Longitudinal Tracking of Objective Measures of Cardiorespiratory Fitness, Body Composition, and Physical Activity in Children

Abstract

Purpose:

Understanding the longitudinal tracking of cardiorespiratory fitness (CRF), adiposity, and moderate-to-vigorous physical activity (MVPA) during childhood is essential for developing effective health interventions. Our study evaluated the tracking of objectively measured MVPA, CRF, and adiposity from early to late elementary school age.

Methods:

Participants were recruited through five elementary schools in the Kanto region in Japan. Of the 247 children assessed at baseline in first grade (ages 6–7), 244 participated in a follow-up in fifth grade (ages 10–11). CRF was assessed as peak oxygen uptake (VO₂peak, mL·kg⁻¹·min⁻¹) during a progressive treadmill test. Body fat percentage was measured using dual-energy X-ray absorptiometry. MVPA was measured using a triaxial accelerometer worn for 14 consecutive days.

Results:

Spearman rank correlations indicated high tracking of body fat percentage in boys (ρ = 0.65, P < 0.001) and girls (ρ = 0.61, P < 0.001), and moderate and high tracking of MVPA in boys (ρ = 0.58, P < 0.001) and girls (ρ = 0.63, P < 0.001). CRF showed moderate tracking in boys (ρ = 0.57, P < 0.001), but weaker tracking in girls (ρ = 0.29, P < 0.01). Generalized estimating equations analyses adjusted for height revealed significant tracking for all three indicators, with a weaker tracking pattern for CRF in girls compared with boys: body fat percentage (β = 0.89 in boys, 0.79 in girls, both P < 0.001), MVPA (β = 0.70 in boys, 0.69 in girls, both P < 0.001), and CRF (β = 0.76 in boys, 0.54 in girls, both P < 0.001).

Conclusions:

These findings support the need for targeted approaches during the early school years to prevent the consolidation of unfavorable profiles in fitness, adiposity, and physical activity. In addition, population-wide approaches throughout the school-age years may be important in supporting fitness development in girls, given the greater variability observed in CRF.

Cardiorespiratory fitness (CRF), body composition, and physical activity (PA) during childhood are key determinants of long-term health (1). Low CRF (2), increased adiposity (3), and insufficient PA (4) during childhood have been linked to elevated risks of cardiovascular and metabolic diseases, including type 2 diabetes, hypertension, and dyslipidemia, both in childhood and later in life. These early-life risk factors—low CRF, high adiposity, and low PA—tend to persist over time, thereby increasing the cumulative burden on lifelong health.

A fundamental concept in this context is tracking, which refers to the stability of an individual’s relative ranking within a population over time (5,6). Strong tracking indicates that individuals tend to maintain their position among others, whereas weak tracking suggests greater variability over time. Understanding these tracking patterns can inform targeted health strategies, such as optimal timing for screening and intervention efforts, and early identification of individuals at risk of cardiometabolic diseases.

Reviews have reported moderate tracking of CRF (7) and overweight status (8), while PA tracks at low to moderate levels, with particularly low tracking in younger age groups (9–11). Longitudinal studies comparing PA and CRF tracking over 5–7 yr have shown that PA tracks less well than CRF from childhood through adolescence (12) but exhibits similar tracking to CRF from preadolescence onward (13,14). These findings suggest that PA is less stable than CRF, particularly during early childhood, potentially due to grater behavioral variability.

However, current evidence is limited by several methodological weaknesses. Most tracking studies have relied on self-reported PA data, which has limited reliability and validity, particularly in younger children (15). Self-report tools are prone to recall bias, social desirability bias, and age-dependent differences in comprehension, making them unsuitable for capturing habitual activity levels, especially in early childhood (15). Similarly, field-based CRF assessments (e.g., shuttle run tests) and anthropometric indicators (e.g., body mass index) are commonly used as proxies for aerobic fitness and adiposity, respectively, but these measures lack the precision of laboratory-based methods. Despite numerous studies on tracking, only a few have comprehensively examined PA, CRF, and adiposity using high-precision, objective measures (12–14,16,17).

Furthermore, many studies focus on a single parameter (CRF, adiposity, or PA) in isolation, without considering the relative tracking strength among them. This limits our understanding of which outcome may serve as the most sensitive marker for identifying at-risk individuals. A more comprehensive comparison of the relative tracking strength among these parameters is therefore warranted. Additionally, few studies have covered the full span of childhood and preadolescence using consistent, high-quality, objective measures.

