Antagonistic effects of Plasmodium-helminth co-infections on malaria pathology in different population groups in Côte d'Ivoire.

INTRODUCTION
Plasmodium spp. and helminths are co-endemic in many parts of the tropics; hence, co-infection is a common phenomenon. Interactions between Plasmodium and helminth infections may alter the host's immune response and susceptibility and thus impact on morbidity. There is little information on the direction and magnitude of such interactions and results are conflicting. This study aimed at shedding new light on the potential interactions of Plasmodium and helminth co-infections on anemia and splenomegaly in different population groups in Côte d'Ivoire.


METHODOLOGY
Parasitologic and clinical data were obtained from four cross-sectional community-based studies and a national school-based survey conducted between 2011 and 2013 in Côte d'Ivoire. Six scenarios of co-infection pairs defined as Plasmodium infection or high parasitemia, combined with one of three common helminth infections (i.e., Schistosoma mansoni, S. haematobium, and hookworm) served for analysis. Adjusted logistic regression models were built for each scenario and interaction measures on additive scale calculated according to Rothman et al., while an interaction term in the model served as multiplicative scale measure.


PRINCIPAL FINDINGS
All identified significant interactions were of antagonistic nature but varied in magnitude and species combination. In study participants aged 5-18 years from community-based studies, Plasmodium-hookworm co-infection showed an antagonistic interaction on additive scale on splenomegaly, while Plasmodium-Schistosoma co-infection scenarios showed protective effects on multiplicative scale for anemia and splenomegaly in participants aged 5-16 years from a school-based study.


CONCLUSIONS/SIGNIFICANCE
No exacerbation from co-infection with Plasmodium and helminths was observed, neither in participants aged 5-18 years nor in adults from the community-based studies. Future studies should unravel underlying mechanisms of the observed interactions, as this knowledge might help shaping control efforts against these diseases of poverty.


Introduction
Plasmodium spp. and helminths are co-endemic in many parts of the tropics; hence, coinfection is a common phenomenon. Interactions between Plasmodium and helminth infections may alter the host's immune response and susceptibility and thus impact on morbidity. There is little information on the direction and magnitude of such interactions and results are conflicting. This study aimed at shedding new light on the potential interactions of Plasmodium and helminth co-infections on anemia and splenomegaly in different population groups in Cô te d'Ivoire.

Methodology
Parasitologic and clinical data were obtained from four cross-sectional community-based studies and a national school-based survey conducted between 2011 and 2013 in Côte d'Ivoire. Six scenarios of co-infection pairs defined as Plasmodium infection or high parasitemia, combined with one of three common helminth infections (i.e., Schistosoma mansoni, S. haematobium, and hookworm) served for analysis. Adjusted logistic regression models were built for each scenario and interaction measures on additive scale calculated according to Rothman et al., while an interaction term in the model served as multiplicative scale measure.

