Objective. To describe growth during initial hospitalization for very small premature infants fed intravenous hyperalimentation, then calcium supplemented 1460 mg/L (36.5 mmol/L) 81 kcal preterm formula.
Population. A total of 109 survivors whose <1000 g birth weight was appropriate for gestational age. Mean gestational age was 25.8 weeks.
Results. Graphs were constructed for weight, length, and head circumference by week of age. Mean and ± 2 SD lines were depicted, with mean intrauterine growth lines for comparison. Separate graphs showed mean weight, length, and head circumference growth by 100 g birth weight cohorts.
Mean Z scores based on normal intrauterine growth curves were calculated. Weight Z scores were −.35 at birth, −1.79 at 14 days, and −1.87 at 56 days. Length Z scores were −.32 at birth, −1.29 at 14 days, and −2.24 at 56 days. Head circumference Z scores were 0.01 at birth, −1.26 at 14 days, and −1.06 at 56 days. (Z score = [measured parameter − intrauterine mean for gestation]/intrauterine SD for gestation).
Repeated-measures multivariate ANOVAs showed the following significantZ score changes. There were decreases in Zscores for weight, length, and head circumference between birth and 14 days and an additional decrease for length between 14 and 56 days. Head circumference Z scores increased from day 14 to day 56, but remained smaller at day 56 than at day 0. Initially, head circumferenceZ scores were better than weight or length (possibly because of late head measurement timing). At day 14, the Zscores for weight were lower than those for length and head circumference. At day 56, the head circumference Z scores were higher than those for length or weight.
Conclusion. Compared with intrauterine standards, weight, length, and head circumference were all worse at day 56 than at birth, although there was relative head-sparing and weight growth paralleled intrauterine growth after 14 days. Length worsened from day 14 to day 56 in spite of the use of calcium and phosphorus-enriched formula.
Study of the growth of <1000-g birth weight premature infants has become of interest recently as increasing numbers of infants survive. This study was designed to describe retrospectively the growth of a group of survivors whose <1000 g birth weight was appropriate for gestational age, unselected for degree of medical illness. The study group was fed intravenous hyperalimentation progressed to calcium and phosphorus-supplemented preterm formula according to a specific protocol that was introduced in 1981. Weekly length and head circumference measurements were available. Therefore, it was possible to construct feeding method-specific growth graphs for weight, length, and head circumference by week of age. Growth could be compared with intrauterine norms.
All infants born at the Royal Victoria Hospital in Montreal, Canada, between January 1981 and March 1993 whose <1000 g birth weights were appropriate for gestational age were potential study subjects. Appropriate for gestational age was defined as having a birth weight ratio of ≥.75. Weight ratio was calculated as weight divided by mean intrauterine weight for the gestational age according to revised sex-specific Usher–McLean growth standards.1According to our previous work, a birth weight ratio of .75 was on average at −2 SDs of weight for gestational age. This provided a cutoff that prevented inclusion of the infants most likely to have clinical consequences of intrauterine growth retardation.
Gestational age was determined from the due date predicted by the mother's date of onset of last menstrual period, unless this differed from an early ultrasound estimation by >7 days, in which case the ultrasound estimation was used. Infants with neither known maternal menstrual dates nor early ultrasound dates did not meet study inclusion criteria.
During the study period, infants were routinely started on intravenous dextrose and water at birth. The aim was to increase to 140 mL/kg·day at stability, although more fluid might be required from day 2 to 7, depending on degree of weight loss. Infants were nursed in a thermoneutral environment from birth and placed in closed incubators after the first hours of life. Humidity was not added to incubators. Amino acid solution and lipid emulsion each were started at 0.5 g/kg·day when the infants were metabolically stable. Both were increased by 0.5 g/kg·day increments as tolerated to a maximum of 3.0 g/kg·day amino acid and 3.5 g/kg·day lipid. Protein and lipid tolerance were verified with blood urea nitrogen and triglyceride measurements, respectively. Intravenous glucose was increased to 10 to 12 g/kg·day (range, 8 to 15) according to blood glucose levels. The aim was 85 to 90 intravenous kcal/kg·day from combined carbohydrate, fat, and protein intake. Calcium was added to the nonlipid solution at 100 mg (2.5 mmol)/L in the first 3 weeks, then 400 mg (10 mmol)/L. Vitamin D (1000 IU/L) was also added.
