Version classiqueVersion mobile

La lutte antivectorielle en France

 | 
Didier Fontenille
, 
Christophe Lagneau
, 
Sylvie Lecollinet
, 
et al.

Tools for risk management and anticipation

From vector risk assessment to epidemic risk assessment1

Texte intégral

  • 1 In this chapter, references to the CD-ROM concern question 10: “What contribution does vector risk (...)

1What are the objectives, principles, methods and system of analysis of vector risk? Does an entomological alert necessarily imply an epidemiological alert? What can modelling bring to this question? Is it possible to predict the size and gravity of an epidemic risk? These questions concern governments in the first instance and from them the public expects better anticipation of the risks and more rapid response to any alert.

VECTOR RISK ANALYSIS: OBJECTIVES AND METHODS

2Risk assessment arose in the early 1980s in the United States. The approach was developed by the National Research Council which defined it in 1983 as being “the utilization of scientific facts to define the effects on health of exposure of individuals or populations to dangerous materials or situations”. It involves therefore the methodical use of research results for scientifically based management of risk and for taking into account gaps in knowledge by following an approach that is as structured and quantified as possible”. Risk assessment presents the results of the characterization of risk(s) along with a discussion of the associated uncertainties. The risk analysis groups together the whole of the actions of assessment, management and communication relative to these risks.

3Such analysis as applied to vector-borne infections (vector risk, see Box) entails examination of the indicators and characteristics of the “vector system” within a complex context (hosts, pathogen, environment, society), gaining and understanding of the dynamics of transmission and the consequences of modifications of one or more elements of the vector system and/or one of its interactions. In a complementary way, it is also a question of analysing the positive and certain possible negative effects of actions that could be implemented. The aim is to provide aid with decision-making, upstream of health repercussions (anticipation) or during an alert, even an epidemic (analysis of management options). This essentially multi-disciplinary process can in return stimulate new questions for research.

Scope of application of risk analysis
■ Anticipation by analysis of potential consequences of the introduction of a competent vector into a given environment.
■ Analysis of the introduction of infectious subjects into an environment where the theoretically required conditions coincide or could coincide.
■ Analysis of the medium- to long-term consequences of changes in an ecosystem that could favour the expansion of an endemic situation.
■ Analysis of factors and conditions that enable a vector system to generate an epidemic.
■ Assessment of epidemic management options according to the potential size of the outbreak.

4In France there is no agency, ministry or organization in charge of this specific kind of investigation. Such analyses have nevertheless been conducted on several occasions: several control plans have thus been produced, such as the West Nile plan for the Mediterranean rim (beginning of the 2000s), the chikungunya-dengue diffusion control plan in metropolitan France (2006), updated on the basis of the risk assessment conducted following the 2007 epidemic in Italy.

Dengue-chikungunya dissemination control plan

5With the aim of formulating a plan, the Minister of Health assigned the InVS to lead an assessment of the risk of the introduction of chikungunya and of the outbreak of such epidemics in metropolitan France and hitherto unaffected overseas departments. A qualitative study resulted in a number of conclusions, including for metropolitan France. A risk of diffusion of the virus appeared to exist in the summer period (for further information, see CD-ROM).

6This assessment led to the drawing-up of surveillance and early warning measures in the framework of a dissemination control plan. Moreover it proved to be predictive both of the risk of expansion of the mosquito’s range, which is currently observed, and of the risk of diffusion of the virus as testified by its circulation as an autochthonous organism, observed for the first time in 2007 in Italy.

Risk analysis of chikungunya and dengue epidemic

7In the European areas where Aedes albopictus is installed, the analysis conducted at national level, like the similar one run at European level by the European Centre for Disease Prevention and Control (ECDC) in 2006, yielded similar conclusions for these zones on potential introduction and autochthonous transmission of chikungunya. The epidemic in Italy of 2007 prompted an update of this risk assessment, notably on the basis of modelling of the potential for diffusion and implantation of this vector in the United Kingdom, which was extended to European scale (see further on in this chapter “Contribution of modelling”).

