Abstract
Objective:
The aim of this study was to investigate the relationship of body mass index (BMI) with subsequent statural growth among children born in the era of the obesity epidemic.
Methods:
Among 18,271 children from Belarus (n=16,781, born 1996-1997) and USA (n=1,490, born 1999-2002), we used multivariable linear and ordinal logistic regression to analyze associations of BMI z-score from infancy to adolescence with subsequent standardized length/height velocity, standing height and its components (trunk and leg lengths), and pubertal timing.
Results:
The prevalence of early adolescent obesity was 6.2% in Belarus and 12.8% in USA. In both Belarussian and USA children, higher BMI z-scores in infancy and childhood were associated with faster length/height velocity in early life, while higher BMI z-scores during mid-childhood were associated with slower length/height velocity during adolescence. We observed associations with greater standing height, trunk length and earlier pubertal development in adolescence that were stronger for BMI z-scores at mid-childhood than BMI z-scores at birth or infancy.
Conclusions:
Our findings in both Belarus and USA support the role of higher BMI in accelerating linear growth in early life (taller stature and longer trunk length), but earlier pubertal development and slower linear growth during adolescence.
Keywords: body mass index, linear growth, pubertal development, birth cohort
Introduction
The current obesity epidemic is a global public health challenge. The increasing prevalence of childhood obesity is a major concern (1), given that excess weight tracks into adulthood (2) and is linked to chronic disorders such as metabolic syndrome, type 2 diabetes, coronary artery disease and stroke (3, 4). Few studies have considered the effects of early life adiposity on stature, which is also known to predict human health and capital (5).
Studies from the 1960’s-70’s reported that a high body mass index (BMI) in childhood is positively associated with linear growth and stature (6, 7). Recent longitudinal studies in the US (8) and Sweden (9) have observed inverse associations between high BMI-for-age in childhood and linear growth during adolescence. Furthermore, recent evidence suggests a relationship between childhood obesity and the timing of pubertal milestones in boys and girls (10, 11), which is known to influence adolescent linear growth. Previous studies, however, had short follow-up periods, lacked data on linear growth across distinct phases in the lifecourse (infancy, childhood and adolescence) and lacked measures of different components of stature (trunk and leg length). To our knowledge, few prospective cohort studies have addressed the association of BMI with linear growth across the lifecourse from infancy to adolescence, especially among children born in the era of the obesity epidemic (mid-1990s and early 2000’s (12)).
To address these gaps, we used prospective cohort data from two countries [USA (Project Viva (13)) and the Republic of Belarus (Promotion of Breastfeeding Intervention Trial; PROBIT (14))] with different childhood obesity rates (15) to examine the relationship of BMI measured from infancy to adolescence to subsequent statural growth velocity, standing height and its components and pubertal timing. We hypothesized that children with higher BMI in bothsettings would demonstrate faster linear growth during early life and greater standing height and its components, but earlier pubertal timing and slower linear growth during adolescence.
Methods
Study populations
Project Viva:
Project Viva is an ongoing prospective cohort study of pre- and perinatal influences on maternal, fetal and child health. As detailed elsewhere (13), we recruited eligible pregnant women during their first prenatal appointment between April 1999 and November 2002 from obstetric practices at Atrius Harvard Vanguard Medical Associates in eastern Massachusetts. Mothers provided written informed consent at enrollment and follow-up visits, and children provided verbal assent at the mid-childhood and early adolescent visit. The Institutional Review Board of Harvard Pilgrim Health Care approved the project in line with ethical standards established by the Declaration of Helsinki. Of 2,128 live singleton births, in this analysis we studied 1490 (70.0%) children who had at least two measures of weight and length/height between birth and early adolescence.
PROBIT:
PROBIT is a cluster-randomized trial of a breastfeeding promotion intervention in the Republic of Belarus, conducted between June 1996 and December 1997 (14). Briefly, 17,046 children were recruited from 31 hospitals that were cluster-randomized to either a control or breastfeeding promotion intervention based on the World Health Organization (WHO) and United Nations Children’s Fund Baby-Friendly Hospital Initiative. The trial and subsequent follow-ups were approved by the Belarusian Ministry of Health and received ethical approval from the McGill University Health Centre Research Ethics Board, the Institutional Review Board at Harvard Pilgrim Health Care, and the Avon Longitudinal Study of Parents and Children Law and Ethics Committee. A parent or legal guardian provided written informed consent in Russian at enrollment and at the follow-up visits, and all children provided written assent at the 11.5- and 16-year visits. Of 17,046 healthy, singleton, term live births, we studied 16,781 (98.5%) children who had at least two measures of weight and stature between birth and mid-adolescence.
Exposure: Body mass index z-score
In Project Viva, trained research assistants measured weight (to the nearest 0.1 kg) and length/height (to the nearest 0.1 cm) using standardized protocols (16, 17) at research visits in infancy (median 6.3; range 5.2–9.8 months), early childhood (3.2; 2.8–6.2 years), mid-childhood (7.7; 6.6–10.9 years) and early adolescence (12.9; 11.9–16.6 years). In PROBIT, polyclinic pediatricians measured weight (to the nearest 0.1 kg) and length/height (to the nearest 0.1 cm) at follow-up visits in infancy (at 1, 2, 3, 6, 9 and 12 months), mid-childhood (6.5; 5.5–8.5 years), early adolescence (11.5; 10.2–14.5 years), and mid-adolescence (16; 14.8–18.9 years), using uniform research specific equipment with standardized training and quality assurance procedures (18).
We used length (<2 years) or standing height (≥2 years) and weight measurements to calculate body mass index (BMI), derive age- and sex-specific BMI z-scores and define underweight, normal weight, overweight and obesity based on the WHO Child Growth Standards (19). We used BMI z-score at the median ages during research visits as our main exposures: Project Viva - birth, 6.3 months, 3.2 and 7.7 years; PROBIT - birth, 3 and 12 months, 6.5 and 11.5 years.
