Modelling the geographical distribution of co-infection risk from single-disease surveys.

BACKGROUND: The need to deliver interventions targeting multiple diseases in a cost-effective manner calls for integrated disease control efforts. Consequently, maps are required that show where the risk of co-infection is particularly high. Co-infection risk is preferably estimated via Bayesian geostatistical multinomial modelling, using data from surveys screening for multiple infections simultaneously. However, only few surveys have collected this type of data. METHODS: Bayesian geostatistical shared component models (allowing for covariates, disease-specific and shared spatial and non-spatial random effects) are proposed to model the geographical distribution and burden of co-infection risk from single-disease surveys. The ability of the models to capture co-infection risk is assessed on simulated data sets based on multinomial distributions assuming light- and heavy-dependent diseases, and a real data set of Schistosoma mansoni-hookworm co-infection in the region of Man, Côte d'Ivoire. The data were restructured as if obtained from single-disease surveys. The estimated results of co-infection risk, together with independent and multinomial model results, were compared via different validation techniques. RESULTS: The results showed that shared component models result in more accurate estimates of co-infection risk than models assuming independence in settings of heavy-dependent diseases. The shared spatial random effects are similar to the spatial co-infection random effects of the multinomial model for heavy-dependent data. CONCLUSIONS: In the absence of true co-infection data geostatistical shared component models are able to estimate the spatial patterns and burden of co-infection risk from single-disease survey data, especially in settings of heavy-dependent diseases.

Spatial dispersion and characterisation of mosquito breeding habitats in urban vegetable-production areas of Abidjan, Côte d'Ivoire.

Although urban agriculture (UA) in the developing world may enhance nutrition and local economies, it may also lead to higher densities of mosquito breeding sites and, consequently, to increased transmission of malarial parasites. If targeted interventions against malaria vectors are to be successful in urban areas, the habitats that support Anopheles breeding need to be identified and detected. Mosquito breeding sites have recently been characterised, and the factors associated with productive Anopheles habitats identified, in market gardens of Abidjan, Côte d'Ivoire. Two surveys were conducted in seven vegetable-production areas, one towards the end of the rainy season and one during the dry season. A standardized methodology was used for habitat characterisation and the detection of Anopheles larvae and mosquito pupae. Overall, 454 and 559 potential mosquito-breeding sites were recorded in the rainy-season and dry-season surveys, respectively. In the rainy season, Anopheles larvae and mosquito pupae were found in 29.7% and 5.5% of the potential breeding sites, respectively, whereas the corresponding percentages in the dry season were 24.3% and 8.6%. The potential breeding sites in an agricultural zone on the periphery of Abidjan were those least likely to be positive for Anopheles larvae and mosquito pupae whereas 'agricultural trenches' between seedbeds were the sites most likely to be positive. In a spatially-explicit Bayesian multivariate logistic-regression model, although one out of every five such wells was also found to harbour Anopheles larvae, irrigation wells were found to be the least productive habitats, of those sampled, for pupae. In the study area, simple and cost-effective strategies of larval control should be targeted at agricultural trenches, ideally with the active involvement of local stakeholders (i.e. urban farmers and urban agricultural extension services).

Effectiveness of dog rabies vaccination programmes: comparison of owner-charged and free vaccination campaigns.

We investigated the percentage of dogs that could be vaccinated against rabies by conducting a pilot campaign in N'Djaména, Chad. Owners were charged US$4.13 per dog vaccinated, and 24% of all dogs in the three city districts covered by the campaign were vaccinated. Total campaign costs were US$7623, resulting in an average of US$19.40 per vaccinated dog. This is five times more expensive than the cost per animal vaccinated during a previous free vaccination campaign for dog-owners, conducted in the same districts. The free campaign, which vaccinated 2605 more dogs than this campaign, cost an additional US$1.45 per extra dog vaccinated. Campaigns in which owners are charged for vaccinations result in lower vaccination rates than in free campaigns. Public health officials can use these results when evaluating the costs and benefits of subsidizing dog rabies vaccination programmes.

Bayesian geostatistical modelling for mapping schistosomiasis transmission.

