Swai_211-217.qxd INTRODUCTION Cattle ticks of the genus Boophilus are the only vectors of Babesia bovis infection (Mahoney 1979), and of these, Boophilus microplus larvae have been singled out as the most efficient transmitting agents of Babesia bovis in areas were this infection exists (Riek 1966). The classic epidemiological model of babesiosis in cattle is based on the long lasting immunity induced by primary infection in calves up to 7–9 months old (Riek 1968) when they are natu- rally resistant to the clinical effects of primary infec- tion. In those herds subjected to Babesia bovis infections are from 0.0005 to 0.005 infective bites/ head/day, are in endemic instability, while the risk is low, below or above those limit (Mahoney & Ross 1972). In this article we present the results of a cross-sec- tional study estimation of the prevalence of Babesia bovis antibodies in dairy cattle on the two contrast- ing and diverse smallholder dairying regions in Tan- zania where Boophilus microplus is often thought not to exist (Lynen, Bakuname & Sanka 1999). The objective were to quantify the occurrence of Babe- sia bovis through a serological survey, estimate the rate of infection of it and explore the possible rela- tionship between the it and some animal/ farm level variables. 211 Onderstepoort Journal of Veterinary Research, 71:211–217 (2004) Cross-sectional estimation of Babesia bovis antibody prevalence in cattle in two contrasting dairying areas in Tanzania E.S. SWAI1, E.D. KARIMURIBO2, N.P. FRENCH3, N.H. OGDEN4, J. FITZPATRICK5, D. KAMBARAGE2 and M.J. BRYANT6 ABSTRACT SWAI, E.S., KARIMURIBO, E.D., FRENCH, N.P., OGDEN, N.H., FITZPATRICK, J., KAMBARAGE, D. & BRYANT, M.J. 2004. Cross sectional estimation of Babesia bovis antibody prevalence in cat- tle in two contrasting dairy areas in Tanzania. Onderstepoort Journal of Veterinary Research, 71: 211–217 The crude prevalence of antibodies to Babesia bovis infection in cattle was estimated by serology using indirect ELISA during the period January to April, 1999. Sera were obtained from 1 395 dairy cattle (of all ages, sexes and breeds) on smallholder farms, the majority being kept under a zero grazing regime. The crude prevalence of antibodies to Babesia bovis was 6 % for Tanga and 12 % for Iringa. The forces of infection based on the age sero-prevalence profile, were estimated at six for Iringa and four for Tanga per 100 cattle years-risk, respectively. Using random effect logistic regres- sion as the analytical method, the factors (variables) of age, source of animals and geographic loca- tion were hypothesised to be associated with sero-positivity of Babesia bovis in the two regions. Keywords: Babesia bovis, dairy cattle, epidemiology, force of infection, sero-prevalence, small- holder, Tanzania 1 Veterinary Investigation Centre, Box 1068 Arusha, Tanzania. E-mail: emasw@yahoo.co.uk 2 Faculty of Veterinary Medicine, Sokoine University of Agri- culture, Morogoro, Tanzania 3 Department of Veterinary Clinical Science and Animal Hus- bandry, University of Liverpool, UK 4 Department of Veterinary Pre-clinical Science, University of Liverpool, UK 5 Department of Veterinary Clinical Studies, University of Glas- gow, Glasgow, UK 6 Department of Agriculture, University of Reading, UK Accepted for publication 23 February 2004—Editor MATERIALS AND METHODS Area and study population The study was carried out in five administrative dis- tricts of Tanga region and two in the Iringa region of Tanzania. Tanga region is situated on the North Eastern corner of Tanzania lying between longitude 36 and 38E and latitude 4 and 6S. The region has heterogeneous physical and climatic features vary- ing from hot, humid coastal lowlands in the east to the cool Usambara Mountains in the north and semi- arid plain in the southwest. There are two rainy sea- sons, the long rains occurring between March and May and the short rains occurring between Septem- ber and November. Rainfall varies widely from 500 mm in semi-arid areas to 1 400 mm in coastal areas and up to 2 000 mm in some inland mountain areas. Daytime temperatures vary