REGULAR ARTICLES

Risk factors for bovine mastitis in the Central Provinceof Sri Lanka

Suraj Gunawardana & Dulari Thilakarathne &

Indra S. Abegunawardana & Preeni Abeynayake &

Colin Robertson & Craig Stephen

Accepted: 1 May 2014# Springer Science+Business Media Dordrecht 2014

Abstract A study of the risk factors associated with mastitisin Sri Lankan dairy cattle was conducted to inform riskreduction activities to improve the quality and quantity ofmilk production and dairy farmer income. A cross-sectionalsurvey of randomly selected dairy farms was undertaken toinvestigate 12 cow and 39 herd level and management riskfactors in the Central Province. The farm level prevalence ofmastitis (clinical and subclinical) was 48 %, similar to whathas been found elsewhere in South and Southeast Asia. Fivecow level variables, three herd level variables, and eightmanagement variables remained significant (p<0.05) in thefinal logistic regression analysis. Expected risk factors relatingto unhygienic environments and inadequate knowledge orpractice of mastitis control were found. Other factors includedparity, milk yield, milking practices, access to veterinaryservices, use of veterinary products, stall structure, and stallhygiene. Many of the risk factors could be addressed bystandard dairy cattle management techniques, but implemen-tation of mastitis control programs as a technical approach is

likely to be insufficient to achieve sustainable disease controlwithout consideration of the social and political realities ofsmallholder farmers, who are often impoverished.

Keywords Cattle . Epidemiology .Mastitis . Risk . SriLanka . Dairy

Introduction

The dairy sector provides a small part of the Sri Lanka’s grossdomestic product (GDP), but it is an important agribusinessbecause dairy cattle provide rural smallholder families with acrucial source of high quality protein and livestock serves asan economic reserve for many families (Perera and Jayasuriya2008). The Sri Lankan dairy sector only produces 30 % ofdomestic demand (Perera and Jayasuriya 2008). Even thoughper capita consumption of milk in Sri Lanka is low (3.6 kg/year) compared to other southern Asian countries (Anon2013), purchases of dairy products to fill the gap betweendomestic production and demand result in a considerableproportion of GDP going toward purchase of foreign-produced milk products. The government has set a goal ofmeeting 50 % of the total milk requirement of the country by2015 (Perera and Jayasuriya 2008). The dairy industry has tobe expanded and improved if Sri Lanka is to become selfsufficient in milk production and to reduce expenditure onimports.

The Sri Lankan dairy industry is composed of local indig-enous cattle (which have low milk yields) and high yieldingdairy breeds which are found most commonly in mid countryand up country zones. The dairy sector is mainly based onsmallholder farmers keeping 2–5 cows. Cattle are reared in allfive climatic zones of Sri Lanka; however, approximately 70–75 % of the cattle and buffalo population can be found in thedry zone (dry zone rainfall averages 900 mm/year; the wettest

S. Gunawardana :D. Thilakarathne : P. AbeynayakeDepartment of Pharmacology and Public Health, Faculty ofVeterinary Medicine and Animal Science, University of Peradeniya,Peradeniya, Sri Lanka

I. S. AbegunawardanaDepartment of Basic Sciences, Faculty of Veterinary Medicine andAnimal Science, University of Peradeniya, Peradeniya, Sri Lanka

C. StephenDepartment of Ecosystem and Public Health. Faculty of VeterinaryMedicine, University of Calgary, Calgary, Alberta, Canada

C. RobertsonDepartment of Geography and Environmental Studies, Faculty ofArts, Wilfred Laurier University, Waterloo, Ontario, Canada

C. Stephen (*)Centre for Coastal Health, Nanaimo, British Columbiae-mail: cstephen@ucalgary.ca

Trop Anim Health ProdDOI 10.1007/s11250-014-0602-9

parts of the country average 5,000 mm/year (Department ofMeteorology 2013)). The total land area of Sri Lanka is65,610 sq km, of which only around two million hectares or30 % are used for agriculture (Perera and Jayasuriya 2008).