Progress has been made in mapping and predicting the risk of schistosomiasis using Bayesian geostatistical inference. Applications primarily focused on risk profiling of prevalence rather than infection intensity, although the latter is particularly important for morbidity control. In this review, the underlying assumptions used in a study mapping Schistosoma mansoni infection intensity in East Africa are examined. We argue that the assumption of stationarity needs to be relaxed, and that the negative binomial assumption might result in misleading inference because of a high number of excess zeros (individuals without an infection). We developed a Bayesian geostatistical zero-inflated (ZI) regression model that assumes a non-stationary spatial process. Our model is validated with a high-quality georeferenced database from western Côte d'Ivoire, consisting of demographic, environmental, parasitological and socio-economic data. Nearly 40% of the 3818 participating schoolchildren were infected with S. mansoni, and the mean egg count among infected children was 162 eggs per gram of stool (EPG), ranging between 24 and 6768 EPG. Compared to a negative binomial and ZI Poisson and negative binomial models, the Bayesian non-stationary ZI negative binomial model showed a better fit to the data. We conclude that geostatistical ZI models produce more accurate maps of helminth infection intensity than the spatial negative binomial ones.

Remote sensing, geographical information system and spatial analysis for schistosomiasis epidemiology and ecology in Africa.

Beginning in 1970, the potential of remote sensing (RS) techniques, coupled with geographical information systems (GIS), to improve our understanding of the epidemiology and control of schistosomiasis in Africa, has steadily grown. In our current review, working definitions of RS, GIS and spatial analysis are given, and applications made to date with RS and GIS for the epidemiology and ecology of schistosomiasis in Africa are summarised. Progress has been made in mapping the prevalence of infection in humans and the distribution of intermediate host snails. More recently, Bayesian geostatistical modelling approaches have been utilized for predicting the prevalence and intensity of infection at different scales. However, a number of challenges remain; hence new research is needed to overcome these limitations. First, greater spatial and temporal resolution seems important to improve risk mapping and understanding of transmission dynamics at the local scale. Second, more realistic risk profiling can be achieved by taking into account information on people's socio-economic status; furthermore, future efforts should incorporate data on domestic access to clean water and adequate sanitation, as well as behavioural and educational issues. Third, high-quality data on intermediate host snail distribution should facilitate validation of infection risk maps and modelling transmission dynamics. Finally, more emphasis should be placed on risk mapping and prediction of multiple species parasitic infections in an effort to integrate disease risk mapping and to enhance the cost-effectiveness of their control.

The reliability of diagnostic techniques in the diagnosis and management of malaria in the absence of a gold standard.

The accuracy of techniques for the diagnosis of malaria are usually compared with optical microscopy, which is considered to be a gold standard. However, microscopy is prone to error and therefore makes it difficult to assess the reliability of other diagnostic techniques. We did a systematic review to assess the specificity and sensitivity of diagnostic techniques in different settings, using a statistical method that avoided defining a gold standard. Performance varied depending on species of the malaria parasite, level of parasitaemia, and immunity. Overall, histidine-rich protein 2 (HRP2)-based dipsticks showed a high sensitivity (92.7%) and specificity (99.2%) for Plasmodium falciparum in endemic areas. The acridine orange test was more sensitive (97.1%) in detecting P falciparum in epidemiological studies, with a specificity of 97.9%. In the absence of a gold standard, HRP2 dipsticks and acridine orange could provide an alternative for detecting falciparum infections in endemic areas and epidemiological studies, respectively. Microscopy still remains more reliable in detecting non-falciparum infections.

Prevalence of geohelminths in savana and forest areas of Côte d'Ivoire.

BACKGROUND: Côte d'Ivoire has large regional variation in intestinal helminth prevalence. STUDY DESIGN: In a large cross-sectional study conducted from 1997 to 1999, stool samples from 6952 children aged 4-15 years were examined for helminth eggs by the Kato/ Katz technique from 24 villages in the savanna (North) and from 21 villages in the forest (West) in Côte d'Ivoire. RESULTS AND CONCLUSION: Ascariasis lumbricoides (0.7%) was the only species present in the savana area. In the forest area, ascariasis occurred most frequently (18.9%), followed by trichuriasis (2%), Strongyloidiasis (0.1%) and oxyuriasis (0.03%). Ascariasis prevalence was not significantly associated with age within each ecological zone, but was significantly higher in boys in the forest zone. This study confirms that in the more arid savanna, the conditions are less suitable for helminth transmission than in the forest zone.

