Serologic survey of cats and dogs during an epidemic of West Nile virus infection in humans

Serologic survey of cats and dogs during an epidemic of West Nile virus infection in humans James C. Kile, DVM, MPH, DACVPM; Nicholas A. Panella, MSc; Nicholas Komar, ScD; Catherine C. Chow, MD, MPH; Adam MacNeil, MPH; Brent Robbins, DVM; Michel L. Bunning, DVM, MPH Objective To estimate West Nile virus (WNV) infection rates, assess environmental variables that correlated with seropositivity in dogs and cats, and assess whether pets should be considered as possible sentinels for WNV and therefore of potential human exposure. Design Cross-sectional serosurvey. Animals 442 dogs and 138 cats. Procedure Serum samples were screened for seropositivity against WNV by use of the plaque reduction neutralization test. Results 116 (26%) dogs and 13 (9%) cats yielded positive results. The odds of seropositivity against WNV for outdoor-only family dogs were almost 19 times as great as those for indoor-only family dogs and almost twice as great for stray dogs as for family dogs. Family dogs not receiving heartworm medication were 2.5 times as likely to yield positive results for antibodies against WNV as family dogs receiving heartworm medication. Conclusions and Clinical Relevance Seropositivity was greater for outdoor family dogs than for indoor family dogs. Further investigation of the potential use of stray dogs as sentinel indicators for WNV infection and the potential risk of human exposure is warranted. (J Am Vet Med Assoc 2005;226:1349 1353). In 2002, the largest epidemic of West Nile virus (WNV) meningoencephalitis in humans ever recorded occurred. 1 There were 4,155 laboratory-confirmed cases in humans (case fatality rate, 6.8%). 2 The Centers for Disease Control and Prevention (CDC) also reported WNV infection in numerous avian and mammalian species, including 16,739 birds reported from 41 states and the District of Columbia, and From the CDC, Epidemic Intelligence Service, 1600 Clifton Rd SE, Atlanta, GA 30333 (Kile); CDC, the National Center for Infectious Diseases, 1300 Rampart Rd, Fort Collins, CO 80521 (Panella, Komar, Chow, Bunning); the National Immunization Program, 1600 Clifton Rd SE, Atlanta, GA 30333 (MacNeil); and the St Tammany Parish Department of Animal Services, 25026 Hwy 36, Abita Springs, LA 70420 (Robbins). Dr. Kile s present address is USDA, Food Safety and Inspection Service, Landmark Center, Suite 300, 1299 Farnam St, Omaha, NE 68102. Dr. Chow s present address is Hawaii State Department of Health, 1132 Bishop St, Honolulu, HI 96813. Mr. MacNeil s present address is Harvard School of Public Health, 677 Huntington Ave, Boston, MA 02115. Dr. Bunning s present address is Air Mobility Command, USAF, 203 W Losey St, Scott Air Force Base, IL 62225. Dr. Kile was an Epidemic Intelligence Service Officer at the time of the study. Address correspondence to Dr. Kile. 14,571 mammals, including 1 cat, 7 dogs, and 24 animals of other species; the remainder of the infected animals were horses reported from 41 states. 2 Little information has been published with regard to WNV infection in domestic pet mammals such as dogs and cats. After the 1999 epidemic in New York City, 5% of domestic dogs tested had antibodies against WNV. 3 Additionally, a human serosurvey carried out in Queens, NY, during the same time revealed that 3% (weighted) of the humans surveyed were seropositive against WNV. 4 When the results were compared, the seroprevalence in dogs from the same area in Queens where the human serosurvey was completed was 11%; this was > 4 times that in humans, suggesting that domestic animals may be useful sentinel indicators for WNV and the potential risk of human exposure. The purposes of the study reported here were to estimate WNV infection rates, assess environmental variables that correlated with seropositivity in dogs and cats, and assess whether pets should be considered as possible sentinels for WNV and therefore of potential human exposure. The survey was performed during an epidemic of WNV infection in the vicinity of Slidell, St Tammany Parish, La, during summer and fall of 2002. Materials and Methods Study design This study used a cross-sectional convenience sample of domestic dogs and cats from St Tammany Parish and the city of Slidell, the largest city in the parish, to determine the prevalence of WNV antibodies and identify and evaluate potential environmental variables and other factors that correlated with seropositivity during a WNV epidemic in humans. Study site During 2002, without adjusting for the number of residents per state, Louisiana had the fourth highest number of laboratory-confirmed WNV cases in humans in the United States and St Tammany Parish had the third highest rate of WNV-induced human encephalitis in the state. 