WEST NILE VIRUS (WNV) associated disease

Transcription

1 VECTOR-BORNE AND ZOONOTIC DISEASES Volume 7, Number 3, 2007 Mary Ann Liebert, Inc. DOI: /vbz West Nile Virus Outbreak Detection Using Syndromic Monitoring in Horses A. LEBLOND, 1 P. HENDRIKX, 2 and P. SABATIER 1 ABSTRACT Recent outbreaks of West Nile virus associated (WNV) diseases, both in the old World and Americas, underline the importance for early warning systems that rapidly identify emerging and re-emerging diseases and thus help in their control. Traditional approaches of disease monitoring become less reliable and increasingly costly when used for rare health-related events, such as WNV outbreaks in southern France. The objective of this work was to discuss methodological issues related to syndromic monitoring of WNV-associated disease in Camargue horses by veterinary practitioners. Tracking cases of equine encephalitis by veterinarians is an example of such syndromic monitoring of an emerging disease. Signs of illness, observed prior diagnostic confirmation, can be of interest because they may provide an early warning for WNV circulation in a given area and allow authorities to take appropriate preventive measures for public health. Key Words: Monitoring Epizootic Syndrome West Nile virus Veterinary practitioners Horses. Vector-Borne Zoonotic Dis. 7, INTRODUCTION WEST NILE VIRUS (WNV) associated disease is a mosquito-borne flaviviral infection transmitted in natural cycles between birds and mosquitoes. Infection can cause febrile, sometimes fatal, illnesses in horses, birds and humans (Komar 2000). The frequency and severity of WNV outbreaks have increased during recent decades. In 1999, the virus reached the North American continent, with subsequent rapid geographic spread resulting in 3000 human cases with 119 deaths in 2005 (for United States Geological Survey 2006, see In Europe, even if WNV outbreaks are of little concern for public health, they can have an important economic impact. Tourist activity tends to fall when the virus circulates along the Mediterranean coast during the end of summer. In the south of France, outbreaks of WNV were observed in horses in 2000, 2003 and 2004, and human cases were diagnosed east of Marseille in Serological studies conducted in 2000 among blood donors living in the South of France showed a significant level of WNV-IgG and IgM prevalence (Gallian et al. 2005). At the present, even though circulation of the virus among horses or in the avian population does not always predict an outbreak in the surrounding human population, the French blood banks exclude blood donors living in areas at risk from July to October. For these reasons, public authorities are interested in implementing reliable monitoring systems that would be capable of identifying rapidly WNV outbreaks, and thereby helping in their control. The WNV outbreak in New York State (NYS), in the summer and fall of 1999, resulted in an increasing number of dead birds. So, the 1 UMR CNRS 5525 TIMC, Unit Environnement et Prévisions de la Santé des Populations, Ecole Nationale Vétérinaire de Lyon, Marcy l Etoile, France. 2 Direction des Services Vétérinaires du Gard, Nîmes, France. 403

2 404 Centers for Disease Control and Prevention (CDC) guidelines consider that a system based on the active monitoring of sentinel birds, and/or the observation of the mortality rate of birds, may provide a sensitive method of detecting WNV (CDC 2003). WNV has been monitored using captive sentinels in several locations around the world (United States, South Africa, Australia, and Romania). But even if seroconversion rates in sentinel chicken flocks did eventually exceed 1% in some locations, these seroconversions were generally detected late in the transmission season, after the onset of human cases (Komar 2001). In Europe, in contrast to the United States, birds usually do not show any symptoms when infected with WNV (Guptill et al. 2003). During the three recent outbreaks in the south of France, the passive monitoring system of mortality in wild birds, the SAGIR network (AFSSA Nancy), detected no abnormal or massive mortality of wild birds in the area (Hars et al. 2001). In 2004, seven seroconversions on sentinel birds were confirmed by the National Reference Center for Arboviruses on the 17th of September, when the epizootic reached its peak in horses. The objective of our work is to evaluate and discuss methodological issues related to syndromic monitoring of WNV-associated disease in horses. The fundamental objective of syndromic surveillance is to identify illness clusters early, before diagnoses are confirmed, and to mobilize a rapid response, thereby reducing morbidity and mortality. An early warning system for WNV outbreaks may help focus limited response resources and improve risk communication (Lewis et al. 2002). METHODS Horse population