Dogs as Sentinels for Lyme Diseasein Massachusetts

Joann M. Lindenmayer, DVM,B. Onderdonk PhD

MPH, Du,wood Marshall, BS, and Andrew

IntroducionSummary statistics of incident hu-

man cases collected by the Centers forDisease Control and state departments ofpublic health have been used to highlightnational and regional distributions andchanges in Lyme disease epidemiology. 1-3Passive surveillance systems, however,are well known for underestimating actualcases of any disease. The implementationof measures designed to minimize humanexposure to ticks and ageographicallyvari-able index of suspicion for Lyme diseasemay affect numbers of case reports, butthey have little to do with changes in theenvironmental burden of B. buigdorferi

Testing dog populations for the pres-ence of antibody to B. burgdorfen hasbeen used as an epidemiologic tool to sup-plement surveillance based on passivecase reporting.4-10 Dogs in endemic areashave been shown to be almost six times aslikely as people to be exposed to infectedticks,1' making them a sensitive indicatorofthe presence ofB. burgdorferi The pre-sent study was undertaken to determinewhether a relationship exists between se-roprevalence of antibodies to B. burgdor-feri in the dog population and incidentcases of human Lyme disease in Massa-chusetts. If so, dogs might serve as senti-nel animals for Lyme disease.

Methods

Canine sera

Thirty-two veterinary practices lo-cated in eight Massachusetts counties par-ticipated in this study (Figure 1). Practiceswere chosen on the basis of the followingcriteria: routine use ofTufts Veterinary Di-agnostic Laboratory (TVDL), willingnessto participate in the study for the period of

1 year, and location. Practiceswere chosento represent counties that were known,based on Massachusetts Department ofPublic Health records, to be endemic forLyme disease (Dukes, Nantucket, Essex,Barnstable, and Plymouth counties), torepresent areas where Lyme disease activ-ity has recently been reported (Hampshireand Hampden counties), or to be located inareas where Lyme disease had not previ-ously been reported (Worcester).

At each veterinary practice, veteri-narians were asked to choose a convenientday of the week, and on that day to takeserum samples from the first 5 to 10 con-secutive dogs that were seen, regardlessof the reason for presentation. If a clientdid not permit blood to be drawn, theveterinarian skipped that dog and contin-ued to sample consecutively until the pre-determined number of samples had beencollected. For each dog sampled, veteri-narians filled out brief forms providinginformation on the dog's breed, age, sex,owner's name and address, reason forpresentation to the veterinarian, date thesample was obtained, and whether or notthe dog was a Lyme disease suspect.

Sampling was undertaken from Au-gust 1, 1988, through December 31, 1988,

Joann M. Lindenmayer is with the departmentof Comparative Medicine, Tufts UniversitySchool ofVeterinary Medicine. Andrew B. On-derdonk is Director, Microbiology Laboratory,Brigham and Women's Hospital, Boston, MA.Durwood Marshall is with Tufts University Ac-ademic Computer Services.

Requests for reprints should be sent toJoann M. Lindenmayer, DVM, MPH, TuftsUniversity School of Veterinary Medicine, De-partment of Comparative Medicine, 200 West-boro Road, North Grafton, MA 01536.

This paper was submitted to the journalMay 30, 1990, and accepted with revisionsMarch 6, 1991.

November 1991, Vol. 81, No. 11

Canine Sentinels for Lyme Disease

FIGURE 1-Lyme Study Veterinary Pracdces In Massachusets Counties.

suspended for the winter, begun again onApril 1, 1989, and continued through July31, 1989. Sampleswere collected in serumseparator tubes and sent to the TVDLwhere theywere centrifuged and frozen atminus 20C until testing could be per-

formed. Serum samples from dogs sus-

pected of having Lyme disease were

tested within 7 days of arrival, and serum

samples from dogs not suspected of hav-ing Lyme disease were tested within 4months of arrival at TVDL. The presenceof antibodies to B. burgdorfen was mea-

sured using an enzyme-linked inmmuno-sorbent assay (ELISA) as previouslydescribed.12 The cutoff value for sero-

positivity of0.170 optical density units is 3standard deviations (SDs) above the meanoptical density for serum samples takenfrom 34 apparently healthy dogs residentin areas not endemic for Lyme disease.

Human Lyme Disease CasesStatistics on Lyme disease cases for

the years 1985 to 1989 were obtained from

the Surveillance Service of the Division ofEpidemiology in the Massachusetts De-partment of Public Health.

StafstialAnalysis

Canine seroprevalence was calcu-lated as the number of dogs seropositivedivided by the number of dogs tested foreach category of interest. A logistic re-

gression modelwas developed to estimaterelative risks for canine exposure associ-

ated with five factors thatwere suspected,a priori, of being predictive of seroposi-tivity.

1. Alttude of the town where eachdog was resident. Several environmentalfactors have been proposed to explain therange of Iwdes dammniu. 1-16 The distri-bution of incident cases of human Lymedisease in Massachusetts suggests an as-

sociation between altitude, either directlyor as a proxy for other factors, and inci-dence. Altitude was modeled on two lev-

els depending on whether a town was lo-cated at <200 feet or 2200 feet above sea

level.2. Date the sample was drawn.

