Introduction

Postnatal growth failure, which is associated with impaired neurocognitive and renal development [1,2,3,4], was observed in nearly 100% of very low birth weight infants at the end of the 1990s and still occurs in more than half of these infants [5]. In the Swedish EXPRESS cohort of ELBW infants, the median Z-scores for body weight were −0.66 at birth and −1.84 at 36 weeks PMA. Importantly, 44% of these infants had a Z-score below -2 standard deviation (SD) at discharge [6]. Although length growth deficit is also very common, the vast majority of children gradually catch up between the ages of 2 and 8 years and are within normal height ranges as adults [7, 8]. A deficit in fat-free mass (FFM) at discharge has been associated with suboptimal neurological outcomes, and the proportion of fat mass (FM) is known to be higher in premature infants compared to foetuses of the same gestational age (GA) [9, 10].

Human milk fortification strategies used in neonatal units worldwide are highly variable [11, 12]. Individualised fortification, whether adjustable or targeted, achieves better postnatal growth than standardised fortification [13, 14]. Adjustable fortification consists in modulating protein enrichment according to serum urea levels, while targeted fortification is based on the analysis of breast milk to adjust protein or energy content [11, 13]. The latter is time- and resource-consuming and its superiority over adjustable fortification have not been demonstrated [14].

The aim of this study was to evaluate the frequency of postnatal growth deficit and assess body composition at discharge in ELBW infants using an individualised nutritional care approach combining standardised fortification followed by adjustable fortification adapted to weight gain and serum urea.

Population and methods

This single-centre retrospective observational study included infants born with a birth weight less than 1000 g, admitted within the first 24 h of life to the neonatal intensive care unit of the Croix-Rousse University Hospital in Lyon (France), and still hospitalised at 35 weeks PMA. Infants with serious congenital malformations were excluded.

Data were prospectively recorded in the patient’s electronic files (ICCA, Philips, Boblingen, Germany). Daily protein and energy intakes were assessed on the first day of each week of life and compared to recommended intakes (protein: < 1 kg: 4.0–4.5 g/kg/day, 1–1.8 kg: 3.5–4.0 g/kg/day, energy: 110–135 kcal/kg/day) [15]. Serum urea was measured weekly. Bronchopulmonary dysplasia (ventilatory support or oxygen therapy at 36 weeks PMA), intraventricular haemorrhage grade 3 or 4, periventricular leukomalacia, retinopathy of prematurity stage ≥ 3, and necrotising enterocolitis stage ≥ 2 were collected.

Body weight was measured daily during the first week of life, and then weight, crown-heel length, and head circumference (HC) were measured weekly. The length was measured using a rigid measuring board suitable for premature newborns (Premie Stadiometer, Ellard instrumentation, Monroe, USA). Anthropometric data were expressed in absolute values and Z-scores, and differences in Z-scores between birth and discharge were calculated [16]. Infants were considered to be small for GA (SGA) when the Z-score for body weight was ≤  −1.28 (10th percentile equivalent) [16]. PNGF was considered when the Z-score loss between birth and discharge was higher than 1 or when the Z-score for body weight at discharge was ≤  −1.28. Air displacement plethysmography (PEA POD®, Cosmed France, Brignais, France) was performed between 35 and 41 weeks PMA. Both FM% and absolute values of FFM were collected. Since at 35 to 41 weeks, infants were 2 to 4 months old, the data from infants born at 35–41 weeks as well as the data obtained in 2-month-old term infants were used as a reference [10].