from 23–28 °C during the cool season (May to September) to 30–33 °C during the hot season (December to March). Iringa region is one of three in the Southern high- land zone of Tanzania and lies between latitude 7 and 8S and longitude 35 and 36E. The region lies between 1 340 and 2 090 m above sea level. Rain- fall is annually bimodal and ranges between 600 and 1 600 mm per annum, with most rain falling between March and June and occasional light rain between August and September. In both regions, type of cattle kept includes Bos taurus breed (Fries- ian, Ayrshire, Jersey, Simmental) and crosses of these breeds with Bos indicus breeds (Tanzania shorthorn zebu, Boran and Sahiwal). Study design A sample size of farms and animals was estimated using Epi-Info version 6.04b (CDC, Atlanta, USA) in order to provide 80 % power, with a confidence of α = 0.05, to estimate disease prevalence and detect associations between dependent and independent variables (Gitau, McDermott, Walter-Toews, Lisse- more, Osumo & Muruki 1994; French & Tyrer 1997). The farms in each study region were randomly (Epi- Info version 6.04b) selected in October 1998, from a sampling frame of 3 001 and 500 (in Tanga and in Iringa, respectively) using the databases of the Tanga and Iringa Dairy Development Projects. The Dutch and Swiss governments have been support- ing dairy schemes in Tanga and Iringa regions, re- spectively. Over the last 15 years (1983–1998) of support, huge data bases have been generated. As the antibody prevalence was not known a priori, a 50 % prevalence was assumed when calculating the total number of farms required for the study, with a 10 % allowable error. The farms in both study sites were estimated (on previous experience) to have an average of three to four dairy stock of any age, breed or sex. A farm sample size of 200 in each study region was considered necessary to provide between 600 and 800 animals for the study and to allow for a “design effect” of 2.0. Farms having more than ten animals were excluded from the selection process because such herds are not considered as “smallholder” enterprises (Tanga Dairy Development Programme 1999). Questionnaire design and administration One person in each region collected most farm- level and some animal-level data using a structured questionnaire, which was completed by the small- holder on all the selected farms on a single visit. The questionnaire was designed to comprise most- ly closed ended (categorical) questions to ease data processing, minimize variation and improve precision of responses (Thrusfield 2000). Question- naire administration and collection of data from ani- mals were carried out by two separate teams, dur- ing the period of January to April 1999, in the two regions. Although comparisons are made between the two regions, the data were not combined for analyses to allow any unexplained variation due to variations in the precise way that data were collect- ed to be estimated. The information gathered concerned farm and ani- mal events that had occurred during 1998 and included cattle location, source (homebred or brought-in), mode of acquisition of dairy stock, sex, level of exotic blood (Filial generation), breed codes, age and housing practices, as well as whether or not a system of zero-grazing was practised of if the cattle had been allowed to graze on pastures in the 3 months prior to sampling. The detailed variables studied have been described by Swai (2002). The responses to many of these questions were inves- tigated as explanatory variables in the analyses of sero-conversion to Babesia bovis. Collection of sera and their analysis During the visit to each farm, blood samples were collected by jugular venipuncture into 10 ml “Vacu- tainer” tubes (Becton Dickson, UK) from all animals on the farm. These were labelled and transported in a refrigerated cool box to local laboratories where aliquots of sera were obtained by centrifugation at 3 000 g for 20 min after which they were then stored 212 Babesia bovis antibody prevalence in cattle in Tanzania at –20 °C at Sokoine University of Agriculture (SUA) prior to dispatch in refrigerated containers to the International Livestock Research Institute (ILRI) in Nairobi, Kenya where they were subjected to indi- rect enzyme-linked immunosorbent assay (ELISA) in order to evaluate the level of antibodies to Babe- sia bovis (Katende, Goddeeris, Nkonge, Morzaria & Musoke 1990; Katende,Toye, Skilton, Nene, Mor- zaria & Musoke 1998). The ELISA has shown a sensitivity of 99 % and a specificity of 98 % (Ka- tende et al. 1998). The results were expressed as percent positivity (PP) values of optical densities (Wright, Nilsson, Van Rooij, Lelenta & Jeggo 1993), relative to that of a strongly positive control serum. Each test serum sample was analysed in duplicate and that of the control sera in triplicate. The thresh- old level of PP for positivity was 25 % for Babesia bovis. Statistical methods Descriptive statistics for the animal- and farm-level explanatory variables examined in the study were developed using Epi-Info, version 6.04d. The rela- tionships between explanatory variables and out- come response (sero-conversion to Babesia bovis) were investigated in two steps by logistic regression (using Egret for Windows version 2.0, Seattle, USA) with “farm” as a random effect because animals on one farm may not have been statistically independ- ent of one another (Kristula, Curtis, Galligan & Bar- tholomew 1992). In the first step, the relationships between each explanatory and outcome variable were individually investigated. In the second step, any variables that were significantly associated at the P < 0.25 level were included in multivariable models producing, by forwards and backwards sub- stitution and elimination, the most parsimonious models in which all explanatory variables remained significant at the P < 0.05 level. The criteria for inclu- sion and exclusion were a change of in deviance significant at the 5 % level according to the maxi- mum likelihood ratio test—Chi square distribution. Forces of infection were estimated from age sero- prevalence profiles using Maximum Likelihood Meth- ods (MLM) in Excel (Microsoft, USA) with solver add-in (Thrusfield 2000). Assuming a stable popu- lation size and age structure and a constant force of infection across all age groups, the log likelihood was derived using the following formula: where Ri = number of sero-positive in group i, Ni = number tested in age group i and λ = the force of infection RESULTS Farm participation All selected 200 farms from each of Tanga and Iringa regions were visited and farmers interviewed during the period of January 1999 to April 1999 (a 100 % response rate). In Tanga, a total of 697 ani- mals kept on 185 (92.5 %) farms were examined. Fifteen farms (7.5 %) had no animals during the actual survey period. In Iringa, a total of 698 animals from 195 (97.5 %) farms were examined. Three farms (1.5 %) had no animals and two farms (1 %) could not be sampled because the owner could not be traced. The number of animals examined per herd was 3 ranged from 1 to 9 animals. The distri- bution of cattle amongst categories of each variable investigated is summarised in Table 1. Serological responses to Babesia bovis The crude sero-prevalence of antibodies to Babesia bovis was 6 % (4.7, 8.6) and 12 % (9.6, 14.7) for Tanga and Iringa, respectively (Table 2). The estimated force of infection was 0.04 and 0.06 per animal years-risk for Tanga and Iringa, respec- tively. The graphical forces of infection are shown in Fig.1 and 2. The estimated force of infection was slightly higher in Iringa than in Tanga. The results of the final logistic regression model are detailed in Tables 3 and 4. In both Tanga and Iringa regions sero-prevalence was significantly greater in grazed animals than in zero-grazed animals. In Iringa region animals locat- ed in Iringa urban district were significantly more likely to be sero-positive for Babesia bovis than ani- mals on farms in Iringa rural district (OR = 5.83, P = 0.010). In both regions cattle brought onto the farms were significantly more likely to be sero-pos- itive than animals born on-farm (OR = 2.43, P = 0.035 and OR = 3.1, P = 0.016, respectively for Iringa and Tanga regions). DISCUSSION The prevalence of Babesia bovis infection as reflect- ed by ELISA appears to be comparatively higher in Iringa (12 %) than in Tanga (6 %). However, the 213 E.S. SWAI et al. Loglikelihood L = ∑ Ri lne–λi + (Ni – Ri) ln (1 – e–λi) a i = 1 214 Babesia bovis antibody prevalence in cattle in Tanzania TABLE 1 The proportions of cattle in each category of each variable investigated during the study No. of animals (%) Variable Categories Iringa Tanga Animal-level variables Sex Male 182 (26) 146 (21) Female 516 (74) 551 (79) Source of animal Homebred 406 (58) 436 (63) Brought-in 292 (41) 261 (37) Filial generation F1 350 (50) 217 (31) F2 347 (49) 459 (66) F3 1 (0.1) 21 (3) Breed codes Ayrshire cross 403 (58) 169 (24) Friesian cross 305 (44) 604 (86) Jersey cross 1 (0.1) 12 (2) Simmental cross 0 5 (1) Sahiwal cross 0 12 (2) TSHZ cross 150 (22) 541 (77) Boran cross 549 (78) 121 (17) Age < 3 years 440 (63) 396 (57) 3 to < 6 years 165 (24) 214 (31) > 6 years 93 (13) 87 (12) Grazing history in 1998 Zero grazing 489 (70) 631 (90) Semi/free grazing 209 (30) 66 (10) Farm-level variables Farm classification Peri-urban 109 (16) 117 (17) Urban 391 (56) 318 (46) Rural 198 (28) 262 (37) District (Iringa) Iringa urban 461 (66) NA Iringa rural 237 (34) District (Tanga) Tanga NA 341 (48.8) Muheza 185 (26.5) Pangani 26 (4) Korogwe 64 (9) Lushoto 81 (12) Mode of acquisition Bought cash 471 (67) 268 (38) Credit (HIT) 117 (17) 356 (51) Others (gift) 110 (16) 73 (11) TABLE 2 The prevalence (with +/– 95 % confidence intervals) of cattle sero-positive for Babesia bovis in the study regions (January to April 1999) Prevalence 95 % CI Region No tested No positive (%) Lower Upper Tanga 697 42 6 4.7 8.6 Iringa 698 84 12 9.6 14.7 estimated prevalences were low when compared to the findings of 88 % by Woodford, Jones, Rae, Boid & Bell-Saikyi (1990) in Pemba, Tanzania; over 60% by Perez, Herrero, Jimenez, Carpenter & Buening (1994) in Costa Rica; and 73 % of dairy cattle by Maloo, Thorpe, Kioo, Ngumi, Rowland & Perry (2001) in coastal Kenya. There was also evidence of wide spread of this pathogen within the studied districts, particularly in Iringa. The distribution of Babesia bovis infection was limited to some districts in Tanga. This distribu- tion pattern may suggest that a situation of “endem- ic instability” prevails and that a similar distribution of the vector ticks (Boophilus spp.) occurs in the district where the Babesia infection is known to exist. The low serological prevalence for Babesia bovis infection may be due to its poor infectivity for Babesia (Mahoney & Mirre 1971) and subsequent- ly the occurrence of lower inoculation rates in sus- ceptible cattle (Mahoney 1979). The lower observed inoculation rates (force of infection), in the light of these findings, may predict an enhanced tick resist- ance (Mahoney 1979) or a genetically based immu- nity to Babesia bovis characteristic of zebu crosses, which results in low inoculation rates of this Babesia parasite. As in other studies, in both Tanga and Iringa regions, a history of recent grazing prior to sampling was 215 E.S. SWAI et al. TABLE 3 Final logistic binomial multiple regression for sero-prevalence of Babesia bovis in dairy cattle, Tanga, Tanzania (January to April 1999) Factor β(SE) OR Lower to upper Wald P Likelihood ratio P 95 % CI Constant –4.33 (0.56) Grazing vs zero grazing 1.84 (0.57) 6.34 2.05–19.56 0.001 < 0.001 Brought vs Homebred 1.14 (0.47) 3.15 1.23–8.03 0.016 < 0.001 Random term 1.44 (0.38) TABLE 4 Final logistic binomial multiple regression for sero-prevalence of Babesia bovis in dairy cattle, Iringa, Tanzania (January to April 1999) Factor β(SE) OR Lower to upper Wald P Likelihood ratio P 95 % CI Constant –5.57 (0.82) Grazing vs zero grazing 1.29 (0.48) 3.65 1.40–9.49 0.007 < 0.001 Brought vs Homebred 0.88 (0.42) 2.43 1.06–5.56 0.035 < 0.001 Iringa Urban vs Iringa Rural 1.76 (0.68) 3.83 1.51–22.53 0.010 0.032 Age centred 0.008 (0.006) 0.062 Random term 2.06 (0.40) FIG. 1 Force of infection estimates for Babesia