Sharing essential knowledge with health researchers and policy makers

Communities of practice

Advances And Challenges In Predicting The Impact Of Lymphatic Filariasis Elimination Programmes

21 Jan 2008

Source: WHO/TDR

Wilma A Stolk , Sake J de Vlas , J Dik F Habbema

Department of Public Health, Erasmus University Medical Center, Rotterdam, P.O. Box 1738, 3000 DR Rotterdam, The Netherlands

Working paper for the Scientific Working Group meeting on Lymphatic Filariasis Research, convened by the Special Programme for Research and Training in Tropical Diseases, Geneva, 10–12 May 2005

Full text source: Scientific Working Group, Report on Lymphatic Filariasis, 10–12 May 2005, Geneva, Swizterland, Copyright © World Health Organization on behalf of the Special Programme for Research and Training in Tropical Diseases, 2005, http://www.who.int/tdr/publications/publications/swg_lymph_fil.htm

Introduction

Lymphatic filariasis is a mosquito-borne parasitic disease and an important cause of chronic morbidity in tropical countries. In 1998, the Global Programme to Eliminate Lymphatic Filariasis (GPELF) was initiated, aiming at worldwide elimination of this parasitic disease as a public health problem [1]. The main strategy in the global programme is to interrupt transmission by annual population treatment with antifilarial drugs (diethylcarbamazine or ivermectin plus albendazole). In addition, morbidity management should reduce the suffering of patients who have chronic manifestations. Thirty-two countries had started elimination programmes in 2002 [2] and this number is still growing.

The goal of elimination is ambitious. Past mass treatment programmes had varying degrees of success. In some areas transmission was apparently interrupted.3 In other areas elimination was not achieved, in spite of long-term control programmes [4,5]. How strategic choices, and operational or biological factors, contribute to success or failure is poorly understood. It is unknown what coverage and duration of mass treatment (and possible additional measures) are required to achieve elimination and how this depends on the vector and parasite strain, endemicity level, and drugs used. Mathematical models can help to clarify these issues and application of such models is considered important for support of GPELF [6].

Mathematical models have been used widely in parasitology. They help to understand the complex transmission dynamics of parasitic diseases and are useful tools for planning and evaluating control programmes [7,8]. Models have also played an important role in lymphatic filariasis research [9,10]. Targeted models, which consider part of the processes involved in transmission, helped, for example, to clarify the role of acquired immunity[11,12] and the macrofilaricidal effects of treatment [13,14]. This paper concentrates on so-called ‘full transmission models’, which relate the rate of transmission to the intensity and distribution of infection in a human population and can be used to predict the impact of interventions on transmission and the probability of elimination.

To our knowledge, three full transmission models have been described in the literature. The first was specifically developed for the evaluation of a vector control programme and is not considered here [15]. The two other models, called EPIFIL [16,17] and LYMFASIM,18 are both being used for planning and evaluation of elimination programmes. After a brief introduction of the processes involved in transmission and control of lymphatic filariasis, we describe the basic structure of these models, compare and discuss some critical model predictions, and outline future research priorities.

Processes in lymphatic filariasis transmission and control

Models for lymphatic filariasis control basically describe the main biological processes involved in transmission (fig. 1). To study the dynamics of transmission and how intervention affects transmission, it is specifically important to take account of density dependence and heterogeneities [19,21].

Figure 1

Transmission cycle of lymphatic filariasis with density dependent mechanisms. This figure shows the life cycle of Wuchereria bancrofti, the main parasitic cause of lymphatic filariasis. The adult worms (macrofilariae) are located in the lymphatic system of the human host, where they live for 5–10 years [26,35]. After mating with male worms, female worms can produce millions of microfilariae (mf), which can be found in the bloodstream and have a lifespan of 6–24 months [13]. A mosquito that takes a blood meal may engorge some mf. Inside the mosquito, mf develop in about 12 days into L3 stage larvae (L3), which are infectious to humans. When the mosquito takes another blood meal, the L3 can enter the human body and some will migrate to the lymphatic system and develop into mature adult worms. The immature period lasts about 6–12 months [39]. Mf cannot develop into adult worms without passing through the developmental stages in the mosquito. Larval development and mosquito survival are density dependent [23,24]. Two possible mechanisms of acquired immunity are shown.11