Principal findings
All identified significant interactions were of antagonistic nature but varied in magnitude and species combination. In study participants aged 5-18 years from community-based studies, Plasmodium-hookworm co-infection showed an antagonistic interaction on additive scale on splenomegaly, while Plasmodium-Schistosoma co-infection scenarios showed protective effects on multiplicative scale for anemia and splenomegaly in participants aged 5-16 years from a school-based study. PLOS  Introduction Anemia is multifactorial. However, in low-and middle-income countries it is largely attributable to parasitic diseases, such as malaria and helminth infections [1]. Yet, the etiology of anemia differs between malaria and helminthiases. Plasmodium spp., the malaria causing parasite, affects iron and hemoglobin (Hb) levels during different stages of its development within the human host. An increased red blood cell (RBC) death related to Plasmodium asexual replication, reduced iron absorption, and decrease in recycling of RBC iron in individuals with inflammation are potential causes of anemia during a malaria episode [2][3][4][5]. Helminth infections are less important with regard to systemic inflammation [5]. However, helminth infections may lead to anemia through blood loss from ingestion of blood cells and lacerations on digestive mucosa by hookworms and through Schistosoma eggs passing through blood vessels and bladder and intestinal tissues that lead to hematuria or blood in stool [6,7]. In addition to anemia, splenomegaly is a frequent clinical manifestation of chronic malaria [4] and advanced Schistosoma mansoni infections [8].
Helminths and Plasmodium spp. are co-endemic in many parts of the tropics. Hence, helminth-Plasmodium co-infection is a common phenomenon [9]. In Côte d'Ivoire, for example, a national school-based survey showed a prevalence of 13.5% for Plasmodium-soil-transmitted helminth co-infection and 5.6% for Plasmodium-Schistosoma co-infection [10]. The impact of such co-infections on the pathophysiology is poorly understood. Effects on the host's health and immune response due to concomitant infection of Plasmodium and helminths are complex and depend on various factors (e.g., stage, intensity, force of transmission, sociodemographic characteristics, immune status, and co-morbidities) [9,11]. To effectively combat Plasmodium infection, a timely and strong pro-inflammatory and type 1 immune response is induced in infected humans. Concomitant chronic helminth infection may, however, alter the immune response by down regulating the Th1 pathway toward anti-inflammatory and type 2 responses [11,12]. This shift may prevent individuals from clinical immunopathology due to inflammation but may, on the other hand, affect malaria parasite clearance [13]. Apart from immunomodulation, other mechanisms such as cohabitation and competition for resources may also influence clinical manifestations in the co-infected host [11]. Potential interactions on pathophysiology, such as anemia and splenomegaly, might be bidirectional, showing antagonism and synergism in co-infected individuals, which might explain conflicting results of recent studies [14]. While several studies conducted in West Africa identified antagonistic effects or reduced odds for malaria-related pathology in Plasmodium-helminth co-infected individuals [15][16][17][18], synergism with potential exacerbation of anemia and splenomegaly was reported from studies in Kenya and Zimbabwe [19][20][21].
Interactions on outcomes in individuals with co-morbidities may be determined on additive or multiplicative scales. Most studies looking at co-morbidities do not look at actual interactions but rather compare estimates from regression analysis between morbidity groups and the few studies exploring actual interactions mainly looked at statistical interaction on multiplicative scale introducing a product term into regression analysis and may thus have missed out potential interaction on additive scale. In clinical research, biologic interactions are often assessed within case-control studies through interaction measures on additive scale, as proposed by Rothman et al. [22]. It is further argued that in etiologic epidemiologic research focus should be put on the assessment of biologic interactions expressed as departure form additivity rather than statistical interaction looking at departure from multiplicativity [23]. The measures proposed by Rothman et al. not only identify the direction but also the magnitude of potential departure from additivity (i.e., antagonism or synergism) and are expressed as synergy index (SI), attributable proportion (AP), and relative excess due to interaction (RERI). In practical use a RERI of 0, an AP of 0 and a SI of 1 would indicate no biologic interaction and exact additivity of the individual effects of Plasmodium and helminth infection on the disease outcome meaning Plasmodium and helminth infections are independent in causing a certain disease outcome. Yet, if the combined effect of Plasmodium and helminth infection is larger (or smaller) than the sum of the individual effects there is interaction on an additive scale or at least departure from additivity that would then translate into a RERI>0, AP>0 and SI>1 or a RERI<0, AP<0 and SI<1, respectively, in case of an antagonistic interaction. This approach has found only limited application in infectious disease epidemiology, although methodologies for cross-sectional designs adjusting for other influencing factors by use of logistic regression analysis have been proposed [24]. With regard to helminth co-infections, a former study revealed significant synergistic effects on anemia among S. japonicum-hookworm co-infected individuals in The Philippines [25]. To date, such interaction measures for helminth and Plasmodium co-infections are lacking. The aim of the current study was to shed new light on the potential interactions on additive and multiplicative scale for Plasmodium and helminth coinfections on anemia and splenomegaly among two population groups (school-aged children/ adolescents and adults) in Côte d'Ivoire.
Recherches Scientifiques en Côte d'Ivoire (CSRS; Abidjan, Côte d'Ivoire). Ethical clearance was received from the ethics committees of Basel (EKBB, reference no. 30/11) and Côte d'Ivoire (reference no. 09-2011/MSHP/CNER-P). In addition, permission for conduct of the national school-based survey was obtained from the Ministry of National Education in Côte d'Ivoire. Directors and teachers of the schools, village chiefs of the study communities, and all concerned district, health, and education authorities were informed about the purpose and procedures of the study. Written informed consent was obtained from each participant, while parents/guardians signed on behalf of children aged below 18 years. Children assented orally. Participation was voluntary; hence, individuals could withdraw from the study at any time without further obligation. All participants benefitted from treatment against soil-transmitted helminths with albendazole (single oral dose of 200 mg for children aged 1-2 years and 400 mg for all participants >2 years) given free of charge. Participants diagnosed with Schistosoma eggs were treated with praziquantel (40 mg/kg). In schools where the prevalence of Schistosoma infection was above 25%, all school children received praziquantel. Individuals with clinical malaria (i.e., tympanic temperature �38˚C and a positive malaria rapid diagnostic test (RDT)) and/or severe anemia were offered artemisinin-based combination therapy together with paracetamol and an anti-anemic treatment, respectively, if the RDT result was negative.