Outcomes
i. Standardized length/height velocity:
Our primary outcome was sex-specific standardized length/height velocity at median follow-up intervals during the research visits: Project Viva – birth to 6.3 months, 6.3 months to 3.2 years, 3.2–7.7 years and 7.7–12.9 years; PROBIT - birth to 3 months, 3–12 months, 12 months to 6.5 years, 6.5–11.5 years and 11.5–16 years. We first calculated length/height velocity as total length/height change divided by the duration of each time interval described above:
where i indexes the research visit, Hi is length/height at research visit i and di is the difference in exact age (years) between visit i and i–1.
In PROBIT, we chose the infancy timepoints of 3 and 12 months, which were derived from a cubic spline analysis demonstrating knots with linear changes in length from birth to 3 months and 3 to 12 months. (20). As the time intervals are of unequal duration, we standardized these velocities as z-scores with a mean of 0 and SD of 1:
where v1,i is the length/height velocity between visit i and i-1, is the mean length/height velocity during the same time interval and SD(v1,i) is the corresponding standard deviation. This approach allowed direct comparison of length/height velocities across different time intervals. We calculated these standardized values using internally derived cohort- and sex-specific means and SDs. While each child’s standardized velocity was derived from his or her individual follow-up time, inferences were based on the median follow-up interval.
ii. Trunk and leg length:
At the early adolescent visit in Project Viva and the 11.5-year visit in PROBIT, trained research assistants (in Project Viva) and pediatricians (in PROBIT) measured sitting height using calibrated stadiometers as previously described (21, 22). We used sitting height as a measure of trunk length and calculated leg length as the difference between standing and sitting height.
iii. Pubertal development:
In Project Viva, we evaluated pubertal development at the early adolescent visit via a validated written pubertal development scale (PDS) filled out by parents. The PDS has been strongly correlated with physician breast tanner staging in girls (23) and pubic hair staging in boys (24). PDS questions for boys include 4 items: voice deepening, facial and body hair growth, acne, and growth spurt. PDS questions for girls include 5 items: breast development, body hair growth, acne, growth spurt and menarche. The response options for each item (except for menarche) were: “not yet started” (1 point), “barely started” (2 points), “definitely started” (3 points), “seems complete” (4 points) and “I don’t know” (coded as missing). A “yes” answer on the menarche question receives 4 points, while a “no” answer receives 1 point. We derived a pubertal score for each participant by summing the point values and averaging across all items as detailed previously (25). In PROBIT, pediatricians received special training to assess genital development in boys, breast development in girls and pubic hair development in boys and girls at 11.5 years (range 10.2–14.5 years), using Tanner’s criteria for stage of maturation (stages 1 to 5) (26, 27). We also obtained self-reported pubic hair Tanner staging through pictograph questionnaires in early adolescence in Project Viva and at 16 years (range 14.8–18.9 years) in PROBIT.
Statistical analysis
We used multivariable linear regression to examine associations of BMI z-score (or overweight/obesity) at each exposure timepoint with the following outcomes: length/height velocities at subsequent time intervals; standing height, and trunk and leg length in adolescence. In both cohorts, we adjusted for maternal age, gestational age at delivery, parental height and BMI as continuous variables and parental educational attainment as a categorical variable (non-university- or university-educated). In Project Viva, we additionally adjusted for maternal smoking history (never, smoked prior to pregnancy or smoked during pregnancy), parity (nulliparous or multiparous), breastfeeding initiation (yes or no) and child race/ethnicity (white, black, Hispanic, Asian or others) as categorical variables. In PROBIT we additionally adjusted for maternal smoking during pregnancy (yes or no) and geographic region of residence (east rural, east urban, west rural or west urban) as categorical variables. We selected these covariates based on prior publications linking BMI and linear growth (6, 9, 28, 29).
In all models, we adjusted for BMI z-score in all previous age periods, but not in subsequent age periods. In models with length/height velocity as the outcome, we also adjusted for length/height velocities prior to BMI at each exposure timepoint (for example: associations between BMI z-score at 3 months and length/height velocity at 3–12 months were adjusted for length/height velocity at 0–3 months). Additionally, we used multivariable ordinal logistic regression, adjusting for the covariates listed, to examine associations of BMI z-score at each timepoint with pubertal development at early- and mid-adolescence.
For all PROBIT analyses, we accounted for clustered measurements within hospitals/polyclinics by including a random effect term for hospital/polyclinic. We did not adjust for intervention vs. control arms, as earlier analyses showed no differences in height at childhood and adolescence between the two study arms (30). We decided a priori to conduct all analyses separately in boys and girls, given the known sex differences in linear growth.
We used chained equation multiple imputation to impute values for children with missing covariate and outcome data, generating 50 imputed data sets including all participants with live births. The imputation model included all exposures, outcomes and covariates under study. In final analytic models after imputation, we combined imputed data sets using MI ESTIMATE in Stata after excluding participants without at least one outcome measure. Lastly, to assess robustness of our study findings, we repeated all analyses in subjects without missing covariate data. We analyzed all data using Stata version 15 (StataCorp, Texas, USA).
Results
Cohort descriptions
Compared to mothers from PROBIT, mothers from Project Viva were older (mean 32.7 vs. 24.5 years) and primarily university-educated (73.5% vs. 13.9%) (Table 1). The prevalence of early adolescent obesity was 6.2% (at 11.5 years) in PROBIT and 12.8% (at 12.9 years) in Project Viva. Compared to boys, girls had slower length/height velocities during infancy and adolescent periods, shorter heights and longer trunks in early adolescence, and earlier pubertal development in early adolescence (Tables 2–3).
