Article Fctors ssocited with West Nile virus disese ftlities in horses Tsh Epp, Cheryl Wldner, Keith West, Hugh Townsend Astrct In 2003, the occurrence nd loction of horses with clinicl signs of West Nile virus infection were identified in the southern portion of Ssktchewn with the help of veterinrins, owners, nd the regionl lortory. A totl of 133 clinicl cses were reported etween July 30 nd Septemer 19, 2003; however, postseson surveillnce suggests tht the numer of cses ws underestimted. The cse ftlity rte ws 43.8% (95% CI 35.2, 52.4). Fctors ssocited with ftlity in clinicl cses included sex, week of onset of clinicl signs, nd cot color. Reported clinicl cses clustered within regionl helth uthority districts, suggesting regionl differences in geogrphic fctors, potentilly including climte nd mosquito control, tht could contriute to the risk of disese. However, most of the vrition in the risk of ftlity in clinicl cses is explined t the individul level rther thn the Regionl Helth Authority level, which suggests the outcome of clinicl disese is primrily determined y chrcteristics of, or mngement fctors ffecting, the individul horse. Résumé Fcteurs ssociés à l mortlité reliée u virus du Nil occidentl chez les chevux. En 2003, l pprition des signes cliniques d infection u virus du Nil occidentl chez les chevux et le lieu où ils étient situés ont été identifiés dns le sud de l Ssktchewn vec l ide des vétérinires, des propriétires et du lortoire régionl. Un totl de 133 cs cliniques ont été rpportés entre le 30 juillet et le 19 septemre 2003 et les oservtions de fin de sison lissent présger que le nomre de cs été sous-estimé. Le tux de mortlité étit de 43,8 % (95 % IC 35,2, 52,4). Les fcteurs ssociés à l mortlité dns les cs cliniques comprenient le sexe, l semine du déut des signes cliniques et l couleur de l roe. Les cs cliniques rpportés étient regroupés dns certins districts des utorités régionles de l snté, lissnt présger des différences régionles dns les fcteurs géogrphiques, dont le climt et le contrôle des moustiques, qui urient pu ffecter le fcteur de risque. Cependnt, l mjorité de l vrition du risque de mortlité dns les cs cliniques semle relié à l individu plutôt qu ux utorités régionles de l snté, ce qui lisse présger que l pprition des cs cliniques est en mjeure prtie déterminée pr des crctéristiques propres à l niml ou pr des fcteurs de gestion qui l influence. (Trduit pr Docteur André Blouin) Cn Vet J 2007;48:1137 1145 Introduction West Nile virus (WNV) is n rovirus (rthropod-orne virus) tht ffects the nervous system of humns, horses, nd irds, cusing mild to severe illness nd sometimes deth (1). It ws first reported in Afric in 1937 (2) nd hs een identified spordiclly in Europen countries from the erly 1960s to Deprtment of Lrge Animl Clinicl Sciences (Epp, Townsend, Wldner), Western College of Veterinry Medicine, University of Ssktchewn, Ssktoon, Ssktchewn S7N 5B4; Pririe Dignostic Services (West), Ssktoon, Ssktchewn S7N 5B4. Address ll correspondence to Dr. Tsh Epp; e-mil: tsh.epp@ussk.c Reprints will not e ville from the uthors. The uthors grtefully cknowledge funding nd support from the Western College of Veterinry Medicine Equine Helth Reserch Fund, Ssktchewn Helth, nd Wyeth Animl Helth. present time (3 5). Since eing dignosed in New York stte in 1999, WNV hs spred cross the North Americn continent (6 9). In Cnd, WNV infection ws first seen in irds in southern Ontrio in 2001 (10). Since 2001, throughout Cnd, the dte nd loction of irds found ded tht tested positive for WNV were systemticlly recorded y the Cndin Coopertive Wildlife Helth Centre. In 2002, the first equine cses (those showing clinicl signs consistent with WNV infection) were identified in Ontrio, Mnito (MB), nd Ssktchewn (SK) (10,11). In SK, there ws only pssive system to monitor the occurrence of clinicl disese in horses, which, unfortuntely, resulted in the numers reported vrying, depending on the source of informtion. The virus is mplified in nturl cycle etween irds nd mosquitoes nd incidentlly infects humns nd horses. In western Cnd, Culex restuns nd Culex trslis re the mjor mosquito species responsile for trnsmission in the mplifiction cycle. Culex trslis is considered the most importnt vector for trnsmission of WNV to humns nd horses (12). Horses CVJ / VOL 48 / NOVEMBER 2007 1137
cn ecome infected when itten y mosquito crrying the virus, ut they do not contriute to the spred or mplifction of the virus in the nturl cycle (13). Once itten, horse my eliminte the virus uneventfully or show clinicl signs including fever, depression, muscle tremors, wekness, lck of coordintion, inility to rise, nd prlysis (14). The incution period of infection to mnifesttion of clinicl signs is 5 15 d (15). Most reports suggest tht round 25% to 45% of those horses tht show clinicl signs die or require euthnsi (4 8,14). Fctors significntly ssocited with ftlity from WNV clinicl disese include nonvccinted sttus for WNV; ge; clinicl signs, such s inility to rise; erly seson onset of clinicl disese; sex; nd reed (8,16). Horses given either 1 or 2 doses of vccine (even if it ws not given ccording to the