While the majority of existing studies have been conducted in Western countries, particularly in North America and Europe (7,8,11–14), there is limited evidence from Asian populations. This gap presents a clear opportunity to extend the research to Asian populations, including Japanese youth (17). This is essential to develop population-specific strategies for promoting lifelong health.

Therefore, the aim of our study was to address these gaps by examining the longitudinal tracking of CRF, adiposity, and PA throughout childhood and preadolescence using objective, gold-standard measures. By comparing the relative stability of these indicators, our study provides novel insights into which parameter may serve as the most robust early marker of future health risk, especially in an Asian population. Our findings may inform the development of more effective, population-specific interventions to improve lifelong health trajectories.

METHODS

Participants

We conducted a 4-yr longitudinal study that followed first-grade elementary school students (ages 6–7). Baseline data were collected in 2012–2013, and follow-up data were collected in 2016–2017. Participants were recruited from five elementary schools in the Kanto region of Japan. Of the 247 students (152 boys and 95 girls) who completed the baseline assessments, 244 participated in at least one of the follow-up assessments at fifth grade (ages 10–11), yielding an overall follow-up rate of 98.8%. The remaining three participants were excluded due to loss to follow-up or withdrawal from the study. Of the 244 participants, 234 attended the follow-up site visit and underwent CRF assessments, and 231 completed body composition measurements. In total, 244 participants—including 10 who did not attend the site visit—completed PA monitoring. For each outcome variable, analyses were restricted to participants with valid measurements at both baseline and follow-up (Fig. 1).

FIGURE 1.

Participant flow diagram. Two participants had invalid PA data at both baseline and follow-up; therefore, 19 unique participants were excluded, resulting in a final sample of 225 children for PA analyses. CRF, cardiorespiratory fitness; DXA, dual-energy X-ray absorptiometry; PA, physical activity.

Our study represents a secondary analysis of a previously conducted cohort study (18,19), focusing on the long-term tracking of physical fitness and body composition, an aspect that was not examined in the original study. The study protocol was approved by the ethics committees of the National Institute of Health and Nutrition in Japan and the Chiba Institute of Technology. All participants and their parents received both written and verbal explanations of the study procedures. Written informed consent was obtained from the parents before participation.

Anthropometry and Body Composition

We assessed weight, height, and body composition during the school summer vacation (July–August) in both first and fifth grades at designated facilities within the authors’ research institutions (20). Body composition was assessed using dual-energy X-ray absorptiometry (DXA) (Hologic QDR-4500, Hologic, Inc., Waltham, MA). Body fat percentage was calculated as the ratio of fat mass to total body weight. Fat-free mass (FFM) was also obtained from the scan. We conducted all measurements with participants wearing light indoor clothing and no shoes.

Cardiorespiratory Fitness

On the same day as the anthropometric measurements, we assessed VO₂peak (mL·kg⁻¹·min⁻¹) using an incremental maximal exercise test on a treadmill (19). The test for first-grade children began with 2 min of walking at 4.2 km·h⁻¹ (70 m·min⁻¹), followed by running at a constant speed of 6.0 km·h⁻¹ (100 m·min⁻¹) with a progressive increase in gradient. After the running phase began, the gradient was increased by 2.5% every 2 min. The test for fifth-grade children began with 2 min of walking at 7.2 or 8.4 km·h⁻¹ (120 or 140 m·min⁻¹), followed by running at a constant speed of 12.0 or 13.2 km·h⁻¹ (200 or 220 m·min⁻¹) with a progressive increase in gradient. To ensure that maximal exertion was reached within 9–12 min, the initial speed and gradient were adjusted based on each participant’s estimated fitness level, which was based on their exercise habits as well as their heart rate and physiological responses during the initial stages of the progressive increase. During the treadmill test, standardized verbal encouragement was provided only by trained research staff to ensure consistency across participants. Exhaustion was determined according to Armstrong’s guideline (21), which considers criteria such as 1) facial flushing, sweating, hyperpnea, unsteady gait; 2) heart rate leveling off at about 200 beats· min⁻¹; and 3) respiratory exchange ratio >1.0. Respiratory gases were analyzed using a gas analyzer (Aero Monitor, AE310S, Minato Medical Science, Osaka, Japan). Participants wore a silicone mask covering the mouth and nose, connected to the gas analyzer for breath-by-breath analysis. VO₂ data were averaged every 30 s, and VO₂peak was determined based on the highest recorded value during the test. VO₂peak was also expressed as per kilogram of FFM (mL·kg⁻¹ FFM·min⁻¹).

