Human milk is the recommended nutritional source for full-term infants for at least the first six months of postnatal life (54th WHA). It is known that in this group of infants, breast milk supplies adequate substrate to meet the infant's nutritional demands, as well as supplying the infant with other substances that may afford some physiological advantage (for example, immunoglobulins and gastrointestinal hormones). Breast feeding may also contribute to maternal-infant bonding.
However, the role of human milk in premature infants is less well defined. The nutrient content of premature human milk provides insufficient quantities of protein, sodium, phosphate and calcium to meet the estimated needs of the infant (Schanler 2001). In addition, large fluid volumes may be required to provide sufficient calories to maintain adequate growth.
Observational studies have shown that premature infants fed human milk have lower growth rates than infants fed term or preterm infant formulae (Atkinson 1983; Cooper 1984, Roberts 1987). Serum albumin and blood urea nitrogen concentrations may decline in premature infants as a result of inadequate dietary protein intake. Premature infants are born with low skeletal stores of calcium and phosphate, and have very high requirements for these minerals if they are to attain adequate postnatal skeletal growth. Poor radiological bone mineralisation, rickets, and fractures have been described in premature infants receiving inadequate dietary intakes of calcium and phosphate, as may be supplied by breast milk alone.
Despite these apparent inadequacies of human milk, other studies have demonstrated that feeding human milk to premature infants may lead to benefits in both the short-term (for example, a lower risk of necrotising enterocolitis (Lucas 1990)) and long-term (for example, improved cognitive and neurodevelopmental outcomes (Morley 1998)).
Commercially-produced multicomponent fortifiers are available for the supplementation of breast milk. These fortifiers provide additional nutrients in the form of protein, calcium, phosphate, and carbohydrate, as well as vitamins and trace minerals. However, many of the nutrients contained within commercial preparations have not been studied either individually or in combination. This review includes trials where infants received more than one nutrient supplement (that is, protein and/or fat and/or carbohydrate and/or minerals). This intervention was prespecified prior to the literature search although it is appreciated that this would lead to a range of potential combined interventions. Other reviews have evaluated the effects of individual components given alone - that is, protein (Kuschel 1999a), carbohydrate (Kuschel 1999b), fat (Kuschel 1999c) or minerals (Kuschel 2001).
To determine if addition of multicomponent nutritional supplements to human milk leads to improved growth, bone metabolism and neurodevelopmental outcomes without significant adverse effects in premature infants.
Criteria for considering studies for this review
Types of studies
Controlled trials utilising either random or quasi-random patient allocation.
Types of participants
Premature infants receiving care within a nursery setting.
Types of interventions
All randomized controlled trials evaluating the supplementation of human milk with multiple nutrients (more than one of the following components: protein, fat, carbohydrate, or minerals [calcium and/or phosphate]), in which treatment was compared with unsupplemented human milk, are included. Supplementation with electrolytes, vitamins, or trace minerals in addition to only one of the above has not been classified as multicomponent fortification for the purposes of this review.
Types of outcome measures
1. Primary outcomes
a. Growth to discharge
b. Size at 12-18 months
c. Bone metabolism
Serum alkaline phosphatase (ALP)
Bone mineral content (BMC)
d. Neurodevelopmental outcomes
Neurodevelopmental outcomes at 18 months
2. Secondary Outcomes
a. Bone metabolism
b. Nitrogen retention studies
c. Adverse effects
Significant hypercalcemia (>2.85mmol/l)
Necrotizing enterocolitis (NEC)
Search methods for identification of studies
Searches of the Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library, Issue 2, 3003), MEDLINE up to August 29, 2003 (using the search terms infant, premature and milk, human), previous reviews including cross references, abstracts, conferences and symposia proceedings, expert informants, journal handsearching mainly in the English language.
Data collection and analysis
The criteria and standard methods of the Cochrane Collaboration and its Neonatal Review Group were used to assess the methodological quality of the included trials.
Additional information was requested from the authors of each trial to clarify methodology and results as necessary.
Each author extracted the data separately, compared data, and resolved differences.
The standard method of the Neonatal Review Group was used to synthesize the data.