Table 1:
Maternal and paternal characteristics of participating children in PROBIT and Project Viva
| PROBIT n=12,206a | Project Viva n=1,311a | |
|---|---|---|
| Maternal age | 24.5 (4.9)b | 32.7 (4.6) |
| Maternal smoking history | ||
| Never smoked | 12009 (98.4)b | 928 (70.8) |
| Smoked before pregnancy | –c | 271 (20.7) |
| Smoked during pregnancy | 197 (1.6) | 112 (8.5) |
| Parity | ||
| Nulliparous | – | 627 (47.8) |
| Multiparous | – | 684 (52.2) |
| Maternal education level | ||
| Not university educated | 10509 (86.1) | 347 (26.5) |
| University educated | 1697 (13.9) | 964 (73.5) |
| Paternal education level | ||
| Not university educated | 10583 (86.7) | 418 (31.9) |
| University educated | 1623 (13.3) | 893 (68.1) |
| Maternal height (cm) | 164.4 (5.6) | 165.1 (7.0) |
| Maternal BMI (kg/m2) | 24.5 (4.4) | 24.7 (5.2) |
| Paternal height (cm) | 176.1 (6.6) | 179.5 (7.5) |
| Paternal BMI (kg/m2) | 25.7 (3.3) | 26.5 (3.9) |
| Gestational age at delivery (weeks) | 39.4 (1.0) | 39.5 (1.8) |
| Child race/ethnicity | ||
| White | 12206 (100) | 916 (69.9) |
| Black | – | 159 (12.1) |
| Hispanic | – | 50 (3.8) |
| Asian | – | 50 (3.8) |
| Others | – | 136 (10.4) |
| Breastfeeding initiation | ||
| No | – | 123 (9.4) |
| Yes | – | 1183 (90.6) |
| Geographic region | ||
| East urban | 3715 (30.4) | – |
| East rural | 1971 (16.2) | – |
| West urban | 2960 (24.3) | – |
| West rural | 3560 (29.2) | – |
Sample sizes restricted to subjects with at least one outcome measure and no missing covariates.
Mean (SD) for continuous variables or n (%) for categorical variables.
Data not available.
Table 2:
Growth and pubertal characteristics of children in PROBIT
| Boys |
Girls |
|||
|---|---|---|---|---|
| na | Mean (SD) or % | na | Mean (SD) or % | |
| BMI z-score (SD units) | ||||
| 0m | 8682 | –0.54 (0.93) | 8099 | –0.64 (0.95) |
| 3m | 8602 | –0.21 (1.04) | 8022 | –0.13 (0.94) |
| 12m | 8509 | 1.14 (0.96) | 7932 | 1.14 (0.88) |
| 6.5y | 7172 | 0.12 (1.05) | 6702 | –0.05 (0.97) |
| 11.5y | 7105 | 0.20 (1.17) | 6738 | –0.05 (1.14) |
| 16y | 6918 | 0.12 (1.02) | 6534 | 0.00 (0.94) |
| Obesity at 11.5y | 565 | 8.0 | 290 | 4.3 |
| Obesity at 16y | 318 | 4.6 | 160 | 2.5 |
| Length/Height velocity (cm/y) | ||||
| 0–3m | 8606 | 37.1 (9.6) | 8024 | 34.9 (8.9) |
| 3–12m | 8475 | 19.9 (3.6) | 7886 | 20.0 (3.4) |
| 12m – 6.5y | 7155 | 7.9 (0.8) | 6688 | 8.0 (0.8) |
| 6.5–11.5y | 6634 | 5.6 (0.9) | 6311 | 6.1 (1.0) |
| 11.5—16y | 6509 | 5.9 (1.3) | 6188 | 3.2 (1.4) |
| Outcomes at 11.5 years | ||||
| Standing height (cm) | 7123 | 149.0 (7.6) | 6750 | 150.7 (7.9) |
| Trunk length (cm) | 7111 | 77.5 (4.2) | 6744 | 78.8 (4.6) |
| Leg length (cm) | 7111 | 71.5 (4.9) | 6744 | 71.8 (4.8) |
| Genital development at 11.5 years | ||||
| Stage 1 | 948 | 13.4 | –b | – |
| Stage 2 | 3818 | 54.0 | – | – |
| Stage 3 | 2047 | 28.9 | – | – |
| Stage 4 | 240 | 3.4 | – | – |
| Stage 5 | 21 | 0.3 | – | – |
| Breast development at 11.5 years | ||||
| Stage 1 | – | – | 767 | 11.5 |
| Stage 2 | – | – | 3441 | 51.4 |
| Stage 3 | – | – | 2149 | 32.1 |
| Stage 4 | – | – | 302 | 4.5 |
| Stage 5 | – | – | 35 | 0.5 |
| Pubic hair development at 11.5 years | ||||
| Stage 1 | 3553 | 50.2 | 1871 | 28.0 |
| Stage 2 | 2895 | 40.9 | 3142 | 46.9 |
| Stage 3 | 571 | 8.1 | 1491 | 22.3 |
| Stage 4 | 47 | 0.7 | 175 | 2.6 |
| Stage 5 | 8 | 0.1 | 15 | 0.2 |
| Outcomes at 16 years | ||||
| Height (cm) | 6948 | 175.7 (7.0) | 6489 | 164.9 (6.1) |
| Pubic hair development at 16 years | ||||
| Stage 1 | 16 | 0.2 | 101 | 1.6 |
| Stage 2 | 51 | 0.7 | 578 | 8.9 |
| Stage 3 | 466 | 6.8 | 1981 | 30.5 |
| Stage 4 | 2700 | 39.2 | 2064 | 31.8 |
| Stage 5 | 3651 | 53.0 | 1777 | 27.3 |
Sample sizes restricted to subjects with at least one outcome measure
Data not available.
Table 3:
Growth and pubertal characteristics of children in Project Viva.