Density dependence means that the outcome of a process depends on the abundance of the parasite stages involved. Several limitation mechanisms may reduce transmission when the average worm burden increases. For example, the proportion of microfilariae (mf) that develop into infectious L3 larvae saturates in Culex quinquefasciatus when the mf intake is higher, limiting the transmission of infection [22,23]. Further, the survival probability of mosquitoes is reported to reduce with their infection load [24]. Acquired immunity may limit infection intensity in the human host. Different mechanisms for this have been proposed [28], but evidence for the operation of such immunity is inconclusive [11,25]. These limitation mechanisms all negatively affect the impact of interventions, because transmission becomes relatively more efficient when infection levels are lower. Density dependence, however, may also occur in the opposite direction (called facilitation). The probability that a female worm mates with a male worm increases with higher worm burdens. Further, in some anopheline mosquito species, larval development might increase with higher mf intake [22]. It is unknown whether density dependence, either limitation or facilitation, occurs in parasite establishment and survival in humans, their fertility, and mf survival.

The term heterogeneity points at variation between individuals. Individuals differ for example in genetic background, nutritional status and behaviour, which may cause differences in exposure to mosquitoes, susceptibility to infection, and survival, maturation and fecundity of parasites. Therefore individuals may be predisposed to heavy or light infection, leading to an aggregated or overdispersed distribution of parasites (with a few hosts harbouring the majority of the parasites). Individuals also differ in compliance and responsiveness to treatment, which may also contribute to aggregation of parasites [13,14]. This aggregation enhances transmission because it increases the probability that female and male worms mate. Heterogeneity may also occur in the parasite population, e.g. with respect to the lifespan and resistance to treatment.

Available models

The two available models for lymphatic filariasis transmission and control, EPIFIL and LYMFASIM, mainly differ in the amount of detail included. Specific variants of both models have been developed for Wuchereria bancrofti transmitted by Culex quinquefasciatus, using data from an integrated vector management control programme carried out in Pondicherry, India, 1981–1985 [17,26]. These ‘Pondicherry model variants’ are described below. Table 1 gives the quantification of several key biological parameters of the models. Figure 2 illustrates the good fit of both models to the precontrol (1981) data from Pondicherry.

Figure 2

Comparison of model predictions with microfilaraemia prevalence by age observed before the start of vector control in Pondicherry, India, in 1981. (A) LYMFASIM predictions for models with anti-L3 immunity (solid line), anti-fecundity immunity (dashed line), and a model variant without immunity (dot-dashed line); the latter model does not fit the data and was therefore rejected [26]. (B) EPIFIL predictions of a model with acquired immunity [17]. Symbols in both graphs indicate the observed prevalence levels with corresponding confidence intervals.

Table 1

Quantification of several key biological parameters in the EPIFIL and LYMFASIM model variants for Pondicherry, where Wuchereria bancrofti is transmitted by Culex quinquefasciatus.

	EPIFIL	LYMFASIM ^a
		Anti-L3 immunity		Anti-fecundity immunity
Parasite lifecycle
Average adult worm lifespan in years (type of distribution)	8 ^b	10.2 ^c		11.8 ^c
Average mf lifespan in months (type of distribution)	10 ^b		10 ^b
Premature period in months	-		8
Exposure variation by age
Exposure at age zero as fraction of maximum exposure	0	0.26		0.40
Age in years at which maximum exposure is achieved	9	19.1		21.3
Density dependence in mosquitoes
Maximum number of L3 larvae that can develop in mosquitoes at high mf intensities	6 ^d		6.6 ^e
Acquired immunity
Duration of acquired immunity in years	lifelong	9.6 ^f		11.2 ^f
Other parameters
Monthly biting rate	5760		2200
Proportion of L3 larvae in mosquitoes that enter the human host when a mosquito bites	0.414x0.32 = 0.13		0.1
Proportion of inoculated L3 larvae that develop successfully into adult worms (x103)	0.113	1.03 ^g		0.42
Mf production per worm	2	0.61 ^h		4.03 ^h ⁱ

- Not considered in the model

a When only one number is given, this is the same for both model variants.

b Assuming a negative exponential distribution.

c Assuming a Weibull distribution with shape parameter α=2.

d Exponential saturating function with initial increase when mf intake increases from zero = 0.047

e Hyperbolic saturating function with initial increase when mf intake increases from zero = 0.09

f This parameter defines the period in which the strength of the immune response is halved in the absence of boosting

g in the absence of anti-L3 immunity

h in the presence of at least one male worm, scaled to the number of mf per 20 µl peripheral blood

i in the absence of anti-fecundity immunity

Adapted from: Subramanian S et al. The dynamics of Wuchereria bancrofti infection: a model-based analysis of longitudinal data from Pondicherry, India. Parasitol, 2004, 128:467–482, with permission from Cambridge University Press.

