Extrauterine Growth Restriction and Catch-up Growth Following NICU Discharge: a Tale of two Standards

Objectives: To assess extrauterine growth restriction (EUGR) after neonatal intensive care unit (NICU) discharge and identify factors associated with catch-up growth (CUG) and neurodevelopmental outcomes at age 2. Study Design: Growth parameters at birth, 36 weeks postmenstrual age (PMA), and two neurodevelopmental follow-up visits were reviewed retrospectively. EUGR was assessed using Fenton and Intergrowth-21 standards. Factors associated with outpatient growth and neurodevelopmental outcomes at age 2 were evaluated using logistic and linear regression. Results: 369 infants comprised the cohort. EUGR prevalence was 14.9% for Intergrowth-21 vs. 56.4% for Fenton (p < 0.001). There was a signi�cant decrease in weight z-score from birth-36 weeks PMA. Weight z-scores returned to birth values by the 2nd clinic visit. Formula-fed infants had faster CUG. Higher Bayley-III at age 2 were associated with human milk feeding. Conclusions: Fewer infants were EUGR using Intergrowth-21. CUG with return to birth weight z-score was seen. Human milk feeding positively impacted Bayley-III scores, despite slower CUG.


Introduction
Appropriate postnatal nutrition is crucial to optimize long term neurodevelopmental outcomes while avoiding the unintended consequences of aggressive nutritional approaches (1)(2)(3).Extrauterine growth restriction (EUGR) is the failure of preterm infants to reach their potential growth (4) and can partially be explained by periods of inadequate nutrition, feeding intolerance and a range of mild to severe morbidities associated with preterm birth.Prenatal and postnatal growth rates have been shown to be important for long-term neurodevelopmental outcomes (5).A cross-sectional de nition of EUGR includes newborns with weight, length, and head circumference less than the 10th percentile based on standard postnatal growth curves (6).Another de nition considers EUGR as the decrease in weight z-score between birth and a given time point of greater than one standard deviation; this resembles a longitudinal de nition (5).
Accurate monitoring of postnatal growth in preterm infants is critical for guiding nutritional protocols and for identifying EUGR.Currently, there is no consensus as to which growth assessment tool is best for monitoring the postnatal growth of preterm infants.The commonly used Fenton growth standard is based on intrauterine growth rates up to 36 weeks post menstrual age (PMA) (7).Ideally, a postnatal growth standard would not be based on intrauterine growth rates but would describe longitudinal postnatal growth of healthy preterm infants.New international standards for newborns without risk factors for fetal growth restriction born to healthy mothers have been described as part of the Intergrowth-21 project (8, 9).
When the postnatal growth standards used to monitor infant growth during the birth hospitalization are based on intrauterine growth rates, many neonates discharged from the NICU qualify as having EUGR, even if there has been catch-up growth (CUG).CUG can be de ned as achievement of a growth percentile ≥ 3rd or ≥ 10th for either weight, length or head circumference (10).Others use cut-offs of change in weight-for age z-score of > 0.67 (11) or a return to their birth weight percentile.Westerberg et al (12) de ned CUG as a positive z-score change greater than 0. The obvious issue with variability in de ning CUG is that comparison of ndings across studies is limited.
The aim of our current study was to assess the prevalence and progression of EUGR, using both Fenton and Intergrowth-21 growth standards, and to identify factors associated with improved CUG following NICU discharge and with neurodevelopmental outcomes at 2 years of age as assessed by the Bayley Scales of Infant and Toddler Development (Bayley-III).