| Boys |
Girls |
|||
|---|---|---|---|---|
| na | Mean (SD) or % | na | Mean (SD) or % | |
| BMI z-score (SD units) | ||||
| Birth | 418 | 0.54 (0.98) | 442 | 0.61 (0.93) |
| 6.3m | 559 | 0.70 (1.08) | 544 | 0.63 (0.97) |
| 3.2y | 603 | 0.80 (0.97) | 590 | 0.62 (0.94) |
| 7.7y | 542 | 0.61 (1.21) | 539 | 0.54 (1.13) |
| 12.9y | 521 | 0.60 (1.23) | 513 | 0.45 (1.25) |
| Obesity at 12.9y | 70 | 13.4 | 62 | 12.1 |
| Length/Height velocity (cm/y) | ||||
| Birth to 6.3m | 350 | 32.3 (3.9) | 364 | 30.7 (4.1) |
| 6.3m - 3.2y | 473 | 11.2 (1.1) | 465 | 11.4 (1.1) |
| 3.2 – 7.7y | 483 | 6.7 (0.7) | 492 | 6.7 (0.7) |
| 7.7– 12.9y | 450 | 6.0 (0.9) | 447 | 5.9 (0.9) |
| Outcomes at 12.9y | ||||
| Standing height (cm) | 523 | 160.8 (10.1) | 514 | 158.9 (7.4) |
| Trunk length (cm) | 522 | 82.1 (5.6) | 513 | 82.8 (4.2) |
| Leg length (cm) | 522 | 78.7 (5.9) | 513 | 76.1 (4.6) |
| Pubertal score | 584 | 2.2 (0.7) | 552 | 2.9 (0.7) |
| Pubic hair development at 12.9y | ||||
| Stage 1 | 21 | 3.9 | 12 | 2.4 |
| Stage 2 | 85 | 15.9 | 52 | 10.3 |
| Stage 3 | 149 | 27.8 | 118 | 23.3 |
| Stage 4 | 203 | 37.9 | 188 | 37.2 |
| Stage 5 | 78 | 14.6 | 136 | 26.9 |
Sample sizes restricted to subjects with at least one outcome measure
Associations of BMI z-score with subsequent length/height velocity
In PROBIT, higher BMI z-scores at birth, 3 and 12 months were more strongly associated with length/height velocity at the next interval (i.e., at 0–3, 3–12 months and 12 months-6.5 years respectively) than at later intervals. For example, in boys, a 1-SD increase in BMI z-score at 3 months was associated with increases in length/height velocity of 0.11 SD units (95% CI 0.09,0.13) at 3–12 months, 0.04 SD units (0.01,0.06) at 12 months-6.5 years and 0.03 SD units (0.00,0.05) at 6.5–11.5 years (Figure 1A). Similarly in Project Viva, a higher BMI z-score at 6.3 months was more strongly associated with length/height velocity at 6.3 months to 3.2 years than at 3.2–7.7 or 7.7–12.9 years (Figure 1B).
Figure 1:
Associations of child BMI z-score with standardized length/height velocity in (A) PROBIT and (B) Project Viva.
In PROBIT, higher BMI z-scores at 6.5 years was positively associated with height velocity at 6.5–11.5 years, but negatively associated with height velocity at 11.5–16 years (Figure 1A). A similar observation was noted in Project Viva: a higher BMI z-score at 3.2 years was positively associated with height velocity at 3.2–7.7 years (0.18 units; 0.08,0.28) but negatively associated with height velocity at 7.7–12.9 years (−0.14 SD units; −0.25,−0.03) in girls (Figure 1B).
Associations of BMI z-score with standing height, trunk and leg length
In PROBIT, associations with standing height at 11.5 years were stronger for BMI z-scores at 6.5 years than at birth or 3 or 12 months: for example, in girls, a 1-SD higher BMI z-score at 6.5 years was more strongly associated with standing height (1.64 cm; 1.46, 1.81) than was BMI z-score 12 months (0.27 cm; 0.07, 0.46) (Table 4). Similar results were observed in Project Viva: for example, in girls, a 1-SD higher BMI z-score at 7.7 years was more strongly associated with standing height at 12.9 years (1.14 cm; 0.43, 1.85) than was BMI z-score at 6.3 months (0.31 cm; −0.25, 0.87) (Table 5). Furthermore in PROBIT, the associations of BMI z-scores with standing height at 16 years were weaker than at 11.5 years (Table 4) and appear to be partly mediated by pubertal development at 11.5 years. For example, in girls, a 1-SD higher BMI z-score at 6.5 years was positively associated with standing height at 16 years (0.10 cm; - 05, 0.25), which was attenuated when adjusted for breast development (0.03 cm; −0.13, 0.18) or pubic hair stage (0.04 cm; −0.12, 0.19) at 11.5 years (Table S1).
Table 4:
Associations of BMI z-score with standing height, trunk and leg length, and pubertal development at 11.5 and 16 years in PROBIT
| β (95% CI)a |
OR (95% CI)a,b |
||||||
|---|---|---|---|---|---|---|---|
| 11.5y | 16y | 11.5y | 16y | ||||
| BMI z-score | Standing height (cm) | Trunk length (cm) | Leg length (cm) | Standing height (cm) | Genital development | Pubic hair stage | Pubic hair stage |
| Boys (n=8,682) | |||||||
| 0m | 0.39 (0.22,0.56)c | 0.23 (0.13,0.33) | 0.16 (0.05,0.28) | 0.56 (0.40,0.73) | 1.05 (1.00,1.12) | 1.03 (0.97,1.09) | 1.02 (0.96,1.08) |
| 3m | 0.12 (−0.03,0.26) | 0.11 (0.03,0.20) | 0.00 (−0.09,0.10) | ‒0.01 (−0.16,0.13) | 1.05 (1.00,1.10) | 1.09 (1.04,1.14) | 1.06 (1.00,1.11) |
| 12m | 0.30 (0.13,0.46) | 0.27 (0.17,0.37) | 0.03 (−0.08,0.14) | 0.14 (−0.02,0.30) | 1.05 (1.00,1.11) | 1.10 (1.04,1.16) | 1.03 (0.98,1.09) |
| 6.5y | 1.37 (1.22,1.52) | 1.03 (0.94,1.12) | 0.34 (0.24,0.45) | 0.27 (0.12,0.42) | 1.19 (1.13,1.26) | 1.26 (1.19,1.33) | 1.10 (1.05,1.16) |
| 11.5y | 1.63 (1.46,1.80) | 1.14 (1.05,1.24) | 0.49 (0.38,0.61) | 0.29 (0.12,0.46) | 1.29 (1.22,1.37) | 1.43 (1.35,1.52) | 1.21 (1.15,1.28) |
| 16y | — | — | — | 0.04 (−0.16,0.25) | — | — | 1.23 (1.15,1.32) |
| Girls (n=8,099) | Breast development | ||||||
| 0m | 0.36 (0.18,0.54) | 0.16 (0.05,0.27) | 0.20 (0.08,0.31) | 0.53 (0.38,0.68) | 1.00 (0.94,1.05) | 0.97 (0.92,1.03) | 0.98 (0.93,1.04) |
| 3m | 0.03 (−0.15,0.21) | 0.14 (0.03,0.25) | ‒0.11 (−0.22,0.01) | ‒0.07 (−0.21,0.08) | 1.08 (1.02,1.14) | 1.11 (1.05,1.17) | 1.04 (0.99,1.10) |
| 12m | 0.27 (0.07,0.46) | 0.24 (0.12,0.36) | 0.03 (−0.10,0.15) | 0.03 (−0.13,0.19) | 1.11 (1.05,1.18) | 1.15 (1.08,1.21) | 1.01 (0.96,1.07) |
| 6.5y | 1.64 (1.46,1.81) | 1.18 (1.08,1.29) | 0.45 (0.34,0.56) | 0.10 (−0.05,0.25) | 1.65 (1.56,1.75) | 1.54 (1.46,1.63) | 1.07 (1.02,1.13) |
| 11.5y | 1.63 (1.44,1.81) | 1.26 (1.15,1.37) | 0.37 (0.25,0.49) | ‒0.13 (−0.29,0.04) | 2.22 (2.09,2.37) | 1.89 (1.78,2.00) | 1.08 (1.02,1.14) |
| 16y | — | — | — | ‒0.50 (−0.69,−0.30) | — | — | 1.09 (1.03,1.16) |
Adjusted for maternal and paternal education level, height and BMI; maternal age and smoking during pregnancy; gestational age at delivery; geographic region; BMI z-score at all previous time points and age at outcome.
