Association of birth weight with general obesity, central obesity, and MASLD in a nationally representative sample of US adolescents: A cross-sectional study from NHANES 1999 to 2020

Abstract

Birth weight (BW) may influence subsequent risk of obesity and metabolic dysfunction-associated steatotic liver disease (MASLD); however, the conclusions are controversial and lack exploration in US adolescents. We aimed to explore the association of BW (including low BW [LBW], normal BW [NBW], and high BW [HBW]) with body mass index (BMI), waist circumference (WC), waist-to-height ratio (WHtR), and fatty liver index (FLI), as well as general obesity, central obesity, and MASLD, in adolescents using National Health and Nutrition Examination Survey 1999 to 2020. BW was obtained from participants’ self-reports. Obesity and MASLD were diagnosed based on their respective specific cutoff values in adolescents. Multivariate linear regression and logistic regression analyses were used to explore these associations and calculate β and odds ratios. A total of 6867 adolescent participants were enrolled. After adjusting for all confounders, BW was positively associated with BMI, WC, WHtR, and FLI (β of 0.639, 1.872, 0.005, and 2.128, respectively). Compared to NBW, HBW was associated with significantly increased BMI, WC, WHtR, and FLI (β of 1.205, 3.387, 0.012, and 4.745, respectively), whereas LBW was not. Similarly, compared to NBW, HBW was associated with significantly increased odds of general obesity, central obesity (as defined by WC/WHtR, respectively), and MASLD (odds ratios 2.629, 1.713, 1.618, and 1.960, respectively). However, LBW was not significantly associated with obesity and steatosis. Race/ethnicity partially influenced these associations. HBW, but not LBW, was associated with increased prevalence of general obesity, central obesity, and MASLD among U.S. adolescents. These findings underscore that adolescents with HBW are at risk for obesity and steatosis and may require early screening and intervention, especially among other Hispanic ethnic groups.

1. Introduction

Birth weight (BW) is one of the most important clinical indicators in neonatology, and it is an important determinant of the health status of newborns. Low BW (LBW) is associated with an increased risk of neonatal mortality; moreover, a large body of well-establishedepidemiologic research suggests that LBW is associated with poor health outcomes in later childhood, adolescence, and/or adulthood.^[1–3] In addition, accumulating clinical evidence also suggests that high birth weight (HBW) may predict certain health trajectories later in life, such as increased risk of bone tumors, obesity, and autonomic nervous system disorders.^[4–6] In addition, the association of HBW or LBW with subsequent health outcomes in life may shift with age,^[7–9] emphasizing that the impact of abnormal BW on health outcomes requires an age-specific perspective.

Obesity is one of the major health problems among adolescents. The prevalence of obesity among adolescents has now reached epidemic proportions, with the prevalence of severe obesity increasing at least fourfold over the past few decades.^[10] Adolescent obesity is associated with the development of multiple short- and long-term complications, such as psychosocial stress and obesity-related metabolic disorders and cardiovascular disease.^[11,12] In addition, 80% of obesity in adolescence can be maintained into adulthood, leading to increased metabolic and cardiovascular risk and higher mortality in adulthood.^[13,14] Early identification of adolescents at risk for obesity and timely intervention can help reduce subsequent adverse obesity-related health outcomes. A previously published systematic review and meta-analysis demonstrated that HBW was associated with a significantly increased risk of obesity in preschool children, school-aged children, and adolescents compared to control populations, whereas LBW was associated with a decreased risk of obesity.^[15] More recent observational clinical studies have suggested that HBW may be associated with an increased risk of general obesity and abdominal obesity in children and adolescents; however, the association of LBW with the prevalence of obesity in pediatric or adolescent populations is controversial.^[16–21] Several limitations were present in the previous evidence. First, there was significant heterogeneity in the definitions of LBW, NBW, and HBW among these studies, and the lack of standardized definitions may have hindered conclusions. Second, most of the observational studies were from Asian cohorts and there was a lack of exploration in other countries/regions such as the United States. In addition, most of the studies included children and adolescent populations, and studies focusing on adolescent samples were lacking.