and area of study The target population included all horses living in Camargue, a region situated in the south of France, along the Rhone river delta. The last census of the livestock, conducted by the Ministry of the Agriculture in year 2000, estimated the horse population of the Camargue at approximately 7,000 individuals, located in an area of approximately kilometres. LEBLOND ET AL. From an ecological point of view, the Camargue presents a diversified environment, consisting of dry areas irrigated by canals and ditches, and wetlands (Hoffmann et al. 1968; Blondel et al. 1981). These biotopes are favourable to both mosquitoes, which are potential vectors for WNV, and populations of sedentary and migratory birds, which are potential reservoirs. Passive monitoring system Within the study area, WNV is considered a potential etiology for all equine clinical cases that include neurological symptoms, and occur between July and November. It is mandatory that the practitioner reports such suspect cases to the health authorities. To improve the early detection of a WNV outbreak, we implemented a web-secure, electronic system for recording clinical data by the veterinary practitioners. A medical software company (Calystene) developed an in situ method of automatic reporting using a personal digital assistant (PDA) with mobile communication (GSM/GPRS). In terms of timeliness, ambulatory record information could be rapidly made available, within 1 h of the case encounter (Lazarus et al. 2001). Data were transmitted to a secure website that provided on-line information on the epizootic. The network could also provide retrospective information from other electronically interconnected practitioners, laboratories, agencies and institutions. Seventeen veterinarians were known to have an equine practice within the study area. Among them, four were selected to test the electronic data registry system on the basis of the size and the location of their practice and of their expressed interest in the project (Leblond et al. 2005). The geographic distribution of their clientele allowed us to obtain a good coverage of the study area. The 13 nonequipped practitioners transmitted their declarations by fax to the health authorities. The web-site was updated daily. Case definition and reporting of surveillance data Neurological symptoms in horses were the target data for field reporting, including a brief datasheet that could be completed within 5

3 WEST NILE SYNDROMIC SURVEILLANCE IN HORSES 405 min. A list of the major symptoms observed was provided with yes/no response boxes. Neurological syndromes were detailed according the presence or absence of dysphagia, facial paralysis, blindness, tremors, hyperesthesia, grinding teeth, abnormal behavior, and locomotors abnormalities. Ataxia, paresis and paralysis were described as symmetric or asymmetric, with the option to specify which limbs were most affected (Snook et al. 2001, Castillo-Olivares et al. 2004). Other data collected for each horse concerned physical characteristics and activity during the last two weeks (recording of displacements in or out of the study area). Data were collected between the 1st of August, 2004 and the 31st of July, A suspect case was defined as a horse residing in the area of study which developed neurological syndrome. All suspect cases were tested for WNV- IgG and WNV-IgM antibodies by the Agence Française de Sécurité Sanitaire des Aliments (AFSSA). Suspect cases were classified as confirmed WNV cases when the ELISA IgM test performed was positive. Determination of epidemic at an early stage We used the number of new cases per week as an index to express the disease incidence in the horse population of the study area. The epidemic period was determined by this index. We assumed that an epidemic started during the first week when the index exceeded a predetermined positive threshold value, and ended a week before the indices for three successive weeks became lower than the negative threshold value. The threshold value for the onset of an epidemic (CV O ) was set between 95% and 99% of the distribution of the number of cases per week and the threshold value for the end of an epidemic (CV E ) was set to 90 percentiles of the distribution of the number of cases per week (Hashimoto et al. 2000). We assumed that if the number of cases per week in a rural district exceeded a critical value for an epidemic warning (CV W ), an epidemic could begin in this area in the following four weeks. The sensitivity, specificity and positive predictive value for epidemic warnings were evaluated by setting several critical values for epidemics. The sensitivity, Se, was defined as the proportion of epidemic-periods with true positive warnings in the pre-epidemic period (4 weeks) and the specificity, Sp, as the proportion of non-epidemic weeks without a