In response to a variety of helminthic infections, the non-pregnant host develops a strong Th2/regulatory T cell (Treg) response with interleukin-4 (IL-4), IL-5, IL-10, and IL-13 synthesis. Similar to pregnancy, the Th2 cytokine production from chronic helminthic infection could also downgrade the initial Th1 responsiveness required for control of malaria infection.^10,11 However, epidemiologic studies evaluating helminthic coinfection with malaria among non-pregnant adults and children have provided conflicting results, suggesting that immune responses may be helminth-dependent, with other factors, such as stage and intensity of infection and age of the host, playing an important role.^12–17 These complex relationships remain even less clear among pregnant or human immunodeficiency virus (HIV)-infected patients.

Participants provided blood samples for rapid HIV testing and peripheral malaria blood films and midday urine for Schistosoma haematobium. When possible during the study visit, participants also provided stool for hookworm, Ascaris, Trichuris, and Schistosoma mansoni. Malaria blood films were stained with 3% Giemsa for 30 minutes and examined by two independent expert microscopists for P. falciparum. The microscopists counted the total number of asexual malaria forms per 300 white blood cells (WBCs). We multiplied this count by 20 to estimate the number of malarial parasites per 6,000 WBCs to approximate 1 μL blood. To assess S. haematobium infection, we centrifuged 10 mL urine and resuspended the pellet. We then placed the suspension on a glass slide (with glass coverslip) and counted the total number of S. haematobium eggs. For stool samples, we used the Kato–Katz method published previously.²⁵ A trained microscopist counted the total number of hookworm eggs less than 2 hours after collection. Other helminth eggs (including S. mansoni, Trichuris, and Ascaris) were counted 4–8 hours after collection. Egg counts are reported as eggs per gram (epg) of stool (for hookworm, S. mansoni, Trichuris, and Ascaris) or eggs per 10 mL (for S. haematobium). We defined light, moderate, and heavy infections according to WHO classifications as follows: hookworm, 1–1,999, 2,000–3,999, and ≥ 4,000 epg; Ascaris, 1–4,999, 5,000–49,999, and ≥ 50,000 epg; Trichuris, 1–999, 1,000–9,999, and ≥ 10,000 epg; S. mansoni, 1–99, 100–399, and ≥ 400 epg.²⁶ Light and heavy intensity S. haematobium infections were defined as < 50 and ≥ 50 eggs/10 mL, respectively.²⁶ We performed HIV tests on all consenting participants using Determine HIV-1/2 (Abbott Diagnostics, Abbott Park, IL) and Unigold (Trinity Biotech, Bray, Ireland) rapid test kits in parallel. We considered two positive or negative results as being confirmed positive or negative, respectively. In the event of discordant results, we performed a third rapid test (HemaStrip; Saliva Diagnostics System, Brooklyn, NY); all discordant results were confirmed using enzyme-linked immunosorbent assay (ELISA) antibody testing.

We clinically managed all participants in accordance with Malawi Ministry of Health and Population (Malawi MOHP) guidelines at the time. Specifically, we screened all participants for anemia (using calibrated Hemocue equipment; Mission Viejo, CA) and syphilis by Venereal Disease Research Laboratory (VDRL) testing. All participants received ferrous sulfate and folic acid if noted to be anemic (hemoglobin < 11 g/dL). Participants who were VDRL positive received intramuscular benzathine penicillin. In addition, we provided all HIV-1–infected participants and their exposed infants with single-dose nevirapine administered as per the HIVNET012 protocol.²⁷ All participants noted to have soil-transmitted helminthes (STH) or schistosomiasis at any point during the study received a treatment course of mebendazole (500 mg) or praziquantel (600 mg), respectively, soon after delivery as recommended by the Malawi MOHP.

We entered all data locally using Epi Info version 6.03 (Centers for Disease Control and Prevention) and used SAS version 9.1 (SAS Institute Inc., Cary, NC) for all analyses. We normalized data on parasite and helminth infection intensity using log transformation and compared differences in geometric means using Student t test and one-way analysis of variance. In univariate analysis, we used Cochran-Mantel-Haenszel χ² statistics to characterize factors associated with the most common parasitic infections (malaria, hookworm, and S. haematobium). To ensure that mixed helminthic infections did not mask significant findings from individual helminthic infections, we repeated analyses with mixed hookworm and S. haematobium infection being categorized separately. We subsequently used multivariate logistic regression to characterize factors associated with malaria infection, including in our model all variables potentially associated with infection on univariate analysis (P ≤ 0.20). We restricted the model to cases for which information on all variables was available (including only women who provided blood, urine, and stool at the initial visit) and checked for two-way interactions. For presentation of final results, we considered P values < 0.05 to be statistically significant.