mnufcturer s recommendtions) seemed less likely to die thn unvccinted horses (8,16). In 1 of these studies, deth ws more likely if clinicl signs occurred first from Mrch 1 to August 19, the niml ws femle, the niml ws over 18 y of ge, or ws unle to rise t ny time during clinicl illness (8). Given the pttern of WNV spred cross the continent, the numer of horses in SK showing clinicl disese ws expected to e higher in 2003 thn in 2002. The 1st ojective of this study ws to document the dte of onset, geogrphic loction, nd outcome of clinicl cses of WNV infection in horses in SK in 2003. The primry ojective ws to identify risk fctors tht could e ssocited with ftlity in infected horses, including mngement, mosquito control, vccintion, individul horse chrcteristics, nd environmentl conditions. Mterils nd methods Study popultion nd protocol Veterinrins in privte prctice, horse owners, nd the stff of the regionl dignostic lortory [Pririe Dignostic Services (PDS), Ssktoon, SK] provided informtion on cliniclly ffected horses in SK during the summer nd erly fll of 2003. In June 2003, ll lrge nd mixed niml prctices in the southern portion of SK were contcted y fcsimile nd sked 1) to send ny serum smples collected from horses where WNV infection ws suspected to PDS in Ssktoon for immunogloulin (Ig)M enzyme-linked immunosornt ssy (ELISA), nd 2) to llow the lortory to notify the reserch tem of ny positive result from the smples sumitted. Of the 79 veterinry clinics contcted, 51 responded to the fcsimiles signed in June 2003. Twelve more prctices were contcted y PDS t the time sumitted smple ws declred positive, giving totl coopertion rte of 80% (63/79). For the purposes of this study, the definition of clinicl cse ws the presence of 1 or more of the clssic signs of WNV clinicl disese (txi, recumency, presis, or prlysis) or deth; nd positive IgM ELISA. Once clinicl cse ws identified, the veterinrin ws contcted to otin consent to contct the owner. If consent ws given, owners were telephoned 2 to 4 wk fter the onset of clinicl signs nd sked for dditionl informtion. The telephone survey included questions out loction, chrcteristics of the ffected horse, mngement of tht horse, vccintion sttus, other resident equids (horses, ponies, mules, donkeys), nd mosquito control mesures pplied t oth the frm nd the individul niml level. In Octoer, fter the mosquito seson, lrge or mixed niml veterinry clinics in ech Regionl Helth Authority (RHA) were contcted to estimte the numer of untested prole cses of WNV infection per region (postseson survey). A rndom numers procedure ws used to rnk ll of the veterinry clinics in ech RHA. Clinics from ech RHA were then contcted y fcsimile in rndom order until t lest 1/3 hd responded. The prticipting veterinrins were sked to estimte the numer of horses tht were known or reported to hve shown clinicl signs consistent with WNV ut were not tested for IgM ntiodies during the 2003 seson. Horse dt The legl lnd loction of the clinicl cse t the estimted time of exposure (up to 2 wk efore the 1st clinicl signs) ws requested for ll cses where follow-up ws otined. Legl lnd loction included qurter, section, township, rnge, nd meridin. In one instnce, glol positioning system (GPS) unit ws used to otin the ltitude nd longtitude. For those cses in which the owner ws unsure or unwilling to give the loction, directions from the nerest town or township/rnge dt were collected. This informtion ws cross-referenced with n electronic mp of SK. By using commercil geogrphicl informtion systems (GIS) softwre (ArcView GIS version 3.2 nd ArcGIS version 8.0; Environmentl Systems Reserch Institute, Redlnds, Cliforni, USA), sptil loction ws displyed s point y using the centroid-of-the-polygon for the lnd loction. Confidentility ws mintined in ll pulictions y summrizing loction dt t the RHA level. Sttistics Cnd s 2001 Census of Agriculture estimted the numer of horses nd ponies (horse dt) in SK to e pproximtely 71 300 nd provided horse dt y Census Consolidted Sudivisions (CCS). The CCS hs similr oundries to rurl municiplities (RM), llowing estimtion of the horse popultion y RHA (ggregtions of RMs) nd the mpping of cses per 1000 horse popultion y RHA. Dt provided y the postseson survey were ggregted nd mpped y RHA to compre nd ssess the ccurcy of surveillnce, using only confirmed cses of WNV for surveillnce in 2003. The nimls were clssified y sex, color, ge, nd reed type, ut not y specific reed. Breed type ws divided into light horses, drft horses nd crosses, ponies, nd other. The ctegory of other ws composed of 3 miniture horses nd 1 mule. Sex ws reported in 3 ctegories: mre, gelding, nd stllion. Color ws divided into 3 ctegories: drk (chestnut, sorrel, rown, lck, or y), light (gry, white, plomino, uckskin, or dun), nd multicolored (ron, pint, pinto, nd pploos). Age ws clssified into 3 ctegories: young (# 5 y), dults (6 18 y), nd geritrics (. 18 y). Age ws ctegorized ecuse the odds of outcome ws not thought to increse on liner scle nd so tht the results were comprle with those of other studies (8). Dy of onset of clinicl signs ws ggregted into week of onset with Sundy s the strt of the week. To llow for esier interprettion, week of onset ws further ctegorized into weeks 1138 CVJ / VOL 48 / NOVEMBER 2007