warning (true negative). The positive predictive value, PPV, was defined as the proportion of true positive warnings among total warnings (Hashimoto et al. 2000). RESULTS Surveillance data and electronic reporting The personal digital assistants (PDA) were distributed to the four selected practitioners on the 10th of August After a short test phase, the first neurological case was declared on the network on the 23rd of August, the next two being registered on the 27th of August by the same practitioner. Among the 17 veterinarians practising in the area of study, twelve reported a total of 72 neurological syndromes in horses, from August 23rd, 2004 to July 21st, Three of the four equipped-veterinarians provided electronic follow-up on 30 horses with neurological syndrome up to the end of the study period (mean 10 syndromes per practitioner). Among the non-equipped veterinarians, nine recorded 42 syndromes (mean 5 syndromes per practitioner), from the 3rd of September to the 14th of October Among the 72 suspect neurological cases, 32 cases of WNV infection were laboratory confirmed (44%). There was no statistical difference between electronic and non-electronic reporting in the rate of confirmed cases (CHI , p 0.3). There was no statistical difference between confirmed and non-confirmed cases in the frequency of clinical signs observed (CHI , p 0.58; Table 1). The mortality rate in horses was 22% (n 7) for confirmed cases and 9 % (n 4) for non-confirmed cases (CHI , p 0.20). Epidemic curve Figure 1 compares the distribution of outbreaks of nervous syndromes and confirmed cases during the period of study (Spearman

4 406 TABLE 1. LEBLOND ET AL. CLINICAL SIGNS OBSERVED AMONG 72 HORSES SUSPECT OF WEST NILE VIRUS (WNV) ASSOCIATED DISEASE Clinical sign No. confirmed cases (%) No. nonconfirmed cases (%) Ataxia 20 (65) Forelimbs 2 30 (75) Forelimbs 2 Hind limbs 2 Hind limbs 6 Hind and forelimbs 1 Hind and forelimbs 1 Not specified 15 Not specified 21 Abnormal behavior 11 (45) Hyperesthesia 6 8 (20) Hyperesthesia 3 Teeth grinding 3 Teeth grinding 2 Hydrophobia 1 Hydrophobia 1 Anxiousness 1 Aggressiveness 1 Circle walking 1 Paresis, Paralysis 11 (45) 22 (55) Muscular tremors, myoclonia 10 (32) Generalized tremors 7 6 (15) Generalized tremors 4 Localized (lips, eyelids) 1 Myoclonia 2 Trismus, myoclonia 2 Hyperthermia 9 (29) 19 (48) Anorexia, lethargy 8 (26) 11 (28) Cranial nerve deficits 6 (19) Blindness 2 5 (14) Blindness 3 Dysphagia 4 Dysphagia 3 Facial paralysis 2 Facial paralysis 2 Others 9 (29) Colic 2 10 (25) Colic 2 Lameness 3 Lameness 2 Cervical, thoracic pain 2 Cervical pain 1 Anemia 2 Polypnea, dyspnea 2 Glossitis 1 Weight loss 2 Penile prolapse 1 rank correlation coefficient 0.88; p 0.01). From the curve, we deduced the percentiles of the distribution of indices for suspect and confirmed cases: i.e., the number of observed neurological syndromes was 16 for 99% of the weeks in the period of study, 8 for 95%, 3 for 90%, and 1 for 75% of the weeks (Table 2). Applying our method for detecting the epidemic period, we could set CV O 10 and CV E 3. The number of neurological syndromes per week exceeds the CV O in the beginning of the 39th week, and becomes lower than the CV E from the end of the 42nd week. Thus, the epidemic period was 4 weeks. Early stage of the epidemic and warning Figure 2 shows the distributions of the declarations of the syndromes (Fig. 2A) and confirmed cases (Fig. 2B) by dates of the onset of clinical signs. We calculated Se, Sp, and PPV of the surveillance for different CV W (Table 3). Increasing the CV W did not cause a decrease in the Se, but caused an increase in Sp and PPV. Setting the CV W for the clinical syndromes to 3 would have allowed the 2004 outbreak to be predicted in the 35th week, that is to say 4 weeks before the beginning of the epidemic period. If the declaration of a warning was based on confirmed cases only, a CV W of 1 or 2 would have allowed the outbreak to be predicted only one week ahead (Fig. 2). The PPV for the syndromic surveillance at a CV W of 3 is only 75%. In fact, one false positive warning was given in January DISCUSSION For rare health-related events, one objective of the monitoring should be to focus on potential high-risk populations (Doherr et al. 2001). Horses seem to be particularly sensitive to WNV and are often considered indicators of viral transmission. Of horses infected with WNV, approximately 10% present with neurological disorders, as compared to 1% of humans (Petersen 2001). Moreover, horses are distributed throughout the Camargue region, live in both urban and rural areas and are not vaccinated against WNV. We have attempted to evaluate the relevance of a monitoring of the equine population as a sentinel for WNV circulation in the Camargue area. WNV infection in horses may involve both central and peripheral nervous system signs.

