Association of small-for-gestational-age status with mortality and morbidity in very preterm Chinese infants

Abstract Background Very preterm infants born small for gestational age (SGA) are at risk for short- and long-term excess mortality and morbidity resulting from immaturity and deficient intrauterine growth. However, previous findings are inconclusive, and there is a paucity of contemporary data in Chinese population. Objectives To evaluate the excess risks of mortality and morbidity independently associated with SGA birth in very preterm (before 32 weeks of gestation) Chinese infants. Materials and Methods The study population included all very preterm infants admitted to the neonatal intensive care units (NICUs) in our hospital and our medical treatment partner hospitals during a 6-year period. The SGA group consisted of 615 SGA infants, and 1230 appropriate-for-gestation-age (AGA) infants were matched with GA and sex as controls at a ratio of 2:1. The associations between SGA birth and outcomes (in-hospital mortality and morbidity) were evaluated by using multivariate logistic regression analysis after adjustment for potential confounders. The CRIBII score was used to indicate admission illness severity, acting as a covariate in the multivariate analysis. Results The SGA group was associated with increased risks of mortality [odds ratio (OR) 2.12; 95% CI: 1.27–3.54] and BPD [OR 1.95; 95% CI: 1.58–2.41] compared to the AGA group. No significant incidences of respiratory distress syndrome (RDS), severe retinopathy of prematurity (sROP), severe intraventricular hemorrhage (sIVH), and necrotizing enterocolitis (NEC) were observed in the SGA group. Further GA-stratified subgroup analysis showed SGA status exhibited certain patterns of effects on mortality and morbidity in different GA ranges. Conclusions SGA status is associated with excess risks of neonatal mortality and BPD in very preterm infants, but the increased risks of mortality and morbidity are not homogeneous in different GA ranges. The contemporary data can help inform perinatal care decision-making and family counseling, particularly for very preterm SGA neonates.


Introduction
SGA is highly prevalent in low-and middle-income countries, particularly in South Asia [1].SGA birth, commonly used as a proxy for intrauterine growth restriction, increases the risk of mortality and morbidity in the neonatal period and beyond.It is generally acknowledged that infants born before 32 weeks of gestation suffer "double jeopardy," as the combined presence of prematurity and SGA confers a higher risk than that of either characteristic alone [2,3].
Although the association between SGA birth and adverse outcomes among very preterm infants has been widely investigated, the strength and direction are not fully established.The divergence has been commonly attributed to the GA range, sample size, and the presence of confounding factors, including illness severity [4].
Admission illness severity has a substantial impact on mortality and morbidity of preterm infants.However, most studies, especially population-based ones, did not use appropriate measurement to adjust for it.The clinical risk index for babies (CRIB) II is a widely used neonatal illness severity scoring system in NICU which provides a recalibrated and simplified measurement within 12h of admission, avoiding the potential problems of early treatment bias [5].Its applicability has been confirmed by many prediction models [6].
With the improvement of maternal and perinatal care over time, more contemporary data is needed to inform decision-making in NICU, family counseling, and prediction of adverse outcomes associated with SGA birth.Therefore, we conducted the present study to evaluate the excess risk of mortality and morbidity independently associated with SGA status in very preterm Chinese infants.

