Growth standards from birth to age 5: a population-specific retrospective longitudinal “WeChildren-Ningbo” study of 500 000 children and comparative analysis with WHO/national references

Abstract

Background

To establish sex-specific growth references for children aged 0–60 months in Ningbo, China and to compare with the WHO and the Chinese national standards.

Methods

A retrospective longitudinal “WeChildren-Ningbo” cohort study was performed for all registered births between January 2013 and December 2019. Anthropometric measurements (length/height and weight) were obtained at 10 follow-up time points: at birth, 3 m, 6 m, 12 m, 18 m, 24 m, 30 m, 36 m, 48 m and 60 months. Sex-specific growth curves were presented by both percentiles (3rd, 10th, 25th, 50th, 75th, 90th and 97th percentiles) and modified Z-score with median and SD based on empirical data. The generalised additive models for location, scale and shape modelling were applied to construct the percentile curves.

Results

Sex-specific growth curves were presented by both percentiles (3rd, 10th, 25th, 50th, 75th, 90th and 97th percentiles) and modified Z-score with median and SD. A total of 477 411 subjects were enrolled and 20 standardised sex-specific growth curves were constructed for weight-for-age, height-for-age and body mass index-for-age. Boys exhibited averagely higher height and weight at each follow-up than girls, and the largest sex differences were observed at 3 months. Ningbo children’s growth curves aligned well with the Chinese national standard, but were higher than the WHO standard, especially in children younger than 36 months.

Conclusion

This study established population-based, sex-specific growth references for children aged 0–60 months in Ningbo, China. These regional references may complement existing national standards for local clinical assessment and contribute data for future multi-regional studies.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Current research results show that China mostly uses the WHO Child Growth Standards or growth references issued by the National Health Commission of China.

• Nevertheless, both standards have critical limitations: they lack comprehensive longitudinal tracking of a large-scale birth cohort (WHO’s longitudinal coverage is limited to 0–24 months), and their data collection periods (WHO: 1997–2003; China: 2015) are outdated for current growth patterns in Chinese children.

WHAT THIS STUDY ADDS

• This large-scale population-based longitudinal study, WeChildren-Ningbo cohort study (2013–2019; n=477 411) established sex-specific growth curves for children aged 0–60 months and for each subgroup with or without low birth weight (LBW), preterm birth (PTB) and vaginal and caesarean section delivery, respectively.

• These growth curves represent the development characteristics in the new generation of children which fills the gap in the current standards (WHO and Chinese national references).

• Our results show high concordance with China’s Growth Standards for Chinese Children Under 7 Years of Age (2023) (validated by percentiles/median±SD), but present significantly higher growth levels than WHO standards, in particular in children younger than 36 months.

• Comparing the growth trajectories between subgroups stratified by LBW, PTB and delivery modes, LBW is the most pronounced key determinant of growth and development.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• The updated growth references and growth curves for children aged 0–5 years will promote the accuracy of early growth assessments in clinical and public health practices.

• It will facilitate precise assessment in those with LBW, PTB and different delivery modes.

• Moreover, the significant impact of LBW on long-term growth development suggests the necessity of implementing specialised health management for this group, in particular the nutritional support during the neonatal period.

Introduction

Infants and children’s growth represents a fundamental indicator of overall health and development and serves as a crucial proxy for the well-being of populations and future capital potential of individuals.¹ The first 5 years constitute a critical window characterised by rapid physical growth and neurocognitive maturation that establish foundational trajectories for cognitive functioning in early life and even for lifelong health.^{2 3} A substantial body of evidence has indicated the preschool growth patterns were associated with educational achievement and health economic productivity later in life.⁴,⁹ Childhood stunting during such a critical period has been linked to increased risks of non-communicable diseases later in life, including cardiovascular diseases, metabolic disorders and compromised immune functions.^{46 10},¹³ In this early life stage, growth trajectories are sensitive to multiple factors regarding nutrition, socioeconomics, environmental exposures and healthcare accessibility which deserve substantial attention and targeted prevention.¹⁴,¹⁶

Thus, assessing early childhood growth is a necessity and its significance extends far beyond the physical anthropometric measures.⁴ Consequently, surveillance on childhood growing serves dual purposes both as a sensitive indicator for population health status and a predictive tool for individual health outcomes across the life course. This paradigm underscores the significance of comprehensive growth surveillance in practice.

