Associations of routine breakfast and napping habits with early adiposity rebound by age 3 years: a population-based cohort study in Japan
Article information
Abstract
Background
Early adiposity rebound (AR) is a key predictor of later obesity and metabolic risk, yet modifiable factors related to early AR remain understudied in large populations.
Purpose
To quantify the prevalence of early AR at age 3 years and identify modifiable correlates in a population‑based cohort of Japanese preschool children.
Methods
We retrospectively analyzed health-check records for 74,466 children who attended both 1.5- and 3-year examinations (2014–2019). Body mass index (BMI) values were converted to World Health Organization z scores; early AR was defined as any increase in BMI between 1.5 and 3 years. Multivariable logistic regression adjusted for birth weight category, sex, household structure, sleep duration, and behavioral factors.
Results
Early AR occurred in 18,673 children (25%), whereas obesity (BMI z score ≥1.64) was present in 4% at 3 years. After controlling the adjustments, routine breakfast consumption (odds ratio [OR] 0.88; 95% confidence interval [CI], 0.81–0.97) and regular napping at 1.5 years (OR, 0.84; 95% CI, 0.79–0.90) were independently associated with reduced odds of early AR, while obesity at 1.5 years strongly predicted early AR (OR, 4.32; 95% CI, 4.00–4.67). Routine juice intake or fast-food consumption showed no significant associations.
Conclusion
In this population‑based cohort, one in 4 preschoolers had early AR by age 3. Simple daily routines—eating breakfast and maintaining regular sleep—may help delay AR and offer actionable targets for early obesity prevention.
Key message
In a population‑based cohort of 74,466 children, 25% experienced early adiposity rebound (AR) by age 3. Daily breakfast and routine napping at 1.5 years were independently associated with lower odds of AR, while obesity at 1.5 years was a strong predictor. These modifiable routines could help delay AR and enable early identification during routine child health checks.
Graphical abstract. OR, odds ratio.
Introduction
The rising prevalence of childhood obesity has become a pressing global health concern. A comprehensive analysis reported a substantial increase in obesity rates among children and adolescents across more than 200 countries between 1990 and 2020, with prevalence reaching 7.5% in boys and 5.4% in girls by 2020 [1]. The prevalence of overweight or obesity among children in OECD (Organization for Economic Co-operation and Development) countries has reached approximately 25%, posing a significant public health challenge [2]. Obesity in children and adolescents, akin to its effects in adults, is linked to a range of adverse outcomes—including type 2 diabetes, metabolic dysfunction-associated steatotic liver disease, sleep apnea, and mental health conditions—while also increasing the risk of type 2 diabetes and cardiovascular disease in early adulthood [3,4]. Importantly, children with overweight or obese are more likely to become adults with obesity, facing a heightened risk of numerous serious health complications [5,6].
Adiposity rebound (AR), first described by Rolland-Cachera et al. [7], represents the point in childhood when body mass index (BMI), after declining during infancy, reaches its lowest point (nadir) around 5 to 7 years of age before beginning to increase again through adolescence. This shift from a decreasing to an increasing BMI trajectory is defined as AR. When AR occurs earlier than expected, a phenomenon termed early AR, it is recognized as a significant risk factor for obesity in childhood, adolescence, and adulthood. Early AR has also been strongly linked to an elevated risk of metabolic disorders, including type 2 diabetes [7-10]. These associations are mediated by several interrelated mechanisms, such as early proliferation of adipocytes, metabolic programming [11], persistent unhealthy dietary and physical activity patterns, hormonal dysregulation [12], genetic predisposition, and cumulative environmental exposures [11]. These interconnected processes can lead to lasting alterations in body composition and metabolic function, predisposing individuals to obesity and related metabolic disorders later in life. Recent longitudinal cohort studies have demonstrated that early AR, occurring between 1.5 and 3 years of age, is associated with increased BMI and waist-to-height ratio in adulthood [13]. Moreover, additional studies have linked early AR to obesity and metabolic syndrome in middle age, particularly after the age of 40 [8]. Evidence further indicates that early AR is associated with elevated triglycerides, apolipoprotein B, and blood pressure, with sex-specific differences observed—boys tend to exhibit higher blood pressure, while girls are more prone to lipid abnormalities [8]. Early AR has also been shown to accelerate fat accumulation during childhood, with significant implications for long-term metabolic health [7]. These effects are especially pronounced in children born small for gestational age or preterm, who face heightened risks of obesity and cardiovascular complications in adulthood [14,15]. This growing body of evidence underscores the importance of identifying and addressing early AR to mitigate its long-term impact on metabolic health.