Metabolic dysfunction-associated steatotic liver disease (MASLD), formerly known as nonalcoholic fatty liver disease (NAFLD), is another emerging health concern in the adolescent population.^[22] Under the previous NAFLD definition, the prevalence of this condition in children and adolescents has been estimated at 36.1%.^[23] Interestingly, some studies have shown that LBW and/or HBW compared to NBW are associated with increased prevalence of NAFLD or MASLD in adults, however controversial findings exist.^[24–27] Some sparse observational evidence suggested an association of LBW and/or HBW with MASLD in children and adolescents, with similarly inconsistent findings.^[28–30] Whether LBW/HBW is associated with the prevalence of MASLD in US adolescents remains largely unexplored.

This study utilized a nationally representative sample of adolescents from the National Health and Nutrition Examination Survey (NHANES) to explore the association of LBW and HBW with the prevalence of general obesity, central obesity, and MASLD. Exploring these associations could help reveal whether abnormal BW contributes to early detection, risk stratification, and timely intervention for obesity and MASLD among adolescents.

2. Methods

2.1. Study design and population

NHANES is a large, national, population-based, multiracial series of cross-sectional surveys. Since 1999, NHANES has been conducted in biennial cycles, sampling approximately 5000 cases per year to assess the health and nutritional status of noninstitutionalized children and adults in the U.S. NHANES is a major program of the National Center for Health Statistics (NCHS), and all cycles were approved by the NCHS Ethics Review Board and written informed consent was obtained from participants. A total of 9812 adolescent participants aged 12 to 15 years from the NHANES 1999 to 2020 March. Participants with missing BMI data (n = 461), WC data (n = 83), FLI data (n = 802), BW information (n = 470), and covariate information (n = 1129) were excluded. A total of 6867 eligible adolescent participants were included in further analyses (Fig. 1).

Figure 1.

Flowchart of study population selection, NHANES 1999 to 2020. NHANES = National Health and Nutrition Examination Survey.

2.2. BW assessment

BW data were obtained from the NHANES Early Childhood Questionnaire, which was available to children and adolescents ≤15 years of age in the US.^[31–33] Participants were asked, “How much did you weigh at birth?” and self-reported their weight at birth (pounds or ounces). This question was asked by trained interviewers to home-based participants through the computer-assisted personal interview system.^[32] Participants’ self-reported BW (pounds or ounces) to grams and categorized them according to World Health Organization criteria as LBW (BW <2500 g), normal BW (NBW, BW at 2500–4000 g), and HBW (BW >4000 g).^[31,32]

2.3. Assessment of outcomes

Study outcomes included body mass index (BMI), waist circumference (WC), waist-to-height ratio (WHtR), and fatty liver index (FLI). BMI was used to define overall obesity. WC and WHtR were used to define central obesity. FLI was used to define MASLD. The FLI was calculated using the formula FLI = (e^{0.953*ln (TG) + 0.139*BMI + 0.718*ln (GGT) + 0.053*WC −15.745})/ (1 + e^{0.953*ln (TG) + 0.139*BMI + 0.718*ln (GGT) + 0.053*WC}−^15.745) × 100.^[34] These markers were first explored as continuous variables. Second, they were used as categorical variables using age- and sex-specific cutoffs for diagnosis of the disease of interest. BMI-defined overall obesity was defined as an individual’s BMI ≥ the age- and sex-specific 95% percentile.^[35] WC > 75th percentile for age and sex suggested central obesity.^[36] WHtR ≥ 0.5 indicated central obesity.^[37] FLI ≥ 60 suggested the presence of MASLD.^[38]

2.4. Covariates

Several important covariates including age, gender, race/ethnicity (Mexican American, non-Hispanic White, non-Hispanic Black, other Hispanic, or other races), household income-poverty ratio (PIR), mother’s age when born, maternal smoking history during pregnancy, total daily dietary energy intake, serum triglycerides (TG), total cholesterol, and high-density lipoprotein cholesterol (HDL-C) were included. Mother’s age when born and maternal smoking history during pregnancy was obtained from self-reports on the Early Childhood Questionnaire.^[33] Daily energy intake was derived from dietary interview data, which was calculated from the USDA’s Food and Nutrient Database for Dietary Studies. Serum lipid/lipoprotein data were obtained from NHANES laboratory tests.

