In malaria endemic regions people are commonly infected with multiple species of malaria parasites but the clinical impact of these Plasmodium co-infections is unclear. Differences in transmission seasonality and transmission intensity between endemic regions have been suggested as important factors in determining the effect of multiple species co-infections.
In order to investigate the impact of multiple-species infections on clinical measures of malaria we carried out a cross-sectional community survey in Malawi, in 2002. We collected clinical and parasitological data from 2918 participants aged >6 months, and applied a questionnaire to measure malaria morbidity. We examined the effect of transmission seasonality and intensity on fever, history of fever, haemoglobin concentration ([Hb]) and parasite density, by comparing three regions: perennial transmission (PT), high intensity seasonal transmission (HIST) and low intensity seasonal transmission (LIST). These regions were defined using multi-level modelling of PCR prevalence data and spatial and geo-climatic measures. The three Plasmodium species (P. falciparum, P. malariae and P. ovale) were randomly distributed amongst all children but not adults in the LIST and PT regions. Mean parasite density in children was lower in the HIST compared with the other two regions. Mixed species infections had lower mean parasite density compared with single species infections in the PT region. Fever rates were similar between transmission regions and were unaffected by mixed species infections. A history of fever was associated with single species infections but only in the HIST region. Reduced mean [Hb] and increased anaemia was associated with perennial transmission compared to seasonal transmission. Children with mixed species infections had higher [Hb] in the HIST region.
Citation: Bruce MC, Macheso A, Kelly-Hope LA, Nkhoma S, McConnachie A, Molyneux ME (2008) Effect of Transmission Setting and Mixed Species Infections on Clinical Measures of Malaria in Malawi. PLoS ONE 3(7): e2775. https://doi.org/10.1371/journal.pone.0002775
Editor: Colin J. Sutherland, London School of Hygiene & Tropical Medicine, United Kingdom
Received: February 15, 2008; Accepted: June 20, 2008; Published: July 23, 2008
Copyright: © 2008 Bruce et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This research was funded by a Wellcome Trust Junior Biodiversity Fellowship to Marian Bruce, Reference 060446.
Competing interests: The authors have declared that no competing interests exist.
Infection with multiple Plasmodium species is common in malaria endemic regions. It has been suggested that interactions between different species, in conjunction with differences in seasonality and intensity of malaria transmission may underlie variation in the epidemiology and clinical presentation of malaria . Of the four human malaria species, Plasmodium falciparum causes the greatest morbidity and mortality but most malaria endemic regions are co-endemic for some or all of the other three human species: P. malariae, P. vivax and P. ovale. Individually, these species cause less severe morbidity and fewer deaths than P. falciparum but they are commonly found as co-infections with P. falciparum , . The effect of multiple species co-infections on the clinical outcomes of malaria is unclear. Most clinical surveys of malaria focus on P. falciparum without reference to the potential effects of co-infecting species.
Only a few studies have documented the effects of co-infection on uncomplicated clinical malaria –. Prior or co-infection with P. malariae has been implicated in protecting against fevers caused by P. falciparum in children in Cote d'Ivoire  and in reducing the density of asexual stages and preventing fevers in Tanzanian children . These epidemiological studies of P. malariae are supported by experimental data from malaria-therapy infections in humans carried out in the 1940's, in which prior infection with P. malariae reduced fever caused by P. falciparum . Prior P. vivax infection reduced morbidity from subsequent P. falciparum infections in Vanuatu  and in Papua New Guinea, prior infection with P. vivax or P. malariae protected against P. falciparum fevers . In contrast to these reports of the protective clinical effects of co-infection, an adverse effect has been described in at least one report; carriage of multiple species was associated with greater levels of anaemia than single-species infections in Nigerian children . As well as affecting clinical outcome, interactions between co-infecting species can modify the within-host dynamics of co-infecting malaria parasites ,  and alter the transmission potential of human hosts  thereby impacting on the epidemiology of malaria.
In this study we examined the relationship of Plasmodium co-infections and clinical measures of malaria under different epidemiological conditions. We compared two districts of Malawi with perennial (Mangochi) and seasonal (Dedza) malaria transmission and examined the effect of transmission intensity by comparing two distinct transmission regions within Dedza District. We used PCR detection of parasites to examine the prevalence and age-distribution of multiple Plasmodium species in these diverse regions. In Malawi, as in most of sub-Saharan Africa, the most common sympatric malaria combination is P. falciparum, P. malariae and P. ovale. Previous studies using PCR diagnosis have demonstrated that the prevalence of P. malariae and P. ovale are greatly underestimated by microscopy especially when these are present in mixed infections with P. falciparum –. We investigated the associations between mixed-species infections and fever, history of fever, haemoglobin concentrations and Plasmodium densities, to determine how carriage of multiple Plasmodium species relates to clinical outcomes of malaria.
Materials and Methods
Study sites and population
The study was carried out between March and April 2002 in two districts of Malawi: Dedza in Central Region (14°22′South 34°19′East) and Mangochi in Southern Region (14°28′South 35°15′East). Dedza district lies in a mountainous, semi-forested region at an altitude of >1000 m on the Rift Valley escarpment, bordering Mozambique. Mangochi lies 100 km to the east of Dedza, adjacent to the south western shore of Lake Nyasa, at a lower altitude of approximately 500 m. Annual rainfall is approximately 900 mm in both districts, with the rainy season November–March, during which >90% of rain falls. Dedza temperatures range from 10°C to 28°C, humidity 75% in the wet season and 3°C to 25°C, humidity 60% in the dry season. Annual mean temperature minima and maxima in Dedza are 7°C lower than for Mangochi and wet and dry season humidity values are 77% and 58% respectively .
The total population of the two districts at the time of the study was 544,334 for Dedza and 671,102 for Mangochi, within a total Malawi population of 11 million. Outside of the district towns, the populations of both districts reside in dispersed rural settlements where most dwellings are of traditional mud and wood construction with thatched roofs. The predominant occupation in both districts is subsistence farming and in villages adjacent to Lakes Malawi and Malombe, fishing provides additional income. The predominant ethnic group in Dedza is Chewa, who are part of the Nyanja group of Bantu, whilst the population in Mangochi is predominantly of the Yao tribe , .
Malaria transmission in Dedza district is restricted to the wet season. During the dry season in Dedza, temperature falls below the minimum for development of Plasmodium in the vector , . An. arabiensis and An. gambiae s.s. are the major vectors . In contrast, malaria transmission in Mangochi is perennial. Mosquito breeding sites are present throughout the year, owing to the presence of low-lying marshy areas near the lakeshore and the cultivation of rice. The number of infectious bites per person per year has been estimated at 27 but this was made in a year where rainfall in Mangochi was lower than usual . A value of 180 infectious bites per person per year, as measured in other low altitude areas of Malawi, may be more representative . Although transmission is perennial in Mangochi, the transmission intensity varies markedly throughout the year owing to seasonal variation in rainfall, being as much as 6 times greater during the wet season than the dry. The main vectors are An. gambiae s.l. and An. funestus .
Official statistics show that the number of clinical malaria cases and malaria deaths were more than three times greater in Mangochi than in Dedza (annual figures 2004, clinical cases Dedza 116,661, Mangochi 374,995, deaths Dedza 150, Mangochi 498 ). Retrospective analysis of paediatric ward admission books from Dedza and Mangochi district hospitals for January–December 2004 (Mangochi n = 1579; Dedza n = 1165) showed that there were also significant differences in the features of malaria-related admissions between the two districts. Children admitted with malaria in Mangochi were younger (Mangochi, mean age 19.4 months, Dedza, 31.9 months, p<0.0001), had more anaemia diagnoses (Mangochi, 39.8%; Dedza, 22.4%, p<0.0001) and had more blood transfusions (Mangochi, 10.3%; Dedza, 4.3%, p<0.0001) than children in Dedza. Children admitted to hospital for malaria in Dedza were older, had more cerebral malaria diagnoses (Dedza, 3.1%; Mangochi, 1.7%, p = 0.016) and had higher hospital case fatality rates (Dedza, 9.6%; Mangochi, 7.3%, p = 0.036) than children admitted in Mangochi (S. Nkhoma, M. Bruce, M. Molyneux, unpublished observations).
Participating villages were the largest settlements in randomly selected census enumeration areas (each containing approximately 1000 residents) within a 25 km radius of the district hospitals of Dedza and Mangochi (Figure 1). Sixteen and 8 study villages were selected in Dedza and Mangochi districts, respectively.