Tle 1. Postseson survey results nd cse ftlity rtes y Regionl Helth Authority (RHA) Surveyed clinics Rtio Cse (totl numer of (estimted Confirmed ftlity clinics in the Totl Estimted proles: cses rtes RHA RHA) proles proles confirmed) ( ) c (%) d Sun Country 4 (11) 20 55 5.5:1 10 50 Five Hills 3 (10) 58 193 16:1 12 75 Cypress 3 (5) 36 60 3:1 17 (1) 47 Regin Qu Appelle 4 (10) 40 100 5.6:1 18 39 Sunrise 2 (6) 21 63 4.8:1 13 23 Ssktoon 5 (8) 7 11 0.5:1 22 41 Hertlnd 3 (6) 8 16 1.6:1 10 30 Kelsey Tril 1 (5) 2 10 2:1 4 (1) 75 Prince Alert Prklnd 1 (4) 12 48 6:1 8 63 Pririe North 2 (7) 21 74 4.4:1 16 (1) 31 Totls 28 (72) 225 630 4.8:1 130 (3) 44 From surveyed clinics Bsed on proportion of surveyed clinics c Numers in prentheses re cses with no further follow-up otined d Cse ftlity rtes re clculted on confirmed cses only of the yer, 31 33 (eginning of August), weeks 34 35 (epidemic pek, end of August), nd weeks 36 38 (Septemer). Housing ws clssified s housed in n enclosed rn, ccess to simple 3-sided shelter with corrl or psture, or no ccess to shelter in open psture with or without trees. The method of mosquito control on frm ws recorded s smudges (smoky fire used to wrd off mosquitoes), wter removl, seled rn, use of fns or specil lighting in rn, use of generl insecticides, or no mosquito control ttempted. The use of insecticides sprys on individul horses ws recorded s yes (ny mount or frequency) or no. Mosquito dt Mosquito dt were otined from the 2003 Ssktchewn Helth mosquito trpping progrm. Center for Disese Control (CDC) trps were used for the collection of live mosquitoes for virus testing, nd New Jersey Light trps (NJLT) were used for the collection of ded mosquitoes for the determintion of mosquito species. Culex trslis nd C. restuns dt were provided on weekly sis from June to Septemer s the verge numer of mosquitoes trpped per night. Vriles of interest were () the highest weekly verge for C. trslis from the whole seson (numer of C. trslis), () the highest weekly verge for C. restuns from the whole seson (numer of C. restuns), (c) whether WNV ws found in pooled mosquitoes of ech species (positive pool), nd (d) the weekly verge for C. trslis from the likely week of exposure for ech cse. For ech vrile of interest, the results of the nerest mosquito CDC nd NJLT trp were linked to ech horse cse. Figure 1. Numer of clinicl cses of West Nile virus infection in horses y week of onset of clinicl signs in 2003. The rs represent the totl numer of cses y week. The lines represent the numer of cses y helth outcome per week. Weeks egin on Sundy nd end on Sturdy (Week 31 runs from July 27 to August 2. Weeks 31 35 correspond to the end of the month of July nd the whole month of August 2003, nd weeks 36 39 correspond to the month of Septemer 2003). Environment dt Dily verge temperture nd precipittion dt for ll climte sttions in SK were otined from Environment Cnd. These vlues were used to crete seson verge for temperture nd seson totl for precipittion. These vriles were clculted for ech climte sttion, using dily vlues for June, July, nd August 2003. The seson verge temperture ws determined y dividing the sum of the dily tempertures y the numer of dys; the seson totl for precipittion ws the sum of the dily mounts of precipittions. The vriles from the nerest climte sttion were linked to ech horse cse. Cumultive growing degree dys (GDD) re the sum of the positive dily differences etween 16 C (the threshold temperture for ctivity of C. trslis nd virus trnsmission) nd the men temperture for tht dy (17). Seson GDD ws clculted for ech climte sttion y using temperture dt from My 1 to Septemer 30. Cses were ssigned dt from the nerest climte sttion with complete dily temperture nd precipittion informtion. Sttisticl nlysis The ssocition etween ech risk fctor nd the finl helth outcome (ftlity) ws exmined y using generlized liner mixed models with inomil distriution nd logit link function (18). The clcultions were performed y using penlized CVJ / VOL 48 / NOVEMBER 2007 1139
Figure 2. Confirmed positive IgM clinicl cses of West Nile virus infection per 1000 horses t risk reported y Sttistics Cnd in ech Regionl Helth Authority (RHA). Figure 3. Comined numer of dignosed positive IgM clinicl cses nd undignosed prole clinicl cses per 1000 horses t risk reported y Sttistics Cnd in ech Regionl Helth Authority. qusi-likelihood estimtes nd the 2nd order of the Tylor series expnsion (MLwiN version 2.0; Centre for Multilevel Modeling, Institute of Eduction, London, UK). The strength of the ssocition etween outcome nd exposure ws reported s odds rtio (OR) with 95% confidence intervls nd P vlues. A 2-level hierrchicl (RHA, Horse) logistic regression model ws used to ccount for geogrphic clustering of oservtions nd ecuse the method of surveillnce used clinics within RHAs (18). The linerity ssumption ws ssessed for ll continuous risk fctors considered in the modeling process. Exposures were renlyzed fter eing clssified into qurtiles nd the linerity ssumption ws exmined y looking for n