Although mild anemia (hemoglobin < 11 g/dL) was commonly observed (77.5%), severe anemia (hemoglobin < 7 g/dL) was identified in only 34 (4%) participants. Presence of mild anemia was not associated with any of the three most common helminthic infections (S. haematobium, hookworm, and S. mansoni). On univariate analysis, severe anemia was more common among participants infected with hookworm (7.4% versus 3.4%; P < 0.05). However, when other variables associated with anemia were incorporated into a multivariate logistic regression model, only malaria infection, first pregnancy, and elevated temperature (> 37.5°C) at presentation were associated with severe anemia (adjusted odds ratio [aOR] = 2.7, 95% confidence interval [CI] = 1.2–6.1; aOR = 3.7, 95% CI = 1.5–9.1; aOR = 3.3, 95% CI = 1.6–6.8, respectively]; neither helminthic nor HIV infection were significant.

The demographic and clinical characteristics associated with having P. falciparum, hookworm, or S. haematobium infection on univariate analysis are shown in Table 2. The characteristics associated with malaria infection in order of the strength of association are first pregnancy, having a pit latrine or no sanitary facility, earthen flooring at home, age < 20 years, exposure to lakes and rivers, mid-upper arm circumference (MUAC) ≤ 25 cm, having a primary school education or less, and weight < 50 kg. Having a job and sleeping under an insecticide-treated bed net (ITN) were associated with a significant reduction in the likelihood of having malaria infection (relative risk [RR] = 0.5, 95% CI = 0.3–0.8; RR = 0.7, 95% CI = 0.5–0.8, respectively). Although 26.9% of participants reported fever during pregnancy, there was no association between reporting fever and malaria parasitemia (47.1% with subjective fever versus 38.2% without subjective fever had malaria parasitemia; P = 0.36). Only 3 of 332 (0.9%) participants with malaria parasitemia had a temperature ≥ 37.5°C.

Associations in prevalence and parasite densities among those with coinfections are shown in Tables 3 and ​and4.4. HIV-infected participants were more likely to have malaria parasitemia (RR = 1.2, 95% CI = 1.00–1.4) and had higher geometric mean malarial parasite densities (721 versus 449 parasites/μL blood; P < 0.005) than those without HIV. However, HIV-positive participants were less likely to have S. haematobium infection (RR = 0.7, 95% = CI 0.5–0.9) and had significantly lower S. haematobium egg counts (17 versus 24 eggs/mL urine; P < 0.05). Malaria infection was more common among participants with hookworm coinfection (RR = 1.6, 95% CI = 1.4–1.9). However, hookworm coinfection was not associated with differences in malaria parasite density (557 versus 473 parasites/μL blood; P = 0.5). In contrast, participants with S. haematobium coinfection had lower malaria parasite densities (410 versus 528 parasites/μL blood; P < 0.05) but were not more likely to be infected with malaria (RR = 1.1, 95% CI = 0.99–1.3).

We further categorized heminthic infections into single or mixed infections. Hookworm infection alone remained associated with malaria infection (RR = 1.6, 95% CI = 1.3–2.1) compared with no helminthic infection. This same effect was noted even with very light infections (< 100 epg; RR = 1.6, 95% CI = 1.2–2.3). In addition, S. haematobium infection alone or mixed hookworm/S. haematobium infection was associated with greater odds of having malaria infection (RR = 1.2, 95% CI = 1.02–1.5; RR = 1.9, 95% CI = 1.5–2.4, respectively) compared with no helminthic infection.

Hookworm infection alone or mixed hookworm/S. haematobium infection had no apparent association with geometric mean malarial parasite densities compared with no helminthic infection (524 versus 557 parasites/μL blood; P = 0.85; 625 versus 557 parasites/μL blood; P = 0.75, respectively). However, S. haematobium infection alone remained significantly associated with lower geometric mean malarial parasite densities (344 versus 557 parasites/μL blood; P < 0.05) compared with no helminthic infection. No difference in mean malaria parasite density was identified between light and heavy S. haematobium infections.