Given the long-term risks associated with early AR, early identification and intervention targeting modifiable lifestyle factors—such as reducing screen time and promoting healthy dietary habits and physical activity—are considered [8,10,16]. Despite the global rise in childhood obesity, large-scale cohort studies focusing on factors related to early AR remain limited [11]. This study leverages a robust, large-scale population-based cohort from Japan’s health check-up system to assess the prevalence of early AR at age 3 years and to explore individual and social factors that may be associated with its early occurrence.
Methods
1. Study design and participants
This study analyzed longitudinal population-based cohort data collected from health check-up records in Okinawa, Japan, from Jan 2014 to Dec 2019 for children at 2 times during their childhood growth, at ages 1.5 years and 3 years. Health screenings for preschool children are a standard health care practice in Japan, and almost all infants and preschool children receive free health checks, which are legally required at ages 3 months, 9–10 months, 1.5 years, and 3 years [2]. In Okinawa, over 90% of the eligible population participated in a comprehensive health checkup in-person at age 3 years, a practice that continued until December 2019, before the onset of the coronavirus disease 2019 pandemic [17]. All eligible children were required to visit designated health examination centers, where they received a thorough assessment. Expert public health nurses and office staff collected detailed health questionnaire information, while pediatricians and dentists conducted physical examinations, ensuring a standardized evaluation for all participants. For this study, data for these children were collected at ages 1.5 years and 3 years and then linked across time. The study excluded data from children without complete health records at 1.5 or 3 years to maintain analytical integrity. Given Okinawa’s high population mobility, with past statistics indicating that approximately 10% of residents move or relocate each year, imputing missing data could have introduced inaccuracies and misrepresented the health status of the population.
This study was approved by the Research Ethics Board of the Okinawa Society of Child Health (No. 007). This study was conducted according to Ethical Guidelines for Medical and Health Research Involving Human Subjects in Japan. Written informed consent was obtained from the children’s parents or legal guardians, and we posted information regarding our research on the institution’s website.
2. Measurements
Data were collected on birth records, family structure, parental child-rearing practices, dietary habits, and dental health during child health examinations. Variables assessed only at age 3 years included siblings, single-parent households, family employment status (none, only father, only mother, or both parents working), attendance at childcare facilities, mandatory vaccination rates (≥80%), presence of a primary care physician, children’s dental hygiene (e.g., ability to brush teeth alone), daily juice consumption, and regular fast-food consumption (at least once every 2 weeks). Municipal immunization records were used to derive vaccination coverage; ≥80% was chosen a priori as a programmatic performance threshold commonly used in public health to indicate adequate uptake.
In contrast, information on caries and screen time (≥1 hour per day, including routine screen habits) was collected at both 1.5 and 3 years. Public health nurses conducted interviews during pediatric health examinations to gather detailed insights into child-rearing practices, providing a comprehensive understanding of factors influencing early childhood development. Questionnaire items, response options, and analytic coding were shown in Supplementary Table 1.
3. Definition of the obtained variables
According to the standards of the World Health Organization (WHO), pediatric weights were defined as follows: macrosomia > 4,000 g, normal birth weight 2,500–3,999 g, low birth weight 1,500–2,499 g, very low birth weight 1,000–1,499 g, and extremely low birth weight <1,000 g. Based on age- and sex-specific BMI z scores, overweight was defined as a BMI z score ≥1.04 and <1.64, and obesity was defined as a BMI z score ≥1.64 [16]. In this study, ‘early AR’ was defined operationally as any increase in BMI (kg/m²) between the 18‑month and 3‑year health checks, recognizing that 2 time points cannot determine the exact BMI nadir.11) For descriptive plots and subgroup tables, we classified non–BMI-increasing children into stable (-0.10 <ΔBMI [kg/m2] ≤0) and decrease (ΔBMI ≤-0.10 kg/m²) to avoid misclassifying changes within routine measurement error. The z scores of the BMI were calculated based on the WHO data.