incresing (or decresing) series of coefficients. The effect of exposure in ech higher ctegory, s determined y the odds rtio, ws compred with the level of the outcome in the seline or referrent exposure ctegory to identify ptterns consistent with liner, monotonic, or threshold effect. All exposure vriles where the ssocition with the odds of ftlity ws significnt t P, 0.25 were considered in developing the finl multivrile model. Nonexposure risk fctors were defined s confounders, if removing or dding the fctor chnged the effect estimte for the exposure y more thn 10%, in which cse, the fctor ws retined in the finl model. Mnul ckwrds elimintion of vriles ws used to chieve finl model contining only the sttisticlly significnt exposure vriles nd ny nonexposure risk fctors tht were either significnt or cted s importnt confounders. Biologiclly resonle interctions were ssessed etween significnt risk fctors (P, 0.05) in the finl model. An pproximtion method sed on ltent vriles ws used to estimte the intrclss (intrregion) correltion coefficient [r = 2 r /( 2 r 1 p2 /3)] to exmine the clustering of clinicl cses of WNV within n RHA (18). Results Onset dte nd loction The dte of onset of clinicl signs for the 1st WNV IgM- positive clinicl cse ws July 30, 2003. The lst recorded dte of onset of clinicl signs in horse ws Septemer 19, 2003. Of the 130 horses for which questionnire ws completed out of the totl 133 clinicl cses, 57 died or were euthnized ecuse of complictions ssocited with clinicl disese (43.8% cse ftlity rte, 95% CI 35.2, 52.4) nd the reminder recovered completely. Cse ftlity rtes differed y RHA (Tle 1). The epidemic curve ppered to pek 1 wk erlier for horses tht died from WNV clinicl disese thn for horses tht survived, suggesting tht there could e difference in risk of ftlity ssocited with time of onset (Figure 1). This oservtion ws explored in susequent nlyses. For ll 133 confirmed clinicl cses, loction ws reported t the RHA level, sed on the loction of the owner (n = 130) or of the sumitting clinic (n = 3). For the 130 clinicl cses with further follow-up, exct legl lnd loction ws otined for 1140 CVJ / VOL 48 / NOVEMBER 2007
Tle 2. Summry of individul risk fctors y survivl outcome nd cse ftlity rtes (n = 130) Helth outcome Cse Cses ftlity Odds Vrile (n) Ded Alive rte rtio 95% CI Week of onset 31 33 27 17 10 63% 4.6 1.7, 12.5 34 35 53 26 27 49% 2.5 1.1, 5.6 36 38 50 14 36 28% ref Age, 5 yers 44 22 22 50% ref 6 18 yers 69 28 41 41% 0.7 0.3, 1.5. 18 yers 17 7 10 41% 0.7 0.2, 2.2 Gender Gelding 60 27 33 40% 1.4 0.7, 3.0 Mre 59 21 38 36% ref Stllion 11 9 2 82% 8.5 1.6, 43.5 Breed Pony 6 2 4 33% 0.6 0.1, 3.6 Light 106 48 58 45% ref Drft 11 4 7 36% 0.9 0.2, 3.3 Cross 3 2 1 67% 2.6 0.2, 29.8 Other 4 1 3 25% 0.4 0.04, 4.0 Color Drk 89 32 57 36% ref Light 19 14 5 74% 4.4 1.5, 13.1 Multi-color 16 8 8 50% 1.9 0.6, 5.4 Individul shelter No shelter 86 42 44 49% 2.7 1.2, 6.1 Simple shelter 41 29 12 71% ref Individul No 108 51 57 47% 2.3 0.7, 7.4 insecticide use Yes 19 5 14 26% ref Vccine No vccine 121 53 68 44% 1.0 0.3, 3.9 Vccinted c 9 4 5 44% ref Primry use Plesure 108 49 59 45% ref Breeding 13 6 7 46% 1.6 0.5, 4.9 Frm work 5 2 3 40% 1.3 0.3, 5.4 Competition 4 0 4 0% d Herd size 10 or less 92 53 39 58% ref 11 25 21 10 11 48% 1.1 0.4, 2.5 25 or more 13 8 5 62% 1.1 0.5, 2.7 Frm mosquito None 97 42 55 43% 0.9 0.4, 2.1 control Used t lest one 29 13 16 45% ref Univrile nlysis djusted for sptil loction (RHA) y inclusion of single rndom effect ref = reference ctegory c Vccinted refers to fully vccinted horses with the stndrd 2 vccintions. Those horses with only 1 vccintion re considered nonvccintes d Not enough dt entries to converge 117 cses (90%), exct GPS coordintes for 1 (1%), nd prtil legl lnd loction (township nd rnge) for 12 (9%). Clinicl cses of WNV infection were reported in ll 10 RHAs in the southern portion of the province (Figure 2). When the rw dt were djusted for potentil regionl differences in reporting, sed on the results of the postseson survey, the RHA with the highest numer of clinicl cses per 1000 horses ws in the south centrl portion of the study re (Figure 3). Postseson results differed y RHA (Tle 1), with the overll numer of estimted untested prole clinicl cses exceeding tht of the tested clinicl cses y pproximtely 4.8 to 1. Description of ffected horses Most clinicl cses occurred in horses # 10 y; the ftlity rte for this ge group ws 46% (38/83). The oldest horse tht survived ws 25 y old, while the oldest horse tht died ws 32 y old. Most horses tht died (s opposed to eing euthnized) did so during the month of August (11/12 or 92%). The ftlity rte ws 36% (21/59) for mres, 45% (27/60) for geldings, nd 82% (9/11) for stllions. Drk colored horses were most commonly ffected (89/124), ut they hd lower ftlity rte (36%) thn did those in other color clssifictions. Most of the clinicl cses were light horse reeds nd most were used for plesure riding (Tle 2). Of the 130 clinicl cses with follow-up, only 9 were reported to hve een fully vccinted ccording to the mnufcturer s recommendtions (2 doses given 3 to 6 wk prt, t lest 2 to 3 wk efore the pek mosquito seson), while 3 were vccinted only once (Tle 2). Initil vccintion dtes for the 9 fully vccinted horses were Ferury (n = 2) nd April (n = 7). The vccintion dtes for the 3 horses given only 1 dose were My, August, nd Septemer. Of the 9 fully vccinted horses, 5 recovered, while 4 died or were euthnized (44% ftlity rte). None of the 3 horses tht were vccinted once died or were euthnized. Five of the 130 (3.8%) horses hd een tken more thn 15 km (10 miles) from their home loction in the 3 wk efore onset of clinicl signs. Of these, 2 styed within the sme RHA, 2 trvelled to different RHA in SK, nd 1 trvelled to BC. In 2003, BC hd no evidence of WNV infection. Five horses hd chnge of residence 2 3 d efore onset of clinicl signs. CVJ / VOL 48 / NOVEMBER 2007 1141