Methods
In this single center retrospective study, preterm infants born at 24-32 weeks gestation and admitted to the neonatal intensive care unit (NICU) at Mount Sinai Hospital (a level IV NICU with ~ 8000 deliveries per year) between January 2017 and March 2022 were included.Infants born at gestational ages below 24w0d or above 32w0d were excluded from this study, as were patients who did not follow up at our Mount Sinai Hospital NICU neurodevelopmental follow-up clinic.This project was approved by the Program for the Protection of Human Subjects at the Icahn School of Medicine at Mount Sinai and was performed in accordance with the Declaration of Helsinki.
For this study, we used the de nition of EUGR as a decrease in weight z-score from birth to 36 weeks PMA of more than one standard deviation.To evaluate growth, weight, length and head circumference, and their corresponding z-scores at birth, 36 weeks PMA, and rst and second NICU neurodevelopmental follow-up clinic visits were recorded, and analyzed with repeated measures ANOVA.Growth was compared using Fenton (up to 50 weeks PMA) and Intergrowth-21 growth charts (up to 64 weeks PMA) and analyzed using the McNemar test.Growth curves and z-scores at 64 weeks PMA correspond to the WHO growth curves for 0-2 years of age (9), and were used for follow-up data for PMA > 64 weeks.
To identify factors that are associated with improved CUG after NICU discharge and improved neurodevelopmental outcomes at two years of age, multivariable logistic regression analyses were conducted.Linear regression models were constructed by sequential addition of covariates with p values < 0.10 in the univariable analysis and retention of those covariates with p < 0.05 in the multivariable analysis.The following factors were analyzed: sex; GA at birth; maternal age at delivery; maternal body mass index (BMI) at delivery; number of antenatal steroid doses; singleton vs. multiple gestation; mode of delivery; IUGR; length of NICU stay; discharge disposition (home, rehabilitation center, transfer to a different hospital); outpatient diuretic, re ux, and/or respiratory medications; PMA at discharge; duration of postnatal steroid therapy; duration of total parenteral nutrition (TPN); duration of NPO; percent of birth weight lost at nadir; days to regain birth weight; duration of mechanical ventilation; bronchopulmonary dysplasia (de ned as treatment with oxygen > 21% for at least 28 days), continuous positive airway pressure use at 36 weeks PMA, and/or discharged from NICU on supplemental oxygen; discharged home on tube feedings; medical insurance status; home zip code; mean income from home zip code; intraventricular hemorrhage (IVH) grade III/IV; culture positive sepsis; patent ductus arteriosus treated medically or surgically; necrotizing enterocolitis (NEC) Bell stage II/III; retinopathy of prematurity (ROP) treated with either bevacizumab injection or laser photocoagulation; number of packed red blood cell transfusions; number of days of treatment with pressors; receipt of occupational, speech or physical therapy at the time of rst and second NICU follow up clinic visits; and any amount of human milk feeding at discharge, rst and second clinic visits.
As a marker of catch up growth from 36 weeks PMA to rst clinic visit, and from rst to second follow up clinic visit, multivariable linear regression models were used to identify factors associated with an increase in weight z-score.A logistic regression model was used to calculate the odds of associated factors with an increase in weight z-score from 36 weeks PMA to rst clinic visit and the odds of having an increased z-score for weight from rst to second follow up clinic visit.
As a marker for their neurodevelopment, Bayley-III scores were evaluated at two years of age and a multivariable logistic regression model was used to investigate factors associated with a change in Bayley-III scores.