Parenteral nutrition was started within the first 2 h of life, was individualised as soon as possible (within 48 h), and continued until the enteral ration reached 120 mL/kg/day. Enteral nutrition started on the first day of life using donor human milk. Then, the mother’s own milk was introduced as soon as possible, when available. It was pasteurised up to 32 weeks of corrected age [17]. Enteral nutrition increased daily from 15 to 20 mL/kg/day, up to 160 mL/kg/day. Energy supplement started at the end of parenteral nutrition and continued until milk intake reached 160 mL/kg/day: Liquigen® 4 g/100 mL (Nutricia, Saint Ouen, France). Fortification of human milk was started when enteral nutrition reached 70 mL/kg/day. Initially, all infants received a standardised fortification with a powder multicomponent fortifier: Fortipre® 4 g powder/100 mL (Nestlé, Noisiel, France) or Fortema® 3 g powder/100 mL + Nutriprem® 0.5 g powder/100 mL (Bledina-Danone, Limonest, France) (Table 1) [11]. Weight gain calculation and serum urea assessment were performed weekly. When weight gain was insufficient, i.e. < 20 g/kg/day, enteral intake was increased to 180 mL/kg/day. If it remained insufficient after 1 week, the fortification was individualised. Individualised adjustable fortification consisted of the addition of a protein supplement (Nutriprem® 1 g/100 mL) if serum urea was low (< 3.5 mmol/L) or an energy supplement (Liquigen®: 4 g/100 mL) if the serum urea was normal (> 3.5 mmol/L). Additional protein was reduced if serum urea was above 6.5 mmol/L. Fortification of human milk was maintained until body weight reached 1800 g. When the mother had not—or not enough—milk, donor human milk was used to complete or replace the mother’s own milk, and then, when body weight was equal to 1800 g, donor human milk was replaced by a preterm formula.

Table 1 Composition of multicomponent fortifiers and protein supplement (per 100 g powder)
Full size table

For statistical analysis, continuous variables were described by their means and SD, and comparisons were performed using Welch’s t-test or the nonparametric Mann–Whitney test. Categorical variables were described by the number of occurrences and percentages, and comparisons were performed using the chi-square test or Fisher’s exact test, as appropriate. Univariable and multivariable logistic regression analyses were performed to identify factors potentially associated with PNGF. Variables with p < 0.05 in univariable analysis were retained in the multivariable model. Results are presented as odds ratios and their 95% confidence intervals [95%CI]. The search for factors potentially associated with FM% was carried out using univariable and multivariable linear regression analyses. Variables with p < 0.05 in univariable analysis were retained in the multivariable model. Results are presented as an estimate with [95%CI]. The alpha-risk significance level for all analyses was set at 0.05. All analyses were performed using R software version 3.5.3 (R Foundation for Statistical Computing, Vienna, Austria).

The study was approved by the ethics committee Comité de Protection des Personnes Sud-Est IV (IRB: 00009118), and the institutional review board (Comité scientifique et éthique des Hospices Civils de Lyon, n°22_608) was registered in Clinicaltrials.gov (NCT02686801).

Results

Between April 1, 2014, and December 31, 2019, 310 infants among the 490 infants with a birth weight less than 1000 g admitted to the neonatal intensive care unit were included. Body composition was assessed in 112/310 (36%) infants (Fig. 1). At birth, mean ± SD GA was 26.7 ± 1.8 weeks with a minimum of 23 weeks. The mean birth weight was 800 ± 128 g with a minimum of 440 g (Table 2).

Fig. 1
figure 1

Flow chart

Full size image
Table 2 Characteristics of 310 extremely low birth weight infants at birth and during hospitalisation
Full size table

Recommended protein intakes were reached before the end of the first week of life. Recommended energy intakes were reached at the end of the second week and were exceeded between the third and eighth week of life, bringing the protein-energy ratio slightly lower than recommended (Fig. 2). Mean serum urea levels were 4.2 ± 3.3 mmol/L at 1 month of life, 3.8 ± 2.4 mmol/L at 6 weeks of life, and 3.5 ± 1.9 mmol/L at discharge. The threshold value of 6.5 mmol/L was exceeded in 14.8% of infants at 1 month, 8.2% at 6 weeks, and 6.0% at discharge.

Fig. 2
figure 2

Weekly protein intake a, energy intake b, and protein to energy ratio c in 310 extremely low birth weight infants between birth and discharge, expressed as mean and standard deviation. Grey zones represent recommended intakes (Ref. 15)

Full size image

At birth, the mean Z-score was −0.39 for body weight, −0.44 for length, and −0.26 for HC. The mean initial weight loss was 10 ± 4% of birth weight (Table 3). Thereafter, weight gain in both boys and girls closely followed the reference curves (Fig. 3). At 1 month of life, 62 infants (20.0%) had a body weight < 10th percentile. The mean Z-score difference between birth and discharge was almost nil for HC (0.03 ± 1.12), moderately negative for the weight (−0.32 ± 0.75), and more strongly negative for length (−1.21 ± 0.92) (Fig. 4). Overall, 84 infants (27.1%) had a body weight < 10th percentile and 26 (8.4%) had a Z-score for body weight below the 3rd percentile at discharge. A total of 114 infants (36.8%) presented PNGF (Table 3). The multivariable analysis identified SGA at birth and the use of postnatal steroids as independent risk factors for PNGF. The proportion of total milk intake as the preterm formula was a protective factor (Table 4).