4. Statistical analysis
Descriptive statistics were utilized to present proportions for categorical variables and medians with interquartile ranges for continuous variables. Univariate logistic regression analyses were conducted to estimate the odds ratios (ORs) and 95% confidence intervals (CIs) for the relationship between predictor variables and the outcome of early AR. All explanatory variables were categorical. Variables with 3 or more categories included data collection year and weight categorization at birth [11], while dichotomous variables included sex [11], single-parent household status [11], presence of siblings, childcare attendance [11], parental employment status, obesity, short sleep duration (less than 10 hours) [11], caries (more than 3), regular breakfast consumption [18], regular juice and fast-food consumption [19,20], and routine screen time at age 1.5 years [20]. Next, we used binary logistic regression analysis to identify factors associated with the development of early AR. In the univariate analysis, the association between each explanatory variable at 1.5 years and the outcome of early AR was assessed individually. Multivariable logistic regression was performed, adjusting for these variables to identify independent associations with early AR. We assessed multicollinearity using variance inflation factors; all variance inflation factors were <2.
Three‑group descriptive comparison (Supplementary Table 2). For descriptive characterization of growth patterns, children were partitioned into 3 mutually exclusive categories according to the change in BMI between visits (ΔBMI = BMI at 3 years − BMI at 18 months): AR (+) (ΔBMI >0), stable (-0.10 <ΔBMI [kg/m2] ≤0), and decrease (ΔBMI ≤ -0.10 kg/m²). Omnibus differences across the 3 groups were tested using chi-square tests for categorical variables and Kruskal-Wallis tests for continuous variables. A P value of <0.05 was considered statistically significant. All analyses were conducted using Stata/MP 17.0 (StataCorp, USA).
Results
The final analysis included 74,466 children with complete check-up data at both 1.5 years and 3 years, resulting in approximately 10.0% missing data from the 1.5-year cohort and 10.4% missing data from the 3-year cohort. Median BMI was 16.2 (interquartile range [IQR], 15.4–17.0) kg/m2 at 1.5 years and 15.7 (IQR 15.0–16.4) kg/m2 at 3 years. The median BMI z score was 0.46 (IQR. -0.16 to 1.03) at 3 years and 0.12 (IQR, -0.43 to 0.68) at 3 years. At age 3 years, based on WHO age‑ and sex‑specific BMI z scores, obesity was present in 4.0% (n=2,732) and overweight in 9.7% (n=7,198); at 1.5 years, the corresponding prevalences were 7.5% (n=5,596) and 17.2% (n=12,816), respectively.
Early AR, operationalized as ΔBMI >0 between the 2 visits, was observed in 25.0% (n=18,673) of children. As shown in Fig. 1, the distribution of ΔBMI (3 years–1.5 years) was approximately symmetric (skewness≈0.00); 25.0% showed an increase (AR (+)) and 75.0% did not. Among the nonincreasing group (ΔBMI ≤0 kg/m2), 71.0% (n=52,832) had a decrease (ΔBMI ≤-0.10 kg/m2) and 4.0% (n=2,961) were essentially unchanged (-0.10 < ΔBMI [kg/m2] ≤ 0).
Distribution of change in BMI (kg/m2) (ΔBMI=BMI at 3 years − BMI at 1.5 years) in the population‑based cohort. The vertical line at Δ=0 demarcates AR (+) (ΔBMI >0) versus nonincrease. BMI, body mass index; AR, adiposity rebound.
The following other categorical values were observed for the study population (Table 1): 37,975 (51%) male children; median birth weight of 3,000 g (IQR, 2745–3260 g); 6,041 (8%) had single-parent households; 55,119 (83%) had other siblings at home; 64,556 (87%) attended childcare; 67,362 (91%) had a routine nap time; 53,368 (72%) slept <10 hours; 13,809 (18%) were able to brush their teeth alone; 71,707 (96%) routinely ate breakfast; 19,198 (26%) had dental caries at age 3 years, and 1887 (3%) had dental caries at 1 year; 19,479 (26%) routinely drank juice; 63,538 (86%) routinely ate fast food; 14,229 (19%) used a car seat; and 65,818 (88%) had a primary care physician.