Description of studies
Details of the included studies are included in the Table 'Characteristics of Included Studies'. Thirteen studies met the inclusion criteria (Modanlou 1986; Carey 1987; Gross 1987 (1); Gross 1987 (2); Greer 1988; Pettifor 1989; Polberger 1989; Kashyap 1990; Zuckerman 1994; Lucas 1996; Wauben 1998; Nicholl 1999 and Faerk 2000).
The types of human milk fortifier (HMF), as defined in this overview, varied from supplementation with a commercial preparation (containing protein, fat, carbohydrate, minerals, electrolytes, and trace minerals) to supplementation with only two individual components. Modanlou 1986 used a fortifier containing protein, carbohydrate, and minerals (Mead Johnson, preparation not specified). Carey 1987 used HMF containing protein and minerals only, but quantities of these were not specified. Gross 1987 (1) and Gross 1987 (2) used Similac Special Care (Ross Laboratories) preterm formula as supplementation, or a powdered HMF. Greer 1988 used a preparation from Ross Laboratories (powdered fortifier, unspecified), as did Pettifor 1989 (Ross Laboratories Human Milk Fortifier) and Kashyap 1990 (preparation not specified). Polberger 1989 supplemented with both protein and fat, and infants in both the control and treatment arms received mineral supplementation. Zuckerman 1994 used an equal mix of maternal milk and a preterm formula (Alprem, Nestle). Lucas 1996 used Enfamil HMF (Mead Johnson). Wauben 1998 used a non-commercial fortifier produced by Wyeth-Ayerst. Nicholl 1999 used Nutriprem human milk fortifier (Cow and Gate Nutricia). Faerk 2000 used a commercial preparation (Milupa Eoprotin).
There was variation between studies in the entry criteria. All studies based entry on birthweight (generally <1850g), although the limits for these were variable. Almost all studies excluded infants with congenital abnormalities or significant illness. Fortification was commenced in most studies once the infant tolerated a prespecified enteral intake. For almost all studies fortification ceased at a specified weight (generally 1800 to 2000g) or at discharge, although, for some studies, the duration of intervention is unclear.
The daily intakes of the various individual components varied between studies, as did enteral intakes. In some studies, there was little or no difference in caloric intakes between the groups (Gross 1987 (1); Gross 1987 (2); Greer 1988; Zuckerman 1994). Details of the individual components are included in the table "Characteristics of Included Studies". In addition, some studies supplemented the control arm with minerals (Polberger 1989; Lucas 1996; Wauben 1998; Faerk 2000)
Excluded studies are listed in the Table "Characteristics of Excluded Studies". Ronnholm 1982 included a treatment group that received both protein and fat supplementation, and was therefore eligible for inclusion. However, it was impossible to extract data from the published reports. Similarly, the data were not presented in an extractable form in Venkataraman 1988. Boehm 1991 compared a commercial preparation (EOPROTIN) with a control group supplemented with albumin, minerals, and sodium, thereby comparing essentially a different form of protein intake. Moyer-Mileur 1992 compared two fortifier preparations but did not have an unsupplemented control group, as also was the case with Metcalf 1994; Schanler 1995; Sankaran 1996; dos Santos 1997; Porcelli 2000 and Reis 2000. Ewer 1996 and McClure 1996 did not report any of the prespecified clinical outcomes, looking only at gastric emptying. Plath 1988 reported in abstract form the results of nitrogen balance studies. Gupta (unpublished) was excluded because of concerns about randomisation and selection criteria. Lucas 1984 was considered for inclusion in this review, particularly as this study was included in a previous systematic review of infant feeding (Sinclair 1992). Although infant nutrition was supplemented, it was by the substitution of maternal milk when insufficient quantities were available. This was not felt to represent "fortification" as such.
Risk of bias in included studies
Eight of the thirteen studies report results for fewer than 15 infants in each arm. Only Lucas 1996; Wauben 1998; Nicholl 1999 and Faerk 2000 included sample size estimates as part of the study design.