Standardised growth curves primarily rely on anthropometric measurements—particularly height and weight—and are usually presented as standard percentiles or deviations within a reference population.¹⁷ The WHO standards for children’s growth have achieved international recognition, as they represent children’s growth patterns under optimal conditions, irrespective of ethnic background or geographic location.¹⁸,²⁰ While they provide a universal reference, many countries, including China, have developed their own national growth references to better account for population-specific characteristics.²¹ China’s latest national growth references were established based on data from representative samples across multiple survey centres.^{21 22} Nonetheless, regional variations in growth patterns exist within China and updated references are in need of precise assessment of childhood growth.

Ningbo, an economically developed coastal metropolis in eastern China, launched the WeChildren-Ningbo cohort study based on the Women and Children’s Healthcare Electronic Monitoring System in 2009 with the framework structure achieving full operational capacity in 2013.^{23 24} The system has collected comprehensive growth measurements in parents and children throughout Ningbo’s healthcare network, enabling accessibility to growth trajectory analyses. Despite this advanced infrastructure, Ningbo still lacks localised growth references that account for its demographic and socioeconomic characteristics—a critical gap that limits the accuracy in preschool children’s growth monitoring and effectiveness of more targeted interventions.²⁵

With the aim to address the gap in population-specific growth references for children aged 0–5 years old in Ningbo in the total population and in subgroups with different key birth characteristics, this study established a series of sex-specific growth curves. These curves were compared with the Chinese national standard and WHO international standard, and growth variations were examined across subgroups stratified by low birth weight (LBW), preterm birth (PTB) and delivery mode of natural vaginal delivery and caesarean section, respectively.

Methods

Design and setting

This retrospective longitudinal cohort study depicted the dynamic growth track of infant/child growth in Ningbo. Strengthening the reporting of observational studies in epidemiology guidelines was adopted to ensure the rationality of reporting observational data (online supplemental file 1).²⁶ Formal consent was not required for this retrospective de-identified dataset, as approved by the Committee. All procedures were strictly conducted to follow the tenets of the Declaration of Helsinki.²⁷

Study population

A cohort of parent-child pairs was built by linking the parental and offspring anonymised records via a unique serial number from the WeChildren-Ningbo cohort (figure 1). Inclusion criteria were: (1) Infants with at least three follow-up data on growth measurements in 0–60 months (born between January 2013 and December 2019), (2) Infant/mother contained basic demographic variables (eg, infant sex) and (3) Infant contained complete birth parameters (eg, birth height/length and weight, gestational age, birth mode).^{18 20 28} Gestational age was determined using a two-step validation process: the last menstrual period (LMP) date was initially recorded at the first prenatal registration visit and subsequently verified using at least two early pregnancy ultrasound examinations during the first trimester. When discrepancies between LMP-based and ultrasound-based estimates exceeded clinically accepted thresholds (ie, 15 days), the ultrasound-based gestational age was adopted. Birth weight was obtained from delivery records documented at the time of birth using standardised measurement protocols.

Figure 1. Flow chart of the research protocol. Italic fonts stand for the dynamic variables in the data set (eg, Height at 0 months). ANOVA, Analysis of Variance; BMI, Body Mass Index; M, median.

The following records were excluded: records containing biologically implausible values (eg, height of 0 cm or weight of 0 kg).^{18 20} To maintain the representativeness of this population-based cohort, pairs were not excluded on preterm individuals, unhealthy maternal medical conditions (eg, gestational diabetes, hypertensive disorders) or congenital anomalies. After applying the aforementioned criteria, a final sample consisted of 477 411 parent-infants/children pairs who were analysed with multiple follow-up surveillance data from birth to 60 months.

Variables/measurements

Children’s demographic and birth characteristics were described through the following variables: sex, birth year, 5 parental education level (categorised as illiterate and semi-literate or elementary school, junior high school, senior high school or technical secondary schools, college or Bachelor degree and Master degree or higher), maternal and paternal smoking and alcohol consumption habits, and delivery mode (vaginal delivery vs caesarean section).