The six most common cattle diseases in Sri Lanka, basedupon the cases reported to the Department of Animal Produc-tion and Health (DAPH), Ministry of Livestock and RuralDevelopment by government field veterinary surgeons are, inorder, helminthiasis, mastitis, ephemeral fever, foot andmouthdisease, paramphistomiasis, and babesiosis (DAPH 2008).Mastitis is the major production-related disease of Sri Lankancattle, and it is one of the limiting factors impeding govern-ment goals for self-sufficiency in dairy production. Althoughthere is no ongoing surveillance for mastitis in Sri Lanka,cases are reported by national veterinarians to the DAPH. Atotal number of 5,633 mastitis cases were reported in 2009,and out of which, 1,320 cases were from the Central Province(Personnel communication, Department of Animal Health andProduction). However, the actual figure is estimated to behigher as many cases go unreported.Mastitis causes economicloss to farmers by reducing milk production, causing milk tobe discarded due to antibiotic treatment, premature culling dueto chronic infection or damaged udder tissues and by costsassociated with veterinary treatments (Hortet and Seegers1998). Expansion of the smallholder dairy sector will need todeal with food safety concerns linked to mastitis because milkcan act as a source for zoonotic bacteria (Hameed et al. 2006),and inappropriate use of antimicrobials creates risks for emer-gence of drug-resistant pathogens. Previous work in some re-gions of Sri Lanka has revealed great variability in the hygienicquality of milk (Vairamuthu et al. 2010). There is some infor-mation available on prevalence (Rupasinghe and Kulasegaram1978), economic impact (Wickramasuriya 1985), and pathogens(Fujikara et al. 1981) involved in mastitis in Sri Lanka, butrelatively little work has been done to recently or systematicallyreview risk factors despite the importance of mastitis to thedevelopment of profitable and safe dairy production.

Mastitis is a complex disease, involving a variety of path-ogenic microorganisms (Harmon 1994). Risk factors for thedisease are similarly diverse, comprising individual cow fac-tors, management variables and variations in environmentalconditions (Yamane et al. 2003). The collection of epidemio-logical data in Sri Lanka is a challenge as no official surveil-lance system exists, although government field veterinariansdo report cases sporadically. The capacity for diagnosis isoften limited to clinical findings and farm-level diag-nostics such as the California mastitis test (CMT). Thepurpose of this study was to investigate and quantifythe risk factors associated with mastitis in Sri Lankandairy cattle. Understanding the potential causes andimmediate actions to minimize risk factors may help tolimit the incidence of mastitis (Yamane et al. 2003).Local investigation was needed to determine if cow

and herd level mastitis risk factors in Sri Lanka differin their importance or type compared to what has beenestablished in epidemiological studies of mastitis con-ducted elsewhere.

Methods

Study area and animals

A cross-sectional study of randomly selected dairy farms inthe Central Province of Sri Lanka was undertaken from May2010 to June 2011. The Central Province (Fig. 1) was selectedbecause it made the greatest contribution to the national dairyindustry and it had the highest density of cattle, the greatestnumber of high producing cattle, and the highest production.According to information provided by veterinarians to theDAPH, the Central Province was reported to have the highestnumber of mastitis cases. The province is approximately5,674 km2 in area and covers three districts; Kandy, NuwaraEliya, andMatale. The altitude of Central Province ranges fromapproximately 500 m (Kandy and Matale) to 1,889 m (NuwaraEliya). The average ambient temperature, relative humidity, andrainfall in Nuwara Eliya districts are approximately 16 °C,79 %, and 1,900 mm whereas in Kandy and Matale, they were24 °C, 66 %, and 1,840 mm and 28 °C, 62 %, and 1,750 mm,respectively (Department of Meteorology 2013).

Study design and sampling

Data on the number of dairy cattle farms in the study area wereobtained from the DAPH. We were able to recruit 28 Veteri-nary Secretariat (VS) divisions to be part of this study: 14 VSdivisions from Kandy, 6 from Matale, and 8 from NuwaraEliya. Recruitment required the availability of veterinariansand their willingness to cooperate with the study. A samplesize of 386 farms was estimated using the sample size formuladescribed by Cochran (1963):

n ¼ Z2 pq=e2

where Z is the z score from the normal table for the desiredlevel of confidence (1.96 for 95 % confidence), p is theestimate proportion, ±5 % margin of error (e), and q is thecomplement of p [13], assuming p=0.5(maximum variabilityor the prevalence), ±5 % precision, and 95 % confidenceinterval.

This total sample size was distributed across the VS usingproportional simple random sampling. Proportional randomsampling is achieved when the population is divided intosubpopulations (strata), and random samples are taken fromeach stratum. In this case, a VS was a stratum. Specific farms

Trop Anim Health Prod

were randomly selected from the DAPH farm record book. Ifthe farmer was not present or we found there were no lactatingcows on our visit, the next nearest farm was selected for thesurvey. As all farms in the study area were small-scale farms ofroughly the same size and production style and because no dataon variability on management characteristics were availablebefore the study, replacement was based on proximity alone.As the herds were not large, all the lactating cows in a particularfarm were tested for mastitis and all information related to eachcow was collected regardless of their mastitis status. Informa-tion related to herd and management factors were collected at a

farm level. From the 386 farms investigated, 696 cows wereexamined for mastitis.