Pettifor 1989 and Zuckerman 1994 used quasi-random allocation via hospital number. Other studies used sealed envelopes (Modanlou 1986; Gross 1987 (1); Gross 1987 (2); Polberger 1989; Kashyap 1990; Lucas 1996; Nicholl 1999; Faerk 2000) or random number tables (Greer 1988). The method of randomisation is unknown for Carey 1987 and Wauben 1998.
Polberger 1989 conducted a double blind study. The assessment of neurodevelopmental and long term growth outcomes for Lucas 1996 was masked, but short-term outcomes were not. In other studies, there was no masking of the intervention (Modanlou 1986; Greer 1988; Pettifor 1989; Zuckerman 1994; Wauben 1998) or masking was unknown.
Most studies have focussed on relatively well infants. Infants who developed significant illness were frequently not enrolled or not included in results (Modanlou 1986; Carey 1987; Gross 1987 (1); Gross 1987 (2); Greer 1988; Pettifor 1989; Polberger 1989; Zuckerman 1994; Wauben 1998; Nicholl 1999). Polberger 1989 withdrew one control infant for apnoea, and two treatment infants (feed intolerance; apnoea) [additional information provided by author]. Zuckerman 1994 withdrew three infants in the control group because of incorrect feeding. Faerk 2000 withdrew 9 infants randomized (6 in the fortifier group and 3 in the phosphorus group), and a further 18 (9 in each group) because DEXA scans were not technically satisfactory. In only six studies (Modanlou 1986; Pettifor 1989; Kashyap 1990; Lucas 1996; Wauben 1998; Nicholl 1999) are the outcomes reported for all infants enrolled. In other studies, either the number of infants enrolled or the reasons for withdrawal are unknown. Where infants have been withdrawn because of feed intolerance, NEC, or death, results have been included if possible.
Lucas 1996 supplemented the control group with phosphorus and, in both groups, provided premature formula if there was insufficient maternal milk. Similarly, Wauben 1998 supplemented control infants with calcium and phosphate. Polberger 1989, primarily assessing caloric supplementation, provided both treatment and control groups with calcium and phosphate. Faerk 2000 also supplemented the control group with phosphorus. These interventions in the control groups may reduce any differences attributable to treatment. The results of this overview have been subjected to a sensitivity analysis excluding these studies.
Effects of interventions
These studies report results for more than 600 infants. There is significant variability between studies in the outcomes of weight gain and linear growth and blood urea levels (all trials included, but not with the sensitivity analysis), head growth, ALP activity, and BMC. This potentially reflects differences in fortifier composition, study design (for example, exclusion criteria, variable enteral and caloric intakes, duration of intervention), and outcome measures (primarily, timing of outcome measurements).
Short-term growth parameters
All studies evaluated short-term growth. The two largest studies (Pettifor 1989; Lucas 1996) did not demonstrate a statistically significant increase in weight gain in the fortification group. Nevertheless, the overall analysis demonstrated greater weight gains in infants receiving fortification (WMD 2.3g/kg/day, 95% CI, 1.7 to 2.9g/kg/day). The difference in daily weight gain remained statistically significant at 3.6g/kg/day (95% CI, 2.7 to 4.6g/kg/day) when the studies where control infants received mineral supplementation are excluded.
For those studies that reported weight gain as g/day, there was a significant increase in the infants receiving fortified feeds (WMD 4.7 g/day, 95%CI 2.8 to 6.7 g/day).
Infants receiving fortifier had greater length gain by 0.12cm/week (95% CI, 0.07 to 0.18cm/week). When the sensitivity analysis was performed, the difference in weekly length gain remained statistically significant at 0.18cm/week (95% CI, 0.08 to 0.28cm/week).
Head growth was also greater in those infants receiving HMF (WMD 0.12cm/week, 95% CI 0.07 to 0.16cm/week). The sensitivity analysis did not significantly alter this finding (WMD 0.14cm/week, 95% CI, 0.09 to 0.20cm/week).
Long-term growth parameters
Two studies (Lucas 1996; Wauben 1998) evaluated long term growth. Wauben 1998 did not demonstrate any differences in weight, length or head circumference at 12 months of corrected age. Similarly, Lucas 1996 did not demonstrate any differences in growth parameters at 18 months corrected age.