When the infants were clinically stable, on or off respirators, oxygen, or continuous positive airway pressure, bolus indwelling orogastric feeds of 81-kcal preterm formula supplemented with 1460 mg (36.5 mmol)/L calcium, 730 mg (23.2 mmol)/L phosphorus, and 1220 IU/L vitamin D (Special Care Similac, Ross Laboratories, Columbus, OH) were started every 2 hours at 0.5 to 1.0 mL per feed. Formula composition appears in Table 1. Medium-chain triglycerides made up 50% of the lipid content. Enteral feeding volumes were increased gradually and parental intake reduced as tolerated, maintaining total fluid intake at 140 mL/kg·day. Gastric residuals and abdominal distension were monitored. When full enteral feeds were attained, a change was made to bolus gavage feeds every 3 hours, and vitamin D 250 IU/day supplement was started. At 2 kg weight, or later for babies requiring prolonged intravenous alimentation, feeds were changed to regular 81-kcal infant formula. Before discharge, 68-kcal infant formula or breast milk was substituted, and feeds were initiated every 4 hours. During the study period, occasional fresh breast milk feeds might be substituted.
An additional study2 addresses clinical correlates of growth. Here, for the purpose of post hoc comparison, caloric intakes are calculated as intravenous calories plus 85% of enteral calories, to give metabolizable calories, making the assumption that only 85% of enteral intake is absorbed from the intestine.3-5
Data for growth graphs came from chart review. During the study period, infants were weighed daily before a feed. Length and head circumference were measured weekly. This was usually performed by the assistant head nurse or another senior nurse in her absence. Crown to heel length was measured using an infant measuring tray, and occipitofrontal head circumference was measured with a disposable tape.
Weekly data were graphed for days 0, 7, 14, etc, on growth charts. For length and head circumference, interpolations were made to 0, 7, 14, etc, days from weekly data. Length and head circumference measurements were excluded if either surrounding point was unavailable for interpolation. Graph lines were terminated as being unrepresentative when >50% of patients had been discharged. Comparison lines came from revised Usher–McLean intrauterine growth grids (made in the same hospital) using pooled-sex values.
Weight, length, and head circumference measurements were entered into a Dbase III file. Computations were performed using SAS statistical software. Harvard Graphics was used to construct growth graphs.
Weight, length, and head circumference values for 0, 14, and 56 days of life were converted to Z scores for each patient, thus, growth across the different parameters and times could be compared. For the purposes of this analysis, the length and head circumference values obtained in the first week were extrapolated to day 0. Zscores were calculated as the mean minus the intrauterine mean for gestation, divided by the intrauterine SD for gestation, using revised sex-specific Usher–McLean intrauterine growth norms. Fifty-six days was chosen for the final Z score comparison, because the first infants were discharged soon after this time.
Repeated-measures multivariate ANOVA were performed to account for inter- versus intrasubject variability and variability across theZ scores for the different growth parameters and times. Nine comparisons were planned among Z scores for day 0, day 14, and day 56, for weight, length, and head circumference. A separate multivariate repeated-measures ANOVA was performed for nine planned comparisons among weight, length, and head circumference at each of day 0, day 14, and day 56. Results were compared using an α level of 0.05 and a Bonferroni correction of 0.05/2 = 0.025 to account for the two separate ANOVAs.
A total of 256 infants whose <1000 g birth weights were appropriate for gestational age were born at the Royal Victoria Hospital during the study period; 121 survived to discharge home. Of these, 2 were transferred from the hospital permanently before 34 weeks' gestation. For an additional 10 patients, either hospital charts or growth charts could not be located. The remaining group of 109 patients comprised the study infants.
Comparison of the 12 excluded patients with the 109 remaining patients showed that respiratory support duration, maximum Fio2 required for treatment of respiratory distress syndrome, incidence of patent ductus arteriosus, birth weight ratio, gestational age, calendar time, incidence of necrotizing enterocolitis, and 5-minute Apgar score were not significantly different between the two groups. There was a significant difference in sex distribution, because 11 of the 12 excluded patients were female (P = .01).
For gestational age determination, 20 (19%) study patients had maternal dates only, 7 (6%) had early ultrasound dates only, and 82 (75%) had both dates. A total of 26 gestational ages were changed from the maternal date estimation to the ultrasound estimation because of >7-day discrepancies between the two estimations.
Intestinal complications associated with this feeding regime were reviewed. One infant, who was excluded because of permanent transfer before 34 weeks' gestation, required bowel resection for necrotizing enterocolitis. Ten of the study patients developed clinical necrotizing enterocolitis (intramural air). None of the study patients required surgery. Using the study feeding approach, no death in the original 256 patients could be attributed to feeding. One infant who developed necrotizing enterocolitis died later of bronchopulmonary dysplasia. Three who were never fed died of gastrointestinal complications.
