点击显示 收起
1 From The Carman and Ann Adams Department of Pediatrics, Wayne State University and Hutzel Hospital (WWKK), and Computing and Information Technology, Wayne State University (EMH), Detroit, MI
2 Supported in part by Ross Products Division, Abbott Laboratories, Columbus, OH. 3 Reprints not available. Address correspondence to WWK Koo, Department of Pediatrics, Hutzel Hospital, 3980 John Road, Detroit, MI 48201. E-mail: wkoo{at}wayne.edu.
ABSTRACT
Background: Despite the theoretical benefits of nutrient-enriched formula given to preterm infants after hospital discharge, its role in reversing growth deficits after hospital discharge remains poorly defined.
Objective: The aim was to determine the effect of different formulas on the growth, bone mass, and body composition of preterm infants after hospital discharge.
Design: This was a randomized, double blind comparison of a nutrient-enriched formula (EF) and a formula for term infants (TF) given for 1 y after hospital discharge. Compared with the TF, the EF had a higher energy density and higher contents of protein, calcium, and phosphorus (by 10%, 21%, 44%, and 11%, respectively) and higher contents of almost all other nutrients (by 10%).
Results: Birth weights of the infants were 630–1620 g (median: 1250 g) and gestational ages were 24–34 wk (median: 29 wk). TF resulted in significantly greater weight, length, head circumference measurements, and their respective z scores on the basis of age- and sex-specific norms. At the end of the study, the mean z scores for the corrected age of infants in the TF group were –0.37 for weight, 0.001 for length, and 0.50 for head circumference. The TF group also had significantly greater dual-energy X-ray absorptiometry measured bone and lean and fat mass than did the EF group (P < 0.05 for all comparisons).
Conclusions: The use of EF for preterm infants after hospital discharge shows no advantage over TF in growth, bone mineralization, and body composition. More studies are needed to determine the optimal postdischarge nutrition support for preterm infants.
Key Words: Nutrition support • dual-energy X-ray absorptiometry • infants • bone • lean • fat
INTRODUCTION
The growth delay of small preterm infants at hospital discharge and during early infancy is well known (1). This is often associated with the presence of bone demineralization and even radiographic rickets, fractures, or both in severely affected infants (2, 3). Most preterm infants are discharged before they reach term (ie, postmenstrual age 40 wk), and, to attain the growth measures of term infants of the same postmenstrual age, they must achieve catch-up growth. Thus, their overall nutrient requirements are higher than those of infants born at term when taking into account the need for catch-up growth (4). An early report of improved growth in preterm infants after hospital discharge with the use of nutrient-enriched formulas (5) has led to the development and commercial availability of postdischarge formulas that appear more suited to the relatively higher nutritional requirements of the small preterm infant. These infant formulas have an energy density and nutrient content between those of the special formulas used in the hospital and the standard infant formulas used for term infants. Their use has been recommended for preterm infants, especially for those with very-low-birth-weights (VLBW) of 1500 g, until 9 mo (6) or 12 mo (4) chronological age. However, the optimal duration for the use of postdischarge formulas is unknown, and there is minimal data on the quality of growth, specifically bone mass and body composition, in preterm infants receiving these formulas for prolonged periods. The aim of the present study was to compare growth, bone mass, and body-composition data in preterm infants fed a nutrient-enriched postdischarge formula with those who were fed a standard infant formula.
SUBJECTS AND METHODS
This was a randomized controlled, double-blinded, parallel feeding study evaluating growth, bone mineralization, and body composition in 2 groups of formula-fed preterm infants with gestational age 34 wk. Infants enrolled into the study had intact gastrointestinal tracts, tolerated full enteral feeding, and were expected soon to be ready for hospital discharge. Exclusion criteria included the presence of major congenital malformation, history of gastrointestinal surgery, or severe postnatal complications that prevented normal suck and swallow actions and adequate enteral feeding at the time of enrollment. On enrollment, infants were randomly assigned to receive 1 of 2 formulas until 12 mo after hospital discharge. The randomization schedule was stratified by sex and birth weight categories (<1001 g, 1001 to 1300 g, and >1300 g). The formula assignment for each subject was placed in a sealed envelope, and an envelope was selected according to the stratified groups on enrollment. In the event that a subject dropped out of the study, the next eligible subject that met the same stratification criteria was assigned to the same formula. Multiple births were not excluded from the study, but all infants from the same set received the same assigned formula.
The study protocol was approved by the Institutional Review Board for Human Investigations at Wayne State University, Detroit, MI. Written informed consent was obtained from one parent of each subject. All study personnel and the subjects' parents were unaware of study formula assignment. Data management and statistical analysis were performed by the investigators. The study sponsor (Ross Products Division, Abbott Laboratories, Columbus, OH) supplied the coded study formulas and maintained the code. Group assignment was not revealed until study completion.
