In our study we found that rates of EUGR varied significantly according to charts and definitions used, resulting higher when using INeS charts both when applying a cross-sectional and a longitudinal-EUGR-1 definition. At multivariate analyses, LOS and a longer time for reaching 100 ml/kg enteral feeding were associated with a higher risk of EUGR at discharge, while maternal preeclampsia reduced this risk.
In line with other studies, in our population frequency of SGA was higher, though not significantly so, when using the Intergrowth-21st growth charts [9, 10, 18]. This result is expected, as Intergrowth-21 charts were plotted with a standard population, with the lowest risk factors known to affect prenatal and postnatal growth (low-risk women, non-smokers, with a normal pregnancy history with normal growing fetuses) and this may explain a tendency to overestimate SGA newborns.
On the contrary, when using Intergrowth-21th charts, EUGR prevalence was lower when compared to Fenton and INeS charts, both with the cross-sectional and the longitudinal definition. This result is consistent with previous studies [9, 10, 18-20] and this difference may be expected, as the Intergrowth charts are based on longitudinal growth of preterm babies, who grow in a completely different environment and with different metabolic responses compared to fetuses in the intrauterine environment. The difference between Intergrowth-21 and INeS charts is even more significant than between Intergrowth and Fenton and this could reflect the more homogeneous population that were used to design them.
Ideal growth for preterm babies is far from being clearly defined. One of the widely accepted goals of neonatal nutritional care is to try to replicate the intrauterine growth as close as possible [21]- and, for this reason, the most commonly used growth charts have been the cross-sectional charts based on in utero growth. However, it is still debated whether it is appropriate to expect preterm babies to grow at the same rate as those in utero [22]. Benefits of improved growth on neurodevelopmental outcomes and chronic lung disease are well known, but the optimal pattern of growth in preterm infants to achieve good long-term health should also take into account the potential impact of excessive growth, associated with the risk of metabolic and cardiovascular disease in later life. The Intergrowth 21th Project therefore developed new growth standards from a cohort of uncomplicated pregnancies with normal growing fetuses, uncomplicated postnatal period and up-to-date nutritional support [8], and explain why EUGR prevalence – regardless of its cross-sectional or longitudinal definition – turns to be lower when compared to other reference charts.
These trends to lower rates of EUGR when using Intergrowth-21 charts have been reported by several other studies, both for cross sectional EUGR definition [9, 10] and longitudinal EUGR [18]. In a European multicountry cohort, cross-sectional EUGR rates were reduced from 24% (Sweden) and 60% (Portugal) when using Fenton charts to 13% (Sweden) and 43% (Portugal), when using Intergrowth-21 [23].
Our study was done because knowing and monitoring the prevalence of EUGR in our Unit, is considered to be a quality measure of care for preterm infants [3]. In our population EUGR prevalence ranged from 40.9% using cross-sectional definition with INeS charts to 4% using longitudinal-1 definition with Intergrowth-21 standards. However, to draw comparisons between different centers is difficult not only because definition of EUGR and reference growth charts differ, but also because maternal and clinical characteristics of the study population, time at EUGR evaluation (36 weeks, 40 weeks PMA, discharge) and choice of ΔSDS threshold when applying the longitudinal-EUGR definition (> 1, >2 SDS), changes. EUGR rates from NICUs of high-income countries similar to ours, evaluated at discharge, varied from 17% to 77.2% for the cross-sectional EUGR [4, 25-29], from 29.8-39.1% for longitudinal-1 definition [30,31] and from 5.2-13% for longitudinal-2 EUGR [30-32].
When looking for risk factors, we confirmed that the characteristics of the study population are determinant to EUGR at discharge. Other studies demonstrated that the degree of longitudinal EUGR is influenced by birth weight z score and gestational age: the lower the birthweight centile, the lower the probability to lose 1 or 2 SDS; on the other hand, the lower the gestational age, the higher the probability to lose z-score [18, 24, 31]. In our series we found a significant role of lower gestational age and, consequently, longer duration of hospitalization on EUGR. Male sex was also significatively related with poor growth, as reported in other studies [18, 33, 34]. Therefore, we decided to adjust our logistic regression for low z-score at birth (SGA), male sex and duration of hospitalization.
Independently of the above factors, in our population we found that LOS more than doubled the risk for EUGR, although not statistically significant. In other studies [18, 24, 34, 35] other comorbidities associated with prematurity had a significant effect on the incidence of growth restriction: patent ductus arteriosus, broncopulmonary dysplasia, necrotizing enterocolitis and late-onset sepsis, need for assisted ventilation, exposure to postnatal steroids and major brain lesions. These risk factors may only be markers for severity of illness: sick infants are often fed less than healthier infants, have increased metabolic demands, and their nutritional needs are rarely met, all of which result in malnutrition and poor growth. However, the presence of LOS as an independent risk factor underlines the necessity for implementing interventions targeted to reduce the incidence of neonatal sepsis [36, 37].
In our analyses longer time to reach 100 ml/kg/die of enteral feeding was found to be a potential risk factor even though only with a borderline statistical significance. Again, this may be an indirect sign of severity of illness, as increasing oral nutrition may be harder in sick infants. Intervention studies showed that optimising nutrition (such as introducing guidelines for increasing feeds and weaning PN) reduced the incidence of EUGR [38-40]
In our study it was surprising to find that preeclampsia was protective for EUGR, reducing it of 83% . As far as we know, no other study found preeclampsia as a protective factor for EUGR. However, IUGR, which is frequently associated with preeclampsia, was found to be protective on longitudinal EUGR in a Spanish cohort of preterm babies [11]. This finding may be explained by the fact that having both a low z-score at birth and losing >1 SD may be difficult to occur. Risk factors for EUGR may vary according to the definition used with a low z-score at birth as a possible risk factor for cross-sectional-EUGR, while higher z-score at birth for longitudinal-EUGR [10, 11,18].
Limitations of our study are inherent to the retrospective observational nature of the study. Moreover, our cohort is small and includes a wide range of gestational ages. Another limitation may lie in the choice of discharge as a time point for assessing EUGR, as there is a wide range of time of evaluation. However this was taken into account in the multivariate analysis by correcting for duration of hospitalization. Moreover, as this study was conducted in a single medical centre, generalization of the data is limited.
In conclusion, our study confirms a wide variability of EUGR rates when using different charts and definitions. It highlights the need to standardize criteria and the evaluation method for EUGR, which would facilitate comparisons between studies and help to improve nutrition in neonatal units and to perform studies on its long term implication.