bovis infection in dairy cattle—Iringa, Tanzania (January to April 1999) FIG. 2 Force of infection estimates for Babesia bovis infection in dairy cattle—Tanga, Tanzania (January to April 1999) �������� �������� ���������������������� � � �� �� �� � � � � � � ��������� � � � �� � � � �� � � � � � � � � �� � � �� �� �� � � �� �� �� � � � � � � ��� ��� ��� ��� � ��� ��� ��� ��� � � � � � � � � � �� � � � � � � � � � �� �� � ��� ��� ��� ��� � ��� ��� ��� ��� �������� �������� ���������������������� � � �� �� �� � � � � � � � � �� � � � �� � � � � � � � � �� � � �� �� �� � � �� �� �� � � � � � � ��������� associated with a significantly higher likelihood of an animal being sero-positive to Babesia bovis infec- tion, compared to zero-grazed animals (Maloo et al. 2001). Sero-prevalence varied with the mode of acquisi- tion of the animals. Accounting for age, brought-in animals were more likely to be sero-positive for Babesia bovis than were homebred animals. This would be consistent with the fact that most dairy stock available for sale in local markets are preg- nant heifers from specialized large-scale cross- breeding ranches where they are mostly grazed rather than housed (Swai 2002). Allowing for variations in sero-prevalence related to age and management factors, such as grazing, geographic variation in sero-prevalence of Babesia bovis infection was also determined. This variation was more conspicuous in the Iringa study region with the odds of infection being three fold for cattle situated in Iringa urban areas when compared to that in the cattle in the Iringa rural district. This implies that there must be geographical variations in either the density of tick vectors or the preva- lence of Babesia bovis infection in host-seeking ticks, or both. Variations in sero-prevalence may partly be owing to cattle immune responses (Ma- honey 1979). Geographic variation in sero-preva- lence was more uniform and generally higher in the Iringa region. Babesiosis has not been reported to be a major clinical problem in both study regions (TDDP 1999; Southern Highland Dairy Develop- ment Programme 1995) but there might be under- estimations of clinical babesiosis in these regions. A diagnosis of bracken poisoning is frequently made by farmers and extension workers when cattle show clinical signs of “red water” (haemoglobinuria or haematuria) and in the light of the findings in this survey, some or many of these cases may, in fact, be misdiagnosed cases of babesiosis. An increased trend of sero-positivity associated with age was evident in this study. These findings are consistent with previous reports (Hugh-Jones, Busch & Jones 1988) which showed young animals to be more resistant to primary infections. We decided a priori to allow for intraherd correlation in the analysis by incorporating “farm” as a random- effect term in all models. However in common with similar studies (Gitau et al. 1994), there was little evidence of clustering at the level of farm, probably owing to the small number of animals per farm (McDermott & Schukken 1994) CONCLUSION The results of this study can be summarized as fol- lows: • Sero-prevalence of antibodies to Babesia bovis was low in smallholder dairy cattle in the study regions, most likely due to zero-grazing man- agement, but there was evidence of a wide dis- tribution in both the study regions. • Consistent with this the force of infection varied geographically. It was, however, comparatively low and the likelihood of a bovine encountering the infection increased significantly with age. • Farmer reporting of grazing significantly in- creased the likelihood of contact of cattle with infective ticks. • The source of animals, particularly recently pur- chased animals, may be more likely to have en- countered and recovered from Babesia bovis infection. ACKNOWLEDGEMENTS We thank the Government of UK through DFID\ NRRD Animal Health Research Programme for financing this work which formed part of a study for the Ph.D. degree on the epidemiology of tick-borne diseases in small-scale farming systems in Tan- zania. The generous cooperation of