Results
Between January 2017 and March 2022, there were 647 NICU graduates who had at least one neonatal follow-up clinic visit.Of these, 369 were born between 24 and 32 weeks gestation and comprised the study population.The clinical characteristics of the study population are shown in Table 1.A comparison of weight z-scores according to Fenton vs. Intergrowth-21 is presented in Fig. 1.Weight and length z-scores at 36 weeks were signi cantly different (p < 0.001 and p = 0.002, respectively) between the two growth standards; head circumference was not signi cantly different.More infants were identi ed as EUGR for weight when the Fenton growth curve was used: 56.4% of our patient population had a decrease in weight z-score ≤-1 from birth to 36 weeks PMA using Fenton vs. 14.9% using Intergrowth-21.
To analyze weight progression, the Intergrowth-21 growth standard was used to calculate z-scores up to 64 weeks PMA, with the WHO standard used thereafter.Comparison of weight z-score over time is shown in Fig. 2. Each weight z-score was signi cantly different when compared with the z-score at previous time points.Interestingly, the average weight z-score decreased signi cantly during the NICU hospitalization, but increased after 36 weeks PMA.Average weight z-scores at the follow up clinic visits were not signi cantly different from the birth weight z-scores.
To identify factors associated with improved CUG, two time periods were analyzed: from 36 weeks PMA to the rst neonatal follow up clinic visit, and from rst to second clinic visit.For both time periods, the data was modeled in two ways and the same results were produced when using multivariable linear or logistic regression models.The rst time period was examined to identify factors associated with a change in weight z-score from 36 weeks PMA to the rst follow-up clinic visit (average PMA of 56.5 weeks [SEM 0.53]).Male gender was signi cantly associated with increased change in weight z-score (coe cient = 0.305 ± SEM 0.089, p < 0.001).Some human milk and exclusive human milk feeding at the rst follow-up clinic visit was associated with decreased weight z-scores when compared with infants who were formula fed (some human milk: coe cient = -0.36 ± SEM 0.133, p = 0.007; exclusive human milk: coe cient = -0.463,± SEM 0.11, p < 0.001).Furthermore, the regression showed that the greater its deviation was from its birth weight z-score, the more it returned to its birth weight z-score (coe cient = -0.38,± SEM 0.042, p < 0.001).For the second time period, from rst to second neonatal follow up clinic visit (average PMA of 86.5 weeks [SEM 1.03]), multivariable linear and logistic regression models were used to examine factors associated with change in weight z-score.The duration of longer total parenteral nutrition (TPN) was signi cantly associated with a positive change in weight z-score (coe cient = 0.008 ± SEM 0.003, p = 0.01).As was the case in the rst time period, the greater the weight z-score deviation was at the rst clinic visit, the more it returned to its birth weight z-score (coe cient = -0.161± SEM 0.042, p < 0.001).
Factors associated with Bayley-III cognitive, language and motor scores at two years of age are shown in Table 2. Gestational age at birth and private insurance or self-pay were associated with signi cantly higher Bayley cognitive scores.Babies who had some human milk feeding at discharge and those who received exclusive human milk feedings at the rst visit had higher Bayley cognitive scores than formula fed infants.Higher grade IVH was associated with lower Bayley cognitive scores.
Table 2. a) Average Bayley scores at two years of age and b) factors associated with Bayley-III cognitive, language and motor scores in the linear regression models.
Infants with private insurance status were more likely to have higher Bayley language scores when compared with infants receiving Medicaid.Bayley language scores were also higher for infants residing in zip codes with higher average incomes.Additionally, infants receiving exclusive human milk feeding vs. formula feeds at the rst clinic visit were more likely to have higher language scores.Male gender, the need for ROP treatment during their NICU admission and need for occupational therapy at the time of the rst follow up visit were associated with lower Bayley language scores.
Increased Bayley motor scores at two years of age were associated with either full or partial human milk feeding vs. formula alone at the rst follow up clinic visit, while decreased motor scores were associated with more severe EUGR during the NICU stay, BPD, and a higher increase in weight z-score from 36 weeks to the second follow up clinic visit.