Enrolment and sample collection
Malaria community surveys were carried out between 8 March and 7 April 2002 in Dedza and Mangochi Districts of Malawi. Consenting individuals aged >6 months from each village were enrolled in the study. Oral consent for village participation was obtained from village elders and individual written consent was obtained from adults after explanation of the aims and methods of the study. Written consent for the participation of children was given by their parents or guardians.
From a single finger prick, thick and thin blood smears and a filter paper blood sample for molecular analysis were prepared and haemoglobin concentration was measured using a portable haemoglobinometer (HemoCue, Sweden) according to the manufacturer's instructions. Axillary temperature was measured with a digital thermometer and a questionnaire was administered to collect the following information from each participant: age (years), sex, occupation, religious denomination, history of fever in previous 2 weeks, medicine taken in the previous 2 weeks, overnight stay away from home in previous 4 weeks, use of bednets and other anti-malarial prevention measures.
Following the analysis of blood smears by microscopy, parasite positive participants with axillary temperature ≥37.5°C, or [Hb]≤8.0 g/dl, or ≥5000 parasites/µl were treated within 24 hours of blood sampling with sulphadoxine-pyrimethamine (SP) and ferrous sulphate, according to government guidelines at the time of the study. Participants with overt signs of severe malaria or with haemoglobin≤5.0 g/dl or parasite density ≥100,000 parasites/µl of blood were taken to the district hospital. We did not follow up the clinical outcomes of treated or hospital referred participants. Ethical approval for the study was granted by the National Health Sciences Research Committee, Ministry of Population and Health, Government of Malawi and Glasgow University Ethics Committee for Non-clinical Research Involving Human Subjects.
Mapping and Remotely Sensed Data
Village location (latitude and longitude) was determined using ground GPS readings (Etrex 12 channel, Garmin Ltd, Kansas, USA) and maps were drawn using ArcInfo 9.1 (Environmental Systems Research Institute, Inc., California, USA). Data on national boundaries, district hospitals, elevation, forested areas, water bodies, marshland and tributaries were obtained from scanned maps , datasets and satellite imagery (NASA/ Earthsat 1 km and Global 2′ Elevation Data, ETOPO2) available from Environmental Systems Research Institute, Inc. (ESRI; Redlands, CA) and public websites (http://earth.google.com/; http://www.geographynetwork.com). As no climate data were available, vegetation cover was used as a proxy of greenness (leaf-top photosynthesis) and precipitation availability . Data were obtained from Normalised Difference Vegetation Index (NDVI) values, extracted from MODIS images at 250 m spatial resolution (NASA. http://modis.gsfc.nasa.gov/). Differences between wet and dry seasons were examined and based on values from images 14–29 September 2001 (dry season), NDVI 30 September–15 October 2001 (dry season), NDVI 6–21 March 2002 (wet season), NDVI 22 March–6 April 2002 (wet season).
All epidemiological and environmental data were imported and geo-referenced in the geographical information system software ArcInfo 9.1. Maps were drawn showing each village location and P. falciparum prevalence in relation to district hospitals and key environmental characteristics. The following geo-climatic variables were available for each village and extracted for statistical analysis: elevation, latitude, longitude, distance from district hospital, mean dry season NDVI, mean wet season NDVI, and difference between mean wet and dry season NDVI. Additional analysis in relation to elevation was undertaken by visualising topographical/contour differences between each village on a 3D wireframe map created in the surface mapping programme Surfer 7.0 (Golden Software Inc., Golden, CO).
Detection of Plasmodium species
Primary malaria diagnosis was carried out within 24 hours of blood sampling by microscopic examination of thick and thin blood smears. Blood smears were fixed with methanol (thin smear only) and stained with giemsa. Smears were initially graded for the presence/absence of Plasmodium and approximate parasite density, then parasite density counts per 200 leukocytes were made from positive smears. Parasite density per µl of blood was calculated assuming a standard leukocyte count of 8000 per µl of blood. In order to assess the reproducibility of microscopy, a random selection of 10% of smears from each district were re-read.
To confirm microscopy results and detect parasites below the limit of microscopy (around 40 parasites/µl of blood) a retrospective, species-specific PCR ,  was carried out using DNA extracted from filter paper blood samples, as previously described . Samples from 3 villages from each district were tested for all 4 human Plasmodium species (P. falciparum, P. vivax, P. malariae and P. ovale) whilst the remainder were tested only for P. falciparum, P. malariae and P. ovale. An estimated parasite density of 10 parasites per µl of blood was assigned per species detected by PCR in those samples that were microscopy negative. [The value of 10 parasites per µl is the approximate parasite detection level of PCR. In microscopy negative samples containing all 3 species detected by PCR in the study area (P. falciparum, P. malariae and P. ovale) the use of this estimate results in an density value below the microscopy sensitivity level of 40 parasites per µl]. To assess the reproducibility of PCR detection a random selection of 5% of samples from each district were re-analysed (from the sample DNA extraction) for the presence of P. falciparum, P. malariae and P. ovale.
Data were analysed using SPSS v10.0 (Chicago, USA). Data for each district were compared using Pearson's Chi squared test for categorical variables and Mann-Whitney U test for continuous variables. Log10 transformation was carried out to normalise parasite density data. Univariate and bivariate regression models (accounting for district and transmission region) were used to determine which variables were associated with the following clinical outcomes measured during community surveys: haemoglobin concentration ([Hb] g/dl), moderate anaemia ([Hb]≤8.0 g/dl) and fever (axillary temperature ≥37.5°C). Linear regression was used for [Hb] and logistic regression for the binary outcomes of anaemia and fever. In the linear regression analyses continuous variables were included as co-variates whilst categorical variables were included as fixed factors.
Mixed effects logistic regression models were fitted for whether or not each individual had PCR-detectable infection, using the statistical software package MIwiN v2.01 (Centre for Multilevel Modelling, University of Bristol, UK). The log odds of being infected was allowed to vary at random between villages and between households within villages, according to a Normal distribution. A series of models was fitted: firstly, without covariates, to estimate the degree of variation between and within villages; secondly, allowing for individual-level fixed effects of sex and age (adult vs. child); thirdly, allowing for village-level fixed effects of elevation above sea level and the mean wet and dry season normalised difference vegetation index (NDVI) in each village; and lastly, allowing for both individual- and village-level fixed effects. Additionally, the final model was refitted allowing for separate random and fixed effects in each district (Dedza or Mangochi). For each model, the variation between villages and between households within villages are reported in terms of the variance in log odds of being infected, and in terms of the estimated odds ratio between units (villages or households) at the 75th and 25th percentiles of the assumed distribution of units. Also reported are the fixed effect estimates, in terms of the associated odds ratio estimates, with 95% confidence intervals. Finally, for the model allowing separate fixed and random effects in each region, p-values are reported for the comparisons of model parameters between villages.
The multiple-kind lottery model was applied to determine if the distribution amongst individuals of Plasmodium species (P. falciparum, P. malariae and P. ovale) detected by PCR, differed from an independent random distribution . The following categories were used for analysis of multi-species data: negative, one species, mixed species (ie >1 species). Data for P. malariae and P. falciparum were also analysed in the absence of the least prevalent species, P. ovale (categories: negative, P. malariae only, P. falciparum only, P. malariae and P. falciparum). We calculated ROE Mixed, ROE Negative or ROE Pm the ratio of observed/expected (and 95% CI) amongst values that deviated from expected mixed, negative or P. malariae samples, as a measure of the magnitude of the deviation from randomness.
In order to evaluate all possible associations between clinical measurements and the presence of multiple Plasmodium infections in three transmission settings it has been necessary to use multiple testing. Unadjusted p values are reported throughout in order not to obscure possible associations and therefore individual p values should be interpreted with this in mind.
Clinical measurements and blood samples for parasitological analysis were collected between March and April 2002, from 2918 participants residing in 16 villages within Dedza District and 8 villages within Mangochi District, Malawi (Figure 1). Demographic and malaria data obtained by questionnaire were available from 2840 out of 2918 (97.3%) participants. Table 1 shows a summary of the characteristics of the two study populations. Overall, a greater proportion of females than males were enrolled in both districts (62.0% and 62.7% for Dedza and Mangochi respectively). This resulted from a deficit in the enrolment of adult males, who were absent from households because sampling took place during daylight hours (Table 1). Amongst children, the sex ratio was equitable between districts (% female children, 52.2% and 51.0% for Dedza and Mangochi respectively, p = 0.660) but the proportion of children in the study was greater in Mangochi (54.8%) than in Dedza (48.6%, Table 1). There was a difference in the religious denomination of the two populations. Dedza was predominantly Christian, whilst the majority faith in Mangochi was Islam, reflecting differences in the geopolitical history in the two regions.
