A

LDa

b

a

ARRA

KEACC

1

TglTtcalr

vT

0d

Veterinary Parasitology 179 (2011) 84–91

Contents lists available at ScienceDirect

Veterinary Parasitology

journa l homepage: www.e lsev ier .com/ locate /vetpar

n agent-based model for control strategies of Echinococcus granulosus

iang Huanga,b,∗, Yan Huanga,b, Qian Wangb, Ning Xiaob, Deyou Yib, Wenjie Yub,ongchuan Qiub

West China School of Public Health Sichuan University, No. 17, Section 3, Renmin South Road, Chengdu, PC 610044, Sichuan, ChinaSichuan Center for Disease Control and Prevention, No. 6, Zhongxue Road, Chengdu, PC 610041, Sichuan, China

r t i c l e i n f o

rticle history:eceived 8 August 2010eceived in revised form 17 January 2011ccepted 27 January 2011

eywords:chinococcus granulosusgent-based modelontrol optionsontrol strategies

a b s t r a c t

Cystic echinococcosis is a widespread zoonosis, caused by Echinococcus granulosus. Thedefinitive hosts are carnivores and the intermediate hosts are grazing animals. Becausehumans are often accidentally infected with the cystic stage of the parasite, a controlprogram is being developed for Western China. Western Sichuan Province in China is ahighly endemic area. In this study, we built an agent-based model (ABM) to simulate andassess possible control strategies. These included dog dosing, control of livestock slaugh-ter, health education, vaccination of intermediate hosts, vaccination of definitive hosts,slow-released praziquantel injections for dogs, removing unproductive old livestock, dogpopulation reduction. These strategies were examined singly and in various combinations.The results show that vaccination based control strategies and also combined control strate-

gies (dog dosing, slaughter control, removing old livestock, dog population reduction) canachieve a higher efficiency and be more feasible. Although monthly dog dosing achievedthe highest efficiency, it required a high frequency and reliability, which were not feasibleor sustainable. The model also indicated that transmission would recover soon after thechosen control strategy was stopped, indicating the need to move from a successful attack

ble conCrown

phase to a sustaina

. Introduction

Cystic echinococcosis (CE) is a widespread zoonosis.he causative agent is the larval stage of Echinococcusranulosus (E. granulosus) growing usually in the liver andungs of intermediate hosts (grazing animals and humans).he definitive hosts (carnivores, including dogs) harborapeworms in their small intestine. Eggs are passed in fae-

es and infect the intermediate hosts orally (Thompsonnd McManus, 2001). Western Sichuan Province in Chinaocated in the southeast of Qinghai-Tibet Plateau waseported as a highly endemic area (Schantz et al., 2003).

∗ Corresponding author at: Sichuan Center for Disease Control and Pre-ention, No. 6, Zhongxue Road, Chengdu, PC 610041, Sichuan, China.el.: +86 13708205110; fax: +86 2885589532.

E-mail address: liangdeep@gmail.com (L. Huang).

304-4017/$ – see front matter. Crown Copyright © 2011 Published by Elsevier Boi:10.1016/j.vetpar.2011.01.047

solidation phase. Copyright © 2011 Published by Elsevier B.V. All rights reserved.

Yaks and sheep/goats serve as the intermediate hosts anddogs serve as the major definitive hosts. The infection rateswere 12.8% in humans, 21% in dogs and 54% in sheep/goats(Wang et al., 2008; Yang et al., 2009).

In 2006 China launched the National EchinococcosisControl program. The control options were based on themonthly treatment of dogs with an anthelmintic (“dog dos-ing” with praziquantel) and slaughter control (MOH, 2006).The monthly dog dosing is an effective way to controltransmission but the problem is the dog dosing coverage(Craig et al., 2007), which is made difficult by distance,the harsh environment, and the lack of commitment bythe dog-owner. The “slaughter control” is also a key to this

program. It depends on reducing feeding offal of interme-diate hosts to the dogs. There are Government-controlledslaughter houses, but many unsupervised slaughters occur,and also there are many animals that die in the field fromdisease, cold or starvation. Lack of commitment by the

.V. All rights reserved.

y Parasi

L. Huang et al. / Veterinar

dog-owner exacerbates this problem. Besides the two con-trol options, the recombinant vaccine of intermediate hostswere used in some countries, it is effective in protect-ing against natural infections, according to a field trial inXinjiang Autonomous Region (Heath et al., 2003). In thispaper, we built an agent-based model (ABM) to simulateand assess the control strategies in order to find out themost feasible control strategies in this region.