Physical Activity

PA was measured using a triaxial accelerometer (Active style Pro HJA-350IT, Omron Healthcare, Kyoto, Japan), which participants wore on the waist for 14 consecutive days during the autumn school term (October–November) in both first and fifth grades (18,19,22). Participants were instructed to wear the devices during waking hours, except when sleeping, bathing, or swimming.

Data were recorded in 10-s epochs and processed using a macro developed by the Japan Physical Activity Research Platform (http://paplatform.umin.jp [in Japanese]). We analyzed data from 7:00 am to 9:00 pm (11). Nonwear time was defined as ≥20 consecutive minutes of zero counts. Data were considered valid if wear time was ≥10 h·d⁻¹ on at least 2 weekdays and 1 weekend day (23). After applying these criteria, data from 7 participants were excluded at baseline and 14 at follow-up due to insufficient accelerometer data. Among these, 2 were duplicates, resulting in the exclusion of 19 unique participants from the 244 individuals who participated in the PA monitoring at follow-up. Consequently, PA data from 225 participants were included in the analysis (Fig. 1).

Based on the ratio of unfiltered to filtered synthetic acceleration (24), activities were categorized as locomotive (e.g., walking, running) and nonlocomotive activities (e.g., playing games, cleaning, building blocks, ball play, aerobic dance). Raw acceleration signals were converted into metabolic equivalents (METs) using validated equations for primary school children (25):

• Locomotive activity: 0.6237 × METs + 0.2411

• Nonlocomotive activity: 0.6145 × METs + 0.5573

Moderate-to-vigorous physical activity (MVPA) was defined as ≥3 METs and vigorous physical activity (VPA) as ≥6 METs. Daily time (min·d⁻¹) spent in each activity was calculated as a weighted average to account for weekday and weekend differences:

• ([weekday mean × 5] + [weekend mean × 2])/7

Statistical Analysis

Descriptive statistics were used to summarize participants’ physical characteristics, with all measured and calculated values presented as means and standard deviations (SDs). Differences in characteristics between boys and girls were examined using independent-samples t tests. Spearman rank correlation coefficients (ρ) were calculated to assess the stability of measurements between first and fifth grades, stratified by sex. This method is commonly used to evaluate the tracking of measurements over time (14,26). To visually complement the analysis, the correlations of VO₂peak, body fat percentage, and MVPA between first and fifth grades were illustrated using scatter plots for boys and girls separately, including regression lines and R² values. The stability of measurements was interpreted based on the following thresholds for Spearman’s correlation coefficient: <0.30 = low stability, 0.30–0.60 = moderate stability, and >0.60 = high stability (6,7).

In addition to rank correlations, generalized estimating equations (GEE) were used to assess the stability of VO₂peak, body fat percentage, and MVPA from baseline to follow-up. While rank correlations assess the stability of individual rankings over time, GEE models account for within-subject correlation inherent in repeated-measures data and provide population-averaged estimates. Each outcome (body fat percentage, VO₂peak, and MVPA) was modeled as a continuous dependent variable, with time (baseline vs follow-up) included as a fixed effect. Height was included as a covariate to account for differences in skeletal growth. Wear time was not added as a covariate in the MVPA models, as all data had already been filtered using strict wear-time criteria and restricted to the 7:00 am to 9:00 pm period, minimizing potential bias due to nonwear time. Body composition was not included as a covariate in the CRF models, as the objective was to evaluate the tracking of CRF itself. An exchangeable working correlation structure was assumed. Results are presented as standardized (β) and unstandardized (B) coefficients with 95% confidence intervals (CIs). Statistical significance was set at P < 0.05. All analyses were performed using STATA 18.0 (StataCorp LLC, College Station, TX).

RESULTS

Table 1 presents the baseline (first grade) and follow-up (fifth grade) characteristics of the participants. The mean age at baseline and follow-up was 6.9 (SD: 0.3) and 10.9 (SD: 0.3) yr, respectively, for both boys and girls. At baseline, boys had significantly greater height, weight, FFM, and VO₂peak than girls, while girls had a higher body fat percentage. Boys also accumulated more MVPA and VPA than girls. At follow-up, similar sex differences persisted, with boys showing higher VO₂peak, FFM, and levels of MVPA and VPA, whereas girls maintained a higher body fat percentage.