Relationship of malaria prevalence with geographical features
The mean prevalence of malaria parasites (any Plasmodium species) detected by microscopy in the two districts was remarkably similar during our wet season survey, given the variation in geography and difference in malaria transmission in the two areas. In Dedza, where transmission is seasonal, the mean microscopy prevalence was 23.6% whilst in Mangochi where perennial transmission occurs, prevalence was only slightly higher, 27.3%, a difference significant only at the 0.05 level (Table 1). PCR detection of parasites increased prevalence in Dedza by a factor of 2.3 and in Mangochi by 2.8, giving PCR prevalences of 54.4% and 76.7% respectively (Table 1). Reproducibility of PCR detection was 93.8%. Reproducibility of microscopy was 86.4% for presence/absence of parasites and 79.4% for density grading. The variation in prevalence observed between villages was greater in Dedza than in Mangochi when either microscopy (Dedza: range 2.2–58.4%, standard deviation (S.D.) 13.6; Mangochi: range 15.8–42.3%, S.D. 8.4) or PCR detection was used (Dedza range 17.6–90.3, S.D. 22.4; Mangochi range 65.5–84.6, S.D. 7.7). Examination of the spatial distribution of villages in relation to prevalence showed that in Mangochi, increased prevalence was not associated with proximity to the large water bodies of Lake Malawi or Lake Malombe, nor to marshes or rivers (Figure 2a). However, in Dedza high prevalence was closely associated with proximity to marshy, river run-off areas in the west of the study area, whilst lower prevalence was found in eastern villages located further from the marshes (Figure 2b), and at higher elevation in the east of the study area (Figure 3).
Multi-level regression models were used to investigate the relationship between geo-climatic variables and malaria prevalence, allowing for demographic factors (Table 2). Between-village and between-household variations in prevalence were largely unchanged when adjusting for individual-level effects. Allowing for village-level factors accounted for approximately 25% of the between-village variation, due to a negative association with elevation above sea level. However, between-village variation in prevalence (after accounting for demographic and geo-climatic factors) was observed only in Dedza region, with villages in Mangochi being far more homogeneous. Between-household variation was similar in both regions. There were differences between districts in the effect of elevation and infection probability (p<0.0001), with the odds of infection reducing by about a half for each 100 m elevation in Dedza, but with no corresponding association found in Mangochi. We used remotely sensed normalized difference vegetation index (NDVI) data as a proxy measurement of vegetation greenness and rainfall as this is linked with malaria intensity in some settings in Africa . The mean wet- and dry-season NDVI values for each village showed no association with infection rates.
The close association of prevalence with elevation and the spatial relationship of prevalence to geographical features that were likely to provide mosquito breeding sites, led us to conclude that in Dedza, prevalence was an indicator of variation in levels of transmission during the wet season. In order to determine the effect of variation in transmission intensity and transmission seasonality on the distribution and clinical effect of mixed infections, the study population was divided into 3 transmission regions: 1) low intensity seasonal transmission (LIST) (eastern, elevated Dedza villages, n = 10, PCR prevalence = 43.0, 95% CI 40.4–45.6); 2) high intensity seasonal transmission (HIST)(western, low lying Dedza villages, n = 5, PCR prevalence = 84.1, 95% CI 81.6–86.7), 3) perennial transmission (PT) (all Mangochi villages, n = 8, PCR prevalence = 76.7, 95% CI 74.1–79.2, Table 3).
Age- and geographical-related variance in prevalence of Plasmodium species
Three out of four human malaria species (P. falciparum, P. malariae and P. ovale) were detected by PCR in both districts (Table S1). P. vivax was not detected in either district. The species diagnostic results from replicate microscopy readings were not reliable and therefore only PCR speciation was used. Microscopy density readings were therefore considered as total Plasmodium density. As with most other malaria endemic regions, P. falciparum predominated in both districts (Dedza, 53.2%; Mangochi 75.7%), whilst P. malariae (Dedza, 7.7%; Mangochi 13.2%) and P. ovale (Dedza, 3.3%; Mangochi 7.8%) were observed less frequently and usually only in combination with P. falciparum (Table S1).
For all species, prevalence was greatest in Mangochi. As seen with total Plasmodium prevalence, the relative abundance of the three Plasmodium species showed striking variation between villages in Dedza but was less variable in Mangochi (Figure 4). The five Dedza villages in the high intensity transmission region had P. falciparum prevalence values within the range of those for Mangochi villages, and had a correspondingly high prevalence of P. malariae (Figure 4a) and P. ovale infections (Figure 4b). One Mangochi village (Matenganya) showed more than double the P. ovale prevalence of the other Mangochi villages (Figure 4b). The age-prevalence profiles of each of the Plasmodium species show marked differences between the two districts (Figure 5a and b). P. falciparum prevalence in Dedza increased with age from approximately 35–60% in the years 1–4, thereafter maintaining a stable level of between 50 and 60% (Figure 5a). In contrast, P. falciparum prevalence in Mangochi was around 90% in infants and children and declined with age during adolescence, reaching similar levels amongst adults to those found in Dedza. P. malariae and P. ovale age profiles are similar in Dedza, showing a slight but not significant rise in children aged 5–14 years. In Mangochi, the two species also gave similar patterns but with a pronounced peak amongst children aged 5–9 years (Figure 5b). When the Dedza data were separated into LIST and HIST regions, significant differences were observed between the regions in the age-prevalence pattern, with the profile for the LIST region showing a peak-shift towards older individuals for P. malariae and P. ovale infection and the HIST region profile closely resembling that of Mangochi (Figure 5c and d).
Names are shown for the 5 Dedza villages in the HIST region of Dedza and 1 Mangochi village (Matenganya) with the highest P. ovale prevalence, for ease of comparison.
Total populations of Dedza district (A), Mangochi district (B), and for Dedza district sub-divided by transmission region, LIST (C) and HIST (D). Data for each Plasmodium species are as follows: P. falciparum (orange), P. malariae (blue) and P. ovale (pink). Error bars show 95% confidence intervals.
Distribution of multiple species infections
To determine if the three Plasmodium species were randomly distributed within human hosts we used the multiple-kind lottery model  which detects deviations from a random distribution of multiple parasite species within multiple hosts. Because of the significant age and spatial differences in prevalence, data from children and adults were analysed separately within each of the three transmission regions. No deviation from a random distribution was seen in children from any transmission region when all species (P. falciparum, P. malariae and P. ovale) were considered or when P. falciparum and P. malariae were considered in isolation from the least prevalent species, P. ovale (Table 4). A greater number of mixed infections in the LIST region and fewer P. malariae infections in Mangochi were observed amongst adults compared with that expected from a random distribution of each species (Table 4). Combining data from children and adults led to the abolition of randomness in all of the regions where randomness had been found in children, indicating that differences in the number and distribution of infections between age groups can confound analyses of whole populations. In addition, combining data for the two Dedza transmission regions also resulted in apparent deviation from randomness amongst children.
Generally, parasite density is closely linked with malaria fever, with higher densities more likely to result in elevated temperature. Parasite density measured by microscopy was greater in Mangochi (mean log10 density per µl of blood, 3.2) than in Dedza, (mean log10 density, 3.0, Table 1). When estimates of the density of sub-microscopic parasites, detected only by PCR were included, this difference was abolished (Table 1). In all 3 transmission regions mean log10 parasite density was greater in children compared with adults (LIST region, p<0.0001; HIST region, p = 0.027; Mangochi PT region, p<0.0001, Figure 6a). Amongst adults there was no difference in parasite density between transmission regions but in children, density was lower in the HIST region compared with both the LIST (p = 0.0060) and the Mangochi PT regions (p<0.0001, Figure 6a). Children with mixed species infections detected by PCR, had lower mean parasite density than those with single infections in the PT region (p = 0.0021) but not in the HIST region (p = 0.071, Figure 6b). Children from the LIST region and adults from all regions showed no difference in parasite density between single or mixed infections but in all these comparisons the number of individuals with mixed infections was less than 10, limiting the strength of these comparisons.
We assessed fever as the principal clinical feature of malaria, by measuring temperature at the time of blood sample collection and we also estimated the occurrence of fever during the previous 2 weeks using a questionnaire. In order to prevent bias in comparisons across regions and between single and mixed infections, we chose not to define regional- and age-specific clinical thresholds of fever and parasite density , , instead comparing fever alone and fever along with standard density thresholds.