The ABM, also known as the individual-based model(IBM), is a computational model. It simulates the actionsand interactions of autonomous agents with a view toassessing their effects on the system as a whole (Wikipedia,2010a). It has been widely used in ecology and other fields(Grimm et al., 2006; Parry et al., 2006)

2. Materials and methods

2.1. The characteristics of the agents

There is recent evidence that perhaps some immunitycan develop in definitive hosts and influence transmission(Torgerson, 2006). Within the intermediate host, E. gran-ulosus uses two mechanisms to subvert immune responseof hosts: developing into a cyst and immunomodulation(Siracusano et al., 2008). Therefore, the model ignored theimmunity of the intermediate hosts. Roberts et al. (1986)defined two parameters to characterize the immunity ofthe definitive hosts, the probability of immunity on expo-sure (˛) and the probability of loss of immunity (�). Themodel had simulated definitive hosts with presence ofimmunity by using these two parameters.

Eggs are highly resistant to environmental factors.Thevenet et al. (2005) reported that the survival time ofeggs of E. granulosus in canine feces in an arid climate wasup to 41 months. The WHO/OIE reported that the eggsremained infective for many months or up to about oneyear in a moist environment at lower ranges of tempera-tures (about +4 ◦C to 15 ◦C), as well as long time survival infreezing status (Thompson and McManus, 2001).

Usually, the slaughters of grazing animals in local com-munities are seasonal, and therefore most of yaks andsheep/goats are killed in winter and spring. Relatively, yaksand sheep/goats in Western Sichuan have a longer lifetimeand lower slaughtering rate than other places in China.According to the field surveys, the slaughtering rates ofyaks and sheep/goats are about 10% and 18% per annumrespectively, and local people tend to kill older livestock,rather than send them to a Government slaughterhouse.These old livestock are highly infectious since they mayhave more fertile hydatid cysts (Torgerson, 2003).

The dogs are infected by the offal of intermediate hostscontaining fertile cysts. The fertile cysts contain many pro-toscoleces, which are infective for the dogs. Therefore, theage-related cyst fertility rate affects the transmission. Thisis an important consideration in this model, and some con-flicting data were reported. The estimated data are listed in

Table 1. The cyst fertility rates in infected sheep/goats andyaks were estimated from existing publications (Duegerand Gilman, 2001; Torgerson, 2003; Torgerson et al., 2009;Yang et al., 2009), and those in yak were obtained fromparasitologists in Sichuan Province. There are almost linear

tology 179 (2011) 84–91 85

positive correlations between the ages of the intermedi-ate hosts and fertility rates. The biological and sociologicalcharacteristics of the agents influencing the transmissionof E. granulosus are stated in this model. The model mayneed to be modified if used for other regions.

2.2. Model construction

The ABM model contains three components: agent,object and the environment (Devillers et al., 2008).The agents have four types in this model: the dogs,yaks/sheep/goats, parasites and the egg contaminations.They are autonomous and characterized by their rules. Theobjects in this model are the number of egg contaminationsand number of infectious intermediate hosts deaths. Theyare not autonomous and calculated by the agents’ statesand populations, and the infection risks of hosts are thefunctions of them. The model has defined a pasturing com-munity in Western Sichuan Province as the Environment.

The ABM is not reliable if it has a complex structure.For simplicity, the model did not simulate a single parasitein a dog intestine; the parasite agent represented a groupof parasites, which a dog acquired after a single infection.Similarly, the egg contamination represented the contam-ination point, which was formed by infectious eggs in theexcretions of infected dogs, and it was considered as theinfection risk for human and livestock.

The model was constructed for simulating the behav-iors and properties of the agents. For example, if the dogaged, the age would be added one time step; if the dog wasinfected, the infection state would be changed to “infected”.

2.3. The basic assumptions of the model

In this model, the population of hosts were constantover time, The mortality and natality were adjusted by thedifferences of the population quantities and the carryingcapacities of the environment. The carrying capacities ofthe study area were estimated from the local county annual(The committee of compiling Shiqu County annals, 2000).

ABM is a discrete time model. The time step model wasdefined as one week, and the simulation started from thefirst week and ended in the 1300th week. The time spanof this model was about 25 years, since 25 year was longenough to assess whether the control strategy is successful(Torgerson, 2003).

The target area of this model situated in the pasturingregion of western Sichuan Province, China, a part of theQinghai-Tibet Plateau, a highly endemic area. It has an aver-age elevation of over 4500 m, year-round temperaturesaverage −4 ◦C, dipping to −40 ◦C in winter (Wikipedia,2010b). In this area, this model defines a pasturing com-munity with grazing grassland of 60 km2, 105 dogs, 35households, 525 sheep/goats and 1650 yaks. The datawere estimated from the Shiqu annals (The committee ofcompiling Shiqu County annals, 2000).

2.4. The initial data and parameters

The data of the infection rates of hosts were estimatedby the previous field surveys and the papers (Wang et al.,

86 L. Huang et al. / Veterinary Parasitology 179 (2011) 84–91

Table 1Age-related cyst fertility rates in infected sheep/goats and yaks.