Efficacy of 10-day melarsoprol schedule 2 years after treatment for late-stage gambiense sleeping sickness.

In 2000, we reported that a new short treatment schedule of melarsoprol was not worse than the longer and demanding standard treatment for late-stage human African trypanosomiasis. This alternative schedule was assessed in an open, randomised clinical equivalence trial of 500 patients in Angola. 24 h after treatment, all patients were parasite free. Of 442 patients, 12 (3%) had relapsed after 1 year, of whom seven (3%) had had standard treatment and five (2%) the alternative treatment. After 2 years, 23 (5%) relapsing patients were reported, 11 (5%) in the standard treatment group and 12 (6%) in the new group. The results at the 2-year follow-up support and strengthen our previous findings.

Spatial modelling of multinomial data with latent structure: an application to geographical mapping of human gene and haplotype frequencies.

We develop hierarchical models for spatial multinomial data with missing categories, to analyse a database of HLA-A and -B gene and haplotype frequencies from Papua New Guinea, with a highly variable number of samples per spatial unit. The spatial structure of the multinomial data is incorporated by adopting conditional autoregressive (CAR) priors for the random effects, reflecting extra-multinomial variation. Different spatial structures are investigated, and covariate effects are evaluated using a novel model selection criterion. Tables and maps reveal strong spatial association and the importance of altitude, a covariate anticipated to be significant in explaining genetic variation. Our approach can be used in identifying associations with environmental factors, linguistic or epidemiological patterns and hence potential causes of genetic diversity (population movements, natural selection, stochastic effects).

Rice irrigation and schistosomiasis in savannah and forest areas of Côte d'Ivoire.

Prevalence and intensity of infection of Schistosoma haematobium and Schistosoma mansoni were studied in relation to irrigated rice cultivation in Côte d'Ivoire. Urine and stool samples were collected from 4 to 15-year-old children in 24 villages in the savannah zone and 21 villages in the forest zone. Villages were classified according to surrounding inland valleys into three agro-ecosystems: (R2) full or partial water control allowing two rice cycles per year; (R1) no or partial water control allowing one harvest per year and (R0) absence of rice growing. In the savannah zone, S. haematobium prevalence was 4.8%, 2.3% and 0.7% and S. mansoni prevalence was 16.1%, 11.9% and 2.1% in R2, R1 and R0, respectively. In the forest zone, S. haematobium prevalence was 0.9%, 4.4% and 1.7% and S. mansoni prevalence was 61.3%, 46.6% and 17.5% in R2, in R1 and R0, respectively. Prevalences of S. mansoni adjusted for village effects were significantly different between agro-ecosystems in both zones. Significance of differences between agro-ecosystems of S. haematobium infection were strongly influenced by outlying villages. In savannah rice growing villages, negative binomial regression on infection intensity of each species showed significant positive relations to the surface of rice cultivated inland valleys, whereas uncultivated inland valleys showed no significant relation. However, in forest rice growing villages, S. mansoni infection intensity showed significant positive relations to the surface of uncultivated inland valleys, whereas surface water on rice cultivated land showed significant negative relations with infection intensity of each schistosomiasis species.

A model of animal-human brucellosis transmission in Mongolia.

We developed a dynamic model of livestock-to-human brucellosis transmission in Mongolia. The compartmental model considers transmission within sheep and cattle populations and the transmission to humans as additive components. The model was fitted to demographic and seroprevalence data (Rose Bengal test) from livestock and annually reported new human brucellosis cases in Mongolia for 1991-1999 prior to the onset of a mass livestock-vaccination campaign (S19 Brucella abortus for cattle and Rev 1 Brucella melitensis for sheep and goat). The vaccination effect was fitted to livestock- and human-brucellosis data from the first 3 years of the vaccination campaign (2000-2002). Parameters were optimized on the basis of the goodness-of-fit (assessed by the deviance). The simultaneously fitted sheep-human and cattle-human contact rates show that 90% of human brucellosis was small-ruminant derived. Average effective reproductive ratios for the year 1999 were 1.2 for sheep and 1.7 for cattle.

A potential impact of climate change and water resource development on the transmission of Schistosoma japonicum in China.