Study population
The medical records of all very preterm (before 32 weeks of gestation) infants admitted to the NICUs at the First Affiliated Hospital of Shantou University Medical College and its medical treatment partner hospitals between 1 January 2015 and 31 December 2020 were reviewed.Demographic and clinical data were retrieved from clinical progress notes and discharge summaries in the electronic patient record database.The records were identified using the following ICD-9 codes: preterm (644.21),premature (644.21), and small for gestational age (656.5x).Infants with a GA under 32w were included.An infant was defined as SGA if the birth weight was below the 10th percentile according to the latest sex-specific birth weight for GA standards in Guangdong Province, China [7,8].Exclusion criteria included congenital anomalies and neonatal death during delivery.
The SGA group consisted of 615 SGA infants with a GA between 24w2/7 � 31w 6/7 , and 1230 appropriatefor-gestation-age (AGA) infants were matched with controls of the same GA and sex at a ratio of 2:1.The GA in weeks and days was determined by the last menstrual period or by prenatal ultrasound if the last menstrual period was uncertain.
The ethical committee of the First Affiliated Hospital of Shantou University Medical College approved the study with a waiver of consent.
Mortality was defined as death before hospital discharge.RDS was defined as: room air partial pressure of oxygen (PaO 2 ) >50mm Hg, or supplemental oxygen to maintain a pulse oximeter saturation over 85%; and a chest radiograph consistent with RDS within the first 24 h of life [9].BPD was defined as continuous use of supplemental oxygen at 36 weeks' post-menstrual age or on oxygen at discharge at 34 to 35 weeks if discharged before 36 weeks [10].Severe ROP was defined as stages 3 to 5 on the basis of a retinal examination before hospital discharge [11].Severe IVH was defined as grades 3 or 4 by using Papile's classification within 28 days of birth [12].NEC was diagnosed based on >1 clinical signs and >1 radiographic findings [13].

Statistical analysis
Comparisons of the independent and outcome variables, in either the entire population or its GA subgroups, were made using Student's t-test or Wilcoxon rank-sum tests according to the data distribution.Normality test was applied to determine the data distribution.Continuous variables were expressed as the mean ± SD. or median with interquartile range if the data were skewed.A value of p < 0.05 was considered to be statistically significant.Post hoc sample size calculations were performed to ensure statistical power.
The following neonatal and maternal characteristics were included as covariates: sex, gestation age, birth weight, head circumference at birth, length at birth, multiple birth, in-vitro fertilization (IVF), cesarean section, use of antenatal steroids, Apgar scores at 1 min, Apgar scores at 5 min, mechanical ventilation, surfactant administration, the CRIBII score, sepsis, patent ductus arteriosus (PDA), length of hospitalization (LOS), premature rupture of membranes (>18 h) (PROM), and maternal age.
Multivariate logistic regression models were developed and performed following adjustment for the potential confounding variables.The results are presented as odds ratios (OR) plus 95% confidence intervals (CI).The variables to be adjusted included most of the above-mentioned ones, particularly CRIBII as a proxy of admission illness severity.The dependent variable was considered to be significantly associated with the outcome if the odds ratio (OR) differed from 1.0 and p < 0.05.
In regression models predicting specific outcomes, the included independent variables were those factors that might possibly affect the outcome variables, based on clinical experience and previous literature.For example, the model predicting BPD included variables such as use of antenatal steroids, mechanical ventilation, surfactant administration, and prevalence of RDS.
GA-stratified subgroup analysis can explore unique mortality and morbidity characteristics that differ from those of the group as a whole.To investigate the association of SGA with outcome variables in different GA subgroups, the logistic analysis was further performed in three GA strata: 24-27 weeks (from 24w 2/7 to 27w 6/7 exactly), 28-29 weeks (28w to 29w 6/7 ) and 30-31 weeks (30w to 31w 6/7 ).
Statistical analysis was performed by using Stata version 12 (Stata Corporation, College Station, TX, USA).

Results
A total of 1845 infants with a GA of 24-31 weeks were hospitalized in our NICUs during the 6-year period.Of them, 615 were SGA, and 1230 were AGA.Demographic, perinatal, maternal, baseline clinical data were compared between the SGA and AGA groups, as shown in Table 1.The mortality and morbidity rates were compared between the SGA and AGA groups, as presented in Table 2.
Overall adjusted odds ratios with 95% CI of outcome variables for the SGA group are presented in Table 3.The SGA group was significantly associated with increased odds of mortality [OR 2.12; 95% CI: 1.27-3.54]and BPD [OR 1.95; 95% CI: 1.58-2.41]compared with the AGA group.The SGA group did not show statistically significant odds of RDS, sROP, sIVH, or NEC.The GA-stratified subgroup analysis found that increased odds of mortality remained significant in the subgroup with a GA of 28-29 weeks.The significance of increased odds of BPD remained in the subgroups with a GA of 24-27 weeks and 30-31 weeks.Different from the insignificant results in the full GA range, the SGA birth was significantly associated with increased odds of RDS in the lowest GA subgroup.