1 12

1 10.1 0

2coos

ipies

Tbodhsttilwpuwsklii

mM(iom

2

uTfi(latfst

Age (years) 1 2 3 4 5 6 7 8 9 10 1

Sheep/goats 0.1 0.1 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1Yaks 0 0 0 0 0 0 0.01 0.03 0.05 0.07

008; Yang et al., 2009). To make this model available,ross-sectional data must be obtained. The infection ratef sheep/goats started from about 54%, the infection ratef yaks started from about 50%, the infection rate of dogstarted from about 21%.

The values of the parameters used in the ABM are listedn Table 2. Torgerson and Heath (2003) had estimated therobability of immunity on exposure (˛) and the probabil-

ty of loss of immunity (�) from the maximum likelihoodstimation (MLE). This ABM used these parameters in theimulations of dogs.

The infection risks for hosts are important parameters.he infection risk for dogs (Rd) was estimated by the num-er of the infective intermediate hosts deaths. Assumingne infective intermediate host was able to infect threeogs, as each householder had on average three dogs, andouseholders feed the offal to all of them after a singlelaughter, the algorithm is detailed in Table 2. The infec-ion risk of yak and sheep/goat (Ri) was the function ofhe number of egg contaminations, grassland area of thentermediate hosts weekly contacted, total grazing grass-and area and the probability of being infected after contact

ith eggs, and the algorithm is detailed in Table 2. Therobabilities of being infected after contact with eggs werenknown parameters. The parameters were varied over thehole range of possible values, and suitable values were

elected by using the selection criteria. The criteria were:eeping the infection rates of hosts at a relatively stableevel within 25 years and keeping a reasonable age-relatednfection rate. The criteria enabled selecting an appropriatenfection pressure for intermediate hosts.

The life expectancy of the egg contamination was esti-ated by the information of WHO/OIE (Thompson andcManus, 2001) and from a study on the egg infectivity

Thevenet et al., 2005) .The environment of the target areas suitable for egg survival, therefore, the model supposedne year for the average egg survival, with an S.D. of twoonths.

.5. Simulations

The ABM is a computer program, which was written bysing R language and environment (R Development Coreeam, 2009). Although there are many good tools designedor ABM, such as the Repast, however, the R languages more flexible and easy for data management. Petzoldt2003) built an ABM prototype of the Daphnia to simu-ate the population dynamics by using R. This model has

similar structure to it. The model was designed as a loop

o simulate 100 trials, and recorded the data of each trialor later analysis. It was very easy to simulate the controltrategies through changing the agent states or popula-ions.

13 14 15 16 17 18 19 20

.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.50

The model simulated the optional control strategies,included dog dosing, slaughter control, vaccination andhealth education. The dog dosing is a traditional controloption and one of the most important control options inChina. The model simulated the dog dosing with differ-ent reliabilities, the reliabilities of dog dosing being set as50%, 70% or 100% in this model. In this model the reliabili-ties was the coverage, for example, 70% reliability indicated70% of dogs were treated with praziquantel, and each doghad the equal probability of being treated. Besides that,the model simulated the dog dosing with different fre-quencies. The dog dosing was simulated through deletingthe parasites within the dosed dogs and changing the doginfection status to “uninfected”. The slaughter control wassimulated by reducing the offal feeding. Consequently, itreduced the infection pressure for dogs. We supposed 70%reliability in slaughter control, indicated 70% offal feedingwas reduced. Old livestock are highly infectious to dogs,removing them was considered to be effective. It was sim-ulated by killing the sheep/goats older than 4 years and theyaks older than 10 years. The population control of dogswas simulated by reducing the population by 50% within 10years. It was gradually reducing but not suddenly dropping.The vaccinations for hosts were simulated by preventingnew infection but not eliminating the existed parasites.The health education persuaded the people to support thecontrol program of the combination of the dog dosing,slaughter control and dog population reduction. The slow-released praziquantel (SRP) was simulated by protectingdogs from infections.

3. Results

The outcomes of the model are the numbers of theagents and their states. The infection rates were calculatedfrom the outcomes. The averages were calculated from 100trials and illustrated the variations of the infection rates ofhosts and the number of egg contaminations. The resultsare displayed on Fig. 1. It is noteworthy that the infectionrates of hosts and the number of egg contaminations froma single trial varied over a wide range, the averages variedless. It indicates that the stochastic of the model is consid-erable. From this point of view, the average infection rateis more meaningful, since it is able to find out the varia-tion trend through the many trials. However, the stochasticis in need of attention. The simulation results of controlstrategies are summarised in Table 3.

3.1. The effect of dog dosing on the infection rates of hosts

To explore the effects of the dosing frequencies andreliabilities on the infection rates of hosts, the model simu-lated the transmission with different dosing frequenciesand reliabilities. The monthly dog dosing with 50% reli-

L. Huang et al. / Veterinary Parasitology 179 (2011) 84–91 87

Table 2The parameters.

