Original research•

Comparative analysis of INTERGROWTH-21st and Fenton growth charts for birthweight classification in a multiethnic Asian cohort: a cross-sectional study

Abstract

Objective The objective of this study is to evaluate the impact of using International Fetal and Newborn Growth Consortium for the 21st Century (INTERGROWTH-21st) (IG-21) growth standards compared with Fenton growth charts on birthweight classification in a multiethnic newborn cohort in Singapore.

Design Cross-sectional study.

Setting KK Women’s and Children’s Hospital, Singapore.

Patients Study population included 2541 babies born between 16 December 2019 and 16 March 2020.

Interventions None.

Main outcome measures Birthweight classifications of small for gestational age (SGA), appropriate for gestational age (AGA) and large for gestational age (LGA) were assessed using IG-21 and Fenton growth charts. The level of agreement between the two charts was measured using Cohen’s kappa coefficient (K).

Results Of 2541 neonates, 171 (6.7%) had discordant birthweight classifications. The kappa coefficient indicated moderate overall agreement (K=0.79) between the charts, with decreasing agreement from preterm (K=0.88) to full-term categories (K=0.71). The largest discordance was observed in 98 (60.5%) neonates classified as LGA by IG-21 but AGA by Fenton. In comparison, 60 (2.9%) neonates classified as AGA by IG-21 were SGA by Fenton, while 13 (4.6%) were SGA by IG-21 but AGA by Fenton.

Conclusions The study found discrepancies in birthweight classification between IG-21 and Fenton growth charts, with Fenton charts overclassifying SGA and underclassifying LGA in our study population. These findings suggest the potential need to integrate IG-21 growth standards into local practice to improve accuracy in neonatal growth assessment. Further research is necessary to evaluate the clinical implications of these discordant classifications on neonatal outcomes.

What is already known on this topic

  • Birthweight classification is critical for identifying newborns who are at risk of both perinatal and long-term metabolic health issues. The Fenton 2013 growth references are commonly used as a method of making this classification, but there are limitations in its generalisability to Asian contexts. The International Fetal and Newborn Growth Consortium for the 21st Century (INTERGROWTH-21st) (IG-21) standards have been proposed to provide a more globally applicable growth standard.

What this study adds

  • This study reveals significant discrepancies in birthweight classification between IG-21 and Fenton 2013 growth charts within a multiethnic Asian cohort. There was overclassification of small for gestational age (SGA) newborns and underclassification of large for gestational age (LGA) newborns when using the Fenton reference as opposed to IG-21 growth standards.

How this study might affect research, practice or policy

  • Our study’s findings suggest that the Fenton 2013 growth charts, in comparison to the IG-21 standards, may lead to overclassification of SGA and underclassification of LGA infants. This misclassification highlights the need to consider integrating IG-21 standards into routine clinical practice to more accurately identify infants at risk of perinatal complications, thereby enhancing early intervention strategies and improving outcomes. Additionally, these findings could drive further research into the clinical implications and predictive value of different growth standards, potentially influencing neonatal care policies globally.

Introduction

Classification of a baby’s size at birth needs to be done accurately, with critical implications for the child’s health. Birth size reflects in utero growth, which can be affected by uteroplacental insufficiency, maternal medical conditions and social factors such as socioeconomic status.1 Being born small for gestational age (SGA) places a newborn at risk of perinatal mortality and morbidity, including respiratory distress syndrome, hypothermia and neonatal hypoglycaemia.2 On the other hand, being born large for gestational age (LGA) places a newborn at risk of birth injuries such as shoulder dystocia and hypoglycaemia. Apart from immediate postnatal implications, birth weight impacts long-term metabolic health. Being born SGA and LGA are both associated with an increased risk of developing obesity and metabolic syndrome in childhood,3 4 correlating with higher cardiometabolic risk as an adult.5 Birthweight classification is currently based on newborn growth charts and reference curves, according to the baby’s sex and gestational age.