Despite the differences in Plasmodium prevalence and species distribution between districts, the proportion of individuals with measured fever was very similar (Dedza, 11.5%, Mangochi, 11.9%, Table 1). Likewise, there was no difference in fever rates between the HIST and LIST transmission regions (LIST, 11.6%; HIST, 11.1%, p = 0.770). In both districts fever rates were higher in children than in adults (Dedza: children, 14.0%; adults 9.1%, OR 1.64 (95% CI 1.23–2.17) p = 0.0005); Mangochi: children 14.3%, adults 9.0%, OR 1.69 (95% CI 1.11–2.56), p = 0.013). However, there was no difference between districts or transmission regions in the child or adult fever rates (data not shown).
Fever was seen in 209 out of 1426 (14.6%) people with P. falciparum only, compared with 30 out of 203 (14.8%) people with mixed P. falciparum & P. malariae infections and 8 out of 79 (10.1%) people with mixed P. falciparum and P. ovale infections. These data showed no deviation from expected fever rates in either district when data from all age groups or children (aged <1–14 years) only were analysed (data not shown). There were insufficient data from each of the two Dedza transmission regions for these to be analysed separately. No significant deviations from expected fever rates amongst P. falciparum only and mixed-species infections were seen when clinical malaria was defined more stringently as temperature ≥37.5°C or 38.0°C along with a positive microscopy reading or along with parasite density ≥1000 parasites/µl (data not shown). Only when clinical malaria was defined more stringently as temperature ≥37.5°C along with parasite density ≥5000 parasites/µl, was there a negative association of mixed infections with clinical malaria, and this was seen in children only (OR 0.43 (95% CI 0.19–0.96) p = 0.035).
Univariate logistic regression was used to investigate the variables associated with measured fever. The variables district and transmission region were not significantly associated but 14 out of 25 other variables were (Table S2). Multivariate logistic regression modelling revealed that the best predictor of fever was log10 parasite density in the absence of all other factors but this model accounted only for 8.2% of the variability in fever (Table 5).
A history of fever during the 2 weeks previous to the community survey was self-reported by more than 40% of people in both districts (Dedza, 45.8%, Mangochi, 42.5%, Table 1). The HIST region showed a slightly greater rate of history of fever (50.6%) than both the LIST (43.9%) or PT regions (42.5%) (HIST cf LIST, OR 1.31 (95% CI 1.07–1.60), p = 0.0076; HIST cf PT, OR 1.39 (95% CI 1.12–1.72), p = 0.0028). The LIST region showed a higher rate of history of fever in adults (47.5%) compared with children (40.2%), (OR 1.34 (95% CI 1.09–1.66), p = 0.0057) whereas no difference between age groups was seen in the other transmission regions (data not shown). People reporting a history of fever were less likely to have mixed species infection than expected by chance in the HIST region (OR 0.63 (95% CI 0.42–0.94), p = 0.025) but this was not the case in the LIST or Mangochi regions (data not shown). Almost one quarter of the variability in the history of fever (23.0%) could be predicted using multivariate logistic regression modelling (Table 5). This is in contrast to measured fever, which was poorly predicted by such analysis.
Antimalarial treatment, drug and bednet use
During the community surveys, we gave antimalarial treatment (see Materials and Methods for clinical treatment criteria) to a greater proportion of people in Mangochi (26.2%) than in Dedza (18.3%, Table 1). Children were more likely than adults to meet treatment criteria in both districts (Dedza: children 24.0%, adults 13.0%; Mangochi: children 33.6%, adults 17.3%) and both children and adults in Mangochi were more likely to require treatment than equivalent individuals in Dedza (children, OR 1.60 (95% CI 1.27–2.03), p = <0.0001; adults, OR 1.40 (95% CI 1.03–1.91), p = 0.033). There was no association between meeting treatment criteria and carriage of multiple species infection; individuals who were treated had no more or less mixed species infections than expected by chance alone when data were analysed by district, transmission region or age (data not shown).
Drug use by each participant in the two weeks previous to the community survey was estimated via a questionnaire. There was no difference between districts in the proportion of people that had taken the anti-malarial drugs SP or quinine (Dedza 7.4%, Mangochi 8.9%) but more people in Mangochi had taken painkillers (Dedza 21.6%, Mangochi 26.7%, Table 1). In both Dedza and Mangochi districts, people who had taken antimalarial treatment in the previous 2 weeks were less likely to have a mixed species infection (Dedza, OR 0.10 (95% CI 0.02–0.40), p<0.0001; Mangochi, OR 0.31 (95% CI 0.11–0.89), p = 0.022).
Bednet usage varied between the two districts. Only 1.4% of individuals in Dedza slept under a bednet regularly, compared with 14.2% in Mangochi (Table 1). In Mangochi, people who slept regularly under a bednet were less likely to have a Plasmodium infection detected by PCR (OR 0.43, 95% CI 0.29–0.65, p<0.0001) and had fewer mixed species infections than expected by chance alone (OR 0.22, 95% CI 0.09–0.55, p = 0.0004). There was a near-significant association between bednet use and reduced current fever rate (OR 0.52, 95% CI 0.26–1.07, p = 0.07) but surprisingly, those using bednets were more likely to have a history of fever than those without (OR 1.82, 95% CI 1.24–2.69, p = 0.002). The low numbers of bednet users prevented similar analyses for Dedza district.
Haemoglobin concentration and anaemia
Overall, mean haemoglobin concentration ([Hb]) in Mangochi (10.6 g/dl) was significantly lower than Dedza (12.0 g/dl, Table 1). Correspondingly, levels of anaemia were different in the two districts. Mild anaemia ([Hb]≤11 g/dl) affected 55% of people in Mangochi, but was only half as prevalent in Dedza. Levels of moderate anaemia ([Hb]≤8 g/dl) in Mangochi were three times higher than in Dedza (Table 1). Mean haemoglobin concentration increased with age in both districts, stabilising in adulthood but was consistently lower (by approximately 1.5 g/dl) across all age groups in Mangochi (Figure 7a). When the two Dedza transmission regions were considered separately, a higher mean [Hb] was found in the LIST region compared with the HIST region in children less than 5 years old, but in older children and adults there was no difference (Figure 7b).
Error bars show 95% confidence intervals. Where error bars are absent, the datum point represents a single individual. Note different y-axis scales on E and F.
Mean haemoglobin concentration was lower in parasitaemic individuals than in those without parasitaemia across all age groups in Dedza (Figure 7c). When analysed separately, the LIST region mean [Hb] was reduced by around 1 g/dl in parasitaemic individuals across ages except for the youngest age group (children <1 year, Figure 7e). In contrast, mean [Hb] in parasitaemic people of all ages was indistinguishable from those without parasitaemia in the HIST region (Figure 7f). In Mangochi, there was no difference in mean [Hb] between parasitaemic and non-parasitaemic individuals, except for children aged 1–4 in whom mean [Hb] was around 2 g/dl lower amongst parasitised children (Figure 7d). There was no difference in mean [Hb] in adults with single compared with mixed species infections in all transmission regions and in children in the LIST and PT transmission regions. However, in the HIST region, children with mixed species infections had greater mean [Hb] compared with those carrying a single species (single, 10.7 g/dl, n = 147, mixed 11.4 g/dl, n = 101, p<0.01).
Adults sleeping regularly under a bednet had higher mean [Hb] than those without bednets in Mangochi (mean [Hb] without net, 11.0 g/dl; with net, 11.7 g/dl, p = 0.018). In children, mean [Hb] did not differ with bednet use (mean [Hb] without net, 10.1 g/dl; with net, 10.2 g/dl, p = 0.540). The impact of bednet usage on [Hb] could not be analysed within Dedza district because so few people in this district used bednets (n = 28).
In order to determine associations between haemoglobin concentration and any of the variables that we measured, regression analyses were carried out. In a univariate regression analysis of [Hb], transmission region was strongly associated (R2 = 0.094, p<0.0001) therefore we carried out bi-variate regression analyses accounting for transmission region. All but 3 of the 24 measured variables showed significant association with haemoglobin level. There were significant interactions between transmission region and several variables (Table S3). Modelling of [Hb] using multivariate linear regression resulted in a best-fit model that explained 37.2% of the variability in [Hb] (Table 6).