Tle 3. Summry of individul environmentl dt (continuous vriles) y helth outcome. These vriles were clculted y using only those climte sttions or mosquito trps for which the dt ws mtched up with the nerest cse Helth Cses Temperture ( C) Overll Odds rtio Vrile outcome (n) Min Mx Men Men ( C) (95% CI) Seson verge temperture Ded 57 16.7 20.3 18.7 18.7 1.0 (0.7, 1.5) ( C/dy) Alive 73 16.4 20.3 18.6 Seson totl precipittion Ded 57 39.2 206 118 123 1.0 (1.0, 1.0) (mm) Alive 73 39.2 215 128 Seson GDD Ded 57 199 490 350 357 1.0 (1.0, 1.0) (degree dys) Alive 73 178 507 355 Exposure week verge Ded 57 14.8 26.5 21.3 20.9 1.1 (1.0, 1.2) temperture Alive 73 14 26.5 20.6 ( C/dy) Exposure week Ded 57 0 44.6 7.65 7.30 1.0 (0.9, 1.0) totl precipittion Alive 73 0 39.4 6.85 (mm) Exposure week Ded 57 0 4.80 0.38 0.5 0.8 (0.5, 1.3) C. trslis Alive 73 0 4.40 0.51 (numer per night) Numer of C. trslis Ded 57 0 10.40 1.67 1.9 0.9 (0.8, 1.1) (numer per night) Alive 73 0 15.00 1.98 Numer of C. restuns Ded 57 0 6.00 0.41 0.3 1.4 (0.8, 2.4) (numer per night) Alive 73 0 2.60 0.25 In this tle Growing degree dys (GDD) is the cummultive growing degree dys for the whole seson (My to Septemer), seson verge temperture is clculted with dily vlues from June to August divided y the totl numer of dys nd seson totl precipittion is the cumultive dily dt from June to August Univrile nlysis djusted for sptil loction (RHA) y inclusion of single rndom effect The loction t which they were most likely exposed (given n incution period of 5 15 d) ws, therefore, recorded s the loction of the previous residence. Most of the clinicl cses were in the sme RHA s the veterinry clinic tht sumitted the lood smple (94%). No horses were housed in rn for ny period of time nd most horses (68%) did not hve ccess to shelter. Nineteen of the horses hd een spryed periodiclly with insecticide during the mosquito seson. None of the horses were lnketed for ny period of time. The mjority of clinicl cses were kept on frms tht did not use ny on-frm mosquito control methods (97/126). For horses on frms tht did use t lest 1 method, the primry choice ws smudges (18/29, 62%). Seven of the 8 horses on frms tht used wter removl survived nd 9 of the 18 horses on frms tht used smudges survived (Tle 2). Herd size rnged from single horse to 300 horses. The mjority of horses were in smll herds of 10 or fewer horses (73%). On most of the frms, ll horses were kept under the sme conditions; however, t 1 frm, the horses were housed in rn, with the exception of the cse horse. There were 5 frms tht hd 2 clinicl cses nd 5 frms tht hd t lest 1 undignosed neurologic cse, in ddition to the tested cse. Description of environmentl nd mosquito dt for the reported cses The men seson verge temperture ws 18.7 C nd the men seson totl precipittion ws 123 mm (Tle 3); oth of these vlues differed y climte sttion nd y region. The men seson GDD ws 357. There ws no ovious or sttisticlly significnt difference etween these vlues for those horses tht died nd those tht survived. Averge temperture nd totl precipittion for the week prior to the estimted exposure period were lso clculted, ut, gin, there ws no pprent or sttisticlly significnt difference etween vlues for those horses tht died nd those tht survived (Tle 3). The medin distnce etween horse cses nd the nerest climte sttion ws 21 km (min, mx = 2, 49 km). Vriles summrizing mosquito dt were similr for those horses tht died or survived. The highest numers of C. trslis nd C. restuns over the whole seson were 1.9 nd 0.3, respectively. The numer of C. trslis mosquitoes during the estimted exposure week of ech cse rnged from 0 to 4.8 (men, 0.5) (Tle 3). Culex trslis numers were higher for those horses tht survived, while C. restuns numers were higher for those horses tht died; however, the comprisons were not sttisticlly significnt. A WNV-positive C. trslis nd C. restuns mosquito pool ws found in the nerest CDC trp for 54% (70/130) nd 39% (51/130) of clinicl cses, respectively (Tle 4). The medin distnce etween horse cses nd the nerest mosquito trp ws 29 km (min, mx = 4, 92 km). Assocition etween individul horse risk fctors, environmentl vriles, nd the odds of ftlity in clinicl cses of WNV Vriles where there ws sttisticlly significnt ssocition with ftlity in the initil univrile mixed models tht djusted for clustering y RHA included week of onset, cot color, gender, nd vilility of individul shelter. Horses ffected in August hd 3.1 times (OR 95% CI 1.4, 6.8) greter odds of 1142 CVJ / VOL 48 / NOVEMBER 2007