Adapted from: Norman RA et al. EPIFIL: the development of an age-structured model for describing the transmission dynamics and control of lymphatic filariasis. Epidemiol Infect, 2000, 124:529–541, with permission from Cambridge University Press.

Epifil

EPIFIL simulates the average course of infection over age and time in a human population by a set of differential equations. The human population is constant in size and age-structure. Limitation in the transmission of infection by culicine mosquitoes is taken into account, so that the number of infectious L3 larvae that can develop in mosquitoes saturates at higher mf intensities. Acquired immunity is included as a second limitation mechanism: it is triggered by incoming L3 larvae and reduces the probability that new larvae develop into adult worms. Heterogeneity is only included by age-related exposure to mosquitoes, i.e. the risk of infection increases with age, until a maximum level is reached at the age of 9 years. The mf prevalence is calculated using a negative binomial distribution, assuming a certain amount of aggregation of parasites in the human population.

The model can be used to simulate the impact of vector control or mass treatment. Vector control is assumed to reduce the mosquito biting rate. Mass treatment leads to killing of a proportion of adult worms or mf and to temporal infertility of worms, depending on the proportion of the population that receives treatment and characteristics of the treatment regimen.

The design of this population-based, deterministic model is based on a general differential equation framework describing the dynamics of macroparasitic infections [19,27,28].

Lymfasim

LYMFASIM simulates the acquisition and loss of worms over age and time in a discrete number of human individuals, using stochastic microsimulation. Individuals interact through biting mosquitoes and together they form a dynamic population of which the size and age-structure may change over time. Like EPIFIL, LYMFASIM takes account of limitation in the proportion of engorged mf that develop into L3 larvae inside the mosquito and of acquired immunity in human hosts. Two model variants were developed for Pondicherry, which differed with respect to the type of acquired immunity: ‘anti-L3’ immunity is triggered by incoming L3 larvae and reduces the probability of successful adult worm establishment; ‘anti-fecundity’ immunity is triggered by the presence of adult worms and reduces the rate of mf production by female worms. By considering individual worms in individual hosts, the model automatically takes account of the declining mating probability of female and male worms with lower average infection intensities. Age-dependent exposure is included, assuming that exposure increases until a maximum is reached at about 20 years of age. Other factors contributing to heterogeneity are variation in exposure to infection within age groups, inclination to participate in treatment programmes, the response to treatment, and the ability to develop immune responses. Parasites may vary with respect to their lifespan (about ten years on average). Individual mf intensities are translated into the number of mf that would be counted in a 20 µl blood smear, taking account of random variability in these counts and reduced sensitivity of diagnostic tests at lower mf densities. The mf prevalence and (geometric or arithmetic) mean mf intensity can be directly calculated from the smear counts, using data from all simulated individuals or specific subgroups.

Similar to EPIFIL, LYMFASIM simulates the impact of vector control by reducing the mosquito biting rate. Treatment takes place at the individual level, and results in killing (part) of adult worms or mf and a temporal or permanent reduction in the fertility of female worms. Selective or mass treatment can be simulated.

This individual-based model uses the technique of stochastic microsimulation, which was earlier applied in the modelling of onchocerciasis transmission and control [29,30].

Comparison of model predictions

Both EPIFIL and LYMFASIM have been used to predict the impact of control measures [9,10,31,32]. In this report, we focus on predictions of the coverage and duration of annual mass treatment programmes that will be required for elimination. All published predictions were based on the Pondicherry variants of the model, although acquired immunity was left out of the model in the EPIFIL predictions. From the predictions of both models we can conclude that it is possible to eliminate lymphatic filariasis by yearly mass treatment, but the number of treatment rounds largely depends on coverage, precontrol mf prevalence, and the macrofilaricidal effects of drugs. This is illustrated in tables 2 and 3, and figure 3. Often the required number of yearly treatment rounds is predicted to be higher than 4–6, which was hoped to be sufficient when GPELF was initiated. As an alternative to longer programmes, one might consider more frequent mass treatment (e.g. half-yearly) or applying vector control in addition to mass treatment (fig. 4).