8This new assessment, part of a collaboration between the ECDC, InVS and the Istituto Superiore di Sanita (Italy), was concretised by a quantitative assessment of the risk of introduction of persons carrying the chikungunya or dengue viruses in an area where Aedes albopictus is present. It was estimated for 2006 that 565 “viraemic person-days” for the dengue virus were introduced into the European areas infested by Aedes albopictus, 228 of whom came in the first three months. In Europe, most of the risk of diffusion of these two viruses is concentrated in Italy at the time when the vector is most active, owing to its presence in most parts of the peninsula.

Limits of risk anticipation

9There are limits to the extent risks can be anticipated. The case of Bluetongue disease is a good illustration. Until recently this disease, although widespread over the world, had spared Europe (except Cyprus), even though incursions had been observed here or there. However, since 1998, different serotypes have been transmitted throughout the Mediterranean Basin. In 2000 a bluetongue outbreak was detected for the first time in metropolitan France (in Corsica), a short time after the presence of Culicoides imicola had been confirmed there. Transmission of these serotypes is linked to the northward extension of the vector’s distribution range, possibly associated with changes in Mediterranean ecosystems under the effects of global warming. Nevertheless, it is clear that bluetongue is transmitted by autochthonous vectors in areas where Culicoides imicola is absent, like the Balkans, or rare, as in certain areas of Italy.

10From 2002, an entomological surveillance system was set up under an agreement between the Ministry of Agriculture and CIRAD. This system has proved to be effective. Indeed, it revealed the colonization of Culicoides imicola in the Var in 2004. It was reinforced in 2005 in departments newly at risk (Var, Alpes-Maritimes, Pyrénées-Orientales, Pyrénées-Atlantiques). In 2006, serotype 8 of the bluetongue virus, absent from the Mediterranean Basin, was introduced into Belgium before spreading rapidly to many other European countries where Culicoides imicola is absent. It causes clinical forms of bluetongue in cattle. Moreover, at the end of 2007 serotype 1 was introduced into the Basque country, on both the Spanish and French sides.

11These episodes show indisputably that the European species of Culicoides are capable of intensively transmitting bluetongue virus. Above all, however, they show the limits of risk anticipation. It is difficult to predict the emergence of a serotype in an area where the disease had never struck, borne by vectors hitherto considered as secondary. More broadly, it illustrates the difficulty of thinking of risks that deviate far from the established theoretical pattern.

ENTOMOLOGICAL AND EPIDEMIOLOGICAL EARLY WARNING

12The presence of the vector of the pathogen responsible for a vector-borne disease is evidently essential for the effective transmission of that disease, but it is far from being sufficient. Other conditions must coincide:

  • presence of a pathogen;
  • possibilities for effective contact between vector and pathogen reservoir;
  • possibilities for effective contact between vector and amplifier host (if that is different from the vector in the transmission cycle);
  • possibilities for effective contact between vector and hosts;
  • sufficient susceptibility of the host population.

Role of the vector

13The vector’s role in transmission is itself dependent on a range of parameters. It should be regarded beyond the simple question as to whether or not it is present. The installation of a means of circulation of the pathogen in a given geographical area can be analysed by following a series of stages (for concrete examples of each of these stages, see CD-ROM).

14The introduction of the vector and then its installation depend on the climatic, hydrological and biological (ability to adapt to new ecosystems) conditions. This installation also depends on competition with other vector species (for example, Aedes aegypti against Aedes albopictus in the south of the North American continent or in Mayotte for dengue) or confrontation with predators and parasites. Entomological studies conducted during the chikungunya epidemic moreover showed that on reunion Island Aedes albopictus benefited from a high degree of ecological flexibility which enabled it to colonize different ecosystems.