There is growing consensus among climate modellers that the unusual global warming observed in the last decades of the 20th century is primarily forced by human activities, namely greenhouse gas increases in the atmosphere. Global warming will trigger alterations in physical and biological systems, including shifts in the spatio-temporal distribution of disease vectors, but the nature and extent of these changes are poorly understood. The purpose of the present study was to assess the potential impact of climate change and water resource development on the distribution of Oncomelania hupensis, the intermediate host snail of Schistosoma japonicum. We employed two 30-year composite datasets comprising average monthly temperatures collected at 623 observing stations throughout China, spanning the periods 1961-1990 and 1971-2000. Temperature changes were assessed spatially between the 1960s and 1990s for January, as this is the critical month for survival of O. hupensis. Our database shows that January temperatures increased at 590 stations (94.7%), and that China's average January temperature in the 1990s was 0.96 degrees C higher than 30 years earlier. The historical 0-1 degrees C January isotherm, which was considered the approximate northern limit of S. japonicum transmission, has shifted from 33 degrees 15' N to 33 degrees 41' N, expanding the potential transmission area by 41,335 km2. This translates to an additional 20.7 million people at risk of schistosomiasis. Two lakes are located in this new transmission area that form part of the proposed South-North water transfer project. Climate change, coupled with water resource developments in China, may pose additional challenges for the control of schistosomiasis.

Evolution of Schistosoma haematobium-related pathology over 24 months after treatment with praziquantel among school children in southeastern Tanzania.

Little is known about the dynamics of pathology due to schistosomiasis following treatment. Public health authorities in endemic areas require such information to decide on the timing of treatment and re-treatment schedules. A study to assess the rate of clearance and reappearance of pathologic lesions due to Schistosoma haematobium using ultrasound has now been carried out in two schools in southeastern Tanzania, an area of moderate-to-high transmission. Baseline data collection found urinary tract pathology in 67% of 533 children. Lesions of the bladder were significantly associated with egg positivity and microhematuria. The attributable fraction estimate of major bladder lesions due to S. haematobium was 75%. In a cohort study, 224 infected children were examined by ultrasound and then treated with a standard dose of 40 mg of praziquantel/kg of body weight. They were re-examined at two, four, six, 12, 18, and 24 months after treatment. Before treatment, 76% had pathologic lesions of the urinary tract. The proportion showing lesions decreased sharply during the first months after treatment to 11% at six months. At 24 months, lesions were detected in 57%, and 11% had developed new severe pathology. In 18 cases, pathology was present throughout, and 34 did not show any pathology throughout the study. This study provides the first detailed report on the evolution of urinary tract pathology due to S. haematobium infections at the community level. The results will help in making decisions on treatment and re-treatment schedules and more generally will provide a basis for designing control strategies in areas of moderate-to-high transmission.

Effect of iron supplementation and malaria prophylaxis in infants on Plasmodium falciparum genotypes and multiplicity of infection.

During a randomized placebo-controlled trial of chemoprophylaxis against Plasmodium falciparum malaria and iron supplementation, in infants living under conditions of intense transmission, all samples of P. falciparum obtained from children aged 5 and 8 months were genotyped by polymerase chain reaction-restriction fragment length polymorphism analysis for the msp2 locus. One hundred and six blood samples were analysed for the number of concurrent infections (multiplicity), and the allelic family of each msp2 genotype was determined. Mean multiplicity of infection was, overall, 2.76 infections/child, and it was significantly reduced in infants receiving chemoprophylaxis. This finding might help to explain the rebound effect in morbidity observed after prophylaxis was ended. Iron supplementation did not affect multiplicity of infection. In infants receiving placebo only, or placebo and iron supplementation, a significant positive association was observed between the number of infections and parasite densities (Spearman's p = 0.25, P-0.047). This association was lost in the group receiving chemoprophylaxis alone, or in combination with iron. This study showed a significant association of FC27-like msp2 alleles with prospective risk of clinical malaria in children (relative risk = 1.487, P = 0.013). Such an association was also found for the present risk of clinical malaria in infants receiving prophylaxis (odds ratio = 3.84, P = 0.026), which might imply that chemoprophylaxis may impair the development of premunition.

Area effects of bednet use in a malaria-endemic area in Papua New Guinea.