Figure 3

LYMFASIM: prediction of the duration of yearly mass treatment with ivermectin required to reach elimination (zero microfilaraemia prevalence 40 years after the start of treatment) with 99% certainty, in relation to coverage. Ivermectin is assumed to permanently sterilize 77% of female worms and to kill all microfilariae. Results are shown for two variants of the LYMFASIM model for Pondicherry that differ in the type of acquired immunity assumed, assuming a precontrol microfilaraemia prevalence of 8.5% [31].

Figure 4

EPIFIL: the impact of different control strategies on the mean microfilaraemia prevalence in an endemic community with precontrol prevalence of 10%. The plot shows the impact of mass treatment alone (five rounds of annual mass treatment with diethylcarbamazine + albendazole, with a coverage of 80%), vector control alone (assuming a 90% reduction in biting rate during five years), and the combination of the two.

Table 2

LYMFASIM: predicted number of annual rounds of mass drug treatment required to achieve elimination in 99% of the simulation runs in an area like Pondicherry, for four different drugs or drug combinations and two coverage levels. Predictions are based on the anti-L3 variant of the model for Pondicherry, with a precontrol microfilaraemia prevalence of 8.5%. Elimination is defined as zero microfilaraemia prevalence 40 years after the start of treatment.32

	Assumed treatment effects (proportion killed)		Predicted number of rounds required for elimination with coverage of:
Drug (combination)	adult worms	microfilariae	65%	80%
Ivermectin + albendazole	35%	100%	10	6
Diethylcarbamazine	50%	70%	8	5
Diethylcarbamazine + albendazole	65%	70%	6	4
Doxycycline	80%	0%	6	4

* Reprinted from: Stolk WA, de Vlas SJ, Habbema JDF. Anti-Wolbachia treatment for lymphatic filariasis. The Lancet, 2005, 365:2067–2068, with permission from Elsevier.

Table 3

Prediction of number of yearly mass treatment rounds required to reach a 0.5% microfilaraemia prevalence threshold, using a combination of diethylcarbamazine plus albendazole in relation to endemicity and coverage. The combination treatment is assumed to kill 55% of all adult worms and 95% of the microfilariae, and to interrupt microfilaria production for six months. EPIFIL simulations were published [10] and concerned a model without acquired immunity. LYMFASIM results from the model with anti-L3 immunity were added for comparison for an average pretreatment microfilaraemia prevalence of 10%.

		Coverage
Pretreatment mf	60%	70%	80%	90%
prevalence
EPIFIL
2.5%	7	6	5	4
5%	9	7	6	5
10%	10	8	7	6
15%	12	9	8	7
LYMFASIM ^a
10%	10	8	6	5

a Based on the average trend in microfilaraemia prevalence of 100 simulation runs.

Reprinted from: Stolk WA et al. Prospects for elimination of bancroftian filariasis by mass drug treatment in Pondicherry, India: a simulation study. J Infect Dis, 2003, 188:1371–1381, with permission from the University of Chicago Press.

Reprinted from: Michael E et al. Mathematical modelling and the control of lymphatic filariasis. The Lancet Infect Dis, 2004, 4:223–234, with permission from Elsevier.

The predictions of EPIFIL and LYMFASIM cannot be compared directly because the original publications reported results for different treatment regimens, with different assumptions on efficacy of the drugs, and different precontrol mf prevalence levels. Further, different criteria for elimination were used: in EPIFIL elimination was assumed to occur if the mf prevalence after treatment was below 0.5%; in LYMFASIM elimination was defined as a zero mf prevalence 40 years after the start of control in 99% of the runs. To allow better comparison of the models, we did a series of additional simulations with LYMFASIM for mass treatment with the combination of diethylcarbamazine plus albendazole, using the same assumptions on drug efficacy and the same criterion for elimination as in published EPIFIL predictions (table 3). Simulations were done with the anti-L3 variant of the LYMFASIM model.