Nymphal ticks, the developmental stagemost responsible for transmission of B.burgdorferi to people, have been demon-strated to be most active in late spring andearly summer, whereas adult ticks havebeen most active in the fall and winter.17The sampling date was thus modeled on

two levels, depending on whether a sam-

ple was obtained between August 1 andDecember 31, 1988, or between April 1and July 31, 1989. These dates roughlycorrespond to fall/winter and spring/summer seasons.

3. Sex This factor was modeled on

two levels, male and female, regardless ofwhether a dog was intact or neutered/spayed. Previous studies indicate thatmale dogs may be predisposed to expo-

sure to B. burgdorferz.94. Breed. Three levels of breed were

chosen to reflect the relative probability of

American Joumal of Public Health 1449

LYME STUDY VETERINARY PRACTICES

* PRACTICE LOCATIONS Bristol DukesJ<c;)

Nantucket

November 1991, Vol. 81, No. 11

lindenmayer et al.

exposure to ticks in the environment-sporting and large mixed breed dogs (highprobability of exposure), toy and smallmixed breed dogs (low probability of ex-posure), and all others, includingworking,hound, terrier, nonsporting, mediummixed breeds, and dogs ofunknown breedstatus (intermediate probability of expo-sure).

5.Age ofthe dog at the tine ofsam-pling. The risk for dogs of testing positivefor antibody to B. buwgdorfen has beenfound to be associated with the cumula-tive time of I. dammini exposure,11 al-though other studies have reported agreater degree of seropositivity and higherantibody titers among younger thanamong older dogs.6.9 Age of dogs, as aproxy for cumulative time of exposure,was modeled on two levels, <2 years ofage and >2 years of age.

Logistic regression analyses werecarried out with BMDP and SAS, andPearson x2 analysis was performed withStatXact (Cytel Software Corporation,Cambridge, Mass).

ResultsCharactenstics ofthe StudyPopulation

During the period August 1, 1988, toJuly 31, 1989, 3011 serum samples ob-tained from 2941 Massachusetts dogs

were tested for the presence ofantibody toB. burgdorferL Seventy dogs, or2% ofthetotal population, were sampled twice, thesecond time for reasons unrelated to thepresent study. Males and femaleswere ap-proximately equally represented (46.4%vs 49.5%, respectively). Sex of the re-maining 4.1% of the study population wasnot noted byveterinarians. Sporting, largemixed breed, andworking dogs comprisedthe majority of the breeds represented(29.4%, 24.3%, and 19.6%, respectively).All other breeds were approximatelyequally well represented. The percentageof dogs in each age category ranged be-tween 5.7% and 8.9% up to 11 years ofage. Above that, the percentage in eachage category ranged between 0.1% (>15years old) and 4.7%.

SeroprevakenceAmong all dogs tested, unconditional

seroprevalence was 20.3%.Comparison ofincidence andpreva-

lence. Canine seroprevalence was calcu-latedby dogs' place ofresidence, the mostlikely site of exposure to B. bugdorfeniThis permitted inclusion of 262 of 351Massachusetts communities in the analy-sis.

Greatvariation exists among incidenthuman cases by county from year to year(Table 1). By contrast, canine seropreva-lence datawere available for every county

in the state except one, and, because somany dogswere sampled, reasonable con-fidence intervals could be calculated foreach point prevalence estimate. A com-parison of incidence and prevalence bytown for a single county (Essex) likewiseconfirms the completeness and stability ofthe canine data (Table 2).

Regression analyses of the relation-ship between canine seroprevalence in1988 to 1989 and human incidence per100 000 persons for each year between1985 and 1989 by county were unable tostatistically confirm the utility of canineseroprevalence as a predictor of the num-ber of human cases. This is likely attib-utable to the small number ofdocumentedcounty-specific cases in anygivenyear. Inorder to overcome the variability inherentin the human data, cases of Lyme diseasewere averaged by county for the years1985 through 1989. When the relationshipbetween loglo (mean incidence in humans1985-1989) and canine seroprevalencewas explored, canine seroprevalence wasfound to be highly predictive of incidence(R2 = 0.80, P <.0001) (Figure 2).

Seroprevalence by month. Seroprev-alence by month ofsample submission didnot vary with the exception of November1988, when a particularly high proportionof samples exhibited seropositivity (Fig-ure 3).

1450 American Journal of Public Health November 1991, Vol. 81, No. 11

Canine Sentinels for Lyme Disease

Seroprevalence by breed Seroposi-tivity was highest among sporting andlarge mixed breed dogs and lowest amongtoy and small mixed breed dogs (Figure 4).

Seroprevalence by age. Degree of se-ropositivity was lowest among dogs <2years of age. Above 2 years of age, sero-positivity remained relatively stable (Fig-ure 5).Sensivity ofthe Carnie Model

Using seropositivity alone, far moredogs tested positive for exposure to B.burgdorferi than incident human caseswere reported. Comparing mean inci-dence for 1985 to 1989 with canine sero-prevalence by county, seropositivity esti-mates were 513 to 17 857 times greaterthan incident human cases.