Parameters Values and algorithms

DogsInfection risk (Rd) Rd = IL

D

I: number of infective intermediate hosts deaths; D: dog total; L: average dognumber of each household kept (3)

Probability of immunity on exposure(˛)

0.3 (Torgerson, 2003)

Probability of loss of immunity perannum(�)

0.02 (Torgerson, 2003)

Yak, sheep/goatsInfection risk (Ri)b � = A

G , X∼B(n, �)(binomial distribution), p = P(X ≥ 1)a

Ri = pc; n: number of egg contaminations. A: estimated grassland area of theintermediate hosts weekly contacted (212 M2 for sheep/goats, 636 M2 foryaks) (Xu and Zhao (2005)). G: Grazing grassland area (60 km2). p: probabilityof contacting with eggs per week. c: probability of being infected after contactwith eggs for yaks (0.05)c and sheep/goats (0.28)c

ParasitesLife expectancy 10 months (Budke et al. (2005)), S.D.:2 monthsd

Prepatent period 6 weeks (Conchedda et al., 2006)Egg contaminationsLife expectancy Mean: 12 months (Thompson and McManus (2001)), S.D.: 2 monthsd

a This formula calculates the probability of that at least one egg contamination exists in the yak or sheep/goats contacted grassland within one week.b Ri includes infection risk of yaks and sheep/goats. They have same algorithm.c The estimation method was described in Section 2.4.d The S.Ds. are estimated values.

Fig. 1. (a) No intervention. (b) Monthly dog dosing with 50% reliability. (c) Slaughter control with 70% reliability. (d) Removing old livestock & 6-monthlydog dosing with 75% reliability. (e) Removing old livestock & vaccination of intermediate hosts with a coverage of 90% & 6-monthly dog dosing with 75%reliability. (f) Vaccination of intermediate hosts with a coverage of 90% & 6-monthly dog dosing with 75% reliability. (g) Vaccination of dogs with a coverageof 70% & 6-monthly dog dosing with 75% reliability. (h) Vaccination of intermediate hosts with a coverage of 90% & vaccination of dogs with a coverageof 70%. (i) Health education (dog population reduced by 50% & 3-monthly dog dosing at 70% reliability & slaughter control). (j) Slow released praziquantelwith 70% reliability. Discontinuous line: infection rates of yaks. Solid line: infection rates of dogs. Gray line: infection rates of sheep/goats. X-axis: infectionrates. Y-axis: time (weeks).

88 L. Huang et al. / Veterinary Parasitology 179 (2011) 84–91

Table 3The simulation results of the control strategies.

Control strategies Years to infection rate of hosts<1% OR the values of infectionrates at the end of 25 yearsa

Reduction in eggcontamination

Monthly dog dosing with 50% reliability (Fig. 1b) S: 2%; Y: 2%; D: <1% >95%Monthly dog dosing with 70% reliability 20th year >95%3-monthly dog dosing with 100% reliability S: 2%; Y: 2%; D: <1% >95%6-monthly dog dosing with 100% reliability S: 16%; Y: 14%; D: 4% 83%6-monthly dog dosing with 75% reliability S: 28%; Y: 25%; D: 7% 61%Slaughter control with 70% reliability (Fig. 1c) S: 16%; Y: 16%; D: 2% 86%Slaughter control with 70% reliability & 3-monthly dog dosingwith 70% reliability

20th year >95%

Removing old livestock & 6-monthly dog dosing with 75%reliability (Fig. 1d)

S: 11%; Y: 13%; D: 4% 78%

Removing old livestock & vaccination of intermediate hostswith a coverage of 90% & 6-monthly dog dosing with 75%reliability (Fig. 1e)

10th year >95%

Vaccination of intermediate hosts with a coverage of 90% &6-monthly dog dosing with 75% reliability (Fig. 1f)

20th year >95%

Vaccination of dogs with a coverage of 70% & 6-monthly dogdosing with 75% reliability(Fig. 1g)

S: 6%; Y: 6%; D: 1% 93%

Vaccination of intermediate hosts with a coverage of 90% &vaccination of dogs with a coverage of 70% (Fig. 1h)

20th year >95%

Health education (dog population reduced by 50% &3-monthlydog dosing at 70% reliability & slaughter control) (Fig. 1i)

15th year >95%

Slow released praziquantel with 70% reliability (Fig. 1j) S: 21%; Y: 19%; D: 4% 76%

e.

aodTicsrlte

3s

ahd7whd

3

i4bywr

Slow released praziquantel with 70% reliability & Vaccinationof intermediate hosts with a coverage of 90%

a S: sheep/goat infection rate; Y: yak infection rate; D: dog infection rat

bility showed a moderate efficiency. The infection ratesf the intermediate hosts decreased to about 2%, and theog infection rate decreased to <1% at the end (Fig. 1b).he monthly dog dosing with 70% reliability decreased thenfection rates of the hosts to <1% within 20 years. The effi-iency of 3-monthly dog dosing with 100% reliability wasimilar to the efficiency of monthly dog dosing with 50%eliability. Six-monthly dog dosing with 100% reliabilityowered the dog infection rate to about 4%, and the infec-ion rates of the intermediate hosts had achieved a newquilibrium of approximately 14–16%.