2.5. Statistical analysis

Data processing and statistical analyses were performed using R (version 4.2.3) and EmpowerStats, and a 2-sided P-value of <.05 reported statistical significance. All analyses were properly weighted according to the NHANES analytic guidelines to consider the sophisticated study design of NHANES. Baseline analyses were conducted based on the BW status of the included adolescent population. Data for continuous variables were reported as mean ± standard error and analyzed using weighted ANOVA, and categorical variables were reported as number (percentage) and analyzed using weighted chi-square tests. Multivariate linear regression analyses were performed to explore the association of LBW and HBW (compared with NBW) with BMI, WC, WHtR, and FLI among US adolescents aged 12 to 15 years and to calculate β and 95% confidence intervals (95% CIs). For BMI, WC, and WHtR, the crude model did not adjust for any covariates; model 1 partially adjusted for age, sex, race/ethnicity and PIR; and model 2 adjusted for age, sex, race/ethnicity, PIR, mother’s age when born, maternal smoking history during pregnancy, total energy intake, TG, total cholesterol, and HDL-C. For FLI, the crude model did not adjust for any covariates; model 1 partially adjusted for age, sex, race/ethnicity, and PIR; and model 2 adjusted for age, sex, race/ethnicity, PIR, mother’s age when born, maternal smoking history during pregnancy, and total energy intake. We used multivariate logistic regression analyses to explore the association of LBW/HBW with general obesity as defined by BMI, central obesity as defined by WC/WHtR, and MASLD as defined by FLI and to calculate odds ratio (OR) and 95% CI. These logistic regression models were consistent with the adjustment variables used in the respective linear regressions. Restricted cubic spline (RCS) was applied to explore nonlinear associations between BW (continuous) and BMI, WC, WHtR, and FLI in adolescents and to pick suitable knots for smoothed curve fitting. In addition to including sex as a covariate in all multivariable models, stratified analyses by sex were conducted to explore potential effect modification. Stratified analyses were conducted to explore whether these associations (between BW and BMI, WC, WHtR, and FLI) differed within subgroups (gender and race/ethnicity) and to identify effect modifiers through interaction analyses. We also explored whether the association of BW with obesity and MASLD remained stable across subgroups.

3. Results

3.1. Baseline characteristics

A total of 6867 adolescent participants with a mean age of 13.508 years were enrolled. There were 881, 5342, and 644 participants in LBW, NBW, and HBW, respectively. As BW increased, participants had higher PIR, energy intake, and mother’s age when born, lower HDL-C, and were more likely to be male, non-Hispanic White, and have no history of maternal smoking during pregnancy. Participants with HBW had higher BMI, WC, WHtR, and FLI (continuous and categorical variables) compared to participants with LBW (Table 1).

Table 1.

Baseline analysis of adolescent participants according to BW status, NHANES 1999 to 2020.

Variables	Total (n = 6867)	LBW(n = 881)	NBW(n = 5342)	HBW(n = 644)	P-value
Age, year	13.508 ± 0.017	13.522 ± 0.050	13.501 ± 0.019	13.550 ± 0.063	.702
PIR	2.591 ± 0.038	2.267 ± 0.070	2.588 ± 0.040	2.972 ± 0.098	<.0001
Energy intake, kcal/day	2142.944 ± 18.709	2033.530 ± 44.434	2146.053 ± 20.012	2240.743 ± 63.115	.014
mother’s age when born	26.671 ± 0.143	26.223 ± 0.308	26.529 ± 0.155	28.296 ± 0.341	<.0001
TG, mmol/L	1.040 ± 0.013	1.053 ± 0.037	1.040 ± 0.013	1.031 ± 0.027	.893
TC, mmol/L	4.072 ± 0.014	4.094 ± 0.042	4.077 ± 0.015	4.005 ± 0.044	.229
HDL, mmol/L	1.324 ± 0.005	1.337 ± 0.014	1.328 ± 0.006	1.283 ± 0.015	.013
BMI	22.577 ± 0.100	22.711 ± 0.285	22.447 ± 0.110	23.460 ± 0.292	.002
WC, cm	78.804 ± 0.266	78.645 ± 0.778	78.443 ± 0.290	81.846 ± 0.796	<.001
WhtR	0.484 ± 0.002	0.488 ± 0.004	0.482 ± 0.002	0.493 ± 0.004	.023
FLI	15.931 ± 0.418	17.033 ± 1.269	15.284 ± 0.442	19.823 ± 1.373	.003
High BMI
No	6527 (96.084)	844 (95.803)	5099 (96.585)	584 (92.431)	<.0001
Yes	340 (3.916)	37 (4.197)	243 (3.415)	60 (7.569)
High WC
No	5151 (77.170)	700 (77.272)	4022 (78.014)	429 (70.375)	.005
Yes	1716 (22.830)	181 (22.728)	1320 (21.986)	215 (29.625)
High WHTR
No	4377 (67.016)	597 (66.626)	3398 (67.414)	382 (64.301)	.032
Yes	2490 (32.984)	284 (33.374)	1944 (32.586)	262 (35.699)
High FLI
No	6209 (91.989)	812 (90.776)	4856 (92.771)	541 (87.150)	<.001
Yes	658 (8.011)	69 (9.224)	486 (7.229)	103 (12.850)
Sex
Male	3505 (51.884)	403 (48.882)	2685 (50.269)	417 (68.023)	<.0001
Female	3362 (48.116)	478 (51.118)	2657 (49.731)	227 (31.977)
Race/ethnicity
Mexican American	1930 (12.616)	196 (12.528)	1525 (12.613)	209 (12.741)	<.0001
Non-Hispanic Black	1881 (13.425)	354 (23.362)	1407 (12.688)	120 (8.139)
Non-Hispanic White	1957 (59.590)	193 (50.651)	1541 (59.807)	223 (67.872)
Other Hispanic	514 (7.117)	56 (6.226)	412 (7.416)	46 (5.749)
Other Race	585 (7.252)	82 (7.233)	457 (7.477)	46 (5.499)
Maternal smoking history during pregnancy
No	5914 (83.587)	701 (74.979)	4619 (83.826)	594 (91.324)	<.0001
Yes	953 (16.413)	180 (25.021)	723 (16.174)	50 (8.676)