Cows were categorized as a case of subclinical mastitiswhen they had visibly normal udder secretions but with atleast one positive CMT quarter. Standard procedure forconducting the CMT was followed. In brief; after cleaningteats and stripping a few squirts of milk onto the ground,several milliliters were collected from each quarter into therespective wells of the CMT plate. The plate was tilted, and anequal volume of CMT solution was added to each well. TheCMT solution and milk were mixed by swirling the paddle.

Fig. 1 Map of studied farms inCentral Province of Sri Lanka (theinsert is a map of Sri Lankalocating Central Province). CHMfarm with a clinical mastitis casenonresponsive to treatment, CMfarm with a clinical mastitis case,SM farm with a subclinicalmastitis case

Trop Anim Health Prod

Clinical mastitis was diagnosed at the quarter level, based onvisible and palpable clinical signs (hard and swollen quarter,kick on touching the udder, changes in milk such as waterysecretions, clots in milk, and blood-tinged secretions) andreduction in milk production. Interpretation of CMTwas doneaccording to the standards given by the National MastitisCouncil (1999); scores were given as 1 for negative (mixtureremained liquid, homogeneous, no coagulation or gel forma-tion), 2 for weak positive reactions (coagulation of milk), 3 fordistinct positive (mixture coagulated, did not stick and dis-tinctly form a gel), and 4 for strong positive (mixture coagu-lated and stuck, tended to form jelly; swirling the cup movedmixture toward the center exposing the outer edges of thecup).

A pretested standardized questionnaire consisting of 51questions was used to collect information on each cow’sclinical history, farmer knowledge about mastitis, farm demo-graphics, and farm-level management and environmental fac-tors. Twelve cow-level risk factors and 39 herd and manage-ment risk factors were studied. These variables had beenidentified as risk factors for bovinemastitis in the internationalliterature. Dichotomous predictor variables included the fol-lowing: pregnancy status, milk fever, body condition (<3 or>3), presence of teat/udder injuries, management factors(feeding concentrates, stall feeding, animals in tie stalls, feed-ing minerals, legs tied during milking, milk removal fromudder, feed offered after milking), housing factors (use ofbedding, ventilated stalls, animal stall sunlit, roof porous,functional dung canal, use of pit or canal for waste removal,floor elevated), and milking management practices (washedthe udder with water before milking, washed the udder withhot water before milking, use of cloth for cleaning the udderbefore milking, use of coconut oil before milking, cleaned theudder with water after milking, cleaned with soap aftermilking, introduced the calf after milking). Nondichotomousvariables are listed in Table 1. A total number of 130 days offield visits were conducted during the study period, and 386people (366 farm owners and 20 laborers) were interviewed.All interviews were conducted in the farmers’ native language(Sinhala) except in 25 cases where a translator was hired forTamil-speaking farmers. Clinical examinations and the surveywere performed by the same investigator at all farms.

Data analysis

The information gathered in the questionnaires was enteredinto a Microsoft Excel worksheet. The prevalence of mastitiswas calculated as the percentage of mastitis affected cows outof the total lactating cows and was stratified on the basis oftypes of mastitis. Association between the disease and variousfactors was determined by calculating chi-square value, andthe degree of association was analyzed using the odds ratio(OR). The potential risk factors were initially screened using

univariate logistic regression using p<0.1 to identify signifi-cant relationships, and then binary logistic regression analysiswas undertaken using these significant variables. The regres-sion procedure was done under stepwise logistic regressionusing the backward Wald test. The p values for data inclusionand exclusion where; a variable was entered if p<0.01 and

Table 1 Description of nondichotomous variables used in the analysis ofrisk factors for mastitis in dairy cows in Sri Lanka

Variable Levels used in the analysis

Cow factors

Age 1 (3–6 years), 2 (7–10 years), 3(>10 years

Breed 1 (Fresian), 2 (Jersey), 3 (Ayashire), 4(cross)

Parity 1 (<3 calves), 2 (4–7 calves), 3 (>7calves)

Month of calving January–December

Milk yield 1 (<8 l), 2 (9–17 l), 3 (>17 l)

Lactation period 1 (0–4 weeks), 2 (5–9 weeks), 3 (10–14 weeks), 4 (>14 weeks)

Calving interval (days) 1 (282–382), 2 (383–483), 3 (484–584),4 (>584)

Dry period 1 (0–1.5 months), 2 (1.5–3 months), 3(>3 months)

Herd factors

Education of farmers 1 (no), 2 (1–5 grade), 3 (6–10 grade), 4(>10 grade)