.2. The effect of the slaughter control based controltrategies on the infection rates of hosts

The outcome of slaughter control only with 70% reli-bility showed that the infection rates of intermediateosts decreased to about 16%, the dog infection rate hadecreased to about 2% (Fig. 1c). The slaughter control with0% reliability combined with the 3-monthly dog dosingith 70% reliability decreased the infection rates of theosts to <1% within 20 years, which is similar with monthlyosing with 70% reliability.

.3. Removing old livestock

Removing old livestock decreased infection rates ofntermediate hosts to 11–13% and infection rate of dogs to

% if combined with 6-monthly dog dosing with 75% relia-ility (Fig. 1d). It decreased infection rates to <1% within 10ears if combined with vaccination of intermediate hostsith a coverage of 90% and 6-monthly dog dosing with 75%

eliability (Fig. 1e). At the beginning of the simulations, the

18th year >95%

dog infection rate suddenly increased to about 30%, sincemany old livestock were killed at the start of the controlstrategy, more dogs were infected.

3.4. The effect of vaccination-based control strategies onthe infection rates of hosts

The vaccination of intermediate hosts was assessed inthis model. We had simulated a vaccination with a coverageof 90%, at the same time dosing the 6-monthly dogs with75% reliability. The results show that the infection rates ofall hosts decreased to <1% within 20 years (Fig. 1f).

A vaccine for dogs is being researched (Zhang andMcManus, 2008). This model simulated this control optionwith the coverage of 70%. The coverage was lower thanvaccination of intermediate hosts, since the vaccination fordogs is more difficult. The 6-monthly dog dosing with 75%reliability was included while vaccinating dogs. Accord-ing to the results, the infection rate of intermediate hostsdecreased to about 6%, and the dog infection rate decreasedto about 1% (Fig. 1g) at the end of the simulation.

The combination of those two kinds of vaccination wasassessed in the model. The infection rates of both types ofhosts decreased to <1% within 20 years (Fig. 1h).

3.5. The effect of health education based control optionson the infection rates of hosts

The health education modelled to change the people’sbehavior to achieve a more comprehensive control strat-egy. The local people are educated to make sure that theydo not allow dogs to eat uncooked offal from intermediatehosts. We assumed that people were successful in reducing

y Parasi

L. Huang et al. / Veterinar

feeding offal to dogs by 70%, and also simultaneously edu-cated the people to reduce dog population by 50%, and tocarry out 3-monthly dog dosing with 70% reliability. Thiscontrol strategy was effective since it had decreased hostinfection rates to <1% within 15 years (Fig. 1i).

3.6. Slow-released praziquantel (SRP) based controlstrategies

The SRP is being researched (Wei et al., 2005; Chenget al., 2009). It is subcutaneously implanted in dogs. Themodel had simulated this control option. The reliabilitywas 70%, which is a low reliability, since SRP may be diffi-cult for implantation in unwanted dogs. At the end of thesimulation, the infection rates of the intermediate hostswere about 19–21%, the infection rate of dogs was about4% (Fig. 1j).

The model also simulated the combination of SRP andvaccination of intermediate hosts. The infection rates allachieved <1% within 18 years.

3.7. The effect of control strategies on the number of eggcontaminations

The degree of egg contamination is an important vari-able as it indicates the risk for human and intermediatehosts. Many control strategies effectively reduced the riskfor human and intermediate hosts by reducing the eggcontaminations. The strategies, which had successfullydecreased host infection rates to <1% within 25 years, hadreduced the egg contaminations by >95%. In addition, othercontrol strategies had reduced them by 61–93%.

4. Discussions

4.1. Control strategies

The population of this parasite has a low stability(Gemmell et al., 1986), and therefore the parasites shouldbe sensitive to the control options. Information is avail-able from the target area, where monthly dog-dosing hasbeen carried out for 4 years. The dog infection rate hadnot decreased to zero or even close to zero in the tar-get area after four years of monthly dog dosing (accordingto the control program in this area). The dog dosing mayhave a very low reliability and may be lower than 50%.Since the monthly dog dosing with 50% reliability hadachieved a moderate efficiency in simulations, and the real-istic control efficiency in Western Sichuan Province wasworse, therefore, we can suppose that the reliability waslower than 50%. In conclusion, the reliability of dog dosinginfluenced the control efficiency, and this is an impor-tant factor in a control program. Money and time can bewasted if procedures are not carried out correctly. How-ever, reports from the fields suggest that the high efficiency

of dog dosing is almost impossible to achieve. Further addi-tions to the control by dog-dosing are needed to achievea cost-effective outcome. For instance, health educationand slaughter control; vaccination of intermediate hosts;removing old livestock; and other procedures.