Data for continuous variables were reported as mean ± standard error and analyzed using weighted ANOVA, and categorical variables were reported as number (percentage) and analyzed using weighted chi-square tests. FLI, Fatty Liver Index; MASLD, metabolic dysfunction-associated steatotic liver disease, defined as FLI ≥ 60.

BMI = body mass index, BW = birth weight, FLI = fatty liver index, HDL = high-density lipoprotein, MASLD = metabolic dysfunction-associated steatotic liver disease, NHANES = National Health and Nutrition Examination Survey, PIR = poverty income ratio, TC = total cholesterol, TG = triglycerides, WC = waist circumference, WHtR = waist-to-height ratio.

3.2. Association of BW with BMI, WC, WHtR, and FLI in adolescents

In fully adjusted model 2, BW was significantly and positively associated with BMI, WC, WHtR, and FLI among adolescents (BMI: β = 0.639, 95% CI = 0.343–0.934, P < .0001; WC: β = 1.872, 95% CI = 1.051–2.693, P < .0001; WHtR: β = 0.005, the 95% CI = 0.001–0.010, P = .0207; FLI: β = 2.128, 95% CI = 0.861–3.394, P = .0012). Notably, compared to NBW, HBW was associated with significantly increased BMI (β = 1.205), WC (β = 3.387), WHtR (β = 0.012), and FLI (β = 4.745), whereas there was no significant association for LBW (Table 2).

Table 2.

Association of BW with BMI, WC, WHtR, and FLI among adolescents.

	Crude model β (95% CI) P-value	Model 1 β (95% CI) P-value	Model 2 β (95% CI) P-value
BMI
BW (kg)	0.441 (0.098–0.783) .0125	0.669 (0.335–1.004) .0001	0.639 (0.343–0.934) < .0001
BW
LBW	0.264 (−0.348 to 0.877) .3987	−0.005 (−0.621 to 0.610) .9871	−0.174 (−0.721 to 0.372) .5330
NBW	Ref.	Ref.	Ref.
HBW	1.013 (0.442–1.585) .0006	1.321 (0.759–1.882) < .0001	1.205 (0.696–1.714) < .0001
P for trend	.0963	.0031	.0008
WC
BW (kg)	1.799 (0.870–2.729) .0002	1.958 (1.039–2.876) < .0001	1.872 (1.051–2.693) < .0001
BW
LBW	0.202 (−1.435 to 1.839) .8093	−0.062 (−1.690 to 1.566) .9405	−0.490 (−1.925 to 0.945) .5043
NBW	Ref.	Ref.	Ref.
HBW	3.404 (1.816–4.992) < .0001	3.681 (2.104–5.257) < .0001	3.387 (1.951–4.822) < .0001
P for trend	.0088	.0024	.0007
WHtR
BW (kg)	0.003 (0.000–0.008) .0449	0.006 (0.000–0.011) .0329	0.005 (0.001–0.010) .0207
BW
LBW	0.005 (−0.004 to 0.014) .2591	0.004 (−0.006 to 0.013) .4388	0.001 (-0.007 to 0.009) .8540
NBW	Ref.	Ref.	Ref.
HBW	0.007 (0.001–0.016) .0201	0.013 (0.005–0.022) .0033	0.012 (0.004–0.020) .0034
P for trend	.8037	.1762	.0800
FLI
BW (kg)	1.801 (0.316, 3.285) 0.0185	2.100 (0.606, 3.593) 0.0065	2.128 (0.861–3.394) 0.0012
BW
LBW	1.749 (−0.911 to 4.409) .1992	1.200 (−1.454 to 3.853) .3768	0.229 (−2.040 to 2.497) .8435
NBW	Ref.	Ref.	Ref.
HBW	4.540 (1.847–7.233) .0012	4.819 (2.136–7.502) .0006	4.745 (2.444–7.047) .0001
P for trend	.0047	.0049	.0135