Transmission region was also strongly associated with the binary clinical outcome of moderate anaemia ([Hb]≤8.0 g/dl) in univariate logistic regression (p<0.0001) and therefore we carried out bi-variate logistic regression analysis accounting for transmission region. When transmission region was included, 15 out of 24 measured variables were associated with anaemia (Table S3). Modelling of moderate anaemia using multivariate logistic regression resulted in a best-fit model that accounted for one third of variability in anaemia outcome (Table 5).
In this study we have investigated the impact of infection with multiple species of Plasmodium, on the clinical manifestations of malaria in different epidemiological settings. We described the epidemiology of multiple Plasmodium infections in two districts of Malawi with seasonal (Dedza) and perennial (Mangochi) malaria transmission. Within the seasonal transmission zone, we have explored the relationship of malaria prevalence with geophysical features and defined two regions in which prevalence varies markedly, almost certainly arising from elevation-linked variation in transmission intensity.
There were marked differences in the spatial and age-distribution of the three sympatric Plasmodium species detected (P. falciparum, P. malariae and P. ovale) between the three transmission regions. Prevalences of parasitaemia for each species were higher and peaked at a younger age in areas with high intensity seasonal or perennial transmission. In Dedza, high parasite prevalences were associated with low altitude and proximity to marshy ground. Higher P. falciparum prevalence was generally reflected in higher prevalence of minority species (P. malariae and P. ovale) raising the possibility of multiple Plasmodium species being co-transmitted. The relative abundance of these species showed micro-geographic variation. This could result from variation in the presence of mosquito species or sub-species, which may have Plasmodium species-specific transmission efficiencies. Alternatively, these differences could arise from stochastic variation in the seeding of infections by incoming human carriers within villages. The human populations in our two study areas come from different ethnic groups, the Chewa and Yao. Different ethnicity or religion can affect housing type, sleeping arrangements and bednet use – factors which may impact on mosquito exposure. However, we know of no evidence to suggest that these groups differ in their genetic predisposition or ability to raise immune responses to malaria infections.
This study is the first to describe the prevalence of P. ovale in Malawi, although its presence has been noted previously . P. vivax was not detected in our study populations, despite its presence in neighbouring Mozambique  and the potential for importation via Malawi's trade links . The distribution of the three Plasmodium species amongst the human population showed randomness amongst children in all regions but was not random in adults from the PT and LIST regions. A non-random distribution is thought to result from the effects of acquired immunity. Our data contrast with those from Papua New Guinea, a holoendemic region where all four co-endemic human malaria species show random distribution across all age groups .
Fever rates were remarkably similar between the three transmission regions despite differences in the seasonality and intensity of transmission and the age-prevalence profiles of each Plasmodium species. The lack of difference could not be explained by differences in anti-malarial drug use, which was similar between districts. People with a history of fever were more likely to have a single infection than those without but they were also more likely to have taken anti-malarial drugs. In 1998, SP (the national first line anti-malarial during study) had a 2-week efficacy of 80% in symptomatic P. falciparum cases in Blantyre (170 km from our study sites) . However, 4-week parasitological failure rates were ∼30–60%, which may account for our observation that many individuals were parasitaemic despite recent self-treatment. In both Dedza and Mangochi, people who had taken antimalarial treatment in the previous 2 weeks were less likely to have a mixed species infection, suggesting that treatment was effective at least against P. malariae and P. ovale co-infections, if not always against P. falciparum. In vivo drug resistance of P. malariae has been described only for chloroquine  and genetic and in vitro studies failed to find evidence of anti-folate resistance in this species in Africa . Differences in drug sensitivity resulting from variation in drug selection pressure and/or mutability between Plasmodium species could explain these data.
Fever was equally likely in people with single or mixed infections. Only when clinical malaria was defined as fever ≥37.5°C with parasite density ≥5000 parasites/µl were there fewer mixed infections than expected, amongst children only. Parasite density was lower amongst children with mixed infections than those with single-species infections, but only in the PT region. This phenomenon may result from a greater level of acquired immunity under perennial transmission, resulting from the earlier age at infection and greater number of infections, acting to reduce density in more heavily infected children. Changes to the immunological control of parasite density and tolerance under different transmission settings can account for differences in P. falciparum morbidity  and similar mechanisms for each species may reduce all species densities. A reduction in parasite density in children with mixed-species infections described in Tanzania  was accompanied by a reduction in fevers, but we found no reduction in fevers in our study. It is possible that lowering of the fever rate (resulting from a reduction in parasite density) in Mangochi was negated by the effect of greater or sustained transmission. The absence of a protective effect of multiple infection against fever, except amongst children with the highest parasite densities, is also in contrast to previous findings from Cote d'Ivoire where children with fevers had fewer mixed P. falciparum-P. malariae infections than asymptomatic children .
A possible explanation for these differences is that in children with very high P. falciparum densities, the detection of lower density species by PCR is masked. This phenomenon is known to occur when detecting multiple P. falciparum genotypes using common primers . The nested species-specific PCR of Snounou et al.,which we employed in this study, also uses common primers in the first round of amplification. This problem could be overcome by designing new PCR methods for malaria diagnosis in which each species was detected independently, without any amplification primers in common. This “common primer” explanation is given weight by the results of our analysis using different density thresholds and the less compelling results given in the Cote d'Ivoire study for children selected on less stringent clinical criteria, who may have had lower parasite densities. Differences in case-definitions of clinical malaria may underlie a similar discrepancy with data from Papua New Guinea .
Anaemia has long been recognised as a complication of malaria, but only recently have studies have confirmed the causal link at the community level , . Lowered [Hb] in malaria infection is due to the destruction of infected red blood cells during asexual multiplication, immune-mediated destruction of non-infected red blood cells  and dyserythropoiesis . In our study, people living under perennial malaria transmission (Mangochi) had a lower mean [Hb] across all age groups than those in the seasonal transmission district (Dedza), and had a greater incidence of anaemia. The lowered [Hb] in Mangochi is probably due to perpetual malaria challenge, although from this study we cannot exclude alternative and unrelated explanations related to the multifactorial etiology of anaemia. Previous studies have determined the effect of micro-nutrient deficiencies, haemoglobinopathies and G6PD variants in Malawi . Cultural differences between the Chewa and Yao can affect height-for-age in children under 5 years in a complex interaction with other factors and there are differences in the levels of growth stunting in these two ethnic groups, suggestive of differences in nutrition . HIV and helminth infections are also known to be important factors in anaemia. There is little difference in the prevalence of soil transmitted helminths between the two study districts but there are higher levels of schistosomiasis in lakeside compared with highland areas . There is a large difference in the prevalence of HIV, which is twice as high in Mangochi as in Dedza, according to official statistics .
An association of anaemia with perennial malaria transmission  is consonant with a greater proportion of severe anaemia and greater rate of blood transfusion amongst paediatric admissions in Mangochi compared with Dedza (unpublished observations). It is likely that persistent malaria pressure resulting in lowered [Hb] puts children at risk of life-threatening anaemia, especially when compounded with other factors such as micronutrient deficiency. The presence of malaria parasites was a useful indicator of [Hb] at the individual level in the LIST region but not in the HIST or PT regions. The apparent lack of effect of parasitaemia on [Hb] in the higher transmission regions (other than in young children) is likely to be because detection of parasites in cross-sectional sampling is a poor predictor of parasitaemia status in older children and adults, in whom low density, chronic infections are not always detectable –.
The higher parasite density found in children compared to adults in all transmission regions may account for the lower mean [Hb] in children relative to adults. Children with mixed species infections had on average greater [Hb] than children with single infections, suggestive of a protective effect of multiple infections. However, this finding was only seen in the HIST region. The functional mechanism for this may be specific to immunity generated through seasonal, compared to perennial infection. The counter-intuitive, lower parasite density found in mixed species infections could result from the inherent bias in the PCR species diagnosis, in which mixed infections may be more likely to be detected in patients where the density of each species is equitable. P. malariae and P. ovale produce lower density parasitaemia than P. falciparum and therefore mixed infections with P. falciparum may be more likely to be detected at lower overall densities.
The use of insecticide treated bednets is now a central focus of anti-malarial campaigns . Bednet use at the time of our study in 2002 was very low although this increased considerably in subsequent years . Bednet users in Mangochi had fewer mixed species infections than those not using nets, in keeping with previous data indicating the role of bednets in preventing new infections . Use of a bednet was associated with a higher mean [Hb] in adults but surprisingly not in children in Mangochi, despite this effect having been described in children in other studies in Malawi  and other African countries . Bednet use gave no protection against fever and those with bednets were more likely to have a history of fever than those without, in contradiction to the multiple studies showing the clinical benefit of bednets against malaria. These results may be due to poor power in this study resulting from the limited number of bednet users.