Baseline characteristics of the study population at age 3 years (N=74,466)
At 3 years of age, 25% of children had early AR. Besides the factors detailed in Table 2, a factor significantly associated with early AR on logistic regression analysis was obesity at age 1.5 years (OR, 4.32; 95% CI, 4.00–4.67). On multiple logistic regression analysis, obesity, birth weight categorization, and mandatory vaccine rate >80% at 1.5 years of age were positively associated with early AR. The following factors, however, were negatively associated with early AR for children: male sex, single-parent status, worker’s role in a family, routine nap time, and routine consumption of breakfast, juice, or fast food. Among these children, male sex (OR, 0.59; 95% CI, 0.57–0.61) and routine napping habits (OR, 0.84; 95% CI, 0.79–0.90) were significant factors that decreased the risk of early AR, while obesity at 1.5 years strongly predicted early AR (OR, 4.32, 95% CI 4.00–4.67). Routine juice intake or fast-food consumption showed no significant associations. Supplementary Table 2 documents the presence of a stable group with similar BMI at both time points and compares characteristics across AR (+), stable, and decrease categories.
Factors associated with early adiposity rebound (AR) by age 3 years
Discussion
This study is the first to investigate the prevalence and risk factors of early AR for children at age 3 years in a large pediatric population. Although the prevalence of obesity was only 4% in this large pediatric cohort of Japanese children at age 3 years, 25% of children had already experienced AR. After controlling for factors related to early AR, obesity, birth weight categorization, and mandatory vaccine rate >80% were positively associated with early AR. By contrast, male sex, a single-parent household, parental employment status, and routine napping were associated with lower odds of early AR, whereas regular juice and fast‑food consumption showed no significant associations.
The mechanisms driving the development of early AR are complex and multifaceted, arising from intricate interactions between neurodevelopment, metabolic demands, endocrine fluctuations, and genetic and environmental factors [11]. One hypothesis suggests that changes in the brain's energy requirements during critical developmental periods may result in excess energy being stored in adipose tissue, thereby contributing to early AR [20]. Hormonal variations, including changes in growth hormone and insulin-like growth factor-1 levels, are also thought to influence adipose tissue metabolism and AR timing [12,21]. Additionally, early-life factors such as nutritional status, physical activity levels, and genetic predispositions likely play essential roles in determining the onset of early AR [19]. Understanding why early AR occurs is vital for identifying at-risk populations and for informing potential prevention strategies. Further research in diverse populations is necessary to explore how genetic, cultural, and environmental factors collectively influence the timing and consequences of early AR.
Our study is the first to specifically examine early AR among 3-year-old children, revealing that 25% had already experienced early AR by this age. However, the long-term consequences of early AR in this population remain unclear. Previous research in Japan indicates that AR typically occurs around the age of 5 years [8], which is earlier than what has been observed in European populations [7,15]. This variability highlights the potential influence of genetic, cultural, and environmental differences on the timing of AR. Furthermore, a recent retrospective cohort study from a Japanese academic university hospital suggested that some children may already show signs of early AR by their 3-year health check-ups [13]. These findings underscore the need for further international studies to investigate early AR in different populations, as they may uncover new insights into its mechanisms and implications, particularly when considering variations across regions and cultures.
Our study demonstrated that routine consumption of breakfast and naptime habits were associated with lower odds of early AR. It has been widely reported that routine breakfast consumption is associated with the prevention of childhood obesity [18]. No previous studies have been conducted to investigate whether napping prevents early onset of AR or obesity in children younger than 3 years. In another regard, short sleep time has been linked to obesity in preschool children [22]. More than 70% of the study participants did not meet the WHO recommendation of 10 h of sleep or more [22], suggesting that adequate total sleep duration may play a more critical role than napping itself in mitigating early AR.