Different reference charts have been used to classify birth weight according to a baby’s gestational age. Available charts include those developed by Babson and Benda,6 Lubchenco et al,7 Dancis et al8 and Usher and McLean.9 The Fenton 2003 Preterm Growth Chart was developed by updating Babson and Benda’s 1976 chart using population-based data with large sample sizes from Canada,10 Sweden11 and Australia,12 and combining with Centers for Disease Control and Prevention term infant data to 10 weeks’ post-term.13 A decade later, these charts were replaced by the 2013 Fenton growth charts as we know today, harmonising the 2003 version with WHO Growth Standards. In our institution in Singapore, the 2013 Fenton growth charts are currently used for classification of birth weight. However, as its data were derived from predominantly European populations, its suitability to be used in an Asian country like Singapore is questionable, given distinct genetic and phenotypic variations.14 Infants of European descent have been shown to be 225.5 g and 254.6 g heavier than infants of Chinese and South Asian descent, respectively.15 There have been efforts to develop newborn growth charts within Singapore. However, they are not population based and have only used data acquired from single institutions.16 17 Given methodological challenges and limitations in the development and use of newborn growth references, it may be prudent to adhere to a prescriptive international growth standard.18 To overcome the limitations of 2013 Fenton charts, the International Fetal and Newborn Growth Consortium for the 21st Century (INTERGROWTH-21st) has developed growth standards for fetuses, newborn infants and postnatal growth of preterm infants, collecting birth anthropometric measurements from a healthy and well-nourished sample population.19 This was designed to augment the WHO Child Growth Standards and facilitate anthropometric comparisons across diverse ethnic groups.19 In a multiethnic population such as Singapore, mainly comprising Chinese, Malays and Indians, there has been no study to our knowledge which examines the implications of adopting INTERGROWTH-21st (IG-21) standards on birthweight classification locally.

The primary objective of this study is to assess the agreement between the IG-21 and Fenton growth charts in classifying birth weight into SGA, appropriate for gestational age (AGA) and LGA within a multiethnic cohort of neonates in Singapore. The secondary objectives are to evaluate whether classification discrepancies vary by gestational age and sex. Given the wide implications of birthweight classification and limitations of existing standards, the overarching goal is to determine if there is a need to consider integrating the use of IG-21 growth standards into local practice over the Fenton growth charts.

Methods

Study design

This was a cross-sectional study, conducted between 16 December 2019 and 16 March 2020 at KK Women’s and Children’s Hospital, Singapore.

Data collection

All neonates born within the study period were eligible for inclusion. Data regarding the neonate’s gestational age, birth weight and sex were collected. Maternal data collected included age, ethnicity, height and prepregnancy weight.

Statistical analysis

The overall distribution of baseline characteristics was reported using descriptive statistics. Continuous variables were presented as mean and standard deviation (SD), while categorical variables were reported using counts and percentages. All statistical analyses were performed by using R programming software.

Birthweight classification

Neonates with missing data or those born before 33 weeks’ gestation were excluded from the analysis, as the lower gestational age limit for reliable birthweight assessment using the IG-21 growth charts is 33 weeks. Birth weight was measured using SECA 334 Weighing Scale, recorded to the nearest gram. Birthweight classification was based on birthweight percentiles for gestational age and sex, according to both the IG-2119 and Fenton 201320 growth charts. SGA was defined as birth weight below the 10th percentile, AGA as birth weight between the 10th and 90th percentiles and LGA as birth weight above the 90th percentile.21 Gestational age was determined using first-trimester ultrasound dating scans or calculated from the date of the last menstrual period.

Assessing overall agreement in classification between Fenton and IG-21 growth charts

To evaluate the primary objective—assessing the level of agreement between the IG-21 and Fenton growth charts in classifying birth weight into SGA, AGA and LGA—confusion matrices were created. These matrices compared birthweight classification distributions between the two charts. Cohen’s kappa (κ) coefficient was calculated to measure the level of agreement, with kappa values of less than 0.40 deemed as unacceptable.22

Assessing agreement and discrepancies in classification by gestational age and sex

To assess whether discrepancies in birthweight classification between the IG-21 and Fenton growth charts vary by gestational age, the analysis was stratified by four gestational age categories: (1) preterm (33 to <37 weeks), (2) early term (37 to <39 weeks), (3) full term (39 to <41 weeks) and (4) late term (≥41 weeks).23 Cohen’s kappa coefficient was calculated for each gestational age subgroup to determine the level of agreement between the two charts in classifying birth weight into SGA, AGA and LGA. Additionally, to assess discrepancies in birthweight classification based on sex, subgroup analyses were conducted for males and females. The level of agreement between the two growth charts was compared between male and female neonates, using Cohen’s kappa to quantify the differences.

