disease

David A Osborn ¹, John KH Sinn ^2,^✉, Lisa J Jones ^3,⁴

Editor: Cochrane Neonatal Group

PMCID: PMC6517017 PMID: 30338526

Abstract

Background

Infant formulas containing hydrolysed proteins have been widely advocated for preventing allergic disease in infants, in place of standard cow’s milk formula (CMF). However, it is unclear whether the clinical trial evidence supports this.

Objectives

To compare effects on allergic disease when infants are fed a hydrolysed formula versus CMF or human breast milk. If hydrolysed formulas are effective, to determine what type of hydrolysed formula is most effective, including extensively or partially hydrolysed formula (EHF/PHF). To determine whether infants at low or high risk of allergic disease, and whether infants receiving early short‐term (first few days after birth) or prolonged formula feeding benefit from hydrolysed formulas.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL 2017, Issue 11), MEDLINE (1948 to 3 November 2017), and Embase (1974 to 3 November 2017). We also searched clinical trials databases, conference proceedings, and the reference lists of retrieved articles and previous reviews for randomised controlled trials and quasi‐randomised trials.

Selection criteria

We searched for randomised and quasi‐randomised trials that compared use of a hydrolysed formula versus human milk or CMF. Outcomes with ≥ 80% follow‐up of participants from eligible trials were eligible for inclusion.

Data collection and analysis

Two review authors independently selected trials, assessed trial quality and extracted data from the included studies. Fixed‐effect analyses were performed. The treatment effects were expressed as risk ratio (RR) and risk difference (RD) with 95% confidence intervals and quality of evidence using the GRADE quality of evidence approach. The primary outcome was all allergic disease (including asthma, atopic dermatitis, allergic rhinitis and food allergy).

Main results

A total of 16 studies were included.

Two studies assessed the effect of three to four days infant supplementation with an EHF while in hospital after birth versus pasteurised human milk feed. A single study enrolling 90 infants reported no difference in all allergic disease (RR 1.43, 95% CI 0.38 to 5.37) or any specific allergic disease up to childhood including cow's milk allergy (CMA) (RR 7.11, 95% CI 0.35 to 143.84). A single study reported no difference in infant CMA (RR 0.87, 95% CI 0.52 to 1.46; participants = 3559). Quality of evidence was assessed as very low for all outcomes.

No eligible trials compared prolonged hydrolysed formula versus human milk feeding.

Two studies assessed the effect of three to four days infant supplementation with an EHF versus a CMF. A single study enrolling 90 infants reported no difference in all allergic disease (RR 1.37, 95% CI 0.33 to 5.71; participants = 77) or any specific allergic disease including CMA up to childhood. A single study reported a reduction in infant CMA of borderline significance (RR 0.62, 95% CI 0.38 to 1.00; participants = 3473). Quality of evidence was assessed as very low for all outcomes.

Twelve studies assessed the effect of prolonged infant feeding with a hydrolysed formula compared with a CMF. The data showed no difference in all allergic disease in infants (typical RR 0.88, 95% CI 0.76 to 1.01; participants = 2852; studies = 8) and children (typical RR 0.85, 95% CI 0.69 to 1.05; participants = 950; studies = 2), and no difference in any specific allergic disease including infant asthma (typical RR 0.57, 95% CI 0.31 to 1.04; participants = 318; studies = 4), eczema (typical RR 0.93, 95% CI 0.79 to 1.09; participants = 2896; studies = 9), rhinitis (typical RR 0.52, 95% CI 0.14 to 1.85; participants = 256; studies = 3), food allergy (typical RR 1.42, 95% CI 0.87 to 2.33; participants = 479; studies = 2), and CMA (RR 2.31, 95% CI 0.24 to 21.97; participants = 338; studies = 1). Quality of evidence was assessed as very low for all outcomes.

Authors' conclusions

We found no evidence to support short‐term or prolonged feeding with a hydrolysed formula compared with exclusive breast feeding for prevention of allergic disease. Very low‐quality evidence indicates that short‐term use of an EHF compared with a CMF may prevent infant CMA. Further trials are recommended before implementation of this practice.

We found no evidence to support prolonged feeding with a hydrolysed formula compared with a CMF for prevention of allergic disease in infants unable to be exclusively breast fed.

Plain language summary

Formulas containing hydrolysed protein for prevention of allergic disease in infants

Review question

Does feeding infants with a formula containing hydrolysed protein result in decreased risk of developing allergic disease such as asthma, dermatitis/eczema, hay fever and food allergy during infancy and childhood?

Background

Allergic disease is responsible for a substantial health burden among infants, children and adults. Early dietary intake may influence the development of allergic disease. When babies are not exclusively breast fed, use of hydrolysed formula instead of ordinary cow's milk formula may reduce allergic disease among babies and children, although additional studies are needed to confirm this. Infant formulas have been designed to lower the chance of infants developing allergic disease. These include hydrolysed cow's milk and soy milk formulas. Hydrolysed formulas break down milk proteins into smaller, potentially less allergy‐producing proteins.

Results

This review of trials found no evidence to support feeding with a hydrolysed formula to prevent allergic disease in preference to exclusive breast feeding. This review also found that for infants who are unable to be exclusively breast fed, there is no evidence that prolonged infant feeding with a hydrolysed formula compared with a cow's milk is associated with any difference in allergic disease, asthma, eczema, rhinitis, food allergy or cow's milk formula at any time point. However, limited data in infants who are exclusively formula fed suggest that feeding with a hydrolysed formula instead of a cow's milk formula may reduce infant allergic disease. Concerns regarding quality of the evidence and consistency of the results indicate that continued study is needed. The evidence in this review comes from literature searches updated until November 2017.

Conclusions

We found no substantial evidence to support short‐term or prolonged feeding with a hydrolysed formula compared with a cow's milk formula for prevention of allergic disease in infants unable to be exclusively breast fed.

Summary of findings

Background

Description of the condition

Food allergy and other allergic disease are prevalent and represent a substantial health problem that may be increasing in developed countries (Burr 1989; Halken 2004; Prescott 2005; Schultz Larsen 1996). Although less than half of those who develop childhood allergic disease have a first‐degree relative with a history of allergic disease, the risk of allergic disease increases substantially with a positive family history (Bergmann 1994; Sears 1996; Tariq 1998). Approximately 10% of children without an allergic first‐degree relative develop allergic disease compared with 20% to 30% with an allergic first‐degree relative (parent or sibling) and 40% to 50% with two affected relatives (Arshad 2005; Bergmann 1997; Hansen 1993; Kjellman 1977). The predictive value of family history is increased with the addition of cord blood immunoglobulin (Ig) E antibody testing, although its accuracy may not be adequate for population screening (Bergmann 1997; Bergmann 1998; Tariq 1998).

Manifestations of allergic disease are age dependent. Infants commonly present with symptoms and signs of atopic eczema, gastrointestinal symptoms and recurrent wheezing. Asthma and rhino conjunctivitis become prevalent in later childhood. Sensitisation to allergens tends to follow a characteristic pattern (Halken 2004), with sensitisation to food allergens in the first two to three years of life, followed by indoor allergens (e.g. house dust mite, pets) and subsequently outdoor allergens (e.g. rye, timothy grass). The cumulative prevalence of allergic disease during childhood is high, with up to 7% to 8% developing a food allergy, 15% to 20% atopic eczema and 31% to 34% asthma or recurrent wheezing (Halken 2004). Of these, 7% to 10% will continue to have asthma symptoms beyond five years of age (Halken 2004). Food hypersensitivities affect approximately 6% of infants younger than three years, and prevalence decreases over the first decade (Osterballe 2005; Sampson 2004).

Allergic disease may be diagnosed by questionnaire or clinician assessment, and the diagnosis may be confirmed by specific skin or serological testing, or by allergen challenge. Diagnostic criteria for different allergic diseases are not uniform, and the mode of ascertainment of allergic disease is variable. Although tests of bronchial hyper‐responsiveness, challenge tests and classical tests of IgE‐mediated allergy have an imperfect correlation with allergy symptoms and clinical signs (Darsow 2000; Peat 2000), they are associated with an increased likelihood of allergic disease (Ronmark 2001; Sears 1998; Sly 1999; Strachan 1996). In addition, some evidence suggests that questionnaires, although compromised by selection and recall bias (Peat 2001), are suitable for allergic disease screening (Kilpeläinen 2001; Ravault 2001). This review includes trials that diagnosed allergic disease by questionnaire or by clinician assessment, with or without confirmation by laboratory testing. Criteria for diagnosis of allergic disease should include typical symptoms and/or signs, with evidence of precipitants, persistence or recurrence typical of allergic disease, or with test evidence confirming atopy or bronchial hyperreactivity.

The World Allergy Organization 2003 consensus (Johansson 2004) recommended that the term 'hypersensitivity' should be used to describe objectively reproducible symptoms or signs initiated by exposure to a defined stimulus at a dose tolerated by normal persons. 'Allergy' is a hypersensitivity reaction initiated by specific immunological mechanisms. The term 'food allergy' is used when immunological mechanisms have been demonstrated. Food‐specific IgG antibodies in serum are not of clinical importance but merely indicate previous exposure to a specific food. If IgE is involved in the reaction, the term 'IgE‐mediated food allergy' is appropriate. Food hypersensitivity is diagnosed by resolution of typical symptoms with elimination from the diet, with confirmation by blinded challenge. Around 2% to 3% of babies develop hypersensitivity to a particular food. Principal symptoms among infants with proven cow's milk protein hypersensitivity (CMPH) are gastrointestinal (˜ 50%), dermatological (˜ 31%) and respiratory (˜ 19%) in nature (Høst 1994; Høst 1995; Schrander 1993). Two of every three infants with CMPH have a family history of atopy (Schrander 1993). CMPH is strongly associated with feeding of cow's milk formula (CMF) to infants during the first month of life (Høst 1991). Many infants with CMPH become tolerant over time, with approximately 30% at one year, 50% at two years and 70% at three years tolerant to cow's milk challenge. The risk of persisting hypersensitivity is increased with evidence of atopy (Høst 1995).

Description of the intervention

Measures to prevent allergic disease including food allergy have included maternal allergen avoidance during pregnancy (Custovic 2000; Custovic 2001; Kramer 2012; Zeiger 1989) and/or lactation (Custovic 2000; Custovic 2001; Zeiger 1989), periods of exclusive breast feeding (Custovic 2000; Custovic 2001; Gruskay 1982; Oddy 1999; Saarinen 1995; Saarinen 2000), and avoidance of potential allergens including food and environmental antigens during the first year of life and beyond (Custovic 2000). Formulas prescribed for infants with the intention of preventing allergic disease including food allergy include hydrolysed cow's milk and elemental formula, as well as soy or hydrolysed soy formula. These formulas may be produced from cow's milk or soy milk, may be derived predominantly from whey or casein proteins and may be partially or extensively hydrolysed. Protein modification is performed through a variety of physiochemical processes including ultra heating and enzymatic cleavage, most often with trypsin and chymotrypsin.

The American Academy of Pediatrics (AAP) recommends that hypoallergenic formulas should be tested in trials in which investigators examine human infants for toxicity and suitability to maintain a positive nitrogen balance, while attempting to predict whether infants allergic to cow’s milk will react adversely to these formulas (AAP 2000). These formulas are studied in infants with cow’s milk or cow’s milk‐based formula allergic reactions verified by double‐blind placebo‐controlled challenge (DBPCC) (Bock 1988). At a minimum, these tests should ensure with 95% confidence that 90% of infants with documented cow’s milk allergy respond to treatment and do not react during challenge (Kleinman 1991). Protein particle size does not appear to be a prerequisite for defining a formula as hypoallergenic, although amino acid‐based formulas and those with more extensive hydrolysis are less likely to produce reactions among infants with cow's milk allergy (CMA) (Hill 2007). Although universal agreement has not been reached on the definition (Chaffen 2010; Høst 1999), for the purposes of this review an extensively hydrolysed formula (EHF) will be regarded as one meeting the AAP definition for hypoallergenic formula (AAP 2000), and those with less extensive hydrolysis will be regarded as partially hydrolysed.

How the intervention might work

Infants' immune systems become sensitised or tolerant to allergens in the order in which they are exposed. Early life exposure to allergens occurs frequently through ingested protein, particularly cow's milk protein in formula (Muraro 2004). Amino acid‐based and extensively hydrolysed protein formulas are produced so as to substantially reduce the antigenicity of the protein and prevent sensitisation of infants to commonly ingested antigens, including cow's milk protein. However, low concentrations of food allergens, especially cow's milk proteins, are present in human milk. It has been suggested that the low incidence of cow's milk protein allergy among exclusively breast fed infants ‐ at 0.5% in unselected infants and 1.3% in high‐risk infants ‐ in prospective birth cohort studies was due to low‐level exposure‐induced tolerance rather than to disease (Halken 2004). It has been proposed that prolonged exposure to allergenic proteins or to proteins with reduced but not absent allergenicity may induce tolerance over time (Allen 2009). Although most infants with cow’s milk hypersensitivity exhibit this in the first year of life, more than 80% subsequently develop clinical tolerance (Katz 2011; Sampson 2004). The concern is that early avoidance of cow's milk protein may reduce the likelihood that infants will subsequently develop tolerance to the allergen (Katz 2010).

Why it is important to do this review

The aim of this review is to gather evidence on the use of hydrolysed formulas for prevention of allergic disease including food allergy. This review does not include treatment of infants with clinically recognised allergic disease.

Objectives

To compare effects on allergic disease including food allergy when infants are fed a hydrolysed formula versus cow's milk formula (CMF) or human breast milk. If hydrolysed formulas are effective, to determine what type of hydrolysed formula is most effective, including extensively or partially hydrolysed formula (EHF/PHF). To determine which infants at low or high risk of allergic disease and which infants receiving early, short‐term or prolonged formula feeding may benefit from hydrolysed formulas.

Methods

Criteria for considering studies for this review

Types of studies

We searched for randomised and quasi‐randomised trials that compared the use of a hydrolysed formula versus human milk or cow's milk formula (CMF). Randomised and quasi‐randomised (e.g. using alternation) trials with ≥ 80% follow‐up of participants were eligible for inclusion.

Types of participants

Infants in the first six months of life without clinical evidence of allergic disease.

Types of interventions

Hydrolysed formulas included:

hydrolysed cow's milk and soy formulas; and
extensively and partially hydrolysed formulas (EHF/PHF).

