The COVID-19 pandemic represented an unprecedented global public health disruption, substantially altering daily routines, movement behaviors, and health-related practices worldwide (1–3). Population-level evidence consistently documents marked reductions in physical activity, increases in sedentary behavior, and widespread interruptions to structured exercise opportunities during periods of lockdown and social restriction, particularly in East Asia, where school closures and mobility controls were often prolonged (4–6). Adolescents constitute a population of particular public health concern, as this developmental stage is characterized by rapid physical growth, consolidation of behavioral habits, and heightened sensitivity to environmental constraints (7, 8). Importantly, disruptions occurring during adolescence may exert lasting influences on health trajectories beyond the immediate pandemic period (9) and into young adulthood. Against this background, examining phase-specific changes in physical fitness among high school-aged (upper-secondary) adolescents aged 15–18 years provides critical insight into the population-level health impact of pandemic-related restrictions and informs strategies for post-pandemic recovery in this population.

Within this broader health context, physical fitness represents a particularly informative indicator of adolescent health. As an integrative construct encompassing cardiovascular, respiratory, musculoskeletal, and metabolic capacities, adolescent physical fitness reflects both functional health status and longer-term health prospects across the life course (10). Unlike many self-reported health behaviors, physical fitness can be objectively assessed through standardized school-based testing, making it particularly well-suited for population surveillance and policy-relevant evaluation and monitoring (11). In China, large-scale fitness monitoring embedded in the Chinese National Survey on Students' Constitution and Health (CNSSCH) has generated near-census data, providing a unique opportunity to examine long-term shifts in youth growth and physical fitness patterns (12, 13). Analyses of successive national surveys have revealed substantial changes in anthropometric indicators since the mid-1990s, accompanied by declines in multiple fitness components, with endurance- and respiratory-related outcomes showing particular vulnerability (11, 13).

Accumulating evidence suggests that the COVID-19 pandemic was associated with unfavorable changes in youth physical fitness, including increases in body mass index and deterioration in cardiorespiratory fitness and flexibility (14–16). However, two important gaps remain in the literature. First, most studies focus on pre-and during contrasts and lack a clearly defined post-pandemic phase, limiting insights into recovery dynamics. Second, adolescents are often treated as a homogeneous population, with limited attention paid to structural heterogeneity within the education system. In China, these limitations are particularly significant. Upper-secondary education is organized through a strong academic–vocational tracking system following the high school entrance examination (Zhongkao), which is associated with systematic differences in institutional resources, curriculum structure, and daily activity environments (17, 18). Beyond academic outcomes, emerging evidence suggests that vocational-track students differ from their general academic peers regarding health-related behaviors and opportunities for physical activity (16, 19, 20). These structural differences raise the possibility that pandemic-related disruptions did not affect all adolescents uniformly, but instead interacted with educational stratification to shape divergent patterns of fitness change and recovery. Accordingly, the present study aimed to characterize phase-specific changes in anthropometric indicators and multiple domains of physical fitness among Chinese upper-secondary students before, during, and after the COVID-19 pandemic, and to examine whether these patterns differed systematically between general academic and vocational education tracks within the same surveillance system. Leveraging large-scale school-based fitness surveillance data collected from 2016 to 2024, this analysis provides a structured assessment of overall population-level shifts and school-type-specific recovery dynamics following pandemic-related disruption.

Using large-scale school-based fitness surveillance data collected before, during, and after the COVID-19 pandemic, this study provides a phase-based assessment of changes in anthropometric and physical fitness outcomes among Chinese upper-secondary school students, with a particular focus on differences between general academic and vocational education tracks within the same surveillance system. Across the overall sample, significant phase-related variation was observed for all measured outcomes, indicating that the pandemic period coincided with measurable shifts in population-level anthropometric and physical fitness patterns. However, these changes were not uniform across the educational tracks. Statistically significant interaction effects between the pandemic phase and school type were identified across outcomes, suggesting that students attending general academic and vocational schools exhibited distinct phase-associated population-level patterns across the pandemic and post-pandemic periods, with differences in magnitude or timing. Although the Phase and School Type interactions were statistically significant, their effect sizes were generally small; therefore, practical significance should be interpreted cautiously.