Tle 4. Summry of individul environmentl dt (ctegoricl vriles) y survivl outcome Cses Helth outcome Odds rtio Vrile (n) Ded Alive (95% CI) Positive pool no 60 28 32 ref c Culex trslis yes 70 29 41 0.8 (0.4, 1.7) Positive pool no 79 35 44 ref c Culex restuns yes 51 22 29 1.0 (0.5, 2.0) Univrile nlysis djusted for sptil loction (Regionl Helth Authority) y inclusion of single rndom effect A positive pool of mosquitoes (collection of species of mosquito designted) t the nerest to the trp loction of the cse horse c ref = reference ctegory dying thn those ffected in Septemer. The month of August ws further roken down into 2 ctegories, weeks 31 33 nd weeks 34 35. The odds of dying in oth weeks 31 33 nd weeks 34 35 (August) were higher thn the odds of dying in weeks 36 38 (Septemer) (Tle 5). The odds of dying ws greter for light colored horses thn for drk colored horses, while multicolored horses were t similr risk to drk colored horses (Tle 5). The odds of dying for mles (stllions nd geldings comined) ws 1.9 times greter (95% CI 0.9, 3.8) thn the odds of dying for femles (Tle 5). Specificlly, the odds of dying for stllions ws 8.5 times greter (95% CI 1.6, 43.5) greter thn the odds of dying for femles (Tle 5). The odds of dying ws greter for those with no shelter thn for those with simple shelter. The finl mixed model included week of onset, gender, nd cot color (Tle 5). Shelter did not remin sttisticlly significnt in the presence of the other vriles nd ws removed from the finl model. The finl model ws mixed model with rndom intercept to ccount for geogrphic clustering with RHA. The vrince ccounted for t the region level ( 2 r ) ws 0.35 (stndrd error, 0.36) or pproximtely 9.5% of the overll vrince in the model. Discussion West Nile virus infection nd the development of ssocited clinicl disese re influenced y mny fctors, including environment, mosquito popultions, nd individul susceptiility (3 5,9). The onset dtes of WNV clinicl disese reflect the time necessry for development of climtic conditions required for mosquito reproduction, nturl mplifiction of the virus, nd shift in mosquito feeding ptterns to llow trnsmission of the virus to horses (12). The loction of the cses corresponded to res where mosquito popultions cple of trnsmitting WNV nd susceptile horse popultions coexist. Distriution of clinicl cses Trvel history ws essentil in determining where horses were potentilly exposed to WNV. However, s few horses hd trvel histories, exposure could e ssumed to hve occurred ner their plce of residence, even without complete trvel history. Clinicl cses nd the sumitting clinic were within the sme RHA. Thus, on the sis of this study, we could ssume WNV ctivity in the loclity of either the sumitting clinic or of the Tle 5. Risk fctors ssocited with survivl outcomes in horses with clinicl infection with West Nile virus (WNV). Finl mixed multivrile model Cses Vrile Ctegory (n) Odds rtio 95% CI P Week of Onset Week 31 33 27 6.3 1.9, 20.4 0.003 Week 34 35 53 2.4 0.9, 6.2 0.08 Week 36 38 50 ref Gender Mre 59 ref Gelding 60 1.2 0.5, 2.9 0.70 Stllion 11 14.7 2.5, 87.4 0.005 Cot Color Drk 89 ref Light 19 6.4 1.9, 21.6 0.004 Multicolor 16 1.9 0.6, 6.5 0.31 Adjusted for Regionl Helth Authority (RHA) y the inclusion of single rndom effect ref = reference ctegory individul horse, depending on the degree of ville loction ccurcy. Most clinicl cses were recorded in the 4 most southerly RHAs, while the fewest occurred in the most northerly RHAs of the study re. This distriution corresponds minly to differences in climte nd mosquito popultions; however, veterinrins nd horse owners locted in the Ssktoon RHA might hve een more inclined to pursue veterinry services nd dignosis, s this is the loction of the PDS. No clinicl cses were dignosed from the northern portion of the province (outside the study re). This ws proly ecuse hitt nd climte do not fvor the principl vector, Culex trslis (12). The postseson survey of veterinry clinics suggested tht there were numerous cses of neurologic symptoms in horses for which owners did not seek dignosis. Dignosis y clinicl signs lone could e mistken for other neurologicl conditions in horses, such s ries, woler syndrome, equine herpesvirus encephlomyelopthy (EHV1), nd western equine encephlomyelitis (14); however, ll of these re considered reltively unlikely in this region. Thus, typicl neurologic symptoms in n re known to hve ctive WNV re resonly predictive of WNV clinicl disese presenttion. A mixed model ws used to ccount for regionl distriution of clinicl cses, ecuse WNV hs een shown to cluster y geogrphic re in other studies, including 1 recent study tht identified sttisticlly significnt clusters of oth horse nd humn cses of WNV in 2003 (9,19,20). In the current study, RHA ws chosen s the unit to represent geogrphic loction in the mixed model, ecuse it ws resonle geogrphic proxy for the veterinry clinics tht were the min source of our informtion. Individul veterinry clinics were not used in the nlysis, s most of these only smpled 1 or 2 cses. The oundries of the RHAs often reflect ptterns of trvel from rurl res to lrger centers for helth services, nd sed on the mps, pper to reflect resonle pttern of ccess to, nd use of, veterinry services. In ddition, cse ftlity rtes differed y RHA, with the highest cse ftlity rte nd highest postseson survey rtio in the Five Hills RHA. Decisions on the form nd timing of pulic eduction cmpigns relted to WNV nd mosquito control progrms re often mde t the level of the RHA or y rurl municiplities locted within the RHA. CVJ / VOL 48 / NOVEMBER 2007 1143