Veterinary suspicion for Lyme dis-ease was associated with serologic test re-sults. Of 203 dogs that were noted to beLyme disease suspects based on clinicalpresentation, 91 (44.8%) tested positivefor exposure to B. burgdorferi, whereasonly 18.4% (465/2530) of nonsuspects and20.1% (56/278) of dogs whose Lyme dis-ease status was not noted tested positivefor exposure. The Pearson x2 test statisticwas 141.7 (df = 2, P = .0011) and wasbased on a Monte Carlo simulation of2000tables generated from fixed marginal dis-tributions of the observed data.

Lgisic regressionL Ofthe dogs stud-ied, 2789 met the criteria that comprise thepopulation profiles for the logistic regres-sion model and were included in the anal-

ysis. The standard logistic regressionmodel had the following form:

ln(p/1-p) = -4.74 + 1.64T - 0.20D- 0.05S + 0.82B1 + 1.41B2 + 1.03A,

wherep = probability of testing positive

1-p = probability of testing negativeT = altitude of town where dog was

residentD = season serum sample was ob-

tainedS = sex of dog

B1 = indicator variable for the differen-tial effect of moderate vs low ex-posure by breed

B2 = indicator variable for the differen-tial effect of high vs low exposureby breed

American Journal of Public Health 1451November 1991, Vol. 81, No. 11

Lind_~ennaar et al.

° 1

_ 1.

0000

0

0 10 20 30 40 50 60 70Seroprevalence for canines 1988-89

FIGURE 2-Li_t-Squares RegressIon Unie for loglo Mean Intercept = 0.13, Slope =0.03, Adjusted R2 = 0.86 (P <.001). Human nlene 1985-89 vs CanineSe5oprevalence, for 13/14 Massachusets Counties

30-35-. ...................................................... ............................ .................................................................................................................... ..............25- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~................................ .........

.............................. . ...............

15-........ ..................................................... ..... ........................................................................................... ........

10-....................................................................................................... ......................................................................................................

25

10.

June JulySept Oct Nov Dec April MayMonth of Submission

FIGURE 3-Canine Valence by Month (mean ± 95% eInterval).

A = age of dog at time of sampling

The x2 goodness-of-fit statistic for theequation sum[2 Oi x ln(OX/Ei)] was 34.37(df = 39, P = .68), suggesting that themodel adequately explained the test out-come. The 48 possible cells resulting fromthe coding of the five factors yielded twocells without any observed data due toincomplete records. Twoofthe 46 remain-ing cells proved to be particularly unusualand notwell accommodatedby the logisticmodel. These two cells were low-expo-sure breed female dogs <2 years of age,resident at low altitudes and sampled inthe spring or summer (n = 3), and low-exposure breed male dogs .2 years of

age, resident at high altitudes and sampledin the fall or winter (n = 5). These cells,because of their small counts, are unlikelyto be representative of the correspondingpopulation profiles. For this reason, theywere removed from the analysis. The re-

calculated x2 goodness-of-fit statistic was29.84 (df = 37,P = .825). All final resultsare based on the latter model.

Odds ratios for factors included in themodel (Table 3) revealed that dogs resi-dent at or near sea level were approxi-mately five times as likely to test positivefor exposure as dogs resident at higherelevations. Breeds with moderate proba-bility of exposure and those with high

probability of exposure were more thantwo times and more than four times aslikely, respectively, to test positive thanbreeds with low probability of exposure.Linear contrast analysis for differencesbetween breeds with high probability ofexposure and all others indicated that theformer were far more likely to be exposedthan the latter (X2 = 33.01, P = .0001).Dogs >2 years of age were almost threetimes as likely to test positive as dogs c2years ofage. Dogs sampled in the springorsummerwere slightly less likely than dogssampled in the fall or winter to test posi-tive, although this difference was not sig-nificant. Sex of the dog had no effect onthe probability of testing positive.

In the logistic regression model, cellswith the highest probabilities for seropos-itivity (>0.22) represented dogs that livedat altitudes <200 feet, of the sporting orlarge mixed breed types, and older than 2years of age. No differences with respectto date of sampling or sex were noted.Cells with the lowest probabilities for se-ropositivity (<0.017) represented dogsthat lived at altitudes .200 feet, were <2years of age, and were of low and mod-erately exposed breed types. As before,no differences in date ofsampling or in sexwere noted.

DisuwsionThe dog has been proposed for use as

a sentinel animal to detect the presence ofB. bugdorfen for two reasons. Dogs ex-posed to infected ticks develop antibodiesto the spirochete and often exhibit clinicalsigns of the disease similar to those in peo-ple. Second, dogs are more likely to beexposed to infected ticks because theycome into more physical contact with tickhabitats than do most people, and because

1452 American Journal of Public Health

0a)I..a

November 1991, Vol. 81, No. 11

Canine Sentiels for Lyme Disease

measures to prevent exposure to ticks are

difficult to implement in dogs. Results ofthe present study indicate that under cer-

tain conditions canine seroprevalence es-

timates are highly predictive of incidenthuman cases, further strengthening theusefulness of dogs as sentinel animals.