TABLE 1.

Participant characteristics.

	Boys (N = 143)		Girls (N = 91)		P value
Baseline (first grade)	n		n
Age (months)	143	83 (4)	91	83 (3)	0.77
Height (cm)	143	119.8 (5.1)	91	117.9 (4.3)	0.003
Weight (kg)	143	22.0 (3.1)	91	21.1 (2.6)	0.025
VO2peak (mL·kg−1·min−1)	143	50.9 (5.0)	91	47.1 (5.1)	<0.001
VO2peak (mL·kg−1 FFM·min−1)	143	65.3 (5.3)	91	63.0 (6.2)	0.003
Body fat percentage (%)	142	20.4 (4.2)	89	23.6 (4.1)	<0.001
Fat-free mass (kg)	142	17.0 (2.0)	89	15.7 (1.5)	<0.001
MVPA (min·d−1)	134	87.2 (22.8)	91	68.7 (17.3)	<0.001
VPA (min·d−1)	134	8.6 (4.7)	91	5.2 (2.7)	<0.001
Wear time (min·d−1)	134	771.6 (27.5)	91	769.9 (31.1)	0.67
Number of valid days	134	12.3 (2.1)	91	12.3 (1.8)	0.81
Follow-up (fifth grade)	n		n
Age (months)	143	131 (3)	91	131 (3)	0.83
Height (cm)	143	141.8 (6.1)	91	142.2 (6.0)	0.64
Weight (kg)	143	34.6 (6.0)	91	34.2 (5.7)	0.60
VO2peak (mL·kg−1·min−1)	143	52.6 (6.6)	91	47.5 (5.1)	<0.001
VO2peak (mL·kg−1 FFM·min−1)	142	68.1 (6.0)	89	63.3 (5.2)	<0.001
Body fat percentage (%)	142	21.4 (5.7)	89	23.2 (4.7)	0.014
Fat-free mass (kg)	142	26.5 (3.6)	89	25.5 (3.5)	0.034
MVPA (min·d−1)	143	66.2 (19.5)	92	44.5 (12.3)	<0.001
VPA (min·d−1)	143	8.1 (5.0)	92	4.2 (3.3)	<0.001
Wear time (min·d−1)	143	783.2 (30.8)	92	779.9 (29.8)	0.42
Number of valid days	143	12.0 (2.6)	92	12.5 (2.3)	0.18

Values are means and standard deviations (SDs).

VO₂peak, peak oxygen uptake.

We summarized the tracking of CRF, body composition, and PA using Spearman’s rank correlation coefficients, as shown in Table 2. All variables exhibited significant positive correlations over time. Body fat percentage showed high tracking in both boys (ρ = 0.65, P < 0.001) and girls (ρ = 0.61, P < 0.001). MVPA demonstrated moderate tracking in boys (ρ = 0.58, P < 0.001) and high tracking in girls (ρ = 0.63, P < 0.001). In contrast, CRF tracking differed by sex: boys showed moderate tracking (ρ = 0.57, P < 0.001), while girls showed weaker tracking (ρ = 0.29, P < 0.01), indicating less stability in CRF over time among girls. The accelerometer data used for PA analysis encompassed an average wear time of over 13 h·d⁻¹ on more than 12 d for both baseline and follow-up assessments, with no significant differences observed between boys and girls.

TABLE 2.

Spearman correlation coefficients for tracking of CRF, body composition, and physical activity.

	Boys		Girls
	n	ρ	n	ρ
Height (cm)	143	0.917**	91	0.778**
Weight (kg)	143	0.828**	91	0.726**
VO2peak (mL·kg−1·min−1)	143	0.571**	91	0.287*
VO2peak (mL·kg−1 FFM·min−1)	142	0.382**	89	0.246*
Body fat percentage (%)	142	0.646**	89	0.606**
Fat-free mass (kg)	142	0.851**	89	0.732**
MVPA (min·d−1)	134	0.576**	91	0.629**
VPA (min·d−1)	134	0.573**	91	0.539**

VO₂peak: peak oxygen uptake.

^*p < 0.01;

^**P < 0.001.