Study design and subjects
Four community-based surveys and a national school-based survey were carried out. Surveys followed a cross-sectional design, including parasitologic and clinical examination and questionnaire interviews for sociodemographic information. The four communities surveyed were situated in southern Côte d'Ivoire and had less than 1,000 inhabitants each with main economic and livelihood activity in subsistence farming and cash crop production (i.e., cacao, coffee, and rubber). The village Sahoua (geographic coordinates: 6˚19'20" N latitude, 5˚10'30" W longitude) and Ancien Carrefour (5˚37'40" N, 4˚01'15" W) were visited in August and September 2011, as described in more detail elsewhere [17]. Two rural communities, namely La Scierie (6˚49'23" N, 3˚24'53" W) and Azaguié CNRA (5˚36'10" N, 4˚00'51" W), were surveyed between May and August 2013. For recruitment, all community members were invited for participation.
For the national school-based study, 94 school localities were visited between November 2011 and February 2012 (i.e., dry season), whereof 93 participated in the survey. In one school the children received deworming shortly before the current study; all analyses were thus based on the results from 92 schools. The localities were selected applying a lattice plus close pairs design [26] and taking into account the ecologic zones, residential area (i.e., a minimum of 20% urban settings), population density, and the presence of a primary school with a minimum of 60 children attending grades 3 to 5 [10,27,28]. In each school, a subsample of 60 children was asked to participate in the study.
Côte d'Ivoire is highly endemic for malaria with perennial transmission of Plasmodium infections [28,29], while infections with soil-transmitted helminths (predominantly hookworm) and Schistosoma spp. are more focally distributed [10]. In all surveyed communities overall Plasmodium prevalence is expected to be high (�60%) and prevalence of soil-transmitted helminths to be of moderate endemicity (range: 10-50%) [17]. Sahoua has been identified as endemic zone for S. haematobium infection, while the two communities situated in the Azaguié district are considered to be S. mansoni endemic [17,30].

Field and laboratory procedures
The study procedures were described elsewhere [10,17,27,28]. In brief, every participant was invited to provide a single stool, urine, and finger-prick blood sample. Duplicate Kato-Katz thick smears [31] were prepared from each stool sample and microscopically examined for the presence of eggs of S. mansoni and soil-transmitted helminths (i.e., Ascaris lumbricoides, hookworm, and Trichuris trichiura). Urine samples were either processed by a filtration technique for detection of S. haematobium eggs or, in case of the national school-based study, using reagent strips (Hemastix; Siemens Healthcare Diagnostics GmbH, Eschborn, Germany) to detect microhematuria [32]. From finger-prick blood samples, thin and thick blood films were prepared on microscope slides. Slides were stained with 10% Giemsa and examined under a microscope (x100) with immersion oil by experienced laboratory technicians. Parasitemia (parasites/μl of blood) of the three endemic Plasmodium species (i.e., P. falciparum, P. malariae, and P. ovale) were recorded [33]. Additionally, finger-prick blood samples were subjected to an RDT for P. falciparum (ICT ML01 Malaria Pf kit; ICT Diagnostics, Cape Town, South Africa). Ten percent of the Kato-Katz thick smears, urine filters, and blood films were subjected to quality control and re-examined by a senior laboratory technician. In case of discrepancies of parasite counts, the slides were re-examined by another senior technician and results discussed until agreement was reached. Clinical examination involved palpation of spleen to assess organ enlargement. Hb was measured using a HemoCue device (HemoCue Hb 301 system; Angelholm, Sweden). Anthropometric measurements, including height and weight, were taken from each participant for subsequent analysis of nutritional status. Sociodemographic information of each participant was collected through administration of pre-tested individual rapid appraisal [34] or a household-based questionnaire [35] for school children and community members, respectively.