Total number of animals 1 (3–5 animals), 2 (6–8 animals), 3 (>9animals)

Lactating cows 1, 2, 3 (≥3)Dry cows 0, 1, 2 (≥2)No. of calves <1 year 0, 1, 2, 3 (≥3)No. of heifers 0, 1, 2 (≥2)

Farm management practices

Source of water 1 (pipe), 2 (well), 3 (other)

Cleanings per day 1 (<1/day, 1/day), 2 (2/day), 3 (>2/day)

Dung removals per day 1 (1/day), 2 (2/day), 3 (>2/day)

Method adopted for milk letdown

1 (no), 2 (calf), 3 (coconut oil), 4 (calfand coconut oil)

Retain milk in udder 1 (no), 2 (in 1 quarter), 3 (>1 quarter), 4(change in time)

Housing

Evenness of floor 1 (floor cracks and crevices), 2 (no)

Type of bedding 1 (no), 2 (wet dirty), 3(dry clean)

Removal of bedding perday

1 (1), 2 (2), 3 (>2)

Length of cow standing area 1 (<6 ft), 2 (7–9 ft), 3 (>9 ft)

Width of cow standing area 1 (<5 ft), 2 (6–8 ft), 3 (>8 ft)

Distance from dung file tostall

1 (<1 m), 2(1–3 m), 3(>3 m)

Environment

Presence of flies 1 (no), 2 (few), 3 (average), 4 (many)

Trop Anim Health Prod

removed if p>0.05. The effects were considered significantwhen p<0.05. The exponential of the regression coefficientcalculated in logistic regression is the odds ratio associatedwith a one-unit increase an exposure variable.

Multicollinearity of independent variables was tested withthe variance inflation factor (VIF). The VIF tests the degree towhich each independent variable can be predicted by otherindependent variables in the model. AVIF threshold of 10wasused to detect problematic multicollinearity (Cohen et al.1983). Model fit was assessed with the Hosmer andLemeshow (2000) (HL) goodness of fit test. The HL test is achi-squared test used to detect problems with model specifi-cation in logistic regression models such as nonlinearity ormissing covariates. All descriptive and inferential analyseswere performed using two statistical software packages,namely SPSS 16 Copyright © SPSS Inc., 1989–2007 andWin Episcope 2.0 (http://infecepi.unizaries/ratio/spft_sp.htm).

This study was conducted as part of research that receivedethical review and approval from the University of CalgaryConjoint Health Research Ethics Board and the University ofPeradeniya Faculty of Veterinary Medicine and Animal Sci-ence Ethics Review Committee. All participants had the rightto withdraw or refuse and provided informed consent prior toinclusion. No personal information was collected.

Results

The main dairy breeds found were the following: Friesian(53 %, n=366), Jersey (28 %, n=196), Ayrshire (9 %, n=59), and Zebu or crossbrad cows (10 %, n=72). Temperatebreeds such as Friesian, Jersey, Ayrshire, and their crosseswere more common in Nuwara Eliya while Zebu and cross-bred cows were frequently present in Matale. The averagenumber of lactating cows per farm was 1.8 (696 lactatingcows/386 farms). The overall farm prevalence of mastitis(clinical and subclinical) was 48 % (n=185 farms). The pro-portion of farms affected was similar in all three districts:Nuwara Eliya 50 %, n=75/150; Matale 47 %, n=22/47;Kandy 46%, n=88/189. Of the 696 lactating cows examined,37 % (n=258) had mastitis (Nuwara Eliya 39 % [105/271],Kandy 37% [124/348], and Matale 37 % [29/77]). Only 10%of the affected cows were classified as having clinical mastitis(28/258), 19 % had clinical mastitis nonresponsive to treat-ment (n=48/258) and 71 % had sub-clinical disease (n=182/258). The patterns of mastitis were similar in all three districts:The ratio of subclinical/clinical cases was 2.3:1 for NuwaraEliya, 2.5:1 for Kandy, and 2.2:1 for Matale. Seventy onecows had a past history of mastitis, and 47%, (n=33/71) wereexperiencing negative consequences of mastitis including asignificant reduction in milk yield (17 %, n=12/71) and blindquarters (30 %, n=21/71). Farmers reported four cases of cowdeath due to systemic illness associated with mastitis.