Relationships between area coverage with insecticide-free bednets and prevalence of Plasmodium falciparum were investigated in 7 community-based surveys over a 33-month period in 1990-93 in 6 villages in the Wosera area of Papua New Guinea. Spatial patterns in circumsporozoite rates for P. falciparum, P. vivax isomorphs K210 and K247, and P. malariae, and the proportions of mosquito blood meals positive for specific human, goat, cat, dog and pig antigens were determined using ELISAs. P. falciparum prevalence in humans was better explained by bednet coverage in the immediate vicinity than by personal protection alone. Circumsporozoite rates for both P. falciparum and P. vivax were also inversely related to coverage with bednets. There was some increase in zoophagy in areas with high coverage, but relatively little effect on the human blood index or on overall mosquito densities. In this setting, protracted use of untreated bednets apparently reduces sporozoite rates, and the associated effects on prevalence are greater than can be accounted for by personal protection. Even at high bednet coverage most anophelines feed on human hosts, so the decreased sporozoite rates are likely to be largely due to reduction of mosquito survival. This finding highlights the importance of local vector ecology for outcomes of bednet programmes and suggests that area effects of untreated bednets should be reassessed in other settings.

Species-specific field testing of Entamoeba spp. in an area of high endemicity.

Entamoeba histolytica has been separated in recent years into 2 morphologically identical species: the apathogenic E. dispar and the pathogenic E. histolytica, only the latter being pathogenic. Although various laboratory techniques allow discrimination between the 2 species there is a lack of field data about the suitability of available diagnostic tests for use in epidemiological studies and few epidemiological studies using species-specific diagnosis have been performed at community level in endemic areas, especially in sub-Saharan Africa. We conducted a repeated cross-sectional study of 967 schoolchildren in central Côte d'Ivoire to compare and evaluate light microscopy, 2 different antigen detection assays, and one polymerase chain reaction (PCR) assay. Microscopy and a non-specific antigen capture Entamoeba enzyme-linked immunosorbent assay (ELISA) were used for the primary screening of all children (time t0). The prevalence of the E. histolytica/E. dispar species complex at t0 was 18.8% by single microscopical examination and 31.4% using the non-specific ELISA. Approximately 2 months after the initial screening, fresh stool specimens were collected on 2 consecutive days (t1 and t2) from (i) all the children who were positive by microscopy at t0 (n = 182) and (ii) 155 randomly selected children who were negative at the primary screening. These samples were tested with a second antigen detection ELISA specific for E. histolytica (n = 238) and with a species-specific PCR assay (n = 193). The second and third examinations (t1 and t2) revealed an additional 43 infections with the species complex E. histolytica/E. dispar, so that the cumulative microscopical prevalence for t1 and t2 was 27.7%. The overall prevalence of E. histolytica by species-specific ELISA antigen detection was low (0.83%), while the prevalence of E. dispar was 15%. When analysing only microscopically positive samples by PCR (n = 129), the ratio E. histolytica: E. dispar was very low (1:46), suggesting that the vast majority of Entamoeba infections in this area were apathogenic. Both species-specific tests performed well but the ELISA was easier to use for large-scale field screening.

Assessing seasonal variations and age patterns in mortality during the first year of life in Tanzania.

Lack of birth and death registries in most of developing countries, particularly those in sub-Saharan Africa led to the establishment of Demographic Surveillance Systems (DSS) sites which monitor large population cohorts within defined geographical areas. DSS collects longitudinal data on migration, births, deaths and their causes via verbal autopsies. DSS data provide an opportunity to monitor many health indicators including mortality trends. Mortality rates in Sub-Sahara Africa show seasonal patterns due to high infant and child malaria-related mortality which is influenced by seasonal features present in environmental and climatic factors. However, it is unclear whether seasonal patterns differ by age in the first few months of life. This study provides an overview of approaches to assess, capture and detect seasonality peaks and patterns in mortality using the infant mortality data from the Rufiji DSS, Tanzania. Seasonality was best captured using Bayesian negative binomial models with time and cycle dependent seasonal parameters and autoregressive temporal error terms. Seasonal patterns are similar among different age groups during infancy and timing of their mortality peaks do not differ. Seasonality in mortality rates with two peaks per year is pronounced which corresponds to rainy seasons. Understanding of these trends is important for public health preparedness.