First, consistent with the majority of previous studies examining the impact of the COVID-19 pandemic on youth physical fitness (16, 28, 29), our results indicate that anthropometric indicators (height, body mass, and BMI) and performance measures related to explosive power and endurance were affected among Chinese high school students aged approximately 15–18 years. However, a closer examination of phase-wise patterns suggests that the observed increases in body mass and BMI among students in regular schools should be interpreted cautiously and may partly reflect the long-term secular upward trajectory. Instead, these increases appear to align with the long-term upward trajectory of anthropometric indicators that have been consistently reported among Chinese adolescents over the past decades (13, 30). From this perspective, the pandemic did not substantially disrupt the pre-existing secular trends in body mass and BMI within the RS population. In contrast, mean height among vocational-school students did not show a clear increase from the pre-pandemic to the during-pandemic phase in our repeated cross-sectional samples. This divergence suggests that vocational-school students may have been more affected during the pandemic period. One possible explanation is unmeasured differences in activity opportunities and routines and other lifestyle factors during the pandemic; however, these mechanisms cannot be tested in our data due to the lack of physical activity and contextual measures. Taken together, these findings imply that while overall anthropometric growth trends among Chinese adolescents may be relatively resilient to large-scale societal disruptions, such resilience is not uniformly distributed across school types, with students in vocational education appearing more sensitive to external shocks, such as the COVID-19 pandemic.

Regarding explosive power, phase-wise comparisons indicated that the COVID-19 pandemic was associated with changes in adolescents' motor performance, consistent with previous reports of pandemic-related fluctuations in youth physical fitness (31–33). Standing long jump performance improved across phases in both school types, suggesting no sustained deterioration during the pandemic period in this outcome, while persistent between-track differences remained substantial. This pattern suggests that explosive performance did not show sustained deterioration across phases and may recover as activity opportunities normalize; however, the role of structured physical education and habitual physical activity cannot be directly evaluated in our data (34, 35). Between-school comparisons further highlight the persistent influence of school types. Across all phases, RS students consistently outperformed their VS counterparts in the standing long jump, with large and highly significant differences observed before, during, and after the pandemic (Supplementary Table S3). The stability of these gaps implies that structural factors associated with the educational context may exert a stronger and more enduring influence on explosive power development than short-term pandemic-related disruptions, in line with previous evidence linking school type to systematic differences in adolescent physical fitness (36). Trajectory patterns provide additional insight into this disparity: RS students exhibited a generally increasing trend in long jump performance across phases, indicating relative robustness despite pandemic-related disruption, whereas VS students exhibited a flatter improvement slope (smaller phase-to-phase gains) relative to RS, consistent with the statistically significant but small Phase × SchoolType interaction. Together, these divergent trajectories suggest that explosive power development among VS students may be more vulnerable to external disturbances, consistent with earlier studies highlighting fitness disadvantages and activity constraints among students in vocational education settings (36, 37).

In contrast to explosive power, endurance- and flexibility-related fitness outcomes showed more pronounced and heterogeneous phase-related changes during the pandemic. Importantly, these changes did not exhibit a uniform decline–recovery pattern.

For forced vital capacity, students in regular schools (RS) exhibited an increase from the pre-pandemic to the pandemic phase, followed by a partial decline thereafter, whereas vocational school (VS) students showed relatively stable values during the pandemic and a clearer increase in the post-pandemic phase (Figure 3A; Supplementary Table S2). A similar pattern was observed for endurance running: RS students demonstrated a rapid improvement after the pandemic, whereas VS students showed a more gradual increase relative to pre-pandemic levels (Figure 3B; Supplementary Table S2). Taken together, the forced vital capacity and endurance patterns are compatible with potential differences in unmeasured activity opportunities, routines, and PE exposure between school types. However, these mechanisms cannot be tested in our data because physical activity and PE exposure were not measured. Despite these school-type-specific nuances, endurance-related outcomes were characterized by substantial phase effects, indicating sensitivity to prolonged disruptions in habitual physical activity and structured physical education (31, 32, 38). This interpretation is consistent with the detraining literature showing that endurance adaptations may emerge with a delay and that recovery trajectories do not necessarily mirror the initial disruption (39, 40). Flexibility displayed a different, yet informative, pattern. Although RS students outperformed VS students in the sit-and-reach test before the pandemic, the between-school gap narrowed during and after the pandemic (Figure 3C; Supplementary Table S3). This convergence is compatible with evidence that COVID-19–related restrictions can disproportionately reduce training exposure among individuals with higher baseline activity levels, thereby temporarily compressing between-group differences in flexibility (37, 41).