Bold values indicate statistical significance (all P < .05) in the fully adjusted model (Model 2).

For BMI, WC, and WHtR, the crude model did not adjust for any covariates; model 1 partially adjusted for age, sex, race/ethnicity and PIR; and model 2 adjusted for age, sex, race/ethnicity, PIR, mother’s age when born, maternal smoking history during pregnancy, total energy intake, TG, TC, and HDL-C. For FLI, the crude model did not adjust for any covariates; model 1 partially adjusted for age, sex, race/ethnicity, and PIR; and model 2 adjusted for age, sex, race/ethnicity, PIR, mother’s age when born, maternal smoking history during pregnancy, and total energy intake.

BMI = body mass index, BW = birth weight, CI = confidence interval, FLI = fatty liver index, HBW = high birth weight, HDL-C = high-density lipoprotein cholesterol, LBW = low birth weight, NBW = normal birth weight, OR = odds ratio, PIR = poverty income ratio, TC = total cholesterol, TG = triglycerides, WC = waist circumference, WHtR = waist-to-height ratio.

3.3. Association of BW with general obesity, central obesity, and MASLD among adolescents

Similarly, in Model 2, BW was positively associated with BMI-defined general obesity. WC-defined central obesity, WHtR-defined central obesity, and FLI-defined MASLD (ORs of 1.473, 1.277, 1.169, and 1.311, respectively). Compared to NBW, HBW was associated with an increased prevalence of overall obesity, central obesity, and MASLD (overall obesity: OR 2.629, 95% CI 1.831–3.776; central obesity [WC]: OR 1.713, 95% CI 1.212–2.421; central obesity [WHtR]: OR 1.618, 95% CI 1.252–2.092; MASLD: OR 1.960, 95% CI 1.436–2.677). However, LBW was not associated with odds of obesity and MASLD (Table 3).

Table 3.

Association of BW with general obesity, central obesity, and MASLD among adolescents.

	Crude model OR (95% CI) P-value	Model 1 OR (95% CI) P-value	Model 2 OR (95% CI) P-value
General obesity
BW (kg)	1.398 (1.045–1.870) .0253	1.440 (1.082–1.918) .0135	1.473 (1.108–1.957) .0083
BW
LBW	1.101 (0.675–1.794) .7009	1.067 (0.662–1.720) .7900	1.021 (0.634–1.645) .9314
NBW	Ref.	Ref.	Ref.
HBW	2.412 (1.708–3.407) < .0001	2.569 (1.799–3.669) < .0001	2.629 (1.831–3.776) < .0001
P for trend	.0082	.0042	.0022
Central obesity (WC)
BW (kg)	1.211 (1.050–1.398) .0095	1.246 (1.077–1.441) .0035	1.277 (1.102–1.480) .0014
BW
LBW	1.029 (0.805–1.315) .8194	0.998 (0.780–1.276) .9863	1.059 (0.826–1.357) .6537
NBW	Ref.	Ref.	Ref.
HBW	1.512 (1.187–1.926) .0010	1.590 (1.238–2.041) .0004	1.713 (1.212–2.421) .0027
P for trend	.0329	.0128	.0045
Central obesity (WHtR)
BW (kg)	1.104 (1.001–1.259) .0398	1.139 (0.996–1.303) .0589	1.169 (1.021–1.338) .0250
BW
LBW	1.031 (0.815–1.304) .8014	0.997 (0.786–1.265) .9814	0.945 (0.737–1.211) .6537
NBW	Ref.	Ref.	Ref.
HBW	1.251 (1.001–1.579) .0014	1.321 (1.037–1.682) 0.0253	1.618 (1.252–2.092) .0003
P for trend	.0082	.0282	.034
MASLD
BW (kg)	1.246 (0.980–1.584) .0741	1.288 (1.013–1.637) .0402	1.311 (1.029–1.671) .0299
BW
LBW	1.270 (0.858–1.880) .2343	1.217 (0.822–1.803) .3282	1.171 (0.790–1.737) .4331
NBW	Ref.	Ref.	Ref.
HBW	1.823 (1.348–2.465) .0001	1.936 (1.424–2.631) < .0001	1.960 (1.436–2.677) < .0001
P for trend	.1728	.0970	.0698