Hydrolysed formulas may be used for:

early, short‐term supplementary or sole formula feeding of infants unable to be exclusively breast fed in the first days of life;
prolonged supplementation of breast fed infants or infants fed solely with formula in the first months of life; and
weaning from the breast with infant formula.

The control group may include infants who receive:

exclusive human milk (breast fed or expressed); and
cow's milk formula (CMF).

Study authors had to pre specify the following comparisons.

Early short‐term hydrolysed formula versus human milk.
Prolonged use of a hydrolysed formula versus human milk.
Early short‐term hydrolysed formula versus CMF.
Prolonged use of a hydrolysed formula versus CMF.

Prespecified subgroup analyses included the following (see Subgroup analysis and investigation of heterogeneity for definitions).

Infant risk of allergic disease.

- Low‐risk infants (no family history of allergic disease among first‐degree relatives).
- High‐risk infants (family history of allergic disease among first‐degree relatives or high cord IgE level).

Extent of protein hydrolysis.

- EHF versus CMF.
- PHF versus CMF.
- EHF versus PHF.

Indication for use.

- Prolonged sole formula feeding.
- Supplemental feeding or weaning from the breast using infant formula.

Method of ascertainment of allergic disease.

- Allergic disease confirmed by test.
- Blinded measurement for allergic disease.

Type of protein hydrolysate used.

- Partially hydrolysed whey formula versus cow's milk formula.
- Partially hydrolysed casein formula versus cow's milk formula.
- Extensively hydrolysed whey formula versus cow's milk formula.
- Extensively hydrolysed casein formula versus cow's milk formula.
- Hydrolysed soy formula versus cow's milk formula.

We excluded studies that included other allergy prevention interventions (e.g. maternal dietary avoidance measures, environmental allergy reduction measures) in the treatment group and not in the control group. Studies that provided other allergy prevention interventions for both treatment and control groups were eligible.

Types of outcome measures

Primary outcomes

All allergic diseases, including asthma, atopic dermatitis, allergic rhinitis and food allergy

Secondary outcomes

Asthma
Atopic dermatitis/eczema
Allergic rhinitis
Cow's milk or soy protein allergy
Food allergy
Urticaria
Anaphylaxis

In the 2017 review update, we no longer reported previously reported food hypersensitivity and potential harms, including growth parameters, cost and infant feed refusal. Definitions of allergic disease must be compatible with the World Allergy Organization 2003 consensus (Johansson 2004).

Researchers may have diagnosed a specific allergic disease on the basis of:

history of recurrent and persistent symptoms typical of the allergic disease;
clinician diagnosis of allergic disease; or
clinical allergy confirmed by testing including detection of allergen sensitisation by skin testing or by serological testing for specific IgE (e.g. radioallergosorbent (RAST), enzyme‐allergosorbent (EAST)), or asthma confirmed by respiratory function testing for the presence of bronchial hyper‐responsiveness confirmed by elimination/challenge.

Investigators used the following definitions of age of allergic disease.

Infant allergic disease incidence: allergic disease occurring up to two years of age.
Childhood allergic disease incidence: allergic disease occurring up to 10 years of age (or up to age of latest report between two and 10 years).
Childhood allergic disease prevalence: reported allergic disease that was present between two and 10 years of age.
Adolescent allergic disease: allergic disease present from 10 to 18 years of age.
Adult allergic disease: allergic disease present after 18 years of age.

Search methods for identification of studies

Electronic searches

We conducted a comprehensive search including: Cochrane Central Register of Controlled Trials (CENTRAL 2017, Issue 11) in the Cochrane Library; MEDLINE (1948 to 3 November 2017) and Embase (1974 to 3 November 2017). We also searched citations of authors of included studies and citation lists of articles and reviews. We did not apply language restrictions.    We searched clinical trials registries for ongoing or recently completed trials (clinicaltrials.gov; the World Health Organization’s International Trials Registry and Platform www.whoint/ictrp/search/en/, and the ISRCTN Registry).

We documented the search strategies in Appendix 1,Appendix 2, and Appendix 3.

Searching other resources

We performed a search of previous reviews including cross references (all articles referenced), abstracts, conferences (Paediatric Academic Societies 2003 to 2016; Perinatal Society of Australia and New Zealand 2003 to 2017), recent review citations and expert informants.

We also updated in November 2017 our search of clinical trials registries for ongoing or recently completed trials (clinicaltrials.gov; controlled‐trials.com; who.int/ictrp). Search strategies are detailed in Appendix 4 and Appendix 5.

Data collection and analysis

This review updates previous versions (Osborn 2003; Osborn 2006a; Osborn 2006b; Osborn 2017a; Osborn 2017b).

Each review author independently assessed eligibility of studies for inclusion. We included only studies with ≥ 80% reporting of randomised infants. We used the criteria and standard methods of the Cochrane Neonatal Review Group to assess the methodological quality of included trials regarding adequacy of randomisation and allocation concealment, blinding of parents or caregivers and assessors to intervention and completeness of assessment of all randomised individuals. We used a data collection form to aid extraction of relevant information and data from each included study. Each review author extracted the data separately, and review authors compared data and resolved differences by consensus. We used the standard methods of the Cochrane Neonatal Review Group to synthesise the data and expressed effects as risk ratio (RR), risk difference (RD) with 95% confidence intervals (CIs) for categorical data, and planned to use mean difference (MD) and 95% CIs for continuous data. We used the Chi² test to examine data for heterogeneity, and we quantified heterogeneity using the I² statistic. We used the fixed‐effect model for meta‐analysis when enrolled infants and interventions were similar and no significant heterogeneity was found. We explored sources of heterogeneity by performing subgroup analysis.

The term 'hydrolysed formula' used without a reference to type refers to both extensively and partially hydrolysed formulas (EHF/PHF). We did not pool studies that used hydrolysed formula for early (first few days of life) supplemental or sole infant feeding with studies that used hydrolysed formula for prolonged feeding. We performed all comparisons by including only studies with no different co‐interventions prescribed for prevention of allergic disease in either study arm (e.g. in treatment group but not in control group). Allergic disease‐preventing co‐interventions included modifications to mother's diet when pregnant or breast feeding and environmental modifications such as avoidance of pet hair and host dust mite reduction measures. The protocol did not originally pre specify that we should restrict analyses to studies with no differential co‐interventions.

Selection of studies

We included all randomised and quasi‐randomised controlled trials fulfilling the selection criteria described in the previous section. Each review author reviewed the search results and separately selected studies for inclusion. Review authors resolved disagreements by discussion.

Data extraction and management

Each review author extracted the data separately. Review authors compared data and resolved differences by consensus.

We obtained additional method details and data from the authors of two studies (Hill 2000; von Berg 2003). For one study (Lam 1992), we extracted methods and data from a thesis.

For the 2016 update, we performed all analyses using Review Manager software (Review Manager 2014).

Assessment of risk of bias in included studies

We used the criteria and standard methods of the Cochrane Neonatal Review Group to assess the methodological quality of included trials. We evaluated the quality of included trials in terms of adequacy of randomisation and allocation concealment, blinding of parents or caregivers and assessors to intervention and completeness of assessment in all randomised individuals.

For the 2016 and 2017 updates, we incorporated previous assessments into RevMan 5 'Risk of bias' tables. We assessed risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017) by examining the following.

Sequence generation (checking for possible selection bias)

Adequate (any truly random process, e.g. random number table; computer random number generator)
Inadequate (any non‐random process, e.g. odd or even date of birth; hospital or clinic record number)
Unclear

Allocation concealment (checking for possible selection bias)

Adequate (e.g. telephone or central randomisation; consecutively numbered sealed opaque envelopes)
Inadequate (open random allocation; unsealed or non‐opaque envelopes, alternation; date of birth)
Unclear

Blinding (checking for possible performance bias)

Adequate, inadequate or unclear for participants
Adequate, inadequate or unclear for personnel
Adequate, inadequate or unclear for outcome assessors

Incomplete outcome data (checking for possible attrition bias through withdrawals, dropouts, protocol deviations)

Adequate (< 20% missing data)
Inadequate
Unclear

Selective reporting bias

Adequate (when it is clear that all of the study’s prespecified outcomes and all expected outcomes of interest to the review have been reported)
Inadequate (when not all of the study’s prespecified outcomes have been reported; one or more reported primary outcomes were not prespecified; outcomes of interest were reported incompletely and so cannot be used; study failed to include results of a key outcome that would have been expected to have been reported)
Unclear

Other sources of bias

We assessed the possibility of other sources of bias (e.g. early termination of trial due to data‐dependent process, extreme baseline imbalance) as follows.

Yes.
No.
Unclear.

Overall risk of bias

We made explicit judgements about whether studies were at high risk of bias, according to the criteria given in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017). With reference to the criteria above, we assessed the likely magnitude and direction of bias, and whether it was likely to have an impact on study findings. We explored the impact of the level of bias by undertaking sensitivity analyses (see Sensitivity analysis section).

Measures of treatment effect

We used the standard methods of the Cochrane Neonatal Review Group to synthesise data and expressed effects as risk ratio (RR) and risk difference (RD) with 95% confidence intervals (CIs) for categorical data, and planned to use mean difference (MD) with 95% CIs for continuous data.

Unit of analysis issues

The unit of analysis was the individual participant (infant).

Dealing with missing data

We recorded missing data in 'Risk of bias' tables and assessed the effect of missing data by performing sensitivity analysis.

We performed all analyses by 'intention to treat' when data were available. When intention‐to‐treat data were not available, we reported data as infants assessed by group of assignment as well as losses after randomisation.

Assessment of heterogeneity

We used the two formal statistics described here.

Chi² test: to assess whether observed variability in effect sizes between studies is greater than would be expected by chance. This test has low power when the number of studies included in the meta‐analysis is small, so we planned to set the probability at the 10% level of significance.
I² statistic, to ensure that pooling of data is valid. We graded the degree of heterogeneity as follows: none (< 25%); low (25% to 49%); moderate (50% to 74%); or high (≥ 75%).

When we found evidence of apparent or statistical heterogeneity, we planned to assess the source of the heterogeneity by conducting sensitivity and subgroup analyses to look for evidence of bias or methodological differences between trials.

Assessment of reporting biases

We documented in the Characteristics of excluded studies table all studies that reported use of a prebiotic in a potentially eligible infant population but did not report allergic disease‐related outcomes. We assessed reporting and publication bias by examining the degree of asymmetry of a funnel plot.

Data synthesis

We used the fixed‐effect model and Mantel‐Haenszel methods for meta‐analysis.

Quality of evidence

We used the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach, as outlined in the GRADE Handbook (Schünemann 2013), to assess the quality of evidence for the following (clinically relevant) outcomes: all allergic disease; specific allergies including asthma, atopic dermatitis/eczema, allergic rhinitis, cow's milk or soy protein allergy, food allergy, urticaria and anaphylaxis.

Two review authors independently assessed the quality of the evidence for each of the outcomes above. We considered evidence from randomised controlled trials as high quality but downgraded the evidence one level for serious (and two levels for very serious) limitations on the basis of the following: design (risk of bias), consistency across studies, directness of evidence, precision of estimates and presence of publication bias. We used the GRADEpro GDT Guideline Development Tool to create ‘Summary of findings’ tables to report the quality of the evidence.

The GRADE approach yields an assessment of the quality of a body of evidence by one of four grades.

High: we are very confident that the true effect lies close to the estimate of effect.
Moderate: we are moderately confident in the effect estimate. The true effect is likely to be close to the estimate of effect but may be substantially different.
Low: our confidence in the effect estimate is limited. The true effect may be substantially different from the estimate of effect.
Very low: we have very little confidence in the effect estimate. The true effect is likely to be substantially different from the estimate of effect.

Subgroup analysis and investigation of heterogeneity

We performed subgroup analyses according to the following.

Infant risk of allergic disease: low‐risk infants (no family history of allergic disease in first‐degree relatives); high‐risk infants (family history of allergic disease in first‐degree relatives or high cord blood IgE level).
Extent of protein hydrolysis: EHF versus CMF; PHF versus CMF; EHF versus PHF. An EHF should meet the definition provided by the AAP Committee on Nutrition (AAP 2000) ‐ extensively hydrolysed proteins derived from cow's milk in which most of the nitrogen is present in the form of free amino acids and peptides ≤ 1500 kDaltons ‐ and should, at a minimum, ensure with 95% confidence that 90% of infants with documented cow's milk allergy (CMA) will not react with defined symptoms to the formula under double‐blind, placebo‐controlled conditions.
Indication for use: prolonged sole formula feeding; supplemental formula feeding; weaning from the breast with infant formula.
Method of ascertainment of allergic disease: clinical allergic disease confirmed by challenge testing or by testing for atopy (e.g. skin testing or serological testing for specific IgE, asthma confirmed by testing for presence of bronchial hyper‐responsiveness, food allergy confirmed by elimination/challenge). Included in this definition is clinical allergic disease in a patient for whom atopy has been confirmed by testing (e.g. asthma when atopy has been confirmed by skin prick testing or RAST for specific IgE); blinded measurement for allergic disease ‐ when measurement of outcome was blinded to treatment allocation (this analysis was not prespecified).
Type of protein hydrolysate used: partially hydrolysed whey formula versus CMF; partially hydrolysed casein formula versus CMF; extensively hydrolysed whey formula versus CMF; extensively hydrolysed casein formula versus CMF; hydrolysed soy formula versus CMF.

Sensitivity analysis

We prespecified a sensitivity analysis to determine whether review findings were affected by including only studies at low risk of bias, defined as those with adequate randomisation and allocation concealment and < 10% loss to follow‐up.

Results

Description of studies

This report updates a previous publication that was withdrawn (Osborn 2017a; Osborn 2017b). A data entry error in the review was identified that had potential to impact the review conclusions. This review corrects the data error and updates the search.

Results of the search

We performed searches on 22 January 2016 (see Figure 1 for study flow diagram). We searched MEDLINE using OVID and retrieved 198 reports (Appendix 1); CENTRAL and retrieved 242 reports (Appendix 3); and Embase and retrieved 199 reports (Appendix 2). We identified eight ongoing or unpublished studies (NCT01987154; NCT00936637; NCT01156493; NCT01735123; NCT01700205; NCT01210391; NCT01036243; Yin 2015) (see Appendix 4 and Appendix 5). We updated the search on 3 November 2017 and found one additional excluded study (Boyle 2016) and an additional 15‐year follow‐up report from a previously included study (von Berg 2003).