Study formulas and feeding procedures
By design, the desired energy density of reconstituted formula used for term infants (term formula, TF) had an energy density of 20 kcal/oz (2800 kJ/L), and the other formula (enriched formula, EF) had an energy density of 22 kcal/oz (3080 kJ/L). EF also contained higher protein, calcium, and phosphorus by 21%, 44%, and 11% respectively, and 10% higher content of almost all other nutrients than did TF (Table 1). All study formulas were previously commercially available cow milk–protein based formulations (Similac with Iron and Similac NeoCare, respectively) from the same manufacturer (Ross Products, Abbott Laboratories, Columbus, OH) packaged in identical paper pouches and labeled with coded identification. Specific written instructions on the reconstitution of milk powder as well as containers for the preparation and storage of reconstituted milk were provided to each parent. The reconstitution procedure was identical for both formulas. In addition, the written instructions were repeated verbally and the parental understanding of these instructions was confirmed by the research personnel.
View this table:
TABLE 1. Content of selected nutrients in the term (TF) and nutrient-enriched (EF) formulas1
Parents were encouraged by the principal investigator and research personnel to offer ad libitum feeding of the study milk to the infants and to discard any residual formula in the bottle after each feeding. The use of and tolerance to the study formulas were confirmed verbally with the parent for all but one subject at each follow up. Parents were requested not to give mineral or vitamin supplementation. Other dietary intake was at the discretion of the family.
Study assessment
Weight, length, and head circumference were measured by using standard methods (7, 8) at enrollment and at 2, 4, 6, 9, and 12 mo after discharge. The infants were weighed in the nude to the nearest 5 g with an electronic scale (Seca, Toledo, OH), which was calibrated daily. Length was measured in duplicate to the nearest 0.1 cm with the infant in a recumbent position by using O'Leary Lengthboards (Ellard Instruments Ltd, Seattle, WA). Head circumference was measured in duplicate as the maximum occipital frontal circumference to the nearest 0.1 cm by using a disposable paper tape measure (Ross Products Division, Abbott Laboratories).
Total body bone mass as bone mineral content (BMC) and fat and lean mass were measured by dual-energy X-ray absorptiometry (DXA) at enrollment and at 2, 4, 6, and 12 mo after discharge. Details of scan acquisition techniques have been reported elsewhere (9, 10). In brief, the technique employed a 2 platform system (Hologic QDR 2000+; Hologic Inc, Bedford, MA) with a foam-covered rigid aluminum platform placed on top of another cloth-covered foam table pad, and an external calibration standard was used. Each infant was wrapped in a cotton blanket for the scan. The use of a diaper with or without a light undergarment for the infant was allowed before bundling the infant in the cotton blanket. However, all coverings were weighed with an electronic scale and the weight recorded. Scan analysis used infant whole-body software version 5.73p, which has been validated independently by different investigators (11–14). Only scans with no significant movement artifacts (9, 10) were included in the data analysis. In our laboratory, the precision error from duplicate infant whole-body scans for BMC and fat and lean mass were 3.3%, 5.6%, and 1.7%, respectively.
Statistical methods
The present study was designed with an estimate of 30 subjects per group to complete the study. On the basis of the 10% difference in energy and nutrient content between the formulas, the sample size was calculated to detect a 10% difference between group means in weight or length (15) at study completion with an of 0.05 (2 tailed) and power of 0.80. The secondary outcome was to determine any changes in DXA-measured bone mass (16) or body composition (17).
A comparison of birth and baseline data between the completers and the dropouts was used to test for attrition bias. Univariate or multivariate (MANOVA) analyses of variance were used to compare the various continuous measures. Chi square was used to compare the discrete variables.
The absolute values of growth data were normalized by expression as z scores by using the age- and sex-matched anthropometric data from the National Center for Health Statistics (18). Serial measurements and z scores for growth measures were used in the statistical analyses. For DXA measurements, bone, fat, and lean mass in absolute amounts and as a percentage of total weight were used in statistical analyses.
Analysis of the serial growth and DXA data between the feeding groups used repeated-measures ANOVA with linear contrast. Birth weight was treated as a covariate to eliminate its effect on growth. Sex and formula groups were treated as fixed factors. However, formula group was the only fixed factor used in the comparison of z scores because the z scores had already corrected for sex. For all subjects with 1 follow-up data point, the intraindividual rate of change (slope) for each growth and DXA measurement was computed by regression on chronologic age at each study time point. The slopes for each measured variable of the subjects from each formula group were compared by using Student's t test and by ANOVA with control for birth weight and sex. In addition, analyses of serial data were repeated post hoc by using race (dichotomized as either African Americans or non-African Americans) instead of sex as a fixed factor.