This study took place during the peak of malaria transmission in the wet season. Dry season data would allow a greater understanding of the differences between the perennial and seasonal transmission regions. It would be useful to clarify the relationship between the HIST and LIST regions to establish the dynamics of carriage of each species during the dry season in these populations. An understanding of how the prevalence of each Plasmodium species decays during the dry season, and the corresponding effect on [Hb] and fever would give valuable insights into the dynamics of the relationships between these factors. Our study suggests that the interaction of multiple Plasmodium species can have protective effects against some clinical outcomes of malaria. This protection appears to be highly dependent on the level and seasonality of malaria transmission and most likely results from differences in the dynamics of acquired immunity. Our data will be informative in predicting the effects of anti-malarial intervention strategies on the clinical outcomes of malaria and on Plasmodium population dynamics across multiple transmission settings.
The number of samples in which each combination of Plasmodium species was detected by PCR for each of the 3 transmission regions.
(0.05 MB DOC)
Univariate logistic regression analysis on fever (axillary temperature ≥37.5°C).
(0.07 MB DOC)
Our thanks go to the people of Dedza and Mangochi Districts who enthusiastically agreed to participate in the study. We are grateful to the Ministry of Health and Population of the Government of Malawi for the provision of vehicles and drivers and to the District Health Officers of Dedza and Mangochi District hospitals for the use of laboratory and office accommodation during the study. We are greatly indebted to the nurses, laboratory staff and environmental officers of Dedza and Mangochi Districts who carried out the community surveys; we thank them for their hard work and dedication. We are also indebted to Keith Scott, Liz Peat and Fiona McMonagle for excellent technical support in Glasgow. We thank Pietro N. Ceccato from The International Research Institute for Climate and Society, The Earth Institute, Columbia University, USA for downloading MODIS images and extracting NDVI values from the satellite data and we are grateful to Andy Tait, University of Glasgow and Phil McCall, University of Liverpool, for useful discussions.
Conceived and designed the experiments: MB AM MEM. Performed the experiments: MB AM SN. Analyzed the data: MB LAKH SN AM. Contributed reagents/materials/analysis tools: MB MEM. Wrote the paper: MB LAKH AM MEM.
- 1. Ritchie TL (1988) Interactions between malaria parasites infecting the same vertebrate hosts. Parasitology 96: 607–639.TL Ritchie1988Interactions between malaria parasites infecting the same vertebrate hosts.Parasitology96607639
- 2. Collins WE, Jeffery GM (2007) Plasmodium malariae: parasite and disease. Clin Microbiol Rev 20: 579–592.WE CollinsGM Jeffery2007Plasmodium malariae: parasite and disease.Clin Microbiol Rev20579592
- 3. Mueller I, Zimmerman PA, Reeder JC (2007) Plasmodium malariae and Plasmodium ovale - the “bashful” malaria parasites. Trends Parasitol 23: 278–283.I. MuellerPA ZimmermanJC Reeder2007Plasmodium malariae and Plasmodium ovale - the “bashful” malaria parasites.Trends Parasitol23278283
- 4. Black J, Hommel M, Snounou G, Pinder M (1994) Mixed infections with Plasmodium falciparum and P. malariae and fever in malaria. Lancet 343: 1095.J. BlackM. HommelG. SnounouM. Pinder1994Mixed infections with Plasmodium falciparum and P. malariae and fever in malaria.Lancet3431095
- 5. Alifrangis M, Lemnge MM, Moon R, Theisen M, Bygbjerg I, et al. (1999) IgG reactivities against recombinant Rhoptry-Associated Protein-1 (rRAP-1) are associated with mixed Plasmodium infections and protection against disease in Tanzanian children. Parasitology 119: 337–342.M. AlifrangisMM LemngeR. MoonM. TheisenI. Bygbjerg1999IgG reactivities against recombinant Rhoptry-Associated Protein-1 (rRAP-1) are associated with mixed Plasmodium infections and protection against disease in Tanzanian children.Parasitology119337342
- 6. Collins W, Jeffery G (1999) A retrospective examination of sporozoite- and trophozoite-induced infections with Plasmodium falciparum in patients previously infected with heterologous species of Plasmodium: effect on development of parasitologic and clinical immunity. Am J Trop Med Hyg 61: 36–43.W. CollinsG. Jeffery1999A retrospective examination of sporozoite- and trophozoite-induced infections with Plasmodium falciparum in patients previously infected with heterologous species of Plasmodium: effect on development of parasitologic and clinical immunity.Am J Trop Med Hyg613643
- 7. Maitland K, Williams TN, Newbold CI (1997) Plasmodium vivax and P. falciparum: Biological interactions and the possibility of cross-species immunity. Parasitol Today 13: 227–231.K. MaitlandTN WilliamsCI Newbold1997Plasmodium vivax and P. falciparum: Biological interactions and the possibility of cross-species immunity.Parasitol Today13227231
- 8. Smith T, Genton B, Baea K, Gibson N, Narara A, et al. (2001) Prospective risk of morbidity in relation to malaria infection in an area of high endemicity of multiple species of Plasmodium. Am J Trop Med Hyg 64: 262–267.T. SmithB. GentonK. BaeaN. GibsonA. Narara2001Prospective risk of morbidity in relation to malaria infection in an area of high endemicity of multiple species of Plasmodium.Am J Trop Med Hyg64262267
- 9. May J, Falusi AG, Mockenhaupt FP, Ademowo OG, Olumese PE, et al. (2000) Impact of subpatent multi-species and multi-clonal plasmodial infections on anaemia in children from Nigeria. Trans R Soc Trop Med Hyg 94: 399–403.J. MayAG FalusiFP MockenhauptOG AdemowoPE Olumese2000Impact of subpatent multi-species and multi-clonal plasmodial infections on anaemia in children from Nigeria.Trans R Soc Trop Med Hyg94399403
- 10. Bruce MC, Day KP (2003) Cross-species regulation of Plasmodium parasitaemia in semi-immune children from Papua New Guinea. Trends Parasitol 19: 271–277.MC BruceKP Day2003Cross-species regulation of Plasmodium parasitaemia in semi-immune children from Papua New Guinea.Trends Parasitol19271277
- 11. Bruce MC, Donnelly CA, Alpers MP, Galinski MR, Barwell JW, et al. (2000) Cross-species interactions between malaria parasites in humans. Science 287: 845–848.MC BruceCA DonnellyMP AlpersMR GalinskiJW Barwell2000Cross-species interactions between malaria parasites in humans.Science287845848
- 12. McKenzie FE, Jeffery GM, Collins WE (2002) Plasmodium malariae infection boosts Plasmodium falciparum gametocyte production. Am J Trop Med Hyg 67: 411–414.FE McKenzieGM JefferyWE Collins2002Plasmodium malariae infection boosts Plasmodium falciparum gametocyte production.Am J Trop Med Hyg67411414
- 13. Kawamoto F, Liu Q, Ferreira M, Tantular I (1999) How Prevalant are Plasmodium ovale and P. malariae in East Asia? Parasitol Today 15: 422–426.F. KawamotoQ. LiuM. FerreiraI. Tantular1999How Prevalant are Plasmodium ovale and P. malariae in East Asia?Parasitol Today15422426
- 14. Mehlotra RK, Lorry K, Kastens W, Miller SM, Alpers MP, et al. (2000) Random distribution of mixed species malaria infections in Papua New Guinea. Am J Trop Med Hyg 62: 225–231.RK MehlotraK. LorryW. KastensSM MillerMP Alpers2000Random distribution of mixed species malaria infections in Papua New Guinea.Am J Trop Med Hyg62225231
- 15. Tobian AA, Mehlotra RK, Malhotra I, Wamachi A, Mungai P, et al. (2000) Frequent umbilical cord-blood and maternal-blood infections with Plasmodium falciparum, P. malariae, and P. ovale in Kenya. J Infect Dis 182: 558–563.AA TobianRK MehlotraI. MalhotraA. WamachiP. Mungai2000Frequent umbilical cord-blood and maternal-blood infections with Plasmodium falciparum, P. malariae, and P. ovale in Kenya.J Infect Dis182558563
- 16. Scopel KK, Fontes CJ, Nunes AC, Horta MF, Braga EM (2004) High prevalence of Plamodium malariae infections in a Brazilian Amazon endemic area (Apiacas-Mato Grosso State) as detected by polymerase chain reaction. Acta Trop 90: 61–64.KK ScopelCJ FontesAC NunesMF HortaEM Braga2004High prevalence of Plamodium malariae infections in a Brazilian Amazon endemic area (Apiacas-Mato Grosso State) as detected by polymerase chain reaction.Acta Trop906164
- 17. The National Atlas of Malawi. Malawi (1984) Department of Surveys, Government of Malawi.The National Atlas of Malawi. Malawi1984Department of Surveys, Government of Malawi.