It is reassuring that both models come to comparable conclusions regarding the number of treatment rounds required to achieve elimination, although LYMFASIM predictions are slightly more optimistic than EPIFIL predictions when population coverage is high. This finding of nearly equal predictions is not straightforward. The LYMFASIM model contains several assumptions and mechanisms, which, relative to EPIFIL, limit the impact of the intervention on transmission: 1) a longer adult worm lifespan (about 10 years vs. 8 years); 2) acquired immunity; 3) heterogeneities in exposure to mosquitoes, compliance to mass treatment, and adult worm lifespan. However, the limiting effect of these assumptions and mechanisms on the impact of mass treatment is apparently counteracted by the enhancing effect of a reduced mating probability of worms at lower average worm burdens in LYMFASIM.

Criteria for elimination

EPIFIL predictions were based on the assumption that transmission will not continue when the mf prevalence falls below 0.5%. The choice for this threshold is somewhat arbitrary in the absence of evidence from the field. Given its individual-based structure, LYMFASIM is more suitable for examining in how many runs infection is ‘truly’ eliminated, as indicated by zero mf prevalence 40 years after the start of control. For example, in the runs with 10% precontrol prevalence, 8 rounds were required to bring the average mf prevalence below 0.5% (table 3). However, in only 87% of the runs did this result in zero mf prevalence 40 years after the start of control. It is clear that to be 99% certain of elimination (as was the criterion in table 2), much longer continuation of mass treatment would be required.

More extensive simulation studies are required to determine a more precise threshold level below which elimination would occur. This threshold level (or threshold levels) will depend on local transmission dynamics and mosquito biting rates, inmigration of parasite carriers or infected mosquitoes, but also on heterogeneities and population size in view of the stochastic processes involved.

Application of models for other regions

The existing model variants were all quantified for transmission of W. bancrofti by Culex quinquefasciatus and tested against data from Pondicherry [17,26]. The basic structure of the models is generalizable to other areas, but various model parameters may take different values. Most importantly, this concerns the relationship between mf density in the human blood and the number of L3 larvae developing in mosquitoes. Unfortunately, few data are available to quantify this relationship for the different mosquito species involved [33]. Especially for the anopheline mosquito species responsible for transmission in large parts of Africa, more field research is needed. Other parameters that may need requantification relate to the composition of the human population, mosquito biting rates and heterogeneity in exposure, and operational characteristics of interventions.

Biological parameters are not expected to vary much between regions. However, our understanding of the biology of infection (in spite of in-depth model-based analysis of the Pondicherry data) is incomplete and there is uncertainty about the quantification of several key parameters, such as the parasite lifespan or role of acquired immunity. Therefore, it is crucial to continue testing the validity of existing and new model variants against epidemiological data. Testing models against age-specific data may help to determine the role of acquired immunity or other processes [34]. Trends during vector control are especially informative about the adult worm lifespan [26,35]. Trends during mass treatment may give information about the effects of drugs on worm survival and productivity. And trends after cessation of control may help to determine whether density-dependent mechanisms have appropriately been included in the model. Better information on all these aspects should eventually come from field research: using combinations of available diagnostic tests (mf and antigen detection, ultrasound to visualize adult worms), it may be possible to further increase the validity of our existing models. Some work has already been done to prepare models for use in other areas. The LYMFASIM model has been applied to age-patterns observed in an area of South-East India that has the same vector-parasite combination and presumably the same transmission dynamics as Pondicherry. This led to the development of new model variants with less strong or no immunity (Subramanian, unpublished data). Comparison of predictions from the new LYMFASIM model variant and EPIFIL with observed trends during mass treatment in this region indicated that assumptions regarding efficacy of drugs or possibly coverage and compliance patterns had to be adapted (Subramanian, unpublished data) [10]. Using published data of uptake and development of mf in Anopheles mosquitoes [22,36,38], LYMFASIM was adapted for transmission in Africa (Stolk, unpublished data). Model parameters were adapted so that the predicted age-prevalence reflects the observed data from this region [25].

Challenges in the evaluation of current elimination programmes

The available models soon have to face new challenges in the ongoing programmes for elimination of lymphatic filariasis. Predictions of the number of treatment rounds required for elimination were only a first step. However, specific programmes also need to be monitored and evaluated. For example, the observed results can be compared with model predictions to see whether progress is as expected. If results lag behind, programmes can be adapted. Also, the models could help to determine when mass treatment can be stopped with low risk of recrudescence, taking account of the specific local conditions, local coverage and compliance levels, and the achieved reduction in mf prevalence and intensity. Analogously, models can help to determine cost-effective surveillance strategies for early detection of recrudescence of infection after cessation of control, and measures to be taken to stop this recrudescence.