Risk of ftlity for clinicl cses The only significnt predictors of risk of ftlity in the finl model were week of onset of clinicl signs, sex, nd color, while controlling for region. A smll mount of vrition (9.5%) ws ccounted for y geogrphic region in the finl mixed model (18). This could reflect vriility of the RHA mosquito progrms nd other inherent differences etween the regions. Differences in clinic policy could hve influenced the vrition, ut, overll, most clinics followed similr guidelines for dignosis or tretment. Therefore, progrms to control the outcome of clinicl disese should e directed towrd the individul horse. Unfortuntely, the vriles found to e significnt in this study re difficult to mnge for disese control purposes. The cse ftlity rte ws similr to rtes from other studies conducted in North Americ (6 8). The incresed survivl rte of horses with clinicl signs s the seson progressed ws similr to findings in the western United Stted (US). Slzr et l (8) found tht, in 2003, horses were 1.7 times more likely to die if they showed clinicl signs erly (efore August 19) in the seson. The incresed risk of deth in August could e result of erlier recognition of clinicl cses s the seson progressed. Owners my hve checked their horses for clinicl signs more often when they herd of clinicl cses in nery res, sought veterinry ssistnce sooner (sutler signs), nd egun trement erlier. This ssumes tht horses hve stedy progression of clinicl signs nd tht owners of horses tht died merely missed the onset of mild signs. However, in this study, mny owners reported seeing nothing unusul with their horses during the dy efore the onset of moderte to severe clinicl disese (recumency, presis, or prlysis of lims). An lternte theory is tht the incution period for different doses of virus corresponding to different time periods results in the difference in immune response nd the vrying presenttions nd outcomes of clinicl disese. Further clrifiction of this finding is wrrnted. Stllions were more likely to die s result of WNV clinicl disese thn either mres or geldings. In some mmmls, it hs een noted tht while testosterone oosts sexul chrcteristics, it lso simultneously impirs immune system functioning (21). Therefore, stllions my hve poorer immune response thn either femles or geldings. To compre our results to those of other studies from the US, we rectegorized the dt s mles (oth stllions nd geldings) nd femles (mres) for renlysis. In oth the other studies, the univrile nlysis suggested tht femles were more likely to die, ut this result ws not sttisticlly significnt (8,16). In our nlysis, no sttisticlly significnt difference existed etween mres nd geldings, ut rther etween stllions nd mres or geldings. Therefore the percentge of mles tht were intct stllions could influence this ssocition. This ws reported s 8% for 1 study (8), compred with 15% for our study. Prevention eduction progrms for humns list wering light colored clothing s method to decrese the possiility of WNV infection (22). When seeking out potentil explntions, no cler evidence for the resoning ehind the use of light colors s prevention of infection could e found. In this study, we looked t risk of outcome once infection nd clinicl disese hd occurred. We found the light colored horses hd greter odds of dying from WNV clinicl disese thn did drk colored horses. A possile explntion could e tht the immune response is linked to color geneticlly. However, no references to justify this were found. Further informtion is needed to determine the role of cot color in the outcome of WNV clinicl disese. Other fctors ssessed in this study, prticulrly ge nd vccintion, were not significntly ssocited with the risk of ftlity in clinicl cses. In other studies, older ffected horses hve een more likely to die (8,16). In this study, the cse ftlity rte ws highest in horses # 5 y; however, this ws not sttisticlly significnt. In this study, only 7% of horses reported to hve clinicl signs in 2003 were vccinted, so there ws little power to ssess vccine effectiveness in prevention of ftlity. Other studies hve shown tht the vccine hs protective effect ginst ftlity in cliniclly ffected horses (8,16). The environmentl vriles (temperture, precipittion, GDD, nd mosquitoes) exmined lso were not ssocited with individul horse risk of ftlity. Clcultion of these vriles involved verging 3 to 5 mo of dt, which my not represent the exposure period of ech cliniclly ffected horse or the developmentl period of the mosquitoes prior to exposure. However, when temperture nd precipittion were clculted y using