Serum alkaline phosphatase
There was no effect on mean ALP activity in the infants studied (WMD 0.2IU/l, 95% CI -34.0 - 34.4IU/l). This result did not change with the sensitivity analysis (WMD -43.2 IU/l, 95% CI -98.3 to 11.8 IU/l).
Bone mineral content
Modanlou 1986; Gross 1987 (1);Gross 1987 (2) found that BMC values were not statistically different between control and treatment groups, although no absolute values are available. BMC has been recorded in two different measurement units. From the two studies where data of radius BMC are available in the format of mg/cm, infants receiving HMF had higher BMC than those receiving unsupplemented milk (WMD 8.3mg/cm, 95% CI 3.8 to 12.8mg/cm). This result is heavily influenced by the study by Pettifor 1989 which contributed 59 of the 79 infants and demonstrated a difference of 12.0 mg/cm (95% CI 6.3 to 17.7mg/cm). The lack of absolute data from those individual trials where there was no difference between groups considerably reduces the confidence of this result. Wauben 1998 and Faerk 2000 - both of whom supplemented their control groups with phosphorus - evaluated whole body BMC and found no difference (WMD 1.7g, 95% CI -1.7 to 5.0g). Wauben 1998 also demonstrated no difference in BMC at 12 months of age.
Only Lucas 1996 evaluated developmental performance at 18 months. There was no statistically significant difference between intervention and control groups.
No studies addressed this outcome. Zuckerman 1994 evaluated wrist radiographs taken at hospital discharge and at the final follow-up visit, and found no difference between the supplemented and supplemented groups in the frequency of periosteal reaction, osteopenia, or rickets.
Nitrogen retention studies
Two studies (Kashyap 1990 and Wauben 1998) have demonstrated increased nitrogen retention in infants receiving HMF containing protein (WMD 66mg/kg/day, 95% CI 35 to 97mg/kg/day). Sensitivity analysis does not significantly change this result.
Although most studies evaluated serum calcium levels, only Lucas 1996 and Wauben 1998 evaluated absolute hypercalcemia (>2.85mmol/l and >2.7mmol/l, respectively). There was no difference between the treatment and control groups, although both studies supplemented the control groups with minerals.
Many studies withdrew infants with feed intolerance and did not report results. Modanlou 1986 and Lucas 1996 both evaluated feed intolerance, finding no difference between the groups, but the outcomes could not be numerically analysed. On the basis of the small number of infants for whom this outcome is reported, there is a non-significant trend towards an increased risk of feed intolerance in treated infants (RR 2.85, 95% CI 0.62 to 13.1).
No study specifically addressed this outcome. Lucas 1996 found that infants receiving HMF were more likely to have "hard stools" than the control group.
There is no significantly increased risk of NEC in infants receiving fortified human milk (RR 1.33, 95% CI 0.7 to 2.5). Sensitivity analysis does not significantly alter this result.
Lucas 1996 demonstrated a statistically significant reduction in pH in infants receiving HMF (pH 7.33, vs. pH 7.34 in controls - WMD -0.01, 95%CI -0.02 to 0.00) which is unlikely to have any clinical significance. Wauben 1998 withdrew one control infant because of metabolic acidosis.
Urea levels are significantly increased in infants receiving HMF (WMD 0.27mmol/l, 95% CI 0.14 to 0.40mmol/l). When the studies evaluating mineral supplementation in the control group are excluded, this difference is increased (0.96mmol/l, 95% CI 0.56 to 1.36mmol/l).
Death as a specific outcome is reported by Pettifor 1989 and Lucas 1996. Other studies included only relatively well infants. There does not appear to be any increased risk of death associated with fortification of human milk (RR 1.48, 95% CI 0.66 to 3.34), although all 7 infants who died in one study (Pettifor 1989) were assigned fortifier. Sensitivity analysis, excluding Lucas 1996, results in inclusion of Pettifor's study only with a trend towards increased risk of death (RR 13.3) but with very wide confidence intervals (95% CI, 0.78 to 227).