Our results provide further evidence to show that S. haematobium coinfection may be associated with reductions in malaria parasite density in pregnant women.^13,14 In addition, consistent with prior studies, our findings suggest that hookworm infection may exacerbate malaria infections among pregnant women.^23,28 Hillier and others²³ showed that hookworm and Mansonella perstans infections were significantly associated with higher prevalence of asymptomatic malaria parasitemia, although this difference depended on the geographic location of residence in Entebbe, Uganda. Yatich and others²⁸ also showed that hookworm coinfection was associated with increased prevalence of malaria infection among pregnant Ghanaian women, although women with Ascaris coinfection had an even greater likelihood of malaria infection. In contrast, van Eijk and others²⁹ showed that Kenyan pregnant women in their second and third pregnancy were less likely to have malaria infection if infected with Ascaris.

The reasons for these conflicting results remain uncertain but are likely multifactorial. In animal models, the stage and intensity of helminthic infection affect host response.³⁰ In human epidemiologic studies, other factors, such as spatial variations in helminth density (even over short distances), have been suggested as significant contributors to malaria–helminthic interactions.²³ Clearly, the host response may also depend on the specific type of helminthic infection and the age of the host. Although the vast majority of helminths and their antigens are able to stimulate a Th2 host response, tissue-dwelling helminths such as Schistosoma spp. may require additional mechanisms to modulate the host immune system given that infection may persist for years. Of late, research has focused on helminth-derived compounds as effectors of the host immune system. Omega-1, a glycoprotein secreted by S. mansoni, has recently been identified as a potent inducer of Th2 immune responses, even in the absence of IL-4 signaling.³¹ Although additional research is needed, potent compounds such as omega-1 could contribute to the protective effect of Schistosomiasis on malaria infection.

However, all the above-mentioned epidemiologic studies (ours included) are observational studies and therefore, are limited in determining cause and effect between these coinfections. Thus far, only one randomized clinical trial has looked at the effect of helminthic treatment on malaria infection.³² Although this clinical trial noted that treatment of Ascaris was associated with a two-fold increase in malaria parasitemia in adults (suggesting a protective effect of the Ascaris coinfection), one must interpret these results with some caution, because the medication used (levamisole) has intrinsic immunomodulatory properties.³³ Of note, Ascaris treatment in children showed neither a beneficial nor detrimental effect on malaria infection. Given these limited results, additional clinical trials are needed.

Although this substudy was designed to evaluate the impact of helminthic coinfection on malaria, we also showed that HIV infection was associated with increased malarial prevalence and geometric mean malaria parasite density, but it reduced S. haematobium prevalence and egg counts. Neither of these findings was unexpected. A recent meta-analysis of 12 clinical trials showed that HIV-infected pregnant women are 1.6 times more likely to have malaria peripheral and placental parasitemia than those who are uninfected. Two- to three-fold increases in malaria peripheral parasitemia were also reported.³⁴ In contrast, prior observational studies have consistently shown reduced excretion of S. haematobium and S. mansoni eggs among non-pregnant adults coinfected with HIV, potentially leading to underdiagnosis of these infections among HIV-infected persons. This may be related to findings from animal studies that have shown that maturation of adult worms and transposition of S. mansoni eggs to the intestine are T-cell–dependent processes.³⁵

However, our study has several limitations. As with all observational studies, the results can only describe potential associations and not detail cause and effect of these coinfections. In addition, because host immune responses vary throughout pregnancy and with additional pregnancies, these findings might be different in multigravid pregnant women or those in their first or third trimesters. Similarly, because CD4 counts were not available from the HIV-infected women in this study, we cannot confirm that these results can be extrapolated across all varying levels of immunosuppression. Although we inquired about a multitude of potential socioeconomic factors that could potentially confound these results, it is possible that there were additional factors about which participants were not asked but that might explain this effect. However, despite the finding that type of floor in the home was significantly associated with malarial infection (suggesting SES did have a role in malaria infection), hookworm infection still remained statistically significant on multiple logistic regression analysis. Because we did not receive stool samples from all women providing blood and urine samples, it is also possible that unidentified STH infection could potentially confound the results. Given the infrequent occurrence of other STH infections, we are also unable to conduct further analysis of Ascaris or Trichuris coinfections. However, this was not unexpected, because Ascaris and Trichuris infections are rarely found in Malawi.³⁶ In addition, it is possible that our results were biased to those who were able to provide stool samples (such as those with gastrointestinal illness), although the majority of stool samples were formed. Lastly, both Kato–Katz stool examination and midday urine collection for S. haematobium have low sensitivity for detecting helminthiasis on a single sample, and therefore, some infections may be missed. However, this would suggest that the prevalence of STH and S. haematobium in this population is likely underestimated.