OR (odds ratio) reflects relative odds of being in a higher Tanner stage category.
Bold values indicate statistically significant associations at p<0.05.
Table 5:
Associations of BMI z-score with standing height, trunk and leg length, and pubertal development at 12.9 years in Project Viva
| BMI z-score | β (95% CI)a | OR (95% CI)a,b | |||
|---|---|---|---|---|---|
| Standing height (cm) | Trunk length (cm) | Leg length (cm) | Pubertal score (units) | Pubic hair stage | |
| Boys (n=758) | |||||
| Birth | 0.09 (−0.69,0.88) | 0.21 (−0.26,0.69) | −0.12 (−0.62,0.38) | −0.02 (−0.09,0.06) | 1.03 (0.83,1.27) |
| 6.3m | 0.32 (−0.27,0.92) | 0.34 (−0.02,0.70) | −0.01 (−0.39,0.36) | 0.03 (−0.02,0.09) | 1.11 (0.93,1.32) |
| 3.2y | 0.88 (0.10,1.67)c | 0.77 (0.31,1.22) | 0.12 (−0.39,0.63) | 0.01 (−0.05,0.08) | 0.93 (0.75,1.16) |
| 7.7y | 1.62 (0.84,2.40) | 0.71 (0.24,1.19) | 0.91 (0.39,1.42) | 0.11 (0.05,0.18) | 1.49 (1.19,1.87) |
| 12.9y | 0.91 (−0.01,1.82) | 1.17 (0.63,1.71) | −0.27 (−0.85,0.32) | 0.12 (0.04,0.20) | 1.37 (1.04,1.79) |
| Girls (n=732) | |||||
| Birth | 0.03 (−0.72,0.79) | −0.05 (−0.51,0.42) | 0.08 (−0.39,0.55) | −0.02 (−0.09,0.04) | 1.05 (0.84,1.31) |
| 6.3m | 0.31 (−0.25,0.87) | 0.41 (0.05,0.77) | −0.10 (−0.47,0.27) | 0.05 (−0.01,0.10) | 1.28 (1.05,1.56) |
| 3.2y | 0.59 (−0.14,1.31) | 0.72 (0.27,1.17) | −0.13 (−0.59,0.32) | 0.05 (−0.01,0.12) | 1.09 (0.86,1.37) |
| 7.7y | 1.14 (0.43,1.85) | 0.88 (0.44,1.32) | 0.26 (−0.20,0.72) | 0.17 (0.10,0.24) | 1.60 (1.26,2.03) |
| 12.9y | 1.43 (0.60,2.27) | 1.50 (0.98,2.03) | −0.07 (−0.60,0.46) | 0.30 (0.23,0.38) | 1.84 (1.38,2.45) |
Adjusted for maternal and paternal education level, height and BMI; maternal age, smoking history and parity; gestational age at delivery; breastfeeding initiation; child ethnicity; BMI z-score at all previous time points and age at outcome.
OR (odds ratio) reflects relative odds of being in a higher Tanner stage category.
Bold values indicate statistically significant associations at p<0.05.
In PROBIT, associations of BMI z-scores with trunk length at 11.5 years were stronger than associations with leg length: for example in girls, a 1-SD increase in BMI z-score at 6.5 years was more strongly associated with trunk length (1.18 cm; 1.08, 1.29) than leg length (0.45 cm; 0.34, 0.56) at 11.5 years. We noted similar observations in Project Viva: for example in girls, a 1-SD increase in BMI z-score at 7.7 years was more strongly associated with trunk length (0.88 cm; 0.44, 1.32) than leg length (0.26 cm; −0.20, 0.72) at 12.9 years.
Associations of BMI z-score with pubertal development
In PROBIT, children with higher BMI z-scores demonstrated more advanced genital (for boys), breast (for girls) and pubic hair development at 11.5 years. These associations were generally stronger for BMI z-scores at 6.5 years than at birth, 3 or 12 months (Table 4). Similarly in Project Viva, associations with pubertal score and pubic hair development at 12.9 years were stronger for BMI z-score at 7.7 years than at birth or 6.3 months (Table 5). Furthermore in PROBIT, the associations of BMI z-scores with pubic hair development at 16 years were weaker than with pubic hair development at 11.5 years.
Associations of overweight or obesity with length/height velocity, standing height trunk and leg length and pubertal development
Our findings for overweight or obesity showed similar patterns of associations to those of BMI z-score, i.e., children who were overweight or obese (vs. those who were of normal weight or underweight) had accelerated linear growth in early life, with taller stature and longer trunks, but earlier pubertal development and slower linear growth during adolescence (Tables S2–S5).