Sample size

From our literature review, studies examining discrepancies in SGA and LGA classification between IG-21 and Fenton growth charts have reported observed differences of approximately 10% for SGA and 2% for LGA.24 25 Therefore, the minimum sample size required, with a power of 80% and a 5% level of significance, was calculated to be 2282. During the study period, 2541 neonates met the inclusion criteria with complete data, providing sufficient power to detect the anticipated differences in classification for further analysis.

Patient and public involvement

It was not appropriate or possible to involve patients or the public in the design, or conduct, or reporting, or dissemination plans of our research.

Results

Figure 1 shows the participant flow. Out of 2552 live births during the study period, 8 newborns with incomplete data (missing gestational age, sex or birth weight) were excluded from the analysis. Additionally, 3 neonates born before 33 weeks’ gestation were excluded, leaving a final study population of 2541 neonates.

Figure 1
Request permissionrequest permission icon
Figure 1

Flow of participants.

Table 1 presents the baseline characteristics. The study population consisted of 52.0% male newborns. Regarding ethnicity, the largest proportion (38.0%) was Chinese. Most neonates were born at term, with 47.2% classified as early term, 46.0% as full term and 0.4% as late term. The mean birth weight was 3076±428 g. For maternal characteristics, the mean prepregnancy body mass index was 24.6±5.3 kg/m².

Table 1
|
Baseline characteristics of the study population

Overall agreement between IG-21 and Fenton growth charts

Figure 2 compares birth weight for gestational age using IG-21 (dotted lines) and Fenton (solid lines) centile curves. Notably, discordance between the two charts in classifying LGA and SGA was more evident in males starting from early term. For females, the discordance in classifying LGA was observed from early term, while SGA discordance was less prominent.

Figure 2
Request permissionrequest permission icon
Figure 2

Sex-specific distribution of birth weight against gestational age at birth. These graphs outline the distribution of birth weights against gestational age at birth for (A) males and (B) females. The dotted lines represent the trendlines for the predicted birth weight against gestational age at birth at the 10th and 90th percentiles with the IG-21 chart while the solid lines represent that of the Fenton chart. The red data point represent infants classified differently in both Fenton and IG-21 charts. IG-21, INTERGROWTH-21st.

Figure 3 illustrates the discordant classifications in greater detail. The proportion of discordant LGA classifications (LGA by IG-21 but AGA by Fenton) was highest among neonates born at early term (4.0%) and full term (3.9%), compared with preterm (1.9%) (Figure 3A). Discordance was more frequent in females compared with males for LGA.

Figure 3
Request permissionrequest permission icon
Figure 3

Changes in agreement between Fenton and IG-21 charts across gestational age at birth by infant sex. This figure shows the changes in (A) the percentages of LGA infants in IG-21 classified as AGA in Fenton, (B) the percentages of AGA infants in IG-21 classified as SGA in Fenton and (C) the changes in agreement between Fenton and IG-21 across gestational age group at birth and sex. The percentages presented in this represent the percentage of infants in IG-21 that were classified otherwise in Fenton for that particular gestational age group and sex. AGA, appropriate-for-gestational age; IG-21, INTERGROWTH-21st; LGA, large-for-gestational age; SGA, small-for-gestational age.

Discrepancies in classification, by gestational age and sex

Discordance in classifying SGA (SGA by Fenton but AGA by IG-21) was only observed in early-term and full-term neonates (Figure 3B), with the discrepancy being more pronounced in full-term newborns (4.0%), nearly four times greater than in early-term infants (1.1%). No SGA discordance was observed in late-term neonates. When analysed by sex, males showed more discordance in SGA classification than females.