smallholder farmers, extension staff and laboratory personnel at ILRI/SUA is also acknowledged. Our thanks are extended to the Director of Veterinary Service, Tan- zania for permission to publish this work. REFERENCES CYTEL SOFTWARE CORPORATION 1999. Statistics and Epi- demiology Research Corporation, ver. 2.0. Seattle, USA. EPI-INFO 1996. Epi-info. Centre for Disease Control, version 6.04d, Atlanta, USA and Geneva, Switzerland. FRENCH, N.P. & TYRER, J. 1997. Birth and death of cattle on small scale dairy farms in Zimbabwe. Society of Veterinary Epidemiology and Preventive Medicine, Chester. Poster presentation. GITAU, G.K., MCDERMOTT, J.J., WALTNER-TOEWS, D., LIS- SEMORE, K.D., OSUMO, J.M. & MURIUKI, D. 1994. Fac- tors influencing calf morbidity and mortality in smallholder dairy farms in Kiambu District, Kenya. Preventive Veterinary Medicine, 21:167–177. HUGH-JONES, M.E., BUSCH, R.C. & JONES, F. 1988. Sero- prevalence survey of Anaplasma marginale card–test reac- tors in Louisiana, USA, cattle. Preventive Veterinary Medicine, 6:143–153. 216 Babesia bovis antibody prevalence in cattle in Tanzania KATENDE, J.M., GODDEERIS, B.M., NKONGE, C.G., MOR- ZARIA, S.P & MUSOKE, A.J. 1990. Identification of a Theile- ria antigen for use in an antibody and antigen detection ELISA. Parasite Immunology, 12:419–433. KATENDE, J.M., TOYE, P., SKILTON, R.A., NENE, V., MOR- ZARIA, S.P & MUSOKE, A.J. 1998. An ELISA for detection of T. parva antibodies using recombinant immunodominant molecule. Parasitology Research, 84:408–416. KRISTULA, M.A., CURTIS, C.R., GALLIGAN, D.T & BARTHOL- OMEW, R.W. 1992. Use of a repeated measures logistic regression model to predict chronic mastitis in dairy cows. Preventive Veterinary Medicine, 14:57–68. LYNEN, G., BAKUNAME, C. & SANKA, P. 1999. Tick and TBD survey in Northern Regions of Tanzania. Proceedings of Tan- zania Veterinary Association Conference, Arusha, Decem- ber, 1999. MAHONEY, D.F. & MIRRE, G.B. 1971. Annals of Tropical Med- icine and Parasitology, 65:309–317. MAHONEY, D.F. & ROSS, D.R. 1972. Bovine babesiosis: Com- puter simulation of Babesia argentina rates in Bos taurus cattle. Annals of Tropical Medicine and Parasitology, 68: 385–392. MAHONEY, D.F. 1979. Babesiosis in domestic animals, in Para- sitic protozoa, Vol. IV, edited by J.P. Kreier. New York: Aca- demic Press. MALOO, S.H., THORPE, W., KIOO, G., NGUMI, P., ROWLAND, G.J. & PERRY, B.D. 2001. Sero-prevalence of vector trans- mitted infections of small-holder dairy cattle in coastal Kenya. Preventive Veterinary Medicine, 52:1–16. McDERMOTT, J.J & SCHUKKEN, Y.H. 1994. A review of the methods used to adjust fro cluster effects in explanatory epi- demiological studies of animal populations. Preventive Veterinary Medicine, 18:155–173. PEREZ, E., HERRERO, M.V., JIMENEZ, C., CARPENTER, T.E. & BUENING, G.B. 1994. Epidemiology of bovine anaplas- mosis and babesiosis in Costa Rica. Preventive Veterinary Medicine, 20:23–31. RIEK, R.F. 1966. The life cycle of Babesia argentina in the tick vector Boophilus microplus. Australian Journal of Agricultural Research, 17:247–254 RIEK, R.F. 1968. Babesiosis, in Infection blood disease of man and animals, edited by D.F. Weinmann & M. Ristic. New York: Academic Press. SOUTHERN HIGHLAND DAIRY DEVELOPMENT PRO- GRAMME. 1995. Annual Progress Report, Iringa. SWAI, E.S. 2002. Epidemiological studies of tick borne diseases in small scale dairy farming systems in Tanzania. Ph.D. the- sis, University of Reading. TANGA DAIRY DEVELOPMENT PROGRAMME 1999. Annual Progress Report, Tanga. THRUSFIELD, M. 2000. Veterinary Epidemiology. UK, Blackwell Science WOODFORD, J.S., JONES, T.W., RAE, P.F., BOID, R. & BELL- SAIKYI 1990. Sero-epidemiological studies of bovine babesiosis on Pemba Island, Tanzania. Veterinary Parasitol- ogy, 37:175–184. WRIGHT, P.F., NILSSON, F., VAN ROOIJ, E.M.A., LELENTA, M. & JEGGO, M.H. 1993. Standardisation and validation of enzyme linkage immunosorbent assay technique for detec- tion of antibody in infectious disease diagnosis. Revue Sci- entifique Technique office international des Epizooties, 12: 435–450. 217 E.S. SWAI et al.