Children’s growth parameters were obtained by standardised anthropometric measurements at 10 longitudinal follow-up timepoints: birth (0 months), 3 m, 6 m, 12 m, 18 m, 24 m, 30 m, 36 m, 48 m and 60 months.^{18 20 22} Trained professionals performed all measurements following established protocols, collecting primary growth parameters including: (1) Recumbent length (measured to 0.1 cm precision for infants 0–24 months), (2) Standing height (measured to 0.1 cm precision for children 30–60 months) and (3) Body weight (measured to 0.1 kg precision in light clothing).^{23 24} Recumbent length and weight for infants (0–24 months) were measured using integrated infant measuring boards (accurate to 0.1 cm for length and 0.01 kg for weight). Standing height for children aged 30–60 months was measured using wall-mounted stadiometers (accurate to 0.1 cm), and body weight was measured using calibrated electronic scales (accurate to 0.1 kg). All equipment was calibrated monthly using reference standard weights following manufacturer specifications. Infants were weighed without diapers, while older children wore light indoor clothing and were measured without shoes. From these measurements, we derived secondary growth indices by calculating body mass index (BMI) at each timepoint using the standard formula (weight (kg)/height (m²)).²⁹,³¹ The study implemented rigorous quality control, with all measurements conducted by certified staff using calibrated equipment according to WHO-standardised techniques.

Statistical analysis

We performed all data processing and statistical analyses with R (V.4.4.3) and its compiler RStudio (V.2023.09.1 Build 494). Key packages were used including tidyverse (V.2.0.0), psych (V.2.4.12), dplyr (V.1.1.4), MVN (V.5.9), ufs (V.0.5.12), gamlss (V.5.4-22) and ggplot2 (V.3.5.1) were loaded and used.³²,³⁶

The core output consisted of three components:¹⁸,²⁰ (1) Sex-specific growth curves in the total Ningbo children aged 0–60 m and in subgroups with or without LBW, PTB and in two delivery modes (natural vaginal vs caesarean section), (2) Comparisons of growth curves in Ningbo children, the Chinese national and the WHO standard and (3) Comparison of growth parameters between subgroups with and without LBW, PTB and between two delivery modes in Ningbo children.

First, sex-specific growth curves for body length/height-for-age, weight-for-age and BMI-for-age were generated presenting both percentiles, modified Z-scores and the generalised additive models for location, scale and shape (GAMLSS) framework. For descriptive percentiles, 7 percentiles at the third, 10th, 25th, 50th, 75th, 90th and 97th were depicted. For the modified Z-score, the median ±1 SD (SD), ±2 SD and ±3 SD were graphed. For the GAMLSS framework, the Box-Cox t distribution was selected as the fitting distribution, which models four parameters: μ (median), σ (coefficient of variation), ν (skewness, ie, Box-Cox transformation power) and τ (kurtosis). Model convergence was set at a criterion of 0.01 with a maximum of 100 iterations.³⁷ From the fitted models, seven percentiles (3rd, 10th, 25th, 50th, 75th, 90th and 97th), and L (nu), M (mu) and S (sigma) parameters were derived at each age point (online supplemental eTables 2–11). These methods were selected in accordance with the Chinese national and WHO recommendations. A series of sex-specific growth curves were constructed in the total subjects.

Second, the growth curves of Ningbo children were compared with the Chinese national and WHO growth references, in both percentiles and modified Z-score, respectively. Percentile-for-percentile differences (3rd, 10th, 25th, 50th, 75th, 90th and 97th) were computed by subtracting Ningbo growth values from the Chinese national and WHO reference standard values at each corresponding follow-up timepoint. The same method was applied in comparing by the modified Z-score data.

Finally, to evaluate the impact of key parameters at birth on children’s growth development after birth, the body weight, length/height and BMI at all follow-up points were compared between subgroups of female or male, LBW or normal birth weight (NBW), PTB or term birth (TB), and in two delivery modes. In detail, bootstrap resampling (5 replicates × 2000 iterations) was performed in each subgroup at a monthly interval of age. Relative differences (%) of anthropometric measurements (body length/height, weight, BMI) were calculated between each pair of subgroup as: [(comparison group − reference group)/reference group] × 100. The reference groups were girls, those born with LBW, PTB and vaginal delivery, respectively. The median and percentiles (P10–P90) for such differences were derived from the bootstrap distributions, and group differences were tested using independent-sample t-tests.

Results

Participants

Of all recruited registered subjects in 2013–2019, 47.7% were girls, 3.2% were with LBW, 4.8% were PTB and 42.8% were delivered by caesarean section (table 1). The mean (median) birth weight and body length in girls were 3.3 kg (3.3 kg) and 49.7 cm (50.0 cm), while in boys, they were 3.4 kg (3.4 kg) and 49.8 cm (50.0 cm), respectively (online supplemental eTables 2 and 3). The mean (SD)/median (IQR) gestational age was 38.9 weeks (SD 1.4)/39.0 weeks (IQR (38.0 to 40.0)), respectively.

Table 1. Summary of demographic characteristics in all subjects.