Unless otherwise indicated, all values are means ± SEMs. Statistical tests were performed with SPSS version 13.0 for WINDOWS (SPSS Inc, Chicago, IL) at an adopted significance level of 0.05 and were two-tailed.
RESULTS
Eighty-nine infants (44 in the EF group) were enrolled into the study. Baseline measurements were performed at 41 ± 0.2 d, and feeding of study formula began thereafter; 67 infants (31 in the EF group) completed the 1-y follow-up. The ranges of birth weights were 630 to 1620 g (median: 1250 g) and gestational ages were 24 to 34 wk (median: 29 wk). All but 5 infants had birth weights between the 10th and 90th percentiles for gestational age (19). Five infants with birth weight <10th percentile satisfied the inclusion and exclusion criteria and were randomly allocated to the EF group. Two of these 5 infants did not return for follow-up after the baseline measurements; the remaining 3 infants had birth weights above the 5th percentile for gestation and did not develop any conditions that precluded the continuation of the study. Four of the 7 sets of twins were randomly allocated to the EF group, and 1 set of triplets was randomly allocated to the TF group. Clinical and demographic details are shown in Table 2.
View this table:
TABLE 2. Comparison of clinical data and reasons for dropout between term (TF) and nutrient-enriched (EF) formula groups1
The infants were discharged from the hospital at 43.2 ± 2.1 d. The actual intervals for follow-up occurred at 53 ± 1.0, 109 ± 1.3, 180 ± 1.4, 273 ± 1.8, and 361 ± 2.9 d after hospital discharge. The corrected ages (based on an age of 0 d at the 40 wk postmenstrual date) at baseline and at each follow-up time points were –37 ± 1.4, 19 ± 1.4, 76 ± 1.8, 147 ± 1.7, 240 ± 2.2, and 329 ± 3.0 d. No significant differences in follow-up interval or corrected ages were observed between the groups. Of the study completers, 12 infants (7 in the EF group) had solids or juice introduced to the diet by 2 mo after hospital discharge, and all were receiving other foods in addition to the study formula by 9 mo after discharge. However, details of the volume of study milk and amounts of other nutritional intake were not recorded for the study subjects.
Most dropout subjects had no data beyond the baseline measurements. Thus, determination for attrition bias focuses on birth and baseline measurements. MANOVA in a comparison of all noncompleters and all completers showed the overall omnibus F value was not statistically significant for birth measurements and gestational age and for baseline anthropometric and DXA measurements. The overall omnibus F value also was not significant in a comparison between noncompleters and completers within each feeding group for birth and baseline data. No significant interaction existed among the status of the study completion, feeding group assignment, and any of the variables tested.
Of the 67 subjects who completed the study (31 in the EF group), there was uncertainty of compliance with ingestion of study formula for 1 subject in the EF group. The inclusion or exclusion of this subject did not significantly alter the outcome, and the results shown were based on inclusion of all subjects.
Serial growth measurements and corresponding z scores are shown in Table 3. Repeated-measures ANOVA showed that the birth weight, sex, and formula type had significant independent effects on various growth measures. After control for birth weight and sex, the formula type maintained a significant independent effect on all growth variables for both absolute measurements and z scores. Infants fed TF had significantly higher values for all growth variables than did infants fed EF.
View this table:
TABLE 3. Weight, length, and head circumference of preterm infants who were fed term (TF) or enriched (EF) formula for 1 y after hospital discharge1
The rates of increase (slope) in unadjusted weight, length, and head circumference and their corresponding z scores are shown in Table 4. For the completers, the rates of increase in weight and z scores for weight and length were significantly higher in TF group than in the EF group. When all subjects with any follow-up measurements were analyzed, the rates of increase for weight and length z scores remained significantly higher in the TF group than in the EF group. For the completers, ANOVA comparison of the slopes of growth variables after control for birth weight with sex or race as fixed factors showed a significant main effect of formula type, with the TF group having a higher rate of gain in weight and length and higher z scores than the EF group. No significant interactions between formula type and sex were observed for any growth variables.
View this table:
TABLE 4. Comparison of the rates of change (slopes) in growth of preterm infants fed term (TF) compared with enriched (EF) formulas for 1 y after hospital discharge
Race showed no independent effect for any variable of growth, although there was a significant race-by-formula type interaction effect on the slopes for length z score. Results were similar when data from all subjects were analyzed, and there was no significant interaction between formula type and sex or race for any growth variables.