Table 2 provides a breakdown of classification discrepancies by sex and gestational age, alongside the agreement between IG-21 and Fenton charts, as measured by Cohen’s kappa (K). A total of 171 neonates (6.7%) had discordant classifications, with a slightly higher percentage of discordance observed in females (7.3%) compared with males (6.2%). The largest discordance was found in infants classified as LGA by IG-21 but AGA by Fenton, occurring in 49.3% of males and 69.7% of females.

Table 2
|
Comparison of birth weight for gestational age classifications between Fenton and IG-21 charts

Among neonates classified as LGA by IG-21, 60.5% were classified as AGA by Fenton. This discordance increased with gestational age: 23.1% in preterm neonates, 53.9% in early-term neonates, 78.0% in full-term neonates and 100% in late-term neonates. For neonates classified as AGA by IG-21 but SGA by Fenton, the overall discordance rate was 2.9%, with males (4.0%) more affected than females (1.5%). This discordance increased with gestational age, from 0.0% in preterm neonates to 1.3% in early-term and 4.7% in full-term neonates. No SGA discordance was observed in late-term neonates.

Agreement in classification, by gestational age and sex

The overall kappa coefficient (K) for agreement between the two charts was 0.79 (95% CI 0.76 to 0.82), indicating moderate agreement. The level of agreement decreased as gestational age increased, from strong agreement in preterm neonates (K=0.88, 95% CI 0.81 to 0.96) and early-term neonates (K=0.83, 95% CI 0.79 to 0.87) to moderate agreement in full-term neonates (K=0.71, 95% CI 0.66 to 0.76).22 Among late-term neonates, the kappa coefficient was 0.85 (95% CI 0.63 to 1.00), but this was based on a small sample size (n=10). These trends are further illustrated in figure 3C, which shows the declining agreement across gestational age categories (from preterm to full-term).

When analysed by sex, the same trend in agreement was observed. In males, the agreement remained strong in preterm (K=0.96, 95% CI 0.90 to 1.00) and early-term neonates (K=0.85, 95% CI 0.80 to 0.90) but decreased to moderate in full-term neonates (K=0.66, 95% CI 0.58 to 0.75). Agreement in late-term males was high (K=1.00, 95% CI 1.00 to 1.00), though based on a small sample size. In females, agreement was moderate across all gestational age categories, with the strongest agreement in early-term neonates (K=0.82, 95% CI 0.76 to 0.87) and the weakest in full-term neonates (K=0.75, 95% CI 0.68 to 0.81).

Discussion

Our study of 2541 newborns revealed important differences in birthweight classifications between the IG-21 and Fenton growth charts, exposing inconsistencies that warrant attention. Notably, discordant classifications of both LGA and SGA were more prevalent from early term onwards. Overall, 6.7% of newborns were classified differently by the two charts, with the largest disparity in full-term neonates, where many were classified as LGA by IG-21 but AGA by Fenton. The kappa coefficient showed a clear decline in agreement from preterm to full-term categories, signalling only moderate overall agreement between the methods. These discrepancies may have crucial implications for clinical decision-making and neonatal care.

The finding of a higher birthweight classification as SGA by Fenton charts in comparison with IG-21 is consistent with other cross-sectional studies. In a study that assessed 668 term neonates in India; 313 (46.86%) and 236 (35.33%) were classified as SGA by Fenton and IG-21 charts, respectively.24 Outside of Asia, Tenório et al observed similar differences in detection of SGA among 344 neonates in Brazil, with Fenton classifying more than three times as many SGA neonates as IG-21 (Fenton: 16.9%, IG-21: 4.9%).26 In our study, this discordance (SGA by Fenton but AGA in IG-21) did not significantly affect preterm babies. However, another study by Lebrão et al involving 173 preterm infants showed that the IG-21 charts classified more babies as SGA (39.2%) than the Fenton charts (35.3%).27 The fact that the Fenton charts classify more babies as SGA is not merely a statistical issue—there could be important clinical implications. For example, in a retrospective observational study involving 1371 preterm neonates, those classified as SGA by IG-21 demonstrated higher sensitivity to developing postnatal complications such as hypoglycaemia, bronchopulmonary dysplasia and death compared with those classified as SGA by Fenton or other population-based charts.28 Additionally, being classified as SGA by IG-21 was more strongly associated with poorer neurodevelopmental outcomes as compared with SGA by Fenton.29

