Version classiqueVersion mobile

La dengue dans les départements français d’Amérique

Raymond Corriveau
Bernard Philippon
André Yébakima

Synopsis and Recommendations

Synopsis – point 2. Epidemiological monitoring

Texte intégral

1Three questions in the survey specifications address the theme of dengue epidemiological surveillance:


2What objectives should a national and/or regional dengue surveillance system have, for early warning and for monitoring trends?

3Pascal Chaud, Jean-Paul Gonzalez and Bruno Hubert address this question (article entitled Objectifs d'un système national et/ou régional de surveillance de la dengue dans les domaines de l'alerte et du suivi des tendances).


4What should the intervention thresholds be (suspected or probable cases, confirmed cases, severe forms) and how should they be determined?

5Bruno Hubert addresses this question in a short article entitled Les seuils d'intervention en période pré-épidémique.


6What can virus detection and characterization tools contribute to dengue surveillance and control?

7Jean-Paul Gonzalez and Pascal Chaud address this issue in an article entitled Méthodes et outils de détection du virus et de surveillance de son activité.

8These three questions are linked together. For a clearer overview it appeared better not to answer to them one by one but, first, to remind scientific and empirical fundamentals of a surveillance system of dengue, and then, to present different proposals concerning objectives, indicators, tools, alert thresholds, knowledge gaps and local studies to be conducted at regional level.

9Dengue transmission risk can be addressed indirectly, through entomological data, the behaviour of different social groups, the dynamics of geographical and climatic factors, etc. But epidemiological monitoring primarily depends on the capacity to detect the virus and identify it as accurately as possible.

10The main purpose of detecting and characterizing the dengue virus is to develop specific, sensitive tools for early diagnosis of the disease. This implies further objectives:

  • surveillance and control of the disease and constant improvement of medical management of cases;
  • discovering a differential diagnosis for the early stages of the infection, so that dengue can be distinguished as early as possible from infection by any co-circulating related viruses that cause dengue-like symptoms, which can lead to sub-optimal patient care;
  • identify viral serotypes so as to monitor their order of circulation and assess the risk of introduction of a new serotype into a non-immune population;
  • identify viral genotypes, which give information about the geographical origin of strains (genetic epidemiology) and their pathogenicity. The results can be used to build up a regional or international data repository on genotypes.

The dengue virus and its expression in humans

11The dengue virus is a member of the flaviviruses. It is a complex, fragile organism that can survive only for a short time outside the protective environment of its host. To survive and multiply, it must parasite target cells in its hosts, mainly mosquitoes and humans.

12The virion is the physical part of the virus, composed mainly of structural proteins (envelope proteins and nuclear proteins) and non-structural proteins which are mostly enzymes playing an essential role for the survival of the virus. It is these viral proteins that provoke antibody production in the host as a means of defence against the infection.

13The dengue virus has a single-strand RNA genome. This simple genome shape is common to all four serotypes, and also to other flaviviruses. It favours recombination between strains; the implications of this are still being studied, but it may be a factor for rapid evolution and/or the emergence of a new type of strain. The dengue virus has been shown to mutate readily, and this makes humans more vulnerable. Because of this risk of variability, laboratory research is still needed to achieve a better understanding of the infectivity, virulence, pathogenicity and evolution of the virus.

14Viremia and immune response are separate phases in the evolution of dengue virus infection in man. Viremia is the presence of the virus in the blood of an infected person. This provokes an immune response, i.e. production of antibodies (immunoglobulins or Ig). Like the virus, the antibodies are evidence of infection. A significant increase in M class immunoglobulins (IgM) are a sign of infection by the dengue virus; an increase in A or G class immunoglobulins indicates infection by a virus that may be either dengue or a related virus.

Laboratory diagnosis

Usefulness and principle

15Laboratory analysis is used to diagnose individual cases; on the public health side, it is a decisive tool for assessing the extent of the infection in the population.

16Serological data are essential for clinicians, epidemiologists and public health managers, particularly those responsible for dengue control. Confirmation of clinical diagnoses, surveillance and control of the disease, study of its incidence and study of its clinical gravity all depend on the specificity and degree of sensitivity of serological examinations.