The search of MEDLINE, CENTRAL and Embase was updated from January 2016 to November 2017. After de‐duplication, 32 additional reports were reviewed. No additional reports of included studies were found. Five additional reports were found that were added to excluded studies (Boyle 2016; Knip 2010; Knip 2011; Scalabrin 2009; Xinias 2017). An expert informant identified an additional unpublished study, the author was unable to get sponsor approval for data release for (Arshad 1990) (see Characteristics of studies awaiting classification).

Included studies

We assessed 16 studies as eligible for inclusion ‐ see Characteristics of included studies table. A previously included study (Sorensen 2007) has been moved to excluded studies as it did not reported an allergic disease outcome. The search revealed an additional study that we assessed as an included study (Kwinta 2009). We found an additional publication (Hill 2000) of a study that we had previously assessed as an excluded study as it had reported a per‐protocol analysis with excess losses. We combined a new report of an 'intention‐to‐treat' analysis with trial data obtained from study authors and have assessed this trial as an included study. Data from sequential publications of the GINI study (von Berg 2003) beyond three years remain ineligible for inclusion in this review owing to excess losses to follow‐up. For the GINI study, data from a single centre (Wessel) has been used in this review as other centres had excess losses (> 20%) at all time points. We have moved to the excluded studies list five previously included studies (Maggio 2005; Picaud 2001; Sorensen 2007; Szajewska 2001; Vandenplas 1993) that have reported no outcome data able to be used in this review.

Types of infants enrolled

High risk of allergic disease: a total of 13 studies (Chirico 1997; de Seta 1994; Halken 2000; Hill 2000; Lam 1992; Mallet 1992; Marini 1996; Nentwich 2001; Oldaeus 1997; Tsai 1991; Vandenplas 1992; von Berg 2003; Willems 1993) enrolled infants at high risk of allergic disease on the basis of a history of allergic disease in a first‐degree relative and/or a high cord IgE level, although Lam 1992 did not report 'high‐risk' criteria.
Risk of allergic disease not specified: three studies did not enrol infants on the basis of risk of allergic disease; Juvonen 1996 enrolled healthy term infants, although 62% had a family history of allergic disease; Kwinta 2009 enrolled very low birth weight infants (≤ 1500 g); and Saarinen 1999 enrolled healthy term infants requiring supplemental feeding in hospital.
Low risk of allergic disease: no study reporting allergic disease outcomes enrolled infants at low risk of allergic disease.

Types of interventions

See Characteristics of included studies for types of formula used in each study.

Early short‐term feeding: hydrolysed formula versus human milk feeding ‐ low‐risk infants

Two studies (Juvonen 1996; Saarinen 1999) compared a hydrolysed formula versus pasteurised donor human milk used for early short‐term infant feeding. Juvonen 1996 gave sole bottle feeds for three days, then all infants were exclusively breast fed. Saarinen 1999 gave supplemental feeds when required while infants were in hospital (average four days). Mothers were then encouraged to breast feed.

Early short‐term feeding: hydrolysed formula versus cow's milk formula (CMF) ‐ low‐risk infants

Two studies (Juvonen 1996; Saarinen 1999) compared a hydrolysed formula versus CMF for early short‐term infant feeding. Juvonen 1996 gave sole bottle feeds for three days, then all infants were exclusively breast fed. Saarinen 1999 gave supplemental feeds when required while infants were in hospital (average four days). Mothers were then encouraged to breast feed.

Prolonged feeding: hydrolysed formula versus human milk feeding

We found no studies for this comparison.

Prolonged feeding: hydrolysed formula versus cow's milk formula (CMF)

Twelve studies compared prolonged supplemental or sole feeding with a hydrolysed formula versus CMF without differential co‐interventions (Chirico 1997; de Seta 1994; Hill 2000; Kwinta 2009; Lam 1992; Mallet 1992; Marini 1996; Oldaeus 1997; Tsai 1991; Vandenplas 1992; von Berg 2003; Willems 1993). Three studies (Chirico 1997; Marini 1996; Oldaeus 1997) reported additional allergy avoidance measures in both hydrolysed formula and CMF groups.

Prolonged feeding: hydrolysed formula versus cow's milk formula (CMF) ‐ low‐risk infants

Kwinta 2009 compared prolonged feeding with a hydrolysed formula versus CMF in low‐risk infants.

Prolonged feeding: hydrolysed formula versus cow's milk formula (CMF) ‐ high‐risk infants

Eleven studies (Chirico 1997; de Seta 1994; Hill 2000; Lam 1992; Mallet 1992; Marini 1996; Oldaeus 1997; Tsai 1991; Vandenplas 1992; von Berg 2003; Willems 1993) compared prolonged feeding with a hydrolysed formula versus CMF in high‐risk infants.

Prolonged feeding: partially hydrolysed formula (PHF) versus cow's milk formula (CMF)

Eleven studies (Chirico 1997; de Seta 1994; Kwinta 2009; Hill 2000; Lam 1992; Marini 1996; Oldaeus 1997; Tsai 1991; Vandenplas 1992; von Berg 2003; Willems 1993) compared prolonged feeding with a PHF versus CMF.

Prolonged feeding: extensively hydrolysed formula (EHF) versus cow's milk formula (CMF)

Four studies (Kwinta 2009; Mallet 1992; Oldaeus 1997; von Berg 2003) compared prolonged feeding with an EHF versus CMF.

Prolonged feeding: extensively hydrolysed formula (EHF) versus partially hydrolysed formula (PHF)

Four studies (Halken 2000; Nentwich 2001; Oldaeus 1997; von Berg 2003) compared prolonged feeding with an EHF versus a PHF.

Prolonged exclusive feeding: hydrolysed formula versus cow's milk formula (CMF)

Seven studies reported prolonged exclusive hydrolysed formula versus CMF (Chirico 1997; de Seta 1994; Kwinta 2009; Lam 1992; Marini 1996; Vandenplas 1992; Willems 1993).

Prolonged feeding: hydrolysed formula versus cow's milk formula (CMF) ‐ studies with blinded measurement

Four studies assessed allergic disease without knowledge of participant allocation (Kwinta 2009; Oldaeus 1997; Vandenplas 1992; von Berg 2003).

Prolonged feeding: hydrolysed formula versus cow's milk formula (CMF) ‐ studies at low risk of bias

We assessed no studies as eligible for inclusion in the sensitivity analysis of adequate study methods (adequate randomisation, allocation concealment and < 10% losses to follow‐up).

Prolonged feeding: partially hydrolysed whey formula versus cow's milk formula (CMF)

Nine studies (Chirico 1997; de Seta 1994; Hill 2000; Lam 1992; Marini 1996; Tsai 1991; Vandenplas 1992; von Berg 2003; Willems 1993) compared a partially hydrolysed whey formula versus CMF.

Prolonged feeding: partially hydrolysed casein‐containing formula versus cow's milk formula (CMF)

Two studies (Kwinta 2009; Oldaeus 1997) compared a PHF containing casein versus CMF.

Prolonged feeding: extensively hydrolysed whey formula versus cow's milk formula (CMF)

One study (von Berg 2003) compared an extensively hydrolysed whey formula versus CMF.

Prolonged feeding: extensively hydrolysed casein formula versus extensively hydrolysed whey formula

Two studies (Halken 2000; von Berg 2003) compared an extensively hydrolysed casein formula versus CMF.

No study compared a hydrolysed soy formula versus CMF. No study compared a hydrolysed soy formula versus hydrolysed CMF.

Types of outcomes

Definitions for allergic disease varied between studies, but usually required persistent or recurring symptoms and signs in the absence of another obvious clinical explanation. For definitions of 'all allergic disease' and type of allergic disease for each study, see Characteristics of included studies.

Studies (with timing) reporting clinician‐diagnosed allergic disease included Chirico 1997 (six months); de Seta 1994 (six and 24 months); Halken 2000 (six, 12 and 18 months); Hill 2000 (two and six to seven years); Juvonen 1996 (three years); Kwinta 2009 (five to seven years); Lam 1992 (six months); Mallet 1992 (one, two and four years); Marini 1996 (six months, one and three years); Nentwich 2001 (six and 12 months); Oldaeus 1997 (nurse examination at three, six, nine, 12 and 18 months and doctor visit at 18 months); Saarinen 1999 (mean age at follow‐up of 27 months, range 18 to 34 months); Tsai 1991 (one, two, four, six and 12 months); Vandenplas 1992 (12 months); and von Berg 2003 (12 months and three, six, 10 and 15 years ‐ note excess losses from three years of age).

Three studies reported using questionnaire‐diagnosed allergic disease: Hill 2000 (six to seven years); Kwinta 2009 (five to seven years); and Willems 1993 (three months and one year).

Seven studies reported specific food allergy (Halken 2000; Hill 2000; Juvonen 1996; Oldaeus 1997; Saarinen 1999; Vandenplas 1992; von Berg 2003).

Excluded studies

We excluded 86 studies and documented reasons for exclusion in the Characteristics of excluded studies table. Here, we document controlled trials of hydrolysed formula according to types of participants and reasons for exclusion.

Preterm or low birth weight infants

The following studies did not report allergic disease: Agosti 2003; Baldassarre 2017; Florendo 2009; Maggio 2005; Mihatsch 1999 (cross‐over trial); Mihatsch 2002 (excess losses); Pauls 1996 (abstract format only); Picaud 2001; Raupp 1995; Riezzo 2001; Rigo 1994b (unclear allocation); Szajewska 2001; and Yu 2014 (likely non‐random, cross‐over).

Term healthy infants

Akerblom 2005 enrolled infants at high risk of diabetes and did not report allergic disease; Akimoto 1997 reported allergic disease (non‐random); NCT00936637 did not report allergic disease; Berseth 2009 (multiple formula differences) did not report allergic disease; Borschel 2013 did not report allergic disease; Borschel 2014a (multiple formula differences) did not report allergic disease; Borschel 2014b (multiple formula differences) did not report allergic disease; Burks 2008 (excess losses) did not report allergic disease; de Jong 1998 used protein‐free formula; Decsi 1992 did not report allergic disease; Decsi 1998 did not report allergic disease; Exl 1998 reported allergic disease (non‐random); Fukushima 1997 reported allergic disease (excess losses); Hartman 1994 reported intolerance not allergic disease and reported unclear losses (abstract format only); Hernell 2003 (unclear allocation) did not report allergic disease; Keller 1996 (unclear allocation) reported allergic disease; Knip 2010 enrolled infants at high risk of diabetes and did not report allergic disease; Lasekan 2006 did not report allergic disease; Medjad‐Guillou 1992 (cross‐over trial) did not report allergic disease; Mennella 2011a did not report allergic disease; Moran 1992 reported excess losses; Paronen 2000 enrolled infants at risk of diabetes and did not report allergic disease; Rigo 1994a (unclear allocation) did not report allergic disease; Scalabrin 2009 (excess losses) reported adverse events including allergic disease; Schmelzle 2003 (excess losses) did not report allergic disease; Schmitz 1992 did not report allergic disease (excess losses); Schrander 1993 was observational; Silva Rey 1996 (unclear allocation; excess losses) reported in a thesis only; Staelens 2008 did not report allergic disease; Tariq 1998 was an observational trial; Knip 2011 enrolled infants at high risk of diabetes and reported adverse events at 10 years including asthma and other allergic disease (excess losses); Vaarala 1995 did not report allergic disease; Vaarala 2012 (enrolled infants at risk of diabetes) did not report allergic disease; and Vandenplas 1993 did not report allergic disease.

Term infants at high risk of allergic disease

Arshad 1990 was unable to get approval from the study sponsor to release data so remains unpublished. Arshad 1992a reported multiple differential allergic disease‐reducing co‐interventions; Barberi 1993 reported unclear allocation, excess losses, and allergic disease (abstract format only); Bergmann 1996a reported allergic disease and non‐random allocation; Boyle 2016 reported multiple differential allergic disease‐reducing co‐interventions (prebiotic and hydrolysed formula) and reported allergic disease; Chan 2002 reported allergic disease and excess losses; Chan‐Yeung 2000 reported multiple differential allergic disease‐reducing co‐interventions; Chandra 1989a reported data that could not be verified after allegations of fraud; Chandra 1989b reported data that could not be verified after allegations of fraud; D'Agata 1996 reported unclear allocation, imbalances between groups and allergic disease; Giovannini 1994 reported excess losses and did not report allergic disease; Halken 1992 reported allergic disease and excess losses; Hattevig 1989 trial of maternal allergen avoidance only; Kuo 2011 was observational; Iikura 1995 reported unclear allocation and imbalances between groups (abstract format only); Martinez‐Valverde 1993 unclear allocation reported in thesis only; Moran 1992 reported excess losses and allergic disease; Nentwich 2003 was an observational trial; Odelram 1996 reported excess losses and allergic disease; Porch 1998 reported excess losses; Schmidt 1995 was an observational trial; Shao 2006 reported multiple differential allergic disease‐reducing co‐interventions; Sorensen 2007 has not reported allergic disease data to date; Szajewska 2004 reported excess losses and allergic disease; Vandenplas 1988 reported unclear allocation and losses; Wopereis 2014 has not reported allergic disease to date (abstract format only); and Zeiger 1989 reported multiple co‐interventions and excess losses.

Infants with allergic disease, infantile colic, gastro‐oesophageal reflux symptoms or feed intolerance

Arikan 2008 did not report allergic disease; NCT01987154 enrolled infants with CMPI; Campbell 1989 did not report allergic disease (non‐random); Corvaglia 2013 did not report allergic disease; Hill 1995b did not report allergic disease; Høst 1991 challenge test in infants with reactions ‐ observational study; Lucassen 2000 did not report allergic disease; Nocerino 2012 did not report allergic disease (abstract format only); Savino 2003 was observational; Savino 2006 reported multiple formula differences; Taubman 1988 did not report allergic disease; Xinias 2017 did not report allergic disease; and Zeiger 1989 reported allergic disease and excess losses.

Risk of bias in included studies

We have summarised risk of bias in included studies in Figure 2; and Figure 3. Overall, we considered no studies to be at 'low risk' of bias.

'Risk of bias' graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

'Risk of bias' summary: review authors' judgements about each risk of bias item for each included study.

