Limits...
Multi-host transmission dynamics of Schistosoma japonicum in Samar province, the Philippines.

Riley S, Carabin H, Bélisle P, Joseph L, Tallo V, Balolong E, Willingham AL, Fernandez TJ, Gonzales RO, Olveda R, McGarvey ST - PLoS Med. (2008)

Bottom Line: Our data provided substantially more support for model structure (a) than for model structure (b).Fitted values for the village-level transmission intensity from snails to mammals appeared to be strongly spatially correlated, which is consistent with results from descriptive hierarchical analyses.These findings have implications for the prioritization of mitigating interventions against S. japonicum transmission.

View Article: PubMed Central - PubMed

Affiliation: Department of Community Medicine and School of Public Health, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong Special Administrative Region, China. steven.riley@hku.hk

ABSTRACT

Background: Among the 6.7 million people living in areas of the Philippines where infection with Schistosoma japonicum is considered endemic, even within small geographical areas levels of infection vary considerably. In general, the ecological drivers of this variability are not well described. Unlike other schistosomes, S. japonicum is known to infect several mammalian hosts. However, the relative contribution of different hosts to the transmission cycle is not well understood. Here, we characterize the transmission dynamics of S. japonicum using data from an extensive field study and a mathematical transmission model.

Methods and findings: In this study, stool samples were obtained from 5,623 humans and 5,899 potential nonhuman mammalian hosts in 50 villages in the Province of Samar, the Philippines. These data, with variable numbers of samples per individual, were adjusted for known specificities and sensitivities of the measurement techniques before being used to estimate the parameters of a mathematical transmission model, under the assumption that the dynamic transmission processes of infection and recovery were in a steady state in each village. The model was structured to allow variable rates of transmission from different mammals (humans, dogs, cats, pigs, domesticated water buffalo, and rats) to snails and from snails to mammals. First, we held transmission parameters constant for all villages and found that no combination of mammalian population size and prevalence of infectivity could explain the observed variability in prevalence of infection between villages. We then allowed either the underlying rate of transmission (a) from snails to mammals or (b) from mammals to snails to vary by village. Our data provided substantially more support for model structure (a) than for model structure (b). Fitted values for the village-level transmission intensity from snails to mammals appeared to be strongly spatially correlated, which is consistent with results from descriptive hierarchical analyses.

Conclusions: Our results suggest that the process of acquiring mammalian S. japonicum infection is more important in explaining differences in prevalence of infection between villages than the process of snails becoming infected. Also, the contribution from water buffaloes to human S. japonicum infection in the Philippines is less important than has been recently observed for bovines in China. These findings have implications for the prioritization of mitigating interventions against S. japonicum transmission.

Show MeSH

Related in: MedlinePlus

Comparison of Model (Red) and Observed (Grey) Prevalence Under Different Hypotheses for Two Classes of Human Infection and Five Potential Animal Reservoir SpeciesHypotheses were as follows. H0 (no site-specific variability, left column), H1 (site specific variability in process of infection from mammals to snails, centre column), and H2 (site specific variability in process of infection from snails to mammals, right column). Each of the 49 villages is indexed (x-axis) according to the prevalence of infection for each particular species, i.e., the ordering of villages is different for charts in different rows in this graph (other than top two rows). For humans (top two rows), the villages are indexed according to estimated overall prevalence. Grey points show the observed prevalence, adjusted jointly for the sensitivity and specificity of parasitological tests and the number of stool samples provided per participant. Binomial confidence intervals are presented for these adjusted data. The red points show the maximum likelihood steady-state values for the transmission model.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC2211559&req=5

pmed-0050018-g002: Comparison of Model (Red) and Observed (Grey) Prevalence Under Different Hypotheses for Two Classes of Human Infection and Five Potential Animal Reservoir SpeciesHypotheses were as follows. H0 (no site-specific variability, left column), H1 (site specific variability in process of infection from mammals to snails, centre column), and H2 (site specific variability in process of infection from snails to mammals, right column). Each of the 49 villages is indexed (x-axis) according to the prevalence of infection for each particular species, i.e., the ordering of villages is different for charts in different rows in this graph (other than top two rows). For humans (top two rows), the villages are indexed according to estimated overall prevalence. Grey points show the observed prevalence, adjusted jointly for the sensitivity and specificity of parasitological tests and the number of stool samples provided per participant. Binomial confidence intervals are presented for these adjusted data. The red points show the maximum likelihood steady-state values for the transmission model.

Mentions: Initially, we assumed that there was no difference between the underlying transmission parameters in each village, i.e., for hypothesis H0 (Table 1), βMS(j) = βMS(k) and βSM(j) = βSM(k) for j,k ∈ 1…49. Therefore, the only source of variation in transmission was the different numbers of hosts of each species in each village. This model variant was not flexible enough to reproduce the observed variation in the data adjusted for measurement error (Figure 2). For each village, for a given class of infection prevalence, the values were very similar. This result suggests that it is not the distribution of potential mammalian reservoir and humans hosts in each village that drives the village-to-village variation in transmission. For example, there is no strong trend of increased infection in humans in those villages with a greater number of water buffaloes: this result is consistent with previous analyses based on a nonmechanistic model of these data [8].


