Types of health-related behaviours: a cluster analysis of the Luxembourgish HBSC data

Background: Although it is known that health behaviours, socio-demographic variables and outcomes correlate, it is rarely investigated if there are typical patterns of these variables among the research subjects.

Objectives: To find out whether the students can be divided into distinct groups based on their health behaviour and whether these groups differ in other ways (outcomes and socio-demographics).

Method: In step 1, a hierarchical cluster analysis was carried out to determine the number of groups and to identify the cluster centres. In step 2, this information was entered as the initial values of a cluster centre analysis. In step 3, the clusters were characterised using additional variables.

Results: The 8065 students surveyed could be divided into 5 distinct groups based on their data on smoking, drinking, soft drinks, exercising, fighting and bullying, with cluster 1 and cluster 5 representing the strongest contrast. Cluster 1 comprises students whose health behaviour is generally positive. It is the largest cluster with 49.5% of students. Cluster 5 comprises students whose behaviour is consistently negative. It is the smallest cluster with 7.1% of students. Students in cluster 2 are close to average on many variables, but their dental health is problematic because they frequently consume soft drinks and rarely brush their teeth. Students in cluster 3 are physically inactive, their mental health is poor, but they are also rarely injured. The students in cluster 4 stand out because of their aggressive behaviour.

Conclusion: With the help of cluster analysis, it is possible to categorise the students into a small number of groups based on their health behaviour. These groups are coherent in terms of health behaviour, many outcome variables and socio-demographic variables.