15Then when the vector is durably installed, the risk of seeing the start of a chain of pathogen transmission depends on its vector competence, in other words the level of coadaptation between pathogen and vector. This can be measured in the laboratory and depends mainly on genetic factors. In addition, there is the effect of inter-species competition which can influence pathogen transmission, positively or negatively.

16As soon as the transmission chain starts, the intensity of transmission to humans or animals depends on the vector capacity (daily average number of secondary infectious bites originating from one infected case). This in turn is governed by the vector density, trophic preference, length of the gonotrophic cycle, parturition rate and the duration of the extrinsic incubation period. The level of vector capacity could result either in a massive epidemic like that of chikungunya in 2005-2006 in Reunion Island, or an out-break like the moderate dengue epidemic in 2004, also on Reunion, or a contained outbreak like the recent one that occurred there (Saint Louis).

17Finally, when the transmission system is established, the mode of circulation varies with a number of parameters, usually extrinsic to the vector: environmental factors and the relation between the host and the vector, a population’s susceptibility and acquired immunity and actions taken for reducing vector pressure. Parameters depending on the vector competences can also affect the situation. Thus, during the Reunion chikungunya epidemic, a mutant genotype of the virus was selected. The hypothesis was put forward that the mosquito’s increased competence for the mutated virus could partly explain the stronger intensity of the second wave of epidemic.

Surveillance of entomological parameters

18What entomological parameters should be watched over for what kind of warning? Two types of parameter should be considered: those giving an assessment of the vector’s part of the risk and those that can help predict this vector risk.

19The first therefore aim to evaluate the part played by the vector following the stages already described:

  • presence or otherwise of the vector searched for in the places where its introduction is plausible, in the knowledge that the increase in the number of exchanges multiplies the opportunities for this;
  • its durable installation and its ecological preferences;
  • its vector competence in relation to the pathogens with which it can come into contact on taking its blood meal (imported human or animal cases);
  • its vector capacity in the situation studied.

20Measurement of these parameters implies the making of field observations and measurements. Those concerning the vector competence and capacity are sometimes highly complex to put into operation and require considerable work on the subject by the scientific teams.

21Gauging the parameters that could help predict the vector risk over extensive areas would entail targeting the zones where there is a risk of introduction, installation or proliferation of the vector. Targeting calls upon the techniques of cartography or modelling or both, possibly using remotesensing satellite data. Some examples are: abundance models of Culex (in order to anticipate the appearance of West Nile virus circulation) in British Columbia or New York, the density of Ixodes pacificus nymphs (vector of Lyme’s disease) in the Californian woodlands.

Pertinence of entomological parameters

22The pertinence of entomological parameters as epidemiological early-warning criteria differs with diseases’ vector transmission cycle and, for the same disease, depending on the endemo-epidemiological situation and the ecosystems.

23For dengue, for example, the situation in metropolitan France is different from that in the French Départements of America (FDAs). In metropolitan France, which is supposedly unaffected, the presence of the vector is a relevant criterion for increasing the level of surveillance and launching preventive actions against any installation of a chain of transmission (of dengue or chikungunya as it is the same vector, Aedes albopictus). In the FDAs, however, entomological surveillance is not much used for epidemiological warning. But the degree of sensitivity to insecticides, a key parameter for gauging the effectiveness of vector control, is closely watched. The same contrast is seen for Rift valley fever between Mayotte and metropolitan France (see these and other examples on CD-ROM).

24For vectors other than mosquitoes, little or no use is made of entomological parameters as warning criteria, but this could change in the future. In this way, if vaccination programmes succeed in strongly reducing the incidence of tick-borne encephalitis in Europe, other indicators than human cases will have to be found for surveillance of the presence of viruses in ticks or animal reservoirs (for vectors of Chagas’ disease and Batonella Quintana, see CD-ROM).

