One of the first models aimed at understanding the complex nature of disease causality was the epidemiological triad which modelled disease occurrence as the result of the interactions between characteristics of pathogens, hosts, and the environment. These models have been expanded in recognition that infectious and non-infectious disease occurrence is the result of the complex interactions between many dynamic socio-economic, behavioural, ecological, environmental, pathogen and other factors (by factors we mean any characteristics or descriptors that can be measured or counted). Identifying these factors, often called risk factors or determinants (here we use the terms interchangeably), has been a primary focus of epidemiology since its emergence (Galea et al„ 2010). Risk factors are most commonly identified using studies that measure the association (positive or negative) between potential risk factors and the occurrence of cases of disease (Joffe et al., 2012). The motivation for identifying these risk factors is to use them to develop strategies and interventions aimed at nullifying (where the effect is negative) or promoting their effect (where the effective is protective or health promoting), thereby improving the health of individuals, groups (e.g. farms), or populations. Risk factor models have been integrated into the practice of human and veterinary medicine and can be found in the epidemiology sections of medical and veterinary textbooks.

Since disease occurrences are known to be the result of the interaction between many factors, it follows that there will be some factors that increase the occurrence of disease, and others that decrease it. Both positive and negative risk factors are in constant operation at varying degrees and time scales. The constant change in risk factors explains much of the random variation that is seen in time series of counts of cases of endemic disease that is common to all surveillance data. Large shifts in the rate of production of endemic disease can result from changes in any single risk factor or a combination of multiple risk factors. Figure 7.4 is an oversimplified illustration of the way in which risk factors influence the occurrence of disease cases that are counted in a surveillance system.

The surveillance models presented up to this point are based on the currently accepted paradigm that the information created by surveillance activities is based on cases of disease or infection, in humans and other species. Case-based surveillance is limited in its timeliness to creating information only after the onset of a disease event in a population. There is a strong motivation for producing information about changing disease risks earlier in the chain of events leading up to a

The effect of determinants of disease case occurrence as seen via a disease surveillance system

FIGURE 7.4 The effect of determinants of disease case occurrence as seen via a disease surveillance system.

significant health event such as rapidly spreading epidemic. Basing decisions on out-of-date information can result in ineffective decisions that can have devastating consequences on humans and animals (Kitching et al., 2006). Earlier information about an impending change in the disease status of a population would be beneficial, as it would provide more time to design interventions to prevent or mitigate the harm caused by the disease. Overcoming the time constraint will require rethinking the way we do surveillance. In addition to producing information about current disease risks, surveillance could also produce information predicting impending change in disease risks or population vulnerability. For example, a well-known risk factor for an epidemic is the proportion of the population that has resistance to infection with the pathogen causing the epidemic. When the proportion drops below the level of herd immunity in the population, the introduction of the pathogen into the population can result in a sustained epidemic. The recent measles outbreaks in North America and Europe occurred because communities of parents decided to withhold vaccination of their children, to the point where community-based epidemics were sustained. The risk factor “reduced proportion of resistant children in the community” is commonly used to model epidemic progression. It is very simple and easy to model but fails to convey the complexity of the situation. A large number of interconnected risk factors for parents failing to vaccinate their children have been identified and grouped into the 5A’s taxonomy (Access, Affordability. Awareness, Acceptance, and Activation) (Bell et al., 2020). Understanding the complexity of these interconnected risk factors is difficult. However, a surveillance system that recorded the number of people refusing measles vaccination for their children would have provided some warning of which communities were reaching a point where an epidemic was imminent. This information would provide additional time for the development of interventions such as campaigns to inform people of the risks of not vaccinating their children.