Allocation

Randomisation

Two studies reported an adequate method of randomisation (Hill 2000; von Berg 2003). Nine studies reported random allocation of infants but not the method of randomisation used (Chirico 1997; de Seta 1994; Kwinta 2009; Lam 1992; Mallet 1992; Marini 1996; Oldaeus 1997; Tsai 1991; Vandenplas 1992), so we assessed their risk as 'unclear'. Five studies reported quasi‐random methods of participant allocation, including Halken 2000 (by date of birth), Juvonen 1996 (by day of month), Nentwich 2001 (odd and even numbers), Saarinen 1999 and Willems 1993 (month of birth), so were assessed as at 'high risk' of bias.

Allocation concealment

Three studies are at 'low risk' of bias due to allocation concealment (Chirico 1997; Hill 2000; von Berg 2003). Nine studies were at 'unclear' risk including de Seta 1994 did not report method of allocation, Halken 2000 used quasi‐random allocation method but blinded intervention, Kwinta 2009 used unclear allocation method, Lam 1992 did not report method, Mallet 1992 did not report method, Marini 1996 did not report method, Oldaeus 1997 did not report method, Saarinen 1999 used quasi‐random allocation method but blinded intervention, Tsai 1991 did not report method, and Vandenplas 1992 (used unclear allocation method). We assessed three studies as having 'high risk' for allocation concealment ‐ Juvonen 1996 quasi‐random allocation, unblinded study, Nentwich 2001 quasi‐random allocation, unblinded prescribing and Willems 1993 quasi‐random allocation, unblinded study.

Blinding

Six studies reported blinding of participants and personnel (Halken 2000; Hill 2000; Kwinta 2009; Saarinen 1999; Vandenplas 1992; von Berg 2003); we assessed four studies as having 'unclear' risk, as they did not report details (Chirico 1997; de Seta 1994; Juvonen 1996; Lam 1992) and six studies as 'high risk as they were unblinded (Mallet 1992; Marini 1996; Nentwich 2001; Oldaeus 1997; Tsai 1991; Willems 1993).

Six studies reported blinding of measurement (Halken 2000; Kwinta 2009; Nentwich 2001; Oldaeus 1997; Saarinen 1999; Vandenplas 1992). Eight studies did not report blinding of measurement of clinical allergic disease (Chirico 1997; de Seta 1994; Hill 2000; Juvonen 1996; Lam 1992; Marini 1996; Tsai 1991; von Berg 2003). Mallet 1992 and Willems 1993 performed unblinded measurements.

Incomplete outcome data

We included in this review only studies that reported < 20% loss to follow‐up. Studies reported the following losses to follow‐up: Chirico 1997 ‐ unclear; de Seta 1994 ‐ none reported; Halken 2000 ‐ 20% at 18 months; Hill 2000 ‐ 7.3% at two years and 20% at six to seven years; Juvonen 1996 ‐ 10% at three years; Kwinta 2009 ‐ 16% at five to seven years; Lam 1992 ‐ 8% at six months; Mallet 1992 ‐ 5% to 8% at four months but > 20% at one to four years; Marini 1996 ‐ 13% at two years and 19% at three years; Nentwich 2001 ‐ 19% at 12 months; Oldaeus 1997 ‐ 9% at 18 months; Saarinen 1999 ‐ unclear, although all infants were reported to be seen routinely in well‐baby clinics; Tsai 1991 ‐ 9% at 12 months; Vandenplas 1992 ‐ 11% at 12 months and > 20% at three and five years; von Berg 2003 ‐ for infants born in Wesel: 14.5% at one year and 19% at three years (all other analyses and time points > 20%); and Willems 1993 ‐ 13% at one year.

Studies with < 10% loss to follow‐up included de Seta 1994 ‐ none reported, Hill 2000 ‐ 7.3% at two years, Lam 1992 ‐ 8% at six months, Mallet 1992 ‐ 5% to 8% at four months, Oldaeus 1997 ‐ 9% at 18 months and Tsai 1991 ‐ 9% at 12 months.

Selective reporting

Eight studies were at 'low risk' of reporting bias with prespecified primary outcomes reported (de Seta 1994; Hill 2000; Marini 1996; Nentwich 2001; Oldaeus 1997; Saarinen 1999; von Berg 2003; Willems 1993). We assessed five studies as being at 'unclear' risk of reporting bias (Chirico 1997; Juvonen 1996; Lam 1992; Tsai 1991; Vandenplas 1992), and three as being at 'high risk' (Halken 2000; Kwinta 2009; Mallet 1992).

Other potential sources of bias

Six studies had imbalances between groups after randomisation, so we considered them to be at 'high risk' of bias (Chirico 1997; de Seta 1994; Halken 2000; Juvonen 1996; Nentwich 2001; von Berg 2003). Six studies did not report sufficient details of baseline characteristics or described differences of 'unclear' importance (Kwinta 2009; Lam 1992; Marini 1996; Oldaeus 1997; Saarinen 1999; Willems 1993). We considered four studies to have well‐balanced groups after randomisation with no other identified source of bias (Hill 2000; Mallet 1992; Tsai 1991; Vandenplas 1992).

We assessed no studies as having 'low risk' of bias overall ('low risk' of selection bias, performance and measurement bias and attrition bias with < 10% loss to follow‐up).

Effects of interventions

See: Table 1; Table 2; Table 3

Summary of findings for the main comparison. Early short‐term feeding: hydrolysed formula versus human milk feeding ‐ low‐risk infants for prevention of allergic disease.

Early short‐term feeding of hydrolysed formula versus human milk for prevention of allergic disease
Patient or population: infants not selected for allergic disease risk.  Settings: hospitals.  Intervention: early short‐term feeding: hydrolysed formula versus human milk feeding for prevention of allergic disease
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect  (95% CI)	Number of participants  (studies)	Quality of the evidence  (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	Early short‐term feeding: hydrolysed formula vs human milk ‐ low‐risk infants
All allergic disease ‐ childhood (incidence)  Follow‐up: mean 3 years	Study population		RR 1.43   (0.38 to 5.37)	90  (1 study)	⊕⊝⊝⊝  very low^a,b,c	Quality of evidence downgraded due to risk of bias, imprecision and potential for reporting or publication bias.
	75 per 1000	108 per 1000  (29 to 405)
Asthma ‐ childhood (incidence)  Follow‐up: mean 3 years	Study population		RR 0.48   (0.05 to 4.41)	90  (1 study)	⊕⊝⊝⊝  very low^a,b,c	Quality of evidence downgraded due to risk of bias, imprecision and potential for reporting or publication bias.
	57 per 1000	27 per 1000  (3 to 250)
Eczema ‐ childhood (incidence)  Follow‐up: mean 3 years	Study population		RR 0.48   (0.05 to 4.41)	90  (1 study)	⊕⊝⊝⊝  very low^a,b,c	Quality of evidence downgraded due to risk of bias, imprecision and potential for reporting or publication bias.
	57 per 1000	27 per 1000  (3 to 250)
Food allergy ‐ childhood (incidence)  Follow‐up: mean 3 years	Study population		RR 1.43   (0.38 to 5.37)	90  (1 study)	⊕⊝⊝⊝  very low^a,b,c	Quality of evidence downgraded due to risk of bias, imprecision and potential for reporting or publication bias.
	75 per 1000	108 per 1000  (29 to 405)
Cow's milk allergy ‐ infancy (incidence)  Follow‐up: mean 27 months	Study population		RR 0.87   (0.52 to 1.46)	3559  (1 study)	⊕⊝⊝⊝  very low^c,d,e	Quality of evidence downgraded due to risk of bias, imprecision and potential for reporting or publication bias.
	17 per 1000	15 per 1000  (9 to 25)
Cow's milk allergy ‐ childhood (incidence)  Follow‐up: mean 3 years	Study population		RR 7.11   (0.35 to 143.84)	90  (1 study)	⊕⊝⊝⊝  very low^a,b,c	Quality of evidence downgraded due to risk of bias, imprecision and potential for reporting or publication bias.
	0 per 1000	0 per 1000  (0 to 0)
The basis for the assumed risk* (e.g. median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI)  CI: confidence interval; RR: risk ratio
GRADE Working Group grades of evidence  High quality: Further research is very unlikely to change our confidence in the estimate of effect  Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate  Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate  Very low quality: We are very uncertain about the estimate

Prolonged feeding: hydrolysed formula versus cow's milk formula for prevention of allergic disease
Patient or population: infants.  Settings: home.  Intervention: prolonged feeding of hydrolysed formula versus cow's milk formula for prevention of allergic disease
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect  (95% CI)	No of Participants  (studies)	Quality of the evidence  (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	Prolonged feeding: hydrolysed formula versus cow's milk formula
All allergic disease ‐ Infancy (incidence)  Follow‐up: 2 years	Study population		RR 0.88   (0.76 to 1.01)	2852  (8 studies)	⊕⊝⊝⊝  very low^1,2,3,4,5	Quality of evidence downgraded due to risk of bias, imprecision and risk of reporting or publication bias. Subgroup analyses found no differences in: low‐risk or high‐risk infants; infants fed a partially hydrolysed formula versus cow's milk formula; infants fed extensively hydrolysed formula versus cow's milk formula; or infants fed partially versus extensively hydrolysed formula. A reduction in infant allergy was found in subgroup analysis of studies that enrolled infants receiving prolonged exclusive formula feeding (typical RR 0.61, 95% CI 0.46 to 0.80; participants = 425; studies = 5; I² = 0%).
	274 per 1000	241 per 1000  (208 to 277)
All allergic disease ‐ Childhood (incidence)  Follow‐up: 3 years	Study population		RR 0.85   (0.69 to 1.05)	950  (2 studies)	⊕⊝⊝⊝  very low^1,2,3,4	Quality of evidence downgraded due to risk of bias, imprecision and risk of reporting or publication bias. Subgroup analyses found no differences in: high‐risk infants; infants fed a partially hydrolysed formula versus cow's milk formula; infants fed extensively hydrolysed formula versus cow's milk formula; or infants fed partially versus extensively hydrolysed formula. A single small study that enrolled infants receiving prolonged exclusive formula feeding reported a reduction in childhood allergy incidence (RR 0.42, 95% CI 0.19 to 0.90; participants = 78).
	353 per 1000	300 per 1000  (244 to 371)
Asthma ‐ Infancy (incidence)  Follow‐up: 2 years	Study population		RR 0.57   (0.31 to 1.04)	318  (4 studies)	⊕⊝⊝⊝  very low^1,3,4	Quality of evidence downgraded due to risk of bias, imprecision and risk of reporting or publication bias. Subgroup analyses found no differences in: high‐risk infants; infants fed a partially hydrolysed formula versus cow's milk formula; infants fed extensively hydrolysed formula versus cow's milk formula; infants fed partially versus extensively hydrolysed formula; or infants receiving prolonged exclusive formula feeding.
	175 per 1000	100 per 1000  (54 to 182)
Eczema ‐ Infancy (incidence)  Follow‐up: 2 years	Study population		RR 0.93   (0.79 to 1.09)	2896  (9 studies)	⊕⊝⊝⊝  very low^1,3,4,5	Quality of evidence downgraded due to risk of bias, imprecision and risk of reporting or publication bias. Subgroup analyses found no differences in: high‐risk infants; infants fed a partially hydrolysed formula versus cow's milk formula; infants fed extensively hydrolysed formula versus cow's milk formula; infants fed partially versus extensively hydrolysed formula; or infants receiving prolonged exclusive formula feeding.
	202 per 1000	187 per 1000  (159 to 220)
Rhinitis ‐ Infancy (incidence)  Follow‐up: 2 years	Study population		RR 0.52   (0.14 to 1.85)	256  (3 studies)	⊕⊝⊝⊝  very low^1,3,4	Quality of evidence downgraded due to risk of bias, imprecision and risk of reporting or publication bias. Subgroup analyses found no differences in: high‐risk infants; infants fed extensively hydrolysed formula versus cow's milk formula; infants fed partially versus extensively hydrolysed formula; or infants receiving prolonged exclusive formula feeding.
	58 per 1000	30 per 1000  (8 to 107)
Food allergy ‐ Infancy (incidence)  Follow‐up: 2 years	Study population		RR 1.42   (0.87 to 2.33)	479  (2 studies)	⊕⊝⊝⊝  very low^1,3,4	Quality of evidence downgraded due to risk of bias, imprecision and risk of reporting or publication bias. Subgroup analyses found no differences in: high‐risk infants; infants fed a partially hydrolysed formula versus cow's milk formula; infants fed extensively hydrolysed formula versus cow's milk formula; or infants fed partially versus extensively hydrolysed formula.
	109 per 1000	155 per 1000  (95 to 254)
Cow's milk allergy ‐ Infancy (incidence)  Follow‐up: 2 years	Study population		RR 2.31   (0.24 to 21.97)	338  (1 study)	⊕⊝⊝⊝  very low^3,6	Quality of evidence downgraded due to risk of bias, imprecision and risk of reporting or publication bias. Subgroup analyses found no differences in: high‐risk infants; or infants fed a partially hydrolysed formula versus cow's milk formula.
	7 per 1000	16 per 1000  (2 to 149)
The basis for the assumed risk* (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).  CI: Confidence interval; RR: Risk ratio;
GRADE Working Group grades of evidence  High quality: Further research is very unlikely to change our confidence in the estimate of effect.  Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.  Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.  Very low quality: We are very uncertain about the estimate.

Date	Event	Description
17 October 2018	New citation required and conclusions have changed	The conclusions of the review have changed.
17 October 2018	New search has been performed	To address a data entry error, the review was updated. The error was corrected and a new search was performed. No new studies for inclusion were found, but one follow‐up report of an included study was located.

