Population description
Following exclusions, we considered 67,968 deliveries for the analysis. Table S1 summarizes each country’s contribution to the overall population. The median maternal age was 26 (IQR: 22 – 31) years, with differences across countries. There were also differences in nulliparity rate, ethnicity, and educational level. The median gestational age at delivery in the study population was 39.5 (IQR: 38.5 – 39.5) weeks. The rate of preterm delivery was 7.9% (5359/67,968). The proportion of neonates classified as SGA was significantly different between the two standards. The WHO growth standard classified the neonates as follows: at or above the 10th percentile: 58,542 (86.1%) and SGA: 9426 (13.9%), while for IG-21st, 63,244 (93%) neonates were at or above the 10th percentile, and 4724 (7%) neonates were identified as SGA. Thus, the rate of neonates classified as SGA by the IG-21st was almost two times lower than that classified by the WHO (7 vs. 13.9%, p<0.001). Similarly, the proportion of neonates classified as FGR was significantly different between the two standards. The WHO growth standard classified the neonates as follows: at or above the 3rd percentile: 63,730 (93.8%) and FGR: 4238 (6.2%), while for IG-21st, 66,517 (97.9%) neonates were at or above the 3rd percentile, and 1451 (2.1%) neonates were classified as FGR. Thus, the rate of neonates classified as FGR by the IG-21st was almost three times fewer than that classified by the WHO (2.1% vs. 6.2%, p<0.001).
Figures 1A and 1B are Venn diagrams describing the classification of newborns according to the percentiles of each standard (at or above the 10th percentile vs. SGA and FGR) using both standards simultaneously. Specifically, 86.1% (58,523/67,968) were considered at or above the 10th percentile by both standards, 6.95% (4721/67,968) of neonates were classified as SGA only by the WHO standard (SGA-WHO only), 0.03% (19/67,968) of neonates were classified as SGA only by the IG-21 standard (IG-21st- only), and 6.92% (4705/67,968) were classified as SGA by both standards (Figure 1A). All neonates identified as SGA by IG-21st alone were preterm births. With respect to FGR, 93.7% (63,718/67,968) were considered above the 3rd percentile by both standards, 4.1% (2799/67,968) of neonates were classified as FGR only by the WHO standard (FGR-WHO only), 0.02% (12/67,968) of neonates were classified as FGR only by the IG-21 standard (IG-21 only), and 2.1% (1439/67,968) were classified as SGA by both standards (Figure 1B).
Table 1 describes clinical characteristics and perinatal outcomes for pregnancies assigned as SGA and FGR by WHO standard alone, by both standards, and those classified as at or above the 10th percentile for both curves. The rate of preterm delivery was higher in the newborns classified as FGR by WHO alone and by both standards than in those classified as above the 10th percentile by the two standards (all p values <0.001). However, the rate of preterm delivery was significantly higher in those classified as SGA by the two curves than in those classified as AGA (16.3% vs. 7.32%, p<0.05). In addition, there were significant differences in the cesarean section rate among the groups, being significantly higher in neonates classified as SGA and FGR by both standards compared to those classified as above the 10th percentile by the two standards (59.4% vs. 46.2%, p<0.001) (Table 1).
The rate of a low APGAR score was significantly higher for neonates classified as SGA and FGR by both standards (1.51% and 2.64%, respectively), followed by neonates classified as FGR only by the WHO (0.82%), being significantly lower in neonates classified as at or above the 10th percentile by both curves (0.46%). Notably, there were no significant differences in the rate of low APGAR scores between those neonates classified as SGA only by WHO and neonates classified as at or above the 10th percentile by both curves (0.40% vs. 0.46%, p=0.64, Table 1). Figures 2A and 2B show the RRs for a low APGAR score or ponderal index, respectively, in neonates identified as SGA and FGR. Neonates classified as SGA and FGR by both standards exhibited the most significant RR for an APGAR score below seven at five minutes (RR: 3.27; [95% CI: 2.52 – 4.24], and 5.57 [3.99 – 7.78], respectively). Importantly, neonates classified as SGA only by WHO alone did not have a significantly higher risk of a low APGAR score (RR: 0.87; 95% CI: 0.55 – 1.39) (Figure 2A).
The median ponderal index was significantly lower in the group of neonates classified as SGA and FGR by both standards than in those classified as above the 10th percentile by both standards (FGR by both standards: 22.5 [IQR: 20.6 – 24.7] & SGA by both standards: 23.5 [IQR: 21.7 – 25.4] vs. 26.7 [24.9 – 28.6]; all p values <0.001). Similarly, the rate of a ponderal index below the 5th percentile was significantly higher in these groups. Neonates classified as SGA and FGR by both standards exhibited the most significant RR for a low ponderal index (RR: 11.95; [95% CI: 10.7 – 13.4], and 14.9 [13.2 – 16.8], respectively) (Figure 2). Furthermore, neonates classified as SGA only by WHO alone also had a significantly higher risk of a low ponderal index (RR: 4.75; 95% CI: 4.1 – 5.53) (Figure 2B). Finally, the cephalization index was significantly higher in neonates classified as SGA by WHO alone and in those classified as SGA and FGR by both standards, displaying, in addition, a trend toward worst values in the latter groups (Table 2). Table S2 shows the odds ratios of SGA by each standard for neonatal outcomes under a hierarchical logistic regression model. In brief, we found that SGA babies only by WHO had an OR of 0.98 (95% CI: 0.61 – 1.57) for a low APGAR score at five minutes and 4.14 (95% CI: 3.52 – 4.86) and a ponderal index below the 5th percentile, respectively.
Table 2 displays the diagnostic performance of the WHO and IG-21st for identifying an APGAR score below seven at 5 minutes and a ponderal index below the 5thpercentile. Both charts exhibited low sensitivities for low Apgar scores (below 30%) and high specificity. We next assessed the diagnostic effectiveness of both fetal growth charts for specific obstetric outcomes, demonstrating that the IG-21st had the highest diagnostic odds ratios (DORs) (Table 2). As an overall measure of diagnostic performance for a low Apgar score, the diagnostic odds ratio was higher when SGA (3.70 vs. 2.02, mean difference: 0.61, 95% CI: (0.45 – 76.6), p <0.001) and FGR (6.22 vs. 3.01, mean difference 0.72, 95% CI: (0.48 – 0.96), p <0.001) were defined by IG-21st than by WHO charts. Similarly, the diagnostic odds ratio for a low ponderal index was also higher when SGA (10.4 vs. 9.01, mean difference 0.14, 95% CI: (0.06 – 0.23) p-value = 0.001) and FGR (14.6 vs. 10.6, mean difference 0.32, 95% CI: (0.2 – 0.42) p<0.001) were defined by IG-21st than by WHO charts. When we applied both fetal growth charts for the identification of a low APGAR score and ponderal index, the IG-21st fetal growth charts marginally improved the prediction of a low APGAR score based on the area under the receiver operating characteristic (ROC) curve (AUC), estimated using 2,000-fold bootstrapping to account for overfitting (Table S3). Specifically, for low APGAR scores, the AUC of the identification of SGA neonates for WHO fetal growth charts were 55.3 (95% CI: 53.1 – 57.5) vs. 57.3 (55.2 – 59.4) for IG-21st, p =0.005 (two-sided) (Table S3).