CONTRIBUTION OF MODELLING

25Modelling is generally based on a simplification of real situations. In the case of the transmission of an infectious disease by a vector, it is therefore founded on a simplification of the biological, epidemiological and/or social aspects of the problem, with a view to predicting but also prior assessment of scenarios and measures of control and prevention. The first attempts to model the dynamics of infectious diseases in the human and animal population, which we owe to the Swiss physician and mathematician Daniel Bernoulli, go back to the second half of the xviii Century. And it was in the xxth Century that the founding principles of modelling of vector-borne infection risks were introduced, thanks to pioneering work (1911) of the British doctor Ronald Ross (Nobel Prize for medicine 1902) and those published in 1957 by G. MacDonald (then director of the Ross Institute of Tropical Hygiene in London). The Ross and MacDonald model defined the bases of the control of vector-borne diseases, including anti-vector measures.

Different types of model
Two types of model are distinguished: qualitative (conceptual) and quantitative ones, which include statistical (empirical) and theoretical models. The theoretical models include deterministic models which are based on a strong hypothesis of homogeneity between the individuals (which is generally not the case in reality), and stochastic models which take into account the het erogeneities between individuals but at the cost of much greater complexity. These heterogeneities can also be taken into account, at least partly, by stratified deterministic models.
The development of powerful information technology tools has helped modeling become indispensable in the face of emerging infectious risks (pandemic influenza, SARS). It gives rise to numerous publications in prestigious scientific journals.
For vector risks, one of the main applications concerns the assessment of the spatio-temporal diffusion of a competent or potentially competent vector able to transmit a pathogen. The other bears on the assessment of the transmission of a vector-borne disease, and in particular:
■ understanding of the dynamics of transmission and identification of deficiencies in knowledge for surveillance and research;
■ assessment of fundamental parameters of transmission (reproduction rate, series interval etc.) with retrospective or even in quasi-real-time;
■ risk assessment;
■ prospective assessment and actions of public health and vector control and their combination;
■ impact assessment of a vector control measure.
For further information on the principles behind these models, their implementation and the bibliography, see CD-ROM.

Spatio-temporal modelling of a vector

26Modelling has given rise to a number of applications bearing on bluetongue disease, dengue, malaria and on the diffusion of Aedes albopictus in Europe, triatoms, vectors of West Nile virus. The objective of these statistical models, sometimes termed “geographical”, is to describe and predict the spatial distribution of vectors in order to identify and map the areas at risk and follow-up their changes and developments. The underlying hypothesis is that the presence of these arthropod vectors and their ecology, and also the spatial distribution of the disease transmitted depend on the environment. The host/vector interaction also corresponds to favourable environmental conditions (for a review of these models, see CD-ROM).

Bluetongue disease

27Models first developed from low spatial resolution meteorological and satellite data have led to the production of probability maps for the presence of Culicoides imicola at continental scale. Progress in satellite imagery (bringing ever finer spatial resolution) has made it possible to predict vector distribution at local scale (at the scale of a region the size of Corsica). A vector like Culicoides imicola has a low active dispersion. Small-scale analysis helps to formulate hypotheses on the location of breeding sites and resting sites and therefore target entomological searches.

28Continental-scale models cannot explain differences in distribution at local level observed on small areas (like Corsica, Var), or characterize in detail the habitat favourable for Culicoides imicola. Conversely, the generalization of results obtained at local level on more extensive zones is difficult or makes it necessary to acquire and standardize data, which is often costly.

Aedes and potential dengue epidemic

29Dengue transmission is subject to climatic factors. Coupling of climatic models with the equation of the vector capacity of dengue has led to the identification of a potential epidemic indicator. Increase in this indicator means that a lower mosquito density can sustain the disease in an endemic mode when the virus is introduced or circulates.