tology 179 (2011) 84–91 89

This model focused on the frequency and reliability ofthe dog dosing. According to the simulation outcomes, thesingle procedure of dog dosing is not recommended by thismodel. Frequency and reliability in dog dosing are equallyimportant, however they are both not likely to be ensured.Firstly, although the outcomes of the monthly dosing with50% reliability and the 3-monthly dog dosing with 100%reliability are similar, the 6-monthly dog dosing with 100%reliability is worse, it indicates that the importance of thefrequency and the reliability in dog dosing are equal. Sec-ondly, monthly dog dosing is a difficult task in this region,as the people who are in charge of the dog dosing admin-istrations find it difficult to access the communities. It isimpossible to treat all the dogs in the current situation oftarget area, and the unwanted dogs decrease the reliabilitysince they are not likely to be treated. Therefore, the dosingfrequencies and reliabilities are both difficult to ensure.

The slaughter control shows a lower efficiency if it isthe only control option in a control program. Although itreduced 70% feeding offal to the dog, the infection rate ofintermediate hosts decreased to about 16%, the infectionrate of the dog decreased to about 2%. According to thesimulation results the 70% reliability is not enough to con-trol the E. granulosus and the single slaughter control wasnot able to control the transmission. However, this controlstrategy had achieved the same result as the monthly dogdosing with 70% reliability when combined with 3-monthlydog dosing with 70% reliability. Both successfully decreasedthe infection rates of hosts within 20 years.

Removing old livestock seems to be very effective, espe-cially if it is combined with vaccination of intermediatehosts and 6-monthly dosing of dogs. The problem of remov-ing old livestock is that the local people are accustomedto keep their livestock until they are very old, allowingthem to die a natural death. Persuading people to kill oldlivestock would seem to be a major challenge. The vaccina-tion of intermediate hosts and 6-monthly dog dosing couldachieve a satisfactory outcome only if carried on until thelast old infectious animals die.

Health education is effective as it is more comprehen-sive, it contains three aspects: dog population reduction,slaughter control and dog dosing. It had lowered thehost infection rates and egg contaminations significantly.Reducing the dog population played a very important role.However, dogs are considered as the indispensable tools forlife and grazing by local people, and the religion is againstkilling dogs. This makes the dog population reduction a bigchallenge.

The vaccinations are important control options, and thevaccinations of intermediate hosts are effective. Accordingto the simulation results, the vaccination of intermedi-ate host is hopeful, since it shortens the control process.However, it required a high coverage and the assistanceof other options to achieve higher control efficiency. Thehigher coverage of vaccination of intermediate hosts is pos-sible, since in the pasturing community, the governors have

established a complete veterinary system and policies foryak and sheep/goats, and the local people pay more atten-tion to the health of animals, which are considered as asymbol of wealth. The government may make the vaccina-tions more accessible for local people. The vaccine for dog

90 L. Huang et al. / Veterinary Parasi

Fig. 2. Infection rates of hosts after control strategy stops in 25th year.Discontinuous line: infection rates of yaks. Solid line: infection rates ofda

iahmvvshoe

4c

t<bcdcibltoobiSmoo

twctdcrsasow

Dueger, E.L., Gilman, R.H., 2001. Prevalence, intensity, and fertility of ovinecystic echinococcosis in the central Peruvian Andes. Trans. R. Soc. Trop.

ogs. Gray line: infection rates of sheep/goats. X-axis: infection rates. Y-xis: time (weeks).

s under development. Compared to vaccine of intermedi-te hosts, the vaccination reduced dogs’ infections more,owever it had less effect on the infection of the inter-ediate hosts. If the dog dosing is very difficult for local

eterinarians, the simulation outcome indicates that theaccination for dogs may be an alternative control option,ince both the combination of vaccination of intermediateosts and twice dog dosing per annum and the combinationf vaccination of intermediate hosts and dogs have similarfficiencies.

.2. The control strategies summary and the risk ofontrol strategies suspending

We have assessed 15 control strategies in the simula-ions. Seven of them had decreased the infection rates to1% within 20 years and decreased egg contaminationsy >95% at the end, and they are considered to be suc-essful. However, some of them are easy for simulation,ifficult for operation. Therefore, the decision-making ofontrol strategies has to be considered with local situationsn mind. In the study area, monthly dosing and high relia-ilities of dosing are not likely to be achieved. Removing old

ivestock and dog population reduction are against the cus-om or religion of local people. In conclusion, vaccinationf intermediate hosts & 6-monthly dog dosing, vaccinationf intermediate hosts & vaccination of dogs are likely toe most feasible. Slaughter control & 3-monthly dog dos-

ng is hopeful if the unsupervised slaughters were reduced.RP shows a good efficiency too, however, the implantationay be difficult to carry out, since there are a large number

f fierce guard dogs in this region, as well as a high numberf unwanted dogs.