Bold values indicate statistical significance (all P < .05) in the fully adjusted model (Model 2).

For general obesity and central obesity, the crude model did not adjust for any covariates; model 1 partially adjusted for age, sex, race/ethnicity and PIR; and model 2 adjusted for age, sex, race/ethnicity, PIR, mother’s age when born, maternal smoking history during pregnancy, total energy intake, TG, TC, and HDL-C. For MASLD, the crude model did not adjust for any covariates; model 1 partially adjusted for age, sex, race/ethnicity, and PIR; and model 2 adjusted for age, sex, race/ethnicity, PIR, mother’s age when born, maternal smoking history during pregnancy, and total energy intake.

BW = birth weight, CI = confidence interval, HBW = high birth weight, LBW = low birth weight, MASLD = metabolic dysfunction-associated steatotic liver disease, NBW = normal birth weight, OR = odds ratio, PIR = poverty income ratio, TC = total cholesterol, TG = triglycerides, WC = waist circumference, WHtR = waist-to-height ratio.

3.4. RCS analysis

RCS analysis showed that BW was nonlinearly correlated with BMI, WC, WHtR, and FLI among adolescents (p for nonlinearity < 0.0001, <0.0001, 0.0403, and 0.0001, respectively) (Fig. 2A–D). Similarly, BW was nonlinearly associated with the odds of general obesity, central obesity (WC), central obesity (WHtR), and MASLD among adolescents (p for nonlinearity was 0.004, 0.0027, 0.0027, and 0.0005, respectively) (Fig. 3A–D). Threshold effect analyses indicated that the association of BW with BMI, WC, WHtR, and FLI was only present for BW > 3 kg (β of 1.453, 4.088, 0.016, and 5.598, respectively) (Table S1). Similarly, the associations of BW with overall obesity, central obesity, and MASLD were significant after their respective inflection points (ORs of 1.915, 1.430, 1.237, and 1.719, respectively) (Table S2).

Figure 2.

RCS analysis of the association of BW with BMI, WC, WHtR, and FLI in US adolescents. (A) BMI; (B) WC; (C) WHtR; (D) FLI. BMI = body mass index, BW = birth weight, FLI = fatty liver index, RCS = restricted cubic spline, WC = waist circumference, WHtR = waist-to-height ratio.

Figure 3.

RCS analysis of the association of BW with obesity and MASLD in US adolescents. (A) general obesity; (B) central obesity (WC); (C) central obesity (WHtR); (D) MASLD. MASLD = metabolic dysfunction-associated steatotic liver disease, WC = waist circumference, WHtR = waist-to-height ratio.

3.5. Stratified analysis

Gender and race/ethnicity were selected for stratified analysis. Interaction analyses indicated race/ethnicity was a significant effect modifier, influencing the association of BW with BMI, WHtR, and FLI (p for interaction 0.037, 0.026, and 0.022, respectively) (Fig. 4). Interestingly, race/ethnicity also influenced the association of BW with central obesity (WC and WHtR) (Fig. S1).

Figure 4.

Stratified analysis of the association of BW with BMI, WC, WHtR, and FLI in US adolescents. BMI = body mass index, BW = birth weight, FLI = fatty liver index, WC = waist circumference, WHtR = waist-to-height ratio.