5,6 When adjusted for the number of residents per state, Louisiana ranked first in the United States for reported human cases, with 6.7 WNV infections/100,000 residents. 7 St Tammany Parish and the Slidell community were chosen as the study sites because of the number of human cases of meningoencephalitis caused by WNV. Of 12 cases reported to the CDC by mid July 2002, 11 were from Louisiana, 4 of which were from St Tammany Parish. Overall, St Tammany Parish had 39 human cases from June 10 to August 29, 2002. 5 St Tammany Parish has a human population of 191,268, with 69,253 households, and Slidell has a population of 25,695, with 9,480 households. 8 Slidell city limits include numerous natural water sources such as sections of Lake Pontchartrain to the south, Pearl River to the east, and numerous bayous and marshes throughout the city. JAVMA, Vol 226, No. 8, April 15, 2005 Scientific Reports: Original Study 1349

Study population, sample collection, and questionnaire Dogs and cats were sampled when evaluated at veterinary facilities (enrolled as family pets) or animal control shelters (enrolled as strays). Participation in the study was offered to all 29 veterinary facilities within the parish, with final enrollment of 15 facilities (52% participation), including 10 from Slidell and 5 from 4 other cities in the parish. There were 33 participating veterinarians. Veterinarians were asked to offer to all clients who resided in St Tammany Parish serologic testing for WNV of all dogs or cats that the client brought to the veterinary facilities. Veterinarians obtained standard consent from pet owners, who then completed questionnaires requesting general information about the animals, including species, the reason the pet was being seen (eg, checkup, sick visit, or surgery), and the number of years the pet had lived at the given address. The questionnaires also gathered information about environmental variables, including geographic locations of the animals within the parish, potential exposure to mosquitoes because of indoor versus outdoor movement, and preventive insecticide usage by the owners on their pets. Additionally, all 3 animal control facilities in the parish were enrolled, including 2 in the parish and 1 in Slidell. Stray dogs and cats were sampled after capture, while awaiting adoption or euthanasia in animal control facilities. Animal control personnel provided a general address for the animal s location at capture. Although the length of time that an animal was stray could not be determined, for the purposes of this study, it was assumed that all stray animals had outdoor rather than indoor exposure. Serologic assay Three to 5 ml of blood was collected from each animal via a standard venipuncture technique and transferred to a serum separator tube that was centrifuged. The resulting serum samples were held at 20 o C until tested. Serum samples were screened for seropositivity against WNV by use of the plaque-reduction neutralization test, which is the most specific test for the arthropod-borne flaviviruses. 9 Briefly, sera were diluted 1:5 and mixed with an equal volume of reference virus (WNV/NY99-4132 strain) for a final dilution of 1:10. Samples were incubated at 37 o C for 1 hour. The virus-serum mixtures were added in duplicate 0.1-mL aliquots to drained wells of cell cultures grown on 6-well plates. Plates were incubated at 37 o C for an hour, and the wells were subsequently overlaid with a nutrient agarose mixture. A second overlay containing a neutral red dye was added 48 hours later. Plates were examined on the following day for plaques (areas of dead cells that appear as colorless areas against a red background of viable cells). Samples with 80% reduction in plaque-forming units, when compared with the back titration, were considered flavivirus-positive and were further titrated to determine end point titers. Positive samples were also screened for neutralizing antibodies against St Louis encephalitis virus. Samples that yielded positive results from the SLE screening were subsequently titrated to determine virus identification. A 4-fold or greater titer for WNV or SLE virus was considered diagnostic for that virus. Geospatial analysis The latitude and longitude were determined for each dog and cat in the study at the home address for family pets and capture address for strays by use of a geographic information system. 