Details of the clinical course appear in Table2.
Consistency of the feeding protocol during the 12 years of the study was confirmed by correlation coefficients between the time from study start and the days of life of starting intravenous protein, intravenous lipid, and enteral feeding, which were, respectively, −.032, 0.012, and −.003 (r values). The intravenous protein solution used was Travasol (Clintec, Canada) throughout the study period, and the lipid emulsion was Intralipid, although both 10% and 20% lipid preparations were in use. Detailed review of enteral feeding for the first 8 weeks of life showed that 90.4% by volume of all feeds were 81 kcal/100 mL-, calcium-supplemented, preterm formula. Progression to regular 81-kcal infant formula occurred during these weeks for 4.3% of feeds and to 68-kcal formula for .7% of feeds. Breast milk comprised 0.9% of feeds. Deviations from the enteral feeding protocol made up 2.7% of feeds, of which all but 8.2% were 91-kcal preterm formula, usually calcium-supplemented. Intravenous and enteral caloric intake is summarized in Fig1.
Mean intake was 60.0 metabolizable kcal/kg·day (SD = 10.0; range, 32.1 to 82.5) for days 0 to 14, then 88.1 metabolizable kcal/kg·day (SD = 6.8; range, 70.1 to 110.5) for days 15 to 56. Estimated mean calcium intake was 68 mg (1.7 mmol)/kg·day orally and 16 mg (0.4 mmol)/kg·day intravenously during the first 56 days of life.
Although the feeding protocol did not change during the study period, there were other changes and improvements in management. Survival rates for infants who were appropriate for gestational age improved during the 12 years, with 42/127 (33%) of infants born before 1987 versus 77/129 (60%) born after 1987 surviving (P < .001). Dexamethasone was introduced in the latter study period and was used in 14 (13%) of subjects. A detailed analysis of the relationship of growth to clinical events appears in a companion article.2 Date of birth was not a significant predictor of growth.
For growth graphs, 99%, 93%, and 93%, respectively, of weight, length, and head circumference values were available.
Figure 2 shows study infant weights plotted by mean ± 2 SDs at 0, 7, 14, etc, days of life. The mean intrauterine weights for mean gestational age (revised Usher–McLean data) are plotted for comparison. Figures3 and 4show similar graphs for length and head circumference. For length and head circumference, day 0 measurements are missing because only figures derived from two consecutive weekly measurements were included in graph computations.
Table 3 shows mean weight, length, and head circumference at 0, 14, and 56 days of life expressed asZ scores with actual Z score SDs. Actual growth parameters also appear in the table with actual standard deviations. Ninety-four percent of values were available for these calculations.
Table 4 shows the repeated-measures multivariate analysis P values for comparisons ofZ scores for weight, length, and head circumference at 0, 14, and 56 days. Results were not changed by using the Bonferroni correction, because no P values between .05 and 0.025 occurred in the analysis. Results showed that weight Zscores were lower at day 14 and day 56 than at birth, although they did not change significantly between 14 and 56 days. Length Zscores were lower at day 14 than at birth and at day 56 than at day 14. Head circumference Z scores were lower at day 14 and day 56 than at birth, although they improved from day 14 to day 56. The initial head circumference Z scores (extrapolated to day 0) were higher than length Z scores (extrapolated to day 0) or (actual) day 0 weight Z scores. At day 14, weightZ scores were lower than length or head circumferenceZ scores. At day 56, head circumference Z scores were higher than length or weight Z scores.
No serum or urine calcium and phosphorus measurements were routinely performed during the study period. No lithiasis or nephrocalcinosis became clinically evident during the period, although no search for these entities was undertaken in asymptomatic patients. Four (3.7%) of the subjects (who achieved full enteral feeds at an average of 66 days) developed rib fractures, whereas these had been fairly frequent before initiation of calcium-supplemented formula in the unit.
Infant feeding is a major component in the care of the extremely premature infant. Feeding methods for very small premature infants vary widely. The literature contains few outcome studies of specific feeding regimens to assist clinicians.