Serial DXA measurements are shown in Figure 1. Formula type, birth weight, and sex but not race showed significant independent influence on DXA measurements. The TF group had significantly higher BMC and fat and lean mass than did the EF group, and the males had significantly higher BMC and lean mass than did the females. No significant interactions between formula type and sex or race were observed for these measurements. The mean weight of the covers used for the infants during DXA measurement was 271 ± 19.9 g and 238 ± 13.7 g for the TF and EF groups, respectively, and these were not significantly different. Bone and fat as a percentage of total mass were significantly increased and the percentage of lean mass significantly decreased for both groups during the study (P < 0.001 for all comparisons). Baseline values for bone, fat, and lean mass percentages were 1.28 ± 0.02%, 13.6 ± 0.48%, and 85.3 ± 0.49%, respectively, in the TF group and 1.29 ± 0.2%, 12.6 ± 0.43%, and 86.1 ± 0.43%, respectively, in the EF group. At the end of study, the respective bone, fat, and lean mass percentages were 2.36 ± 0.03%, 29.3 ± 0.74%, and 68.3 ±0.75% in the TF group and 2.39 ± 0.03%, 26.7 ± 1.24%, and 69.1 ± 2.27% in the EF group. After control for birth weight, the formula type had a significant independent effect on fat and lean mass percentage (P < 0.01 for both), with a slightly higher fat mass percentage and lower lean mass percentage in the TF group. Neither sex nor race had any independent effect on bone, fat, or lean mass percentage. No significant interactions between formula type and sex or race were observed in any comparison.
FIGURE 1.. Mean (±SE) serial bone mineral content and fat and lean mass of male ( and ) and female (• and ) preterm infants fed term (TF; solid lines) or enriched (EF; dashed lines) milk formulas for 1 y after hospital discharge. Independent main effects of formula type on bone mineral content and fat and lean mass were observed (P 0.01 for all comparisons), and male infants had greater bone mineral content and lean mass than did female infants (P < 0.01 for both) based on repeated-measures ANOVA with control for birth weight. No significant formula type x sex x time interactions were observed for bone mineral content, fat mass, or lean mass.
Comparison of the slopes for unadjusted BMC, fat or lean mass, or each as a percentage of total weight was not significant for the completers. However, there was a significantly higher slope for unadjusted BMC in the TF group compared with the EF group (0.592 ± 0.024 compared with 0.530 ± 0.017, P < 0.05l; t test) when data from all subjects were analyzed.
DISCUSSION
Despite the theoretical benefits of nutrient-enriched formula for preterm infants after hospital discharge, its role in the reversal of impaired growth and bone mineralization at hospital discharge remains poorly defined. Short-term studies reported a beneficial growth effect in preterm infants fed nutrient-enriched formula during hospitalization and for a short period after hospital discharge (20, 21). However, the growth benefits of long-term use (6 mo) of nutrient-enriched formula after hospital discharge have been inconsistent (22–24). In comparisons of infants fed enriched formula with those fed term formula after hospital discharge, greater growth was observed only in male infants fed enriched formula: this greater growth consisted of higher weight, length, and head circumference until 6 mo corrected age in one single center study (22); higher weight and length, but not head circumference, until 9 mo corrected age in a multicenter study (23); and higher weight and length at 6 mo and greater head circumference at 12 mo only for a subset of infants with birth weights <1250 g in another multicenter study (24). Bone mass and body composition were measured in only 1 of these 3 cohorts (25).
In contrast to our null hypothesis of no significant difference in growth between the groups, our results showed that preterm infants who were fed regular formula over a 1-y period after hospital discharge outperformed those who were fed nutrient-enriched formula in growth, bone mass, and body composition. The independent formula effect without interaction between formula and sex indicated the growth benefit of the term formula applied to both sexes. The major possible confounders to growth with respect to birth weight (24, 26–28), sex (15, 18, 22–24), and postnatal clinical status (1, 23, 24, 26, 28) were comparable between our study groups. Racial bias in growth response to different nutrient intake during infancy has not been reported, which is consistent with our finding that race had no independent effect on growth.
In the present study, the encouragement for the parent to provide ad libitum feeding to their infant and to discard any residual milk formula from the bottle did not allow for accurate measurements of formula intake, and no attempts were made to monitor the quantity of other food intake, which limited the information on nutrient intake. However, multiple long-term postdischarge feeding studies reported higher volume of consumption of regular formula than of enriched formula by an average of 9% to 23% in one study (24) and of 22% in another study (22); the amounts of regular formula ingested daily were consistently >200 mL/kg (22) and as much as 350 mL/kg (29), which may account for our findings.