5 WEST NILE SYNDROMIC SURVEILLANCE IN HORSES Epidemic period Confirmed cases Syndromes Number of horses Critical Value for Onset (95-99% syndromes per week) 4 2 Critical Value for End (90% syndromes per week) Week number at the beginning of clinical signs FIG. 1. Epidemic curve in 2004: determination of the epidemic period based on the distribution of neurological syndromes in horses (n 72); comparison with the distribution of confirmed cases. These signs may be indistinguishable from those caused by other etiologies. Other common causes of equine encephalitides in France include equine herpesvirus-1, Borna disease, rabies and, rarely, Equine Protozoal Myeloencephalitis. Other diseases of the central nervous system in horses result from trauma, moldy corn poisoning, hepatoencephalopathy or cervical instability. The definitive diagnosis of neurological disorders in horses represents a significant challenge to veterinary practitioners, and only occurs in approximately 50% of cases. In addition, the cost and time involved with such a diagnosis may be dissuasive to the owner (Knottenbelt 1996, Trock et al. 2001). According to our results, clinical signs did not allow practitioners to differentiate confirmed and non-confirmed WNV cases. At the present time, the definition of suspect cases to be declared cannot be improved and it remains valid to report all the cases of neurological syndromes of horses in the monitoring system. For emerging and re-emerging diseases, a passive monitoring system should be the core system used. For monitoring the WNV-associated diseases in the horse, we chose to establish a monitoring system of neurological syndromes reported by veterinary practitioners. Syndromic surveillance aims to identify a threshold number of early symptomatic cases, allowing detection of an outbreak some days or weeks earlier than would conventional re- TABLE 2. PERCENTILES OF THE DISTRIBUTION OF CASES PER WEEK IN THE STUDY AREA (INDICES) IN 2004, CRITICAL VALUES FOR THE ONSET AND END OF AN EPIZOOTIC, AND NUMBER OF CASES AND MEAN DURATION OF THE 2004 WEST NILE VIRUS (WNV) EPIDEMIC IN FRANCE Critical value Percentile of the number of cases/week for epidemic Epidemic 75% 90% 95% 99% Onset End Number of cases Mean duration Suspect weeks Confirmed weeks

6 408 A 18 Epidemic period LEBLOND ET AL Beginning of the clinical signs Syndromes declaration Number of horses True positive warning False positive warning 4 Critical Value for Warning Week number B FIG. 2. (A) Outbreak warning and syndromic surveillance by the veterinary practitioners. (B) Outbreak warning and case confirmation by the laboratory. porting of confirmed cases by the laboratory (Henning 2004). In the case of the 2004 WNV epizootic in Camargue, the syndromic surveillance system could provide a warning four weeks before the epidemic period. In comparison, the laboratory-case confirmation in horses could only provide a warning less than one week before the epizootic. The electronic

7 WEST NILE SYNDROMIC SURVEILLANCE IN HORSES 409 TABLE 3. SENSITIVITY, SPECIFICITY, AND POSITIVE PREDICTIVE VALUE OF EPIDEMIC WARNINGS FOR THE 2004 WEST NILE VIRUS (WNV) OUTBREAK CV W Se (%) Sp (%) PPV Suspect Confirmed CV W critical value for warning; Se, sensitivity; Sp, specificity; PPV, positive predictive value. recording proved essential for a timely and continuous recording of the suspicions all the year round. Non-equipped practitioners declared their suspicions from September to the end of October versus equipped practitioners declared their suspicions from August to July. In the future, the electronic monitoring should allow us to estimate the baseline number of neurological cases