Variables	N (%)
Total subjects	477 411 (100.0)
Birth year
2013	55 163 (11.6)
2014	65 348 (13.7)
2015	62 850 (13.2)
2016	76 606 (16.1)
2017	77 423 (16.2)
2018	69 441 (14.6)
2019	70 580 (14.8)
Gender
Female	227 554 (47.7)
Male	249 836 (52.3)
Birth weight
Low birth weight (<2500 g)	15 295 (3.2)
Normal birth weight (≥2500 g)	462 116 (96.8)
Gestational age
Preterm birth (<37 weeks)	22 734 (4.8)
Term birth (≥37 weeks)	454 677 (95.2)
Delivery mode
Caesarean section	204 357 (42.8)
Vaginal delivery	273 054 (57.2)
Maternal education level
Illiterate and semi-literate or elementary school	16 204 (3.6)
Junior high school	130 575 (28.3)
Senior high school or technical secondary schools	76 400 (16.5)
College or Bachelor’s degree	227 354 (49.3)
Master’s degree or higher	10 586 (2.3)
Paternal education level
Illiterate and semi-literate or elementary school	9623 (2.1)
Junior high school	131 804 (28.7)
Senior high school or technical secondary schools	96 609 (21.0)
College or Bachelor’s degree	208 574 (45.4)
Master’s degree or higher	13 212 (2.9)
Maternal cigarette smoking
No	476 510 (99.8)
Yes	879 (0.2)
Maternal alcohol drinking
No	476 365 (99.8)
Yes	1036 (0.2)
Paternal cigarette smoking
No	310 991 (65.2)
Yes	166 278 (34.8)
Paternal alcohol drinking
No	332 597 (69.7)
Yes	144 707 (30.3)

The mean and median ages of mothers at pregnancy were 28.7 years (SD 4.7) and 28.2 years (IQR (25.6 to 31.6)), respectively. Very few mothers had cigarette smoking (0.2%) or alcohol drinking (0.2%), while over 1/3 of fathers had positive reports on cigarette smoking (34.8%) and alcohol drinking (30.3%). Detailed characteristics of parents and infants are summarised in table 1. The number of children with available measurements varied across follow-up time points and was listed in online supplemental eTable 1.

Sex-specific standardised growth curves

A total of 20 sex-specific growth curves were constructed: 6 curves for the total population (height-for-age, weight-for-age and BMI-for-age, each for males and females), and 14 additional curves for clinically relevant subgroups including children with LBW, PTB and different delivery modes. Sex-specific curves were constructed because of well-established sexual dimorphism in childhood growth patterns.

From birth to age 60 months, the median body weight increased from 3.3 kg to 18.0 kg in girls, and the median body length/height increased from 50.0 cm to 110.0 cm, while in boys, it increased from 3.4 kg to 18.6 kg and from 50.0 cm to 111.0 cm, correspondingly (online supplemental eTables 2 and 3).

At all follow-up points, boys consistently exhibited higher median length/height and weight than girls (figure 2). As illustrated in the comparisons on length/height and weight between sexes, boys had the median body length/height of 50.0 cm at birth, 62.5 cm at 3 months, 68.8 cm at 6 months, 88.5 cm at 24 months and 111.0 cm at 60 months, respectively (figure 2A). For girls, the median length/height followed a similar trajectory but was slightly lower at each time point: 50.0 cm at birth, 61.0 cm at 3 months, 67.0 cm at 6 months, 87.0 cm at 24 months and 110.0 cm at 60 months On body weight, the median weights in boys were 3.4 kg, 6.9 kg, 8.5 kg, 12.5 kg and 18.6 kg, respectively (figure 2B), while the median weights in girls were 3.3 kg, 6.3 kg, 7.9 kg, 12.0 kg and 18.0 kg correspondingly. The relative differences (%) of body length/height and body weight between boys and girls increased from birth to age 3 months where a peak difference was observed, reaching 2.46% and 8.77%, respectively, after which the differences gradually narrowed (figure 2C,D). Detailed reference data at 11 percentiles (3rd, 5th, 10th, 15th, 25th, 50th, 75th, 85th, 90th, 95th and 97th) and modified Z-scores (median±1SD, 2SD and 3SD) are provided in online supplemental eTables 2–11 for comprehensive clinical reference. For graphical presentation, 7 percentiles (3rd, 10th, 25th, 50th, 75th, 90th and 97th) were displayed in the growth curves (figures3 4) to maintain visual clarity, consistent with WHO growth chart conventions.