Discussion
The present study found the GA infants who were very preterm had increased mortality and an increased incidence of BPD compared with their AGA peers with a GA of 24-31 weeks after adjusting for potential   [14].Further GA-stratified analysis confirmed similar patterns in mortality and morbidity risks across the GA range.Substantial studies reveal that premature SGA infants have been shown to be at a greater than 1 to several-fold increased risk for mortality and major neonatal morbidity.It is generally believed that the underlying mechanism is that the continuing effect of suboptimal fetal growth intensifies complications of prematurity.
Increased mortality rates related to SGA birth have been reported in most of previous studies [2, [14][15][16][17][18][19][20][21][22][23].Several lines of research suggest that excess mortality and morbidity among SGA compared with AGA preterm infants stems from an additive "double insult" due to fetal growth restriction and postnatal sequelae attributed to prematurity [3,4].A population-based cohort study proposes that the prevention of neonatal death should focus on combined preterm birth and concomitant severe SGA, which serves as a useful perinatal surveillance indicator [2].
Boghossian et al. reported that SGA infants had 1.02 to 2.87-fold higher risks of mortality compared with AGA peers starting at a GA of 23 weeks, which increased until 29 weeks in a population-based multicenter study [14].
In the analysis stratified by GA, we didn't observe the significant mortality increase that remained in the lowest GA subgroups.This pattern was similar to the findings of a case-control study from Taiwan [17] and a mortality pattern study from Greece [24].We speculate that prematurity predominates growth restriction at the early GA (24-27 weeks), which may shadow the adverse effects of SGA birth.However, such a pattern did not exactly match the findings from an early report [21].
The mechanisms for the increased mortality among premature SGA infants are not fully understood.Presumably, unfavorable intrauterine environment exerts a detrimental effect on various organs, which leads to deprivation of nutrients and oxygen, may trigger off a cascade of adverse respiratory, neurological, and metabolic events.Various adverse morbidities, including RDS, BPD, sROP, NEC, etc., may be interpreted as a constellation of physiologically linked sequelae contributing to the higher risk of mortality observed.

BPD
The need for oxygen supplementation at the age of 28 days or at 36 weeks' postmenstrual age has generally been acknowledged among SGA infants, which reflects a continuum of adverse events, including a systemic inflammatory response secondary to intrauterine hypoxia and acidosis, and more severe early neonatal lung diseases.
Multivariate logistic analysis showed the association between SGA status and BPD was significant [OR 1.95; 95% CI: 1.58-2.41]after adjusting for the confounding factors.The result was in concordance with the findings of most previous studies [14,17,21,[25][26][27].In GA stratified analysis, the increased risks for BPD remained significant in the SGA infants with a GA of 24-27 weeks [OR 2.45; 95% CI: 1. 15 [27].Other studies also found that SGA infants have an increased risk of developing BPD but did not report relative risk data [15,16,20,23,24,28].The higher risk of BPD associated with SGA can be partially explained by the following two reasons.First, SGA infants are more vulnerable to the traumatic effects of mechanical ventilation.Second, SGA infants have less chronologic time ex utero to recover from oxygen dependence.

