Identifying Clusters of Health Behaviors in a Japanese Working Population at Risk for Non-Communicable Diseases: A Latent Class Analysis of 12,168 Individuals

Noncommunicable diseases (NCDs) have become a significant global problem. Health behaviors are associated with NCDs, and characterizing populations using a public health approach can help provide specific interventions according to their characteristics. This study aims to examine the formation of clusters of health behavior combinations in the Japanese working population at risk of NCDs, taking into account the influences of age and gender, using latent class analysis. Participants were individuals at risk for NCDs but had not previously been diagnosed with any. Latent class analysis (LCA) was used to study clustering based on basic characteristics and health behaviors. All statistical analyses were conducted using R (Version 4.0.4) and the “poLCA” package (Version 1.6.0). This study included 12,168 participants. LCA compared models with one to six latent classes. The five-class model was determined to be the most appropriate based on Bayesian Information Criterion, Akaike Information Criterion, and G^2 values, as well as distinguishable cluster characteristics. Cluster 1: “having healthy lifestyles but disliking hospitals”; Cluster 2: “women with healthy lifestyle behaviors”; Cluster 3: “general population”; Cluster 4: “middle-aged group in need of lifestyle improvement”; Cluster 5: “a group receiving treatment for lifestyle-related diseases." This study reveals discernible health behavior patterns in a sample of the Japanese population using large real-world data, suggesting the effectiveness of distinct approaches when considering a population approach to public health. •Latent class analysis reveals distinct health behavior clusters in Japanese working population.•Five clusters identified, considering age, gender, and health behavior patterns.•Findings may inform tailored public health interventions for noncommunicable disease prevention.