Microscopic diagnosis of sodium acetate-acetic acid-formalin-fixed stool samples for helminths and intestinal protozoa: a comparison among European reference laboratories.

The present study aimed to compare the diagnostic performance of different European reference laboratories in diagnosing helminths and intestinal protozoa, using an ether-concentration method applied to sodium acetate-acetic acid-formalin (SAF)-preserved faecal samples. In total, 102 stool specimens were analysed during a cross-sectional parasitological survey in urban farming communities in Côte d'Ivoire. Five SAF-preserved faecal samples were prepared from each specimen and forwarded to the participating reference laboratories, processed and examined under a microscope adhering to a standard operating procedure (SOP). Schistosoma mansoni (cumulative prevalence: 51.0%) and hookworm (cumulative prevalence: 39.2%) were the predominant helminths. There was excellent agreement (kappa > 0.8; p < 0.001) among the reference laboratories for the diagnosis of S. mansoni, hookworm, Trichuris trichiura and Ascaris lumbricoides. Moderate agreement (kappa = 0.54) was found for Hymenolepis nana, and lesser agreement was observed for other, less prevalent helminths. The predominant intestinal protozoa were Entamoeba coli (median prevalence: 67.6%), Blastocystis hominis (median prevalence: 55.9%) and Entamoeba histolytica/Entamoeba dispar (median prevalence: 47.1%). Substantial agreement among reference laboratories was found for E. coli (kappa = 0.69), but only fair or moderate agreement was found for other Entamoeba species, Giardia intestinalis and Chilomastix mesnili. There was only poor agreement for B. hominis, Isospora belli and Trichomonas intestinalis. In conclusion, although common helminths were reliably diagnosed by European reference laboratories, there was only moderate agreement between centres for pathogenic intestinal protozoa. Continued external quality assessment and the establishment of a formal network of reference laboratories is necessary to further enhance both accuracy and uniformity in parasite diagnosis.

The spatial distribution of Anopheles gambiae sensu stricto and An. arabiensis (Diptera: Culicidae) in Mali.

Variations in the biology and ecology and the high level of genetic polymorphism of malaria vectors in Africa highlight the value of mapping their spatial distribution to enhance successful implementation of integrated vector management. The objective of this study was to collate data on the relative frequencies of Anopheles gambiae s.s. and An. arabiensis mosquitoes in Mali, to assess their association with climate and environmental covariates, and to produce maps of their spatial distribution. Bayesian geostatistical logistic regression models were fitted to identify environmental determinants of the relative frequencies of An. gambiae s.s. and An. arabiensis species and to produce smooth maps of their geographical distribution. The frequency of An. arabiensis was positively associated with the normalized difference vegetation index, the soil water storage index, the maximum temperature and the distance to water bodies. It was negatively associated with the minimum temperature and rainfall. The predicted map suggests that, in West Africa, An. arabiensis is concentrated in the drier savannah areas, while An. gambiae s.s. prefers the southern savannah and land along the rivers, particularly the inner delta of Niger. Because the insecticide knockdown resistance (kdr) gene is reported only in An. gambiae s.s. in Mali, the maps provide valuable information for vector control. They may also be useful for planning future implementation of malaria control by genetically manipulated mosquitoes.

Malaria mapping using transmission models: application to survey data from Mali.

Geographic mapping of the distribution of malaria is complicated by the limitations of the available data. The most widely available data are from prevalence surveys, but these surveys are generally carried out at arbitrary locations and include nonstandardized and overlapping age groups. To achieve comparability between different surveys, the authors propose the use of transmission models, particularly the Garki model, to convert heterogeneous age prevalence data to a common scale of estimated entomological inoculation rates, vectorial capacity, or force of infection. They apply this approach to the analysis of survey data from Mali, collected in 1965-1998, extracted from the Mapping Malaria Risk in Africa database. They use Bayesian geostatistical models to produce smooth maps of estimates of the entomological inoculation rates obtained from the Garki model, allowing for the effect of environmental covariates. Again using the Garki model, they convert kriged entomological inoculation rates values to age-specific malaria prevalence. The approach makes more efficient use of the available data than do previous malaria mapping methods, and it produces highly plausible maps of malaria distribution.