RDS
In this study, we did not find that SGA birth is associated with RDS in the full GA range but an increased odds of RDS in the lowest GA subgroup [OR 2.89; 95% CI: 1. 26-6.84].
The association between RDS and SGA status remains inconclusive and mixed.The previous concept that increased pulmonary maturation in response to "stress" from IUGR leads to a decreased incidence of RDS has been challenged by more recently published studies.Giapros et al. and Turitz et al. showed that SGA is not associated with RDS or other adverse respiratory conditions [24,29].Nobile et al. revealed that being SGA was not related to increased risk of RDS, but significantly associated with BPD due to multiple pathophysiologic mechanisms [25].RDS may be a predisposing factor for BPD, but it is widely acknowledged that BPD can develop in the absence of RDS.Sharma et al. found no change in RDS risk in SGA infants at GA � 32 weeks [27].
Several studies have shown an increased incidence of RDS in SGA children [22,30,31].Recently, Boghossian et al. showed 1.15-to 1.43-fold excess risks of RDS among SGA infants in a large population-based study [14].Ley et al. reported 1.98-fold excess risk for RDS in SGA infants at GA 25-28 weeks, but not at GA 29-32 weeks, which is consistent with our findings [32].
It is noteworthy that the odds ratios of SGA birth are quite low even in the studies reporting positive results, which implies a weak relationship between SGA status and RDS.RDS is universal in extremely and very preterm neonates, attributed to reduced or impaired surfactant production.Antenatal glucocorticoid treatment and postnatal surfactant therapy may alleviate the adverse effects of low GA, but their effects on SGA infants developing RDS are still not clear.
Conversely, several other studies have confirmed a decrease in RDS among SGA infants.Bartels et al. found a lower incidence of RDS in SGA preterm infants in a large population-based study [33].

Other major morbidity
In our study, we did not find any significant differences in other major morbidities, including sROP, sIVH, and NEC, between the SGA and AGA groups in the full GA range.

Strength and limitation
The strengths of this contemporary 6-year single-center study are the application of the latest birthweight reference and the CRIBII score.
First, using the latest birthweight standard can minimize misclassification, which is critical for assessing the impact of SGA birth on neonatal mortality and morbidity [34].At present, the Chinese 1988 reference curve of newborn birthweight is still widely used.Obviously, a latest standard is very necessary for evaluation.Second, admission illness severity (CRIBII score) was considered a confounding covariate in the analysis, which was usually ignored in previous studies.Although GA and BW are still decisive factors in evaluating the impact of SGA birth, the effect of illness severity should not be ignored.
Despite the above points, several limitations of the study should be acknowledged.First, it is a retrospective, case-control study which is subject to potential selection biases, so the generalizability of our findings may be limited to similar settings.Second, the study did not apply propensity score matching, which is widely regarded as an effective way to mitigate the differences between experimental and control groups.Third, the relatively small sample size in relation to certain event rates may affect power in subgroup analysis.Fourth, a lack of detailed maternal and fetal growth data limited the effectiveness of conclusions.Therefore, the findings should be interpreted with caution.

Conclusions
In conclusion, SGA status is associated with excess risks of neonatal mortality and BPD in very preterm infants, but the increased risks of mortality and morbidity are not homogeneous in different GA ranges.The contemporary data can help inform perinatal care

Table 1 .
Demographic and perinatal characteristics of the SGA and AGA groups.Presented as median (interquartile range).IVF in-vitro fertilization.CRIBII The clinical risk index for babies version 2; PDA patent ductus arteriosus; PROM premature rupture of membranes. #

Table 2 .
Mortality and morbidity rates of the SGA and AGA groups.
-5.21].Boghossian et al. reported the risks of BPD were 1.84-to 3.56-fold among SGA infants with a GA of 22-29 weeks [14].Nobile et al. discovered that being SGA is significantly associated with BPD [OR 2.69] [25].Tsai et al. found the SGA group is associated with

Table 3 .
Multivariate-adjusted odds ratio (95% CI) of outcome variables for the SGA group and subgroups stratified with gestational age (the AGA groups and subgroups as reference).risks of BPD [OR 2.08] compared to the AGA group [17].Qiu et al. observed that SGA infants have a higher odds of BPD [OR 1.78] [26].Regav et al. reported that SGA infants have a 3.42-fold risk of BPD [21].Sharma et al. observed that SGA infants had a significantly higher risk for developing BPD as compared to AGA infants [OR 2.2] increased