Fig. 2 shows that if the infection rates of hosts after con-rol strategy stops in 25th year, the infection rates of hostsill recover to a high level within several years. It indi-

ates that the system of parasite transmission is inclinedo be in a stable state. A single unexpected infection ofog may lead the recovery of the transmission though theoupling of the transmission cycle. The WHO Guidelinesecommends that the attack phase should move to a con-olidation phase where the low levels of egg contaminationchieved are maintained by the most cost-effective and fea-

ible continuing control procedures. This combination ofngoing procedures cannot be predicted at this stage, butill become apparent as a control program proceeds.

tology 179 (2011) 84–91

4.3. Comparison with an existing model

Torgerson had simulated the control strategies withthe differential equations in population level (Torgerson,2003). To compare the outcomes of these two models, wehad simulated the control strategies, which were simulatedin Torgerson’s model. The outcomes of these two modelsare comparable but the results are not exactly the same. Thepresets were not exactly the same, and the rationales arequite different. The 6-monthly dog dosing, in Torgerson’ssimulations, the infection rate of sheep/goats had achieveda equilibrium of about 20%, while in this model the infec-tion rate of sheep/goats had achieved a equilibrium of about28%. The control strategy of combination of vaccination ofintermediate hosts and 6-monthly dog dosing, in Torger-son’s model, the infection rate of sheep/goats decreased tozero within 15 years. In our model it decreased to about<1%, which was considered as a very low level, within 20years.

4.4. The merits of the ABM and future developments

The ABM consists of the individual agents, however itfocuses on the population dynamics. It has demonstratedthe population dynamics of parasites transmission emerg-ing from the agents, which have only several simple rulesand properties. Besides that, the ABM is flexible for themodifications, even if a new type of agent was foundinvolved in the cycle, and it has no need for reconstruction.Additionally, because it was constructed by simulating thenatural entity, the rationale is relatively simple for under-standing.

The parameters used in this model were estimationsor were taken from published papers. Some of them maynot be reliable. The more reliable parameters should be apriority in the future development.

Acknowledgment

We would like to thank Dr. David Heath for his help.

References

Budke, C.M., Jiamin, Q., Craig, P.S., Torgerson, P.R., 2005. Modeling thetransmission of Echinococcus granulosus and Echinococcus multilocu-laris in dogs for a high endemic region of the Tibetan plateau. Int. J.Parasitol. 35, 163–170.

Cheng, L., Guo, S., Wu, W., 2009. Characterization and in vitro release ofpraziquantel from poly(epsilon-caprolactone) implants. Int. J. Pharm.377, 112–119.

Conchedda, M., Gabriele, F., Bortoletti, G., 2006. Development and sexualmaturation of Echinococcus granulosus adult worms in the alternativedefinitive host Mongolian gerbil (Meriones unguiculatus). Acta Trop.97, 119–125.

Craig, P.S., McManus, D.P., Lightowlers, M.W., Chabalgoity, J.A., Garcia,H.H., Gavidia, C.M., Gilman, R.H., Gonzalez, A.E., Lorca, M., Naquira,C., Nieto, A., Schantz, P.M., 2007. Prevention and control of cysticechinococcosis. Lancet Infect. Dis. 7, 385–394.

Devillers, H., Lobry, J.R., Menu, F., 2008. An agent-based model for pre-dicting the prevalence of Trypanosoma cruzi I and II in their host andvector populations. J. Theor. Biol. 255, 307–315.

Med. Hyg. 95, 379–383.Gemmell, M.A., Lawson, J.R., Roberts, M.G., 1986. Control of echinococ-

cosis/hydatidosis: present status of worldwide progress. Bull. WorldHealth Organ. 64, 333–339.

y Parasi

ulosus infection and options for control of cystic echinococcosis inTibetan communities of Western Sichuan Province. China. PloS. Negl.

L. Huang et al. / Veterinar

Grimm, V., Berger, U., Bastiansen, F., Eliassen, S., Ginot, V., Giske, J., Goss-Custard, J., Grand, T., Heinz, S.K., Huse, G., Huth, A., Jepsen, J.U.,Jørgensen, C., Mooij, W.M., Mueller, B., Peer, G., Piou, C., Railsback,S.F., Robbins, A.M., Robbins, Martha, M., 2006. A standard protocol fordescribing individual-based and agent-based models. Ecolmodel 198,115–126.

Heath, D.D., Jensen, O., Lightowlers, M.W., 2003. Progress in control ofhydatidosis using vaccination – a review of formulation and deliv-ery of the vaccine and recommendations for practical use in controlprogrammes. Acta Trop. 85, 133–143.