Table 3 Anthropometric data and postnatal growth in 310 extremely low birth weight infants
Full size table
Fig. 3
figure 3

Postnatal weight gain (g/kg/day) in 310 extremely low birth weight infants (a boys; b girls). The grey lines represent the Fenton weight gain curves, and the black lines represent the mean weight gain of the present cohort (solid line) + / −1 standard deviation (dotted lines)

Full size image
Fig. 4
figure 4

Differences in Z-score for body weight, crown-heel length, and head circumference between birth (light grey) and discharge (dark grey) in 310 extremely low birth weight infants. Expressed as boxplot (median, P25, P50, min, max)

Full size image
Table 4 Risk factors of postnatal growth failure in 310 extremely low birth weight infants
Full size table

Body composition assessment could not be performed in 198 infants, mainly because of ventilatory support (Fig. 1). The infants who underwent a body composition assessment had similar characteristics to the others, except for their birth weight, which was significantly higher. They were also less sick and had less PNGF (Table 5). Measurement of body composition was performed at a mean of 67 ± 15 days of life, i.e., 38 ± 1 weeks PMA. FM% was 19.8 ± 3.6% of body weight, and FFM was 2314 ± 389 g (Fig. 5). FM% tended to be lower in infants that were SGA at birth than in non-SGA infants (18.8 ± 3.6% vs. 20.1 ± 3.4% p = 0.093). FM% was significantly higher in infants with optimal growth compared to those with PNGF (20.2 ± 3.2% vs. 18.1 ± 4.5%, p = 0.036). Factors influencing FM% were the proportion of milk intake as preterm formula and GA at birth. Those influencing FFM were GA at birth and sex (Table 6).

Table 5 Characteristics of extremely low birth weight infants with or without PEA POD®
Full size table
Fig. 5
figure 5

Fat mass (%) and fat-free mass (g) at discharge in 112 extremely low birth weight infants of the present cohort (dark black) compared to term infants at birth (white) and term infants at 2 months of age (grey). Reference values for term infants were obtained from ref. 10

Full size image
Table 6 Factors associated with fat mass percentage (a) and fat-free mass (b) at discharge in 112 extremely low birth weight infants. Results are expressed as estimates with their 95% confidence interval
Full size table

Discussion

In the present cohort of very high-risk ELBW infants, the individualised nutritional care approach applied prevented postnatal weight loss in most infants, limited length deficit, and supported excellent HC growth.

Protein intakes were close to the recommended intakes [15]. The recommended total energy intake, which may be difficult to achieve in such extremely immature infants, was reached faster than previously reported [15, 18]. Although the protein-to-energy ratio was slightly lower than recommended due to the high energy intake, the former remained higher than previously reported [19]. The high energy intake observed is likely due to the fact that the energy supplementation required after cessation of parenteral nutrition was not stopped as soon as recommended, reflecting the difficulties in fully adhering to protocols in clinical practice. However, such intakes supported early postnatal growth, as only a fifth of infants were SGA at 1 month of life, compared to the 75% previously reported, representing a significant improvement in the prevention of initial growth deficit originally described by Embleton et al. [18, 20]. Good protein utilisation was reflected by rather low serum urea. These results suggest that the slightly excessive energy intake relative to the protein intake avoided the restriction of protein utilisation which could be related to a lack of energy. Although such intakes allowed good postnatal growth in most infants, they also likely favoured high FM% at discharge. These data advocate for close monitoring of protein and energy intakes, but also that of the protein-to-energy ratio.