only dt from the week efore the estimted exposure dte of ech clinicl horse, the results were still not sttisticlly significnt. With so much vrition in environmentl vriles y locl mcro- nd micro-climtes, the use of the nerest climte sttion my not hve represented the conditions t the ctul loction of the cliniclly ffected horse. More likely, however, is the conclusion tht environmentl conditions would not ply role in n individul horse s helth outcome once it hd ecome infected with WNV, rther it would e ssocited with the risk of infection only. The role of vrious other prevention strtegies were lso explored. Mosquito reduction nd voidnce strtegies were not widely used y the owners of cliniclly ffected horses. None of the mosquito reduction methods were significnt predictors of ftlity. Housing horses in rn t night hs een suggested s mens of preventing infection nd disese. In this study, no horses with WNV clinicl disese were housed in rn t ny time. When vriles sex, cot color, nd onset of clinicl signs were included in the nlysis, the protective effect of simple shelter ws not significnt. Even with these findings, voidnce of mosquitoes is still the est method of prevention of clinicl disese, with deth s n possile outcome. In ddition, owners should e wre of the timing nd risk of WNV infection in their geogrphic re, so tht they cn use ll methods ville to them in the prevention of infection nd in hving fvorle outcome of clinicl cses. CVJ References 1. Hyes C. West Nile fever. In: Month TP, ed. The Aroviruses: Epidemiology nd Ecology. Boc Rton, Florid: CRC Pr, 1989: 59 88. 2. Smithurn KC, Hughes TP, Burke AW, Pul JH. A neurotropic virus isolted from the lood of ntive of Ugnd. Am J Trop Med 1940;20:471 492. 3. Hulek Z, Hlouzk J. West Nile fever A re-emerging mosquitoorne virl disese in Europe. Emerg Infect Dis 1999;5:643 650. 1144 CVJ / VOL 48 / NOVEMBER 2007
4. Murgue B, Murri S, Zientr S, Durnd B, Durnd JP, Zeller H. West Nile outrek in horses in southern Frnce, 2000: The return fter 35 yers. Emerg Infect Dis 2001;7:692 96. 5. Autorino GL, Bttisti A, Deuel V, et l. West Nile virus epidemic in horses, Tuscny region, Itly. Emerg Infect Dis 2002;8:1372 1378. 6. Ostlund EN, Crom RL, Pederson DD, Johnson DJ, Willims WO, Schmitt BJ. Equine West Nile encephlitis, United Sttes. Emerg Infec Dis 2001;7:665 669. 7. Porter MB, Long MT, Gertrm LM, et l. West Nile virus encephlomyelitis in horses: 46 cses (2001). J Am Vet Med Assoc 2003;222: 1241 1247. 8. Slzr P, Tru-Drgtz JL, Morley PS, et l. Outcome of equids with clinicl signs of West Nile virus infection nd fctors ssocited with deth. J Am Vet Med Assoc 2004;225:267 274. 9. USDA: APHIS: VS. West Nile virus in equids in northestern United Sttes in 2000 [monogrph on the Internet]. Full report (or highlights). Fort Collins, Colordo: Centers for Epidemiology nd Animl Helth c2001. Aville from: www.phis.usd.gov/vs/ ceh/wnvreport.pdf Lst ccessed 14/5/2007. 10. Weese JS, Bird JD, Dely J, et l. West Nile virus encephlomyelitis in horses in Ontrio: 28 cses. Cn Vet J 2003;44:469 474. 11. Autrush SM, O Connor BP, Clrk C, Smpieri F, Nylor JM. Clinicl West Nile virus infection in 2 horses in western Cnd. Cn Vet J 2004;45:315 317. 12. Curry P. Ssktchewn mosquitoes nd West Nile virus. Blue Jy 2004; 62:104 111. 13. Bunning ML, Bowen RA, Cropp B, et l. Experimentl infection of horses with West Nile virus. Emerg Infect Dis 2002;8:380 385. 14. Ostlund EN, Andresen JE, Andresen M. West Nile encephlitis. Vet Clin North Am Equine Prct 2000;16:427 441. 15. Cndin Food Inspection Agency (CFIA): CFIA response to centrl nervous system signs in horses in West Nile virus endemic res field nd registered estlishment investigtion [monogrph on the Internet]. Ottw, Ontrio c2005. Aville from: http://www.inspection.gc.c/ english/nim/hesn/diseml/wnvvno/horchee.shtml Lst ccessed 14/5/1007. 16. Schuler LA, Khits ML, Dyer NW, Stoltenow CL. Evlution of n outrek of West Nile virus infection in horses: 569 cses (2002). J Am Vet Med Assoc 2004;225:1084 1089. 17. Woody Ornmentls Entomology @ Cornell University [homepge on the Internet] (New York) Growing degree-dy trcker: Explntion of growing degree dys c2003. Aville from: http://www.entomology. cornell.edu/extension/woodys/growingdegreedys.html Lst ccessed 14/5/2007. 18. Dohoo I, Mrtin W, Stryhn H. Mixed models for discrete dt. In: Dohoo I, Mrtin W, Stryhn H. Veterinry Epidemiologic Reserch. Chrlottetown, Prince Edwrd Islnd: AVC Inc, 2003:499 520. 19. Corrign RL. Prediction multi-species evlution of humn risk of West Nile virus infection, Ssktchewn, 2003 [Msters disserttion]. Ssktoon, Ssktchewn: University of Ssktchewn, 2005. 20. Brownstein J, Rosen H, Purdy D, et l. Sptil nlysis of West Nile virus: Rpid risk ssessment of n introduced vector-orne zoonosis. Vect Borne Zoon Dis 2002;2:157 164. 21. Folstd I, Krter AJ. Prsites, right mles, nd the immunocompetence hndicp. Am Nt 1992;139:603 622. 22. Ssktchewn Helth [homepge on the Internet] Mosquito seson nd the West Nile virus: Cover up nd void mosquitoes. c2003. Aville from: http://www.helth.gov.sk.c/rr_wnv_info.html Lst ccessed 14/5/2007. CVJ / VOL 48 / NOVEMBER 2007 1145