The present comparison betweentwo different measures of disease and twodifferent time periods is unavoidable fortwo reasons. Incident canine cases ofLyme disease are difficult to determine.Efforts to induce seroconversion to B.bugdorferi or clinical Lyme disease inlaboratory dogs have not been unifornlysuccessful.18,19 Furthermore, dogs oftenexhibit seropositivity to the spirochetewithout concomitant clinical evidence ofdisease.6-8'1'8-21 It is unlikely, then, thatincident cases among dogs will ever beavailable for comparison with human in-cidence rates. Second, regression analy-ses of canine seroprevalence in 1988-1989and human incidence for each year be-tween 1985 and 1989 failed to reveal a re-

lationship between the two. This is prob-ably due to the great variability inherent inincident human cases from year to year.

Seroprevalence estimates among dogs fora given year are likely to be stable becauseso many dogs can be sampled under an

active surveillance system. Itwould, how-ever, be useful to extend these studiesover a longer period of time in order todetermine whether or not these observa-tions are maintained.

The overall seroprevalence estimateof 20.3% for this study does not reflectseroprevalence among all Massachusettsdogs because participatingveterinary clin-ics were chosen to afford maximal cover-

age of particular counties in Massachu-setts. For this reason, county- and town-specific seroprevalence rates are more

likely than state-wide estimates to be rep-resentative of dogs presented for veteri-nary care in these regions. However, all

seroprevalence estimates may actuallyunderestimate true seroprevalence be-cause they were based on dogs presentedtoveterinary practices rather than the gen-eral canine population. Inclusion of straydogs and dogs that do not receive regularveterinary care might well increase sero-

prevalence estimates because these dogswould be more likely to run free and there-fore be exposed to infected ticks.

Seroprevalence estimates were

based on the assumption that seropositivedogs received exposure in the townswhere they resided. In fact, dogs oftentravel with their owners, particularly inthe summer months. This raises the pos-

30

25r *t F

0

e~~~~~~~~re

10 .....................................................................................

T SM Te N H W MA U LM SBreed

FIGURE4-CanlneeroprealencebyBreed (mean ± 95%confidenceIntal;T=toy,SM=small mixed, Te=terrler, N=nonsporting, H=hound, W=workdng,M=mlxed, U=unknown, LM=Iage mixed, S=sportlng).

sibility that individual seropositive dogswere exposed to infected ticks at locationsother than their towns of residence andmight account for the presence of sero-

positive dogs in towns where Lyme dis-ease is not known to be endemic.

Previous studies in this laboratoryand elsewhere1222-24 have documentedfalse positive ELISA test results for bothcanine and human serum samples. Whilewe recognize that this possibility exsts,test results from an area where Lyme dis-ease is not known to be endemic by Mas-sachusetts Department of Public Healthstandards (i.e., Worcester County) indi-cate that a very small proportion of dogs

(1/78 or 1.3%) tested positive for antibodyto B. buwgdorfefi, presumptive evidencethat the proportion of false positive sam-

ples in this study is low.Use of the canine to detect and quan-

tify Lyme disease risk offers several ad-vantages over the use ofhuman incidencerates alone. Dog populations are particu-larly convenient to test for exposure to B.burgdorfen. Dogs are routinely brought toveterinarians in the spring for heartwonntesting. Blood drawn for this purpose can

likewise be made available for Lyme dis-ease testing, obviating the need for a sep-arate visit to the veterinarian. Further-more, dogs tested at this time of year are

American Journal of Public Health 1453

40.40-....................................................................................................... ............................... .....................

.................................I........... ................ . ...........

20 ................................. ............. ......................................................

0Unk 1 10 11 12 13 14 15 16>16

Age in YearsFIGURE 5--ConirmS evalence by Age (mean ± 95% cog i de intervao.

November 1991, Vol. 81, No. 11

Linenmayer t al.

usually well animals presented for routinecheckups and are thus more likely thandogs presented at other times of the yearto be representative of veterinary cliniccatchment populations.

Second, seroprevalence estimates indogs not only reflect incidence rates inpeople, but they also appear to be a farmore sensitive indicator ofthe presence ofB. burgdorferi than do the latter. Serop-ositivity for every 100 000 dogs tested wasconsistently much higher than Lyme dis-ease incidence per 100 000 persons.

Third, seroprevalence estimates indogs are more likely to reflect the actualenvironmental risk of Lyme disease thanare reports of incident human cases. Fac-tors such as increasing awareness ofLyme disease may falsely increase inci-dence rates, whereas the use of measuresto prevent exposure to ticks may actuallydecrease incidence rates apart from anychanges in the environmental burden ofB.burgdorferi Measures designed to pre-vent dogs from exposure to potentially in-fected ticks are frequently unrewarding.Dogs' propensity to pickup large numbersof ticks and the difficulties associated withfinding tiny I. dammin make tick expo-sure a near-certainty in endemic areas ex-cept, perhaps, for dogs that rarely ventureoutdoors.