30Historical data (climate from 1931 to 1980) and a prediction of a temperature rise of 1.16°C from now to 2050 suggest that the potential epidemic indicator is also rising with a stronger progression in temperate zones. Models indicate that with global warming, incidence will rise first in the regions bordering endemic areas and then also in the latter zones. However, these models still involve uncertainties and considerable scientific debate which make complementary work necessary. Indeed they do not take into account other criteria that could influence the basic reproduction rate (human-vector contact, vector movements, vector control measures and so on). It must be recalled that Europe has already experienced dengue outbreaks, as in Greece in 1927-1928 for example where Athens suffered a million cases.

31Other models have been published in the literature whose objective is to describe the dispersion dynamics of Aedes aegypti (model of biological invasion), or to assess strategies for containing the invasion and predicting the changes and developments in Aedes albopictus, or forecast the abundance of a vector while taking climate fluctuations into consideration (for further information, see CD-ROM).

Potential spread of Aedes albopictus in Europe

32An aim to analyse the risk in order to arrive at proposals for prediction mechanisms prompted the use of a qualitative model associated with a geographical information system in the United Kingdom to gauge Aedes albopictus’s ability to become installed and its probable seasonal activity during the year. A similar European-scale analysis was launched. This revealed a strong potential for the installation of this vector in Europe, well beyond the Mediterranean countries that are already affected.

Anopheles and malaria

33Predictions from a spatial distribution model for Anopheles hyrcanus in the Camargue and its development with various environmental parameters, derived from satellite data, have been studied alongside field observations. The two types of data showed good agreement. An adult index, derived from the larval index, proved to be very well correlated with the observed abundance of the vector. This work indicated that in certain ecosystems, prediction of the distribution of a malaria vector is both possible and feasible by using satellite data, a first step in the assessment of the risk of reemergence of malaria in the south-east of France.

Triatoma

34A model of the geographical distribution of Triatoma, blood-sucking bugs, in the State of Puebla (Mexico) can, for each species, help identify the geographical and climatic conditions favourable for its development. It can correlate the potential presence of Triatomine species with the population under the risk of Chagas disease.

West Nile virus vectors

35A population-dynamics model for Culex tarsalis and Culex pipiens, developed with a view to spatial analysis of the risk of transmission of West Nile virus (before its implantation in British Columbia), has led to the identification of the areas with the highest risk of infection by the virus before its introduction. It has also highlighted some large gaps in knowledge (on the associated ecology, ecosystems). The distribution of the disease’s vectors has also been studied, using a predictive statistical model.

Functioning of the model

36A model of the dynamics of a vector transmission aims to reproduce the transmission of the pathogen within a vector system by numerical methods. Such a model, whether statistical or theoretical, must therefore incorporate many different parameters concerning the infection history in the host, the vector, the vector-host interaction, the reservoir, the environment and the factors influencing its transmission.

Statistical models

37These models make it possible to put forward predictions of the changes and developments of the risk, even the impact of vector control measures, on the basis of pre-established statistical relationships. They can be applied to the detection of epidemics, as is the case in the French Départements of America, the prediction of an epizootic disease outbreak (like Rift Valley Fever), or forecasting of the dynamics of a disease (such as leishmaniasis) with a view to assessing the impact of control measures.

38Such models have been further developed to anticipate outbreaks of Rift Valley fever, particularly in Kenya in an area where temporary pools form (for further information on models, see CD-ROM). A correlation has been established in this region between episodes of exceptionally abundant rain, themselves associated with El Niño events, and epizootic outbreaks of Rift Valley fever. The exceptional rains flood the dambos which then act as an ideal breeding ground for the mosquito vectors (Aedes). In Kenya, for example, outbreak predictions have reached an accuracy of 95 to 100 %, two to five months before their occurrence, by using measurements relating to vegetation and rainfall (including satellite data, field measurements). They make it possible to set up preventive measures (vaccination of vulnerable herds, mosquito-larva control). Another example is seen in Senegal: a stochastic model has simulated the densities of Aedes vexans and Culex poicilipes as a function of rainfall.