Because the MVPA tracking analysis was limited to participants who met a predefined wear-time criterion, we further assessed the robustness of these estimates by reanalyzing the data using a more stringent inclusion threshold. Specifically, we changed the requirement from at least 2 weekdays and 1 weekend day to a stricter threshold of 5 weekdays and 2 weekend days, based on the recommended minimum of 4–5 valid days for school-aged children (27). This led to the exclusion of 23 additional participants, resulting in a final sample of 202 (117 boys and 85 girls). The change had minimal impact on the MVPA tracking estimates: the Spearman rank coefficients (ρ) changed from 0.58 to 0.57 in boys, and from 0.63 to 0.62 in girls. These nearly identical values indicate that the findings were not sensitive to the wear-time criterion.

Figure 2 illustrates the tracking patterns of body fat percentage, CRF, and MVPA between first and fifth grades using scatter plots, stratified by sex. Pearson correlation coefficients (r) for each variable were as follows: body fat percentage (boys: r = 0.69, girls: r = 0.65), CRF (boys: r = 0.54, girls: r = 0.25), and MVPA (boys: r = 0.58, girls: r = 0.63).

FIGURE 2.

Scatter plots with fitted regression lines showing tracking of CRF, adiposity, and MVPA from baseline (first grade) to follow-up (fifth grade). MVPA, moderate-to-vigorous physical activity; VO₂peak, peak oxygen uptake.

GEE analysis (Table 3) further supported the rank correlation findings, confirming tracking of VO₂peak, body fat percentage, and MVPA in both boys and girls (all P < 0.001). However, the tracking patterns varied by sex. Notably, VO₂peak showed weaker tracking in girls (β = 0.54) compared with boys (β = 0.76), indicating lower stability in fitness levels over time among girls. In contrast, body fat percentage (boys: β = 0.89; girls: β = 0.79) and MVPA (boys: β = 0.70; girls: β = 0.67) demonstrated relatively stable tracking in both sexes, with adiposity showing the greatest consistency. These results highlight sex-specific differences in the longitudinal stability of key physical health indicators during the school-age years.

TABLE 3.

Generalized estimating equations (GEE) results for tracking of CRF, adiposity, and MVPA from baseline (first grade) to follow-up (fifth grade).

	Boys							Girls
	β	Β	SE	z	P value	95% CIs		β	Β	SE	z	P value	95% CIs
VO2peak
Baseline VO2peak (mL·kg−1·min−1)	0.76	0.85	0.05	18.35	<0.001	0.76	0.94	0.54	0.61	0.05	12.31	<0.001	0.51	0.71
Time (follow-up vs baseline)	0.59	3.56	1.02	3.48	0.001	1.56	5.57	0.66	3.97	1.35	2.94	0.003	1.32	6.63
Height (cm)	−0.01	−0.08	0.04	−2.01	0.044	−0.16	0.00	−0.02	−0.15	0.05	−2.99	0.003	−0.24	−0.05
Constant	1.60	17.63	5.67	3.11	0.002	6.52	28.74	2.65	35.68	6.52	5.47	<0.001	22.91	48.45
Body fat percentage
Baseline body fat percentage (%)	0.89	0.99	0.04	23.41	<0.001	0.91	1.08	0.79	0.88	0.05	18.96	<0.001	0.79	0.97
Time (follow-up vs baseline)	0.50	2.51	0.79	3.19	0.001	0.97	4.05	0.19	0.96	0.98	0.98	0.326	−0.95	2.87
Height (cm)	−0.01	−0.07	0.03	−2.22	0.027	−0.13	−0.01	−0.01	−0.06	0.04	−1.54	0.123	−0.13	0.02
Constant	1.67	8.65	3.69	2.34	0.019	1.42	15.89	1.36	9.51	4.34	2.19	0.028	1.01	18.01
MVPA
Baseline MVPA (min·d−1)	0.70	0.75	0.03	25.22	<0.001	0.69	0.80	0.67	0.72	0.03	25.87	<0.001	0.66	0.77
Time (follow-up vs baseline)	−0.58	−14.01	3.20	−4.38	<0.001	−20.28	−7.74	−0.64	−15.32	2.70	−5.68	<0.001	−20.61	−10.03
Height (cm)	−0.01	−0.29	0.12	−2.39	0.02	−0.53	−0.05	−0.02	−0.37	0.10	−3.83	<0.001	−0.55	−0.18
Constant	1.99	57.02	14.82	3.85	<0.001	27.97	86.06	2.12	62.42	11.52	5.42	<0.001	39.84	85.01

β, standardized coefficient; B, unstandardized coefficient; SE, standard error; z, z value; 95% CIs, 95% confidence intervals; VO₂peak, peak oxygen uptake.

