There is now almost universal agreement that the first-choice milk for extremely-low-birth-weight (ELBW; < 1,000 g) infant feeding is breast milk expressed by the infant's mother. Human milk (HM), however, is not always available. We have developed a new, high-lactose formula (HLF) for the early enteral feeding advancement of ELBW infants during the first 2 weeks of life if mother's milk is not available. The HLF was not designed to meet all the nutritional needs of ELBW infants (similar to human milk). The HLF was designed primarily to facilitate gastric emptying and intestinal passage while the the infants are still receiving parenteral nutrition to meet all the nutritional needs. Hydrolyzed cow milk protein was used because it has been shown to promote gastric emptying (1,2) and to accelerate the gastrointestinal transport (3) compared with untreated cow milk protein. The overall formula composition was based on the data of Grütte (4), who has shown in rats and infants that a formula (similar to human milk) with at least 55 g/L lactose, protein less than 20 g/L, phosphate less than 450 mg/L (protein bound ≤ 140 mg/L), and calcium less than 400 mg/L induced a stool pH below 6 and frequent loose yellow bowel motions, which are well known in breast-fed infants. There is a phosphate-catalyzed mutarotation between α- andagr;-and β-anomers of lactose in the gut, and Grütte has shown that only α-lactose is easily digested. Given a low phosphate concentration, the mutarotation slows down and lactose enters the large bowel, where it is metabolized to short-chain fatty acids by the large bowel flora and thereby accelerates the intestinal transport (4).
However, it has been suggested that feeding tolerance in premature infants can be improved by replacing lactose with maltose (5). The aim of this study therefore was to compare the effect of a low lactose formula (LLF) with high phosphate with the effect of the new formula, HLF, on the feeding tolerance in ELBW infants. We hypothesized that HLF would increase the cumulative feeding volume in the first 2 weeks of life at least by 20% in at least 60% of matched pairs.
MATERIALS AND METHODS
This randomized, multicenter trial was conducted in the neonatal intensive care units of seven German tertiary care hospitals over a period of 13 months. All (N = 152) ELBW infants born within the study period were eligible for the study. Exclusion criteria were major congenital malformations and known gastrointestinal abnormalities. All infants received parenteral nutrition, including phosphorus and calcium. Parenteral nutrition was gradually decreased with increasing enteral intake. The study period covered the first 2 weeks of life only (the period of feeding advancement) because as soon as feeding was established (150 mL · kg−1 · d−1), human milk was fortified and the formulas were changed to a nutritionally adequate preterm infant formula. In each center, the infants were enrolled within 48 hours of birth, stratified by gestational age (< 28 vs. ≥ 28 weeks), and allocated randomly to HLF or LLF (sealed envelopes, computer-generated randomization lists).
Characterization of the Study Formulae
The HLF (Table 1) was designed as a low-protein, low-phosphate, and low-calcium diet with a lactose content of 55 g/L. Hydrolyzed cow milk protein (whey:casein = 60:40; 75.0% less than 1500 D, 16.5% 1,500–3,500 D, 5.0% 3,500–6,000 D, and 3.5% more than 6,000 D) was used. The HLF was supplied as a sterile liquid formula in 50-mL bottles. A new bottle was taken daily for every individual infant and stored at 4°C. For every feeding, a suitable aliquot was taken from the individual's bottle.
In contrast to HLF, LLF (Table 1) contained a negligible lactose concentration and higher concentrations of calcium, phosphate, and protein (hydrolyzed cow milk whey protein, 63–74% less than 500 D, 12–15% 500–2,500 D, 11–15% 2,500–6,000 D, and 0.1% more than 6,000 D, product information). The LLF was supplied as a powder, and a suitable volume of formula was prepared daily for every infant. Further handling was identical with that for HLF.
At 48 hours of age, bolus gavage milk feeding was started (nasogastric tube). Human milk feeding was encouraged and formula was fed if human milk was not available. The formulas were masked, but color and taste were different. The infants were fed every 2 hours, starting at 12 mL · kg birth weight−1 · d−1. Feedings were advanced every 24 hours by 12 mL · kg birth weight−1 · d−1 whenever more than 50% of the calculated amount was given during the previous 24 hours. Gastric residuals were checked before each feeding. In infants with a birth weight of 750 g or less, the scheduled feed was given at a gastric residual volume of 2 mL or less. In infants with a birth weight of 751 g to 1000 g, the critical gastric residual volume was set to 3 mL. With gastric residuals more than 2 mL (3 mL), the difference up to the scheduled amount had been given. No milk was fed if the gastric residual was more than 2 mL (3 mL), and the volume of the feed was equal to or less than the gastric residual volume. The color of the gastric residuals, the subjective impression of those in charge, and clinical conditions such as hypotension, infection, or Indomethacin therapy did not influence the feeding strategy. After extubation or intubation, feedings were withheld for 6 hours. There were no enemas and no laxatives during the study.