4. Discussion

BW was positively and nonlinearly associated with BMI, WC, WHtR, and FLI as well as general obesity, central obesity, and MASLD among U.S. adolescents in a national, large-sample, multiethnic, serial cross-sectional study. Compared with NBW, HBW was associated with significantly increased prevalence of BMI, WC, WHtR, and FLI, as well as obesity and MASLD, whereas LBW was not significantly associated. Race/ethnicity partially influenced these associations. These findings emphasize that U.S. adolescents with HBW are at risk for overall obesity, central obesity, and MASLD, and require early monitoring and timely prevention of obesity and steatosis. In addition, early maternal prevention of HBW may help reduce the risk of obesity and MASLD in subsequent adolescence.

To the authors’ knowledge, this is the first time that the association of LBW/HBW with general obesity, central obesity, and MASLD among adolescents has been explored in the national, large-sample, multiethnic NHANES database. A large body of observational research has suggested that HBW is associated with an increased risk of overall obesity and central obesity among children and adolescents. A meta-analysis published in 2011 that included 20 observational studies showed that HBW (compared to BW ≤4000 g) was positively associated with the risk of obesity among children and adolescents (OR 2.07, 95% CI 1.91–2.24), whereas LBW (compared to BW ≥2500 g) was associated with a decreased risk of obesity (OR 0.61, 95% CI 0.46–0.80).^[15] Interestingly, when NBW was used as a reference, the inverse correlation between LBW and obesity risk disappeared, while the positive correlation for HBW was maintained.^[15] A large-sample cross-sectional analysis from China demonstrated that HBW (defined as ≥3000 g) was associated with significantly increased odds of central obesity (as defined by WHtR) in children and adolescents (6–17 years old) compared to the reference (BW in the range of 2500–2999 g), while LBW (<2500 g) was not associated with the likelihood of central obesity.^[16] Another cross-sectional study from China that included 6561 participants showed that HBW increased the odds of general obesity in children and adolescents, while LBW decreased the prevalence of overweight.^[17] A cross-sectional analysis including Brazilian school-based adolescents aged 10 to 17 years suggested that BW was significantly and positively associated with BMI and WC, and that biological maturation partially mediated these associations.^[18] A retrospective cohort study from rural India enrolling 756 children aged 7 to 10 years showed that HBW (>3500g) was associated with significantly increased BMI in childhood compared to BW at 2500 to 2999 g.^[19] Shi et al included 10,041 children and adolescents aged 7 to 17 years in a large-sample cross-sectional analysis from China demonstrating a significant increase in the prevalence of general obesity in both participants with a BW ≥ 3500 g and a BW of 2500 to 2999 g compared to the reference (BW of 3000–3499 g).^[20] In a large cross-sectional analysis including 16,580 Chinese children and adults aged 7 to 17 years, Yuan et al demonstrated that participants with HBW (BW of 3500–5000 g) had a significantly increased risk of general obesity compared to those with BW of 3000 to 3499 g, and that those with BW of 4000 to 4499 g also had a significantly increased odds of central obesity (WHtR).^[21] Interestingly, participants with very low BW (< 1500g) had the highest risk of central obesity (OR = 2.03).^[21] A national cross-sectional analysis from Korea revealed a higher prevalence of general obesity in female, but not male, adolescents with BW > 75% compared to the reference with BW at 25% to 75% (OR = 2.13, 95% CI 1.03–4.41).^[39] Zou et al similarly showed in a cross-sectional analysis in China that HBW compared to NBW was associated with a significantly higher prevalence of general obesity among participants aged 6 to 18 years (OR 1.611).^[40] However, another retrospective cohort study from Australia indicated that BW was not associated with BMI in adolescents.^[41] Finally, a national cross-sectional analysis from Iran suggested that compared to NBW (2500–4000 g), HBW (>4000 g) was associated with increased odds of general obesity and central obesity in children and adolescents aged 6 to 18 years, whereas LBW was associated with a decreased prevalence of general obesity.^[42] Overall, these studies collectively suggest that higher BW is associated with an increased prevalence of overall and central obesity in children and adolescents, whereas the association of LBW with obesity risk in the adolescent population remains controversial. Using standard BW classification criteria, this study demonstrated for the first time in a nationally representative sample of adolescents in the U.S. that BW was positively and nonlinearly associated with both BMI, WC, and WHtR as well as general and central obesity. Compared with NBW, HBW was associated with significantly increased BMI, WC, and WHtR and prevalence of obesity, whereas LBW was not significantly associated. These findings were consistent with some of the previous studies in demonstrating that attention to obesity risk and early prevention is required among adolescents with HBW. In addition, this study provided the first indication that race/ethnicity may influence these associations, and the positive associations were more significant in other Hispanic ethnic populations, suggesting the need for individualized screening and prevention strategies.