DISCUSSION

In this study, we investigated the longitudinal tracking of CRF, adiposity, and MVPA in elementary school children from first to fifth grade. Our rank correlation results suggest that the stability of these health indicators over 4 yr was generally moderate to high, with tracking of body fat percentage showing a relatively high level of stability for both boys and girls. MVPA tracking was moderate in boys and high in girls, indicating relatively stable activity patterns, particularly among girls. However, the tracking of CRF in girls was notably weaker compared with boys, indicating lower consistency in individual CRF levels over time. Additionally, the results from the GEE analysis further supported these findings, indicating a slightly lower tracking of CRF in girls.

Our findings indicate that CRF in girls exhibits lower longitudinal stability compared with boys. McMurray et al. (12) observed no apparent sex differences in CRF tracking among U.S. youth, with Spearman correlation coefficients for predicted VO₂max from ages 8 to 16 yr (years 1–7) being similar for boys and girls (ρ = 0.50–0.53). In an Australian longitudinal study using the PWC170 test, Guo et al. (26) reported correlation coefficients of 0.27 for males and 0.30 for females from childhood to young adulthood. The Muscatine Study (13) reported moderate tracking of VO₂peak over 5 yr, starting around age 10, and found correlation coefficients of 0.44 for boys and 0.39 for girls. Similarly, Sasayama et al. (17) examined the tracking of the number of laps in the 20-m shuttle run test over 3 yr, starting around ages 9–10, and reported correlation coefficients of 0.62 for boys and 0.51 for girls. These results suggest that the stability of CRF over time may attenuate with longer follow-up periods and that sex differences in tracking may vary across populations. Compared with previous findings, the particularly low CRF tracking observed among girls in our study (r = 0.25) compared with boys (r = 0.54) highlights the need for further research in Asian cohorts.

The lower stability of CRF tracking in girls may be partly attributable to their narrower distribution of VO₂peak values (Fig. 2). A more constrained range of CRF values may lead to greater rank-order fluctuations, amplifying apparent instability even when absolute changes are modest. In such contexts, minor shifts in performance or physiological state can disproportionately affect an individual’s relative standing. Another potential explanation lies in sex differences in the timing and tempo of growth spurts and sexual maturation. At the time of follow-up, many girls in our cohort were within the developmental window around peak height velocity—typically occurring between ages 12 and 13 in boys and between ages 10 and 11 in girls (28)—and some may already have experienced menarche. Such pubertal changes, accompanied by rapid somatic growth, can alter cardiorespiratory responses and may have contributed to the weaker stability of CRF observed among girls. In addition, although both boys and girls showed a similar absolute decline in MVPA (approximately 22 min·d⁻¹), girls started from a lower baseline and therefore reached substantially lower MVPA at follow-up. This lower absolute level of activity may have been insufficient to consistently support CRF development, further contributing to the weaker tracking of CRF among girls. Furthermore, while body fat percentage and PA tracked similarly in both sexes, CRF may be more sensitive to other modifiable factors. For instance, a recent study of Japanese schoolchildren aged 10–12 found that CRF, measured via the 20-m shuttle run, was associated with dietary quality only in girls (29), suggesting that nutritional changes during this period might have disproportionately affected girls’ CRF stability.

In terms of tracking of MVPA, previous studies that assessed objectively measured PA have generally found weaker tracking in girls than in boys (30,31), although some cohorts report similar tracking patterns for both sexes (32). Sasayama et al. (17) conducted a 3-yr follow-up study on PA among 9-yr-old Japanese schoolchildren using a uniaxial accelerometer and found significant tracking of step count and MVPA only in boys. In contrast, despite our study having a 1-yr longer follow-up period, we found tracking of all PA indicators in both boys and girls. These discrepancies may be partly attributed to variations in the accelerometer models used for measurement. Specifically, we utilized an accelerometer that estimates MET values with distinct formulas for different activities, such as walking/running and other actions (e.g., playing or performing household chores). This approach enhances MET estimation accuracy, whereas other accelerometers have been shown to underestimate MET values (33,34) as well as energy expenditure during free-living days (35). Taken togther, this finding suggests that girls’ PA during early school ages may exhibit greater stability than previously assumed.