Direct comparison of our study with earlier reports is not possible because of multiple differences between the studies. For example, the subjects in one multicenter study had higher mean birth weights and a large proportion (24% to 41%) of small-for- gestational-age infants in various cohorts studied (23) and none of the studies were stratified by sex in the randomization process (20–25, 29, 30). There were also differences in the study duration and contents of multiple micronutrients for both the standard and enriched formulas (22, 23) compared with our study. One previous multicenter study (24) that used similar formulas from the same manufacturer showed inconsistent benefit in growth from the use of enriched formula, although the results were confounded by the limited number (40%) of subjects who completed the study.
The use of standardized growth z scores eliminated any potential bias from absolute growth measures due to minor imbalances in sex, gestation, or the exact age at follow-up between the groups. The use of z scores also allowed for comparison against population norms for healthy infants born at term. In the present study, a large improvement in z scores occurred over the first 9 mo for all anthropometric measurements in both groups, and this catch-up continued until the end of the study for length and head circumference z scores. This is consistent with the postulate that most catch-up growth occurs during infancy (31), although the process of catch-up growth could continue for a much longer period (26, 32). These data indicated that the growth of these infants compared favorably with earlier reports (23, 30) and with population norms (18).
The comparison of the slopes of z scores allowed the use of data from all subjects, including those who had incomplete follow-up measurements. This statistical approach can be considered as an intent-to-treat analysis by using actual follow-up data of all subjects rather than estimated values with potential biased inferences under plausible models for the dropout process (33). In the present study, the comparison of slopes allowed for analysis of data from all subjects that had 1 follow-up measurement, which included >85% of all subjects enrolled; the result did not significantly affect the overall finding.
Consistent with other reports (22–24), our findings suggested that head circumference was the least affected variable by the type of postnatal nutrition support. Our findings were also consistent with the similar neurodevelopment status observed in infants with different nutrient intakes despite significant differences in growth variables including head circumference (23, 24, 34).
Our data showed that sex and formula type had significant independent effects on bone and body composition. The greater bone and lean mass in the boys was not surprising and was consistent with their greater body weights and lengths. The apparently dramatic differences in body composition between the groups, with greater absolute amount of bone, fat, and lean mass in infants who were fed regular formula than in those who were fed EF, reflected the overall greater body weight and length in these infants. Furthermore, the body composition of the same infants were within the ranges reported for healthy infants at the same body weight (16, 17) with the use of the same DXA pencil beam technique. In any case, consistency in the amount of covering and the technique in scan acquisition and analysis in our studies (9, 10, 35) indicated that the differences in body composition between the formula groups were not the result of an artifact due to different coverings during the DXA scan. The lack of formula effect on bone mass and body composition in girls reported by other investigators (25) was presumably associated with a lack of stratification by sex or a lack of statistical power.
In the present study, neither sex nor race had an independent effect on bone, fat, or lean mass percentage. This would support the conclusion that diet is the major determinant of changes in bone mass and body composition in infants. The lack of significant difference in the slopes for the rates of change in bone and body composition measurements may indicate a high degree of homogeneity between the groups or a lack of power, because these aspects were not the primary outcome studied.
Our data, if confirmed, potentially has significant implications for both in-hospital and postdischarge nutrition support practices. In most hospitals, the current practice of arbitrarily weaning preterm infants to fewer feedings per day at a volume of 150 to 180 mL · kg–1 · d–1 may accustom the parents to stop feeding when the infant finishes the "expected" amount, thus inadvertently restricting intake. Many formula-fed preterm infants in the United States are eligible for free formula from the Women, Infant and Children program, which provides a fixed amount of 750 mL infant formula per day, a policy that may arbitrarily restrict milk intake for the infant.
Both the commercial term and postdischarge formulas were reformulated, primarily with the addition of long-chain polyunsaturated fatty acids (36) and nucleotides. However, the overall energy density and the remaining micronutrient contents were essentially unchanged. Preterm infants fed nutrient-enriched in-hospital and postdischarge formulations with or without long-chain polyunsaturated fatty acids fortification showed no significant difference in growth (37, 38). Additional studies are warranted to confirm our findings, because the data may have significant implications on the post–hospital discharge nutrition management and the public nutrition support program for small preterm infants.
ACKNOWLEDGMENTS
The study sponsor placed no limits on the statements in the final manuscript.
WWKK participated in design and execution of the study, analysis and interpretation of the data, and completion of the manuscript. EMH participated in statistical analysis, interpretation of the data, and manuscript writing. The authors had no financial or personal interest in the organization sponsoring the research.
REFERENCES