17As regards public health, one of the main problems is the time lost waiting for a diagnosis. This mainly depends on the stage the disease has reached, the patient's immunology, the quality and preparation of the blood sample to be tested, and also the speed with which the laboratory can return the information needed for proper medical management of the case, monitoring the circulation of the virus and launching control operations if necessary.

18Factors that must be taken into account and properly managed are the methods and materials used, the cost and availability of diagnosis (as techniques become increasingly sophisticated and efficient, they also become more expensive), the associated information systems (for transferring results) and staff qualifications (the work is now highly specialized).

19Biological diagnosis techniques must be able to characterize the dengue virus, differentiate it from related viruses, distinguish between different dengue virus serotypes (serotyping) and even characterize the genomes of circulating strains within the same serotype (genotyping).

Diagnostic methods

20Samples are taken and selected according to purpose (diagnosis, monitoring patients' progress, dengue surveillance, etc.), the chosen technique, the degree of precision sought and the clinical stage of the illness.

21Detecting the virus can be done directly (isolating the virus, detecting antigens) or indirectly through serological traces of antibody-antigen reactions in the host. A virus strain can be detected by PCR or by culturing on mosquito cells. In both cases a biological sample has to have been taken during the viremia phase (the first four days). Antigens are looked for in these cells and in samples of tissue or biological fluids.

22Laboratory methods for detecting human antibodies vary in cost and speed of execution. The most commonly used is ELISA (Enzyme Linked Immunosorbent Assay). Another is the haemagglutinin inhibition (HI) test. An increase in IgM antibodies indicates either a recent infection (the past six to ten months) or an active infection. For seroprevalence surveys, the presence of IgG indicates an old infection, but does not distinguish between the dengue virus and antigenically similar flaviviruses.

23Identifying serotypes can be done either serologically using the seroneutralisation test, which is highly specific but slow and costly, or virologically by characterizing the virus or a serotype-specific part of its genome, using PCR, which can also be used to specify genotype.

Fundaments of an epidemiological monitoring system

24Epidemiological monitoring of a disease like dengue generally has three purposes: knowledge of the disease, evaluating measures and programmes, and early warning and intervention. Early warning and early intervention are a priority because they partly determine the efficacy of measures to control vector populations and prevent the dispersion and expansion of outbreak areas.

25For fast, appropriate and well-targeted activation of the only measures currently available (vector control, including preventive measures in the community), there has to be an efficient epidemiological monitoring system. To be responsive, such a system must have:

  • the capacity to detect and characterize the virus quickly, precisely and reliably throughout the area concerned (i.e. in this case everywhere in the FDAs);
  • a good entomological and epidemiological information gathering system;
  • well-defined, proven indicators of (a) vector risk and (b) the scale and characteristics of the epidemic;
  • as precise as possible a definition of the thresholds for alert and intervention;
  • the means for rapid dissemination of information among all actors concerned.

Existing monitoring systems in the FDAs

26For dengue, all three départements are within the PAHO's highest transmission risk class: "the vector is present and dengue is endemic with periodic epidemics". For comparison, mainland France is in the lowest category: "the vector is absent and its presence highly unlikely".

27Dengue surveillance in the FDA's is organised along three "couples Département-Région”, and all three take part in sub-region arrangements with corresponding coordinating systems in neighbouring countries. This "regional system” overlaps with the national one applied in mainland France, which does not involve systematic monitoring of imported cases of dengue.

28In the French Antilles there have been dengue surveillance systems for ten years now, based on active gathering of data from sentinel networks of doctors in private practice and laboratories performing serological tests. The data are analysed weekly by each department's DSDS. The indicators (gross and mean numbers of suspected cases, per doctor and as a proportion of serological tests) are examined for each département and by commune, to detect any abnormal increase. The vector control services are alerted if a geographical cluster of cases is identified. In French Guiana, where the sentinel network of doctors is currently being formed, dengue monitoring was previously handled entirely by the Pasteur Institute in Cayenne, which is a national reference centre for arboviruses. The system has several limitations:

  • in the Antilles more than half the serological tests performed go unrecorded because they are carried out by unaffiliated laboratories, either locally or in mainland France;
  • neither doctors nor patients are keen to have tests done, partly because of the cost, partly because a period of time is necessary for the detection of IgM, partly because of the long waiting time for a result;
  • test requests and results reports do not include enough information;
  • because the sentinel work is unevenly spread, the surveillance of clinical and confirmed cases is not representative enough to allow early detection of epidemic areas;
  • there is no structured system for monitoring severe or hospitalized dengue cases;
  • monitoring of circulating serotypes is insufficient;
  • gathering, analysis and transmission to the vector control services of useful epidemiological information is far too slow to allow timely, well-targeted control operations. The usual waiting time for laboratory results is two to three weeks;
  • there is insufficient feedback to information providers and different sectors of the public, this being a demanding and time-consuming task.