Figure 2. Sex differences in children’s growth indicators in Ningbo, China, including (A) Body length/height, (B) Weight, (C) Relative difference (%) in body length/height and (D) Relative difference (%) in weight. Note: the different colours represent different groups.

Figure 3. Sex-specific growth curves in percentiles for (A) Body length/height-male, (B) Weight-male, (C) BMI-male, (D) Body length/height-female, (E) Weight-female and (F) BMI-female based on empirical data in Ningbo, China. Note: the different colours represent different percentiles of growth indices. BMI, body mass index.

Figure 4. Sex-specific growth curves in percentiles for (A) Body length/height-male, (B) Weight-male, (C) BMI-male, (D) Body length/height-female, (E) Weight-female and (F) BMI-female using the generalised additive models for location, scale and shape model in Ningbo, China. Note: the different colours represent different percentiles of growth indices. BMI, body mass index.

Sex-specific growth curves for Ningbo children aged from birth to age 60 months were then built for body length/height-for-age, weight-for-age and BMI-for-age, by depicting percentiles (P3, P10, P25, P50, P75, P90 and P97) (figures3 4) and the modified Z-score (medians±1SD, 2SD, 3SD) (online supplemental eFigures 1 and 2). As shown in the growth curves (figures3 4), all three parameters, length/height, weight and BMI, showed the most rapid increase during the first 3 months postpartum, and then a gradual deceleration in growth rate. Despite the rate variations, body length/height and weight maintained a steady increase up to 60 months, while BMI exhibited an inverted V-shaped left-skewed curve which peaked at 6 months. On average, children reached their peak BMI at 6 months, and after that, distinct percentile-dependent trends appeared: the BMI levels at or below the 50th percentile continued to decline from 6 months till 60 months, while BMI levels at or above the 85th percentile began to slightly increase from 48 months to 60 months (figures3 4). Consistent with the sex-difference patterns observed in body length/height and weight, the relative sex differences (%) in BMI peaked at 3 months by 3.39% (online supplemental eFigure 3). After that, the sex differences of BMI gradually narrowed until the differences lowered to almost zero at 60 months of age.

Comparison with the Chinese national and WHO growth standards

Comparison with the Chinese national standard

Ningbo children demonstrated close alignment with the Chinese national growth standards. For body weight at birth, the median birth weight in Ningbo boys (3.4 kg) was slightly smaller than the national median level (3.5 kg). This difference varied higher or lower within 0.1 kg subsequently till 36 m. At 48 m, the median height in Ningbo boys showed a gap of 0.2 kg lower than the national standard, and such gap continued to enlarge till 0.5 kg at 60 m (18.6 kg vs 19.1 kg). Among girls in Ningbo, the median birth weight was 3.3 kg, identical to the national median level. This difference varied within 0–0.1 kg before 36 m. At 48 m, the median weight in Ningbo girls was 0.2 kg less than the national median level (16.0 kg vs 16.2 kg) and the gap increased to 0.4 kg at 60 months (18.0 kg vs 18.4 kg) (online supplemental eTable 2 and online supplemental eFigures 4 and 5).

For body length/height, the differences in median levels were within 1 cm at all percentiles and across all follow-up time points. The difference varied in a relatively smaller range before 36 m but enlarged at 48 m and 60 m. For boys, the median length at birth in Ningbo children (50.0 cm) was slightly lower than the national standard (51.2 cm). It varied higher or lower by 0.1–0.5 cm than the national standard until 36 m. After that, at 48 m, the median length of Ningbo boys was 0.9 cm lower than the national standard and a further 1 cm at 60 m, respectively. For girls, the median of length/height in Ningbo girls was 0–0.3 cm lower than the Chinese national standard before 36 m. At 48 months, the difference in the median height of Ningbo girls slightly enlarged to 0.7 cm lower than the national standard and 0.8 cm at 60 m (online supplemental eFigures 4 and 5).

Comparison with the WHO growth standard

Ningbo infants and children exhibited consistently higher weight-for-age and height-for-age percentiles than the WHO standard, particularly during the first 36 months of life (online supplemental eTables 2 and 3, and online supplemental eFigures 4 and 5). The magnitude of differences varied by age, sex and growth parameter.