Out of 2,784 quarters examined, 444 (16 %) were found tobe affected with mastitis. Of this, 336 quarters were subclini-cally infected while 39 were clinical mastitic quarters and 69were clinical case nonresponsive to treatment. Rear quarterswere more frequently affected (n=242, 54.5 %) compared toforequarters (n=202, 45 %), and the right side (n=257, 58 %)was more frequently affected than the left (n=187, 42 %). Outof 2,784 quarters examined, 16% (n=444) of the quarters from257 cows were CMT positive (single quarter=129, two quar-ters=93, three quarters=11, all four quarters=24). CMT grad-ing of affected quarters revealed that 27 % were CMT 2, 35 %were CMT 3 positive, and 38%were CMT 4 positive quarters.

Nuwara Eliya district farmers typically housed their cows24 h a day and practiced cut and fed system (93 %, n=140). Asimilar practice was observed among 69 % (n=130) of thedairy farmers in the Kandy district and 45 % (n=21) in theMatale district. The rest of the farmers tied up their cows atnight and tethered during the day using a semi-intensivemanagement system.

Variables with p values <0.1 in univariate analysis includedseven cow factors, five housing factors, four herd factors,three factors associated with milking, two with hygiene, andone with the farmer. The variables retained in the multivariatemodel are shown in Table 2. The VIF was lower than 10 foreach independent variable indicating that multicollinearitywas not present in the model. The p values of the HL test forboth the herd level (p=0.078) and cow level (p=0.218)models did not reveal any model misspecification.

Table 2 presents the OR and 95 % confidence interval (95 %CI) for each variable found significant in the final analysis. Fivecow level variables (pregnancy status, parity, lactation length,milk yield, and dry period length), three herd level variables(overall animal numbers, number of lactating cows, and numberof dry cows), and eight management variables (milk removal,feeding system, standing area, floor surfaces, two waste remov-al variables, presence of flies, and access to calves after milking)remained statistically significant. The highest odds ratios werefound for the following; (i) having an average number of flies(OR 5.39, 95 % CI 2.23–13.01); (ii) having >9 animals on farm(OR 4.74, 95%CI 1.62–13.85); (iii) pregnancy (OR 4.03, 95%CI 2.01–8.04); (iv) having ≥3 lactating cows on a farm (OR4.11, 95 % CI 1.89–8.89); (v) cows with milk yield of >17 l(OR 3.89, 95 % CI 1.88–8.05), and (vi) dry period of 0–1.5 months (OR 3.65, 95 % CI 2.06–6.48).

Only 1.8 % of farmers (n=7) did not allow for any dryperiod, and 8.5 % (n=33) adhered to a very short dry period(<1.5 months). The majority of farmers (n=366, 95 %)adopted abrupt cessation of milking for drying off. Dry cowtherapy has been identified as an effective way of preventingnew infections and eliminating existing infections (Eberheart1986), but 99 % of farmers interviewed had no knowledge ofdry cow therapy. Each of the four farmers (1 %) who practicesdry cow therapy used cloxacillin-based products.

Trop Anim Health Prod

In almost all of the farms, the cows were milked byhand, and occasionally, milk letdown was done throughintroduction of calves. Premilking, postmilking, andintermilking sanitations were well below the standards ofhygienic practices and were limited to washing the udderwith normal or warmwater. Followingwashing, 49% (n=190)of farmers use a cloth to dry the udder, and usually, it was the

same cloth used for every animal in the herd for severalconsequent days. Most farmers (75 %, n=288) applied lubri-cants such as coconut oil, sesame oil, gee, or hyco on the udderprior to milking to minimize pressure-induced damage tomammary gland. Only 4 % of the farms (n=14) practicedpremilking teat dipping; the majority commenced this aftermastitis cases arose.

Table 2 Odds rations and 95 %confidence limits for variablesfound to be significantly associ-ated with any type of mastitis indairy cattle in Sri Lanka by mul-tiple logistic regression

Variable and Wald statistic Levels OR (95 % CI OR)