Sex differences appear to influence childhood growth trajectories and subsequent obesity risk. Based on data collected from 1983 to 2010 for 5000 boys and 2000 girls in Poland, AR occurred earlier in girls [23,24]. Other studies have similarly reported that girls, particularly those with higher BMI percentiles, tend to experience earlier AR than boys [25]. Conversely, some studies found negligible or nonexistent sex differences. Of concern, girls with increased BMI before the age of 3 were more likely to develop insulin resistance by age 12 compared to boys [26]. These findings highlight the importance of heightened vigilance among parents and caregivers regarding early AR in daughters.
By contrast, neither routine juice consumption nor fast-food consumption was associated with an increased risk of early AR. The timing of AR has not been demonstrated to be linked to high protein intake in several studies [27,28]. Additionally, no significant associations have been reported between the timing of AR and nutritional intake of energy foods, fats, or carbohydrates [29]. These results underscore the multifactorial nature of early AR and suggest that dietary patterns alone may not fully explain its onset.
The strengths of this study are 3 key factors: it was the first study with a large sample size design conducted in Japan; it had a high rate of child participation in routine health check-ups; and the data accuracy was ensured by data collection performed by trained nurses, physicians, and dentists. This study has several limitations. First, its retrospective design limited the ability to establish temporal relationships and causality. Specifically, we did not have access to information about maternal and paternal BMI or the educational backgrounds of the parents or guardians. Second, information on children’s dietary and sleep habits was based on caregiver self-reports, which may be subject to recall and social desirability bias. Additionally, we did not gather data on physical activity during these health examinations for children younger than 3 years, which is important because early AR appears to reflect a positive energy balance during the preschool years [30]. However, the exact contributions of physical activity, sedentary behavior, and food intake to the positive energy balance and to the timing of AR remain unclear [10]. Another key limitation is that we did not prespecify a threshold for a clinically meaningful increase in BMI (e.g., ΔBMI [kg/m2] ≥0.10 or ≥0.20). Future work should evaluate such thresholds; the absence of this criterion may attenuate effect estimates. Finally, our 2 measurement points (1.5 years and 3 years) cannot establish the true BMI nadir; therefore, some children categorized as early AR may simply be on an upward trajectory without having reached the nadir. This potential misclassification likely biases associations toward the null rather than creating spurious findings. Despite these limitations, this study represents the largest investigation to date on the prevalence and risk factors of early AR in Japanese preschool children. By incorporating a variety of socioeconomic factors commonly observed in developed countries, this research provides valuable insights into early AR. These findings may significantly inform future studies of childhood obesity, particularly in understanding the interplay of socioeconomic, behavioral, and biological factors. Future research should further examine these relationships and assess strategies that may help address early AR. Understanding why early AR occurs is vital for identifying at-risk populations and informing prevention efforts, while recognizing that additional longitudinal studies are needed to clarify causal pathways and underlying mechanisms.
In conclusion, in this population-based cohort of 74,466 children, approximately 25% had early AR by age 3. Female sex and nonroutine napping (shorter sleep) were associated with higher odds of early AR, while routine breakfast consumption and napping habits were associated with lower odds after adjustments, suggesting actionable prevention targets.
Supplementary materials
Supplementary Tables 1-2 is available at https://doi.org/10.3345/cep.2025.01998.
Questionnaire items, response options, and analytic coding 2014–2019
Comparison of characteristics by BMI change between 18 months and 3 years (AR (+) = ΔBMI >0; stable = -0.10 <ΔBMI ≤0; decrease = ΔBMI ≤-0.10 kg/m2)
Notes
Conflicts of interest
No potential conflict of interest relevant to this article was reported.
Funding
This work was supported by JSPS KAKENHI Grant (Grant-in-Aid for Research Activity Start-up) Number 20K23229 and The Japan Foundation for Pediatric Research (JFPR) Number 20-013. The JSPS and JFPR had no role in the design and conduct of the study.
Acknowledgments
We are grateful to the steering committee of the Okinawa Child Health Study Group, including Doctors Masaya Miyagi, Hirotsune Hamabata, Keisuke Katsuren, and Takaya Toma, for their invaluable contributions to this study.
Author contribution
TY is the only contributing author listed for this manuscript.