Aims of epidemiological monitoring

In the FDAs

29The aims were proposed in 1998 in the Guide de la surveillance de la lutte contre la dengue dans les départements français d'Amérique and were set out in detail at an inter-regional workshop of health services in April 2001 in Cayenne. They differ between the three phases of an epidemic, whatever the geographical scale of observation, and can be summed up as follows:

30In endemic or pre-epidemic periods:

  • detect early any increase in surveillance indicators, any change in circulating serotypes (identity, distribution);
  • systematically identify as early as possible any pathogens causing dengue-like febrile syndromes that may cause false alarms and lead to errors in patient care and dengue epidemic forecasting;
  • confirm the occurrence of an epidemic, on whatever geographical scale;
  • trigger and pilot vector control measures;
  • monitor and analyse disease trends (number of cases, clinical presentation, severity, etc.).

31During confirmed epidemics:

  • monitor the evolution of the epidemic in terms of geographical spread, serological and clinical characteristics, etc.;
  • describe clinical pictures and optimize medical management by keeping close watch on severe forms;
  • find out the serotype(s) concerned and its (their) genotype(s);
  • target the first vector control operations on zones with highest disease incidence. In practice, given the time lag before the alarm is raised and the rapid spread of the disease, vector control treatment must cover practically the whole zone concerned from the outset.

32In late epidemic and post-epidemic phases:

  • confirm the end of the epidemic;
  • make a careful description of the epidemic to help improve forecasting and preparedness for controlling future outbreaks;
  • assess the immune status of the population for the different serotypes, in comparison to the pre-epidemic situation (seroprevalence surveys);
  • assess the control measures used during the epidemic, and their results;
  • estimate the cost of the epidemic;
  • make proposals for research to enhance knowledge of the disease and its control.

In mainland France

33Dengue surveillance must obviously focus primarily on the French overseas dependencies where the risk to human health is far higher. On the mainland, dengue should be treated like an imported arbovirus disease and handled by the existing specialist services.

34As travel to countries where dengue is endemic becomes more frequent, so do cases of dengue virus infection identified in mainland France, where dengue is second only to malaria as a cause of febrile illness on return from a tropical country. Surveillance of dengue cases would serve to quantify and characterize imported dengue cases and so inform health workers, travellers and travel agents. One obstacle to this could be unskillful use of the diagnostic kits by analysis laboratories unfamiliar with their use and limitations.

35Dengue monitoring on the mainland could be guided by the following aims:

  • describe the characteristics of imported cases;
  • inform health workers of dengue epidemics in the overseas dependencies and issue recommendations on the vigilance to be used with suspect cases coming from epidemic areas;
  • strengthen the prevention message to travellers and travel agents dealing with regions at risk from dengue and especially those where there is a current epidemic;
  • distinguish imported cases from possible autochthonous cases, which would necessitate rapid action against a vector (e.g. in the event of Ae. albopictus extending its range; this mosquito is currently under surveillance in mainland France).

Choice of indicators

36During epidemics, and in the prospect of early intervention which is the short term main objective of surveillance, priority must be given to the search for indicators sensitive enough to detect any transmission of the virus.

37A greater specificity must also be systematically aimed at through corroborating biological tests (serology and RT-PCR).

38At present, the more sensitive and reactive indicator is the follow-up of clinically suspected cases by the sentinel network, provided that it is associated with a parallel surveillance of the biologically confirmed cases.

39During epidemics the predictive value of clinical diagnosis of the disease is sufficiently increased: the clinical follow-up of suspected cases by the sentinel network allows the description of the epidemic evolution without the need for a systematic serological confirmation.