Statistical analysis
Data were double-entered and cross-checked using EpiInfo version 3.5.3 (Centers for Disease Control and Prevention; Atlanta, United States of America). Statistical analyses were performed in Stata version 14.0 (Stata Corp.; College Station, United States of America). Data from the community surveys and the national school-based survey were analyzed separately. Within the community data set, we divided the sample into school-aged children/adolescents (5-18 years) and adults (>18 years). All school-aged children/adolescents and adults with complete parasitologic data and Hb measurement were considered for the main analysis and, depending on the outcome, further stratified into participants with anemia and splenomegaly records, respectively.
According to the World Health Organization (WHO), anemia was defined as Hb levels below 11.5 g/dl and 12.0 g/dl in children aged 5-11 years and children aged 12-14 years, respectively, while pregnant women (�15 years), non-pregnant women (�15 years), and males (�15 years) had anemia cut-offs of 11.0, 12.0, and 13.0 g/dl, respectively [36]. Participants having a palpable spleen of grade 1 or higher using a Hackett's scale [37] were considered to have splenomegaly. Malnutrition in school-aged children/adolescents was defined as a Zscore <-2 in any of the calculated indicators (i.e., height-for-age, body mass index (BMI)-forage, or weight-for-age) based on STATA macros from WHO child growth standards [38]. In adults, BMI and mid-upper arm circumference (MUAC) were used to define malnutrition according to cut-offs reported by Eddleston et al. [39].
Schools participating in the national school-based survey were classified into different endemicity profiles to address heterogeneity according to the following criteria: (i) Plasmodium-endemic: any positive case per school from microscopy or RDT; (ii) S. haematobiumendemic: any positive case for microhematuria defined as reagent strip positive with an intensity of �1+; (iii) S. mansoni-endemic: any positive case per school detected by microscopy of Kato-Katz thick smears; and (iv) hookworm-endemic: schools with a minimum hookworm infection prevalence of 10% or any case of moderate-or heavy-intensity infection (�2,000 eggs per gram of stool (EPG)) [7]. Since school-aged children/adolescents from the community-based and the school-based surveys showed very high rates of Plasmodium infection (84.4% and 75.0%, respectively), a two-scenario approach was adopted for analysis of interaction measures of the different Plasmodium-helminth co-infection categories. The first scenario used "presence or absence" of Plasmodium infection, while the second scenario classified participants into "no or lightly-infected" versus participants with high Plasmodium parasitemia. Cut-offs for high parasitemia were set at 1,000 and 1,500 parasites/μl of blood for the schoolbased and the community-based surveys, respectively, taking into account the relationship between Plasmodium parasitemia and prevalence of anemia (S1 Fig). Sample size used for interaction analysis thus varied depending on the number of participants with available information of the two clinical outcomes (i.e., anemia and splenomegaly) and belonging to a school endemic to the investigated pair of parasites.
Univariate logistic regression analysis was applied to calculate crude odds ratios (ORs) for anemia and splenomegaly and different explanatories, including sociodemographic, parasitologic (i.e., (co-)infection status), and clinical indicators. For interaction analysis, multivariable logistic regression models were built for each Plasmodium-helminth pair, including all major helminth species identified from parasitologic examination, namely Plasmodium-S. haematobium, Plasmodium-S. mansoni, and Plasmodium-hookworm co-infection. To adjust for concurrent helminth infections, infection status with soil-transmitted helminths or Schistosoma, respectively, was introduced in the multivariable models as a covariate. All models were adjusted for sex, age group, socioeconomic status, and malnutrition that may have an influence on infection status and clinical outcome [34,40]. Specifically, for schistosomiasis that is caused by two species in Côte d'Ivoire (S. mansoni and S. haematobium), we excluded these schoolaged children/adolescents from the sample who were infected with S. haematobium in case interaction between S. mansoni and Plasmodium was assessed and vice versa to obtain speciesspecific estimates. For all Plasmodium-helminth pairs, interaction measures on multiplicative scale were assessed through an interaction term in the respective logistic regression model. RERI, AP, and SI and their respective 95% confidence intervals (CIs) and p-values were calculated to assess departure from additivity using the following formulas incorporated in a readily available calculation mask [23]: and where eb 1 stands for the OR of having condition A (A+B-) relative to the reference category of not having either condition (A-B-), eb 2 stands for the OR of having condition B (A-B+) relative to the reference category (A-B-) and eb 3 is the OR of having both conditions (A+B+) relative to the reference (A-B-).
Interaction measures on additive scale are calculated on the assumption that all infection categories (i.e., single and co-infection categories) present an increased risk for an outcome compared to the reference category. In case ORs were lower than 1 for one or several of these risk categories within a model, AP and SI were considered as non-applicable [41]. Significant antagonistic and synergistic interactions on additive scale were defined as SI <1 and SI >1, respectively, with a p-value below 0.05 and a 95% CI not including 1. The same was applied for interactions on multiplicative scale but using the OR of the respective product term.