Data Collection and Analysis
Demographic variables were recorded for all infants. Every second hour, beginning at 48 hours of age and ending after the fourteenth day of life, regurgitation or vomiting, the volume and color of gastric residuals, and the feeding volume were recorded as secondary outcome variables. Necrotizing enterocolitis (NEC) was diagnosed according to the Bell stages (6). At the end of the study, weight, length, and head circumference were recorded again.
Main outcome measure was the cumulative milk feeding volume (CFV) from day 3 to day 14 to integrate the feeding volume tolerated on every single study day in one parameter. We hypothesized that infants who were fed with HLF would have at least a 20% higher CFV than infants who were fed with LLF in at least 60% of matched pairs. A 20% higher CFV was regarded as clinically significant. The hypothesis was tested using a one-sided sequential plan for matched pairs (7). For analysis, consecutively enrolled infants were paired within each stratum at the coordinating study center. The results were analyzed by intention to treat.
For calculation of the equations of the two boundary lines in the sequential plan, the α error and the β error were set at 0.05. The desired proportion of pairs in which the CFV of HLF was at least 20% higher than the CFV of LLF was π = 0.6. The null-hypothesis was that π = 0.5. The maximum number of pairs required was 171. The study was to continue until the sample path crossed one of the boundary lines. If the upper boundary line was crossed, a difference between the two formulas would be indicated. A sample path crossing the bottom line would indicate that the hypothesis was rejected. The advantage of the sequential design was that in cases where study effects were either pronounced or missing, the sample path would cross one of the boundary lines quite early and the study would stop early, thus substantially reducing the number of study infants required.
The CFV and the feeding advancement (median, 25th and 75th percentile) of the ELBW infants who received less or more than 10% of human milk were described to assess the effect of human milk on early enteral feeding tolerance.
Descriptive data analysis was performed using the Mann-Whitney U test, the χ2 test, or Fisher exact test, as appropriate, without adjustment for multiple testing. The level of significance was set at P = 0.05. The data was analyzed using Excel 97 (Microsoft, Munich, Germany), SAS (SAS Institute Inc., Cary, NC), and Statgraphics plus 7.0 (Statistical Graphics Corporation, Rockville, MD). Data is shown as median, minimum, and maximum. Feeding volume is always presented as milliliters per kilogram of birth weight to correct for the varying postnatal drop in weight, even if not stated separately.
The protocol was approved by the Ethics Committee of the University of Ulm, and informed written parental consent was obtained.
The sample path in the graphic display of the sequential plan crossed the bottom boundary line as soon as 99 ELBW infants were enrolled and the hypothesis was rejected. One hundred fifty-two ELBW infants were admitted to the seven participating units within the enrollment period. A total of 53 infants were excluded for the following reasons: death within the first 2 days of life (n = 19), incorrect randomization (n = 15; 15 consecutively enrolled infants in one study center received the same formula without randomization), nonenrollment for unknown reason (n = 5), bad general condition (n = 6, defined by subjective impression of those in charge), transfer to another hospital within the study period (n = 4), NEC on the first day of life before randomization (n = 1, proved histopathologically), suspected trisomia 18 (n = 1), small number of infants (n = 2; one study center enrolled only two infants; this center was excluded). There was no difference between the two groups with respect to clinical characteristics (Table 2).
The study failed to demonstrate the hypothesized 20% advantage of HLF. The CFV was 720 mL/kg birth weight (range, 0–962 mL/kg birth weight) and 613 mL/kg birth weight (range, 3–1283 mL/kg birth weight; HLF vs. LLF). In infants of fewer than 28 weeks of gestation, the CFV was 671 mL/kg birth weight (range, 59–962 mL/kg birth weight) and 502 mL/kg birth weight (range, 3–895 mL/kg birth weight; HLF vs. LLF), and in infants of 28 weeks of gestation or more, the CFV was 864 mL/kg birth weight (range, 0–953 mL/kg birth weight) and 868 mL/kg birth weight (range, 10–1238 mL/kg birth weight).