  • 1 S2E.Dengue: Spatial Surveillance of Epidemics applied to Dengue. Partners in this project include (...)

40Other indicators which might show great sensitiveness for detecting epidemics should be validated. The Pasteur Institute in French Guiana and the Cayenne hospital have established a correlation between the number of negative searches for Plasmodium on one hand, and the number of requests for serological tests and the number of positive dengue tests on the other hand. These three indicators increase synchronously in the period before a serologically confirmed epidemic. Meanwhile in Martinique, it has been noticed that before the pre-epidemic phase can be detected, there is an increase in requests for full blood count, thrombopenia and leucopenia. It therefore seems important to make a larger-scale assessment of possibilities for predicting epidemics by monitoring (from laboratory data) biological indicators that are not specific to infection by the dengue virus. This kind of monitoring would considerably increase the responsiveness and representativeness of the system and would be a useful addition to monitoring the number of suspected clinical cases reported by sentinel doctors. However, it could only be applied regularly if the laboratories were using a software for automatic storage and processing of the results of biological analyses in the laboratory's computer system; the S2E.Dengue project has a system of this kind in view1.

41The dengue cases in hospitals, especially in intensive care units, must also be included in the list of routinely monitored indicators, in order to establish the severity of the endemo-epidemic, to describe its clinical patterns and to organise the patients nursing.

42Finally, the follow-up of circulating serotypes and the immunity rates of the population against each of the serotypes (considering various factors of age, residence ...) are also useful indicators to guide intervention decisions and to predict epidemic risks.

Proposal for optimizing epidemiological monitoring in the FDAs

43Experience shows that, with a view to fairly fast intervention to halt transmission, data on suspected and biologically confirmed cases should be entered, processed and transmitted to the mosquito control services at least once a week.

44Information on suspected cases (reports from sentinel networks, requests for biological confirmation, hospitalized suspected cases) should be standardized on a single model with at least the contact details of the person reporting, the patient's age and address, the date of the first signs, the existence of signs of the severe form, and the type of test prescribed. Together, these data make it possible to date the infection, locate target populations and monitor the epidemic in terms of numbers of cases, severity and zones at risk. Additional information could be given on travel to endemic areas in the two weeks before symptoms appeared, and whether the normal place of residence is in a non-transmission zone.

45In order to fullfil the surveillance objectives, the clinical and biological patterns shown by hospitalized dengue cases should be systematically documented.

46In view of local emergency measures, a procedure is needed for compulsory reporting of biologically confirmed cases, hospitalized cases and deaths attributed to dengue, to enable the implementation of individual and collective prevention measures (Art. R11-3 of décret 99-362 of 6th May 1999 and Art. D11-2 of décret 99-363 of 6th May 1999). This would provide a legal and institutional framework for monitoring dengue and would legitimate notification of cases to the health authorities. It would also enable France to fulfil its international commitments with regard to hemorrhagic fevers and meet the WHO recommendations.

47Regular feedback information could be automatically drawn up and transmitted and the results sent by mail to the data producers; presented on websites in a suitable form, these results could reach such target audiences as health workers, the general public, travel agents etc.

48The S2E.Dengue programme is being set up in French Guiana and could be adopted and installed to Martinique and then Guadeloupe. It aims to develop computer technology for spatial monitoring of epidemics, mainly epidemiological monitoring of febrile syndromes in general and dengue in particular. As a first stage, the proposed tools would automate as far as possible the tasks of gathering epidemiological data, analysing it and transmitting it back to mosquito control services on one side doctors and biologists in private practice on the other side, hospitals and the army. These new methods should significantly improve the monitoring systems. The entomological indicators could more easily be linked geographically to the epidemiological indicators, provided patients' addresses are allocated to geographical zones defined in co-operation with the vector control services.

49As a second stage, the epidemiological information network built up will gradually be given a set of special algorithms for analysing and predicting outbreaks in high-risk populations. In this way, with the fusion of different kinds of spatio-temporal data (medical, entomological, meteorological and environmental) combined with the development of bio-mathematical models and geographical information systems, it should be possible to forecast and prevent epidemics.

The question of thresholds

50It is essential to define epidemiological alert and intervention thresholds, even though we do not yet have the capacity to correlate the values of epidemiological indicators with those of vector density or virus transmission intensity parameters.