For body weight, the differences between Ningbo and WHO medians were most pronounced during early infancy and gradually narrowed with age. At 3 months, the median weight for Ningbo boys (6.9 kg) was 2.4 kg (53.3%) higher than the WHO median (4.5 kg), and for girls, the Ningbo median (6.3 kg) was 2.1 kg (50.0%) higher than the WHO median (4.2 kg). These substantial differences persisted through 12 months (boys: 10.0 vs 7.9 kg, +2.1 kg (26.6%); girls: 9.5 vs 7.3 kg, +2.2 kg (30.1%)). From 24 months onward, the differences progressively decreased: at 36 months, Ningbo boys exceeded the WHO median by 0.4 kg (14.7 vs 14.3 kg, +2.8%), and by 60 months, the difference was 0.4 kg (18.6 vs 18.2 kg, +2.2%) for boys and −0.1 kg (18.0 vs 18.1 kg, −0.6%) for girls.

For body length/height, the largest differences were observed between 3 and 12 months of age. At 3 months, Ningbo boys (median 62.5 cm) were 7.8 cm (14.3%) longer than the WHO median (54.7 cm), and Ningbo girls (median 61.0 cm) were 7.4 cm (13.8%) longer than the WHO median (53.6 cm). These differences remained substantial at 12 months (boys: 76.2 vs 67.5 cm, +8.7 cm (12.9%); girls: 75.0 vs 65.6 cm, +9.4 cm (14.3%)). After 24 months, the differences narrowed progressively: at 36 months, the difference was 1.3 cm (97.0 vs 95.7 cm, +1.4%) for boys and 1.3 cm (96.0 vs 94.7 cm, +1.4%) for girls. By 60 months, Ningbo children closely approximated the WHO values (boys: 111.0 vs 109.5 cm, +1.4%; girls: 110.0 vs 109.0 cm, +0.9%).

Sex-specific growth curves in subgroups and subgroup disparities

The growth parameters were compared at all follow-up points between subgroups stratified by LBW or NBW, PTB or TB, and two delivery modes—caesarean section or vaginal delivery, respectively (online supplemental eFigures 6–8). The growth curves for each subgroup with LBW or NBW (online supplemental eTabels 6 and 7 and online supplemental eFigures 9–12), PTB or TB (online supplemental eTabels 8 and 9 and online supplemental eFigures 13–16), delivered by caesarean section or vaginal delivery (online supplemental file 2 eTables 10–11 and online supplemental eFigures 17–20) were constructed and presented by both descriptive percentiles (3rd, 10th, 25th, 50th, 75th, 90th and 97th) and modified Z-score (median±1SD, ±2 SD, ±3 SD), respectively.

In detail, compared with LBW children, those born with NBW had approximately 47.00% higher of the median birth weight (online supplemental eFigure 6). This relative percentage difference decreased quickly after birth and narrowed to 15.18%, 8.91%, 5.72% and 5.37% at 3 m, 6 m, 12 m and 18 m, respectively. Such differences remained stable (4%–5%) from 24m to 60m. A similar trend was observed for body length/height, although with a much smaller gap. The birth length of NBW children median had 6.40% higher of the median level in the LBW children. This relative percentage difference decreased to 5.09%, 3.27% and 2.14% at 3 m, 6 m and 12 m, respectively, and further narrowed to 1.38% at 60 months. For BMI, the largest disparity occurred at birth, with NBW children having 30% higher median levels than those born LBW, but such difference declined rapidly with a marginal difference of 1%–2% from 6 m to 60 months.

Compared with PTB children, TB children (≥37 weeks) had a 32.61% higher median birth weight and 3.48% higher median birth length (online supplemental eFigure 7). The relative differences (%) of body weight decreased sharply after birth to 7.84% at 3 months, 3.02% at 6 months and 0.72% at 12 months, respectively, after which the gap became unchanged at around 0.5%–1.0%. For body length/height, this gap reduced from 3.48% at birth to 3.3% at 3 m, 1.72% at 6 m, 0.83% at 12 m and 0.32% at 60 months, respectively. The differences in BMI were quite significant at birth by 23.53%, but they became negligible at 3 m of age and onward. Compared with TB children, PTB showed an obvious catch-up growth after birth.

Finally, children who were born by caesarean section delivery showed slightly higher median weight (0%–1.37%) and BMI (0.68%–1.0%) than those born via vaginal delivery at all ages, but no substantial relative differences were observed (online supplemental eFigure 8).

