Investigators derived three distinct trajectories from two-year changes in multiple anthropometric measures. They applied k-means clustering to changes in BMI, waist circumference, hip circumference, waist-to-hip ratio, waist-to-height ratio, and body weight. The cohort split into Stable (48.0%), Adiposity-Increasing (31.0%), and Adiposity-Decreasing (21.0%) clusters. The cluster labels reflected aggregated directionality across the six indicators rather than a single metric. The study team observed that trajectories captured heterogeneous patterns of change that would be missed by analyzing BMI alone.
Baseline body composition differed across clusters. The authors reported that baseline body fat mass was significantly associated with eventual cluster membership. Participants who later joined non-stable clusters tended to present higher fat mass at baseline compared with those in the Stable cluster. Changes in body composition over the follow-up period—specifically increases in body fat mass and changes in skeletal muscle mass—showed strong associations with cluster assignment. The data indicated that trajectories categorized as adiposity-increasing were linked to meaningful fat accumulation and relative loss, or smaller gains, in muscle mass during follow-up.
The study team compared baseline lifestyle measures and their changes across clusters. Worsening sleep quality, operationalized as an increase of at least 2.5 points in the Pittsburgh Sleep Quality Index (PSQI), was significantly associated with membership in the Adiposity-Increasing cluster. Reduced physical activity, measured by declines on the Global Physical Activity Questionnaire (GPAQ), also correlated with the adiposity-increasing trajectory. The authors noted that these lifestyle shifts—sleep deterioration and decreased activity—co-occurred with fat gain and muscle change, suggesting linked behavioral and compositional processes rather than isolated metric fluctuation.
The analytic sample comprised 1,404 participants from the Korean Medicine Daejeon Citizen Cohort (KDCC) who completed baseline and the first follow-up visit. Mean age in the analytic sample was 44.74 years (SD 6.67); 30.0% were male and 70.0% were female. Follow-up assessments took place between June 2020 and March 2022, a period that overlapped with the COVID-19 pandemic in Korea. The authors used k-means clustering on six change variables to identify trajectory groups, then compared baseline body-shape indicators, body composition, and lifestyle factors across clusters. Multinomial logistic regression was used to identify variables associated with cluster membership. Analyses were performed on a complete-case dataset of participants without missing main-variable data; the authors compared included versus excluded participants and reported no significant differences on several anthropometric measures, though the analytic sample was slightly older with a marginally larger hip circumference.
The authors acknowledged limitations related to the analytic approach and data availability. Analyses used a complete-case dataset; selection bias cannot be ruled out if participants with missing data differed systematically from those included. The first follow-up visit occurred during the COVID-19 pandemic, and the authors reported that follow-up assessments followed standardized protocols under national infection-control guidelines, but they did not quantify pandemic-related behavioral effects beyond what the measured lifestyle variables captured. The source did not report intervention or dosing details because this was an observational cohort. The datasets analyzed are not publicly available due to confidentiality and ethical restrictions; de-identified minimal datasets may be provided to qualified researchers through an institutional request process managed by the Korean Medicine Data Center.
The study team concluded that integrated assessment of body fat mass, skeletal muscle mass, composite body-shape indicators (including WHtR), and modifiable lifestyle factors such as sleep quality and physical activity provides more nuanced characterization of adiposity-related trajectories than single metrics alone. Specific unanswered questions documented by the authors include potential selection bias from complete-case analysis and the need for longer follow-up to determine how early trajectory membership predicts clinical outcomes; the source noted these limitations and did not report subsequent regulatory milestones or interventional trials.