Fourth, canine seroprevalence esti-mates permit a fairly complete represen-tation of areas of potential risk to bedrawn. As evidenced by the Massachu-setts Department of Public Health data,collected under a passive reporting sys-tem, incidence measures for a particularcounty or town are frequently based on asingle case report. Furthermore, markedyear-to-year variability in incidence ratesis frequently observed. Under an activecanine surveillance system such as the oneused here, additional information wasmade available to permit the calculation ofpoint prevalence estimates and reason-able confidence intervals for 31 of 34towns in one particular county (Essex)and for 13 of 14 counties.

The logistic model employed in thepresent study appears to adequately ex-plain the risk ofseropositivity amongMas-sachusetts dogs and has defined certainrisk factors for canine exposure, namely,altitude of town of residence, breed ofdog, and age. Dogs that live at altitudes<200 feet above sea level appear to be atgreater risk for exposure to B. burgdor-fem The fact that seroprevalence varieswidely even at this altitude raises the pos-sibility that ecologic factors such as hu-midity or temperature may affect deer tick

survival and thus the risk of exposure toLyme disease. Lyme disease is alsoknown to occur in noncoastal areas of theUnited States where altitude may not bedefinable as in the present model.

Previous studies have not been ableto confirm a breed disposition among se-ropositive dogs, although mixed breeddogs and hunting breeds were more fre-quently seropositive.9 The present study,perhaps because of a larger sample size,establishes that seropositive dogs aremore likely to be sporting and large mixedbreed dogs. These animals are presum-ably at greater risk of exposure becausethe former are used for hunting and sport-ing purposes and the latter, by virtue oftheir size, may be allowed more freedomto roam.

Dogs >2 years of age appear to be atgreater risk of exposure than youngerdogs. Other studies have documentedmaintenance of antibody titers in dogsmonths to years after initial exposure tothe spirochete.20,21 Under these circum-stances, seroprevalence could be ex-pected to increase with advancing age. Inthe present study, however, age-specificseroprevalence estimates appear rela-tively stable above 2 years of age. It maybe that, for a given dog population, only afixed proportion of its members are at riskfor exposure toB. burgdorferi For exam-ple, a smaller proportion of dogs might beat risk in areas with large urban popula-tions and strict leash laws than in ruralareas with less restrictive leash laws. Dogsat risk may be exposed to infected I. dam-mini and seroconvert at an early age, andthus, after the critical age of 2 years, noincrease in age-specific seropositivitywould be noted.

In the present study, sex appearednot to explain seropositivity. Informationon reproductive status of an animal,whether intact or neutered/spayed, wasrequested but frequently not provided. Itis possible that reproductive status, as anindicator for proclivity to roam, mightwelibe a risk factor for seropositivity, althoughthis could notbe determined in the presentstudy.

The risk factors identified for dogsmay directly or indirectly illuminate cer-tain aspects ofthe epidemiology ofhumanLyme disease. Altitude can certainly be arisk factor for people. Age, as a proxy forcumulative duration of exposure, andbreed, as a proxy for occupational or rec-reational exposure, are other factors thatshould be considered in models investi-gating the risk of Lyme disease in people.

A vaccine againstB. burgdorfen hasrecently been made available for use indogs (Borrelia burgdorferi Bacterin, FortDodge Laboratories, Fort Dodge, Iowa).Widespread use of such a vaccine couldwell compromise the ability ofcanine pop-ulations to serve as sentinel animals forLyme disease. Vaccinated dogs wouldpresumably be protected against field ex-posure to the spirochete and antibody ti-ters in vaccinated dogs would be indistin-guishable from those of naturally exposeddogs. The degree towhich the dog-owningpublic avails itself or the vaccine remainsto be seen. However, the vaccine is likelyto be put to greatest use in Lyme disease-endemic areas where a sentinel systemwould be of least benefit. Dogs in nonen-demic areas might well still serve as anearly warning system for detection of thespirochete if they remain largely unvacci-nated. []

AcknowledgentsThis work was supported by a public healthfellowship from the Massachusetts Health Re-search Institute of the Medical Foundation,Boston, Mass, to Dr. Lindenmayer.

The authors are grateful to the veterinar-ians who participated in this study, and in par-ticular to Dr. Frederick Johnson, Dennis Ani-mal Hospital, Dennis, and Drs. James Dasbachand John Rae, Nantucket Animal Hospital,Nantucket, Massachusetts, for their manyhelpful comments and suggestions. Mr. PaulEtkind, Director, Epidemiology Program of theMassachusetts Department of Public Health,critically reviewed the manuscript.

References1. Schmid GP, Horsley R, Steere AC et al.Lyme disease surveillance in the UnitedStates, 1982. J Infect Dis. 1985;151:1144-1149.

2. Ciesielski CA, Markowitz LE, Horsley R,Hightower AW, Russell H, Broome CV.The geographic distribution of Lyme dis-ease in the United States. Ann N YAcadSci 1988;539:283-288.

3. Tsai TF, Bailey RE, Moore PS. Nationalsurveillance of Lyme disease, 1987-1988.Conn MedL 1989;53:324-326.

4. Teitler J, Madigan J, DeRock E, PedersenN, Carpenter T, Franti C. Prevalence ofBorrelia bwgdorfen antibodies in dogs innorthern California: risk factors andzoonotic implications. AnnN YAcad Sci1988;539:500-503.