Human milk was available in 55 infants, and in 41 infants, the human milk portion was more than 10% of the CFV. A human milk portion of more than 10% of the CFV was associated with an accelerated feeding advancement (Fig. 1), an increased feeding volume on day 14 (130 mL · kg birth weight−1 · d−1 [range, 0–166] vs. 105 [range, 0–150];P < 0.05; human milk > 10% vs. human milk ≤ 10%), and a higher CFV (mL/kg birth weight) compared with a human milk portion of less than 10% (833 mL/kg birth weight [range, 74–1,283] vs. 502 [range, 0–962;P < 0.05). The difference in the CFV was less pronounced in the HLF group (847 mL/kg birth weight [range, 74–953] vs. 586 [range, 0–962]) than in the LLF group (826 [range, 135–1,283] vs. 339 [(range, 3–931]). In more than half of the infants, the feeding volume was advanced according to protocol by 12 mL · kg birth weight−1 · d−1 on at least 10 of 12 study days, and on the final study day (day 14), the feeding volume was 103 mL · kg birth weight−1 · d−1 (range, 0–166).
Necrotizing enterocolitis (Bell stage ≥ 2) occurred in five infants of the LLF group but in none of the HLF group (Fisher exact test, P = 0.027). Three infants died within the study period for the following reasons: severe respiratory distress syndrome (n = 2, one in each group) and intracranial hemorrhage into the posterior fossa (n = 1, LLF group). Laparotomy was performed in three HLF and in two LLF infants for the following reasons: meconium ileus (n = 1), meconium ileus together with perforated microcolon (n = 3), and large subcapsular hematoma of the liver (n = 1). Feedings were withheld at least for 5 days after surgery. The feeding volume on day 14 was 0 in all of them. There was no significant difference between the two groups with regard to the secondary outcome variables (results not shown).
The current study on feeding advancement was innovative for several reasons. It focused on ELBW infants only. Milk feeding was started very early, at 48 hours of age. Clinical conditions like infections, hypotension, patent ductus arteriosus, and Indomethacin therapy or umbilical catheters in place did not influence the feeding strategy. To decrease the influence of potential subjective bias, only one single objective parameter, the preprandial gastric residuals, determined the volume of each feeding and of the feeding advancement. Feedings were not reduced for any of the following criteria: historical factors producing perinatal stress, temperature instability, episodes of apnea or bradycardia, and abdominal distension.
Although the study was a controlled randomized trial, it had some limitations. The study was performed in a double-blind manner, but the different smell and taste of the formulas may have been a bias. Another limitation is that primary outcome analysis was carried out according to the intention to treat and just 44 of the 99 infants received formula only. The study failed to find the hypothesized 20% advantage of HLF, however, a possible difference might have been missed because formula was fed as a supplement to human milk. At least some human milk was available in 55 infants, and 41 infants received more than 10% of human milk. There was virtually no difference between the CFVs of the infants of both groups who received at least 10% of human milk. However, the analysis of the subgroup of 58 infants who received formula only or less than 10% of human milk seemed to support our hypothesized advantage of the composition of HLF ([range, 586 mL/kg birth weight; range, 0–962] vs. 339 [range, 3–931], HLF vs. LLF). A second randomized trial especially focusing on formula-fed ELBW infants would be required to provide the definitive answer.
The design of the study permitted analysis of the influence of human milk on early enteral feeding advancement. Feedings were advanced more rapidly in ELBW infants who received at least 10% of human milk (Fig. 1). This data supported the hypothesis that human milk should be the first diet in ELBW infants.
Our study does not support the hypothesis that there is an association between lactose feeding and feeding intolerance. Our experience with human milk feeding, the experience of others (8,9), and the high CFV with HLF feeding in the subgroup of ELBW infants who received formula only or less than 10% of human milk suggest that lactose should remain the main carbohydrate in preterm infant formulas.
In ELBW infants, the timing and advancement of early enteral nutrition is controversial. Gastrointestinal priming (minimal enteral nutrition) (9) has been proposed as standard of care for very-low-birth-weight (≤1500 g) infants (10), but earlier enteral feeding advancement has been shown to be safe as well (11), and there is no study that compares these two methods in ELBW infants. In the current study, very early feeding advancement with milk according to the standardized protocol was possible without disadvantages for the ELBW infants. A median feeding volume of 100 mL/kg birth weight on day 14 in more than half of the ELBW infants suggests that 14 days of minimal enteral nutrition would be a drawback for these infants. The optimal rate at which feedings should be advanced in ELBW infants still must be defined. In very-low-birth-weight infants, feedings were advanced at rates of up to 35 mL · kg−1 · d−1without adverse effects (11), but in that study, enteral nutrition was initiated later than in the current study.
There are reservations as to whether early enteral feeding increases the incidence of NEC in preterm infants. There have been several reports of randomized trials of early versus late enteral feeding with NEC as a primary or secondary outcome (9,12–16). These studies together show that early enteral feeding does not increase the risk. In the current study, the overall incidence of NEC (5%) seemed to be quite low (17,18). However, this may be related to the fact that the study period covered the first 2 weeks of life only. In a recent American Neonatal Research Network report, the incidence of NEC was 78 of 694 ELBW infants (11%) given for the time until discharge, transfer, death, or age 120 days (19).