51To establish these thresholds, the following must be taken into account:

  • most epidemics are highly localized at the beginning. The surveillance network therefore needs to be very dense;
  • epidemics start slowly; it can take four to six weeks for an outbreak to reach the epidemiologically detectable stage;
  • symptomless forms, or forms with very slight symptoms, are common, especially in young children;
  • the size of the population at risk from an epidemic varies greatly depending on the serotype responsible and exposure to earlier epidemics;
  • several serotypes can be in circulation at the same time in the same area; as a result, epidemics can spread over several years and risk assessment is difficult (French Guiana, 1996-1998);
  • because of the climatic and socio-geographical differences between the three départements, each département requires a separate approach.

52The notion of an alert threshold is both qualitative and quantitative, and a threshold value can only be determined by reference to earlier data.

53The alert threshold should be the signal for community mobilization and targeted mosquito control operations to begin. Whatever the vector control strategy adopted, speed and precision are essential.

54In practice, in places where dengue is only weakly endemic operations should be triggered by a single case confirmed by PCR or a serological test, by one hospitalized suspected case, by a report from a sentinel doctor of two or more suspected cases within a week, or – especially- by the detection of a single case of a serotype new to an area. To refine and combine the thresholds ultimately chosen, the S2E.Dengue project will have a decisive contribution to make.

Research to be done in the FDAs

55The studies required to provide data on local epidemiological patterns are mainly studies for which there are known protocols that need to be transposed to the FDAs or repeated there.

56Retrospective studies of existing data on dengue virus genotypes isolated in the FDAs would establish the order in which the different serotypes have appeared and how they have circulated, for the sub-region first and then planet-wide.

57Retrospective studies of the pattern of appearance of IgM antibodies during the epidemics of 1997 and 2001 should help us understand the outbreak dynamics of epidemics in the FDAs and ultimately identify risk indicators of the shift from endemic to epidemic status.

58Research into early markers of severe dengue is now an urgent task in the FDAs, given the increasing frequency of hemorrhagic and other severe syndromes. These markers are essential for improving patient care and efficient epidemiological surveillance. This requires an exhaustive study of the clinical presentations concerned and a semiological re-consideration of the disease.

59A study of viral load is under way at the Fort-de-France university hospital, and is expected to add to our knowledge of the clinical aspects, evolution and prognostics of the disease. Similar work performed elsewhere indicates that where dengue symptoms are severe, knowledge of the kinetics of viremia can help in deciphering complex clinical pictures and provide a decisive prognostic element for earlier medical management of patients.

60Refining differential diagnosis on the basis of a virological inventory of the FDAs' different ecosystems should produce a reliable etiological diagnosis data repository. This would help in assessing the epidemic risk of dengue or other diseases and establishing an early warning system.

61Identifying candidate strains for a vaccine. Given the lack of forecasting and prevention capability and the lack of a specific therapy, developing a vaccine for dengue is currently a world priority. French pharmaceutical research is well advanced in developing clinical trials of vaccines. This necessitates studies of viral strains from all affected regions and the participation of FDA virology and viral ecology resources.

62While the impact of these studies goes far beyond their borders, it is important that the FDAs take account of them and that the specialist institutes make a direct contribution. It is not simply a matter of taking part in a world scientific enterprise, even though there is a fast-spreading world dengue pandemic to cope with. Most of this applied research is designed to enlarge our knowledge of the mechanisms by which dengue develops into an epidemic, especially the early stages. Its value for prevention and control in the FDAs is obvious. As far as possible, this work must be done locally, in local epidemiological conditions.

63French laboratories are also playing an active part in research into antiviral chemotherapy. It would be a pity if the FDAs –which, along with the French Pacific and Indian Ocean islands, are the only parts of French territory to be affected by the disease– were not more closely involved in extending the range of dengue control strategies, both at département level and within the regional structures they belong to.


1 S2E.Dengue: Spatial Surveillance of Epidemics applied to Dengue. Partners in this project include the Centre national des etudes spatiales, the Lyon veterinary college, the Pasteur Institute in Cayenne and the IRD.

© IRD Éditions, 2003

Conditions d’utilisation :

Cette publication numérique est issue d’un traitement automatique par reconnaissance optique de caractères.
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search