5. Schulze TL, Bosler EM, Shisler JK, WareIC, Lakat MK, Parkin WE. Prevalence ofcanine Lyme disease from an endemic areaas determined by serosurvey. ZentralblBakteriol Mikrobiol Hyg [A]. 1986;263:427434.

6. Magnarelli LA, Anderson JF, SchreierAM, Ficke CM. Clinical and serologicstudies ofcanine borreliosis.JAm VetMedAssoc. 1987;191:1089-1094.

7. Burgess EC. Natural exposure of Wiscon-

1454 American Journal of Public Health November 1991, Vol. 81, No. 11

Canine Sentinels for Lyme Disease

sin dogs to the Lyme disease spirochete.LabAnim Sci 1986;36:288-290.

8. MagnareUli LA, Anderson JF, KaufmannAF, Lieberman LL, Whitney GD. Borre-liosis in dogs from southern Connecticut.JAm Vet MedAssoc. 1985;186:955-959.

9. Cohen ND, Carter CN, Thomas MA, An-gulo AB, Eugster AK Clinical and epi-zootiologic characteristics ofdogs seropos-itive for Borelia burgdorfen in Texas: 110cases (1988). J Am Vet Med Assoc.1990;197:893-898.

10. Greene RT, Levine JF, Breitschwerdt EB,BerkhoffHA. Antibodies to Borrelia burg-dorferi in dogs in North Carolina.AmJ VetRes. 1988;49:473-476.

11. Eng TR, Wilson ML, Spielman A, Lastav-ica CC. Greater risk ofBorreai buwgdorfeninfection in dogs than in people. J InfectDis. 1988;158:1410-1411.

12. Lindenmayer JM, Weber M, Bryant J,Marquez E, Onderdonk AB. Comparisonofindirect inmunofluorescent antibody as-say, enzyme-linked immunosorbent assay,and Western immunoblot for the diagnosisofLyme disease in dogs.JCin MicrobioL1990;28:92-96.

13. Carey AB, Krinsky WL, Main AJ. Lwdesdammnbi (Acari:1Iodidae) and associatedIxodid ticks in south-central Connecticut,USA. JMed EntomoL 1980;17:89-99.

14. Piesman J, Spielman A, Etkind P, RuebushT1, Juranek DD. Role of deer in the epi-zootiologyofBabesia micmti in Massachu-setts, USA. JMed EntotoL 1979;15:537-540.

15. SchulzeI, Bowen GS, Lakat MF, ParldnWE, Shisler JK. Geographical distributionand density of lwdes dammuni (Acari:Ix-odidae) and relationship to Lyme diseasetramission in New Jersey. Yale J BiolMed. 1984;57:669-675.

16. Lastavica CC, Wilson ML, Berardi VP,Spielman A, Deblinger RD. Rapid emer-gence of a focal epidemic ofLyme diseasein coastal Massachusetts. N Engi J Med.1989;320:133-137.

17. Spielman A, Clifford CM, Piesman J, Cor-win MD. Human babesiosis on NantucketIsland, USA: description of the vector kc-odes (Ixodes) dammin4, N. SP. (Acari-na:Ixodidae). J Med EntomoL 1979;15:218-234.

18. Burgess EC. Experimental inoculation of

dogs with Bome?lia bwgdoifen. ZentralblBabktelMwnbiol4HygfA. 1986;263:49-54.

19. Greene RT, Levine JF, Breitschwerdt EB,et al. Clinical and serologic evaluations ofindiced Bomrlia burgdorfe,i infection indogs.Am J Vet Res. 1988;49:752-757.

20. Magnareli LA,AndersonJF, SchreierAB.Persistence of antibodies to Borrelia bw-dorfen in dogs ofNew York and Connect-icut.JAm VetMedAssoc. 1990;196:1064-1068.

21. Bosler EM, Cohen DP, Schulze TL, OlsenC, Bernard W, Lissman B. Host responsestoBorrelia bwgdorfen in dogs and horses.Ann N YAcad Sci 1988;539:221-234.

22. Wilkinson HW. Immunodiagnostic testsfor Lyme disease. Yale J Biol Med.1984;57:567-572.

23. Craft JE, Grodzcki RL, Steere AC. Anti-body response in Lyme disease: evaluationof diag tic tests. J Infect Dis. 1984;149:789-795.

24. Russell H, Sampson JS, Schmid GP, Wil-kinson HW, Plikaytis B. Enzyme-linkedimmunosorbent assay and indirect imnmu-nofluorescence assay for Lyme disease. JInfect Dis. 1984;149:465-470.

I........... -I .-m-N-m-m-mv-.............I----------------------- :. --------------- -.-.-.-.-.-.-..%%.C....!......e.................:...

................".C............ ......: ": :"::.:". :.:"..:: :"

-.A%-.-..%-.-......'e.....".......C.'.-............... .:...... :.:.:. ,:,::: ..:: .,-:. ::-..

.... .- :.:: ,:::. ::,:::::... :-.X .;...:.:. ...:.:... ..... ....

....... ... .... ....