This overview has demonstrated that fortification of human milk with more than one nutritional supplement (caloric, protein, and/or mineral) results in small but statistically significant increases in weight gain, linear growth, and head growth over the short term study periods evaluated. No long term advantage has been shown in terms of either growth (Lucas 1996, Wauben 1998) or neurodevelopmental outcome (Lucas 1996).
Short-term growth is a difficult outcome to assess - particularly if the first two weeks of life, when weight loss is common, are included in the overall weight gain results. Although the differences for these outcomes are small, the effect of these small increases in growth over the short term may be cumulative. For prolonged hospital stays, a small advantage in weight gain or head or linear growth may have a significant impact on growth parameters at discharge or even age at discharge. However, these outcomes were not evaluated in this review. Two studies reported these outcomes (Modanlou 1986; Wauben 1998) and found no difference between the groups.
Fortification of human milk has no effect on ALP levels. BMC has been variably reported and only one study (Pettifor 1989) has individually shown a difference. The two most recent studies (Wauben 1998 and Faerk 2000) did not demonstrate any differences in whole body BMC, although phosphorus was given to the control groups in both trials. Fractures have not been reported as an outcome in any study. Nitrogen retention has been examined in two studies and is significantly increased in infants receiving fortifier.
Potential adverse effects of fortification do not appear to be significantly increased, although the total number of infants studied and the unavailability of results for some infants randomized and subsequently withdrawn makes it difficult to be confident of this finding. There is no evidence of a significantly increased risk of NEC. Urea levels are higher and pH levels lower in infants receiving fortification, but the clinical significance of this is not clear. The increased urea levels in the fortifier group are not above the accepted range of normal . If anything, excluding Lucas 1996, those in the control group are low and the higher levels in fortified infants may reflect improved dietary protein intake. There are insufficient data to evaluate other potential adverse effects.
Implications for practice
There is sufficient evidence to demonstrate that fortification of human milk with more than one nutritional component is associated with short-term improvements in weight gain, linear and head growth. There is no clear effect on bone mineral content. There is no evidence that these short-term gains in growth lead to any demonstrable long-term benefits in growth, bone mineral content, or neurodevelopmental outcomes, although this may well be related to the absence of follow-up in almost all studies. There does not appear to be any increase in clinically significant adverse effects in supplemented infants, although the total number of infants studied is small and the abstractable data from the published studies is limited.
Implications for research
Fortification of human milk has become common practice, based largely on metabolic studies evaluating the composition of human milk and the nutritional requirements of preterm infants. There is an absence of evidence of long-term benefit, and insufficient evidence to be reassured that there are no deleterious effects. Despite this, it is unlikely that further studies evaluating fortification of human milk versus no supplementation will be performed. Indeed, Lucas 1996 felt that it was not ethical to withhold phosphorus supplementation in control infants and other studies since then have also supplemented the control groups (Wauben 1998; Faerk 2000).
Most commercially available fortifiers contain varying amounts of protein, carbohydrate, calcium, phosphate, other minerals (zinc, manganese, magnesium, and copper), vitamins, and electrolytes. The benefits of many of these individual components have not been evaluated in a controlled manner. Further research should be directed toward comparisons between different proprietary preparations and evaluating both short-term and long-term outcomes and adverse effects, in search of the "optimal" composition of fortifiers. This has, in part, been addressed by studies excluded from this overview (Moyer-Mileur 1992; Metcalf 1994; Sankaran 1996; Porcelli 2000; Reis 2000). The number of study subjects required to adequately evaluate these outcomes would be extremely large.
The reviewers wish to thank those authors who were able to provide additional information to assist with this review.
Data and analyses
- Top of page
- Authors' conclusions
- Data and analyses
- What's new
- Declarations of interest
- Sources of support
- Index terms
Last assessed as up-to-date: 28 August 2003.
Protocol first published: Issue 4, 1998
Review first published: Issue 4, 1998
Declarations of interest
Sources of support
- National Women's Hospital, Auckland, New Zealand.
- University of Auckland, Auckland, New Zealand.
- No sources of support supplied
Medical Subject Headings (MeSH)
MeSH check words
Humans; Infant, Newborn
* Indicates the major publication for the study