Over the years many theories have been developed to explain why people engage in certain unhealthy behaviours and how these spread in populations [Reference Glanz and Bishop1]. These theories share the idea that behaviour is in some way influenced by social contacts. Yet, empirical studies of unhealthy behaviours generally investigate behavioural change processes from an individual perspective and until recently paid little attention to social environmental influences on behaviour. An intriguing exception is work from Christakis & Fowler [Reference Christakis and Fowler2, Reference Christakis and Fowler3], who showed that both smoking and obesity spread from person-to-person, that the type of contact matters, and that groups can be distinguished within a social network. This lead to the idea that unhealthy behaviour is contagious and that it spreads in populations like an infectious disease. This has been suggested before conceptually [Reference Gladwell4], but there is a need to further operationalize this concept in ways that can be tested scientifically. Basically, adopting unhealthy behaviour is analogous to acquiring, say, influenza from a family member. Moreover, influenza tends to cluster in schools, which can also be observed for unhealthy behaviours. Tuberculosis and leprosy are even better examples of infectious diseases that show similarities with unhealthy behaviours: they cluster in households and communities, only a minority of those exposed eventually develop disease, and clinical signs may not be visible until several years after infection. Although there is considerable evidence that the spread of behaviours is explained by social influence, it is also true that similarity of behaviours observed in social networks may to some extent be the result of the tendency of people to select others with similar behaviours (homophily). Yet, it is difficult to disentangle homophily from social influence [Reference VanderWeele5, Reference Shalizi and Thomas6].

Apart from contagiousness, other concepts and underlying mechanisms can be identified that are comparable for unhealthy behaviour and infectious diseases. First, an important concept in infectious diseases is heterogeneity, which can concern individual susceptibility to infection, infectiousness of a patient, and mixing patterns in the population [Reference Anderson and May7]. Heterogeneity in susceptibility resembles variation in adopting unhealthy behaviours, such as stated in the theory of Diffusion of Innovations [Reference Rogers8], which indicates that some people are more susceptible to adopt a behaviour than others. The rate of adoption further depends on the number of people in the social network that engage in a certain behaviour. Each individual has his/her own adoption threshold. For instance, some people are more self-efficacious than others, resulting in different levels of resilience. Heterogeneity in infectiousness can be compared with variation in social influence: position within networks, closeness of relationships, and number of contacts may explain why some people are more influential than others [Reference Valente9]. Heterogeneity concerning mixing patterns reflects that individuals tend to cluster within populations, e.g. according to age group or socioeconomic position. Second, a mechanism strongly related to heterogeneity is the presence of so-called super-spreaders. These are individuals that accelerate dissemination of an infection in a population, because of a prominent role in the contact network (i.e. many contacts) and/or high infectiousness. This greatly resembles opinion leaders or peer-role models, which are early adopters and can easily spread behaviours to others, due to their persuasiveness and high number of social contacts [Reference Rogers8]. Third, vaccination is another concept that both fields share. Vaccination induces immunity, reduces the number of susceptible people, and reduces the risk of infectious diseases. In a similar way, social inoculation provides resistance to unhealthy behaviour by emphasizing refusal skills, and thus reducing the risk of adopting a behaviour [Reference Evans, Raines and Hanselka10]. Although vaccination and social inoculation are not exactly the same, they serve the same purpose. A fourth comparable mechanism is the influence of physical environmental factors. The physical environment promotes or discourages the spread of infections and behaviours in social networks through, e.g. climate and availability of fast-food, respectively. However, the availability of fast-food can also trigger a person to start unhealthy eating without any influence from the social environment.

The fact that the principles of infectious diseases and unhealthy behaviour show a remarkable resemblance challenges us to study unhealthy behaviour as an infectious disease. Infectious disease epidemiology has been studied for decades using sophisticated methods, in particular mathematical modelling, to analyse spread within populations, to predict the course of epidemics, and to evaluate interventions. As a major innovative step, Hill et al. [Reference Hill11] recently modelled the obesity epidemic as an infectious disease, using data from the Framingham Heart Study cohort [Reference Dawber12]. The model mimics transitions between two compartments, i.e. susceptible (non-obese) and infected (obese) individuals. It also allows for possible spontaneous infections not resulting from contacts. The study concludes that the obesity epidemic is driven by both contagious and spontaneous infection and will stabilize at 42% of the population being obese within the next 50 years. However, as the authors indicate in an earlier paper [Reference Hill13], the proposed compartmental model is rather simplistic and does not take into account possible heterogeneities.

A major enhancement would be to go from compartmental modelling to a more comprehensive and realistic approach. Individual-based modelling is particularly useful to realistically model networks and individual heterogeneities. It simulates life-histories of individuals and specific interactions between individuals over time. Events, such as birth, death, relationship formation, transfer between social/risk groups, and acquisition of infection (behaviour), are modelled through chance processes. Another advantage is that it is more suitable for analysing the impact of interventions aimed at certain groups, such as households or schools. Individual-based modelling has proven to be very useful for practical decision-making in infectious disease control, starting with the ONCHOSIM model for river blindness control in West Africa [Reference Plaisier14, Reference Remme15]. A more relevant model for sexually transmitted diseases, STDSIM, explicitly models individual contacts (sexual relationships) and formation of (sexual) networks [Reference Korenromp16]. Another recent example is the SIMCOLEP model for leprosy [Reference Fischer17], in which the formation of and movement between households is modelled.

The application of individual-based modelling holds great promise to address some of the major questions in public health regarding health-related behaviours. Why are some people more open for unhealthy behaviours than other people? What are major determinants causing the adoption of certain behaviours? How can we best prevent unhealthy behaviour or promote behavioural change? These questions can only be answered adequately when taking into account the social context in which behaviours take place. Until now behavioural studies have mainly focused on the individual in a static environment. The introduction of infectious disease methodology and in particular individual-based modelling would be a great addition, because it takes into account the dynamics of individual interactions and the social environment at large. This may result in new or revised interventions and policies. For instance, community interventions for behavioural change that only show small individual effects may eventually have substantial indirect public health effects. In contrast, some interventions with large individual effects may ultimately have a small impact on the population, due to a limited reach. Individual-based modelling in particular allows translating individual effects to population impact. Moreover, infectious disease modelling provides useful key concepts, such as the basic reproduction number (R ₀), i.e. the average number of successful transmissions per infectious person in a fully susceptible population. An outbreak of, e.g. smoking in a non-smoking population will occur if R ₀ > 1, which indicates that each smoker will on average trigger at least one other individual to start smoking. The goal is to reduce R ₀ to below 1, to stop further spreading of smoking.

In conclusion, the spread of unhealthy behaviour shows marked similarities with infectious diseases, and hence embracing existing infectious disease methods is beneficial. A first attempt to apply infectious disease modelling for unhealthy behaviours has now been published, but there is substantial room for improvement by including the dynamics and heterogeneities of social networks. The field of research aimed at studying health-related behaviours and at developing interventions and policies to promote health behaviours may benefit substantially from further exploiting models for infectious diseases, in particular individual-based models.