There was a difference between the two groups in nutrition-related gastrointestinal morbidity. Necrotizing enterocolitis was observed only with LLF feeding (P = 0.027). Considering the design of the study, this result may be incidental and would indicate the need for a prospective randomized trial with NEC incidence as a primary end point. However, the unique composition of HLF (high lactose, low protein, low calcium, low phosphate), which had been shown to induce stools very similar to those induced by human milk feeding (4), may be responsible for the different incidence of NEC (Bell stage ≥ 2) as well.
In conclusion, the study failed to find the hypothesized 20% advantage of the new HLF. With regard to the low NEC incidence, very early enteral feeding advancement according to the current feeding protocol appeared to be safe, but further trials are required to show whether overall morbidity, parenteral nutrition time, infusion-related complications, and length of hospitalization will be reduced. The observed advantage of HLM supports the hypothesis that HLM should be the first diet in ELBW infants; however, this hypothesis still must be confirmed in a controlled, randomized trial.
1. Billeaud C, Guillet J, Sandler B. Gastric emptying in infants with or without gastroesophageal reflux according to the type of milk. Eur J Clin Nutr. 1990; 44:577–83.
2. Tolia V, Lin C-H, Kuhns LR. Gastric emptying using three different formulas in infants with gastroesophageal reflux. J Pediatr Gastroenterol Nutr. 1992; 15:297–301.
3. Mihatsch WA, Högel J, Pohlandt F. Hydrolysed protein accelerates the gastrointestinal transport of formula in preterm infants. Acta Paediatr. 2001; 90:1–3.
4. Grütte F-K. [Prospects and limitations of adjusting cow's milk formula to human milk with specific consideration of microbiological effects.] PhD-thesis, Academy of Science, Berlin, German Democratic Republic 1984.
5. Griffin MP, Hansen JW. Can the elimination of lactose from formula improve feeding tolerance in premature infants? J Pediatr. 1999; 135:587–92.
6. Bell MJ, Ternberg JL, Feigin RD, et al. Neonatal necrotizing enterocolitis. Therapeutic decisions based upon clinical staging. Ann Surg. 1978; 187:1–7.
7. Wald A. Sequential tests of statistical hypotheses. Ann Math Statist.
8. Lucas A, Cole TJ. Breast milk and neonatal necrotizing enterocolitis. Lancet. 1990; 336:1519–23.
9. Schanler RJ, Shulman RJ, Lau C, Smith O'B, Heitkemper MM. Feeding strategies for premature infants: randomized trial
of gastrointestinal priming and tube-feeding method. Pediatrics. 1999; 103:434–9.
10. Kliegman RM. Experimental validation of neonatal feeding practices. Pediatrics. 1999; 103:492.
11. Rayyis SF, Ambalavanan N, Wright L, et al. Randomized trial
of “slow” versus “fast” feed advancements on the incidence of necrotizing enterocolitis in very low birth weight infants. J Pediatr. 1999; 134:293–7.
12. LaGamma EF, Ostertag SG, Birenbaum B. Failure of delayed oral feedings to prevent necrotizing enterocolitis in the neonate. Am J Dis Child. 1985; 139:385–9.
13. Glass EJ, Hume R, Lang MA, et al. Parenteral nutrition compared with transpyloric feeding. Arch Dis Child. 1984; 59:131–5.
14. Ostertag SG, LaGamma EF, Reisen CE, et al. Early enteral feeding does not affect the incidence of necrotizing enterocolitis. Pediatrics. 1986; 77:275–80.
15. Wilson DC, Cairns P, Halliday HL, et al. Randomised controlled trial of an aggressive nutritional regimen in sick very low birth weight infants. Arch Dis Child. 1997; 77:F4–11.
16. Yu VYH, James B, Hendry P, et al. Total parenteral nutrition in very low birth weight infants: a controlled trial. Arch Dis Child. 1979; 54:653–61.
17. Kliegman RM, Walker WA, Yolken RH. Necrotising enterocolitis: research agenda for disease of unknown etiology and pathogenesis. Pediatr Res. 1993; 34:701–8.
18. Kanto Jr, WP Wilson R, Breart GL, et al. Perinatal events and necrotizing enterocolitis in premature infants. Am J Dis Child. 1987; 141:167–9.
19. Ehrenkranz RA, Younes N, Lemons JA, et al. Longitudinal growth of hospitalised very low birth weight infants. Pediatrics. 1999; 104:280–9.