The importance of obtaining good early growth for these infants is unclear in the literature. Hack6,7 assessed a group of premature infants whose <1500 g birth weights were appropriate for gestational age at 8 months and 33 months corrected age. Those who were small at 40 weeks' gestation, but caught up by 8 months, did well neurologically, whereas those who had not caught up by 8 months did not. At 33 months, developmental quotients were less for infants who remained small. Ross8,9 also followed a group of premature infants whose birth weights were <1500 g. Holding birth weight constant, good neurologic outcome was associated with larger head circumference by 1 month postterm and greater length and weight by 3 months' postterm. At 3 years of age, an association persisted between neurologic status and height and head circumference. These studies suggest that early catch-up in growth may be important to neurologic outcome. However, the effects of neurologic status on growth and neonatal illness on both growth and neurologic status cannot be dissociated. Hadders-Algra10 showed that appropriate-for-gestational-age premature infants whose weight was less than the 10th percentile at term age remained smaller at 6 years than premature infants whose weight was greater than the 10th percentile at term. This provides evidence that ultimate growth achievement may be related to early catch-up.
Gill11 provided separate weight curves by week of gestation for infants born before 30 weeks' gestation who were fed parenteral glucose, amino acid, and fat emulsion followed by standard 67-kcal or 80-kcal premature formula with or without fresh breast milk. Wright12 provided weight, length, and head circumference curves by week of age for infants whose ≤1500 g birth weight was appropriate for gestational age. Lines were depicted by 250 g birth weight category. Infants were fed hyperalimentation, and dilute enteral feeds progressed to 81-kcal premature formula or fortified breast milk as available. Mean growth in length, as well as in weight and head circumference, plots comparably with our patients. Wright provided a comprehensive review of the growth reported by others. Growth curves were compared with the curves of Dancis,13rather than with intrauterine standards.
There is a lack of data in the literature specifically for infants whose birth weight is <1000 g.
It has not been usual in the literature to compare growth of very premature infants with intrauterine standards. Infants ex utero are subject to stresses not encountered in utero that have an impact on growth. It does not follow automatically that it is most desirable for them to achieve the same growth as is achieved in utero. However, we have no other reasonable standard for their growth. The curves of Dancis13 are based on the actual growth of a group of infants without complications. They have been surpassed because of current techniques of infant care.12 It seems likely that growth will continue to improve as premature care improves. This makes it unproductive to replace the Dancis curves with other similarly derived curves as references for ideal growth. Referencing growth to an intrauterine standard does provide a comparison that can be understood intuitively and provides at least a maximum goal for therapeutic efforts. Curves derived from actual populations whose management is known, such as those presented here, although making no claim to represent ideal growth, may serve as a point of comparison for future efforts.
Our infants differ from infants reported in previous studies in that they received 81-kcal premature formula containing 1460 mg (36.5 mmol)/L calcium from the start of enteral feeding to meet the enhanced calcium needs of a rapidly growing premature infant.
Despite the concern about beginning enteral feeds in small premature infants with concentrated mineral-enriched formula, our incidence of necrotizing enterocolitis is comparable with those of large reported series.14,15 It must be emphasized, however, that feeding progression of our infants was cautious, with monitoring for abdominal distension and increased gastric residuals.
Our study benefits from the availability of intrauterine comparison data from the same hospital population, which was recently revised. Meaningful comparisons thus can be made with intrauterine norms.
Unfortunately, the study population did not have head circumference measurements performed according to uniform technique at birth. To be able to perform the ANOVAs, we extrapolated head circumference measurements performed during the first week to day 0. The standard curves that we used for comparison were obtained from a population that included infants who died early who were measured at birth. One might speculate that this is the reason why analysis showed larger head circumference Z scores than weight or length Zscores at day 0. An alternative would be to speculate that larger head circumference conferred a survival advantage. The magnitude of difference between the day 0 and day 14 and day 56 head circumferenceZ scores is large enough that the extrapolation is unlikely to have created the significant differences between the day 0Z scores and day 14 and 56 Z scores.
In our population, the greatest deviations from intrauterine norms in all growth parameters occurred in the first 14 days of life. Subsequently, head circumference improved and weight grew at expected intrauterine rates, whereas length growth was slower than expected intrauterine growth between 14 and 56 days. Most of the growth deficit incurred by 56 days actually occurred in the first 14 days of life. Potentially, growth improvement could be achieved either by decreasing the early losses or by subsequent catch-up.
Because most of the growth deficit in the study infants occurred in the first 14 days of life, it is logical to examine ways of decreasing the early losses. It is unlikely that caloric intake can be increased during the early days without exceeding metabolic and renal tolerances for fat, glucose, and protein, which are usually decreased during the first 7 to 14 days for extremely premature infants. The recent availability of means of providing high environmental humidity at low infection risk may be one way to reduce initial weight loss not only by lessening insensible water loss but also by minimizing energy lost in thermoregulation.