Parasite infection and clinical status
Adults, in general, had a lower prevalence of Plasmodium and helminth infections than school-aged children/adolescents, with the exception of hookworm that was more often found in older individuals (31.4% in adults compared to 26.0% and 17.2% in school-aged children/ adolescents in the community and the school survey, respectively) ( Table 1). Anemia among adults (28.3%) from the four communities was comparable to the national school-based survey

Sociodemographic, parasitologic, and nutritional determinants of clinical morbidity among school-aged children/adolescents and adults
Univariate relationship analysis between clinical morbidity and sociodemographic, parasitologic, and nutritional factors revealed that the contribution from Plasmodium and helminth infections on anemia in adults was low (  in detail the different multiple species infections observed in the three population cohorts (i.e., adults from communities, school-aged children/adolescents from communities, and school-aged children/adolescents from the school-based survey) and how samples were drawn within to serve for interaction analysis of the different co-infection scenarios. Highest proportions of Plasmodium-helminth co-infections were observed in school-aged children/adolescents from the community-based surveys. Excluded individuals in the communities showed differences in mean age and socioeconomic status but were similar with regard to the outcome variables and other influencing factors. Most individuals excluded from the heterogeneous national school-based repository were eliminated during harmonization for specific helminth species endemicity, while only a small proportion (<3%) was dropped for being co-infected with the second Schistosoma species. While for the Plasmodium-S. mansoni scenario excluded individuals were of older age, lower socioeconomic status and more affected by co-morbidities (here: splenomegaly and malnutrition) the opposite was observed for the Plasmodium-hookworm scenario. Individuals from hookworm non-endemic areas had higher proportions of females, were younger, wealthier, and less affected by anemia and malnutrition.

Interaction measures on anemia
In all three Plasmodium-helminth co-infection models of the adult sample, at least one coinfection had a lower odds (OR <1) compared to the reference category (S1 Table). Consequently, measures of interaction on additive scale were not reliable and their calculation not applicable. We identified a significant antagonistic effect on anemia on multiplicative scale in adult participants co-infected with Plasmodium and S. mansoni (OR 0.39, 95% CI 0.16, 0.95).
Similarly to the findings from univariate regression analysis, single and co-infection categories of Plasmodium and Schistosoma showed significantly higher ORs for anemia in the multivariable models compared to non-infected school-aged children/adolescents from the schoolbased survey, while no such relationships could be shown for hookworm infection ( Table 3). ORs of the co-infection categories remained slightly below the expected sum of the ORs for the single infection categories, thus showing a tendency for antagonism (SI <1). Indeed, for the scenario of high Plasmodium parasitemia and concomitant S. haematobium infection (OR 0.26, 95% CI 0.13, 0.52), we identified a significant antagonistic effect on multiplicative scale for anemia among school-aged children/adolescents from the school-based sample. This finding was underlined by a negative RERI of -1.86 (95% CI -2.75, -0.96). No significant antagonism or synergism on anemia on additive scale (indicated by SI and AP) for Plasmodiumhelminth co-infection was identified.
Interaction analysis in school-aged children/adolescents from the communities revealed no significant interactions, neither on additive nor on multiplicative scale, for anemia as an outcome using six different co-infection scenarios (Table 4). Of note, certain co-infection categories showed significantly higher ORs for anemia compared to the corresponding reference categories such as school-aged children/adolescents positive for Plasmodium and S. haematobium (OR 2.39, 95% CI 1.08, 5.32) and school-aged children/adolescents with high Plasmodium parasitemia and concomitant hookworm infection (OR 2.35, 95% CI 1.19, 4.62).