To address the discussed issues on monitoring and surveillance, the models must be extended to include the results of antigen detection, which is widely used in monitoring and surveillance by ongoing control programmes. Other possibly useful extensions of the model include migration of parasite carriers and infected mosquitoes and development of resistance to available drugs.

Although discussion until now focused on the elimination of transmission, this goal may be difficult to achieve in some areas. In some situations focus may shift to reducing the public health problem without explicitly eliminating infection. To address this with the models, more attention is required for the development of disease. Simple mechanisms of disease development are included in both models, but this has received little attention in published work until now.

Conclusions

There are currently two models for lymphatic filariasis transmission and control, LYMFASIM and EPIFIL, that have been used in predicting the impact of mass treatment programmes. These models give more or less similar predictions on the number of treatment rounds that will be required for elimination, at least in Pondicherry-like situations. The models differ however in defining when elimination occurs, which leads to different advice on the duration of mass treatment. In view of current elimination programmes, it is crucial to obtain better criteria for when to stop control, taking account of stochasticity in the eventual outcome of elimination. Antigen tests should be included in the model, and the disease part of the models may need more attention. Model variants that are adjusted to local situations are powerful tools to aid decision-making in current control programmes.

References

1. Molyneux DH, Zagaria N, (2002) Lymphatic filariasis elimination: progress in global programme development. Ann Trop Med Parasitol 96: S15-S40.

2. WHO, (2003) Lymphatic filariasis - Progress report on the Programme in 2002. Wkly Epidemiol Rec 78: 171-179.

3. Schlemper BR, etal (2000) Elimination of bancroftian filariasis (Wuchereria bancrofti) in Santa Catarina state, Brazil. Trop Med Int Health 5: 848-854.

4. Esterre P, etal (2001) The impact of 34 years of massive DEC chemotherapy on Wuchereria bancrofti infection and transmission: the Maupiti cohort. Trop Med Int Health 6: 190-195.

5. Vector Control Research Centre, (2003) Annual report 2003 Pondicherry, India:VCRC .

6. (2004) Towards a strategic plan for research to support the global program to eliminate lymphatic filariasis. Summary of immediate needs and opportunities for research on lymphatic filariasis. Philadelphia, Pennsylvania, USA, December 9–10, 2003. Am J Trop Med Hyg 71: 1-46.

7. Goodman B, (1994) Models aid understanding, help control parasites. Science 264: 1862-1863. [news]

8. Habbema JD, etal (1992) Epidemiological modelling for onchocerciasis control. Parasitol Today 8: 99-103.

9. Das PK, Subramanian S, (2002) Modelling the epidemiology, transmission and control of lymphatic filariasis. Ann Trop Med Parasitol 96: S153-S164.

10. Michael E, etal (2004) Mathematical modelling and the control of lymphatic filariasis. Lancet Infect Dis 4: 223-234.

11. Michael E, Bundy DA, (1998) Herd immunity to filarial infection is a function of vector biting rate. Proc R Soc Lond B Biol Sci 265: 855-860.

12. Michael E, etal (2001) Transmission intensity and the immunoepidemiology of bancroftian filariasis in East Africa. Parasite Immunol 23: 373-388.

13. Plaisier AP, etal (1999) Efficacy of ivermectin in the treatment of Wuchereria bancrofti infection: a model-based analysis of trial results. Parasitol 119: 385-394.

14. Stolk WA, etal Effects of ivermectin and diethylcarbamazine on microfilariae and microfilaria production in bancroftian filariasis . submitted for publication

15. Rochet MJ, (1990) A simple deterministic model for bancroftian filariasis transmission dynamics. Trop Med Parasitol 41: 225-233.

16. Chan MS, etal (1998) Epifil: a dynamic model of infection and disease in lymphatic filariasis. Am J Trop Med Hyg 59: 606-614.

17. Norman RA, etal (2000) EPIFIL: the development of an age-structured model for describing the transmission dynamics and control of lymphatic filariasis. Epidemiol Infect 124: 529-541.