Cow factors

Pregnancy status (21.9) Yes 4.03 2.01–8.05

Lactation period (26.8) 0–4 weeks 2.54 1.72–3.78

5–9 weeks 1.11 0.78–1.59

10–14 weeks 0.93 0.66–1.30

>14 weeks 0.5 0.36–0.69

Parity (14.1) <3 calves 0.56 0.42–0.76

4–7 calves 1.77 1.31–2.40

>7 calves 1.03 0.55–1.92

Milk yield (9.3) <8 l 0.58 0.43–0.78

9–17 l 1.31 0.97–1.77

>17 l 3.89 1.88–8.05

Dry period (14.9) 0–1.5 months 3.65 2.06–6.48

1.5–3 months 0.7 0.48–1.02

>3 months 0.55 0.32–0.94

Herd factors

Total number of animals (5.8) 3–5 0.25 0.14–0.42

6–8 2.92 1.63–5.22

>9 4.73 1.62–13.85

Number of lactating cows (5.3) 1 0.36 0.22–0.59

2 1.53 0.91–2.58

≥3 4.1 1.89–8.89

Number of dry cows (8.8) 1 0.52 0.29–0.92

≥2 1.91 1.08–3.38

Management factors

Milk removal from udder (12.7) No 1.67 1.07–2.59

Type of feeding (12.9) Stall feeding 2.79 1.73–4.52

Other 0.36 0.22–0.58

Length of cow standing area (11.6) <6 ft 1.69 1.09–2.63

7–9 ft 1.15 0.73–1.81

>9 ft 0.54 0.35–0.82

Evenness of floor (15.2) No cracks and crevices 0.37 0.24–0.57

Cracks and crevices 2.68 1.75–4.09

Waste disposal method (9.9) To pit 0.39 0.25–0.61

Open drain 2.57 1.63–4.06

Functional dung canal (4.4) No 2.12 1.33–3.38

Presence of flies (10.6) None 0.36 0.22–0.59

Few 0.87 0.74–1.03

Average 5.39 2.23–13.01

Many 1.94 1.16–3.25

After milking introduce calf (3.2) Yes 1.92 1.17–3.13

Trop Anim Health Prod

Twenty-one percent of the farmers housed their cows insheds inadequate for dairy cattle. The standard dimensions fordairy cattle were not considered in shed construction, andunsatisfactory waste disposal systems provided habitats forflies. The variables length of cow standing area, waste disposalmethod, functioning dung channel, and presence of flies wereretained in the multivariate model. Forty-nine percent of thefarms visited had not modified the floor of their cowsheds fordairy cattle, and they were either made of stones or not con-structed at all, leaving a muddy standing area. Even among thecemented floors, (n=199, 51 %), 93 (24 %) had cracks andcrevices allowing accumulation of wastewater mixed with dungor urine. We found an association of mastitis with an unevenfloor (OR 2.68, 95 % CI 1.75–4.09). Twenty-one percent (n=101) of farmers cleaned the cattle shed of beddingmaterials onlyonce every few days. Furthermore, 110 farmers (28%) confinedtheir cattle in a closed stall where light and air entered onlythrough the entrance which was an opening roughly about 2m×1.5 m. Since most stalls were covered with metal roofing sheets,these stalls were warm, dark, wet, and humid.

Stall feeding appeared to be linked with higher incidence ofmastitis (OR=2.79 (1.73–4.52)) than cows tethered at nightand loose during day time. Cows fed within their stall spentthe whole day within the cubical and thus were exposed for alonger duration to the unhygienic environment. However, mostfarmers (327 herds, 85 %) offered feed after milking whichallowed the cow to remain standing 30 min–1 h postmilking.

The farmers generally lacked knowledge of mastitis con-trol, did not maintain records, experienced misdiagnosis ofmastitis by themselves or nonveterinarians, treated with indig-enous medicine, delayed seeking veterinary service, treatedwithout laboratory diagnosis, did not adhere to therapeuticregime due to financial difficulties, experienced a shortageof recommended drugs, and had access to limited laboratorydiagnostic facilities and veterinary services.

Discussion

Bovine mastitis is multifactorial disease which involves aninteraction of microorganisms with host management factors(Sharma et al. 2012). It is well known that trends of the diseaseand associated risk factors markedly differ between geograph-ical areas (Fujikara et al. 1981). However, Sri Lankan dairyfarms were not unique in terms of the general types of riskfactors associated with bovine mastitis. Expected risk factorsrelating to unhygienic environments, inadequate knowledge orpractice of mastitis control, and cow-level variables were foundin this study. This included factors such as parity, milk yield,milking practices, access to veterinary services and use ofveterinary products, stall structure, and stall hygiene. The prev-alence of mastitis found in this study was similar to what hasbeen found in South and Southeast Asia in other studies (e.g.,

India 29.3–78.5 % and Pakistan 47.6 % (Sharma et al. 2012)andBangladesh 20.0–44.8% (Rahman et al. 2010)). This studydemonstrated that subclinical mastitis is common in smallhold-er dairying in Sri Lanka. We found that herds with moreanimals had a greater risk of mastitis. This was not consistentwith conclusions of other researchers who found that smallerherds tended to have higher incidence rate of mastitis(Fadlelmoula et al. 2007; Simensen et al. 2010), but all of theherds in this study were small, preventing a good comparisonof herd size as a risk factor. The data do not allow us todetermine if the higher risk in relative larger herds (>9 animals)reflected a statistical anomaly or was an indicator that farmer’scapacity for hygiene and animal care was overwhelmed as herdsize got larger, if physical space for cattle was restricted leadingto more teat injury or other factors were involved.