This study has several limitations. First, phase-based comparisons were derived from repeated cross-sectional data rather than true longitudinal tracking. While this design allows for the identification of population-level temporal patterns and subgroup differences, it limits causal inferences regarding individual developmental pathways and adaptive responses to pandemic-related disruptions. Key behavioral and contextual factors, such as physical activity volume and intensity, training modalities, home-based exercise, and psychosocial stress, were not directly assessed. Second, the post-pandemic period in our dataset is limited to 2023–2024, which restricts inferences about longer-term recovery or adaptation beyond this window. Third, we did not explicitly model the clustered structure of students nested within schools; therefore, standard errors may be underestimated and p values should be interpreted cautiously. Fourth, the data were collected from two cities within a China-specific academic–vocational tracking system, which may limit generalizability to other regions or educational contexts. Finally, outcomes were derived from routine school-based assessments rather than research-grade clinical measurements, and measurement precision may vary across testing conditions. The absence of these measures may partly account for the heterogeneous responses observed across fitness domains and school types, including delayed changes in endurance performance and convergence patterns in flexibility. Despite these limitations, the consistent interaction between the pandemic phase and school type across outcomes indicates that the pandemic did not affect adolescent fitness uniformly but rather reshaped developmental trajectories in a school-type-specific manner. This finding underscores the importance of the educational context when interpreting large-scale environmental shocks. These limitations also indicate directions for future research.

Using large-scale school-based fitness surveillance data spanning the pre-pandemic, pandemic, and post-pandemic periods, this study provides a comprehensive overview of how adolescent anthropometric and physical fitness outcomes in China have varied across pandemic phases. The results indicate that pandemic-related disruptions were associated with measurable changes in multiple fitness domains, but these changes were domain-specific and shaped by the educational track. Anthropometric indicators showed relative stability at the population level, although students in vocational education appeared to be more sensitive to external disruptions. In contrast, explosive power was more consistently differentiated by school type than by pandemic phase, whereas endurance- and flexibility-related outcomes were more strongly influenced by temporal disruptions affecting physical activity opportunities. Taken together, these findings suggest that large-scale societal shocks do not affect adolescent physical fitness uniformly but instead reshape developmental trajectories in domain- and context-specific ways. This heterogeneity underscores the importance of fitness monitoring and intervention strategies that are sensitive to both the fitness domain and the educational context. Future research integrating multivariate and predictive approaches may further clarify the patterns of resilience and vulnerability and help inform more targeted health promotion efforts among adolescents.

Our findings suggest that post-pandemic school fitness strategies may benefit from a track-sensitive approach. First, educational track (academic vs. vocational) could be used as a stratification dimension when safeguarding minimum opportunities for physical education and structured activity during periods of disruption. Second, given that endurance and flexibility exhibited more pronounced phase-dependent changes, whereas explosive power showed more persistent between-track differences, recovery-oriented programming may prioritize endurance- and flexibility-focused activities delivered through short and frequent modules tailored to each track's time constraints and daily routines. Third, routine adolescent fitness surveillance should consider reporting and monitoring key outcomes stratified by educational track to better identify heterogeneous trajectories and inform targeted resource allocation aimed at reducing health inequalities. These implications should be interpreted cautiously because the present study is based on repeated cross-sectional data and lacks direct measures of physical activity and other contextual factors; thus, future intervention and policy evaluation studies are warranted. These results should be interpreted as average phase-associated differences rather than as definitive evidence of long-term consequences.