Funded Grants

Exploring how infectious diseases, beliefs, and behaviors interact on a social network

Infectious diseases, on the one hand, and ideas, norms and technologies, on the other, spread in distinct, but distinctly related, ways. Much attention has been given to investigating these separate processes (although much more remains to be learned), and substantial attention has been given to the parallels between them.

Less attention has been paid, however, to direct interactions between these two processes. In human societies, the spread of infectious diseases can have a profound impact on norms and beliefs, and thus on behaviors. For example, the HIV epidemic has dramatically changed attitudes about sexual relationships, and about safe sex, in many communities, with corresponding impacts on the spread of disease. Rising and falling levels of rabies incidence in sub-Saharan Africa trigger changes in governmental and societal behavior which are likely important to the dynamics of the disease. Flu epidemics (and, more recently, the SARS epidemic) have been correlated with reductions in interactions between individuals, and an increased use of face masks. Population responses to disease are not necessarily always helpful; for example, people fleeing from infectious diseases may be unwittingly helping to spread them. Understanding the interdependence between transmission of ideas and transmission of infectious diseases could point the way to more effective public-health strategies.

The proposed project will combine mathematical modeling, data analysis, and sociological research to develop a basic and applied theory of the interactions between an infectious disease, and related beliefs and behaviors, spreading in a population. We will combine general and specific approaches, both constructing abstract models, and focusing on details of particular cases, in particular the case of male circumcision as an intervention against HIV.

The WHO recently concluded that the research evidence in favor of male circumcision as a protective measure is "compelling", while simultaneously warning that it is crucial to ensure that public-health messages about partial protection and safe sex be attached to the circumcision initiative, and that the "gender implications" of directly protecting males but not females with this campaign must be addressed. We will use available HIV incidence and population-survey data to explore the interaction between the spread of HIV, the idea that male circumcision is protective against HIV, and the association of a wide variety of messages, norms, beliefs and attitudes with the idea of protective circumcision.

An important factor in real diseases is the fact that ideas themselves travel in groups, and these groupings are constantly changing. Public-health officials often strive, in effect, to package medical or public-health technologies with messages that they believe are helpful. Thus, people treated for HIV may be given information about safe sex, or those treated for cholera may be given information about safe drinking water. But ideas associate themselves in other ways, as well. So, for example, a community of individuals under treatment for a sexually transmitted disease might develop a norm of practicing less safe sex, based on the belief that the treatment protects their partners. Which attitudes, beliefs and behaviors spread along with medical and public health interventions will be affected by how infectious diseases spread, and may in turn have dramatic effects on the spread of those diseases.

It is also worth noting that information about the prevalence and severity of an infectious disease, like information about prevention and treatment, is largely mediated through cultural messages. In most cases, individuals do not form their opinions of disease risk primarily through direct experience. People may think that a disease is widespread when it is rare, or vice versa; for example, many people are surprised to learn that rabies causes more human deaths each year than dengue fever. A vivid case in point is that of influenza. The case that there is objectively more reason to worry about pandemic flu now than there was 10 years ago is mixed at best, but there is no question that society is more worried, and that many behaviors have changed as a result.

The proposed research will develop a theoretical framework for studying interactions between the spread of: infectious diseases; technologies which treat disease or prevent its transmission; and norms and behaviors affecting the spread of disease. We will address general questions about how such phenomena are expected to interact, and also specific case studies, using publicly available data, focusing in particular on the case of HIV and circumcision. Crucially, we will work to link our general insights to our specific findings, and vice versa.

At the broadest level, we will look at abstract models of interacting spread on a "network" -- a population linked by connections, representing potential transmission either of information or disease. Much has already been learned about how diseases spread on networks, and about how to abstract key, measurable parameters from the huge amount of detail needed to completely specify a network. Typically, it is at least necessary to measure the mean and variance of individuals' numbers of connections, as well as a measure of how clustered connections are. Increasingly, it is becoming evident that changes in the network are important for disease spread; thus it may also be necessary to have a parameter that measures the rate of this change.

We will work to solidify understanding of simple principles of how diseases spread on networks, and then to expand them to include interactions between disease spread and changes in behavior. Simple models have already explored feedbacks between disease prevalence and behavior changes, but as far as we know this has been done mostly in a phenomenological fashion, not explicitly considering the spread of messages, and mostly without taking network structure into account.

At a more concrete level, we will investigate and characterize available data from Africa about HIV incidence, traditional and changing patterns of male circumcision, and surveys about sexual behavior, attitudes and beliefs. We will investigate what factors determine how well messages (both accurate and inaccurate) spread, and how effective these messages are at changing behavior. Unfortunately, most existing sexual behavior surveys provide limited ability to separate attitudes from behavior. We will therefore supplement available survey data with our own, relatively small-scale surveys in Africa to begin to disentangle issues that cannot be resolved from the broader surveys.

We will work to bridge the abstract and the concrete, both by using insights from modeling to guide data analysis (and, eventually, data collection), and by using results from data analysis to construct more specific models, which we will analyze in parallel with the simpler basic models. For example, we will expand our basic model to account for the fact that the sexual-contact network on which HIV spreads is different from the communication network on which information spreads (although they encompass the same group of people). We will further investigate the effects of "directional" interactions -- i.e., a doctor or journalist may transmit information to many people, but receive information from only a few.

Finally, we will work to synthesize our findings into specific hypotheses and suggestions about public-health messages: what sorts of messages, and what methods of transmission, will be most effective at convincing people to adopt behaviors that will protect themselves and others against the spread of infectious diseases?

In a practical sense, control of infectious diseases depends to a large extent on effectively communicating messages that change people's behavior, both in terms of reducing transmission, and seeking appropriate prevention and treatment. The proposed investigation of how messages, behaviors and infectious diseases interact promises important findings about network dynamics, as well as applications to public health in general, and to controlling HIV in particular.