Interaction measures on splenomegaly
Within the national school-based survey, co-infection with Plasmodium and either of the two Schistosoma species showed antagonistic effects on splenomegaly (Table 3). For the scenarios of high Plasmodium parasitemia and S. haematobium co-infection (OR 0.33, 95% CI 0.13, 0.86) and Plasmodium-S. mansoni (OR 0.32, 95% CI 0.10, 0.99) co-infection, these negative interactions were found to be on multiplicative level. Both single-infection categories of the two scenarios showed significantly higher ORs for splenomegaly; the co-infection categories, however, were below the expected sum or non-significant. No significant antagonism or synergism on splenomegaly for Plasmodium-hookworm co-infection was observed in the schoolbased sample. Table 4 summarizes the six scenarios applied for testing interaction on splenomegaly in Plasmodium-helminth co-infection in school-aged children/adolescents from the communitybased surveys. Neither of the Plasmodium-Schistosoma pairs showed any significant interaction measures or higher ORs for single and co-infection categories compared to the reference Antagonistic effects of Plasmodium-helminth co-infections on malaria pathology Table 3

Discussion
This study presents an analysis of interaction measures on additive and multiplicative scales for Plasmodium-helminth co-infection on two clinical manifestations (i.e., anemia and splenomegaly) using a suite of data from cross-sectional surveys carried out in Côte d'Ivoire in two population groups (i.e., school-aged children/adolescents and adults). Our findings revealed only a limited contribution of (co-)infections on anemia in adults. In adult females, pregnancies and motherhood may have much higher impact on anemia, especially in rural settings of Côte d'Ivoire where a woman has, on average, five or six children in her life [42]. Another explanation for the low contribution may be seen in the low helminth infection intensities in most of the positive cases [17]. In a recent study, Chami and colleagues have shown that general infection status contributes little to the anemia burden in adults, however cure and reduction of heavy intensity infections of S. mansoni and hookworm may reduce anemia in affected adults up to 37.0% and 24.9%, respectively [43]. Yet, in our adult sample S. mansoni-co-   infection showed a significant antagonistic effect on anemia on multiplicative scale; a finding that warrants further investigation. Splenomegaly rates were low in adults, and hence, it was not feasible to pursue any specific analysis. The low rates are likely attributable to an acquired immunity to malaria among this age strata in this high-endemicity area [44]. In the schoolaged/adolescent population, we observed considerable differences depending on whether the study was community-or school-based. School-aged children/adolescents from the four communities in southern Côte d'Ivoire showed higher infection prevalence, levels of parasitemia and clinical morbidity than the national school-based sample. Differences in included and excluded Schistosoma-infected individuals in the communities can be explained by higher peak age in S. mansoni infections and a more developed infrastructure in the village of Sahoua where most of the S. haematobium cases occurred. Relationships between clinical morbidity and single infections with Schistosoma and Plasmodium, yet, were found to be more significant and pronounced among school children from a more heterogeneous sample from whole Côte d'Ivoire. However, all identified significant interactions in both school-aged children/adolescents data sets are of antagonistic nature but vary in magnitude and parasite species combination. Each Plasmodium-helminth pair for the school-aged/adolescents strata is discussed in more detail below.

Plasmodium-hookworm co-infection
Unlike studies conducted in East Africa [45,46] and what is generally expected from two parasite species that have been shown to be positively associated [47] and having different mechanistic pathways of causing anemia [2,14], we could not identify any exacerbation in anemia due to Plasmodium-hookworm co-infection in the school-aged population. What we observed is basically no departure from additivity. This may partly be explained by generally low hookworm infection intensities found in our study samples [10,17]. Our results are in line with earlier findings from Côte d'Ivoire [16] and from rural communities in central Ghana [47] that showed a comparable epidemiologic profile with high Plasmodium rates (�75% of school-aged children/adolescents infected) and mainly low-intensity hookworm infections. In settings with such a high force of Plasmodium transmission, contribution to anemia may thus be largely attributable to malaria, while the effect on anemia from light-intensity hookworm infection seems negligible. Nutritional studies assessing potential contribution of Plasmodiumhookworm co-infection to iron-deficiency anemia undertaken in rural Côte d'Ivoire further emphasize this assumption [5]. Of note, numerous studies on Plasmodium-helminth co-infection and its relationship with anemia reported highest odds of anemia in Plasmodium singleinfected individuals, followed by helminth (co-)infected individuals compared to the uninfected reference group [48][49][50]. However, these prior studies did not have a closer look at potential (antagonistic) interactions, which indeed seem to be present, considering lower risks of anemia in people co-infected compared to Plasmodium single-infected individuals. Within the community-based survey, Plasmodium-hookworm co-infection showed protective effects against splenomegaly, whereas this negative interaction could not be confirmed in the national school-based sample. Other studies on potential interaction from Plasmodium-hookworm coinfection on splenomegaly are still scarce. Mboera and colleagues investigated the effect of concurrent hookworm and Plasmodium infection in Tanzanian school children and revealed enhanced Plasmodium parasitemia in Plasmodium-hookworm co-infected individuals but no exacerbating effect on splenomegaly [49].