Our results suggest that future extension programs mustadvocate for farmers to develop comprehensive plan for mas-titis control. These plans need to target issues of infrastructure,education, and animal management. To be acceptable andeffective, such plans must be practical, easy to understand,effective across dairy herds, affordable, safe, and shown to beeffective. The challenge in Sri Lanka will be how to developeffective strategies to deal with well-known risk factors withinthe context of smallholder farms, with minimal infrastructureand training in a hot, humid tropical country. Smallholder dairyfarmers are some of the poorest people in Sri Lanka. Profitsfrom their cows were vital to the household income of most ofthe farmers in this study. Lost or reduced milk production,premature culling, and cow deaths contribute to severe eco-nomic losses associated with mastitis (Nielsen 2009; Sudhanand Sharma 2010). Dealing with the impact of mastitis is vitalnot just to build a national industry but also to provide incomeand safe protein to poor and vulnerable populations. Many ofthe risk factors for mastitis could be addressed by practicalmanagement of dairy cows. Implementation of mastitis controlprograms as a technical approach at the farm level is likely tobe insufficient to achieve sustainability without integration ofsocial and political realities of smallholder farmers.

Singh and Pundir (2001) saw the following as weaknesses insmallholder dairy production in Sri Lanka: “Small and scatteredanimal holdings; low milk yields; shortages of feed; poorinfrastructural/institutional facilities and support; negativetrends in cattle and buffalo population growth in the last decade;low utilization of installed milk processing capacity; unhealthycompetition among private milk collectors; poor quality ofanimal health care and breeding services; lack of professionalmanagement; and lack of a well-defined national policy fordairy development.” Farm sizes, infrastructure, human capacityand knowledge may limit the effectiveness of applying mastitiscontrol strategies developed for intensive dairy production.Priority is often given in agriculture development programs totechnologies that maximize the productivity of individual ani-mals (Randolph et al. 2007). This might not be appropriate in

Trop Anim Health Prod

the current Sri Lankan context where lack of knowledge andpoor access to animal health services may bemore foundationallimitations to reducing mastitis and boosting production.

Caution is advised in using this study to set priorities formastitis control in Sri Lanka. This is the most recent publishedepidemiological study of the bovine mastitis in this country. Thisstudy selected only one province, and there was some variationin practices, breeds, and climate within districts in that province.The Central Province may not represent all of Sri Lankan dairyfarms. While the sampling plan aimed to provide proportionalrandom sampling of the herds in the Central Province, we did notconfirm that the final sample was a true reflection of the provinceby undertaking census of the provincial herds. We are confidentthat significant biases did not result from our sampling plan dueto the similarity in size and management of farms, but werecognize the potential biases that could result from using farmerresponses to a survey. Given that the results in Sri Lanka were aswould be expected based on fundamental knowledge of mastitisin tropical countries, we do not believe these biases to be signif-icant. There were some expected mastitis risk factors that werenot retained in our final model such as milking interval, methodof milking, and premilking, postmilking, and intermilking sani-tations. Our finding of only 10 % Zebu or Zebu-cross in oursample does not reflect the higher proportion of dairy animals ofthese breeds across the country (Perera and Jayasuriya 2008).The production types are similar to what is described for thecountry, but there are variations in housing, feeding, and breedsof animals in different climate zones in the country. Nevertheless,these results provide important background data that can be usedfor farmer education and planning in the Central Province.

Acknowledgments Wewish to thank the dairy farmers of Sri Lanka fortheir cooperation. This paper has been based on work supported by fromthe Global Health Research Initiative, a collaborative research fundingpartnership of the Canadian Institutes of Health Research, the CanadianInternational Development Agency, Health Canada, the InternationalDevelopment Research Centre, and the Public Health Agency of Canada.

Conflict of interest The authors declare that they have no conflict ofinterest.

References

Anon. 2013. Dairy Development Project. Ministry of Livestock andRural Community Development. Sri Lanka. http://www.livestock.gov.lk/site/images/stories/dairy_deve.project_rev_150_last.pdf.Accessed May 16, 2013

Cochran, W.G., 1963 “Sampling Techniques” 2nd ed.( John Wiley andSons, Inc., NewYork)

Cohen, J., Cohen, P.,West, S. G. andAiken, L. S., 1983. Applied multipleregression/correlation analysis for the behavioral sciences.Lawrence Erlbaum Associates, Incorporated

Department of Meteorology.2013. Sri Lanka. www.meteo.gov.lk/index.Accessed May 16, 2013.