Date	Event	Description
25 May 2017	Amended	A data entry error in the review has been identified that may impact the review conclusions. The review will be re‐published following revision and peer review.
31 October 2016	New search has been performed	See above
31 October 2016	New citation required and conclusions have changed	We identified 8 new ongoing or unpublished studies; we found 1 additional excluded study and an additional 15‐year follow‐up report from a previously included study; for 1 previously excluded study, trial authors provided additional data now included in the review. We added the GRADE method and SoF tables. We downgraded review conclusions.
18 September 2008	Amended	We converted this review to new review format
27 July 2006	New search has been performed	We reviewed the eligibility of all trials. We included several new studies and updated reports. We partially redid comparisons to better meet the objectives and methods specified in the protocol. Additionally, we performed previously specified subgroup analyses according to whether studies had blinded measurement for allergy and whether enrolled infants who were solely formula fed were fed 100% whey formula or casein‐containing formula (according to degree of hydrolysis)  Exclusion of 2 previously included trials and inclusion of a new large trial resulted in substantial changes to the review and to the review conclusions
27 July 2006	New citation required and conclusions have changed	We made substantive amendments

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 All allergic disease	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.1 Childhood (incidence)	1	90	Risk Ratio (M‐H, Fixed, 95% CI)	1.43 [0.38, 5.37]
2 Asthma	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
2.1 Childhood (incidence)	1	90	Risk Ratio (M‐H, Fixed, 95% CI)	0.48 [0.05, 4.41]
3 Eczema	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
3.1 Childhood (incidence)	1	90	Risk Ratio (M‐H, Fixed, 95% CI)	0.48 [0.05, 4.41]
4 Food allergy	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
4.1 Childhood (incidence)	1	90	Risk Ratio (M‐H, Fixed, 95% CI)	1.43 [0.38, 5.37]
5 Cow's milk allergy	2		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
5.1 Infancy (incidence)	1	3559	Risk Ratio (M‐H, Fixed, 95% CI)	0.87 [0.52, 1.46]
5.2 Childhood (incidence)	1	90	Risk Ratio (M‐H, Fixed, 95% CI)	7.11 [0.35, 143.84]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 All allergic disease	7		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
1.1 Infancy (incidence)	7	1729	Risk Ratio (M‐H, Random, 95% CI)	0.77 [0.59, 1.00]
1.2 Childhood (incidence)	2	510	Risk Ratio (M‐H, Random, 95% CI)	0.68 [0.31, 1.52]
2 Asthma	5		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
2.1 Infancy (incidence)	3	177	Risk Ratio (M‐H, Fixed, 95% CI)	0.61 [0.29, 1.28]
2.2 Childhood (incidence)	1	78	Risk Ratio (M‐H, Fixed, 95% CI)	0.38 [0.08, 1.84]
2.3 Childhood (prevalence)	2	727	Risk Ratio (M‐H, Fixed, 95% CI)	0.98 [0.73, 1.31]
3 Eczema	7		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
3.1 Infancy (incidence)	7	1608	Risk Ratio (M‐H, Fixed, 95% CI)	0.96 [0.80, 1.14]
3.2 Childhood (incidence)	2	510	Risk Ratio (M‐H, Fixed, 95% CI)	0.85 [0.61, 1.19]
3.3 Childhood (prevalence)	2	726	Risk Ratio (M‐H, Fixed, 95% CI)	0.94 [0.71, 1.26]
4 Rhinitis	3		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
4.1 Infancy (incidence)	2	115	Risk Ratio (M‐H, Fixed, 95% CI)	0.4 [0.09, 1.70]
4.2 Childhood (incidence)	1	78	Risk Ratio (M‐H, Fixed, 95% CI)	0.48 [0.04, 5.03]
4.3 Childhood (prevalence)	1	295	Risk Ratio (M‐H, Fixed, 95% CI)	0.90 [0.59, 1.37]
5 Food allergy	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
5.1 Infancy (incidence)	1	338	Risk Ratio (M‐H, Fixed, 95% CI)	1.31 [0.75, 2.30]
6 Cow's milk allergy	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
6.1 Infancy (incidence)	1	338	Risk Ratio (M‐H, Fixed, 95% CI)	2.31 [0.24, 21.97]

Methods	Single‐centre randomised controlled trial in Italy.
Participants	Inclusion criteria: infants of mothers with atopy (rhinitis, asthma, eczema or food intolerance) who "could not breast feed".
Interventions	Treatment (n = 21): partially hydrolysed cow's milk whey formula (Vivena HA‐Primigiorni HA).  Control (n = 14): CMF (brand not reported).  Co‐interventions (in all 'at‐risk infants'): avoidance of passive smoking, exposure to pets and mites, avoidance of nurseries, delayed weaning to 6 months of age.
Outcomes	Primary outcome(s): immunogenicity and antigenicity of pHWF in at‐risk infants, including RAST for milk and egg proteins, total and specific IgE and specific IgG and IgG4 subclass antibodies.  Other outcomes: eczema: defined as a pruritic, chronic or chronically relapsing dermatitis with typical features and distribution.  Follow‐up to 6 months (infant eczema incidence).
Notes	Trial of sole prolonged partially hydrolysed whey CMF and environmental allergen avoidance vs CMF.  Conflict of interest: none reported.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Method of sequence generation not reported.
Allocation concealment (selection bias)	Low risk	After informed consent was obtained from parents, infants considered at risk of atopy and who could not be breast fed were randomised (by the sealed‐envelope method) to treatment on the first day of life. Only infants who were exclusively breast fed, special formula fed or traditional formula fed and who were not exposed to passive smoking during the first 6 months of life were included in the study.
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Not reported.
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Not reported for clinical outcomes. Reported for measurement of sensitisation (specific IgE).
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Numbers of enrolled infants not reported. Only infants who received 6 months of allocated formula analysed.
Selective reporting (reporting bias)	Unclear risk	Specific clinical outcomes not reported to be prespecified.
Other bias	High risk	Unequal numbers in groups reported.

Methods	Multi‐centre quasi‐randomised (alternation) controlled trial in Denmark.
Participants	Inclusion criteria: infants with bi‐parental atopy or uniparental atopy and cord IgE ≥ 0.3 kU/L.
Interventions	Supplemental or sole formula feeding with:  Treatment 1 (n = 79): extensively hydrolysed casein formula (Nutramigen).  Treatment 2 (n = 82): extensively hydrolysed whey formula (Profylac).  Control (n = 85): pHWF (NAN‐HA).  Recommended duration of feeding: 4 months.  Co‐interventions (all infants): delay solids and cow's milk to 4 months; avoid smoke, pets, damp housing.
Outcomes	Primary outcome(s): allergic disease.  Other outcomes: physician examination at 6, 12 and 18 months (infant allergic disease).  Definitions  Any atopy: symptoms of asthma, atopic dermatitis, allergic rhino conjunctivitis or at least 2 episodes of allergic urticaria.  Asthma: clinician diagnosed, ≥ 3 episodes of recurrent wheezing needing bronchodilators.  Atopic dermatitis: physical examination, ≥ 3 months' duration.  Allergic rhino conjunctivitis: ≥ 1 month or recurrent symptoms.  Food allergy: confirmed by unblinded elimination/challenge.  CMA/CMPI: confirmed by unblinded elimination/challenge and exclusion of lactose intolerance and infection.
Notes	Trial of supplemental or sole pHWF vs extensively hydrolysed casein formula vs extensively hydrolysed whey formula in high‐risk infants.  Control group of non‐randomly allocated breast fed infants not included in analysis.  Conflict of interest: funded by the Danish Dairy Foundation. Companies provided formula and funding.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quasi‐random. Postnatal allocation by date of birth.
Allocation concealment (selection bias)	Unclear risk	Allocation predictable but intervention blinded.
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Used formula tins labelled 'A, B or C'.
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	‘The blinded coding of the products was not revealed until all the data registration was complete’.
Incomplete outcome data (attrition bias)   All outcomes	High risk	Unclear as to exact numbers in each group not completing the study. Of initial population of 595 infants, 92% were included in study and 80% completed follow‐up. Reasons for losses included parental refusal (19), received other formula in first days (23), 'dropped out' (36), not seen at 18 months, did not fulfil inclusion criteria (4) and non‐compliance (32).
Selective reporting (reporting bias)	High risk	Not intention‐to‐treat analysis.
Other bias	High risk	Some baseline differences between study groups (bi‐parental atopy: Nutramigen 38%, Profylac 22%, Nan HA 39%).

Methods	Randomised controlled trial in Australia.
Participants	Inclusion criteria: Mother‐baby pairs were enrolled if the unborn child had a first‐degree relative with a history of eczema, asthma, allergic rhinitis or food allergy.
Interventions	Treatment 1 (n = 206): pHWF (NAN HA; Nestle, Biessenhoffen, Germany). Treatment 2 (n = 208): soy‐based formula (ProSobee; Mead Johnson Nutrition/Bristol Myers, Melbourne, Australia). Control CMF (n = 206): CMF (NAN; Nestle, Tongala, Australia). Mothers were encouraged to initiate and maintain breast feeding for at least 6 months. Study formulas introduced at cessation or partial cessation of breast feeding or as a breast milk substitute if breast feeding was not intended. Approximately 50% of infants received some allocated formula by 4 months of age; 16.5% never received allocated formula because of continuing breast feeding (13.6%; n = 78/575) or use of a non‐allocated formula (2.9%; n = 17/575).
Outcomes	Childhood outcomes based on parent report during telephone interviews up to age 2 (every 4 weeks until 64 weeks, then at 78 and 104 weeks), and at 6 or 7 years. SPTs were performed at 6, 12 and 24 months according to a standard technique by 1 of 3 allergy‐trained research nurses. Primary aim: to determine incidence of allergic manifestations (eczema and food reactions) up to 2 years of age in high‐risk infants. Definitions Eczema: doctor‐diagnosed eczema or any rash that was treated with topical steroid preparation (excluding rash that affected only the scalp or nappy region).  Food reaction: within 2 hours of ingesting that food, child developed an acute skin rash (urticaria, angioedema, erythematous or morbilliform), a flare of pre‐existing eczema, signs of anaphylaxis or vomiting.  Any allergic manifestation: presence of eczema or food reaction within first 2 years of life.  Positive SPT: a wheal of at least 3 mm (mean) diameter with a positive (histamine) control. Childhood outcomes based on parent report during telephone interviews conducted when children were 6 or 7 years of age were defined as follows. Current childhood eczema: eczema diagnosed by family physician in previous 12 months. Current childhood asthma: asthma diagnosed by family physician in previous 12 months. Persistent childhood asthma: asthma diagnosed by family physician in previous 12 months on at least 2 occasions at follow‐up at 5, 6 or 7 years. Current childhood allergic rhinitis: one or more episodes of nasal discharge and/or congestion in the absence of an upper respiratory tract infection in the previous 12 months that family physician or parent attributed to allergic rhinitis (hay fever) and that was treated with an antihistamine and/or nasal steroid.
Notes	Sponsor: Nestec Ltd, a subsidiary of Nestle Australia, provided the study formula and staff funding for the first 6 years of the study. A.J. Lowe has received research support from Dairy Australia. K.J. Allen has received research support from Wyeth and Nutricia. M.J. Abramson has received research support from the National Health and Medical Research Council. D.J. Hill has received research support from Nestle Australia, SHS International and Nutricia Initial study reports were not intention‐to‐treat. Latest report is an intention‐to‐treat analysis so is now considered eligible for inclusion. Data obtained from study authors excluded enrolments in the first 9 to 10 months, when infants were not randomised to a hydrolysed formula. Executives from Nestlé Australia wrote to the editors of JACI on 12 May 2010 to raise scientific concerns about the study, including alleged discrepancies between the interim report and the publication accepted by JACI Before publication, and with full information on concerns raised by Nestlé, the following organisations reviewed these issues. Editorial Board of JACI. American Academy of Allergy Asthma and Immunology (AAAAI) Board of Directors. Committee on Publication Ethics (COPE – see http://publicationethics.org/). Mercy Hospital for Women Human Research Ethics Committee (HREC). Approving ethics committee for the MACS Project. Mercy Hospital for Women Board of Directors. Study authors documented that internal audits of study data included re‐entry of infant feeding data for all infants and a random audit of outcome data for 100/620 infants.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	An independent statistician created each of the computer‐generated allocation schedules.
Allocation concealment (selection bias)	Low risk	Mother‐baby pairs were allocated to the next sequential number as they were enrolled in the study and were assigned to the formula code allocated to that number.
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Staff were blind to allocation codes and to group of allocation at the time of outcome assessment. Cans of formula were labelled at an independent location. Parents of participants were informed of the identity of the assigned formula only after the child’s second birthday.
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Staff were blind to allocation codes and to group of allocation at the time of outcome assessment. However, random allocation list was available to research staff.
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	2 years: 45/620 (7.3%) excluded; 6 to 7 years: 125/620 (20%) excluded. Excluded as lost contact or parent refused further follow‐up. Low risk 2 years, high risk 6 to 7 years owing to excess losses.
Selective reporting (reporting bias)	Low risk	Prospective protocol registration (ACTRN12609000734268). Primary endpoints stated in the objectives reported in the results.
Other bias	Low risk	Infants allocated to CMF and pHWF groups similar on baseline risk factors. No differences between groups in terms of duration of exclusive breast feeding or age of introduction of solids Quote: "The first 97 infants were randomized to either the CMF or soy study groups. When the pHWFbecame available, a new random allocation series was generated with a higher proportion allocated to the pHWF to obtain equal numbers in each formula group." This review incorporates only infants allocated to the 3 groups contemporaneously.

Methods	Population derived (city of Malmo) quasi‐randomised (alternation) controlled trial in Sweden.
Participants	Inclusion criteria: 144 healthy term infants of pregnant mother volunteers. 62% had family history of atopy.
Interventions	Early sole feeding for 3 days. Subsequently, all infants exclusively breast fed.  Treatment 1 (n = 53): pasteurised human milk feeds from milk bank.  Treatment 2 (n = 38): casein EHF (Nutramigen).  Control (n = 39): CMF (Baby Semp).
Outcomes	Primary outcome(s): macromolecular absorption, antibody production and allergic symptoms.  Other outcomes: serum IgE at 4 days, 8 months, 1 and 2 years; SPT at 1 and 2 years; clinical allergic disease to 3 years (child allergic disease incidence).  Criteria for allergic disease diagnosis not reported.
Notes	Trial of early (first 3 days) sole human milk vs CMF vs extensively hydrolysed casein formula.  Use of volunteers meant possible selection of high‐risk infants.  Conflict of interest: unclear; work supported by several foundations; affiliations not reported.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quasi‐random, infants allocated according to day of the month.
Allocation concealment (selection bias)	High risk	Predictable allocation; method of concealment not reported.
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Not reported.
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Not reported.
Incomplete outcome data (attrition bias)   All outcomes	High risk	Group numbers unequal. 15/144 (10%) lost to 3 years.
Selective reporting (reporting bias)	Unclear risk	Aim to measure 'allergic symptoms'.
Other bias	High risk	Potentially clinically important imbalances between groups (maternal smoking, bi‐parental atopy).