in the study area and improve the calculation of the Critical Warning Value. The validity of the surveillance system should be further evaluated by determining the level of the background occurrence of cases. The incidence of neurological diseases in horses is low. Studies conducted in France and in the USA showed a mortality risk associated with nervous disease of % per year in horses (Kaneene et al. 1997, Leblond et al. 2000). A better knowledge of the size and of the geographic distribution of the population should allow an assessment of the representativity of the system, and give an estimation of the incidence in smaller areas, different from the administrative units as communes. The stability of surveillance systems depends on voluntary or mandatory reporting, generally without compensation. In the future, tools have to be developed to meet the needs for feedback. Regular feedback would stimulate practitioners to provide timely input. Messages can be transmitted automatically when connections are established for data transmission (Martin et al. 1995). Guidelines have been proposed for determining whether a syndromic surveillance system provides reliable, useful information to decision-makers (CDC 2004). The balance between predictive value, sensitivity and timeliness should be assessed (Buehler 2004). Other attributes to be evaluated are the system s usefulness, acceptability, stability and cost-effectiveness. The usefulness of epidemic warning is related to the possibility of taking countermeasures against an impending epidemic, such as vector control or vaccination. A key problem is the definition of an epidemic. The critical values for the onset and end of an epidemic could vary with the disease, season, area, and so on. Finally, the definition of an epidemic warning should be based on the need for a public health response (Hashimoto et al. 2000). In our case, a warning four weeks before the epidemic period might be useful for the control. At this stage, the syndromic surveillance system should allow health authorities to prevent equine morbidity and mortality through the early detection of equine epizootics. Actions for vector control could be undertaken when the system yields an epidemic warning. Moreover, WNV-IgG antibody prevalence in the horse population is low ( 6%) and the vaccination of horses is forbidden, so horses could be considered as good sentinels for the WNV circulation in the area (Leblond et al. 2005). Further studies are still needed to evaluate if the system could be useful to prevent human disease. The attributes of the WNV monitoring system have to be compared with other systems available for outbreak detection. In the United States, in 1999 and 2000, equine cases did not precede WNV detection in mosquitoes or wild birds, nor did it predict human case occurrences. Therefore, at present, horses are unlikely to provide an early warning to WNV transmission in the United States (Trock et al. 2001). In France, since 2001, both passive and active monitoring systems for birds and mosquitoes have been implemented (Hars et al. 2001). In 2004, neither abnormal avian mortality, nor positive pools of mosquitoes were identified. A single seroconversion was detected in a duck that became positive in the end of August and was laboratory-confirmed in the 36th week, i.e., 3 weeks before the epidemic period in horses (Hars 2004). So, in this case, the system did not provide an alert before the syndromic surveillance in horses. Furthermore, a disadvantage of such active monitoring system is that the costs may become greater, as well as harder to justify when the target disease becomes rare (Doherr et al. 2001).