DAPH, 2008. Veterinary Epidemiological Bulletin Sri Lanka,Department of Animal Health and Production. Ministry ofLivestock and Rural Community Development, 1(1),1—8

Eberheart, R.J., 1986. Management of dry cows to reduce mastitis,Journal of Dairy Science, 69(6),1721—1732

Fadlelmoula, A., Fahr, R.D., Anacker, G. and Swalve, H.H., 2007. Themanagement practices associated with prevalence and risk factors ofmastitis in large scale dairy farms in Thuringia, Germany 1: envi-ronmental factors associated with prevalence of mastitis, AustralianJournal of Basic and Applied Sciences, 1(4), 619—624

Fujikara, T., Rupasinghe, T., Ramakrishnaswany, A. Kulasegaram, P.,Santiyago, G. L.,and Prabakaran, P., 1981. Epizootiological studyon bovine mastitis in Sri Lanka. Japan Agricultural ResearchQuarterly, 15, 249—25

Hameed, K.G.A., Sender, G. and Korwin-Kossakowska A., 2006. Publichealth hazard due to mastitis in dairy cows, Animal Science Papersand Reports, 25(2),73—85

Harmon, R.J., 1994. Physiology of mastitis and factors affecting somaticcell counts, Journal of Dairy Science,77(7), 2103—2112

Hortet, P. and Seegers, H., 1998. Calculated milk production lossesassociated with elevated somatic cell counts in dairy cows: reviewand critical discussion, Veterinary Research, 29(6), 497–510

Hosmer, D. W. and Lemeshow, S., 2000. “Applied Logistic Regression”,2nd ed, (John Wiley & Sons. New York)

National Mastitis Council. 1999. Laboratory handbook on bovine masti-tis. (National Mastitis Council Madison. WI)

Nielsen, C., 2009. Economic impact of mastitis in dairy cows, DoctoralThesis Swedish University of Agricultural Sciences Uppsala

Perera B.M.A.O. and Jayasuriya, M.C.N., 2008. The dairy industry in SriLanka: current status and future directions for a greater role in nationaldevelopment, Journal of National Foundation Sri Lanka, 39, 115–126

Randolph, T. F., Schelling, E., Grace D., Grace, D., Nicholson, C.F.,Leroy, J.L., Cole, D.C., Demment, M.W., Omore, A., Zinsstag, J,,and Ruel M., 2007. Invited review: role of livestock in humannutrition and health for poverty reduction in developing countries,Journal of Animal Science, 85, 2788—2800

Rahman, M. M., Islam, M. R., Uddin, M. B. and Aktaruzzama, M., 2010.Prevalence of subclinicalmastitis in dairy cows reared in Sylhet districtof Bangladesh, International Journal BioResearch, 1(2), 23—28

Rupasinghe T. and Kulasegaram, P., 1978. Incidence and aetiology ofsubclinical mastitis in cows in Sri Lanka, Ceylon Veterinary Journal,26(1/4), 22—28

Sharma, N., Rho, G.J. and Hong, Y.H., Kang, T.E., Lee, H.K., Hur. T-Y.,and Jeong, D.K., 2012. Bovine mastitis: an asian perspective, AsianJournal of Animal & Veterinary Advances, 7(6), 454—476

Simensen, E., Østerås, O., Bøe K. E., Kielland, C., Ruud, L.E., and Næss,G., 2010. Housing system and herd size interactions in Norwegiandairy herds; associations with performance and disease incidence,Acta Veterinaria Scandinavica, 52(14), 1—9

Singh, K. and Pundir, R.S., 2001. Problems and prospects of smallholderdairy production and marketing in South Asia: an overview,Proceedings of a South–South workshop held at National DairyDevelopment Board (NDDB) Anand, India, 88—104

Sudhan, N.A. and Sharma, N. 2010. Mastitis—an important productiondisease of dairy animals, SMVS Dairy Year Book, 72–88,

Vairamuthu, S., Sinniah, J. and Nagalingam, K., 2010. “Factors influenc-ing production of hygienic raw milk by small scale dairy producersin selected areas of the Jaffna district, Sri Lanka,” Tropical AnimalHealth and Production, 42(3), 357—362

Wickramasuriya, U.G.J.S., 1985. Evaluation of a mastitis control pro-gramme adopted in small dairies in the district of Colombo, Journalof National Science Council Sri Lanka, 13(1), 17—27

Yamane , I., Eguchi , M., Tsutsui T., and Shimura, K., 2003. Investigationof risk factors associated with the incidence of bovine mastitis, The11th international symposium of theWorld Association of VeterinaryLaboratory Diagnosticians and OIE Seminar on Biotechnology, 1–3

Trop Anim Health Prod