Methods	Three‐centre randomised controlled trial in Italy.
Participants	Inclusion criteria: maternal questionnaire used to identify infants with well‐defined family history of allergic disease in either parent.
Interventions	Infants randomised were those whose mothers did not wish to breast feed or had insufficient milk.  Treatment (n = 48): 'moderately' HF (Nidina HA, Nestle).  Control (n = 47): CMF (Nan, Nestle).  Formula feeding advised to 5 months.  Co‐interventions (both groups): maternal cow's milk and food avoidance measures for breast feeding mothers. For infants, cow's milk and allergenic foods avoided to 1 year. Advice given to modify environmental exposure (smoking, pets, carpets, avoiding infant community care to 2 years).
Outcomes	Primary outcome(s): allergic manifestations and nutritional adequacy of formula.  Other outcomes: weight, length and head circumference at 6 months, 1 and 3 years. Physician‐diagnosed allergic disease.  Definitions  Atopic dermatitis: typical rash in at least 2 areas.  Recurrent wheezing: ≥ 3 episodes and physician diagnosed.  Recurrent urticaria: ≥ 2 episodes after exposure to particular antigen.  GI symptoms: vomiting and/or diarrhoea after exclusion of infection and lactose intolerance, not confirmed by blinded elimination/challenge.  Allergic rhinitis: ≥ 3 weeks rhinorrhoea.  RAST and SPTs also performed in affected individuals.  Follow‐up performed at 1, 2 (infant allergic disease) and 3 years (child allergic disease).
Notes	Trial of prolonged supplemental or sole 'moderately' hydrolysed whey formula vs CMF feeding in high‐risk infants.  Co‐interventions in both groups.  Conflict of interest: none reported.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Postnatal 'random' allocation of infants; method not reported.
Allocation concealment (selection bias)	Unclear risk	Method not reported.
Blinding of participants and personnel (performance bias)   All outcomes	High risk	quote: "Where artificial feeding was required... babies... randomly allocated to formula (a) or formula (b) ... (but the mothers were not blinded to the allocation)".
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Physicians "unaware of the dietary regimen", but insufficient information reported on blinding of personnel to allocation.
Incomplete outcome data (attrition bias)   All outcomes	High risk	Hydrolysed group losses 5 (10%) at 1 year, 6 (13%) at 2 years, 8 (17%) at 3 years.  Cow's milk group losses 6 (13%) at 1 year, 7 (15%) at 2 years, 9 (19%) at 3 years.
Selective reporting (reporting bias)	Low risk	Prespecified allergic disease outcomes. Standardised definitions.
Other bias	Unclear risk	Insufficient data reported at baseline for allergic disease risk factors between study groups.

Methods	Single‐centre quasi‐randomised (alternation) controlled trial in Czech Republic.
Participants	Inclusion criteria: pregnant women who themselves, husbands or children attended an allergic disease or dermatology outpatient clinic (i.e. family history of atopy in first‐degree relative).
Interventions	Mothers encouraged to breast feed for at least 6 months to avoid cow's milk and highly allergenic foods. Allocated sole or supplemental formula if unable to solely breast feed according to prenatal treatment allocation.  Treatment 1 (n = 37): partially hydrolysed whey CMF (Beba HA, Nestle, Denmark).  Treatment 2 (n = 35): extensively hydrolysed whey CMF (Hipp HA, Hipp GnbH, Gmunden, Austria).  Co‐interventions: all mothers encouraged to breast feed for 6 months, avoid cow's milk for first 6 months, introduce solids after 6 months and delay allergenic foods to after 12 months.  At 6 months: 24/37 fed PHF and 21/35 fed EHF.  At 12 months: 31/37 fed PHF and 28/35 EHF.
Outcomes	Primary outcome(s): antigen‐specific reactivity of mononuclear cells to cow's milk protein; cow's milk‐specific IgE and IgG; atopic skin symptoms.  Other outcomes: symptom diaries kept. Blinded paediatrician assessment for atopic dermatitis.  Reported weights up to 12 months (data not given).  Definitions  Atopic dermatitis: typical rash in at least 2 locations relapsing for at least 3 months' duration. Standardised score used (SCORAD).  Allergic disease reported at 6 and 12 months (infant allergic disease).
Notes	Trial of sole or supplemental feeding pHWF vs extensively hydrolysed whey formula in high‐risk infants unable to be completely breast fed in first 6 months. Conflict of interest: supported by research grants. The "study done independently of infant food companies".
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quasi‐random ‐ prenatal randomisation by odd and even numbers.
Allocation concealment (selection bias)	High risk	Prenatal randomisation by odd and even numbers. Postnatal allocation to formula if unable to fully breast feed.
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Paediatrician prescribing treatment aware of allocation. Formula not blinded.
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Second paediatrician unaware of allocation.
Incomplete outcome data (attrition bias)   All outcomes	High risk	1/73 (1%) post randomisation loss. 13/72 (18%) not fed HF and reported in separate group.
Selective reporting (reporting bias)	Low risk	Prespecified atopic skin symptoms in first 12 months of life. Standardised definitions.
Other bias	High risk	Groups appeared well balanced after allocation. However, not intention‐to‐treat analysis.

Methods	Population derived (well‐baby clinics in 3 towns) randomised controlled trial in Sweden.
Participants	Inclusion criteria: term newborn infants with 2 allergic family members or 1 allergic family member and cord IgE ≥ 0.5 U/L.  Mean age of weaning between 3 and 4 months.
Interventions	In infants weaning from the breast:  Treatment 1 (n = 55): extensively hydrolysed casein formula (Nutramigen, Mead Johnson).  Treatment 2 (n = 51): partially hydrolysed formula whey:casein ratio 60:40 (Mead Johnson).  Control (n = 49): CMF (Enfamil, Mead Johnson).  Co‐interventions: both groups advised to not smoke and avoid pets. Solids introduced after 4 months. Avoidance of cow's milk, eggs, fish and citrus until after 9 months.
Outcomes	Primary outcome(s): atopic and allergic disease at 18 months (infant allergic disease incidence).  Other outcomes: nurse examination at 3, 6, 9, 12, 18 months and doctor visit at 18 months. SPTS at each visit and specific IgE RAST at 9, 12, 18 months.  Definitions  Atopic dermatitis: standard scoring system used.  Food reactions: double‐blind placebo‐controlled challenges for formula milk reactions.  Asthma: recurrent wheeze with doctor confirmation.  Allergic rhinitis: doctor verified and allergen sensitisation proved.  Gastrointestinal allergic disease: positive unblinded oral challenge to food to which infant was sensitised.
Notes	Trial of extensively hydrolysed vs partially hydrolysed vs CMF for weaning high‐risk infants.  Conflict of interest: co‐investigator from formula company. Formula supplied by Mead Johnson. Study supported by Bristol‐Myers Inc.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Stratified by age at weaning. Infants randomised when weaning commenced. Method of sequence generation not reported.
Allocation concealment (selection bias)	Unclear risk	Method not reported.
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Identical coded tins used for extensively and partially hydrolysed formulas; no blinding of CMF tin.
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Quote: "Nurses and investigators blinded throughout the study".
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	14/155 post‐randomisation losses (9%) at 18 months. Feeding problems resulted in 11 losses (EHF 3, PHF 6, CMF 2).
Selective reporting (reporting bias)	Low risk	Primary endpoints reported in results.
Other bias	Unclear risk	Incomplete reporting of potential confounders in groups. Potential imbalance (furry animals at home: EHF 22%, PHF 6%, CMF 16%) between groups after allocation, although reported to be not statistically significant.

Methods	Three‐hospital quasi‐random (allocation by month) study in Finland.
Participants	Inclusion criteria: healthy full‐term infants requiring supplemental feeding in hospital.
Interventions	Early supplementary feeding in hospital with:  Control 1 (n = 1758): CMF (Tutteli, Vali, Finland).  Control 2 (n = 1844): pasteurised donor human milk.  Treatment (n = 1715): extensively hydrolysed whey formula (Pepti‐Junior, Nutricia, the Netherlands).  Average duration hospital stay 4 days.  Mothers encouraged to breast feed. Supplemental CMF used after discharge when required. Solids introduced at 4 to 6 months.  No co‐interventions.
Outcomes	Primary outcome(s): CMA.  Other outcomes: CMA ‐ mothers contacted researchers if symptoms suggestive of CMA appeared. Well baby clinics also informed of study (all infants seen average 8 times in first 12 months). Definitions  CMA: unblinded elimination/challenge performed.  Mean age follow‐up 27 months (range 18 to 34 months) (infant allergic disease).
Notes	Trial of early supplemental human milk vs extensively hydrolysed whey formula vs CMF.  Potential ascertainment bias as compliance with reporting not assessed.  Conflict of interest: supported by Nutricia.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quasi‐random allocation by month of birth and hospital born.
Allocation concealment (selection bias)	Unclear risk	Possible if blinding maintained.
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Colour‐coded bottles used.
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Blinding maintained until last follow‐up assessment.
Incomplete outcome data (attrition bias)   All outcomes	High risk	6209/6267 (99%) eligible mothers returned baseline questionnaire. Mothers were asked to call study author if symptoms of CMA appeared. Compliance not assessed.  Diary of infant feeding regimen returned by 118/118 mothers of infants subsequently found to be hypersensitive to CM and 76% CM‐tolerant infants. Well‐baby clinics of the area, in which every infant is seen an average of 8 times during the first 12 months of life, were informed of the study.
Selective reporting (reporting bias)	Low risk	Primary endpoint was an adverse reaction to challenge with CM.
Other bias	Unclear risk	Baseline characteristics not reported.

Methods	Multi‐centre randomised controlled trial in Germany.
Participants	Inclusion criteria: high risk of allergic disease in healthy infants with at least one first‐degree family member with allergic disease  Exclusion criteria: severe acquired or congenital diseases, gestation < 37 weeks, birth weight < 2500 g, > 14 days, intake cow's milk‐based formula before inclusion, incapability of parent to comply with study protocol.
Interventions	Mothers encouraged to breast feed for at least 4 months. Study formula provided for when sole breast feeding no longer continued and provided until infant 6 months of age. Infants (all centres: N = 2252; Wessel: n = 1087) randomised to:  Treatment 1 (all centres: n = 557; Wessel: n = 273): partially hydrolysed 100% whey formula (Beba HA, Nestle, Vevey, Switzerland)  Treatment 2 (all centres: n = 559; Wessel: n = 265): extensively hydrolysed 100% whey formula (Hipp HA, Hipp, Pfaffenhofen, Germany)  Treatment 3 (all centres: n = 580; Wessel: n = 281): lactose‐free, extensively hydrolysed 100% casein formula (Nutramigen, Mead Johnson, Diezenbach, Germany)  Control (all centres: n = 556; Wessel: n = 268): CMF with casein:whey ratio 40:60 (Nutrilon Premium, Nutricia/Numico, Zoetermeer, the Netherlands)  Co‐interventions: all groups received advice about breast feeding for at least 4 months, preferably 6; no dietary restrictions during lactation; not to feed solids during study period, and thereafter to add 1 food a week and avoid common allergenic foods in first year.  58.4% of infants received study formula
Outcomes	Primary outcome(s): allergic disease.  Other outcomes: allergic disease (atopic manifestations), asthma and eczema.  Definitions  Allergic disease (atopic manifestations) diagnosed at 12 months (infant allergic disease) as atopic dermatitis, allergic urticaria or gastrointestinal food allergy.  Atopic dermatitis: typical morphology and distribution of skin lesions; pruritus; chronicity (duration ≥ 14 days, chronically relapsing or both); confirmed on skin examination by a second specially trained allergist; severity rated using the SCORAD method.  Allergic urticaria: at least 2 episodes of itching eruptions or swelling with typical appearance, observed by parents or physician, caused by the same allergen. In case of a single episode, immunological evidence (specific SPT or allergen‐specific IgE level ≥ 0.35 KU/L or positive provocation response).  Gastrointestinal food allergy: suspected if GI symptoms not explained by any other condition, and if unblinded elimination challenge reproduced symptoms. GI allergic disease definite if a positive standardised elimination/challenge procedure. Double‐blind, placebo‐controlled food challenge performed in cases of uncertain reactions.  At 3 years, childhood allergic disease included atopic dermatitis, urticaria, food allergy with manifestation in the gastrointestinal tract and asthma.  Allergic asthma: diagnosed from parental report of relevant symptoms (wheeze and/or cough without infection) or regular use of asthma medication in the child's third year of life. Asthma symptoms included wheezing or cough for at least 2 weeks (acute laryngotracheitis excluded); exercise‐induced wheeze or cough at any time (with crying, laughing or activity); and episodes of wheezing or dry night‐time cough. At 6 years, reported physician‐diagnosed allergic diseases (atopic dermatitis, food allergy, allergic urticaria, asthma and hay fever/allergic rhinitis). At 10 years, reported physician‐diagnosed asthma, allergic or atopic eczema/dermatitis, hay fever/allergic rhinitis, urticaria and food allergy. Also, if present, at 10 years, parents reported a physician’s diagnosis during past 4 years, treatment in past 12 months or both. At 15 years, parent‐reported physician‐diagnosed asthma, allergic rhinitis (AR) and eczema, spirometric indices and sensitisation.
Notes	Data for intention‐to‐treat analyses of all infants (including breast fed infants) according to study centre provided by study authors for all allergic disease, asthma and eczema. Intention‐to‐treat 3‐year data for food allergy not provided.  Analyses meeting inclusion criteria for the review are intention‐to‐treat analyses including breast fed infants for all study centres at 1 year and infants enrolled in Wesel for 3‐year outcomes. Excess losses beyond 3 years, so data not included in analyses. Trial of prolonged breast feeding with supplemental or sole formula feeding when required comparing use of CMF, pHWF, extensively hydrolysed whey formula and lactose‐free extensively hydrolysed casein formula.  Sponsor: study supported by Federal Ministry for Education, Science, Research and Technology and the Child Health Research Foundation. Formulas provided by Nestle, Hipp, Milupa, Numico and Mead Johnson.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated list stratified for single or double parental atopy and study region.
Allocation concealment (selection bias)	Low risk	Random sequence generation and blinded intervention.
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Quote: "Blinding of parents and the study team was guaranteed by using identically labelled tins for the study formula coded with 4 different letters for each of the 4 formulas".
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	None reported.
Incomplete outcome data (attrition bias)   All outcomes	High risk	In intention‐to‐treat analyses for study centres (Munich and Wesel) that included breast fed infants: 1 year: 304/2252 (13.5% lost to follow‐up). 3 years: 692/2254 (31% lost to follow‐up). 6 years: 572/2252 (25% lost to follow‐up). 10 years: 1115 to 1145/2522 (44% to 45% lost to follow‐up). 15 years: Response rates were 61.1% (1377/2252) ‐ lost to follow‐up 875/2252 (39%) Breast fed infants randomised to interventions who did not receive interventions were followed up only in Wesel. In intention‐to‐treat analyses for Wesel only: 1 year: 158/1087 (14.5% lost to follow‐up). 3 years: 206/1087 (19% lost to follow‐up).
Selective reporting (reporting bias)	Low risk	Primary endpoints and specific timings stated in the methods were reported in the results in the original study report (von Berg 2003).
Other bias	High risk	Quote: ".. significant bias in the eHF‐C group because disproportionally more children had to be excluded as a result of noncompliance with the study formula (18% [47/257] in the eHF‐C group vs 10% to 12% in the other study formula groups, P = 0.02)".