We also examined the tracking of body fat percentage and found high stability, with rank correlation coefficients exceeding 0.6 for both boys and girls. These findings suggest relatively high stability in adiposity from early to late elementary school years, with greater consistency than observed for PA levels and CRF, both of which had rank correlation coefficients below 0.6. In the Young Hearts Project, which followed participants from ages 15 to 22 yr over a 7-yr period, moderate tracking of subcutaneous fat (measured by skinfold thickness) was observed in females, whereas tracking was poor in males (14). The reasons for these discrepancies between studies remain unclear; however, possible explanations include differences in the age groups studied, variations in methods for measuring obesity, or racial factors.

A notable strength of our study is the use of objective and highly reliable measurement methods, widely recognized as gold standards. Previous studies have often relied on self-administered or proxy-reported PA questionnaires (9,11), field-based assessments of CRF (e.g., shuttle run tests) (7,14,16,36), and body composition estimations derived from skinfold thickness (14,16). However, these methods have inherent limitations. For example, no existing PA questionnaire for youth possesses acceptable reliability and validity (15); field-based CRF assessments provide less precise estimations of maximal oxygen uptake (37); and skinfold thickness may not accurately reflect total body fat and is susceptible to measurement error (38). In contrast, we employed accelerometer-based assessments of PA, direct measurements of CRF via VO₂peak assessed with a gas analyzer, and precise evaluation of adiposity via DXA, ensuring greater accuracy, validity, and consistency across measurement techniques. By addressing these methodological shortcomings, our study provide a more robust assessment of the longitudinal tracking of these health indicators, contributing valuable insights into their stability over time.

Our study has several limitations. First, the sample size was relatively small, and participants were recruited from five elementary schools in the Kanto region, one of the most urbanized areas of Japan. However, the high follow-up rate (98.8%) enhances the internal validity of our findings, ensuring that the observed tracking patterns are less likely to be affected by selection bias. Future studies should aim to include a more diverse and nationally representative sample to confirm whether these findings generalize to other populations. Second, data collection occured before the onset of the COVID-19 pandemic, and it remains uncertain whether these results apply to children who have experienced pandemic-related lifestyle changes. Evidence suggests that the pandemic has led to declines in PA and fitness, as well as increases in adiposity at the population level (39,40). Given these changes, future research should examine whether tracking patterns in CRF, adiposity, and PA have been affected in postpandemic cohorts. Finally, while the present study focused on the tracking of each indicator separately, future studies may also consider cluster-based or joint trajectory analyses to examine how combinations of CRF, body fat percentage, and PA track over time. Such approaches could help identify distinct developmental profiles and high-risk subgroups, thereby providing more nuanced targets for intervention. Despite these limitations, a major strength of this study lies in its methodological rigor—employing objective, gold-standard measures for all three key health indicators: CRF, adiposity, and PA. This enhances the reliability of our findings and sets our study apart from previous tracking research that relied on self-reported or indirect assessments.

CONCLUSIONS

These findings highlight the need for early efforts prior to school entry, as well as targeted approaches during the early school years to prevent the consolidation of unfavorable profiles in fitness, adiposity, and PA. In addition, girls’ CRF showed greater variability, indicating that population-wide approaches throughout the school-age years may be especially important for supporting fitness development in this group.

This work was supported by a Grant-in-Aid for JSPS Fellows (JP13J07359 to T. A.) and Grants-in-Aid for Young Scientists B (JP24700759 to Y. H. and JP25750373 to M. W.) from the Japan Society for the Promotion of Science. One of the authors (H. W.) was supported by the Yamaha Motor Foundation for Sports.

The authors would like to thank the children who participated in this study and their parents. All authors declare that they have no conflicts of interest.

The datasets generated and analyzed during the current study are not publicly available due to legal and ethical restrictions under Japanese law, including the Act on the Protection of Personal Information (Act No. 57 of 30 May 2003, last amended in 2023). These regulations prohibit the public deposition of datasets containing personal information. Deidentified datasets may be shared with researchers affiliated with academic or research institutions; however, such data sharing will be considered on a case-by-case basis and is conditional on approval from the relevant institutional ethics commitee. Aggregate data or summary statistics may be made available from the corresponding author upon reasonable request.

The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. The results of the present study do not constitute endorsement by the American College of Sports Medicine.