8 410 In conclusion, further evaluation of syndromic surveillance in horses should be undertaken in areas of the Old World where extensive populations of horses are maintained and WNV is a potential threat. ACKNOWLEDGMENTS We thank the CNES (Centre National d Etudes Spatiales), the Haras Nationaux and the DGAl (General Direction of Food, Ministry of Agriculture) for their financial support; the Society Calystene for their scientific and technical help; and the Pasteur Institute of Lyon (CNR Arboviroses), AFSSA (French Agency for Food Security and Hygiene), the veterinary practitioners, and the owners for their help in the collection and analysis of serum samples. REFERENCES Blondel, J, Isenmann, P. [Guide to the Birds of Camargue]. Lausanne: Delachaux and Niestle; Buehler, JW. Review of the 2003 National Syndromic Surveillance Conference lessons learned and questions to be answered. MMWR Morb Mortal Wkly Rep 2004; 53(Suppl): Castillo-Olivares, J, Wood, JLN. West Nile Virus infection of horses. Vet Res 2004; 35: CDC (Centers for Disease Control and Prevention). Epidemic/Epizootic West-Nile Virus in the United States: Guidelines for Surveillance, Prevention and Control, 3rd rev. Atlanta: CDC; CDC (Centers for Disease Control and Prevention). Framework for evaluating public health surveillance systems for early detection of outbreaks: recommendations from the CDC working group. MMWR Morb Mortal Wkly Rep 2004; 53:RR-5. Doherr, MG, Audigé, L. Monitoring and surveillance for rare health-related events: a review from the veterinary perspective. Trans R Soc Lond B 2001; 356: Gallian, P, de Lamballerie, X, de Micco, P, Andreu, G. [West Nile Virus: implication and generalities in blood transfusion]. Transf Clin Biol 2005; 12: Guptill, SC, Julian, KG, Campbell, GL, Price, SD, et al. Early-season avian deaths from West Nile virus as warnings of human infection. Emerg Infect Dis 2003; 9: Hars, J. Surveillance de l Avifaune en 2004 dans la Petite et Grande Camargue Rapport Final. Montpellier: Office National de la Chasse et de la Faune Sauvage; Hars, J, Auge, P, devisscher, M-N, Keck, N, et al. An epidemic of West Nile fever in the south of France. Results of an epidemiologic survey on wild birds. Presented at LEBLOND ET AL. the Wildlife Disease Association Annual Conference, Kwa Maritane, South Africa, Hashimoto, S, Murakami, Y, Taniguchi, K, Nagai, M. Detection of epidemics in their early stage through infectious disease surveillance. Int J Epidemiol 2000; 29: Henning, KJ. Overview of syndromic surveillance what is syndromic surveillance? MMWR Morb Mortal Wkly Rep 2004; 53(Suppl):5 11. Hoffmann, L, Mouchet, J, Rageau, J, Hannoun, C, et al. [Epidemiology of the West Nile virus: study of an outbreak in Camargue. II. Outline of the physical, biological and human environment]. Ann Inst Pasteur (Paris) 1968; 114: Kaneene, JB, Ross, WA, Miller, RA. The Michigan equine monitoring system. II. Frequencies and impact of selected health problems. Prev Vet Med 1997; 29: Knottenbelt, DC. Equine neurological disease and dysfunction: a diagnostic challenge for the practitioner. Part 1: objectives and limitations of a neurological examination. Equine Vet Educ 1996; 8: Komar, N. West Nile Virus Surveillance using Sentinel Birds. Ann NY Acad Sci 2001; 951: Lazarus, R, Kleinman, P, Dashevsky, I, DeMaria, A, et al. Using automated medical records for rapid identification of illness syndromes (syndromic surveillance): the example of lower respiratory infection. BMC Public Health 2001; 1: /1/9. Leblond, A, Villard, I, Leblond, L, Sabatier, P, et al. A retrospective evaluation of the causes of death of 448 insured fench horses in Vet Res Commun 2000; 24: Leblond, A, Zientara, S, Chadoeuf, J, Comby, N, et al. Prévalence de l infection par le virus West Nile chez le cheval en Camargue. Rev Med Vet 2005; 156: Lewis, MD, Pavlin, JA, Mansfield, JL, O Brien, S, et al. Disease outbreak detection system using syndromic data in the greater Washington DC area. Am J Prev Med 2002; 23: Martin, SM, Bean, NH. Data management issues for emerging diseases and new tools for managing surveillance and laboratory data. Emerg Infect Dis 1995; 1: Petersen, LR, Roehrig, JT. West Nile virus: a reemerging global pathogen. Emerg Infect Dis 2001; 7: Snook, CS, Hyman, SS, Del Piero, F. West Nile virus encephalitis in eight horses. J Am Vet Med Assoc 2001; 218: Trock, SC, Meade, BJ, Glaser, A, Ostlund, EN, et al. West Nile virus outbreak among horses in New York State, 1999 and Emerg Infect Dis 2001; 7: Address reprint requests to: Dr. Agnès Leblond UMR CNRS 5525 TIMC Ecole Nationale Vétérinaire de Lyon 1 Ave. Bourgelat Marcy l Etoile, France a.leblond@vet-lyon.fr