The association of BW with MASLD in adolescents remains understudied. A multicenter cross-sectional study demonstrated that HBW was associated with an increased prevalence of biopsy-proven severe steatosis in participants < 21 years of age (OR 1.82, 95% CI 1.15–2.88), whereas LBW was associated with an increased odds of advanced liver fibrosis (OR 2.23, 95% CI 1.08–4.62).^[28] However, another biopsy-proven NAFLD cohort demonstrated that small for gestational age was significantly associated with severe steatosis in children and adolescents aged 6 to 17 years (OR 4.0, 95% CI 1.43–10.9), whereas the prevalence of moderate steatosis was not significantly different between the groups.^[30] A population-based cohort study from Australia demonstrated that BW was not associated with the risk of NAFLD (diagnosed by questionnaire and ultrasound) in adolescents aged 17 years.^[29] Similar to obesity indicators, the results suggested that HBW was associated with increased odds of MASLD among adolescents, whereas LBW was not significantly associated, suggesting that a history of HBW may require prompt attention as a group at risk for MASLD.

The finding that LBW was not significantly associated with obesity or MASLD in this adolescent cohort contrasts with some previous studies that reported positive associations.^[15,21,28] Several factors may explain this discrepancy. First, the effect of LBW on later metabolic health might be more pronounced in adulthood rather than adolescence, as the cumulative effects of postnatal factors such as accelerated catch-up growth and lifestyle habits may take time to manifest.^[7,8] Second, the study population was limited to adolescents, and the relatively short follow-up period since birth may not capture the full trajectory of disease development. Third, the complex interplay between LBW and subsequent obesity might be mediated by postnatal weight gain,^[43] a factor not available in this dataset. Future longitudinal studies tracking participants from birth through adolescence are needed to disentangle these temporal relationships.

Several mechanisms may explain the association of HBW with obesity and MASLD in adolescents. HBW may affect cardiac and circulatory health, leading to metabolic disorders and adipose tissue accumulation. Maternal blood glucose and subsequent fetal hyperinsulinemia may promote accelerated fetal growth.^[5] BW can be assumed to reflect the in-utero environment. Overnutrition during pregnancy affects BW and can lead to persistent epigenomic alterations, resulting in an increased risk of obesity in later life.^[16,44] Some genetic factors may also contribute to these associations.^[43] In addition, HBW may lead to higher levels of growth factors, which may affect the risk of obesity in later life.^[20] These mechanisms may confirm the developmental origins of health and disease, i.e., fetal programming permanently shapes the structure, function, and metabolism of the body and leads to subsequent disease.^[45]

The strength of this study lies mainly in the fact that it is a national, multiethnic, large-sample, population-based study, making the findings potentially generalizable. In addition, this is the first study to explore the association of LBW/HBW with general obesity, central obesity, and MASLD in a representative sample of U.S. adolescents, which filled a research gap and has potential clinical value. However, there are limitations to this study. It was a cross-sectional analysis, and therefore could not draw causal associations and still could not adequately adjust for confounding factors. Due to database limitations, some important influences such as maternal obesity status, genetic susceptibility, and history of catch-up growth were not available, which may have affected the findings. Additionally, data on pubertal status, a potential confounder in adolescent studies, were not consistently available across all survey cycles and thus could not be directly adjusted for; age was used as a proxy in the models. The self-reported nature of BW data, recalled by participants or their parents, is also subject to recall bias.^[31–33] The diagnosis of MASLD was based on noninvasive serologic markers rather than imaging or biopsy, which may have compromised accuracy. However, the FLI has proven its reliability in numerous NHANES and other population-level studies. Future large-sample well-characterized prospective cohort studies are needed to confirm these findings.

5. Conclusions

In a national cross-sectional analysis, BW was positively and nonlinearly associated with BMI, WC, WHtR, and FLI, as well as the prevalence of general obesity, central obesity, and MASLD among US adolescents. HBW but not LBW was associated with significantly increased odds of obesity and MASLD compared with NBW. These findings underline that adolescents with HBW may be at risk of developing obesity and MASLD and require early detection and intervention.

Author contributions

Data curation: Jinjin He.

Methodology: Chunmei Liu.

Writing – original draft: Jinjin He.

Writing – review & editing: Lijuan Zhu.