Discussion
In this study, we demonstrated that more babies are identi ed as having EUGR when using Fenton growth standards as compared to Intergrowth-21, but a return to their birth weight z-score was achieved by the second follow-up clinic visit around 86 weeks PMA.Babies who were human milk fed after discharge from the NICU had slower CUG, but higher Bayley-III scores at 24 months of age.
Horbar et al. demonstrated that postnatal growth failure and severe postnatal growth failure decreased from 2000 to 2013 in North American hospitals in the Vermont Oxford Network (13).However, growth failure in preterm infants during NICU hospitalization remains a serious problem, with a reported incidence ranging from 30 to 60% (6, 14, 15).Our study demonstrated that use of postnatal growth standards based on the assumption that postnatal growth should mimic intrauterine growth (Fenton standard) resulted in a higher prevalence of EUGR than when infants were plotted on a growth standard derived from the actual growth of healthy preterm infants monitored longitudinally after birth (Intergrowth-21 standard); this aligns with previous studies which showed EUGR was less prevalent when Intergrowth-21 was used (16-19).EUGR de ned by Intergrowth-21 categorizes infants with adverse clinical courses more elaborately (19), such that weight-based growth failure may be more strongly associated with poor neurodevelopmental outcomes at 12 and 24 months of age than Fenton (16).
Accurate monitoring of preterm infant growth is critical in guiding nutritional protocols.Nutrition delivery that exceeds infants' actual needs may cause unintended harm, including an increased risk for cardiovascular disease, obesity and metabolic syndrome later in life (20)(21)(22)(23)(24).One potential advantage of the preferential use of Intergrowth-21 is less erroneous identi cation of EUGR, and thereby less risk of overfeeding.Moreover, longitudinal monitoring after discharge is possible on the same growth curve up to 64 weeks PMA, making growth monitoring during the transition from inpatient to outpatient care seamless.An important limitation of the Intergrowth-21 growth standards is that few babies born before 33 weeks' gestation could be included, as it was di cult to identify healthy babies with decreasing gestational age at birth.Therefore, data was derived using a very small cohort of preterm infants at 33-37 weeks gestational age (8, 9) that were later supplemented with measurements from 408 neonates who had been previously excluded (infants born at 24-33 weeks gestational age to mothers with some risk factors for fetal growth restriction) (25).Therefore, no z-scores were available for size at birth for gestational ages less than 24 weeks and we excluded these infants from our study.
The literature suggests that preterm infants fed an exclusive human milk diet experience slower growth compared to those receiving preterm formula, which places them at greater risk for developing EUGR (26-28).Ehrenkranz et al. demonstrated that extremely low birth weight (ELBW) infants with slower growth velocity in the NICU exhibited poorer neurocognitive outcomes at 18-22 months corrected age (29).However, it is worth noting that preterm infants fed human milk demonstrate improved neurodevelopmental outcomes in comparison to those fed preterm formula (30)(31)(32).We found that human milk feeding had a positive impact on cognitive, language and motor developmental outcomes at 24 months of age, despite lower weight z-scores after NICU discharge, which is in alignment with Rahman et al. who found that EUGR in ELBW infants fed an exclusive human milk diet did not negatively impact neurodevelopment at 24 months (33).
Few studies have examined the progression of weight z-scores after NICU discharge.We found that infants born between 24 and 32 weeks GA had a return to their birth weight z-score by the time of their second NICU follow up clinic visit.
McGowan et al. found that infants with ≥ 2 medical morbidities had lower Bayley scores at 2 years of age (34).Our study showed that having a history of IVH grade III or IV, ROP treatment and BPD, being male, needing occupational therapy, and having more severe EUGR were all associated with lower Bayley-III scores at two years of age.In addition to human milk feeding, higher Bayley-III scores were also associated with increasing GA, and private insurance.Similarly, previous studies showed an association of economic vulnerability with an increased risk of adverse neurodevelopmental outcomes at 2 and 5 years of age (35).
Our study had several important limitations.The study reviewed data obtained from patients before and during the Covid-19 pandemic.In 2020 and 2021, many fewer follow-up visits were conducted in person.
Our study may furthermore be biased towards those patients whose parents are presumably more compliant to medical recommendations and not lost to follow up.Telehealth visits could not collect anthropomorphic measurements or perform detailed neurodevelopmental testing.We cannot know what impact stay-at-home policies and social distancing had on the social and learning environments of these infants and toddlers.Many children spent more time at home with close relatives and less time in daycare and school settings; the downstream effects of these differences in socialization on their neurodevelopment is not yet known.
While we did not assess the protein content of nutrition during the NICU hospitalization, improved change in weight z-score between the rst and second follow up visits was signi cantly associated with TPN duration, suggesting that higher protein delivery early in life might be bene cial for later CUG.Interestingly, Stephens et al. found that increased rst-week protein and energy intakes in extremely low birth weight infants were associated with higher Mental Development Index at 18 months of life (36).
Furthermore, we were not able to determine if a mother was nursing her infant directly or bottle-feeding expressed human milk and relied on mothers' reports of feeding type, amount and frequency.Unfortunately, it was also not recorded in the patients' chart when the mothers switched from one feeding type to another.
Despite these limitations, and although infants had decreased weight gain from 36 weeks PMA to the rst NICU follow up visit when they were receiving human milk as opposed to formula, our results suggest that human milk feeding practices may play an important role in language, motor and cognitive developmental outcomes for preterm infants.The identi cation of risk factors such as BPD, IVH, need for ROP treatment and male gender may help clinicians target interventions and support to infants at higher risk for developmental delays.We conclude that formula feeding at the expense of human milk feeding in the interest of improved CUG for babies identi ed as EUGR on Fenton growth curves may lead to unnecessary overfeeding and may instead contribute to lower Bayley scores at two years of age.

Declarations
Con ict of Interest: The authors declare that they have no competing interests.

Ethics approval and consent to participate:
This study was approved by the Program for the Protection of Human Subject of the Icahn School of Medicine at Mount Sinai with waiver of informed consent.The study was performed in accordance with the Declaration of Helsinki.

Funding:
No funding was received for this study.
Author Contributions: Dr. conceptualized and designed the study, designed the data collection instruments, collected data, carried out the initial analysis, and drafted the manuscript.
Dr. Green participated in the planning of the study, performed the statistical analysis, and reviewed and revised the manuscript.
Dr. Weintraub participated in review of the data analysis, drafted, reviewed, and revised the manuscript.
All authors approved the nal manuscript as submitted and agree to be accountable for all aspects of the work.

Figure 1 Comparison
Figure 1

Table 1 .
Characteristics of the study population in mean ± SEM or n with (%)