...:.:....:...:.:-..:.:.:-....-....,." '',.,..,..,.,.'',."

....... .....................................:..:...:.:.: :.:...:.:. .:.:.:.:.:-:.:.:.: :.:.:.:.:.:.,.:-:.:.: :.:.:.:.:.:.:.:-:.:.,.:... -:---..- ............:.:.:.:.:.:.:.:.:.:.:.:.:.:. .....::. .....:.-:-:.:.,.: .:.:.-. ........................... .:-.-:-:-:-.-:-: :-.-:-:-:-:-:-:.:.:.: .:....... ....... ............ ..................... .. -.:.:.:. .:.:-:. .,.:-:...:-:.:...:.:.:.:.: :.:: :::::.: ..: :..::.,.::::::::::,:,::::: ::::.:: .'- -'-

::::::::.: .. ....:.:.:...:.:.:...:.:.:.:.:.:.:.:.: .::;::::: :::::::::.:::::::-:-:.;.:.:-:.:. :::::,::::::::: :.:.:-:-, .:.:.:.:-:-:.:.:.:-:.:.:.:.:.:.:.:.:-:.:.:.:-: .............. .................................. ............................. ............ ................:........... ..:.:.:.:.:.:.:.:.:.:.:-:-, .................. .......... ...... .................... --.-.- -.-,-.-. ...........:.:.:.:.:.:. .:: :::::::.::::::: '-'-'-'

.., ........ .... .:-:.:.:.:- :-:-:-:-:-:-:-:...... ............................... ...................................----------...... .......................,........................:.:..:.:.:.:.:.:...:.:...;.:.:..... .......:: ...................

:.:-: :.:.:.:.:.:.:.:.:.:.:.:.:.:.:-:.:.:.:.:.:.: .................................... .... .:.:.:.:.:.:.:.:-:.,.:.:.:.,.:-:.:.,.:-:-:.,.:-:-:.:.: ..:.:.:.:.:-:.:.:.:.:.:.:.:.:.:.:.:-:.:.:.:-:- .:.:... :.:.:.:.:...... ....... ............... .::;: -:::::::........ .......:.:.:.: :.:.:.:.:-:.:.:.:-:.:..................'....................................:::::::::::::::::::::::-:::::::.:::::::::.:::::::.:::.:.:.:-:.:.:.:.:.:.:.:-:e"-.".".".".".".".-.'.'...................................................................................... ....... :::::::::::::::::::::::::::::::::.:::::::.. ....... ........... ............... .... ...... .......... ................ ... ....... .......................................... ......... ....... .:.:....-:-:....-:-: .......... :::;:::::;:;:::::::::::.:.- .. .. ................,......-, .............. ... ....... ............................. .:.:.:-:.:.:.:-. .... ... ............... ...

.:.:.:.:....'.'. :.:.:.:-:.:.:.:.:.:::::::!:::::::::,- .:.:.:.:.:.:.:.:.:.:.:....... ..................... ..:.:.:.:.:.:.:.:-:.:.:.:.:.:.:.:.:.:.,-. .......................-.......................-.-.....- .....:........................ ..:-:.:.:.:- :-:-:-.-:-:-:-:...:.:.:.:.::.:.:.:.:.:.,.::.:.:.:.:-:.:.:.:-:.;.,.: :.:.:.:.:.:.:.:.:-:-:.,.,.-,-,...,.,-,.,.,.,-.,.,.,.,.......-..............,...........................:.:.:.:.:-:-:.:.:-:.:.:.:.:.:.:.:..............:.:........ ................................... .................. ...................."... ...........................................:.:.:. .:.:.:.:.:.:...:.:.:.:.:.:.:.:.:.: :.:.:.: ................

.................... ..: :::: ....: :.:.:.:.:. .:-:.:...:.:.:.:.: :-:': :":" :"":.:.......:.:.:.:.:.:.:':-: :.:.:.:.:.:.:.:.: :.:.:. ..........- :::::::,:::X ..- ............... ..:.:.:.:...:.:.:. .:.:.:. :-.-.-.-.-.-.-.-:..-:..-:.. ..:.:..:..-..:.:.:........ ........:.:.:.:-:.:.:.,- ................................ ... ......................................................:.: ..... .:-:.....:-......................-....... :.:.:.:.: :.-.-. ......:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.. ... '' ,.,.:.:.-.-.: :.: :.:.,- -:-:.:.:.:.:.:.:-:......-....... .......