- 18. Northrup N (1986) The migrations of Yao and Kololo into Southern Malawi: aspects of migrations in nineteenth century Africa. The International Journal of African Historical Studies 19: 59–75.N. Northrup1986The migrations of Yao and Kololo into Southern Malawi: aspects of migrations in nineteenth century Africa.The International Journal of African Historical Studies195975
- 19. Thorold A (1993) Metamorphoses of the Yao Muslims. Thorold, A. In: Brenner L,, editor. Muslim Identity and Social Change in Sub-Saharan Africa. Bloomington: Indiana University Press. pp. 79–90.A. Thorold1993Metamorphoses of the Yao Muslims. Thorold, A.L, BrennerMuslim Identity and Social Change in Sub-Saharan AfricaBloomingtonIndiana University Press7990
- 20. Haworth J (1988) The global distribution of malaria and the present control effort. In: Wernsdorfer WH, McGregor I,, editors. Malaria: principles and practice. London: Churchill Livingstone. pp. 1379–1420.J. Haworth1988The global distribution of malaria and the present control effort.WH WernsdorferI, McGregorMalaria: principles and practiceLondonChurchill Livingstone13791420
- 21. Garnham PCC (1966) Malaria Parasites and Other Haemosporidia. Oxford: Blackwell. PCC Garnham1966Malaria Parasites and Other HaemosporidiaOxfordBlackwell
- 22. Hawley W, Sexton J, Steketee R (1992) Malaria vector assessment, Malawi October 1991–September 1992. USA: Centres for Disease Control. W. HawleyJ. SextonR. Steketee1992Malaria vector assessment, Malawi October 1991–September 1992.USACentres for Disease Control
- 23. Mzilahowa T (2005) Malaria Transmission intensity and the population structure of Plasmodium falciparum in the lower Shire Valley, Southern Malawi.:. T. Mzilahowa2005Malaria Transmission intensity and the population structure of Plasmodium falciparum in the lower Shire Valley, Southern Malawi.:PhD Thesis, University of Liverpool. PhD Thesis, University of Liverpool.
- 24. The Ministry of Health and Population (2004) HIV/AIDS, malaria, TB and under 5 five diarrhoea by number of cases, admissions and deaths by Districts and Regions. The National Statistical Office of Malawi. The Ministry of Health and Population2004HIV/AIDS, malaria, TB and under 5 five diarrhoea by number of cases, admissions and deaths by Districts and Regions.The National Statistical Office of Malawi
- 25. Hay SI, Snow RW, Rogers DJ (1998) Predicting malaria seasons in Kenya using multitemporal meteorological satellite sensor data. Trans R Soc Trop Med Hyg 92: 12–20.SI HayRW SnowDJ Rogers1998Predicting malaria seasons in Kenya using multitemporal meteorological satellite sensor data.Trans R Soc Trop Med Hyg921220
- 26. Snounou G, Viriyakosol S, Zhu XP, Jarra W, Pinheiro L, et al. (1993) High sensitivity of detection of human malaria parasites by the use of nested polymerase chain reaction. Mol Biochem Parasitol 61: 315–320.G. SnounouS. ViriyakosolXP ZhuW. JarraL. Pinheiro1993High sensitivity of detection of human malaria parasites by the use of nested polymerase chain reaction.Mol Biochem Parasitol61315320
- 27. Singh B, Bobogare A, Cox-Singh J, Snounou G, Shukri Abdullah M, et al. (1999) A genus- and species-specific nested polymerase chain reaction malaria detection assay for epidemiologic studies. Am J Trop Med Hyg 60: 687–692.B. SinghA. BobogareJ. Cox-SinghG. SnounouM. Shukri Abdullah1999A genus- and species-specific nested polymerase chain reaction malaria detection assay for epidemiologic studies.Am J Trop Med Hyg60687692
- 28. Bruce MC, Macheso A, Galinski MR, Barnwell JW (2007) Characterization and application of multiple genetic markers for Plasmodium malariae. Parasitology 134: 637–650.MC BruceA. MachesoMR GalinskiJW Barnwell2007Characterization and application of multiple genetic markers for Plasmodium malariae.Parasitology134637650
- 29. Janovy JJ, Clopton RE, Clopton DA, Snyder SD, Efting A, et al. (1995) Species density distributions as null models for ecologically significant interactions of parasite species in an assemblage. Ecol Modell 77: 189–196.JJ JanovyRE CloptonDA CloptonSD SnyderA. Efting1995Species density distributions as null models for ecologically significant interactions of parasite species in an assemblage.Ecol Modell77189196
- 30. Smith T, Schellenberg JA, Hayes R (1994) Attributable fraction estimates and case definitions for malaria in endemic areas. Stat Med 13: 2345–2358.T. SmithJA SchellenbergR. Hayes1994Attributable fraction estimates and case definitions for malaria in endemic areas.Stat Med1323452358
- 31. Smith T, Hurt N, Teuscher T, Tanner M (1995) Is fever a good sign for clinical malaria in surveys of endemic communities? Am J Trop Med Hyg 52: 306–310.T. SmithN. HurtT. TeuscherM. Tanner1995Is fever a good sign for clinical malaria in surveys of endemic communities?Am J Trop Med Hyg52306310
- 32. Richter J, Hoppenheit B, Muller-Stover I, Gobels K, Haussinger D (2004) Case report: Plasmodium ovale infection acquired in Malawi. Trans R Soc Trop Med Hyg 98: 751–752.J. RichterB. HoppenheitI. Muller-StoverK. GobelsD. Haussinger2004Case report: Plasmodium ovale infection acquired in Malawi.Trans R Soc Trop Med Hyg98751752
- 33. Peragallo MS, Sabatinelli G, Majori G, Cali G, Sarnicola G (1997) Prevention and morbidity of malaria in non-immune subjects; a case-control study among Italian troops in Somalia and Mozambique, 1992–1994. Trans R Soc Trop Med Hyg 91: 343–346.MS PeragalloG. SabatinelliG. MajoriG. CaliG. Sarnicola1997Prevention and morbidity of malaria in non-immune subjects; a case-control study among Italian troops in Somalia and Mozambique, 1992–1994.Trans R Soc Trop Med Hyg91343346
- 34. Rosenberg R (2007) Plasmodium vivax in Africa: hidden in plain sight? Trends Parasitol 23: 193–196.R. Rosenberg2007Plasmodium vivax in Africa: hidden in plain sight?Trends Parasitol23193196
- 35. Mehlotra RK, Kasehagen LJ, Baisor M, Lorry K, Kazura JW, et al. (2002) Malaria infections are randomly distributed in diverse holoendemic areas of Papua New Guinea. Am J Trop Med Hyg 67: 555–562.RK MehlotraLJ KasehagenM. BaisorK. LorryJW Kazura2002Malaria infections are randomly distributed in diverse holoendemic areas of Papua New Guinea.Am J Trop Med Hyg67555562
- 36. Plowe CV, Kublin JG, Dzinjalamala FK, Kamwendo DS, Mukadam RA, et al. (2004) Sustained clinical efficacy of sulfadoxine-pyrimethamine for uncomplicated falciparum malaria in Malawi after 10 years as first line treatment: five year prospective study. BMJ 328: 545.CV PloweJG KublinFK DzinjalamalaDS KamwendoRA Mukadam2004Sustained clinical efficacy of sulfadoxine-pyrimethamine for uncomplicated falciparum malaria in Malawi after 10 years as first line treatment: five year prospective study.BMJ328545
- 37. Maguire JD, Sumawinata IW, Masbar S, Laksana B, Prodjodipuro P, et al. (2002) Chloroquine-resistant Plasmodium malariae in south Sumatra, Indonesia. Lancet 360: 58–60.JD MaguireIW SumawinataS. MasbarB. LaksanaP. Prodjodipuro2002Chloroquine-resistant Plasmodium malariae in south Sumatra, Indonesia.Lancet3605860
- 38. Tanomsing N, Imwong M, Pukrittayakamee S, Chotivanich K, Looareesuwan S, et al. (2007) Genetic analysis of the dihydrofolate reductase-thymidylate synthase gene from geographically diverse isolates of Plasmodium malariae. Antimicrob Agents Chemother 51: 3523–3530.N. TanomsingM. ImwongS. PukrittayakameeK. ChotivanichS. Looareesuwan2007Genetic analysis of the dihydrofolate reductase-thymidylate synthase gene from geographically diverse isolates of Plasmodium malariae.Antimicrob Agents Chemother5135233530