Discussion

Overall

This WeChildren–Ningbo study constructed the first sex-specific and subgroup-specific growth curves for children aged 0–60 months in Ningbo, China, based on a large-scale population cohort of 477 411 children over 4.7 million measurements. These references closely aligned with the Chinese national growth standard throughout 0–60 months, but were higher than the WHO standard in particular at early life stages younger than 36 months. Birth weight emerged as the strongest determinant in affecting postnatal growth. Preterm born children exhibited distinct catch-up growth trajectories. Although the BMI trajectories in this cohort demonstrated normal adiposity rebound timing, long-term follow-up studies are needed to determine whether the relatively higher early-life weight-for-age is associated with later obesity risk. Our study provided the most updated paediatric growth surveillance analysis in Ningbo children and underscored the need for localised growth assessments in China. The findings may not be directly generalisable to regions with differing socioeconomic conditions. However, the close alignment with Chinese national standards suggests reasonable representativeness for well-nourished Chinese children.

At birth, Ningbo children presented similar percentile distribution for height and weight as the Chinese national and WHO standards. However, from 3 months onward, especially between 6 m and 30 m, Ningbo children showed substantially higher values than those in the WHO growth curves, while they still aligned well with the Chinese national standard throughout till 60 months. This was reasonable and several factors could explain the above divergence between Ningbo reference and the WHO standard. First, the WHO growth curves were established based on an international subject sample, in order to reflect “optimal growth” under ideal health and nutritional conditions³⁸ in diverse countries where children come from diverse ethnic and cultural backgrounds. Ningbo children all belonged to the Chinese ethnic group and were dominated by the Han people. This predominantly single ethnic group was not directly comparable with the ethnic composition in the WHO growth curve. Second, Ningbo’s social and economic development was at a relatively middle-to-high level that household income, maternal education level and healthcare accessibility likely contributed to enhanced growth outcomes at the population level.³⁹,⁴¹ As a coastal and economically developed city, Ningbo had the Gross Domestic Product (GDP) per capita of 26 171 US$ in 2024, almost twice the average Chinese national level of GDP per capita of 13 303 US$. Since the WHO standard is still used as a clinical tool in Children’s healthcare and clinical visits, the comparison between Ningbo children and the WHO standard helps promote a more precise guidance in clinical assessment on Ningbo children’s growth.

Our findings on sex-specific growth patterns are consistent with prior studies, demonstrating that male children exhibited higher height and weight than females throughout early life. This was largely due to biological dimorphism.⁴² In males, the postnatal “mini-puberty” peaks at 2–3 months and is finished by around 6 months, whereas in females this hormonal surge can persist until 2–4 years of age. Consequently, the developmental divergence between sexes reached its zenith around 3 months, with subsequent gradual diminishment of this disparity.⁴³ This baseline dimorphism is primed in utero, as evidenced by significantly higher mean testosterone concentrations in the amniotic fluid of male foetuses between 8 and 24 weeks of gestation.^{44 45} Androgen-mediated pathways subsequently potentiate male skeletal muscle development, reinforcing the observed growth differential. It underscored the need for sex-specific growth standards and suggested that early growth trajectories might have implications for long-term health outcomes in males and females which warranted further investigation.⁴⁶

In addition to sex, birth weight was a critical factor influencing children’s growing and development after birth. Between those with and without LBW, the median relative difference of body weight and height still remained 4.65% and 1.38% at 60 months, respectively, much higher than the relative differences between PTB and TB children at the same age (1.18% for body weight and 0.46% for height, respectively), or between vaginal delivery and caesarean section (0.96% for body weight and 0.00% for height). Such pronounced long-term significant influence of LBW on children’s body development was also reported in previous studies. A retrospective longitudinal study analysed medical records from two primary healthcare centres and their affiliates involved 950 children (469 boys, 481 girls) in Vilnius, Lithuania. Effects on growth patterns were larger for birth weight than gestational age. For example, early preterm girls/boys were 0.3/0.8 kg lighter, 0.9/0.9 cm shorter and 0.8/0.8 kg/m² thinner, while low BW girls/boys were 0.5/1.0 kg lighter, 1.5/1.4 cm shorter and 0.8/0.9 kg/m² thinner.⁴⁷ A prospective longitudinal birth cohort in two semi-urban areas in the central region of Gabon, central Africa found that, after adjusting for PTB, infant sex, maternal age and literacy, LBW remained as an independent risk factor associated with higher odds of stunting and underweight with strong statistical evidence at month 1 and month 12 (adjusted OR (aOR) 10.3, 95% CI 5.9 to 17.9; aOR 33.1, 95% CI 12.2 to 83.2 and aOR 2.6, 95% CI 1.4 to 4.9, aOR 4.5, 95% CI 2.5 to 7.2).⁴⁸ This suggested that the small for gestational age component of LBW, which was in utero growth restriction, might be more important in explaining the observed postnatal growth retardation rather than the being born too soon component. The underlying pathological conditions potentially included uteroplacental dysfunction, hypertensive disorders or illicit and toxic substances during pregnancy. We did not have the detailed exposure information in pregnancy, but this finding highlighted the necessity of particular care on LBW that contributed to the development of infants/children.

Methodologically, our study differs from the WHO Multicentre Growth Reference Study in several aspects. The WHO standards were derived from healthy breastfed infants under optimal environmental conditions, with strict exclusion of maternal smoking, gestational complications and infant morbidities.¹⁸ In contrast, our study adopted an inclusive approach without excluding children based on maternal conditions or congenital anomalies. This decision was made to establish growth references that reflect the actual distribution of growth patterns in the general paediatric population, thereby enhancing clinical applicability. Consequently, our references may be considered “growth references” describing how children in Ningbo actually grow, rather than “growth standards” prescribing how children should grow under optimal conditions. Clinicians should consider this distinction when applying these references in practice.

Strengths and limitations

This study featured several notable strengths. First, it leveraged a large, population-based cohort with repeated longitudinal measurements, allowing for a robust characterisation of growth trajectories in the total subjects, as well as in subgroups with different sex, with or without LBW or PT. Second, the WeChildren-Ningbo system ensured standardised anthropometric assessments conducted by trained healthcare professionals which enhanced data accuracy and reliability.²⁴ Third, the comparison with both international (WHO) and Chinese national references could guide local clinical evaluation against a more appropriate reference curve. Fourth, the identification of susceptible subgroups born with LBW in affecting body weight and height after birth provided evidence for effective targeted prevention measures.

Nonetheless, the study has several limitations. First, the study was conducted in a single city, Ningbo, an economically developed city in China. This might limit the generalisability of findings to regions with differing socioeconomic contexts. However, the good alignment with the Chinese national growth curve showed its good representativeness in the same ethnic group. Second, infant feeding practices were not analysed due to the ongoing natural language processing in the WeChildren-Ningbo database. Third, as a retrospective study using routinely collected health surveillance data from multiple centres over an extended period (2013–2019), detailed information on specific equipment brands across all participating health centres was not available. However, all participating centres followed standardised national protocols for paediatric anthropometry, with mandatory staff training and monthly equipment calibration, which ensured measurement quality and consistency. Lastly, unrecognised or undiagnosed health conditions in some children could have influenced their growth trajectories, potentially affecting the growth patterns in a small subset of participants.

Several potential sources of bias warrant consideration. First, although this was a population-based cohort capturing virtually all registered births rather than a selected sample, the declining follow-up rates at later ages could introduce bias if children who continued follow-up differed systematically from those who did not. However, the sex distribution remained stable across all time points, suggesting no major differential attrition. Second, while standardised measurement protocols were implemented across all participating centres, some inter-centre variability in measurement technique is inevitable in large-scale multi-site studies. The use of uniform national standards and mandatory staff training mitigated but could not entirely eliminate this source of variation. Third, although the 7-year study period (2013–2019) was characterised by stability in healthcare infrastructure and protocols, secular trends in childhood growth patterns cannot be entirely excluded. Future studies with time-stratified analyses may help address this question.

Height and weight trajectories during the first 5 years are crucial indicators of early childhood development, influencing long-term health, metabolic outcomes and cognitive function.^{1 49} Both excessive catch-up growth and persistent overweight could increase the risk of short stature, obesity and chronic diseases in adulthood.^{50 51} Future research should further track distinct patterns associated with sex, regional variation and early-life exposures and so on. Complementary analytical approaches, such as superimposition by translation and rotation model, could also be employed to further characterise individual-level growth trajectory variability and identify subgroups with distinct growth patterns. Locally specific growth curves, as developed in this study, could be validated for use in healthcare settings in Ningbo to complement international and national standards. Given that nutrition is a critical determinant of early growth, the higher weight-for-age observed in Ningbo infants compared with WHO standards could be partially attributable to regional feeding practices. Future prospective studies with detailed nutritional data are needed to elucidate these relationships. Moreover, comparative studies across regions in China will shed light on genetic, environmental and sociocultural factors, guiding tailored approaches to promote healthy growth in diverse populations.

Conclusion

These regional growth references can be applied for clinical assessment of children’s growth and development within Ningbo’s healthcare system. The findings may also inform local public health policy-making and contribute as regional data for future multi-centre national studies aimed at developing updated growth references for Chinese children.

Data availability statement

No data are available.