Empirically distinct movement-behaviour profiles showed only partial correspondence with broader lifestyle characteristics, indicating that physical activity level alone should not be treated as a proxy for overall adolescent lifestyle quality.
Key Findings
Results
K-means clustering of Polish adolescents produced two physical activity-sedentary behaviour profiles: a Higher Physical Activity profile and a Lower Physical Activity profile.
The final classification included 1070 adolescents (55.4%) in the Higher Physical Activity profile and 862 (44.6%) in the Lower Physical Activity profile.
Clustering was based on four Z-standardised variables: sedentary time, walking, moderate-intensity physical activity, and vigorous-intensity physical activity.
Solutions from k = 2 to k = 6 were evaluated using the silhouette coefficient, Calinski-Harabasz index, Davies-Bouldin index, and Adjusted Rand Index (ARI).
The study sample consisted of 3307 Polish adolescents aged 15–17 years, of whom 1932 had complete and analytically eligible data for profile derivation.
The study used a cross-sectional, school-based design using the International Physical Activity Questionnaire-Long Form.
Results
Cluster separation between the two physical activity profiles was limited, and cluster membership was strongly influenced by variable scaling.
The silhouette coefficient for the two-cluster solution was 0.294, indicating limited separation between the two groups.
Cluster assignments were almost identical across repeated k-means runs with different random starting centroids (ARI = 0.999), reflecting low dependence on initial centroid configuration rather than robustness to changes in sample composition.
The correspondence between classifications obtained from unstandardised and Z-standardised variables was minimal (ARI = 0.008), demonstrating a strong influence of scaling on cluster membership.
Results
The Higher Physical Activity profile was associated with both higher pro-healthy and higher non-healthy diet index scores compared to the Lower Physical Activity profile.
The Higher Physical Activity profile had higher Pro-Healthy Diet Index scores (r = 0.100, p < 0.001).
The Higher Physical Activity profile also had higher Non-Healthy Diet Index scores (r = 0.090, p < 0.001).
The Higher Physical Activity profile showed slightly higher energy-drink intake (r = 0.080, p = 0.001).
More frequent out-of-home meal consumption was observed in the Higher Physical Activity profile (V = 0.079, p = 0.002).
Effect sizes for all dietary co-occurring behaviours were generally small.
Results
The two physical activity profiles showed no clear differences in school-day sleep duration, television/computer use, or internet-related risk.
School-day sleep duration did not differ meaningfully between the Higher and Lower Physical Activity profiles.
Television/computer use showed no clear differences between profiles.
Internet-related risk did not differ between profiles.
Weekend sleep duration did differ between profiles (V = 0.120, p = 0.001), though effect size was small.
The Lower Physical Activity profile did not show a consistently adverse sleep or digital-behaviour pattern.
Conclusions
Higher physical activity coexisted with both more and less favourable dietary characteristics, demonstrating that physical activity level alone is not a proxy for overall adolescent lifestyle quality.
Higher physical activity was associated simultaneously with higher Pro-Healthy Diet Index (r = 0.100, p < 0.001) and higher Non-Healthy Diet Index scores (r = 0.090, p < 0.001).
The authors conclude that 'physical activity level alone should not be treated as a proxy for overall adolescent lifestyle quality.'
Effect sizes for co-occurring behaviours were generally small across all domains examined.
The findings are described as 'hypothesis-generating rather than as evidence for fixed lifestyle types or profile-specific interventions' given the exploratory, cross-sectional design and method-dependent nature of the profiles.
Methods
A substantial proportion of the original sample was excluded from profile derivation due to incomplete or analytically ineligible IPAQ-LF data.
Of 3307 Polish adolescents initially enrolled, only 1932 participants had complete and analytically eligible data and were included in profile derivation.
This represents a retention of approximately 58.4% of the original sample for the clustering analysis.
Exclusion was based on final processing of the International Physical Activity Questionnaire-Long Form (IPAQ-LF).
What This Means
This research suggests that Polish teenagers who are more physically active do not necessarily lead healthier lifestyles overall compared to their less active peers. The study grouped 1,932 adolescents aged 15–17 into two clusters—a Higher Physical Activity group and a Lower Physical Activity group—based on their reported levels of walking, moderate exercise, vigorous exercise, and time spent sitting. Despite the differences in movement behaviour, the two groups looked surprisingly similar in areas like how long they slept on school nights, how much time they spent watching TV or using computers, and their risk behaviours related to internet use.
One of the most striking findings was that the more physically active teenagers actually scored higher on both a 'pro-healthy' diet measure and a 'non-healthy' diet measure simultaneously. They also consumed slightly more energy drinks and ate out more often. This suggests that being more active may go hand-in-hand with eating more in general—both nutritious and less nutritious foods—rather than simply eating better. The differences found between the groups were statistically significant but small in magnitude across all lifestyle areas examined.
This research suggests that researchers and health professionals should be cautious about assuming that a physically active teenager is also sleeping well, eating healthily, or avoiding problematic screen time. The study also highlights important technical limitations: the way data are processed and variables are scaled can substantially change who ends up in which cluster, meaning these profile groupings are not fixed or definitive categories. The authors emphasise that these results should be seen as generating new questions for future research rather than as a basis for designing specific health programmes targeting particular adolescent 'types.'
Check Your Own Numbers
Upload your bloodwork. We'll cross-reference your results against this study and 4,700 others.
Wojciechowska M, Żurawski A, Cieśla E, Zmyślna A, Kozieł D. (2026). Physical Activity-Sedentary Behaviour Profiles and Co-Occurring Lifestyle Behaviours Among Adolescents: A Person-Oriented Analysis.. Nutrients. https://doi.org/10.3390/nu18172927