Plasmodium-S. haematobium co-infection
In the community-based sample where S. haematobium infection was three times more prevalent than in the national school-based sample, S. haematobium single or Plasmodium-S. haematobium co-infection were negatively associated with anemia, whereas Plasmodium infection was the main driver of splenomegaly. For all four scenarios of Plasmodium-S. haematobium co-infection on anemia and splenomegaly, we found no significant interactions among school-aged children/adolescents from the four communities. Yet, we identified a significant antagonism on multiplicative scale on both morbidity indicators for the scenario of S. haematobium plus concomitant high Plasmodium parasitemia from the national school-based survey. Earlier studies from Tanzania and Zimbabwe identified Plasmodium and S. haematobium single-and co-infections as important predictors of anemia and splenomegaly, while only little information was available on the presence and magnitude of potential interactions from coinfection [20,49,50]. Findings from Mali and Senegal corroborate the assumption of a protective effect of Plasmodium-S. haematobium co-infection on malaria pathology, such as a reduced number of febrile malaria episodes and lower Plasmodium parasitemia in co-infected individuals [15,18,51,52].

Plasmodium-S. mansoni co-infection
While earlier studies conducted in East Africa reported exacerbation of (hepato-)splenomegaly in Plasmodium-S. mansoni co-infection [8,19], we observed the contrary. Although all infection categories (single and co-infection) were significantly positively associated with splenomegaly in school-aged children/adolescents from the national school-based sample, interaction measures clearly indicated an antagonism on multiplicative scale in Plasmodium-S. mansoni co-infected school-aged children/adolescents. Among school-aged children/adolescents from the four communities, we could not identify a significant antagonism, but odds of splenomegaly seemed to be reduced both in school-aged children/adolescents with low and high Plasmodium parasitemia and concomitant S. mansoni infection, compared to schoolaged children/adolescents infected with Plasmodium singly (low and high parasitemia). The odds of anemia were highest among Plasmodium-S. mansoni co-infected, compared to all other infection categories, in school-aged children/adolescents from the school-based sample. Interaction analysis, however, did not show any protective effect nor any exacerbation to the expected risk from concomitant infection. Interestingly, both antagonism as well as synergism in Plasmodium-S. mansoni co-infection might be explained by immunological interactions by down-or up-regulating, respectively, of a pro-inflammatory immune response that is related with the development of (hepato-)splenomegaly and that may also contribute to anemia due to inflammation [1,53,54]. The direction in which the immune response and thus the potential morbidity is induced may depend on the clinical phase of the disease, the duration since infection, the infection intensity of both, Plasmodium and Schistosoma, and the disposition of the host's immune system [53]. In our study, we mainly focused on school-aged children/adolescents, part of whom might not yet have acquired full immunity against these infectious diseases. All S. mansoni-positive individuals were identified as egg excreters and their infection stage thus may be considered as chronic but with mainly low egg counts.