39In parallel, time series models (Chaves model) suggest, with a strong probability, a link between the transmission dynamics (of leishmaniasis, for example) and the climatic factors. The role of temperature was recently observed in the resurgence of leishmaniasis epidemic in the Jerusalem area. Similar work has been conducted on malaria and dengue.

Theoretical models

40These models are based on a preliminary representation of the phenomenon studied, using deterministic models of different elements of the system at whole-population level and stochastic models that can allow the introduction of a random proportion and hence an individual component (see the principles in detail on CD-ROM). In general, they are founded on modelling of the transmission of the infection between the different elements of the epidemiological cycle, which necessitates a schematic representation of the transmission between the different populations involved and their various relevant stages (see Figure 5).

41In the particular case of vector-borne disease, the host reservoir and patient can be purely animal (as in bluetongue) or human (dengue, chikungunya), or can affect both animal and human (infection by West Nile virus, Rift valley fever, leishmaniasis, tick-borne encephalitis, Lyme disease). Figure 5 mainly concerns the first two categories. Whatever the case may be, two parameters play a decisive role in the transmission models (see Box): the base reproduction rate and the times series or intergenerational interval (the interval between two successive cases).

Figure 5. Flow chart representing transmission of a vector-borne infection
Example of a compartment-model flow chart representing transmission of a vector-borne infection (adapted from Tran et al., 2005). The continuous arrows indicate host and vector state flows between the different compartments of infection chronology; the dashed arrows represent pathogen transmission between host and vector and vice versa.

The key parameters of transmission: reproduction rate and series interval
The reproduction rate (R0) measures the number of cases of infection, in a susceptible population, resulting from the contact with an individual already infected (outside any implementation of a control strategy). A rate of less than 1 signifies a fall in the number of cases as the transmission process proceeds, a rate higher than 1, an expansion of the disease and an epidemic situation. In real situations, it may be that only a portion of the population is susceptible (non-immune). To take account of the last element in the transmission, we then talk of the “net” reproduction rate R, the product of R0 and this proportion of the sensitive population. It is then R which must be lower than 1 for the epidemic risk to recede and peter out. Knowledge of R0 is of crucial interest as it can be used to deduce the proportion of individuals who must be immunized or protected or to judge which vector control measures to conduct to prevent an epidemic outbreak.
As for the series interval, it expresses the speed of occurrence of cases in the transmission chain. In such a chain, the series interval depends on several parameters: time interval Tv between the date of infection and the beginning of symptoms in the host (1 to 2 days for chikungunya), time interval TB between the appearance of the effective infection and the bite by the competent mosquito (1 to 6 days for chikungunya), time interval TM between the bite responsible for contaminating the mosquito and the one which will contaminate (4 to 6 days for chikungunya), time interval TI between the date of infection of the new host and the appearance of the symptoms (2 to 7 days for chikungunya). For a vector-borne infection, the last time interval depends on the vector and the interaction of the pathogen with the latter. In general, the shorter the series interval, the more the spread of the infection accelerates (multiplication of the number of cases) and the acceleration is exponential.

42A simple model has been used for example to analyse the risk of transmission of West Nile virus in the United States. The results are still in debate in the scientific community, but suggest that climate change will act on the abundance and geographical distribution of the vector, on its vector competence. This model therefore represents a fundamental tool for risk analysis.

Applications of theoretical models

43In a more general way, a proportion of these models have an essentially cognitive aim, whether it concerns dengue, leishmaniasis, or Rift Valley fever.

44Some of this body of research work focuses on the threshold effect and the state of equilibrium between the compartments of the model in the transmission and/or competition effects between serotypes. Others concentrate on the application conditions, the impact of the heterogeneity of conditions of the transmission of infection on the models and especially on predictions (see the abundant scientific literature on these subjects on CD-ROM).