The 2013 revision of the Fenton charts included a smoothing process to better align them with the WHO growth charts, improving the transition for growth monitoring in preterm infants.30 This adjustment raised curve trajectories to merge with the WHO charts, which may contribute not only to the overclassification of SGA but also to the underclassification of LGA, especially in term infants. Our study confirmed this, with 60.5% of neonates classified as LGA by IG-21 but AGA by Fenton, a discrepancy that was more pronounced in females (69.7%) than males (49.3%). Similar findings were reported by Singamala et al, where IG-21 classified nearly twice as many LGA neonates (n=22, 3.02%) compared with Fenton (n=10, 1.37%).24 In a larger cohort study by Prakash et al involving 2507 neonates, this discordance was even more pronounced, with IG-21 identifying 5.5% as LGA by IG-21, compared with only 1.4% by Fenton.31 Being born LGA is known to predict a higher risk of obesity and metabolic complications.32 In a retrospective study of 173 preterm infants by Lebrão et al, IG-21 charts were better at predicting overweight status at 1 year old than the Fenton charts, for those who were classified as SGA or LGA by IG-21.27

There are implications arising from discordant classifications. With potential overclassification of SGA, this could result in an increased number of newborns receiving monitoring for anticipated complications. These may include blood investigations and closer vital signs monitoring. In Singapore, babies born SGA undergo blood glucose monitoring for detection of hypoglycaemia.33 In a cost analysis study conducted in France that included all livebirths in public and private hospitals, babies born SGA incurred initial hospitalisation costs (€2948) that were more than double those of babies born AGA (€1328).34 These added costs could be related to longer hospitalisation and monitoring. Babies born LGA are also at risk of early postnatal complications, including hypoglycaemia, polycythaemia, hyperbilirubinaemia and respiratory distress.35 With potential underclassification of LGA, this could result in missed opportunities for monitoring of these complications. If Fenton charts were to be replaced by IG-21 for birthweight classification, further studies comparing the clinical performance and cost-effectiveness in detecting postnatal complications for both SGA and LGA would be useful.

The study has notable strengths, including a large sample size of 2541 neonates from a maternity hospital with more than 10,000 annual births. We included newborns from different gestational age categories with subgroup analysis. This provides statistical power and enhances the reliability of our findings. This was a multiethnic cohort, comprising three different Asian ethnicities—Chinese, Malay and Indian. There was a good representation of other ethnicities, including Caucasians and mixed races. This increases the generalisability of our findings.

We recognise our study’s limitations. This was a cross-sectional study design, limiting the ability to correlate IG-21 and Fenton birthweight classification status with important postnatal outcomes, such as respiratory distress and hypoglycaemia. There was no correlation with maternal obstetric history and clinical outcomes, such as pregnancy-induced hypertension, gestational diabetes mellitus or association with the need for caesarean section and neonatal resuscitation. The use of data sourced from a single tertiary neonatal unit may limit the generalisability of our findings to other settings, particularly those with different population demographics or healthcare practices. Data from other hospitals, both in public and private sectors, were not included. This could restrict the external validity of our results and their applicability to a broader population. Lastly, our study methodology excluded newborns less than 33 weeks’ gestation, and only 10 late-term newborns were included. Hence, our findings may not be applicable to premature babies born before gestational age of 33 weeks’ and babies born late term.

In conclusion, findings from our study suggest discrepancies in birthweight classifications between the IG-21 and Fenton growth charts with potential clinical implications, especially for identifying SGA and LGA infants. These discordances could influence the detection and management of at-risk neonates, particularly regarding their susceptibility to postnatal complications and long-term health outcomes. Future research studies should rigorously evaluate the clinical performance of IG-21 versus Fenton charts, focusing not only on their accuracy in predicting complications but also on their cost-effectiveness in guiding interventions for SGA and LGA infants.

Article metrics
Altmetricopen-url
Dimensionsopen-url