Multi-host transmission dynamics of Schistosoma japonicum in Samar province, the Philippines.

Riley S, Carabin H, Bélisle P, Joseph L, Tallo V, Balolong E, Willingham AL, Fernandez TJ, Gonzales RO, Olveda R, McGarvey ST - PLoS Med. (2008)

Comparison of Model (Red) and Observed (Grey) Prevalence Under Different Hypotheses for Two Classes of Human Infection and Five Potential Animal Reservoir SpeciesHypotheses were as follows. H0 (no site-specific variability, left column), H1 (site specific variability in process of infection from mammals to snails, centre column), and H2 (site specific variability in process of infection from snails to mammals, right column). Each of the 49 villages is indexed (x-axis) according to the prevalence of infection for each particular species, i.e., the ordering of villages is different for charts in different rows in this graph (other than top two rows). For humans (top two rows), the villages are indexed according to estimated overall prevalence. Grey points show the observed prevalence, adjusted jointly for the sensitivity and specificity of parasitological tests and the number of stool samples provided per participant. Binomial confidence intervals are presented for these adjusted data. The red points show the maximum likelihood steady-state values for the transmission model.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC2211559&req=5

pmed-0050018-g002: Comparison of Model (Red) and Observed (Grey) Prevalence Under Different Hypotheses for Two Classes of Human Infection and Five Potential Animal Reservoir SpeciesHypotheses were as follows. H0 (no site-specific variability, left column), H1 (site specific variability in process of infection from mammals to snails, centre column), and H2 (site specific variability in process of infection from snails to mammals, right column). Each of the 49 villages is indexed (x-axis) according to the prevalence of infection for each particular species, i.e., the ordering of villages is different for charts in different rows in this graph (other than top two rows). For humans (top two rows), the villages are indexed according to estimated overall prevalence. Grey points show the observed prevalence, adjusted jointly for the sensitivity and specificity of parasitological tests and the number of stool samples provided per participant. Binomial confidence intervals are presented for these adjusted data. The red points show the maximum likelihood steady-state values for the transmission model.
Mentions: Initially, we assumed that there was no difference between the underlying transmission parameters in each village, i.e., for hypothesis H0 (Table 1), βMS(j) = βMS(k) and βSM(j) = βSM(k) for j,k ∈ 1…49. Therefore, the only source of variation in transmission was the different numbers of hosts of each species in each village. This model variant was not flexible enough to reproduce the observed variation in the data adjusted for measurement error (Figure 2). For each village, for a given class of infection prevalence, the values were very similar. This result suggests that it is not the distribution of potential mammalian reservoir and humans hosts in each village that drives the village-to-village variation in transmission. For example, there is no strong trend of increased infection in humans in those villages with a greater number of water buffaloes: this result is consistent with previous analyses based on a nonmechanistic model of these data [8].

Bottom Line: Our data provided substantially more support for model structure (a) than for model structure (b).Fitted values for the village-level transmission intensity from snails to mammals appeared to be strongly spatially correlated, which is consistent with results from descriptive hierarchical analyses.These findings have implications for the prioritization of mitigating interventions against S. japonicum transmission.

View Article: PubMed Central - PubMed

Affiliation: Department of Community Medicine and School of Public Health, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong Special Administrative Region, China. steven.riley@hku.hk

ABSTRACT

Background: Among the 6.7 million people living in areas of the Philippines where infection with Schistosoma japonicum is considered endemic, even within small geographical areas levels of infection vary considerably. In general, the ecological drivers of this variability are not well described. Unlike other schistosomes, S. japonicum is known to infect several mammalian hosts. However, the relative contribution of different hosts to the transmission cycle is not well understood. Here, we characterize the transmission dynamics of S. japonicum using data from an extensive field study and a mathematical transmission model.

Methods and findings: In this study, stool samples were obtained from 5,623 humans and 5,899 potential nonhuman mammalian hosts in 50 villages in the Province of Samar, the Philippines. These data, with variable numbers of samples per individual, were adjusted for known specificities and sensitivities of the measurement techniques before being used to estimate the parameters of a mathematical transmission model, under the assumption that the dynamic transmission processes of infection and recovery were in a steady state in each village. The model was structured to allow variable rates of transmission from different mammals (humans, dogs, cats, pigs, domesticated water buffalo, and rats) to snails and from snails to mammals. First, we held transmission parameters constant for all villages and found that no combination of mammalian population size and prevalence of infectivity could explain the observed variability in prevalence of infection between villages. We then allowed either the underlying rate of transmission (a) from snails to mammals or (b) from mammals to snails to vary by village. Our data provided substantially more support for model structure (a) than for model structure (b). Fitted values for the village-level transmission intensity from snails to mammals appeared to be strongly spatially correlated, which is consistent with results from descriptive hierarchical analyses.

Conclusions: Our results suggest that the process of acquiring mammalian S. japonicum infection is more important in explaining differences in prevalence of infection between villages than the process of snails becoming infected. Also, the contribution from water buffaloes to human S. japonicum infection in the Philippines is less important than has been recently observed for bovines in China. These findings have implications for the prioritization of mitigating interventions against S. japonicum transmission.

Show MeSH
Related in: MedlinePlus