Methods	Single‐centre quasi‐randomised (allocated by month) controlled trial in Belgium.
Participants	Inclusion criteria: infants not breast fed with a family history of allergic disease and cord IgE ≥ 0.5 IU/L.
Interventions	Prolonged sole formula feeding with:  Control: (n = 55): CMF.  Treatment (n = 67): pHWF (Nan HA).  Formula used for first 3 months, then unrestricted diet.  No co‐interventions.
Outcomes	Primary outcome(s): allergic disease.  Other outcomes: paediatrician‐administered questionnaire at 3 months and 1 year for allergic disease (infant allergic disease incidence).  Definitions  Allergic disease included eczema, asthma, recurrent episodes of bronchitis, persistent rhinitis, persistent gastrointestinal symptoms (excluding infection) and sleeping difficulties.  No specific definitions given.
Notes	High rate (45%) of non‐compliance with formula.  Trial of prolonged sole pHWF vs CMF in high‐risk infants.  Conflict of interest: unclear; co‐investigator from FNRS, Brussels.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quasi‐random ‐ infants postnatally allocated by even or odd month of birth.
Allocation concealment (selection bias)	High risk	Quasi‐random and unblinded.
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Formulas not masked.
Blinding of outcome assessment (detection bias)   All outcomes	High risk	None.
Incomplete outcome data (attrition bias)   All outcomes	High risk	17/135 (13%) high‐risk infants did not complete the study.
Selective reporting (reporting bias)	Low risk	Prespecified allergic disease.
Other bias	Unclear risk	Did not report baseline characteristics of groups.

Study	Reason for exclusion
Agosti 2003	AGA preterm infants allocated to low hydrolysed protein formula vs intact protein formula. Method of allocation not reported. Did not measure allergy. n = 20.
Akerblom 2005	242 newborn infants who had a first‐degree relative with type 1 diabetes and carried risk‐associated HLA‐DQB1 alleles. After exclusive breast feeding, the infants underwent a double‐blind, randomised pilot trial of either casein hydrolysate or conventional cow's milk‐based formula until the age of 6‐8 months. Did not report allergic disease or food allergy.
Akimoto 1997	Cohort study.
Arikan 2008	Randomised families with colicky infants to 4 different intervention groups ‐ massage, sucrose solution, herbal tea and hydrolysed formula and control group. Did not report allergy. n = 35 per group.
Arshad 1990	Randomised controlled trial in infants at high risk of allergy (family history of allergy in two or more of mother/father/sibling) receiving formula (n = 60 in each group). Cow’s milk formula versus hydrolysed formula (extensively hydrolysed casein formula Aptamil HA, Milupa) for the prevention of allergy. Reported allergy at 12 months. Completed, however author unable to get approval from sponsor to release data.
Arshad 1992a	Had multiple allergies preventing co‐interventions in treatment group and not in control group
Baldassarre 2017	RCT of infants 28‐33 weeks' gestational age; birth weight of 700 g to 1750 g and AGA within 24 hours of first enteral feeding (n = 65). Cow's milk protein versus EHF feeding. Did not report allergy. Reported feeding advancement and markers of feeding tolerance.
Barberi 1993	Infants at high risk of atopy randomised to HF vs CMF. 440 received allocated formula. Method of allocation not reported. Excess post‐randomisation losses ‐ 278/815 (34%). No intention‐to‐treat analysis. n = 815.
Bergmann 1996a	Infants at risk of atopy were assigned to 1 of 3 nutritional groups and were followed up through the first 6 months of life: 20 infants were breast fed, 28 received a predominantly whey protein‐based infant formula, 53 received an ultrafiltrated partially hydrolysed whey protein formula. Non‐random allocation. n = 101.
Berseth 2009	Double‐blind randomised controlled trial in healthy formula‐fed term infants with no family history of allergy of 70% lactose, partially hydrolyzed cow's milk protein formula supplemented with docosahexaenoic acid and arachidonic acid vs full lactose intact cow milk protein formula supplemented with docosahexaenoic acid and arachidonic acid. Did not report allergy. Measured adverse effects and tolerance. n = 335.
Borschel 2013	Healthy term infants randomised to an amino acid–based formula or an extensively hydrolysed casein‐based formula. Did not report allergy. n = 213.
Borschel 2014a	RCT in healthy term infants fed whey‐based, palm olein oil–free PHF vs whey‐based, palm olein oil‐containing PHF (not eligible comparison). Did not report allergy. n = 209.
Borschel 2014b	RCT in healthy term infants fed ready‐to‐feed or powdered form of an extensively hydrolysed casein‐based formula (not eligible comparison). Did not report allergy. n = 195.
Boyle 2016	RCT in infants at high risk of allergy fed PHF supplemented with prebiotic vs standard CMF without prebiotic. Not an eligible comparison. Reported allergy. n = 863.
Burks 2008	Randomised healthy term infants to amino acid‐based formula vs EHF. Excess post‐randomisation losses (33%). Reported adverse events. Did not report allergy. n = 165.
Campbell 1989	Enrolled infants with colic. Not randomised.
Chan 2002	Trial of sole prolonged partially hydrolysed CMF vs CMF in infants at high risk of atopy. Excess losses (28%). n = 110.
Chan‐Yeung 2000	Infants at high risk of allergy. Had multiple allergy‐preventing co‐interventions in treatment group and not in control group. n = 545.
Chandra 1989a	Original data unable to be verified.
Chandra 1989b	Original data unable to be verified.
Corvaglia 2013	Randomised preterm infants with symptomatic gastroesophageal reflux to extensively hydrolysed preterm formula vs standard preterm formula. Did not report allergy. n = 40.
D'Agata 1996	Infants at high risk of allergy. 30 exclusively breast fed, 50 hypoallergenic milk‐fed, 30 soy milk‐fed and 15 fed with conventional milk formula. Method of allocation unspecified. Substantial imbalances in numbers. Reported allergy.
de Jong 1998	Trial of early supplementation of CMF vs a protein‐free placebo formula (not hydrolysed protein) in breast fed infants
Decsi 1992	Healthy term infants who were breast fed or received conventional formula (Mildibe, EGIS; Pre‐Aptamil, Milupa), or a formula containing hydrolysed proteins (Aptamil H.A., Milupa). Did not report allergy. n = 10 per group.
Decsi 1998	Term infants fed conventional formula or PHF. Excess losses (27% hydrolysed formula group and 21% CMF group). Did not report allergy. n = 11 per group.
Exl 1998	Allocated to intervention (breast feeding, hydrolysed formula and delayed weaning) according to geographical area.
Florendo 2009	Preterm infants randomised to feeds with a standard non‐hydrolysed whey–casein vs partially hydrolysed whey preterm infant formula. Reported feed intolerance. Did not report allergy. n = 80.
Fukushima 1997	Trial of maternal allergen avoidance and infant supplemental or sole hydrolysed or CMF feeding when required. Differential losses with excess losses in maternal and infant hydrolysed formula group (27%) and in maternal hydrolysed formula and infant CMF group (23%).
Giovannini 1994	Study in infants at high risk of atopy given human milk vs soy formula vs whey‐based low‐degree hydrolysate vs casein‐based high‐degree hydrolysate vs soy plus collagen‐based high‐degree hydrolysate formula. Did not report allergy. Excess post allocation losses (56/138) not analysed in group of assignment (solely breast fed infants reported separately). n = 138.
Halken 1992	Trial of prolonged supplementary or sole extensively vs partially hydrolysed ultrafiltrated formula in high‐risk infants. Only infants who received hydrolysed formula included in analysis. Excess losses after allocation (24%). n = 158.
Hartman 1994	Randomised trial in healthy infants allocated to soy formula, CMF or PHF. Abstract only. Losses unclear. Reported intolerance and allergy but data not extractable from abstract. n = 510.
Hattevig 1989	Trial of maternal allergen avoidance. The diet for infants was the same in both groups. During first 6 months, only breast milk and or formula based on hydrolysed casein (Nutramigen) given.
Hernell 2003	Breast fed infants, infants fed hydrolysate formulas, infants fed milk formula. Method of allocation not reported. Allergy not reported. n = 55.
Hill 1995b	Enrolled infants with 'colic'. Randomised to casein hydroIysate or CMF. Did not report allergy. n = 151.
Høst 1991	Cohort study. Casein hydrolysate (Nutramigen) used in elimination/challenge.
Iikura 1995	Abstract form only. Method of allocation unclear. Substantial differences in group sizes.
Keller 1996	Allocation to various feeding regimens including hydrolysed formula performed by nurses 'at random'. "Maternal decision respected". Unlikely to be random allocation. Allergy reported.
Knip 2010	Randomised trial 230 infants with HLA‐conferred susceptibility to type 1 diabetes and at least one family member with type 1 diabetes to receive either a casein hydrolysate formula or a conventional cow’s‐milk–based formula (control) whenever breast milk was not available during the first 6 to 8 months of life. Did not report allergy outcomes.
Knip 2011	Enrolled infants at risk of type 1 diabetes mellitus. Intervention commenced at weaning. Excess losses reported: 2997 / 5156 randomised infants not included in analysis = 58%. Reported adverse events including asthma, other forms of allergy at 10 years of age.
Kuo 2011	Prospective, observational, uncontrolled cohort study in newborns who had at least 1 first‐degree family member with a history of atopy and could not breast feed. Fed with HF or CM for at least 6 months and monitored prospectively at 6, 18 and 36 months of age.
Lasekan 2006	Assessed growth efficacy, gastrointestinal tolerance and plasma biochemical measurements of healthy infants receiving a partially hydrolysed rice protein‐based formula vs CMF for the first 16 weeks after birth. Did not report allergy. n = 65.
Lucassen 2000	Enrolled infants with excessive crying. Randomised healthy, thriving, formula‐fed infants, < 6 months old, crying > 3 hours per day on at least 3 days per week to whey hydrolysate formula or standard formula. Did not report allergy. n = 43.
Maggio 2005	Randomised controlled trial of preterm formula with hydrolysed cow’s milk proteins reporting short‐term growth and urinary and plasma amino acid levels. Did not report allergy. n = 21.
Martinez‐Valverde 1993	No definition of allergic symptoms reported in first 4 months. In Spanish version of thesis, method of treatment allocation not extractable independently.
Medjad‐Guillou 1992	Cross‐over trial of CMF vs hydrolysed formula in healthy term infants. Did not report allergy.
Mennella 2011a	Healthy infants randomly assigned to CMF or PHF between 0.5 and 7.5 months of age. Did not report allergy. n = 79.
Mihatsch 1999	Cross‐over trial examining effects of hydrolysed formula on plasma amino acids and gastrointestinal transit time in preterm infants. Did not report allergy.
Mihatsch 2002	Randomised trial of partially hydrolysed preterm infant formula vs CMF in VLBW infants at low risk of atopy establishing enteral feeds. Excess post‐randomisation exclusions 48/135 (36%). Did not report allergy. n = 135.
Mitchell 1977	Trial of lactose hydrolysed milk, not protein hydrolysed. Enrolled slightly undernourished Aboriginal children < 3 years of age.
Moran 1992	Trial of supplementary or sole hydrolysed formula vs CMF in low‐risk infants. Excessive losses > 20% in both groups. n = 205.
NCT00936637	Enrolled health term infants assigned to one of three nutritional groups: 20 infants were breast fed, 28 received an adapted predominantly whey protein based infant formula, 53 received an ultrafiltrated, partially hydrolysed whey protein formula. Did not report allergy.
NCT01987154	Enrolled infants with presumptive cow’s milk allergic colitis. Randomly assigned to EHCF with LGG (Nutramigen LGG) or without (Nutramigen) (EHCF LGG).
Nentwich 2003	Observational study of infants fed a whey hydrolysate formula compared with exclusively breast fed controls.
Nocerino 2012	Randomised trial of standard formula vs partially hydrolysed whey formula in infants with diagnosed infant colic. Did not report allergy. Published as conference abstract only. n = unclear (52 reported).
Odelram 1996	Trial of extensively hydrolysed vs CMF for weaning of high‐risk infants. Excluded trial, as 13 losses in addition to 9 post‐randomisation exclusions (total 27%). n = 91.
Paronen 2000	Enrolled infants at high genetic risk for diabetes. Did not report allergy. n = 119.
Pauls 1996	Trial of formulas with hydrolysed vs non‐hydrolysed protein for starting enteral feedings in preterm infants < 1500 g. Did not report allergy. Only outcomes to day 6 reported. Reported in abstract format only.
Picaud 2001	Preterm infants randomly assigned to PHF or conventional formula. Did not report allergy. n = 16.
Porch 1998	Infants at high risk of allergy randomly assigned extensively hydrolysed casein formula (Nutramigen); partially hydrolysed whey formula (Good Start); and soy formula. Excess losses 51/181 (28%). n = 181.
Raupp 1995	Trial of sole hydrolysed formula vs CMF in low birth weight infants. Excess post‐randomisation losses. Allergy not reported. n = 39
Riezzo 2001	Randomised trial of standard and hydrolysate formulas in preterm infants. Allergy not reported. n = 36.
Rigo 1994a	Method of treatment allocation unclear. Allocated successive infants to formulas. Trial of 5 different types of hydrolysed formula in healthy term infants. Extent of hydrolysis not reported. Allergy not reported. n = 74.
Rigo 1994b	Trial of hydrolysate formula in preterm infants. Method of allocation not stated. Allergy not reported. n = 19.
Savino 2003	Observational study of whey hydrolysate formula enrolling infants with "minor feeding problems".
Savino 2006	Randomised to receive formula containing partially hydrolysed whey proteins, prebiotic oligosaccharides, with a high beta‐palmitic acid content, or standard formula and simethicone (multiple formula differences).
Scalabrin 2009	RCT of healthy term infants who received extensively hydrolysed casein formula, the same formula supplemented with Lactobacillus rhamnosus GG or partially hydrolysed whey:casein (60:40) formula supplemented with LGG (PHFLGG). Excess losses at all time points. Reported adverse events including allergy. n = 289.
Schmelzle 2003	Randomised healthy formula‐fed term infant trial of partially hydrolysed whey infant formula vs standard infant CMF. Excess losses ‐ 52 (34%). Allergy not reported. n = 154.
Schmidt 1995	Observational study in high‐risk infants (infants allocated formula at parents' discretion).
Schmitz 1992	Trial in normal breast fed infants of early supplementary hydrolysed formula vs CMF. Did not report allergy. Excess losses at 1 year. n = 256.
Schrander 1993	Cohort study of newborn infants to determine incidence of CMPI and response to Pregomin (Milupa) protein hydrolysate formula.
Shao 2006	Multiple dietary interventions including hydrolysed formula in treatment group for infants at high risk of atopy who were not able to be breast fed.
Silva Rey 1996	Trial of partially hydrolysed milk formula. Excess losses ‐ 124/276 (45%) ‐ 42 losses by 6 months and further 82 excluded post allocation. Method of allocation not reported. n = 276.
Sorensen 2007	Infants (n = 242) with a family history of allergy assigned to one of 3 randomised groups: weaned to either CMF, EHW or PHW formula. All infants were breast fed for 3 months, then study formulas was were given for 1 month. Published as abstract only. Allergy not reported to date.
Staelens 2008	Double‐blind randomised cross‐over study in healthy newborns fed CMF vs PHF vs EHF. Allergy not reported. n = 17.
Szajewska 2001	Low birth weight infants were assigned randomly to receive extensive protein hydrolysate preterm formula, partial protein hydrolysate preterm formula and standard preterm formula. Allergy not reported. n = 61.
Szajewska 2004	Randomised trial of extensively hydrolysed preterm formula vs partially hydrolysed preterm formula vs CMF in high risk for atopy preterm infants. Excess post‐randomisation losses at all times ‐ 22/90 (24%) at 4 to 5 months. n = 90.
Tariq 1998	Cohort study.
Taubman 1988	Enrolled infants with excessive crying ('colic'). Randomised to hydrolysed casein formula or counselling. Allergy not reported. n = 20.
Vaarala 1995	Double‐blind trial; 10 infants received cow's milk‐based formula and 10 received a casein hydrolysate formula until the age of 9 months. Allergy not reported. n = 20.
Vaarala 2012	Infants with HLA‐conferred susceptibility to type 1 diabetes randomly assigned to receive CMF (n = 389), whey‐based hydrolysed formula (WHF) (n = 350) or whey‐based FINDIA formula essentially free of bovine insulin (n = 365). Allergy not reported. n = 1104.
Vandenplas 1988	Retrospective study. Embedded intervention study. Trial of formula, breast milk or a hypoallergenic formula. Method of allocation reported to be "chronological". Losses unclear. n = 75
Vandenplas 1993	Randomised healthy term infants to whey intermediate hydrolysed formula vs CMF. Allergy not reported. n = 45.
Wopereis 2014	Infants at high risk of allergy randomised to PHF containing a specific mixture of oligosaccharides or to standard CMF for the first 6 months of life. Allergy not reported to date. Published abstract only. n = 108.
Xinias 2017	Enrolled infants with colic allocated to formula containing partially hydrolyzed whey protein, reduced lactose, Bifidobacterium lactis (B lactis) and galacto‐oligosaccharides. Allergy not reported.
Yu 2014	Likely non‐random allocation. Cross‐over design. Enrolled preterm infants. Did not report allergy.
Zeiger 1989	Trial in mothers and infants at high risk of atopy of maternal dietary avoidance in pregnancy and lactation, and infant allergen avoidance through encouragement of breast feeding with supplemental or weaning formula use and subsequent dietary restriction vs usual maternal diet and infant feeding with use of supplementary or weaning CMF. Excluded as multiple co‐interventions and excess losses.

