Risk Analysis and System Dynamics
A System Dynamics Based Model
Giuseppe Noce - DVM PhD
Servizio Salute - Regione Marche
giuseppe.noce@regione.marche.it
In 2002, the European Union passed Regulation (EC) no 178 of 28 January 2002 based on its own health laws on Risk Analysis (RA), in accordance with the WTO agreement which obliges WTO members to remove barriers to trade unless there is a risk to human, animal or plant health. Barriers may only be implemented when risk is demonstrated through the RA (WTO, 1995; MacDiarmid e Pharo, 2003).
RA is a scientific tool to aid decision-making. RA helps decision-makers to choose between different options to mitigate risk. The RA is composed of four components: hazard identification, risk assessment, risk management and risk communication.
Hazard identification is the first step of RA. This is the step to identify the pathogenic agents which could be introduced and what effect they could have by answering the questions: What can go wrong ? or Which accidents can occur ? The second step is risk assessment, which consists of four phases: release assessment, exposure assessment, consequence assessment and risk estimation. The goal of this step is to identify and to define the consequences of the accidents identified in the previous step. The third step of RA is risk management and is subdivided into risk evaluation, option evaluation, implementation and monitoring and review, in which control measures, to reduce the risk to an acceptable level, are defined and implemented. Cost-effectiveness measures are more helpful when selecting from the several options useful to mitigate risk The last but not least phase is risk communication. In this phase, the risk and the opinions on hazards are gathered from the interested parties and the results of the risk assessment and the risk management are presented to decision-makers. The communication process begins at the start of RA, and continue through all the phases of RA. The fact that communication takes place within the RA process influences the acceptance of the risk management strategy (MacDiarmid e Pharo, 2003; OIE,2004).
To perform RA, a useful model is required to represent a real world system, which can help us to learn something new about the system being studied.
Several techniques are possible to represent reality. A paper plan or an type other physical model might help us to conduct experiments (Ford, 1999). Since the 1980s, diseases have not been viewed as a statistical pattern but rather as a dynamic of host-pathogen interactions; therefore the mathematical model approach is increasingly more useful (Thrusfield, 1995, Smith and coll, 2005). However, it's worth considering that as long ago as 1776, Bernoulli was already using a mathematical model to demonstrate that the variolation immunized and protected human from the pox (Bernoulli,1776).
In the 1960s, Forrester and his colleagues developed their initial ideas by applying concepts from feedback control theory to the study of industrial systems. Forrester's idea is called System Dynamics (SD), a systems thinking for building models of complex situations which can be better understood and managed. SD serves to create a model to study its dynamics and behaviour. SD is concerned with improving and controlling problematic behaviour of systems to facilitate an understanding of the relationship between the behaviour of the system itself over time and through space. In other words, SD models are useful to help understand the patterns of systems and, at the same time, to identify the condition that causes the patterns of systems. This knowledge can indicate to the decision-makers what prescriptions will work or not (Caulfiled and Maj, 2002). Therefore the SD models describe the change throughout the whole system. The system's overall framework is therefore easily demonstrated by SD models (Bagni and coll., 2002).
SD is widespread and involves many health, economic and social sectors (Forrester, 1961; Ford, 1999).
Health risks, such as diseases, are dynamic and complex systems which change in time and space. A model that is able to represent this dynamic and complex system and to predict the disease patterns, is a valid tool in decision-making processes for decision-makers. A model like this supports the identification of the mechanisms and the factors that spread diseases and allows the relative rapid selection of different scenarios. This model becomes essential to allow qualitative and quantitative analysis of defence strategy and its consequences. Through the SD model, the decision-makers can select the best combination of control and eradication strategies and their costs.
Many software packages simulate SD models: spreadsheets focus on one point in time while stock-and-flow software (such as Vensim, Dinamo or Stella) is able to analyse the system's behaviour (Ford, 1999).
In my specialization thesis, I developed a generic SD contagious disease model using Vensim. This SD model is able to overcome the difficulties of traditional mathematical models of diseases which may represent some aspects of the complex system of the diseases.
The flow structure of the model, that I developed, is similar to the SIR (Susceptible-Infected-Recovered) model use in epidemiology. The population (albeit humans or animals) may contract the infection and become infected without symptoms. To simplify the model, the infected are not able to spread the infection. A variable quota of infected is treated and the recovery is possible because of treatment or natural healing. After infection, a periodo of immunity persists. The model simulates the new infection. To reduce the disease's spread and the size of the infected population, model recreates the treatment used to cure the infection. Cost-effectiveness of treatments are also simulated. In picture 1 and 2 the model are represented, while in table 3 the model's equations are shown.
In conclusion, SD models enable a better simulation of the dynamic of host-pathogen interactions while their results mirror the reality.
Acknowledgements
I thank to Prof. R. Berchi of Milan for his irreplaceable and invaluable help to create and develop the SD model of my thesis.
References
Bagni R., Berchi R., Cariello P. (2002) A comparison of simulation models applied to epidemics Journal of Artificial Societies and Social Simulation 5(3),
Bernoulli M.D. (1766) Essai d'une nouvelle analyse de la mortalité causée par la petite vérole st des avantages de l'inoculation pour la prévenir. In Histoire de l'Académie Royale des Sciences, 1760, Avec les Mémoires de Mathématique et de Physique. L'Imprimerie Royale, Paris. [Citato in Thrusfield M. (1995) Veterinary epidemiology Second Edition Blackwell Science Ltd, Oxford (Gran Bretagna)]
Caulfield C.W., Maj S.P. (2002) A case for System Dynamics Global J of Enging Educ 6(1) 35-34.
Ford A. (1999) Modelling the environmental Island Press Washington DC,
Forrester J. (1961) Industrial Dynamics Waltham, MA: Pergasus Communications
MacDiarmind S.C. e Pharo H.J. (2003) Risk analysis: assessment, management and communication Rev Sci Tech Off. Int. Epiz. 22(2), 397-408
OIE (2004) Handbook on Import Risk Analysis for animals and animal products Vol I e II, ed OIE, Paris
Smith K.F., Dobson A.P., McKenzie F.E., Real L.A., Smith D.L., Wilson M.L. (2005) Ecological theory to enhance infectious disease control and pubblic health policy Front Ecol Environ, 3(1), 29-37
Thrusfield M. (1995) Veterinary epidemiology Second Edition Blackwell Science Ltd, Oxford
WTO (1995) Agreement on the Application of sanitary and Phytosanitary measures . Word Trade organization, www.wto.org
|