The postnatal growth observed in this cohort closely followed that of foetal growth, at least for body weight and HC. The present individualised nutritional care approach helped to avoid the postnatal weight deficit as demonstrated by a Z-score loss much lower than previously reported (−1), despite the fact that the infants herein were less mature [20]. This deficit was also lower than that reported more recently by Cormack et al. (−0.48) in a similar population and even lower than the −0.7 to −1 Z-score loss recently proposed as acceptable [11, 21]. In the present cohort, there were four times less infants with a Z-score for body weight below −2 at discharge than in the EXPRESS cohort [6]. Moreover, less than a third of infants herein had a Z-score for body weight below −1.28 at discharge, which is lower than previously reported in even more mature very low birth weight infants [23]. The observed absence of postnatal HC deficit in the majority of infants herein is also noteworthy, as such a deficit has been associated with suboptimal neurological outcomes [1, 24]. In a similar population, Cormack et al. reported a higher Z-score loss of 0.82 [22]. In the EPICure cohort, ELBW infants with a significant deficit in postnatal HC growth had HC below reference values as adults [25]. Although there is no strong evidence supporting that having an HC close to the mean for GA at discharge is associated with better neurodevelopment, it seems rather reassuring for the future of these high-risk infants. The length deficit observed herein was similar to the −1.5 and −1.16 previously reported [21, 26]. Surprisingly, length data are quite rarely reported in studies assessing postnatal growth in ELBW infants [20, 27], and very few authors reported an improvement in the Z-score for length during hospitalisation [28, 29]. It is well known that the final height of premature infants is approximately 1SD lower than that of term infants [30]. Furthermore, since postnatal length growth deficit can be associated with long-term consequences such as osteoporosis, large deficits in length should be avoided as much as possible [31]. The few studies that found such positive length kinetics underlined the central role of protein intake [28, 29]. This represents another reason for optimising protein intakes and protein-to-energy ratio. In summary, and in contrary to previously reported studies, the individualised nutritional care approach used in the present cohort helped limit postnatal growth deficits [32,33,34].

Of note, PNGF might be a more relevant marker of growth deficit than just SGA at discharge [35]. When using PNGF, the postnatal growth deficits in weight, length, and HC were present in more infants than when using SGA at discharge. Neonatologists should aim to reduce the risk of PNGF rather than SGA at discharge. This study confirmed that SGA at birth and postnatal steroids are independent risk factors for PNGF and found that the proportion of milk ingested as the preterm formula was a significant protective factor of PNGF. This could be due to the fact that preterm formula, which is used to supplement or replace absent or insufficient breast milk provides, a more stable nutritional supply than fortified breast milk.

Currently, there is no consensus regarding the body composition objective at the end of hospitalisation. Due to the metabolic adaptation to extrauterine life needed to increase energy storage and improve thermoregulation, the foetal body composition cannot serve as a reference [36]. Given that full-term neonates have an FM% of around 10% at birth and 25% at 2–3 months of life [11, 37] and that the FM% at discharge in the present ELBW cohort was 20%, the objective could be between that of a 36–40 week foetus and that of a full-term infant aged 2–3 months. This relatively high FM% at discharge is similar to that observed in smaller cohorts with similar postnatal weight change but higher than the 15% reported in less immature infants with a more favourable postnatal weight change [37, 38]. Contrary to what has been reported, the results herein showed that each additional GA week at birth resulted in a decrease in FM% [39]. Thus, the more immature the infant, the higher the FM% at discharge, which could reflect an increase in fat storage due to the difficulty in maintaining well-balanced protein and energy intakes throughout hospitalisation. Nevertheless, FM% has been shown to normalise within a few months after discharge [10]. Such a transient excess in FM% could thus be useful for ELBW infants, as it may represent the “price to pay” to avoid postnatal growth deficits, particularly regarding HC.

A deficit in FFM at discharge has been associated with neurological impairment at 2 years of age [9, 40]. Herein, FFM was lower than in term infants (2.8–2.9 kg), confirming the data published by Hamatschek et al. (mean of 2.5 kg), but was 300 g higher than that reported in a very low birth weight cohort [10, 39]. However, as the FFM is expressed in absolute value, it depends directly on body weight, and it is therefore difficult to compare studies, in which nutritional care and body weight at discharge varies greatly.

A limitation of this study is its single-centre design, although this did not prevent the data from a significant number of ELBW infants to be analysed. Furthermore, it avoided the impact of potential inter-centre differences in practices other than nutritional management, which could impact postnatal growth. Moreover, only a subgroup of infants could benefit from the body composition assessment herein. They had less bronchopulmonary dysplasia and therefore, less postnatal steroid treatment. However, even though the most severely ill infants did not undergo a body composition measurement, those who had it still represented a population of very high-risk ELBW infants.

In conclusion, an individualised nutritional care approach using standardised fortification followed by adjustable fortification limited body weight and HC postnatal growth deficits. FM% was higher than that of foetuses of the same GA, possibly representing a necessary adaptation to extrauterine life. Further studies are still needed to determine the growth and body composition objectives in ELBW infants according to their impact on later development.