10 Two human WNV studies a,b were conducted concurrently in the same area as the dog and cat WNV serosurvey described here. For affected humans, the geocoded locations for the seropositive human fever and serosurvey studies were recorded. A geographic information system software program c was used to assess geographic distribution and geographic relationships between dogs, cats, and affected humans. Statistical analyses Statistical analyses were performed by use of a statistical software program. d Crude percentages of positive and negative status for neutralizing antibodies against WNV were calculated for each type of study animal, and the proportions that yielded positive results were further examined. Data were stratified for analysis by environment type, environment exposure, years at address, purpose of visit to veterinary facility, application of flea control products, and heartworm medication status. Pearson χ 2 or Fisher exact tests were used to determine associations. Confounding of variables for observed effect was examined by use of stratified analysis, and when appropriate, Mantel- Haenszel adjusted odds ratios (ORs) and 95% confidence intervals (CIs) were calculated to adjust for confounding variables. For all analyses, a value of P < 0.05 was considered significant. Results Animals Final enrollment in the study included 442 (76%) dogs and 138 (24%) cats. Blood samples were collected from August 29 through October 28, 2002. Of the 580 enrollees, 453 (78%) were family pets and 127 (22%) were strays. Three hundred (52% of all animals) were dogs from Slidell, with 248 family and 52 stray dogs, and 142 (25%) were dogs from elsewhere in St Tammany Parish, with 119 family and 23 stray dogs. Eighty-three (14% of all animals) were cats from Slidell, with 57 family and 26 stray cats, and 55 (10%) were cats from elsewhere in the parish, with 29 family and 26 stray cats. Serologic results Of 580 animals tested, 116 of 442 (26%) dogs and 13 of 138 (9%) cats were seropositive for neutralizing antibodies against WNV, indicating a significantly (P < 0.001) higher seroprevalence in dogs than in cats (OR, 3.42; 95% CI, 1.86 to 6.29). Animals were stratified by environment type or family versus stray status, and this association of higher seroprevalence of dogs, compared with cats, remained significant (P < 0.001) for both groups; family dogs had odds of seropositivity > 5 times those of family cats (OR, 5.11; 95% CI, 2.01 to 13.01), and stray dogs had odds of seropositivity > 3 times those of stray cats (OR, 3.28; 95% CI, 1.35 to 7.95 [P = 0.01]). After adjusting for environment-type status (family vs stray), this overall association remained (OR, 4.21; 95% CI, 2.21 to 8.01). After adjusting for species, stray animals had odds of seropositivity that were twice those of family animals (OR, 2.04; 95% CI, 1.27 to 3.29). In individual strata, stray dogs had odds of seropositivity almost twice those of family dogs (OR, 1.89; 95% CI, 1.12 to 3.20 [P = 0.02]), and stray cats had odds of seropositivity almost 3 times those in family cats (OR, 2.95; 95% CI, 0.91 to 9.55 [P = 0.08]). Level of outdoor exposure was examined for family dogs and cats by use of indoor-only status as the reference group (Table 1). Significant (P < 0.001) differences were observed for outdoor-only family dogs, which were almost 19 times as likely to yield positive results for antibodies against WNV as indoor-only family dogs. Family dogs not receiving heartworm medication were > 2 times as likely to yield positive results as family dogs receiving heartworm medication (P = 0.01). After 1350 Scientific Reports: Original Study JAVMA, Vol 226, No. 8, April 15, 2005

Table 1 Odds ratios (ORs) and 95% confidence intervals (CIs) for seropositivity against West Nile virus (WNV) associated with environmental exposure and other factors in owned dogs and cats during a human epidemic in St Tammany Parish, La, 2002. Dogs Seropositive Seropositive Variable (% [proportion]) OR 95% CI (% [proportion]) OR 95% CI Where pet stays Indoor only 11 (2/19) Ref Ref 5 (2/41) Ref Ref Outdoor only 69 (48/70) 18.55 3.94 87.34 18 (2/11) 4.33 0.54 35.02 Both in and out NA NA NA 3 (1/34) 0.59 0.05 6.81 Pet door 20 (19/94) 2.15 0.46 10.14 NA NA NA Walks only 9 (17/180) 0.89 0.19 4.17 NA NA NA Heartworm medicine Yes 22 (73/328) Ref Ref 11 (2/18) Ref Ref No 42 (13/31) 2.52 1.18 5.39 5 (3/63) 0.40 0.06 2.60 Flea control Yes 24 (73/305) Ref Ref 5 (3/57) Ref Ref No 26 (14/55) 1.09 0.56 2.10 8 (2/25) 1.57 0.25 10.00 Ref = Referent