Predicting increased disease risk for known endemic diseases is a relatively common practice in humans, animals, and plant health, and ecological sciences

(Funk et al., 2019; Kleczkowski et al., 2019; Thompson and Brooks-Pollock, 2019; Lee et al., 2017). It has been especially promoted for predicting changing vector-born disease risk because vector populations are affected by weather (temperature, precipitation, and humidity) (Morin et al., 2018; Lee et al., 2017). It has also been reported for predicting changing risk for diseases (e.g. Ebola or obesity) that are affected by changing social behaviours (Funk et al., 2019; Scarpino and Petri, 2019; Galea et al.. 2010). Epidemic prediction has been likened to weather forecasting (Morin et al., 2018; Moran et al., 2016). We can speculate that epidemic prediction will improve over time as methods and the amount and quality of data improve, similar to the way that weather forecasting has improved over time. Because of the current state of the art of epidemic prediction methods and the associated uncertainties with epidemic prediction, this type of surveillance is not likely to replace conventional surveillance in the immediate future. However, it could be used to provide additional information to supplement conventional surveillance information.

Risk factor surveillance has been proposed for animal health surveillance (Rich et al., 2013), but is not widely practiced. It is more common in public health surveillance. The Behavioural Risk Factor Surveillance System has been used since 1981 in the United States to monitor a variety of risk factors for human disease, including immunization, preventative testing, physical activity, chronic conditions, mental health, obesity, tobacco use, alcohol and substance abuse, and health risks associated with sexual behaviour, injury, and violence (Pierannunzi et al., 2013). It has successfully predicted epidemics (for example diabetes) and has been used by most of the U.S. states to design and implement disease control interventions (including legislation) aimed at reducing disease risk predicted by the surveillance system (Mokdad, 2009).

Both case-based and risk factor surveillance can be practiced at the same time. Surveillance targeting disease cases could be continued as usual, producing information that is used to make decisions about the current situation in the population. Risk factor surveillance could be added and used to produce information predicting future changes in disease risk (Figure 7.5). This information would be used to make decisions about implementing prevention activities aimed at reducing the future risk of disease.

Monitoring several individual risk factors as independent variables is not likely to provide a complete understanding of the processes affecting disease production (Joffe et al., 2012). There are at least two reasons for this. The first is that many risk factors do not act independently. They interact with each other and with other risk factors in ways that make it difficult to understand their behaviour without understanding the behaviour of the complex system in which the species under surveillance resides (Joffe et al., 2012; Galea et al., 2010). The second is that monitoring risk factors independently ignores interconnectedness of the health of humans, animals, and the environment. Understanding the connections between many species and the environment they reside in, as a system, will be essential for understanding how changes in risk factors for disease in one species may affect the health of another species, or how interventions

Adaptive management model of surveillance that monitors disease cases and determinants of disease

FIGURE 7.5 Adaptive management model of surveillance that monitors disease cases and determinants of disease.

aimed at reducing disease in one species may have beneficial or harmful effects on another.

Even though there is a toolbox of methods that have been developed in the field of complexity science that could be used in One Health, the use of complex system methods has not been widely reported in One Health research (see Chapter 13 for more on complexity in On Health). There are studies that have identified the relationships between risk factors and disease in single species using complexity science methods, and complexity methods in combination with participatory approaches have been used to select the most appropriate disease control activities and to develop health policy (Mumba et al„ 2017). At the time of writing, we found no studies that used complexity science methods to select targets for One Health surveillance. The reasons for this are unknown. It may be that complexity science is not well known among One Health researchers and practitioners, or it may be that additional basic research is needed to support the use of complexity science in One Health. It may also relate to the relatively narrow scope of One Health that is currently being implemented. Most reported applications of One Health focus on zoonotic diseases. Relationships between species are included only for those species that act as reservoirs or vectors for human disease, and only risk factors that affect disease in people are considered. This narrow view is a one-way causal model pointing towards people. These are not full models, because they ignore the effects of interventions to protect human health on the health of non-human species, ecosystems, and the environment. Furthermore, they do not consider how the health of animals and the environment effects the health of humans. Complex system approaches have the potential to aid our understanding of all these interactions and to help to develop interventions that are more likely to be considerate of the health of non-human species and the environment. However, complex system approaches will not likely be seen as necessary until organizations responsible for health fully embrace One Health and develop programs that accept responsibility for the health of all species and the environment.

< Prev   CONTENTS   Source   Next >