18. Plaisier AP, etal (1998) The LYMFASIM simulation program for modeling lymphatic filariasis and its control. Methods Inf Med 37: 97-108.

19. Anderson RM, May RM, (1991) Infectious diseases of humans: dynamics and control Oxford:Oxord University Press .

20. Duerr HP, Dietz K, Eichner M, (2005) Determinants of the eradicability of filarial infections: a conceptual approach. Trends Parasitol 21: 88-96.

21. Churcher TS, Ferguson NM, Basáñez M-G, (2005) Density dependence and overdispersion in the transmission of helminth parasites. Parasitol 131: 1-12.

22. Southgate BA, Bryan JH, (1992) Factors affecting transmission of Wuchereria bancrofti by anopheline mosquitoes. 4. Facilitation, limitation, proportionality and their epidemiological significance. Trans R Soc Trop Med Hyg 86: 523-530.

23. Subramanian S, etal (1998) The relationship between microfilarial load in the human host and uptake and development of Wuchereria bancrofti microfilariae by Culex quinquefasciatus: a study under natural conditions. Parasitol 116: 243-255.

24. Krishnamoorthy K, etal (2004) Vector survival and parasite infection: the effect of Wuchereria bancrofti on its vector Culex quinquefasciatus. Parasitol 129: 43-50.

25. Stolk WA, etal (2004) Meta-analysis of age-prevalence patterns in lymphatic filariasis: no decline in microfilaraemia prevalence in older age groups as predicted by models with acquired immunity. Parasitol 129: 605-612.

26. Subramanian S, etal (2004) The dynamics of Wuchereria bancrofti infection: a model-based analysis of longitudinal data from Pondicherry, India. Parasitol 128: 467-482.

27. Anderson RM, May RM, (1985) Helminth infections of humans: mathematical models, population dynamics, and control. Adv Parasitol 24: 1-101.

28. Woolhouse ME, (1992) A theoretical framework for the immunoepidemiology of helminth infection. Parasite Immunol 14: 563-578.

29. Habbema JD, etal (1996) The microsimulation approach to epidemiologic modeling of helminthic infections, with special reference to schistosomiasis. Am J Trop Med Hyg 55: 165-169.

30. Plaisier AP, etal (1990) ONCHOSIM: a model and computer simulation program for the transmission and control of onchocerciasis. Comput Methods Programs Biomed 31: 43-56.

31. Stolk WA, etal (2003) Prospects for elimination of bancroftian filariasis by mass drug treatment in Pondicherry, India: a simulation study. J Infect Dis 188: 1371-1381.

32. Stolk WA, de Vlas SJ, Habbema JDF, (2005) Anti-Wolbachia treatment for lymphatic filariasis. Lancet 365: 2067-2068. discussion

33. Snow LC, Michael E, (2002) Transmission dynamics of lymphatic filariasis: density-dependence in the uptake of Wuchereria bancrofti microfilariae by vector mosquitoes. Med Vet Entomol 16: 409-423.

34. Duerr HP, Dietz K, Eichner M, (2003) On the interpretation of age-intensity profiles and dispersion patterns in parasitological surveys. Parasitol 126: 87-101.

35. Vanamail P, etal (1996) Estimation of the fecund life span of Wuchereria bancrofti in an endemic area. Trans R Soc Trop Med Hyg 90: 119-121.

36. Bryan JH, Southgate BA, (1988) Factors affecting transmission of Wuchereria bancrofti by anopheline mosquitoes. 1. Uptake of microfilariae. Trans R Soc Trop Med Hyg 82: 128-137.

37. Bryan JH, Southgate BA, (1988) Factors affecting transmission of Wuchereria bancrofti by anopheline mosquitoes. 2. Damage to ingested microfilariae by mosquito foregut armatures and development of filarial larvae in mosquitoes. Trans R Soc Trop Med Hyg 82: 138-145.

38. Boakye DA, etal (2004) Vector competence, for Wuchereria bancrofti, of the Anopheles populations in the Bongo district of Ghana. Ann Trop Med Parasitol 98: 501-508.

39. WHO, (1992) Lymphatic filariasis: the disease and its control. Fifth report of the WHO Expert Committee on Filariasis Geneva:World Health Organanization . WHO Technical Report Series, No. 821, pages1–71