45One category of model is used with the objective of evaluating the fundamental parameters of transmission (strength of infection, reproduction rate, series interval) with retrospective applications, even quasi-real time ones. Estimation of the reproduction rate by the Ross and MacDonald formula (mentioned in a preceding section) creates practical problems because the set of necessary parameters is not usually available for a given pathogen in a given situation. There are many published works on the subject (see CD-ROM), but it is important to check the application conditions case by case. It is worthwhile noting that recent biomathematical work conducted in 2004 following the SARS epidemic, makes possible a quasi-real time application of the instant when one has the data available on the cases per unit of time and at a place of occurrence as well as an estimate of the distribution of the series interval.

46One of the major applications of such models bears on risk assessment. For example, it is thanks to them that the ECDC, working jointly with the InVS and the Istituto Superiore di Sanita (Italy), assessed the potential risk of the installation of chikungunya and dengue in the regions of Europe where the Aedes albopictus vector is present. They used a simple theoretical model (see the paragraph above on the “risk analysis for chikungunya and dengue” and details on the CD-ROM).

47Another type of application concerns the prior and post assessments of public health actions and the impact of vector-control methods. The former have been much developed over recent years, and many specific pieces of work have been done on dengue, onchocercosis, malaria, chikungunya and lymphatic filariasis. The OnchoSim Project, fruit of a partnership between Erasmus University of Rotterdam and WHO, is a good illustration of this as it is an integral part of the WHO onchocercosis control programme in West Africa. It served to make a variety of prior assessments of medical and entomological interventions and their combination, showing that the elimination of onchocercosis by mass treatment is possible, with nevertheless a variable treatment duration depending on the level of infestation and of heterogeneity of the epidemic.

48As for the a posteriori assessments of the impacts of vector control, they can help reproduce (simulate) the epidemic without control measures (see detail on CD-ROM). Thus, the work conducted in Reunion Island suggests that without effective control measures 70% of the population, or even more, would have been hit by the chikungunya epidemic of 2005-2006. The fact that the epidemic was limited to an attack rate of about 38 % suggests that the control measures contributed to a reduction in transmission sufficient to change the net reproduction rate and hence to contain the extent of the epidemic.

STRENGTHS AND LIMITATIONS OF MODELS

49A review of the scientific literature brings out a prolific number of models. However, although some of the work mentioned above were integrated right form the start into a prevention programme (as is the case of OnchoSim already cited), this is not the case for most of them.

50A model of the transmission of a vector-borne disease must also incorporate all the social and behavioural aspects: people’s behaviour, social context, dynamics of immunity, effect of treatments and medical prevention measures. The example of dengue transmission in Singapore is symbolic of this. There, a significant discrepancy is observed between the successes of control effort (in terms of entomological indicators) and its relative failure in clinical terms. Account must in fact be taken of the dynamics of the immunity in the population, following a reduction in transmission owing to vector control. A resurgence occurred, paradoxically, probably owing to the very effectiveness of the control measures. All the more close attention must be paid to this aspect now that a vaccine is being developed against dengue.

51The last comment, which also can apply to control against malaria vectors, does raise controversy, however. Complementary research must be done in order to gain a more precise idea of the relations between the vector risk and the epidemiological risk as a function of vector systems;

52Finally, most of the theoretical models are deterministic in character, therefore based on a hypothesis involving a homogeneity that is hardly realistic. Moreover, few of them concern zoonotic vector-borne diseases (West Nile, Rift Valley fever and so on).

Strengths

53The contribution modelling can make stems from its ability to improve understanding of complex phenomena and to throw light on management options. Models provide a way of comparing qualitatively and quantitatively the prediction on the dynamics of a disease according to different scenarios regarding the conditions of interaction between the host, the pathogen and the environment, and as a function of control measures, even if they do not give the possibility of predicting exactly the number of persons infected or who might be.