category. NA = Not applicable. Cats Table 2 Crude (unadjusted), strata-specific, and environment exposure-adjusted associations between exposure to heartworm medication and WNV seropositivity in owned dogs during a human epidemic in St Tammany Parish, La, 2002. Analysis type Received heartworm Seropositive OR and variable medication (% [proportion]) (95% CI) Crude (unadjusted) No 42 (13/31) 2.52 (1.18 5.39)* Yes 22 (73/328) Strata-specific Indoor only No 0 (0/2) NA Yes 6 (1/16) Indoor except for walks No 31 (5/16) 5.64 (1.68 18.92)* Yes 8 (12/161) Dog door No 29 (2/7) 1.58 (0.28 8.84) Yes 20 (17/84) Outdoor only No 100 (6/6) NA Yes 65 (41/63) Environment adjusted 4.24 (1.66 10.85)* *Significant (P 0.05) association. Breslow-Day test for homogeneity of ORs revealed no significant (P 0.05) difference in ORs across strata; therefore, the adjusted (Mantel-Haenszel) OR is reported. NA = Not available (2 X 2 tables contained cells with 0 observations, so an OR could not be calculated). stratifying by level of outdoor exposure (Table 2), this association between heartworm medication and WNV seropositivity was significant (P = 0.002) only for family dogs taken outdoors for walks. However, results of the Breslow-Day test for homogeneity of ORs were not significant, indicating no difference in ORs across these strata of outdoor exposures. After adjusting for outdoor exposure, the overall association remained significant, indicating a consistent association between not receiving heartworm medication in family dogs and WNV seropositivity. There were no significant differences between WNV-positive and -negative seroprevalences with regard to family dogs and family cats and whether their given address was in Slidell or in other parish areas, the number of years the pet lived at the address, the purpose of the visit to the veterinary facility, or the application of flea control products on the pet. Geospatial results Assessment of canine, feline, and human cases in Slidell indicated that there were no WNV-seropositive family dogs or cats in the same household with a WNV seropositive human. However, there was 1 household with a human case and a seronegative dog. Discussion The WNV seroprevalence in domestic dogs in all of St Tammany Parish in 2002 (26%) was higher than that reported in New York City in 1999 (5%), and the seroprevalence in cats in the parish was 9%, although the data from the 2 studies are not suitable for statistical comparison. The higher seroprevalence of infection in the animals that were resident for < 1 year, although not significant, suggests that most infections occurred in 2002. Otherwise, an increased seroprevalence would have been expected in the other animals. The WNV seroprevalence in domestic dogs in all of St Tammany Parish was 15 times as great as the WNV seroprevalence in humans in the same area and 10 times as great as that reported in humans in the 1999 New York City study. Seroprevalence in cats in all of St Tammany Parish was 5.5 times as great as in humans in the same area. Of the 2 human WNV studies that were conducted concurrently in the same area, 1 study a found that 5 of 55 (9%) febrile outpatients had IgM antibodies against WNV. The second study b found that 21 of 1,226 (2%) randomly selected Slidell residents were seropositive against WNV. The high seroprevalence in domestic animals relative to that in JAVMA, Vol 226, No. 8, April 15, 2005 Scientific Reports: Original Study 1351

humans may have been associated with different exposure variables, including feeding preferences of infected mosquitoes and host-vector susceptibility and distribution. After adjustment for environment type and environmental exposure, dogs had higher WNV seroprevalence than cats. This suggests that there were variables related to being canine or feline that were associated with WNV infection, and those variables should be further studied. For instance, studies 11,12 assessing mosquitoes and heartworm infection (also a vector-borne disease) in dogs and cats have revealed that heartworm disease is detected at lower rates in cats than in dogs. Suggested reasons for this difference include cats not tolerating mosquito bites as well as dogs and differing feeding behavior of vector populations with respect to these animals. A higher prevalence of antibodies against WNV in dogs than cats suggests that dogs would make a better sentinel species than cats. The odds of seropositivity were approximately twice as great for stray dogs as for family dogs, and although not significant, stray cats had almost 3 times the WNV seroprevalence as family cats. Furthermore, outdoor-only family dogs and family dogs with a dog door had a higher WNV seroprevalence than indoor-only family dogs. These findings indicate higher WNV seroprevalence for animals with greater outdoor exposure. The reason that the odds of seropositivity were 2.5 times as great in family dogs not receiving heartworm medication as those receiving heartworm medication could not be determined from the available information. Results suggest that environment-exposure variables do not explain this association. After adjusting for outdoor exposure, the overall association remained significant, indicating a consistent association between not receiving heartworm medication in family dogs and WNV seropositivity. Whether heartworm medication is protective against infection with WNV should be assessed. Results of this study substantiate the need for further investigation of the potential use of dogs as sentinel indicators for WNV and the potential risk of human exposure. For arboviral diseases, a useful sentinel species for risk of human exposure would have similar vector-feeding patterns as humans, be highly susceptible to mosquito-borne infection yet resistant to disease, survive infection, develop detectable antibodies but not develop sufficient viremia to infect mosquitoes, and not infect other species. 13,14 Domestic dogs fulfill these criteria. 15,16 Studies 17-20 using domestic pet mammals, such as dogs, as sentinels have evaluated environmental health hazards to humans. The design and findings of such studies reflect the importance of the shared environment of domestic pets and humans and of using varied animal sentinel species in monitoring programs. The studies also report on animal disease cases as early markers or sentinel indicators of disease risk to humans, including using animals to monitor remedial activities against the exposure variables. During 2002, St Tammany Parish used sentinel chicken flocks to detect WNV activity via 30 flocks within the parish and 2 to 3 flocks in Slidell. e Blood samples were collected from birds twice per month. Data indicated that seroconversion in the sentinel chicken flocks was 9% from April to December, which was only a third that in dogs of the same parish. Because the greatest prevalence of antibodies against WNV was detected in stray dogs (48% in St Tammany Parish and 33% in Slidell), an option for public health departments would be testing a random selection of captured stray dogs every week at animal control facilities. This information could be used to provide a warning regarding WNV transmission to mammals in the region. In addition, this information could possibly be used to predict human cases within the area as well as for monitoring the effectiveness of mosquito control measures. a. Chow CC, Kruger J, Asamoa K, et al. West Nile fever among patients with unexplained febrile illness during an epidemic of West Nile meningoencephalitis Slidell, Louisiana, 2002, in Proceedings. 52nd Annu Epidemic Intelligence Service Conf 2003;71. b. Vicari AS, Zielinski-Gutierrez E, Montgomery SP, et al. Household-based seroepidemiological survey of West Nile virus infection Slidell, Louisiana, 2002 (abstr), in Proceedings. 52nd Annu Epidemic Intelligence Service Conf 2003;4. c. ArcView, version 8.2, ESRI, Redlands, Calif. d. SAS, version 8.02, SAS Institute Inc, Cary, NC. e. Palmisano C, St Tammany Parish Mosquito Abatement Control, Slidell, La: Personal communication, 2002. References 1. CDC. Provisional surveillance summary of the West Nile virus epidemic United States, January November 2002. Morb Mortal Wkly Rep 2002;51:1129 1133. 2. CDC. Final West Nile virus update for 2002, reported April 15, 2003. Available at: www.cdc.gov/ncidod/dvbid/westnile/ surv&controlcasecount02.htm. Accessed Jun 2, 2004. 3. Komar N, Panella NA, Boyce E. Exposure of domestic mammals to West Nile Virus during an outbreak of human encephalitis, New York City, 1999. Emerg Infect Dis 2001;7:736 738. 4. Mostashari F, Bunning ML, Kitsutani PT, et al. Epidemic West Nile encephalitis, New York, 1999: results of a householdbased seroepidemiological survey. Lancet 2001;358:261 264. 5. Office of Public Health, Department of Health and Hospitals, State of Louisiana. West Nile virus encephalitis outbreak, 2002. Available at: www.oph.dhh.state.la.us/infectiousdisease/westnileprog/docs/wntab103102.pdf. Accessed Feb 9, 2003. 6. CDC. Available at: www.cdc.gov/od/oc/media/wncount.htm. Accessed Feb 9, 2003. 7. Weathermax D. West Nile virus. CDC EIS Bull 2002;Fall:3 9. 