MOH, 2006. A letter about seeking for opinions in echinococcosiscontrol technical proposal. Last retrieved November 18, 2010 at:http://www.moh.gov.cn/publicfiles/business/htmlfiles/mohjbyfkzj/pgzdt/200804/19138.htm, in Chinese.

Parry, H.R., Evans, A.J., Morgan, D., 2006. Aphid population response toagricultural landscape change: a spatially explicit, indlividual-basedmodel. Ecolmodel 199, 451–463.

Petzoldt, T., 2003. R as a simulation platform in ecological modelling. RNews 3, 8–16.

R Development Core Team, 2009. R: a language and environment for sta-tistical computing. R Foundation for Statistical Computing, Vienna,Austria. ISBN 3-900051-07-0, URL http://www.R-project.org.

Roberts, M.G., Lawson, J.R., Gemmell, M.A., 1986. Population dynamicsin echinococcosis and cysticercosis: mathematical model of the life-cycle of Echinococcus granulosus. Parasitology 92 (Pt 3), 621–641.

Schantz, P.M., Wang, H., Qiu, J., Liu, F.J., Saito, E., Emshoff, A., Ito, A.,Roberts, J.M., Delker, C., 2003. Echinococcosis on the Tibetan Plateau:prevalence and risk factors for cystic and alveolar echinococcosisin Tibetan populations in Qinghai Province, China. Parasitology 127(Suppl), S109–120.

Siracusano, A., Rigano, R., Ortona, E., Profumo, E., Margutti, P., Buttari, B.,Delunardo, F., Teggi, A., 2008. Immunomodulatory mechanisms dur-ing Echinococcus granulosus infection. Exp. Parasitol. 119, 483–489.

Thevenet, P.S., Jensen, O., Drut, R., Cerrone, G.E., Grenovero, M.S., Alvarez,

H.M., Targovnik, H.M., Basualdo, J.A., 2005. Viability and infectiousnessof eggs of Echinococcus granulosus aged under natural conditions ofinferior arid climate. Vet. Parasitol. 133, 71–77.

The committee of compiling Shiqu County annals, 2000. The ShiquAnnals. Sichuan People’s Press, Chengdu, Sichuan Province, China, pp.112–160, in Chinese.

tology 179 (2011) 84–91 91

Thompson, R.C.A., McManus, D.P., 2001. Aetiology: parasites and life-cycles. In: Eckert, J., Gemmell, M.A., Meslin, F.-X., Pawlowski, Z.S.(Eds.), WHO/OIE Manual on Echinococcosis in Humans and Animals:A Public Health Problem of Global Concern. Office International desEpizooties, Paris, pp. 1–19.

Torgerson, P.R., Heath, D.D., 2003. Transmission dynamics and con-trol options for Echinococcus granulosus. Parasitology 127 (Suppl.),S143–158.

Torgerson, P.R., 2003. The use of mathematical models to simulate controloptions for echinococcosis. Acta Trop. 85, 211–221.

Torgerson, P.R., 2006. Canid immunity to Echinococcus spp.: impact ontransmission. Parasite Immunol. 28, 295–303.

Torgerson, P.R., Ziadinov, I., Aknazarov, D., Nurgaziev, R., Deplazes,P., 2009. Modelling the age variation of larval protoscole-ces of Echinococcus granulosus in sheep. Int. J. Parasitol. 39,1031–1035.

Wang, Z., Wang, X., Liu, X., 2008. Echinococcosis in China, a review of theepidemiology of Echinococcus spp. Ecohealth 5, 115–126.

Wei, J., Cheng, F., Qun, Q., Nurbek, Xu, S.D., Sun, L.F., Han, X.K., Muhan, Han,L.L., Irixiati, Jie, P., Zhang, K.J., Islayin, Chai, J.J., 2005. Epidemiologicalevaluations of the efficacy of slow-released praziquantel-medicatedbars for dogs in the prevention and control of cystic echinococcosis inman and animals. Parasitol. Int. 54, 231–236.

Wikipedia, 2010a. Agent-based Model, Last retrieved November 18, 2010aat: http://en.wikipedia.org/wiki/Agent-based model.

Wikipedia, 2010b. Tibetan Plateau, Last retrieved November 18, 2010b at:http://en.wikipedia.org/wiki/Tibetan Plateau.

Xu, S.X., Zhao, X.Q., 2005. Calculation for economic and ecological benefitsof yak and sheep fattening in shed in Changjiang-Yellow Rivers SourceRegion–a case study from Maqin County, Qinghai Province. Chin. J.Eco-Agric. 13 (1), 195–197 (in Chinese).

Yang, Y.R., McManus, D.P., Huang, Y., Heath, D.D., 2009. Echinococcus gran-

Trop. Dis. 3, e426.Zhang, W., McManus, D.P., 2008. Vaccination of dogs against Echinococcus

granulosus: a means to control hydatid disease? Trends Parasitol. 24,419–424.