Implications for public health
Although we mainly identified antagonistic effects on anemia and splenomegaly in Plasmodium-helminth co-infected individuals, our data indicate considerable burden due to these infections. Especially on national scale, multiparasitism represented a double burden among school-aged children/adolescents, yet an exacerbation from concomitant infection was not evident in our setting. Burden assessment due to parasitic infections for co-endemic countries thus warrants adaption as demanded and discussed over the last decade [55,56]. Of note are also the apparent differences in interactions of Plasmodium-helminth co-infections between African sub-regions. While data from East African settings highlight potential exacerbation [19][20][21] the opposite was found in studies investigating etiopathology of malaria in helminth co-infected individuals in West Africa [15,16,18,51,52]. Some of these differences from one country to another may be explained through a different force of Plasmodium transmission (e.g., mostly higher in West Africa), but also varying helminth species strains, related virulence and, to what extent they cause clinical morbidity, may play a role whether Plasmodium-helminth co-infection results in a more antagonistic effect rather than in synergism [9,57]. Moreover, the presence of a second helminth co-infection may steer potential interactions on malaria pathology in the opposite direction or enhance them [58,59]. In the light of national control programs and increased financial means and control efforts targeting helminthiases, a deeper understanding about potential effects on susceptibility to malaria in co-endemic areas should be gained. Removing helminth infections may alter immune responses to Plasmodium infection in previously helminth-infected individuals and thus also influence the number of malaria attacks and severity [58]. Malaria control should thus pay increased attention to the school-aged population that is the major target group of preventive chemotherapy against helminthiases and already neglected in terms of malaria treatment and use of preventive measures compared to other populations (e.g., pregnant women and children under five years of age). Furthermore, for enhancing (cost-)efficiency, concerted efforts against both types of infectious diseases in co-endemic areas are desirable and could for example include education campaigns against malaria during helminth mass drug administration and against helminthiases during mass distributions of insecticide-treated bed nets, respectively.

Strengths and limitations
Our study has several strengths and limitations. The diagnostic approach, particularly for detection of helminth infections, lacks sensitivity, and hence infection rates might be underestimated. Of note, mainly light-intensity infections were missed, thus our results might still be reliable for relationship and interaction analysis between infection status and clinical morbidity [7]. In Côte d'Ivoire, helminthiases other than schistosomiasis and soil-transmitted helminthiasis exist, such as enterobiasis, strongyloidiasis, onchocerciasis, and lymphatic filariasis [60,61]. These diseases were not investigated but they might influence malaria pathology by altering the immune response to Plasmodium [62,63]. With regard to the two clinical outcomes considered in our study-anemia and splenomegaly-especially the former is multifactorial and may be caused by hematological diseases, inflammation, and nutritional deficiencies, in addition to helminth infection and malaria. However, we conjecture that a major contribution to these clinical manifestations are parasitic infections [1] with Plasmodium and helminth infections ranking among the most important ones in terms of prevalence and burden [55]. With regard to methodological and analytical limitations it is worth highlighting that especially the concept of biologic interaction on additive scale, as proposed by Rothman et al. [22], is designed for case-control studies with matched groups and investigating risk exposures (relative risk (RR) or OR >1); conditions that are usually not met in observational studies. Crosssectional studies may thus complicate analysis or produce a high uncertainty and variance in the interaction estimates through (i) unbalanced numbers within the different exposure categories (e.g., low number of Plasmodium-uninfected school-aged children/adolescents in our samples); (ii) relationships between exposure and outcome may not always reveal an actual risk but rather being protective (RR or OR <1) compared to the defined reference category; (iii) the investigated outcomes may be influenced by other covariates (e.g., age, sex, and socioeconomic status) that are potentially not equally distributed among the comparison groups; and (iv) splenomegaly and anemia are manifestations due to chronic infection and longitudinal study designs may better capture relationships with (earlier) infections that may even be missed during cross-sectional surveys. We addressed some of these points by the introduction of low-and high-parasitemia Plasmodium infection scenarios to avoid small reference groups and for investigation of the relationship with parasite burden. Additionally, effects on estimates from unbalanced numbers in each category may be less pronounced in large data sets such as the national school-based study sample in the current analysis. Interaction measures on additive scale for preventive factors were not considered and recoding of exposure variables, as proposed by Knol and colleagues [41], avoided due to complication of subsequent interpretation. Furthermore, we built logistic regression models that allowed for adjustment for potential confounders and Schistosoma species-specific models by excluding individuals with concomitant infections with the second Schistosoma species. To reassure that only individuals from actual co-endemic areas are compared within interaction analysis, we subsampled school-aged children/adolescents from the national cohort according to helminth endemicity. This may have helped to harmonize partly as seen in several differences in parameters between excluded and included individuals. We could, however, not change the study design; a longitudinal design looking at infection status at different time points and thus allowing for discrimination of repeated infections may be more appropriate in future studies assessing clinical outcomes that are related to chronic exposure with the investigated parasites.