....................... ..........................-::.:.:.:..:.:.. --.............. -.--.-.-.-:::" -.,.:-:.:.,- '-'-'-'-'-'-' '-'-' '-'-'-'-'-'-'-'.,.:,.'-""""""""""""""'- ..:.:.:.:.: ..,,:::::::,::: ............................................. ..... .......................: ....:......................... .......... ....... .............,-.....,-, ..:, .:..:::::::::::-.-.-..-..-..-.-..'..-.....:-:.:.,- -:.:.:.:.:.:.:.:-:.:.:.:-;.:.:.:.:.:.:.:.:.:.:-:.:.:.::;:;: :::::::-:.:.:.:-:.:.. .,.,.-.,..:. :::.:-:.:.,.:-: .: ::::............................................................ ................................... .. ......... -.-.:. .. :.:.:.:.:. ........................C.............. :...:.:-:.:.:.:......... .:.:-,... ...................'' :...:.-.-.:.....:.:.:.:.:.:.:........... ..... ............ ..- .....,........... ............................... ..... ... .A ...:.::.: ........:.:...:...:.:.......:...:.:. ...:.:...:.:. -:.,.:-:.....:-:.....:-:.'.'.:-:...:-:-:.:.:.:-:.:.:.. .... ... .. .................:.:.:...:.- :-!-:-:-.-.-:-:-: :...:.:.:...:.. -..-.-.'." ....... ....... .... .... ... ..:.:.:.:.:.:.:.:.:.:.:.:.:. .:.:.:...:.:. .: X..... .. ....... ....... ................... ...... .-... ..-.-.-.-. ...... .-.-.-.-......................... ...... ... ........--............. ................. ....................... ... ............. .-- ........ ......... ........ .. .... ........... ..... .. ............ ........... .. .. .... ......... ............. ... ... ........:...: :.:.:.:.:-:.:.:.:-:-,.,.., ,..,.,"... ......................

........... ..... ,.:....-.-.-.. :.:.:.:-:.:.:..-.-.- ............,.......-.-. .- ----..:....-:.:.:.:.:.:.:.:.:.:.,.:.,.:......... .... :.:.:.:.: ::.: ....... ::.: :.:.:.: :.:.:.: ...:...:.:.:. .: :. ...... .......-- ....................................... ........:.:.:.:.:.:.....:.:.:.:.:..:. "' ..................................................... ...-....... ............ ............ ......-.:-:.:.:.:-:.:.:.: :.:.., ,.,..,.., '.., "..'..'..'.................-. :...:.:.:.:.:.: .:.: :.:....-:-:.:.:.:-:.:.:.: ::'............ ......... .......... ,:::::: .... .............. .................. .......................... -.......,..........- .::::::: ::::::::::::::. ..................

... ... .:.:: :.:.:.:.:.:.:........................................................................-.......-.-. ............................ ............. ....... .......-.-.....-.-...-.-.-.-.-.-. :...:-:.....:-:-, ............... ..... ............. .:.:.:.:.:.:... ..... .......................... ...-.................. .:.:...:.:.:...:.:.:. ,:.......:. -:-:-:-.-:-:-,.,m........................................... ..................................... ................................. '' .,:: ... .....

:'an:.-: .-.:.,.::.-:::.:.-. ::::.....:.:.:.:.:.:.:.:.:.:.:.::.:.:.:-:-,.............-..........-.'.'.-.'.'.'.'... ..........................................-

..--. ,,.-"""--..,-;"::. :I .1 -r................................................................................................................... ..............-.-.-.-.-..-.-................................'.-.'.'.'.'.'...............................'e....... :.:.:.:-:.:.:.:.:.:.:.:-:.:.:.,-..... '''......-......................... ::....... .."'': ....:':d ....:x ................ .......,......" '. ....., A:.:::: :::."- ::ik::n .* :- ..... .. ...

.:i.................... I. .....,- ....w .....

'---" e,-, ",-"--,-"":,:,:,:,......':-I -............:-.................. ..... ... ....P .-I .-:::::::::.,.,.''." .....................

..... ........... .. ::':.':,-it. ................. ... :.:: j .. -:- ..:::::::::::,.:.:::::.:::::::111-In -1,111-oft ..

.................... .... ... .:: :-.. ......-.-..................-.........I......... I. ::::::::::;:..:::.:::::.:::::::.:.::,:::: ....... -F ...":-b:-."":':'o-:':'d -':':':':':':'::':':':'d :'.N':':':':............................ ..... .:.:.

...........................................................................................-.........-... ::..................................- :................................. ............................. .:-:.:.:.:-:.:.:.:.:.:.:.:-:.:.:. :............ ''"-...............--.-.-.-.-.-..-.-.-...--.-.....................:::.- .:.:...:.:-:.:...:-:-:.:.:.:-:.:.:.:. ............................ ....... .:: :.,.....,::.,.,.,* .,.."..,::.......:.:.:.:.:...:.:.:.:.:.:.:..........e.,.,-::::::.:::::;iz.::,:.. ,::::,:::::,?::.:.

.IM ............ ... m.. .:.m- :.. .....: II .. ................ .... ....... .. .. ................ .................... I

............... :.. - ...0 .... .:.:.:.:.::. .:-:-:-: :-:.,.:.:-:.:...:.......:.:. .....:.:.:.:.:..:.:.:.,.:-:-:.:.:- ,:,:.:-:.....:-:.:.:. :...:-:-:...:-:-...- -:. v ........ :..........:.- :-:-:-:-:-:-i-:-:.:.:.:..-.-.-.-.--. ..:......................................... :....... ..::: ..:i::....::.......:.-.-::-:::..:-.,.,K ..".-::: ................... :::,:,::,:,:::,::,:,::;,k l'',.,t.........:: :- t ..e r:::.....: .::......... `:.:%:::..:::::::-j-... ............... ..: