- Email: [email protected]
Case–control study of canine infection by a newly recognised Babesia microti-like piroplasm
Preventive Veterinary Medicine 61 (2003) 137–145
Case–control study of canine infection by a newly recognised Babesia microti-like piroplasm F.J. Guitián a,∗ , A.T. Camacho b , S.R. Telford III c,1 a
Department of Veterinary Clinical Sciences, Royal Veterinary College, University of London, North Mymms, Hertfordshire AL9 7TA, UK b Laboratorio Lema and Band´ın, C./Lepanto 5, Bajo, 36201 Vigo, Spain c Laboratory of Public Health Entomology, Department of Immunology and Infectious Diseases, Harvard School of Public Health, 665 Huntington Avenue, Boston, MA 02115, USA Received 23 September 2002; accepted 8 June 2003
Abstract We did a case–control study to identify risk factors for prevalent infection of dogs by a newly recognised Babesia microti-like piroplasm. Clinical manifestations and haematology of infected dogs also were described. Forty-three laboratory-based cases and 86 individually matched controls were studied. Information on clinical signs and on risk factors was collected by a questionnaire and telephone interviews. Haematology was carried out for all the dogs. Variables were screened in a bivariable conditional logistic regression and checked for colinearity. The final multivariable model was selected by backward stepwise elimination. The odds of a case having ticks when examined at the clinic was 4 times that of a control and the odds of a case being a hunting or a house-guarding dog were, respectively, 24.2 and 2.7 times those of a control. The most consistently reported clinical signs were weakness (79%), tachycardia (43%) and haemoglobinuria (42%). Mean red-blood-cell count, haemoglobin concentration, platelet count, and mean platelet volume of infected dogs were lower than the reference values and those of non-infected dogs—but leukocyte count, mean corpuscular volume and red-blood-cell distribution width were higher. © 2003 Elsevier B.V. All rights reserved. Keywords: Babesia microti; Theileria annae; Dogs; Parasitological disease; Risk factors; Odds ratio; Logistic-regression analysis
∗ Corresponding author. Tel.: +44-170-766-6507; fax: +44-170-765-2090. E-mail address: [email protected] (F.J. Guiti´an). 1 Present address: Division of Infectious Diseases, Tufts University School of Veterinary Medicine, 200 Westboro Road, North Grafton, MA 01536, USA.
0167-5877/$ – see front matter © 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0167-5877(03)00164-8
138
F.J. Guiti´an et al. / Preventive Veterinary Medicine 61 (2003) 137–145
1. Introduction Babesiosis is a parasitic disease of domestic and wild mammals; it is caused by tick-borne haematozoan organisms of the genus Babesia. The parasite normally causes a syndrome characterised by hyperthermia, anaemia and haemoglobinuria (Taboada, 1998). Traditionally, Babesia spp. infecting animals have been classified as “large” or “small” based upon their morphology and elements of their life cycles (Mehlhorn and Schein, 1987; Telford and Spielman, 1998), although all parasitologists have recognised that piroplasm systematics might be greatly revised when more information became available. In dogs, it was assumed that the only Babesia species causing disease were Babesia canis and Babesia gibsoni; solitary ring-like parasites with a diameter of 1–3 m in Romanoswky-stained thin blood smears, routinely have been identified as B. gibsoni (Anderson et al., 1979). Recent publications, however, suggest that the diversity of piroplasm species infecting dogs might be greater than previously appreciated. Recent papers provide case reports of “small” Babesia infections in Spanish dogs (Zahler et al., 2000; Camacho et al., 2001, 2002). In the first report, molecular phylogenetic analysis indicated that the infecting agent was in fact more closely related to Babesia microti, a rodent parasite in the temperate regions of the entire northern Hemisphere and the cause of human babesiosis in eastern North America; because the 18SrDNA sequence was not completely identical to that of B. microti, the new name “Theileria annae” was assigned to the canine agent. Assignment to the genus Theileria reflects a controversial argument by some parasitologists working with piroplasms that the “small” Babesia should be removed from the genus Babesia. We note, however, that elements of a characteristic Theileria life cycle (such as pre-erythrocytic stages) have not yet been reported for this agent, and thus consider the name “T. annae” to be provisional. In our extensive case series from Galicia (Camacho et al., 2001), the agent also was identified definitively by molecular phylogenetic methods as the B. microti-like parasite. The report also suggests that the infection is endemic among dogs of NW Spain, and an important cause of morbidity as reflected by haematological and serum biochemistry analysis of infected dogs. The vector for this emergent canine infection remains undescribed, although the dog-feeding Ixodes hexagonus is suspected based on their presence upon the dogs in NW Spain, and the relative absence of others (Pallas, 1998). A recent report by Camacho et al. (2003) further supports this hypothesis, showing that the odds of an infected dog being parasitised by I. hexagonus adult ticks increased fivefold compared to non-infected dogs. Our objective was to identify risk factors associated with infection of a dog with the parasite in an area previously suggested to be endemic for the infection by the B. microti-like piroplasm.
2. Materials and methods 2.1. Study design and studied animals We designed a case–control study. Cases (n = 48) were defined as dogs presenting to a local diagnostic laboratory between 20 December 2000 and 20 March 2001 for which
F.J. Guiti´an et al. / Preventive Veterinary Medicine 61 (2003) 137–145
139
small ring-shaped B. microti-like parasites were detected within giemsa-stained thin blood smears. Each case was matched with two laboratory-based controls. Controls were those dogs from which whole blood was submitted for haematological count by the same practitioner as the corresponding case, diagnosed as not infected by B. microti, and received at the laboratory within the minimum time of referral of the corresponding case. All cases which met the specified criteria and for which two controls arrived in the laboratory within ±60 days from referral of the case were included in the study. Five cases were excluded due to unavailability of controls. Median time between referral of cases and corresponding controls was 12.5 days. For controls received on the same day selection was based on the closest laboratory code (assigned each day by ascending order as samples were received). 2.2. Factors examined A questionnaire with closed questions (except for the variable “breed”) was used to obtain information on the clinical signs of B. microti-like infection (only for incident cases) and on risk factors that might influence Babesia exposure (on both case and control dogs) (Telford et al., 1993). The questionnaire data were collected via telephone interviews performed by the same interviewer. All the 23 veterinarians contacted agreed to participate. 2.3. Laboratory procedures Blood samples obtained for each of the study animals were used for haematological analysis. Haematology included leukocyte count, red-blood-cell count, haemoglobin concentration, hematocrit, red cell distribution width (RDW), mean corpuscular volume (MCV), platelet count, plateletcrit, platelet distribution width (PDW) and mean platelet volume (MPV), and was carried out by STKS-2 VCS (electrical impedance) automated Coulter counter. All microscopy analyses were performed by the same person and were done on giemsa-stained thin peripheral-blood smears. A dog was considered infected (case) when small ring-shaped intraerythrocytic piroplasms were observed in the sample. The microscopist was blinded to clinical signs, risk factors and practice from which the sample was obtained. 2.4. Data analysis Descriptive statistics for qualitative and quantitative variables under study were computed. Odds ratios were estimated by conditional logistic regression. The model-building strategy involved two steps (Hosmer and Lemesow, 1989). In the first step, all single variables were screened in a simple conditional logistic regression. Only variables significant at P < 0.25 by Wald’s test (two-tailed) were included in the second step of the analysis. Before continuing with the second step of the analysis, all the selected variables were checked for inter-correlation, and in case of a pair of variables whether there was a significant association (P < 0.05 by χ2 -test, two-tailed) and both with P < 0.25, the variable judged as most biologically plausible was used as a candidate in the multivariable analysis. The subsequent
140
F.J. Guiti´an et al. / Preventive Veterinary Medicine 61 (2003) 137–145
selection of the most parsimonious multivariable model was achieved by backward stepwise elimination of non-significant (P > 0.1, two-tailed tests) factors. To assess whether variables originally excluded from the initial (full) model because of lack of association with the outcome in the simple (univariable) model acted as confounders, the final model was re-run making those variables available. The variable age was studied both, as a continuous variable, and as a categorical variable with four categories defined by the quartiles of the ages in the studied population. A forward-selection procedure was also considered and results compared to those of the backwards elimination approach. The signed-rank test was used for comparing median haematological values for cases and controls. For comparison, values for the two matched controls were summarised using the mean. The analyses were carried out using version 7.0 of the statistical package Stata (Stata Corporation, College Station, TX).
3. Results 3.1. Risk factors The distribution of categorical hypothetical risk factors examined is shown in Table 1. Quartiles for the age (in months) were 24, 50 and 108 for non-infected dogs, and 10, 36 and 66 for infected dogs. In the simple conditional logistic-regression analyses, eight variables were associated with B. microti-like infection (P < 0.25) (Table 1). Of these, five variables were not included in the second step of the analysis because of colinearity between variables (breed, housing, habitat, vaccination and deworming) (Table 2). The remaining three variables (presence of ticks, purpose, and age) were used as candidates in the multivariable analysis and after the final step, two variables remained in the model with P < 0.10 (Table 3): presence of ticks and purpose. (The same final model was obtained when a forward-selection procedure was used. Also, the same final model was obtained regardless of the inclusion of the variable age as continuous or categorical variable.) The odds of a case having ticks when examined at the clinic was almost 4 times that of a control. The odds of a case being a hunting or a house-guarding dog were, respectively, 24.2 and 2.7 those of a control. 3.2. Clinical signs The most common clinical signs observed by the veterinarians were weakness (34 cases), haemoglobinuria (visually detected; 17 cases), tachycardia (15 cases), hyperthermia (15 cases), tachypnea (10 cases) and hepatosplenomegaly (8 cases). Controls had variety of conditions, including gastrointestinal (21), genitourinary (19), cardiovascular (15), musculo-skeletal (6), dermatological (4) and neurological (2) disease. Three had tumours and 20 were sampled because of the presence of unspecific signs such as weakness or loss of appetite.
F.J. Guiti´an et al. / Preventive Veterinary Medicine 61 (2003) 137–145
141
Table 1 Distribution of categorical factors examined between dogs infected (n = 43) and not infected (n = 86) by a B. microti-like parasite in NW Spain, 2000–2001 Factor
Infected by B. microti-like piroplasm? Yes
No
Sex Male Female
21 22
50 36
Breeda Herding Hound Toy Working Sport Terrier
8 15 5 10 2 3
19 9 20 18 12 8
Purposea,b Shepherd Hunting House-guarding Others
1 14 17 11
0 6 30 50
Presence of ticks when examined at the clinica Yes No
9 30
7 73
Housinga Indoors Kennel Garden or field
24 10 9
51 6 29
Habitata Urban Semi-urban Rural
8 6 29
32 29 25
Travel historyc Never outside the region Outside the region in the country Outside the country
36 4 1
66 10 1
Vaccinateda Not correctly Correctly
17 23
20 62
Dewormeda Not correctly Correctly
18 23
22 60
Babesiosis previously diagnosed at the clinic Yes No
5 38
14 72
Wald’s P < 0.25 (two-tailed) in bivariable analysis; offered to the multivariable model. Category “shepherd” was collapsed into “others” for subsequent analysis. c Categories “outside the region in the country” and “outside the country” were collapsed into a single category: “outside the region”, for subsequent analysis. a
b
142
F.J. Guiti´an et al. / Preventive Veterinary Medicine 61 (2003) 137–145
Table 2 Inter-correlation between selected variables to be offered to the multivariable model (two-tailed P-values for the χ2 -test of association, 43 cases and 86 matched control dogs in NW Spain, 2000–2001)
Breed Purpose Ticks Housing Habitat Vaccinated Dewormed a
Purposea
Ticksa
Housing
Habitat
Vaccinated
Dewormed
Age
<0.001
0.84 0.56
<0.001 <0.001 0.52
<0.001 <0.001 0.13 <0.001
0.02 0.002 0.72 <0.001 0.28
0.01 <0.001 0.83 <0.001 0.07 <0.001
0.13 0.15 0.20 0.19 0.010 0.18 0.55
Variables selected as candidates in the multivariable analysis.
Table 3 Final multiple conditional logistic-regression model for the outcome presence of infection by a B. microti-like piroplasm (43 cases and 86 matched control dogs in NW Spain, 2000–2001) Factor
β
P
Ticks Yes No
1.37 –
0.04
3.92 1
1.04, 14.8
Purpose House-guarding Hunting Others
1.01 3.19 –
0.04 0.006
2.75 24.2 1
1.03, 7.31 2.52, 233
Odds ratio
95% CI
3.3. Haematological findings Valid haematological counts were obtained for 43 cases and 83 controls; 25th, 50th and 75th percentiles for each of the haematological parameters under study are in Table 4. Median red-blood-cell count, haemoglobin concentration and hematocrit of infected dogs clearly were abnormal compared to the reference values and in all cases significantly lower than corresponding values for the control group. On the other hand, MCV and RDW of infected dogs were significantly higher than those of control dogs.
4. Discussion Our study demonstrates that infected dogs are more likely to have had exposure to ticks or to factors known to be associated to tick exposure (such as hunting or house-guarding activities). These results are consistent with our knowledge on the epidemiology of canine piroplasms and support the previous observation by Camacho et al. (2001) that “T. annae” is endemic among the dogs of NW Spain. The 5-year time interval with which both studies have been conducted supports this interpretation. Another fact supporting endemicity is that 36 of the studied cases had no travel history outside Galicia.
F.J. Guiti´an et al. / Preventive Veterinary Medicine 61 (2003) 137–145
143
Table 4 Results of the haematological count for 43 dogs infected by a B. microti-like parasite and 83 non-infected laboratory-based controls, 2000–2001 Parameter (normal reference range and units)
Group
Percentile
Leukocyte count ((6–17) × 103 cells/ml)
Cases Controls
Red-blood-cell count ((5.5–8.5) × 106 cells/ml)
Cases Controls
Haemoglobin (12–18 g/dl)
Cases Controls
4.7 12.4
6.4 15.1
Hematocrit (37–55%)
Cases Controls
14.1 35.6
RDW (<19%)
Cases Controls
MCV (60–77 fl.)
25th 10.8 8.6
50th
0.11
−3.47
<0.001
9.7 17.2
−8.1
<0.001
18.6 43.5
28.1 49.2
−22.45
<0.001
13.3 12.9
15.4 13.7
18.5 14.7
1.60
0.001
Cases Controls
69.6 66.8
72.6 68.8
71.1 71.0
3.35
<0.001
Platelet count ((120–500) × 103 cells/ml)
Casesc Controls
9 99
Plateletcrit (<0.35%)
Casesc Controls
PDW (<19%)
Casesc Controls
15.8 15.3
16.5 16.1
17.4 16.6
0.200
MPV (<12 mm3 )
Casesc Controls
7.6 8.4
8.7 9.7
10.5 10.8
−0.650
0.01 0.11
2.5 6.3
36 189 0.04 0.18
23.2 19.0
Pb
2.9
2.01 5.33
15.6 12.6
75th
Median difference between cases and controlsa
3.97 7.18
104 268 0.09 0.25
−162 −0.158
<0.001 <0.001 0.15 0.100
Median (value for cases − mean value for matched controls). By the signed-rank test. c N = 42. a
b
The sensitivity and specificity of the microscopic identification of B. microti in human samples were estimated to be 84 and 100%, respectively (Krause et al., 1996). Therefore, the use of this test could result in the inclusion of false negatives among the control group. False negatives are more likely to occur at the initial stages of infection. Lack of information about the prevalence of the infection prevents an assessment of the magnitude of this misclassification. However, its effect would likely be an underestimate of the true associations, being unlikely that the imperfect performance of the diagnostic test contributes to the detection of significant associations. Our results strongly support a significant clinical burden imposed (including intense anaemia) by the B. microti-like infection to dogs. Our red-blood-cells results are consistent with the previously reported values for a series of 157 animals studied during 1996: median red-blood-cell count, haemoglobin concentration and hematocrit of 2.4, 6.3 and 18.9, respectively (Camacho et al., 2001). Comparison with values obtained for the control group provide a crude measure of the likely impact of the exposure; however, caution should be
144
F.J. Guiti´an et al. / Preventive Veterinary Medicine 61 (2003) 137–145
taken when interpreting the differences, because the reference group is a laboratory-based control population. Nevertheless, the diversity of conditions for which samples from the controls were sent for diagnosis makes a systematic error unlikely and this is further supported by the observation that mean values for haematological parameters for these dogs fell within normal ranges. Leukocyte abnormalities in dogs with babesiosis are inconsistently observed (Irving and Hutchinson, 1991; Lobetti, 1995; Taboada and Merchant, 1991; Taboada, 1998). Again, in agreement to the previously published case series, leucocytosis is not a common characteristic of the disease.
5. Conclusions Infection by the newly recognised B. microti-like piroplasm is likely endemic among canines in NW Spain and is associated with severe regenerative haemolytic anaemia and thrombocytopenia. The risk of infection is greater for dogs that have been infested by ticks or are exposed to practices known to be associated with tick exposure, such as hunting.
Acknowledgements We thank Laboratorio Lema and Bandin SL and the veterinarians contacted for their co-operation during this investigation.
References Anderson, J.F., Magnarelli, L.A., Donner, C.S., Spielman, A., Piesman, J., 1979. Canine Babesia new to North America. Science 204, 1431–1432. Camacho, A.T., Pallas, E., Gestal, J.J., Guitian, F.J., Olmeda, A.S., Goethert, H.K., Telford, S.R., 2001. Infection of dogs in northwest Spain with a Babesia microti-like agent. Vet. Rec. 149, 552–555. Camacho, A.T., Pallas, E., Gestal, J.J., Guitian, F.J., Olmeda, A.S., 2002. Natural infection by a Babesia microti-like piroplasm in a splenectomised dog. Vet. Rec. 150, 381–382. Camacho, A.T., Pallas, E., Gestal, J.J., Guitian, F.J., Olmeda, A.S., Telford, S.R., Spielman, A., 2003. Ixodes hexagonus is the main candidate as vector of Theileria annae in northwest Spain. Vet. Parasitol. 112, 157–163. Hosmer, D.W., Lemesow, S., 1989. Applied Logistic Regression. Wiley, New York, pp. 82–134. Irving, P.J., Hutchinson, G.W., 1991. Clinical and pathological findings of Babesia infection in dogs. Aust. Vet. J. 68, 204–209. Krause, P.J., Telford, S., Spielman, A., Ryan, R., Magera, J., Rajan, T.V., Christianson, D., Alberghini, T.V., Bow, L., Persing, D., 1996. Comparison of PCR with blood smear and inoculation of small animals for diagnosis of Babesia microti parasitemia. J. Clin. Microbiol. 34, 2791–2794. Lobetti, R.G., 1995. Leukaemoid response in two dogs with Babesia canis infection. J. S. Afr. Vet. Assoc. 66, 182–184. Mehlhorn, H., Schein, E., 1987. The piroplasms: life cycle and sexual stages. Adv. Parasitol. 23, 37–103. Pallas, E., 1998. Aproximacion al conocimiento de la epidemiologia de la babesiosis en Galicia. Ph.D. Thesis. Universidad de Santiago de Compostela. Santiago de Compostela. Taboada, J., 1998. Babesiosis. In: Greene, C.E. (Ed.), Infectious Diseases of the Dog and Cat. Saunders, Philadelphia, pp. 473–481.
F.J. Guiti´an et al. / Preventive Veterinary Medicine 61 (2003) 137–145
145
Taboada, J., Merchant, S.R., 1991. Babesiosis of companion animals and man. Vet. Clin. N. Am.: Small Anim. Pract. 21, 103–123. Telford III, S.R., Spielman, A., 1998. Babesiosis of humans. In: Cox, F.E.G., Kreier, J.P., Wakelin, D. (Eds.), Topley and Wilson’s Microbiology and Microbial Infections, vol. 5. Edward Arnold Publications, London, pp. 349–360. Telford III, S.R., Gorenflot, A., Brasseur, P., Spielman, A., 1993. Babesial infection in man and wildlife. In: Kreier, J.P. (Ed.), Parasitic Protozoa, vol. 5, 2nd ed. Academic Press, San Diego, pp. 1–47. Zahler, M., Rinder, H., Schein, E., Gothe, R., 2000. Detection of a new pathogenic Babesia microti-like species in dogs. Vet. Parasitol. 89, 241–248.
Case–control study of canine infection by a newly recognised Babesia microti-like piroplasm F.J. Guitián a,∗ , A.T. Camacho b , S.R. Telford III c,1 a
Department of Veterinary Clinical Sciences, Royal Veterinary College, University of London, North Mymms, Hertfordshire AL9 7TA, UK b Laboratorio Lema and Band´ın, C./Lepanto 5, Bajo, 36201 Vigo, Spain c Laboratory of Public Health Entomology, Department of Immunology and Infectious Diseases, Harvard School of Public Health, 665 Huntington Avenue, Boston, MA 02115, USA Received 23 September 2002; accepted 8 June 2003
Abstract We did a case–control study to identify risk factors for prevalent infection of dogs by a newly recognised Babesia microti-like piroplasm. Clinical manifestations and haematology of infected dogs also were described. Forty-three laboratory-based cases and 86 individually matched controls were studied. Information on clinical signs and on risk factors was collected by a questionnaire and telephone interviews. Haematology was carried out for all the dogs. Variables were screened in a bivariable conditional logistic regression and checked for colinearity. The final multivariable model was selected by backward stepwise elimination. The odds of a case having ticks when examined at the clinic was 4 times that of a control and the odds of a case being a hunting or a house-guarding dog were, respectively, 24.2 and 2.7 times those of a control. The most consistently reported clinical signs were weakness (79%), tachycardia (43%) and haemoglobinuria (42%). Mean red-blood-cell count, haemoglobin concentration, platelet count, and mean platelet volume of infected dogs were lower than the reference values and those of non-infected dogs—but leukocyte count, mean corpuscular volume and red-blood-cell distribution width were higher. © 2003 Elsevier B.V. All rights reserved. Keywords: Babesia microti; Theileria annae; Dogs; Parasitological disease; Risk factors; Odds ratio; Logistic-regression analysis
∗ Corresponding author. Tel.: +44-170-766-6507; fax: +44-170-765-2090. E-mail address: [email protected] (F.J. Guiti´an). 1 Present address: Division of Infectious Diseases, Tufts University School of Veterinary Medicine, 200 Westboro Road, North Grafton, MA 01536, USA.
0167-5877/$ – see front matter © 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0167-5877(03)00164-8
138
F.J. Guiti´an et al. / Preventive Veterinary Medicine 61 (2003) 137–145
1. Introduction Babesiosis is a parasitic disease of domestic and wild mammals; it is caused by tick-borne haematozoan organisms of the genus Babesia. The parasite normally causes a syndrome characterised by hyperthermia, anaemia and haemoglobinuria (Taboada, 1998). Traditionally, Babesia spp. infecting animals have been classified as “large” or “small” based upon their morphology and elements of their life cycles (Mehlhorn and Schein, 1987; Telford and Spielman, 1998), although all parasitologists have recognised that piroplasm systematics might be greatly revised when more information became available. In dogs, it was assumed that the only Babesia species causing disease were Babesia canis and Babesia gibsoni; solitary ring-like parasites with a diameter of 1–3 m in Romanoswky-stained thin blood smears, routinely have been identified as B. gibsoni (Anderson et al., 1979). Recent publications, however, suggest that the diversity of piroplasm species infecting dogs might be greater than previously appreciated. Recent papers provide case reports of “small” Babesia infections in Spanish dogs (Zahler et al., 2000; Camacho et al., 2001, 2002). In the first report, molecular phylogenetic analysis indicated that the infecting agent was in fact more closely related to Babesia microti, a rodent parasite in the temperate regions of the entire northern Hemisphere and the cause of human babesiosis in eastern North America; because the 18SrDNA sequence was not completely identical to that of B. microti, the new name “Theileria annae” was assigned to the canine agent. Assignment to the genus Theileria reflects a controversial argument by some parasitologists working with piroplasms that the “small” Babesia should be removed from the genus Babesia. We note, however, that elements of a characteristic Theileria life cycle (such as pre-erythrocytic stages) have not yet been reported for this agent, and thus consider the name “T. annae” to be provisional. In our extensive case series from Galicia (Camacho et al., 2001), the agent also was identified definitively by molecular phylogenetic methods as the B. microti-like parasite. The report also suggests that the infection is endemic among dogs of NW Spain, and an important cause of morbidity as reflected by haematological and serum biochemistry analysis of infected dogs. The vector for this emergent canine infection remains undescribed, although the dog-feeding Ixodes hexagonus is suspected based on their presence upon the dogs in NW Spain, and the relative absence of others (Pallas, 1998). A recent report by Camacho et al. (2003) further supports this hypothesis, showing that the odds of an infected dog being parasitised by I. hexagonus adult ticks increased fivefold compared to non-infected dogs. Our objective was to identify risk factors associated with infection of a dog with the parasite in an area previously suggested to be endemic for the infection by the B. microti-like piroplasm.
2. Materials and methods 2.1. Study design and studied animals We designed a case–control study. Cases (n = 48) were defined as dogs presenting to a local diagnostic laboratory between 20 December 2000 and 20 March 2001 for which
F.J. Guiti´an et al. / Preventive Veterinary Medicine 61 (2003) 137–145
139
small ring-shaped B. microti-like parasites were detected within giemsa-stained thin blood smears. Each case was matched with two laboratory-based controls. Controls were those dogs from which whole blood was submitted for haematological count by the same practitioner as the corresponding case, diagnosed as not infected by B. microti, and received at the laboratory within the minimum time of referral of the corresponding case. All cases which met the specified criteria and for which two controls arrived in the laboratory within ±60 days from referral of the case were included in the study. Five cases were excluded due to unavailability of controls. Median time between referral of cases and corresponding controls was 12.5 days. For controls received on the same day selection was based on the closest laboratory code (assigned each day by ascending order as samples were received). 2.2. Factors examined A questionnaire with closed questions (except for the variable “breed”) was used to obtain information on the clinical signs of B. microti-like infection (only for incident cases) and on risk factors that might influence Babesia exposure (on both case and control dogs) (Telford et al., 1993). The questionnaire data were collected via telephone interviews performed by the same interviewer. All the 23 veterinarians contacted agreed to participate. 2.3. Laboratory procedures Blood samples obtained for each of the study animals were used for haematological analysis. Haematology included leukocyte count, red-blood-cell count, haemoglobin concentration, hematocrit, red cell distribution width (RDW), mean corpuscular volume (MCV), platelet count, plateletcrit, platelet distribution width (PDW) and mean platelet volume (MPV), and was carried out by STKS-2 VCS (electrical impedance) automated Coulter counter. All microscopy analyses were performed by the same person and were done on giemsa-stained thin peripheral-blood smears. A dog was considered infected (case) when small ring-shaped intraerythrocytic piroplasms were observed in the sample. The microscopist was blinded to clinical signs, risk factors and practice from which the sample was obtained. 2.4. Data analysis Descriptive statistics for qualitative and quantitative variables under study were computed. Odds ratios were estimated by conditional logistic regression. The model-building strategy involved two steps (Hosmer and Lemesow, 1989). In the first step, all single variables were screened in a simple conditional logistic regression. Only variables significant at P < 0.25 by Wald’s test (two-tailed) were included in the second step of the analysis. Before continuing with the second step of the analysis, all the selected variables were checked for inter-correlation, and in case of a pair of variables whether there was a significant association (P < 0.05 by χ2 -test, two-tailed) and both with P < 0.25, the variable judged as most biologically plausible was used as a candidate in the multivariable analysis. The subsequent
140
F.J. Guiti´an et al. / Preventive Veterinary Medicine 61 (2003) 137–145
selection of the most parsimonious multivariable model was achieved by backward stepwise elimination of non-significant (P > 0.1, two-tailed tests) factors. To assess whether variables originally excluded from the initial (full) model because of lack of association with the outcome in the simple (univariable) model acted as confounders, the final model was re-run making those variables available. The variable age was studied both, as a continuous variable, and as a categorical variable with four categories defined by the quartiles of the ages in the studied population. A forward-selection procedure was also considered and results compared to those of the backwards elimination approach. The signed-rank test was used for comparing median haematological values for cases and controls. For comparison, values for the two matched controls were summarised using the mean. The analyses were carried out using version 7.0 of the statistical package Stata (Stata Corporation, College Station, TX).
3. Results 3.1. Risk factors The distribution of categorical hypothetical risk factors examined is shown in Table 1. Quartiles for the age (in months) were 24, 50 and 108 for non-infected dogs, and 10, 36 and 66 for infected dogs. In the simple conditional logistic-regression analyses, eight variables were associated with B. microti-like infection (P < 0.25) (Table 1). Of these, five variables were not included in the second step of the analysis because of colinearity between variables (breed, housing, habitat, vaccination and deworming) (Table 2). The remaining three variables (presence of ticks, purpose, and age) were used as candidates in the multivariable analysis and after the final step, two variables remained in the model with P < 0.10 (Table 3): presence of ticks and purpose. (The same final model was obtained when a forward-selection procedure was used. Also, the same final model was obtained regardless of the inclusion of the variable age as continuous or categorical variable.) The odds of a case having ticks when examined at the clinic was almost 4 times that of a control. The odds of a case being a hunting or a house-guarding dog were, respectively, 24.2 and 2.7 those of a control. 3.2. Clinical signs The most common clinical signs observed by the veterinarians were weakness (34 cases), haemoglobinuria (visually detected; 17 cases), tachycardia (15 cases), hyperthermia (15 cases), tachypnea (10 cases) and hepatosplenomegaly (8 cases). Controls had variety of conditions, including gastrointestinal (21), genitourinary (19), cardiovascular (15), musculo-skeletal (6), dermatological (4) and neurological (2) disease. Three had tumours and 20 were sampled because of the presence of unspecific signs such as weakness or loss of appetite.
F.J. Guiti´an et al. / Preventive Veterinary Medicine 61 (2003) 137–145
141
Table 1 Distribution of categorical factors examined between dogs infected (n = 43) and not infected (n = 86) by a B. microti-like parasite in NW Spain, 2000–2001 Factor
Infected by B. microti-like piroplasm? Yes
No
Sex Male Female
21 22
50 36
Breeda Herding Hound Toy Working Sport Terrier
8 15 5 10 2 3
19 9 20 18 12 8
Purposea,b Shepherd Hunting House-guarding Others
1 14 17 11
0 6 30 50
Presence of ticks when examined at the clinica Yes No
9 30
7 73
Housinga Indoors Kennel Garden or field
24 10 9
51 6 29
Habitata Urban Semi-urban Rural
8 6 29
32 29 25
Travel historyc Never outside the region Outside the region in the country Outside the country
36 4 1
66 10 1
Vaccinateda Not correctly Correctly
17 23
20 62
Dewormeda Not correctly Correctly
18 23
22 60
Babesiosis previously diagnosed at the clinic Yes No
5 38
14 72
Wald’s P < 0.25 (two-tailed) in bivariable analysis; offered to the multivariable model. Category “shepherd” was collapsed into “others” for subsequent analysis. c Categories “outside the region in the country” and “outside the country” were collapsed into a single category: “outside the region”, for subsequent analysis. a
b
142
F.J. Guiti´an et al. / Preventive Veterinary Medicine 61 (2003) 137–145
Table 2 Inter-correlation between selected variables to be offered to the multivariable model (two-tailed P-values for the χ2 -test of association, 43 cases and 86 matched control dogs in NW Spain, 2000–2001)
Breed Purpose Ticks Housing Habitat Vaccinated Dewormed a
Purposea
Ticksa
Housing
Habitat
Vaccinated
Dewormed
Age
<0.001
0.84 0.56
<0.001 <0.001 0.52
<0.001 <0.001 0.13 <0.001
0.02 0.002 0.72 <0.001 0.28
0.01 <0.001 0.83 <0.001 0.07 <0.001
0.13 0.15 0.20 0.19 0.010 0.18 0.55
Variables selected as candidates in the multivariable analysis.
Table 3 Final multiple conditional logistic-regression model for the outcome presence of infection by a B. microti-like piroplasm (43 cases and 86 matched control dogs in NW Spain, 2000–2001) Factor
β
P
Ticks Yes No
1.37 –
0.04
3.92 1
1.04, 14.8
Purpose House-guarding Hunting Others
1.01 3.19 –
0.04 0.006
2.75 24.2 1
1.03, 7.31 2.52, 233
Odds ratio
95% CI
3.3. Haematological findings Valid haematological counts were obtained for 43 cases and 83 controls; 25th, 50th and 75th percentiles for each of the haematological parameters under study are in Table 4. Median red-blood-cell count, haemoglobin concentration and hematocrit of infected dogs clearly were abnormal compared to the reference values and in all cases significantly lower than corresponding values for the control group. On the other hand, MCV and RDW of infected dogs were significantly higher than those of control dogs.
4. Discussion Our study demonstrates that infected dogs are more likely to have had exposure to ticks or to factors known to be associated to tick exposure (such as hunting or house-guarding activities). These results are consistent with our knowledge on the epidemiology of canine piroplasms and support the previous observation by Camacho et al. (2001) that “T. annae” is endemic among the dogs of NW Spain. The 5-year time interval with which both studies have been conducted supports this interpretation. Another fact supporting endemicity is that 36 of the studied cases had no travel history outside Galicia.
F.J. Guiti´an et al. / Preventive Veterinary Medicine 61 (2003) 137–145
143
Table 4 Results of the haematological count for 43 dogs infected by a B. microti-like parasite and 83 non-infected laboratory-based controls, 2000–2001 Parameter (normal reference range and units)
Group
Percentile
Leukocyte count ((6–17) × 103 cells/ml)
Cases Controls
Red-blood-cell count ((5.5–8.5) × 106 cells/ml)
Cases Controls
Haemoglobin (12–18 g/dl)
Cases Controls
4.7 12.4
6.4 15.1
Hematocrit (37–55%)
Cases Controls
14.1 35.6
RDW (<19%)
Cases Controls
MCV (60–77 fl.)
25th 10.8 8.6
50th
0.11
−3.47
<0.001
9.7 17.2
−8.1
<0.001
18.6 43.5
28.1 49.2
−22.45
<0.001
13.3 12.9
15.4 13.7
18.5 14.7
1.60
0.001
Cases Controls
69.6 66.8
72.6 68.8
71.1 71.0
3.35
<0.001
Platelet count ((120–500) × 103 cells/ml)
Casesc Controls
9 99
Plateletcrit (<0.35%)
Casesc Controls
PDW (<19%)
Casesc Controls
15.8 15.3
16.5 16.1
17.4 16.6
0.200
MPV (<12 mm3 )
Casesc Controls
7.6 8.4
8.7 9.7
10.5 10.8
−0.650
0.01 0.11
2.5 6.3
36 189 0.04 0.18
23.2 19.0
Pb
2.9
2.01 5.33
15.6 12.6
75th
Median difference between cases and controlsa
3.97 7.18
104 268 0.09 0.25
−162 −0.158
<0.001 <0.001 0.15 0.100
Median (value for cases − mean value for matched controls). By the signed-rank test. c N = 42. a
b
The sensitivity and specificity of the microscopic identification of B. microti in human samples were estimated to be 84 and 100%, respectively (Krause et al., 1996). Therefore, the use of this test could result in the inclusion of false negatives among the control group. False negatives are more likely to occur at the initial stages of infection. Lack of information about the prevalence of the infection prevents an assessment of the magnitude of this misclassification. However, its effect would likely be an underestimate of the true associations, being unlikely that the imperfect performance of the diagnostic test contributes to the detection of significant associations. Our results strongly support a significant clinical burden imposed (including intense anaemia) by the B. microti-like infection to dogs. Our red-blood-cells results are consistent with the previously reported values for a series of 157 animals studied during 1996: median red-blood-cell count, haemoglobin concentration and hematocrit of 2.4, 6.3 and 18.9, respectively (Camacho et al., 2001). Comparison with values obtained for the control group provide a crude measure of the likely impact of the exposure; however, caution should be
144
F.J. Guiti´an et al. / Preventive Veterinary Medicine 61 (2003) 137–145
taken when interpreting the differences, because the reference group is a laboratory-based control population. Nevertheless, the diversity of conditions for which samples from the controls were sent for diagnosis makes a systematic error unlikely and this is further supported by the observation that mean values for haematological parameters for these dogs fell within normal ranges. Leukocyte abnormalities in dogs with babesiosis are inconsistently observed (Irving and Hutchinson, 1991; Lobetti, 1995; Taboada and Merchant, 1991; Taboada, 1998). Again, in agreement to the previously published case series, leucocytosis is not a common characteristic of the disease.
5. Conclusions Infection by the newly recognised B. microti-like piroplasm is likely endemic among canines in NW Spain and is associated with severe regenerative haemolytic anaemia and thrombocytopenia. The risk of infection is greater for dogs that have been infested by ticks or are exposed to practices known to be associated with tick exposure, such as hunting.
Acknowledgements We thank Laboratorio Lema and Bandin SL and the veterinarians contacted for their co-operation during this investigation.
References Anderson, J.F., Magnarelli, L.A., Donner, C.S., Spielman, A., Piesman, J., 1979. Canine Babesia new to North America. Science 204, 1431–1432. Camacho, A.T., Pallas, E., Gestal, J.J., Guitian, F.J., Olmeda, A.S., Goethert, H.K., Telford, S.R., 2001. Infection of dogs in northwest Spain with a Babesia microti-like agent. Vet. Rec. 149, 552–555. Camacho, A.T., Pallas, E., Gestal, J.J., Guitian, F.J., Olmeda, A.S., 2002. Natural infection by a Babesia microti-like piroplasm in a splenectomised dog. Vet. Rec. 150, 381–382. Camacho, A.T., Pallas, E., Gestal, J.J., Guitian, F.J., Olmeda, A.S., Telford, S.R., Spielman, A., 2003. Ixodes hexagonus is the main candidate as vector of Theileria annae in northwest Spain. Vet. Parasitol. 112, 157–163. Hosmer, D.W., Lemesow, S., 1989. Applied Logistic Regression. Wiley, New York, pp. 82–134. Irving, P.J., Hutchinson, G.W., 1991. Clinical and pathological findings of Babesia infection in dogs. Aust. Vet. J. 68, 204–209. Krause, P.J., Telford, S., Spielman, A., Ryan, R., Magera, J., Rajan, T.V., Christianson, D., Alberghini, T.V., Bow, L., Persing, D., 1996. Comparison of PCR with blood smear and inoculation of small animals for diagnosis of Babesia microti parasitemia. J. Clin. Microbiol. 34, 2791–2794. Lobetti, R.G., 1995. Leukaemoid response in two dogs with Babesia canis infection. J. S. Afr. Vet. Assoc. 66, 182–184. Mehlhorn, H., Schein, E., 1987. The piroplasms: life cycle and sexual stages. Adv. Parasitol. 23, 37–103. Pallas, E., 1998. Aproximacion al conocimiento de la epidemiologia de la babesiosis en Galicia. Ph.D. Thesis. Universidad de Santiago de Compostela. Santiago de Compostela. Taboada, J., 1998. Babesiosis. In: Greene, C.E. (Ed.), Infectious Diseases of the Dog and Cat. Saunders, Philadelphia, pp. 473–481.
F.J. Guiti´an et al. / Preventive Veterinary Medicine 61 (2003) 137–145
145
Taboada, J., Merchant, S.R., 1991. Babesiosis of companion animals and man. Vet. Clin. N. Am.: Small Anim. Pract. 21, 103–123. Telford III, S.R., Spielman, A., 1998. Babesiosis of humans. In: Cox, F.E.G., Kreier, J.P., Wakelin, D. (Eds.), Topley and Wilson’s Microbiology and Microbial Infections, vol. 5. Edward Arnold Publications, London, pp. 349–360. Telford III, S.R., Gorenflot, A., Brasseur, P., Spielman, A., 1993. Babesial infection in man and wildlife. In: Kreier, J.P. (Ed.), Parasitic Protozoa, vol. 5, 2nd ed. Academic Press, San Diego, pp. 1–47. Zahler, M., Rinder, H., Schein, E., Gothe, R., 2000. Detection of a new pathogenic Babesia microti-like species in dogs. Vet. Parasitol. 89, 241–248.