Heath-related behaviors such as smoking, alcohol misuse, physical inactivity and unhealthy diets contribute significantly to chronic diseases [1]. Many studies report on interrelationships between some health-related behaviors such as physical activity with healthy eating habits [2], and smoking with eating habits [3]. The interrelationships between health-related behavior are considered to be multidimensional [4–6]. Roysamb et al. [7] suggested a multidimensional model consisting of three groups of behaviors, namely, “high action”, “addiction” and “protection” behaviors. Moreover, the Problem Behavior Theory supports the view that the relationships between problem behaviors are multidimensional in nature [8]. The multidimensional approach assumes that certain health-related behaviors tend to cluster in a number of different patterns among both adolescents and adults [9–12]. For example, Raitakari et al. [13] found that a poor diet, smoking, physical inactivity and excessive consumption of alcohol clustered in young adults, while Neumark-Sztainer et al. [14] found associations between different health-compromising behaviors, namely, unhealthy weight loss, substance abuse, suicide risk, delinquency, and sexual activity. In an extensive systematic review of studies published between 1995 and 2003 to identify the clustering of four health-related behaviors (smoking, alcohol abuse, safe sex and healthy nutrition) in adolescents, Wiefferink et al. [15] identified three patterns of clustering. The largest cluster was adolescents who ate healthily, were not smokers and who did not drink alcohol. The second cluster was adolescents who ate unhealthily, smoked and drank alcohol. The third cluster comprised adolescents who ate unhealthily but did not smoke or drink alcohol. Later, Van Nieuwenhuijzen et al. [9] identified two clusters of behaviors for younger adolescents aged 12–15 years, and three clusters for adolescents aged 16–18 years.

Clustering is important because the co-occurrence of multiple health-compromising behaviors is associated with increased risk of chronic diseases including certain cancers and cardiovascular diseases [16]. The increased risk is the result of accumulation and synergistic adverse effects of behaviors on health [17]. Moreover, behavioral patterns in adulthood are primarily shaped during the adolescence period [18]. Therefore, understanding how health-related behaviors relate to one another in adolescents has important implications throughout the life course [19].

Different types of behaviors encompass different aspects of adolescents’ lifestyle. Behaviors related to healthy eating, oral hygiene practices, physical activity, physical fighting, and smoking have a considerable immediate and longer term effect on the health of adolescents and are related to one another. For example, higher fruit intake is associated with increased physical activity [20] and with lower rates of smoking and alcohol consumption [21]. In terms of dietary behaviors, lower fruit intake goes together with higher consumption of sweets and soft drinks and saturated fat [22]. Hygiene behavior such as toothbrushing frequency, is linked to patterns of smoking [23]. Indeed, smoking is viewed as a “gateway behavior” to other risky behaviors like drug use and drinking alcohol [24]. The Problem Behavior Theory postulates that physical fighting is a reliable predictor of multiple risk behaviors such as carrying weapons, injury [25, 26], and substance abuse [27]. Despite these associations between different behaviors, research has generally focused on a limited number of behaviors at a time, with most studies looking at the clustering of two behaviors, thereby limiting understanding of the inter-relationships between different and diverse health-related behaviors among adolescents. Furthermore, these studies have employed basic statistical techniques that either assess only the associations between specific behaviors in a cluster or look at whether the prevalence of predetermined clusters of behaviors is higher than expected; these are correlation coefficients and observed/expected ratios, respectively. While useful, these techniques can only look at behavioral clusters that are predetermined, rather than explore whether the different behaviors form clusters according to theoretical expectations.