Trial name or title	Digestive tolerance of slightly hydrolyzed starter infant formula with probiotics
Methods	Randomised controlled trial.
Participants	n = 480 Inclusion criteria Healthy infants 0 to 1 month old at point of enrolment. Birth weight < 2500 g and > 4500 g. Gestational age < 37 weeks and > 42 weeks. Infants who cannot be breast fed because of maternal status. Willing to exclusively consume assigned study formula with whey protein. Likely to be compliant. Mother/caregiver demonstrating an understanding of given information and ability to record requested data. Exclusion criteria Congenital illness or malformation. Significant prenatal and/or postnatal disease. Receiving systemic antibiotic treatment at time of enrolment. Infant with symptoms of allergy to cow's milk. Infant's family not expected to comply with treatment (feeding regimen). Participants who cannot be expected to comply with treatment. Currently participating or having participated in another clinical trial during the last month.
Interventions	Hydrolysed formula with probiotics. Acidified hydrolysed formula. Hydrolysed formula without probiotics. Standard infant formula.
Outcomes	Incidence of crying/fussing from 1 to 3 months. Growth and night sleep.
Starting date	December 2009; study completion date: April 2012.
Contact information	A/Prof. Boosba Vivatvakin
Notes	ClinicalTrials.gov Identifier:NCT01036243

Trial name or title	Hydrolized protein formula for premature infants
Methods	Randomised controlled trial.
Participants	n = 118 Inclusion criteria Birth weight 500 g to 1500 g and < 32 weeks' GA who survived more than 3 days. Not yet started enteral full feeds. Study explained and parent/caregiver demonstrating understanding of given information. Informed consent signed. Exclusion criteria Chromosomal anomalies. Major congenital anomalies (complex cardiac anomalies, congenital hydrocephalus, renal dysplasia). Congenital (e.g. jejunal atresia) and acquired (e.g. GI perforation) GI pathology precluding oral feed and/or requiring major surgical or medical intervention. Parental refusal. Prior enrolment into a conflicting clinical trial. Conflicting clinical trials will be those in which the intervention could modify the outcome of the present study, for example, studies conducted to accelerate feeds and or improve tolerance.
Interventions	Protein hydrolysed formula. Standard premature formula.
Outcomes	Primary outcome measure: time to achieve full feeds. Secondary outcome measures: postnatal days to achieve full feeds.
Starting date	July 2010; study completion date: September 2015.
Contact information
Notes	ClinicalTrials.gov Identifier: NCT01156493. No report to date found.

Trial name or title	Growth of infants fed an extensively hydrolyzed infant formula
Methods	Randomised controlled trial.
Participants	n = 282 Inclusion criteria Healthy newborn singleton infant. Full term (≥ 37 weeks' gestation). Birth weight between ≥ 2500 g and ≤ 4500 g. 14 ± 3 days of age on enrolment. Infant's mother elected not to breast feed. Baby exclusively formula fed a minimum of 3 days before enrolment. Study explained and written information provided to parent/caregiver. Informed consent signed (parent/legal representative). Exclusion criteria Congenital illness or malformation affecting infant feeding and/or growth. Suspected or known allergy to cow's milk protein. Significant prenatal and/or postnatal disease. Any readmission to hospital (except for hyperbilirubinaemia) before enrolment. Infant receiving oral or IV antibiotic therapy in the 10 days before enrolment. Infant receiving prescription medication (with exception of topical antibiotics and/or treatment for thrush) or frequently using over‐the‐counter medications except vitamin and mineral supplements. Infant receiving probiotics in the 7 days before enrolment. Infant currently participating in another clinical study. Infant's family in the Investigator's assessment cannot be expected to comply with the protocol.
Interventions	Extensively hydrolysed whey infant formula vs extensively hydrolysed casein infant formula.
Outcomes	Weight gain to 4 months. Mean weight gain (grams/d) from enrolment to 4 months of age.
Starting date	November 2010; study completion date: July 2013.
Contact information	Ricardo Sorensen, MD, Louisiana State University Health Sciences Center in New Orleans.
Notes	ClinicalTrials.gov Identifier: NCT01210391

Trial name or title	"Watch Your Baby Grow" Study (GRO)
Methods	Randomised controlled trial.
Participants	n = 144 Inclusion criteria Healthy term (≥ 37 and ≤ 42 weeks' gestation at birth), singleton infant. Birth weight between 2500 g and 4500 g. At time of enrolment, infant must be ≤ 14 days old (date of birth = day 0). Mother 18 years of age or older. Infant consuming a standard cow's milk protein infant formula and receiving a standard cow's milk protein infant formula for at least the last 2 days before enrolment. Exclusion criteria Mother with gestational diabetes during pregnancy. Infant with condition requiring infant feedings other than feeding CMF from a bottle. Infant with major congenital malformations (i.e. cleft palate, haemangiomas, extremity malformation). Infant with suspected or documented systemic or congenital infection (e.g. human immunodeficiency virus, cytomegalovirus). Infant with evidence of significant cardiac, respiratory, endocrinological, haematological, GI or other systemic diseases. For example, infant must not be receiving insulin or growth hormone. Dyad with relative (son, daughter, niece, nephew, cousin, aunt, uncle, sibling) of ancillary personnel connected with the study.
Interventions	Extensively hydrolysed infant formula. Standard CMF during first year of life.
Outcomes	Primary outcome measures: growth and energy balance.  Secondary outcome measures: Intake and feeding behaviours, genotype.
Starting date	October 2012. Estimated completion date: December 2017.
Contact information	Julie A. Mennella
Notes	ClinicalTrials.gov Identifier: NCT01700205

PERMALINK

Infant formulas containing hydrolysed protein for prevention of allergic disease

David A Osborn

John KH Sinn

Lisa J Jones

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Summary of findings

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Primary outcomes

Secondary outcomes

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Sequence generation (checking for possible selection bias)

Allocation concealment (checking for possible selection bias)

Blinding (checking for possible performance bias)

Incomplete outcome data (checking for possible attrition bias through withdrawals, dropouts, protocol deviations)

Selective reporting bias

Other sources of bias

Overall risk of bias

Measures of treatment effect

Unit of analysis issues

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Quality of evidence

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

Results

Description of studies

Results of the search

1.

Included studies

Types of infants enrolled

Types of interventions

Early short‐term feeding: hydrolysed formula versus human milk feeding ‐ low‐risk infants

Early short‐term feeding: hydrolysed formula versus cow's milk formula (CMF) ‐ low‐risk infants

Prolonged feeding: hydrolysed formula versus human milk feeding

Prolonged feeding: hydrolysed formula versus cow's milk formula (CMF)

Prolonged feeding: hydrolysed formula versus cow's milk formula (CMF) ‐ low‐risk infants

Prolonged feeding: hydrolysed formula versus cow's milk formula (CMF) ‐ high‐risk infants

Prolonged feeding: partially hydrolysed formula (PHF) versus cow's milk formula (CMF)

Prolonged feeding: extensively hydrolysed formula (EHF) versus cow's milk formula (CMF)

Prolonged feeding: extensively hydrolysed formula (EHF) versus partially hydrolysed formula (PHF)

Prolonged exclusive feeding: hydrolysed formula versus cow's milk formula (CMF)

Prolonged feeding: hydrolysed formula versus cow's milk formula (CMF) ‐ studies with blinded measurement

Prolonged feeding: hydrolysed formula versus cow's milk formula (CMF) ‐ studies at low risk of bias

Prolonged feeding: partially hydrolysed whey formula versus cow's milk formula (CMF)

Prolonged feeding: partially hydrolysed casein‐containing formula versus cow's milk formula (CMF)

Prolonged feeding: extensively hydrolysed whey formula versus cow's milk formula (CMF)

Prolonged feeding: extensively hydrolysed casein formula versus extensively hydrolysed whey formula

Types of outcomes

Excluded studies

Preterm or low birth weight infants

Term healthy infants

Term infants at high risk of allergic disease

Trial name or title	Early Dietary Intervention and later signs of beta‐cell Autoimmunity (EDIA)
Methods	Randomised controlled trial.
Participants	Inclusion criteria Infants up to 12 months of age. Infant's parents give signed consent to participate and their HLA genotype is eligible. Exclusion criteria Older sibling of the newborn infant included in the study. Multiple gestation. Parents unwilling or unable to feed the infant cow's milk‐based products for any reason (e.g. religious, cultural). Gestational age of newborn infant less than 35 weeks. Inability of the family to take part in the study (e.g. family had no access to the Study Centre or telephone). Newborn infant has a recognisable severe illness such as those due to chromosomal abnormality, congenital malformation, respiratory failure needing assisted ventilation, enzyme deficiencies, etc. Infant receiving any infant formula other than study formula or Nutramigen at the delivery hospital. No HLA sample drawn before the age of 8 days.
Interventions	Experimental: extensively hydrolysed casein formula. Experimental: cow's milk‐based infant formula.
Outcomes	Primary outcome measure: intestinal permeability determined at the age of 3, 6, 9 and 12 months with the lactulose/mannitol test. Secondary outcome measure: serum metabolic profile. Serum metabolic profile will be analysed with metabolomics at the age of 3, 6, 9 and 12 months.  Other outcome measures: intestinal microbiome.
Starting date	January 2013. Completion date August 2016.
Contact information	Mikael Knip, MD; mikael.knip@helsinki.fi
Notes	ClinicalTrials.gov Identifier: NCT01735123.

Trial name or title	Application effect of extensively hydrolyzed milk protein formula and follow‐up in preterm children with a gestational age of less than 34 weeks
Methods	Randomised single‐blind and controlled clinical trial.
Participants	Preterm children of gestational age less than 34 weeks who cannot be breast fed.
Interventions	Extensively hydrolysed milk protein (100% whey protein) formula vs preterm children’s formula until discharge from neonatal intensive care unit.
Outcomes	First endpoint: food intolerance in preterm children. Second endpoint variables include (1) time to achieve full enteral nutrition; (2) time of parenteral nutrition; (3) time of NICU stay; (4) meconium drainage time; (5) daily spontaneous faecal discharge conditions; (6) growth; (7) total protein, albumin, calcium, phosphorus and alkaline phosphatase; and (8) neonatal necrotising enterocolitis, cholestasis, parenteral nutrition associated with cholestasis and congenital hypothyroidism.
Starting date	2014
Contact information	Li‐Xing Qia; qiao_lixing@163.com
Notes	Chinese Clinical Trial Registry (http://www.chictr.org.cn/); ChiCTR‐IOR‐14005696