Our population of VLBW infants fed predominantly human milk experienced low rates of SGA at discharge, and average loss of one-half of a standard deviation in weight for gestational age between birth and discharge. Although increasing proportions of enteral intake as human milk were not associated with increasing detriment in weight z-score, infants fed >75% human milk experienced significantly larger decrease in weight z-score from birth to discharge when compared with infants fed < 75% (p = 0.03). In addition, among infants fed > 75% human milk, infants fed predominantly donor human milk demonstrated higher rates of SGA at discharge than those fed maternal milk or mixed donor and maternal milk (p = 0.08).
The study infants grew well overall by the definition of being AGA at birth (z-score −0.4, i.e.
percentile), and appropriate for postmenstrual age at discharge (z-score −0.94, i.e.
percentile). In-hospital growth failure, as defined by SGA status at 36 weeks’ postmenstrual age, has previously been described as a problem in the VLBW population. Dusick and colleagues described a population of 1433 infants from the NICHD Neonatal Research Network (NRN) born between 2000 and 2001. They determined that while only 16% of VLBW infants were SGA at birth, 89% met this definition at 36 weeks postmenstrual age [14
]. Those outcomes represented an apparent improvement in rates of growth failure among VLBW infants in the NRN, as 97% of 4438 infants admitted to NRN centers between 1995 and 1996 were SGA at postmenstrual age at 36 weeks [20
]. In contrast, 34% of our subjects were SGA at discharge, which occurred at a median postmenstrual age of 38.7 weeks, and only 21% of infants who were AGA at birth had become SGA by the time of discharge.
As reported by other investigators, predominantly human milk diets in our population (>75%) resulted in significantly slower growth, as defined by a statistically significantly larger decrease in weight z-score between birth and discharge, than diets containing <75% human milk, although no dose–response effect was noted between increasing proportion of human milk and growth. Schanler and colleagues studied a similar population of VLBW infants, comparing those fed an average of 84% of the in-hospital diet as fortified human milk to those fed solely preterm formula. They found that human-milk-fed infants experienced slower weight gain, and were 500 g lighter at discharge than their formula-fed peers [2
]. Using the Fenton chart to calculate z-scores for Schanler’s study results in a weight z-score change of −1.86 for the human milk group compared with a weight z-score change of −1.34 for the formula group. O’Connor and colleagues also found a negative dose–response relationship between human milk intake and in-hospital growth as measured by weight at term-adjusted age. Among their population of larger VLBW infants (mean GA 30 wks, 1275 g birthweight), infants fed > 80% fortified human milk weighed 500 g less at term-adjusted age than those fed solely formula [11
]. Negative change in weight z-score calculated using the Fenton chart from birth to term in their study was also related to human milk, with the largest decline seen among those fed the most human milk (−1.62), and the least decline seen with exclusive formula feeding (−0.64).
A predominantly donor milk diet was associated with higher rates of growth failure as defined by discharge weight < 10th
percentile for postmenstrual age (SGA) in our population, compared with predominantly maternal milk and mixed human milk diets (p = 0.08). Infants fed predominantly mixed human milk diets had the lowest rates of SGA of all infants in the subgroup of infants fed >75% human milk. Although all infants included in the >75% human milk subgroup analysis received >75% human milk, intake varied systematically. Infants in the mixed diet group received 75-80% human milk, while infants in the maternal and donor groups received 85-100% human milk. The rate of SGA in the mixed group (21%) is therefore closer to the rate of SGA for the 50-75% human milk group (25%) than that of the >75% human milk group considered as a whole (39%) (Tables and ). Sullivan and colleagues also reported growth outcomes in a population of VLBW infants fed >70% maternal milk, some of whom received a formula supplement as the remainder of the diet, and some of whom received donor human milk [3
]. Infants receiving donor human milk in this trial also received human milk human milk fortifier, in contrast to the bovine milk human milk fortifiers used both in all other studies discussed and in our subjects. Infants fed the donor milk supplement showed a trend toward slower weight gain than those fed a formula supplement (p = 0.13).
A possible explanation for slower growth rates among infants fed fortified human milk, in general, and donor milk, specifically, is the inadequate protein content of standard fortified human milk for the needs of VLBW infants. Growth of preterm infants is linearly related to protein intake [21
]. Enteral protein requirements to achieve in-utero growth rates in VLBW infants range between 4 g/kg/day for the <1000 g infants and 3.6 g/kg/day for 1200–1500 g infants [22
]. Despite the higher protein content of milk produced by mothers delivering preterm [23
] and routine use of human milk fortifiers, VLBW infants fed human milk experience a protein deficit during the initial months of life. This deficit has been estimated by Arsanoglu et al, who compared assumed protein intake for VLBW infants to actual intake, to be 0.5 to 0.8 g/kg/day [25
Inadequate protein intake is even more pronounced when donor milk is considered. Most donor human milk is obtained from donors who delivered term infants and have been lactating for several months, and thus does not contain the higher preterm protein levels. There is evidence that donor milk contains even lower protein content than standard mature human milk, possibly due to processing and handling. Wojcik et al studied 415 samples of unpooled donor milk from 273 donors and found that the median protein content was 1 g/dl and that 30% of samples contained < 1 g/dl [26
]. Mean protein content in 39 samples from donors to the Mother’s Milk Bank of Ohio was 0.9 mg/dl [27
]. Using standard powdered bovine HMF with donor milk thus results in a diet severely deficient in protein, i.e. 2.8-3 g/kg/day, increasing the risk for growth failure.
Growth can be optimized with human milk in general, and donor milk specifically, if this relative protein deficiency is recognized and diets are modified to provide adequate protein intake. Protein intake can be optimized in either individualized or group methods [28
]. The approach now taken in our unit is to increase protein intake for all infants, with target intake of 3.6-4 g/kg/day, without attempting to individualize intake. Our findings of lower rates of growth failure than other investigators is likely attributable to higher protein intake in our predominantly human-milk-fed subgroup of VLBW infants. However, as multi-component fortifiers were used to achieve target protein intakes (bovine human milk fortifiers and term formula concentrate) in our population, overall caloric content of the milk was also increased in addition to protein. Some of the growth results obtained with our approach may be due simply to higher caloric density of the milk, however, protein intake has been found to be strongly linked to growth in VLBW infants [21
This study’s strengths include the prospectively collected growth and nutritional information and the use of z-scores, which allow for comparison of growth outcomes to a recognized standard. Our study also includes a significant number of VLBW infants fed donor human milk, and allows us to compare predominantly donor milk to predominantly maternal milk diets.
Our study’s weaknesses include the limited sample size and observational nature. Using an observational study design did not allow us to control enteral intake or compare different protein intakes in a randomized fashion. Use of donor milk or specific protein fortification schemes may have been systematically associated with other infant factors that impacted growth, creating residual confounding. In addition, we collected weight data only at birth and discharge, and were not able to assess the timing of the lowest weight z-score experienced by the infants, or how the trajectory of weight growth may have changed from day to day or week to week during hospitalization. We do not assume that the weight z-score at discharge represents the lowest weight z-score, we only report the change between birth and discharge. We were also unable to assess the effects of donor or maternal human milk on linear growth in our population, as standardized length data were not obtained. Ramel and colleagues have recently demonstrated that poor in-hospital linear growth is associated with poor cognitive outcomes in former VLBW infants at age 2, and that length z-scores for gestational age were lower than weight z-scores in their population [30
]. Future studies should include standardized length measurements. Another limitation of our study is that we did not measure caloric intake or protein intake directly, and therefore we are unable to report the intakes needed to achieve the growth results we demonstrated.