54For example, during the latest chikungunya epidemic in Reunion Island, it is justifiable to think that a risk analysis based on a very simple compartmental model and the Ross and MacDonald model would have allowed, right from the Southern winter of 2005, identification of the epidemic wave of the beginning of 2006 from among the highly probable scenarios. From this point of view, modelling is not and cannot remain the prerogative solely of modellers and biomathematicians. It should, on the contrary, involve multidisciplinary teams following a multidisciplinary approach (entomologists, virologists, epidemiologists, biomathematicians, and also specialists on the environment, human and animal health, on public health and so on).

Limitations

55Models must nevertheless always be interpreted with precaution and after a careful examination of the hypotheses and conditions of application. Furthermore, it must be kept in mind that the conclusions of mathematical models depend strongly on the validity of the values and distribution of the parameters used (including incubation, latent and infectious periods, vector capacity and competence, contact rate and heterogeneity, intensity of infection, effectiveness and duration of acquired immunity). Many of these factors are often poorly estimated and imprecise. Sometimes they are unknown, as in the case of an emerging infection. Similarly, specific studies on the effectiveness of envisaged control measures must have been done beforehand.

56The validation of a model in relation to the observed data is an important condition for any wider application. However, even when the model’s internal validity is considered to be good, it does not necessarily mean that its external validity (performance in a different population or environment or another vector system) has been acquired.

RECOMMENDATIONS
Promote a well-organized health structure to coordinate risk analysis
To date there is no structure specifically organized to lead this type of analysis. However, risk analysis work is carried out in agencies (Agence Française de Sécurité Sanitaire des Aliments, AFSSA, InVS), in research institutes (IRD, CNRS, INRA, CIRAD, INSERM), in universities and grandes écoles in France. The application of the concept of risk analysis to vector-borne infections appears completely appropriate and useful as an aid to decision-making (anticipation of a risk, proven risk). The complexity of the analysis of vector systems means that it involves a rigorous multidisciplinary approach, which has been implemented in recent years in several public health organizations.
Develop standardized collection methods for data that is common to human and animal epidemiology, entomology and vector control, all being combined in a geographical information system.
The part played by entomology in epidemiological warning measures results mainly from knowledge stemming from research on population biology and pathogen-vector-hosts interactions, and from the performance of any entomological surveillance set in place. There is much still to be done, in particular: assessment and validation of the entomological indicators most relevant and best adapted for epidemiological early warning mechanisms.
This type of tool will in addition facilitate the follow-up of the appropriate indicators and will allow generation of warning signals with a view to triggering action.
Develop and incorporate models into special plans for the anticipation, management and development of surveillance and/or early warning systems to optimize the actions decision-makers and operational agents can take
The involvement of multidisciplinary teams is necessary to achieve this, both basic-research and applied-research oriented groups, so that vector systems can be analysed in all their complexity, including their environmental and human components (including immunity aspects).
The models are not in themselves the final objective. They are tools for use as aids to understanding observations, to prediction and decision-making. It is therefore important to have available assessments and validations of models in relation to the observed data, in order to ensure the interface between the conceptual development of a model and its possible application to operations: the internal validity of a model does not necessarily imply that it is externally valid.
This is a question of a highly important operational objective which entails bringing closer together the actors involved in surveillance and those in research, with the overall aim of developing and validating operationally useful tools.

Notes

1 In this chapter, references to the CD-ROM concern question 10: “What contribution does vector risk assessment make to epidemic risk assessment?”.

Table des illustrations

Légende Figure 5. Flow chart representing transmission of a vector-borne infectionExample of a compartment-model flow chart representing transmission of a vector-borne infection (adapted from Tran et al., 2005). The continuous arrows indicate host and vector state flows between the different compartments of infection chronology; the dashed arrows represent pathogen transmission between host and vector and vice versa.
URL http://books.openedition.org/irdeditions/docannexe/image/1277/img-1.jpg
Fichier image/jpeg, 55k

Le texte et les autres éléments (illustrations, fichiers annexes importés) sont sous Licence OpenEdition Books, sauf mention contraire.

Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search