If the initial growth deficit cannot be prevented, would it be possible to increase nutritional support thereafter and obtain catch-up growth? This catch-up ideally would be in length and head size, as well as in weight, because these all are less at 56 days than at birth, and length falls increasingly behind from 14 to 56 days. The feeding regimen used in the study was aimed at providing normal intrauterine (not catch-up) rates of growth. In fact, normal intrauterine growth in weight and head size catch-up did occur between 14 and 56 days. Would larger intakes in this period have produced symmetric catch-up growth? Would they have been well tolerated? To answer these questions requires additional systematic study of more liberal feeding regimens.
Linear growth has been associated with calcium status.16,17The study infants' length continued to fall behind intrauterine norms from 14 to 56 days. By 3 weeks of age, our intravenous calcium concentration met the American Academy of Pediatrics recommendations.18 The calcium concentration in the enteral formula met the American Academy of Pediatrics recommendation16,18 and exceeded the recommendation of the European Society of Pediatric Gastroenterology and Nutrition19 from the start of enteral feeding. However, because enteral feeding progression was slow, with many stops and starts dictated by clinical status, the total calcium supplied was on average inadequate to meet fetal accretion rates.20Increasing the intravenous calcium concentration, as suggested by the recent work of MacMahon21 and Dunham,22 would have improved calcium intake. Additional studies are required to determine whether such increased intravenous mineral intake would improve length growth.
In the first 14 days, when weight and length growth fell from intrauterine curves, growth in head circumference also was subnormal. By 56 days, growth in head circumference had exceeded weight and length growth relative to expected rates of growth in utero. This is what might be expected physiologically in an undernourished population. However, although there was some catch-up from 14 to 56 days, the mean head circumference at 56 days was still significantly smaller than intrauterine norms. It is worrisome that reduction in brain size during this period, which is normally one of very active brain growth, might be associated with negative developmental consequences.
In conclusion, this study provides growth data for the <1000-g birth weight, appropriate-for-gestational-age infant fed intravenous hyperalimentation progressed to calcium-supplemented preterm formula. Compared with intrauterine standards, weight, length, and head circumference were all worse at 56 days than at birth, although some degree of head sparing was evident. Although weight growth was comparable with intrauterine norms from 14 to 46 days of life, there was no catch-up growth. Length growth was slower than the expected intrauterine rate of growth during this period.
Additional research is needed to determine how to improve growth in the very small premature population, both to reduce initial losses in all types of growth and to maximize subsequent growth, especially in length. Safer means of delivering high humidity may result in less early growth failure. Recognition of the need for catch-up growth after 14 days may increase caloric goals during this period. However, tolerance for increased feedings needs to be demonstrated. Newer techniques for increasing intravenous calcium intake may result in improved skeletal growth.
We thank Frances McLean for assistance with the database.
- Received March 30, 1995.
- Accepted February 18, 1997.
- Address correspondence to Margaret A. Berry, MDCM, FRCP, Department of Neonatology, Royal Victoria Hospital, 687 Pine Ave W, Montreal, Quebec, Canada H3A1A1.
Reprint requests to (R.H.U.) Department of Neonatology, Royal Victoria Hospital, 687 Pine Ave W, Montreal, Quebec, Canada H3A1A1.
- ↵Usher R, McLean F, Abrahamowicz M, Kramer M. Birth weight distribution by confirmed gestational age and sex. In press
- Berry M,
- Abrahamowicz M,
- Usher R
- De Curtis M,
- Brooke O
- Gill A,
- Yu V,
- Bajuk B,
- Astbury J
- Wright K,
- Dawson J,
- Fallis D,
- Vogt E,
- Lorch V
- Palmer S,
- Biffin A,
- Gamsu H
- ↵Koo W, Tsang R. Calcium, magnesium, phosphorus, and vitamin D. In: Tsang R, Lucas A, Uauy R, Zlotkin S, eds. Nutritional Needs of the Preterm Infant. Baltimore, MD: Williams and Wilkins; 1993:135–155
- American Academy of Pediatrics, Committee on Nutrition
- MacMahon P,
- Mayne P,
- Blair M,
- Pope C,
- Kovar I
- Dunham B,
- Marcuard S,
- Khazanie P,
- Meade G,
- Craft T,
- Nichols K
- Copyright © 1997 American Academy of Pediatrics