8. US Census Bureau. US census 2000. Available at: www.census.gov/main/www/cen2000.html. Accessed Nov 22, 2002. 9. Petersen LR, Marfin AA. West Nile Virus: a primer for the clinician. Ann Intern Med 2002;137:173 179. 10. Geocode.com. Available at: www.geocode.com. Accessed Mar 24, 2003. 11. Labarthe N, Serrao ML, Melo YF, et al. Mosquito frequency and feeding habits in an enzootic canine dirofilariasis area in Niteroi, State of Rio de Janeiro, Brazil. Mem Inst Oswaldo Cruz 1998;93:145 154. 12. Miller M. Feline dirofilariasis. Clin Tech Small Anim Pract 1998;13:99 108. 13. Langevin SA, Bunning M, Davis B, et al. Experimental infection of chickens as candidate sentinels for West Nile virus. Emerg Infect Dis 2001;7:726 729. 14. Apperson CS, Hassan HK, Harrison BA, et al. Host feeding patterns of established and potential mosquito vectors of West Nile virus in the eastern United States. Vector Borne Zoonotic Dis 2004;4: 71 82. 15. Austgen LE, Bowen RA, Bunning ML, et al. Experimental infection of cats and dogs with West Nile virus. Emerg Infect Dis 2004;10:82 86. 1352 Scientific Reports: Original Study JAVMA, Vol 226, No. 8, April 15, 2005

04-05-0227.qxp 3/28/2005 1:33 PM Page 1353 19. Bukowski JA, Wartenberg D, Goldschmidt M. Environmental causes for sinonasal cancers in pet dogs, and their usefulness as sentinels of indoor cancer risk. J Toxicol Environ Health A 1998;54: 579 591. 20. Bukowski JA, Wartenberg D. An alternative approach for investigating the carcinogenicity of indoor air pollution: pets as sentinels of environmental cancer risk. Environ Health Perspect 1997; 105:1312 1319. Selected abstract for JAVMA readers from the American Journal of Veterinary Research Assessment of the hemodynamic effects of lidocaine administered IV in isoflurane-anesthetized cats Bruno H. Pypendop and Jan E. Ilkiw Objective To determine the hemodynamic effects of lidocaine (administered IV to achieve 6 plasma concentrations) in isoflurane-anesthetized cats. Animals 6 cats. Procedure Cats were anesthetized with isoflurane in oxygen (end-tidal isoflurane concentration set at 1.25 times the predetermined individual minimum alveolar concentration). Lidocaine was administered IV to each cat to achieve target pseudo steady-state plasma concentrations of 0, 3, 5, 7, 9, and 11 µg/ml, and isoflurane concentration was reduced to an equipotent concentration. At each plasma lidocaine concentration, cardiovascular and blood gas variables; PCV; and plasma total protein, lactate, lidocaine, and monoethylglycinexylidide concentrations were measured in cats before and during noxious stimulation. Derived variables were calculated. Results In isoflurane-anesthetized cats, heart rate, cardiac index, stroke index, right ventricular stroke work index, plasma total protein concentration, mixed-venous PO2 and hemoglobin oxygen saturation, arterial and mixed-venous bicarbonate concentrations, and oxygen delivery were significantly lower during lidocaine administration, compared with values determined without lidocaine administration. Mean arterial pressure, central venous pressure, pulmonary artery pressure, systemic and pulmonary vascular resistance indices, PCV, arterial and mixed-venous hemoglobin concentrations, plasma lactate concentration, arterial oxygen concentration, and oxygen extraction ratio were significantly higher during administration of lidocaine, compared with values determined without lidocaine administration. Noxious stimulation did not significantly affect most variables. Conclusions and Clinical Relevance In isoflurane-anesthetized cats, although IV administration of lidocaine significantly decreased inhalant requirements, it appeared to be associated with greater cardiovascular depression than an equipotent dose of isoflurane alone. Administration of lidocaine to reduce isoflurane requirements is not recommended in cats. (Am J Vet Res 2005;66:661 668) JAVMA, Vol 226, No. 8, April 15, 2005 April 2005 See the midmonth issues of JAVMA for the expanded table of contents for the AJVR or log onto www.avma.org for access to all the abstracts. Scientific Reports: Original Study 1353 SMALL ANIMALS 16. Blackburn NK, Reyers F, Berry WL, et al. Susceptibility of dogs to West Nile virus: a survey and pathogenicity trial. J Comp Pathol 1989;100:59 66. 17. van der Schalie WH, Gardner HS Jr, Bantle JA, et al. Animals as sentinels of human health hazards of environmental chemicals. Environ Health Perspect 1999;107:309 315. 18. Backer LC, Grindem CB, Corbett WT, et al. Pet dogs as sentinels for environmental contamination. Sci Total Environ 2001;274:161 169.