Complete case analyses
Our complete case analyses in both cohorts showed similar patterns of associations of BMI z-score with length/height velocity (Figure S1A–B), standing height, trunk and leg length and pubertal development (Tables S6–S7) as those observed after multiple imputation.
Discussion
We found that in both cohorts, children with higher BMI during infancy and childhood demonstrated faster subsequent growth in stature, both for standing height and trunk length. Additionally, children with higher BMI showed earlier pubertal development and slower linear growth during adolescence; these associations were more pronounced for BMI measures in later childhood than in infancy.
Few prospective cohort studies have addressed the relationship of BMI with statural growth across the lifecourse from infancy to adolescence. The associations we observed between higher BMI during early life and subsequent increase in length/height velocities corroborate the findings from population-based studies by He et al (28) and Luo et al (29). These studies, however, lacked data on linear growth during early life (<2 years of age) and were based on children born and raised in the 1970s, a time period before the obesity epidemic began and when environmental factors were relatively favorable for growth (28).
Our findings therefore better reflect contemporary children born and raised in the mid-1990s and early 2000’s, when the obesity epidemic was accelerating (12). Furthermore, we found consistent results in both cohorts, despite having different obesity rates (6.2% at 11.5 years in PROBIT, and 12.8% at 12.9 years in Project Viva) and exposures to obesogenic environments. They therefore add support to the notion that high BMI is associated with an accelerated rate of linear growth in early life (31). We speculate that factors known to influence linear growth and previously shown to be positively correlated with BMI, such as insulin-like growth factor I (IGF-I) (32) and fasting insulin (33), may play an important role in the relationship between higher BMI in early life and accelerated subsequent linear growth.
As a consequence of faster linear growth, pre-pubescent children with obesity typically present with taller stature than their normal-weight peers (34), consistent with our observations of larger standing heights during early adolescence among children with higher BMIs during mid-childhood. Furthermore, the greater heights in early adolescence were mainly attributable to longer trunks, rather than longer legs. To our knowledge, only one previous study has reported associations between greater fatness and subsequent longer trunks in prepubescent children (35). Previous findings in pre-pubertal children showed that IGF-I had stronger associations with later growth in trunk, but not leg length (36). Other studies also reported that pre-pubertal BMI is associated with insulin-induced reduction of sex hormone binding globulin and increased sex steroid bioavailability, which has a marked influence on the growth hormone-IGFI axis (37). We speculate that the divergence in components of later height among children with high BMI may be due to differences in exposure to hormonal determinants of femoral and vertebral growth i.e., sex steroids and IGF-I (36, 37). The divergence in height components may have unintended consequences. We have previously shown that trunk length, compared to leg length, had stronger associations with blood pressure and cardio-metabolic risk in mid-childhood independently of BMI (22). A low leg-to-trunk ratio has also been associated with an elevated risk of hypertension in children and adolescents (38).
Previous studies have shown that children with higher pre-pubertal BMI have a smaller pubertal height gain (9), consistent with our findings of slower adolescent growth in stature among pre-pubescent children with higher BMI. As noted in our observations and in other studies (39), higher pre-pubertal BMI is associated with earlier pubertal development, possibly through increased secretion of adipokines (e.g., leptin) that may trigger activity of the gonadotrophin-releasing hormone pulse generator (10). In addition, some genetic signals may influence both BMI and pubertal development (40). Earlier pubertal development typically leads to earlier cessation of linear growth and, therefore, smaller adolescent height gains. More recently, Lee et al reported correlations of delayed pubertal development with obesity in boys (41), although that study was cross-sectional and based on a convenience sample. Furthermore, for Project Viva, we speculate that the observed sex differences in early adolescent linear growth might be due to the fact that girls were further along in puberty than boys at early adolescence.
Although we lacked data on final adult height in both cohorts, we observed weaker associations (which were mostly null) between childhood BMI and height at 16 years in PROBIT children. This finding is consistent with recent evidence suggesting that children with high BMI eventually attain the same final height as their normal-weight peers (9). We acknowledge that childhood overweight and obesity have more serious health consequences (such as metabolic syndrome, type 2 diabetes, coronary heart disease and stroke later in life (2–4)) than deviations in linear growth and stature. Nevertheless, these deviations are a limiting factor to “optimal” body growth which may have consequences for future health outcomes (22).
Strengths of our study include its relatively large sample size of over eighteen thousand children from two prospective cohorts, each with multiple measures of growth from infancy to adolescence obtained by highly-trained research staff. In addition, our study benefits from the fact that we obtained similar results in two very different populations with differing obesity rates (15) and confounding structures (degree of income inequality, health care systems), suggesting that bias due to uncontrolled (residual) confounding is an unlikely explanation for the associations we observed.
Our study is not without limitations, however. First, BMI is an imperfect measure of adiposity; increased BMI is linked with maturational status, driven by gains in bone and muscle as well as fat. Thus, our results may not reflect the true relationship between adiposity and subsequent linear growth. Second, the use of BMI is currently not recommended for children below age 2 years; rather, weight-for-length (WFL) is recommended as a marker of weight status and adiposity for children in this age group (42). Past studies, however, have identified high concordance between WFL and BMI in infancy (43). Furthermore, we have previously shown that choice of WFL or BMI as indicators of overweight in children younger than 2 years does not substantially affect the ability to estimate future adiposity and other cardio-metabolic outcomes (15). These observations suggest that the use of BMI is likely to yield comparable results to those of WFL for children below age 2. Third, we lacked indicators of growth tempo (i.e., age at peak height velocity) and skeletal maturation (i.e., bone age). Fourth, we used self-reported measures of pubic hair staging and parent-reported measures of pubertal score in Project Viva, which may be subject to measurement error. Nevertheless, both pubertal indicators were reasonably correlated with each other (Spearman’s rank correlation = 0.66) and showed similar associations with BMI z-score, suggesting reliability of the reported pubertal measures. Fifth, we investigated multiple outcomes, thereby increasing the risk of false-positive results. We chose not to adjust for multiple comparisons. Instead, the “significance” of our findings is based on the strength and consistency of the associations observed across related outcomes and between the two cohorts (44). Lastly, our study findings may not be generalizable to other populations, since many of our participants were white (both cohorts) or university-educated (in Project Viva). Differences between children who continued vs those lost to follow-up may theoretically have led to selection bias, although our multiple imputation analyses showed very similar findings to those of our complete-case analyses.
Conclusion
In conclusion, our findings from two contemporary prospective birth cohorts with different obesity rates support the role of higher BMI in accelerating linear growth in early life, with taller stature and longer trunk length, but earlier pubertal development and slower linear growth during adolescence.
Supplementary Material
Study Importance.
What is already known about this subject?
Previous studies reported that high body mass index (BMI) correlates positively with linear growth and stature in childhood, but negatively with adolescent linear growth.
Few prospective cohort studies have addressed the association of BMI with linear growth across the lifecourse from infancy to adolescence, especially among children born in the era of the obesity epidemic (mid-1990s and early 2000’s).
What does your study add?
Our findings from two contemporary prospective birth cohorts with different obesity rates support the role of higher BMI in accelerating linear growth in early life, but earlier pubertal development and slower linear growth during adolescence.
The relationship of higher BMI in early life with subsequent taller stature was attributable to longer trunks rather than longer legs.
Acknowledgments
Funding
Project Viva is supported by the National Institutes of Health (R01 HD034568, UH3 OD023286). Emily Oken is additionally supported by P30 DK092924. PROBIT is supported by Canadian Institutes of Health Research (grant MOP-53155) and National Institutes of Health (R01 HD050758). Izzuddin M Aris is supported by the National University of Singapore Overseas Postdoctoral Fellowship (NUS OPF/2017). Seungmi Yang is a young career investigator (chercheur-boursier junior 1) of the Fonds de recherche du Quebec-Santé. Abby F Fleisch is supported by the National Institutes of Health (K23 ES024803). Richard M Martin works in the Integrative Epidemiology Unit supported by the United Kingdom Medical Research Council (MRC) and the University of Bristol (grant MC_UU_12013/1–9). Richard M Martin is also supported by the National Institute for Health Research Bristol Biomedical Research Centre, University Hospitals NHS Trust and University of Bristol, Bristol, UK. The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Footnotes
Disclosure
The authors declare no conflict of interest.
References
- 1.NCD Risk Factor Collaboration. Worldwide trends in body-mass index, underweight, overweight, and obesity from 1975 to 2016: a pooled analysis of 2416 population-based measurement studies in 128.9 million children, adolescents, and adults. Lancet. 2017. October 10;10(17):32129–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Singh AS, Mulder C, Twisk JW, van Mechelen W, Chinapaw MJ. Tracking of childhood overweight into adulthood: a systematic review of the literature. Obes Rev. 2008. September;9(5):474–88. [DOI] [PubMed] [Google Scholar]
- 3.Skinner AC, Perrin EM, Moss LA, Skelton JA. Cardiometabolic Risks and Severity of Obesity in Children and Young Adults. N Engl J Med. 2015. October;373(14):1307–17. [DOI] [PubMed] [Google Scholar]
- 4.Gjaerde LK, Gamborg M, Angquist L, Truelsen TC, Sorensen TIA, Baker JL. Association of Childhood Body Mass Index and Change in Body Mass Index With First Adult Ischemic Stroke. JAMA Neurol. 2017. November 1;74(11):1312–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Carba DB, Tan VL, Adair LS. Early childhood length-for-age is associated with the work status of Filipino young adults. Econ Hum Biol. 2009. March;7(1):7–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Forbes GB. Nutrition and growth. J Pediatr. 1977. July;91(1):40–2. [DOI] [PubMed] [Google Scholar]
- 7.Forbes GB. Lean Body Mass and Fat in Obese Children. Pediatrics. 1964. September;34:308–14. [PubMed] [Google Scholar]
- 8.Sunder M Shrinking due to corpulence? BMI in childhood predicts subsequent linear growth among US children and youth, 1963-1970. Ann Hum Biol. 2008. Jul-Aug;35(4):432–8. [DOI] [PubMed] [Google Scholar]
- 9.Holmgren A, Niklasson A, Nierop AF, et al. Pubertal height gain is inversely related to peak BMI in childhood. Pediatric research. 2017. March;81(3):448–54. [DOI] [PubMed] [Google Scholar]
- 10.Reinehr T, Roth CL. Is there a causal relationship between obesity and puberty? The Lancet Child & adolescent health. 2019. January;3(1):44–54. [DOI] [PubMed] [Google Scholar]
- 11.Frisch RE, Revelle R. Height and weight at menarche and a hypothesis of menarche. Archives of disease in childhood. 1971. October;46(249):695–701. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.James WP. WHO recognition of the global obesity epidemic. International journal of obesity. 2008. December;32 Suppl 7:S120–6. [DOI] [PubMed] [Google Scholar]
- 13.Oken E, Baccarelli AA, Gold DR, et al. Cohort profile: project viva. Int J Epidemiol. 2015. February;44(1):37–48. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Kramer MS, Chalmers B, Hodnett ED, et al. Promotion of Breastfeeding Intervention Trial (PROBIT): a randomized trial in the Republic of Belarus. JAMA. 2001. January 24–31;285(4):413–20. [DOI] [PubMed] [Google Scholar]
- 15.Aris IM, Rifas-Shiman SL, Li LJ, et al. Association of weight for length vs body mass index during the first 2 years of life with cardiometabolic risk in early adolescence. JAMA Network Open. 2018;1(5):e182460. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Perng W, Hajj H, Belfort MB, et al. Birth Size, Early Life Weight Gain, and Midchildhood Cardiometabolic Health. J Pediatr. 2016. June;173:122–130. e1.(doi). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Perng W, Rifas-Shiman SL, Kramer MS, et al. Early Weight Gain, Linear Growth, and Mid-Childhood Blood Pressure: A Prospective Study in Project Viva. Hypertension. 2016. February;67(2):301–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Guthrie LB, Oken E, Sterne JA, et al. Ongoing monitoring of data clustering in multicenter studies. BMC Med Res Methodol. 2012. March 13;12:29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.World Health Organization. WHO child growth standards: length/height-for-age, weight- for-age, weight-for-length, weight-for-height and body mass index-for-age Methods and development WHO (nonserial publication) Geneva: WHO; 2006. [Google Scholar]
- 20.Tilling K, Davies N, Windmeijer F, et al. Is infant weight associated with childhood blood pressure? Analysis of the Promotion of Breastfeeding Intervention Trial (PROBIT) cohort. Int J Epidemiol. 2011. October;40(5):1227–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Patel R, Lawlor DA, Kramer MS, et al. Socioeconomic inequalities in height, leg length and trunk length among children aged 6.5 years and their parents from the Republic of Belarus: evidence from the Promotion of Breastfeeding Intervention Trial (PROBIT). Ann Hum Biol. 2011. September;38(5):592–602. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Regnault N, Kleinman KP, Rifas-Shiman SL, Langenberg C, Lipshultz SE, Gillman MW. Components of height and blood pressure in childhood. Int J Epidemiol. 2014. February;43(1):149–59. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Brooks-Gunn J, Warren MP, Rosso J, Gargiulo J. Validity of self-report measures of girls’ pubertal status. Child development. 1987. June;58(3):829–41. [PubMed] [Google Scholar]
- 24.Schmitz KE, Hovell MF, Nichols JF, et al. A Validation Study of Early Adolescents’ Pubertal Self-Assessments. The Journal of Early Adolescence. 2004 2004/November/01;24(4):357–84. [Google Scholar]
- 25.Carskadon MA, Acebo C. A self-administered rating scale for pubertal development. The Journal of adolescent health : official publication of the Society for Adolescent Medicine. 1993. May;14(3):190–5. [DOI] [PubMed] [Google Scholar]
- 26.Marshall WA, Tanner JM. Variations in pattern of pubertal changes in girls. Arch Dis Child. 1969. June;44(235):291–303. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Marshall WA, Tanner JM. Variations in the pattern of pubertal changes in boys. Arch Dis Child. 1970. February;45(239):13–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.He Q, Karlberg J. Bmi in childhood and its association with height gain, timing of puberty, and final height. Pediatr Res. 2001. February;49(2):244–51. [DOI] [PubMed] [Google Scholar]
- 29.Luo ZC, Cheung YB, He Q, Albertsson-Wikland K, Karlberg J. Growth in early life and its relation to pubertal growth. Epidemiology. 2003. Jan;14(1):65–73. [DOI] [PubMed] [Google Scholar]
- 30.Martin RM, Kramer MS, Patel R, et al. Effects of Promoting Long-term, Exclusive Breastfeeding on Adolescent Adiposity, Blood Pressure, and Growth Trajectories: A Secondary Analysis of a Randomized Clinical Trial. JAMA Pediatr. 2017. July 3;171(7):e170698. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Freedman DS, Thornton JC, Mei Z, et al. Height and adiposity among children. Obes Res. 2004. May;12(5):846–53. [DOI] [PubMed] [Google Scholar]
- 32.Bideci A, Cinaz P, Hasanoglu A, Elbeg S. Serum levels of insulin-like growth factor-I and insulin-like growth factor binding protein-3 in obese children. J Pediatr Endocrinol Metab. 1997. May-Jun;10(3):295–9. [DOI] [PubMed] [Google Scholar]
- 33.Frystyk J, Vestbo E, Skjaerbaek C, Mogensen CE, Orskov H. Free insulin-like growth factors in human obesity. Metabolism. 1995. October;44(10 Suppl 4):37–44. [DOI] [PubMed] [Google Scholar]
- 34.Stovitz SD, Demerath EW, Hannan PJ, Lytle LA, Himes JH. Growing into obesity: patterns of height growth in those who become normal weight, overweight, or obese as young adults. Am J Hum Biol. 2011. Sep-Oct;23(5):635–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Sandhu J, Ben-Shlomo Y, Cole TJ, Holly J, Davey Smith G. The impact of childhood body mass index on timing of puberty, adult stature and obesity: a follow-up study based on adolescent anthropometry recorded at Christ’s Hospital (1936-1964). Int J Obes (Lond). 2006. January;30(1):14–22. [DOI] [PubMed] [Google Scholar]
- 36.Rogers I, Metcalfe C, Gunnell D, Emmett P, Dunger D, Holly J. Insulin-like growth factor-I and growth in height, leg length, and trunk length between ages 5 and 10 years. J Clin Endocrinol Metab. 2006. July;91(7):2514–9. [DOI] [PubMed] [Google Scholar]
- 37.Mauras N Growth hormone and sex steroids. Interactions in puberty. Endocrinology and metabolism clinics of North America. 2001. September;30(3):529–44. [DOI] [PubMed] [Google Scholar]
- 38.Dong B, Wang Z, Ma J. Leg-to-trunk ratio and the risk of hypertension in children and adolescents: a population-based study. J Public Health (Oxf). 2016. December 2;38(4):688–95. [DOI] [PubMed] [Google Scholar]
- 39.Marcovecchio ML, Chiarelli F. Obesity and growth during childhood and puberty. World Rev Nutr Diet. 2013;106:135–41. [DOI] [PubMed] [Google Scholar]
- 40.Bertini V, Fogli A, Bruno R, et al. Maternal Uniparental Disomy 14 (Temple Syndrome) as a Result of a Robertsonian Translocation. Molecular syndromology. 2017. May;8(3):131–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Lee JM, Wasserman R, Kaciroti N, et al. Timing of Puberty in Overweight Versus Obese Boys. Pediatrics. 2016. February;137(2):e20150164. [DOI] [PubMed] [Google Scholar]
- 42.Grummer-Strawn LM, Reinold C, Krebs NF. Use of World Health Organization and CDC growth charts for children aged 0–59 months in the United States. MMWR Recomm Rep. 2010. September 10;59(RR-9):1–15. [PubMed] [Google Scholar]
- 43.Furlong KR, Anderson LN, Kang H, et al. BMI-for-Age and Weight-for-Length in Children 0 to 2 Years. Pediatrics. 2016. July;138(1). [DOI] [PubMed] [Google Scholar]
- 44.Streiner DL. Best (but oft-forgotten) practices: the multiple problems of multiplicity-whether and how to correct for many statistical tests. Am J Clin Nutr. 2015. October;102(4):721–8. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.