- 39. Smith T, Ross A, Maire N, Rogier C, Trape JF, et al. (2006) An epidemiologic model of the incidence of acute illness in Plasmodium falciparum malaria. Am J Trop Med Hyg 75: 56–62.T. SmithA. RossN. MaireC. RogierJF Trape2006An epidemiologic model of the incidence of acute illness in Plasmodium falciparum malaria.Am J Trop Med Hyg755662
- 40. Färnert A, Arez AP, Babiker HA, Beck HP, Benito A, et al. (2001) Genotyping of Plasmodium falciparum infections by PCR: a comparative multicentre study. Trans R Soc Trop Med Hyg 95: 225–232.A. FärnertAP ArezHA BabikerHP BeckA. Benito2001Genotyping of Plasmodium falciparum infections by PCR: a comparative multicentre study.Trans R Soc Trop Med Hyg95225232
- 41. Geerligs PD, Brabin BJ, Eggelte TA (2003) Analysis of the effects of malaria chemoprophylaxis in children on haematological responses, morbidity and mortality. Bull World Health Org 81: 205–216.PD GeerligsBJ BrabinTA Eggelte2003Analysis of the effects of malaria chemoprophylaxis in children on haematological responses, morbidity and mortality.Bull World Health Org81205216
- 42. Korenromp EL, Armstrong-Schellenberg JR, Williams BG, Nahlen BL, Snow RW (2004) Impact of malaria control on childhood anaemia in Africa - a quantitative review. Trop Med Int Health 9: 1050–1065.EL KorenrompJR Armstrong-SchellenbergBG WilliamsBL NahlenRW Snow2004Impact of malaria control on childhood anaemia in Africa - a quantitative review.Trop Med Int Health910501065
- 43. Jakeman GN, Saul A, Hogarth WL, Collins WE (1999) Anaemia of acute malaria infections in non-immune patients primarily results from destruction of uninfected erythrocytes. Parasitology 119: 127–133.GN JakemanA. SaulWL HogarthWE Collins1999Anaemia of acute malaria infections in non-immune patients primarily results from destruction of uninfected erythrocytes.Parasitology119127133
- 44. Phillips RE, Looareesuwan S, Warrell DA, Lee SH, Karbwang J, et al. (1986) The importance of anaemia in cerebral and uncomplicated falciparum malaria: role of complications, dyserythropoiesis and iron sequestration. Q J Med 58: 305–323.RE PhillipsS. LooareesuwanDA WarrellSH LeeJ. Karbwang1986The importance of anaemia in cerebral and uncomplicated falciparum malaria: role of complications, dyserythropoiesis and iron sequestration.Q J Med58305323
- 45. Brabin BJ, Prinsen-Geerligs PD, Verhoeff FH, Fletcher KA, Chimsuku LH, et al. (2004) Haematological profiles of the people of rural southern Malawi: an overview. Ann Trop Med Parasitol 98: 71–83.BJ BrabinPD Prinsen-GeerligsFH VerhoeffKA FletcherLH Chimsuku2004Haematological profiles of the people of rural southern Malawi: an overview.Ann Trop Med Parasitol987183
- 46. Pelletier DL, Msukwa LA (1991) The use of national sample surveys for nutritional surveillance: lessons from Malawi's National Sample Survey of Agriculture. Soc Sci Med 32: 887–898.DL PelletierLA Msukwa1991The use of national sample surveys for nutritional surveillance: lessons from Malawi's National Sample Survey of Agriculture.Soc Sci Med32887898
- 47. Bowie C, Purcell B, Shaba B, Makaula P, Perez M (2004) A national survey of the prevalence of schistosomiasis and soil transmitted helminths in Malawi. BMC Infect Dis 4: 49.C. BowieB. PurcellB. ShabaP. MakaulaM. Perez2004A national survey of the prevalence of schistosomiasis and soil transmitted helminths in Malawi.BMC Infect Dis449
- 48. Carneiro IA, Smith T, Lusingu JP, Malima R, Utzinger J, et al. (2006) Modeling the relationship between the population prevalence of Plasmodium falciparum malaria and anemia. Am J Trop Med Hyg 75: 82–89.IA CarneiroT. SmithJP LusinguR. MalimaJ. Utzinger2006Modeling the relationship between the population prevalence of Plasmodium falciparum malaria and anemia.Am J Trop Med Hyg758289
- 49. Bruce MC, Donnelly CA, Packer M, Lagog M, Gibson N, et al. (2000) Age- and species-specific duration of infection in asymptomatic malaria infections in an endemic population in Papua New Guinea. Parasitology 121: 247–256.MC BruceCA DonnellyM. PackerM. LagogN. Gibson2000Age- and species-specific duration of infection in asymptomatic malaria infections in an endemic population in Papua New Guinea.Parasitology121247256
- 50. Bruce MC, Galinski M, Barnwell JW, Donnelly CA, Walmsley M, et al. (2000) Genetic diversity and dynamics of P. falciparum and P. vivax populations in multiply infected children with asymptomatic malaria infections in Papua New Guinea. Parasitology 121: 257–272.MC BruceM. GalinskiJW BarnwellCA DonnellyM. Walmsley2000Genetic diversity and dynamics of P. falciparum and P. vivax populations in multiply infected children with asymptomatic malaria infections in Papua New Guinea.Parasitology121257272
- 51. Färnert A, Snounou G, Rooth I, Björkman A (1997) Daily dynamics of Plasmodium falciparum subpopulations in asymptomatic children in a holoendemic area. Am J Trop Med Hyg 56: 538–547.A. FärnertG. SnounouI. RoothA. Björkman1997Daily dynamics of Plasmodium falciparum subpopulations in asymptomatic children in a holoendemic area.Am J Trop Med Hyg56538547
- 52. Binka F, Akweongo P (2006) Prevention of malaria using ITNs: potential for achieving the millennium development goals. Curr Mol Med 6: 261–267.F. BinkaP. Akweongo2006Prevention of malaria using ITNs: potential for achieving the millennium development goals.Curr Mol Med6261267
- 53. Stevens W, Wiseman V, Ortiz J, Chavasse D (2005) The costs and effects of a nationwide insecticide-treated net programme: the case of Malawi. Malar J 4: 22.W. StevensV. WisemanJ. OrtizD. Chavasse2005The costs and effects of a nationwide insecticide-treated net programme: the case of Malawi.Malar J422
- 54. Smith T, Felger I, Fraser-Hurt N, Beck HP (1999) Effect of insecticide-treated bed nets on the dynamics of multiple Plasmodium falciparum infections. Trans R Soc Trop Med Hyg 93: Suppl 153–57.T. SmithI. FelgerN. Fraser-HurtHP Beck1999Effect of insecticide-treated bed nets on the dynamics of multiple Plasmodium falciparum infections.Trans R Soc Trop Med Hyg93Suppl 15357
- 55. Mathanga DP, Campbell CH, Taylor TE, Barlow R, Wilson ML (2006) Socially marketed insecticide-treated nets effectively reduce Plasmodium infection and anaemia among children in urban Malawi. Trop Med Int Health 11: 1367–1374.DP MathangaCH CampbellTE TaylorR. BarlowML Wilson2006Socially marketed insecticide-treated nets effectively reduce Plasmodium infection and anaemia among children in urban Malawi.Trop Med Int Health1113671374
- 56. Fraser-Hurt N, Felger I, Edoh D, Steiger S, Mashaka M, et al. (1999) Effect of insecticide-treated bed nets on haemoglobin values, prevalence and multiplicity of infection with Plasmodium falciparum in a randomized controlled trial in Tanzania. Trans R Soc Trop Med Hyg 93: Suppl 147–51.N. Fraser-HurtI. FelgerD. EdohS. SteigerM. Mashaka1999Effect of insecticide-treated bed nets on haemoglobin values, prevalence and multiplicity of infection with Plasmodium falciparum in a randomized controlled trial in Tanzania.Trans R Soc Trop Med Hyg93Suppl 14751