Plasmodium vivax Malaria during Pregnancy, Bolivia

Plasmodium vivax is a major cause of illness in areas with low transmission of malaria in Latin America, Asia, and the Horn of Africa. However, pregnancy-associated malaria remains poorly characterized in such areas. Using a hospital-based survey of women giving birth and an antenatal survey, we assessed the prevalence rates of Plasmodium spp. infections in pregnant women in Bolivia, and evaluated the consequences of malaria during pregnancy on the health of mothers and newborns. P. vivax infection was detected in 7.9% of pregnant women attending antenatal visits, and placental infection occurred in 2.8% of deliveries; these rates did not vary with parity. Forty-two percent of all P. vivax malaria episodes were symptomatic. P. vivax–infected pregnant women were frequently anemic (6.5%) and delivered babies of reduced birthweight. P. vivax infections during pregnancy are clearly associated with serious adverse outcomes and should be considered in prevention strategies of pregnancy-associated malaria.

Plasmodium vivax is a major cause of illness in areas with low transmission of malaria in Latin America, Asia, and the Horn of Africa. However, pregnancy-associated malaria remains poorly characterized in such areas. Using a hospital-based survey of women giving birth and an antenatal survey, we assessed the prevalence rates of Plasmodium spp. infections in pregnant women in Bolivia, and evaluated the consequences of malaria during pregnancy on the health of mothers and newborns. P. vivax infection was detected in 7.9% of pregnant women attending antenatal visits, and placental infection occurred in 2.8% of deliveries; these rates did not vary with parity. Forty-two percent of all P. vivax malaria episodes were symptomatic. P. vivaxinfected pregnant women were frequently anemic (6.5%) and delivered babies of reduced birthweight. P. vivax infections during pregnancy are clearly associated with serious adverse outcomes and should be considered in prevention strategies of pregnancy-associated malaria.
I n Latin America, where malaria transmission is low and mostly unstable, Plasmodium vivax is the most prevalent malaria parasite species. Although ≈3 million pregnant women are exposed to malaria in Latin America each year, the actual number of malaria infections during pregnancy is considerably lower (1). Pregnancy-associated malaria is poorly characterized in such areas of low or unstable transmission, as in most areas in which of P. vivax is predominant (2), but malaria can be severe in all parity groups because most women of childbearing age have low levels of prepregnancy and pregnancy-specific protective immunity to malaria (3).
One of the first studies that demonstrated parasitization of the placenta by P. falciparum was conducted in Latin America (4), and reported serious adverse outcomes, such as miscarriages late in pregnancy or stillbirths. No other study related to pregnancy-associated malaria was conducted in Latin America for ≈80 years until a cohort study investigating P. vivax infection during pregnancy in Honduras (5) and a case-series report of 143 pregnant women infected with P. falciparum in French Guiana (6) were reported. Both studies outlined serious adverse outcomes (anemia, preterm delivery, hypotrophy, and stillbirth) associated with malaria by parasite species during pregnancy. More recent studies in the Amazon regions of Brazil and Peru reported increased incidence rates of infection with P. falciparum, but not P. vivax, in pregnant women (7,8). Outside Latin America, a few studies reported the effect of pregnancy-associated malaria in unstable malaria settings in Africa and Asia (9-12), and described increased risks for low birthweight and for maternal anemia as consequences of P. vivax infection during pregnancy (13,14).
Using a hospital-based survey of women giving birth and an antenatal survey, we assessed the prevalence rates of Plasmodium infection in pregnant women in 2 malaria-endemic areas of Bolivia. We also evaluated the consequences of malaria infection during pregnancy on the health of mothers and newborns.

Study Sites
This study was conducted in 2 malaria-endemic areas in Bolivia: the northern district of Guayaramerín in the Amazon region on the border with Brazil, and the district of Bermejo in the southern region on the border with Argentina. In both areas, malaria transmission occurs during the warm and wet season during November-April and is low and unstable; Anopheles darlingi and An. pseudopunctipennis mosquitoes are the main malaria vectors, respectively (15,16). P. vivax predominates in both areas; P. falciparum is present only in Guayaramerín. The annual parasite incidence rates in 2003 were 21.6 and 106.6 infections/1,000 inhabitants in Bermejo and Guayaramerín, respectively. These 2 districts are targeted by routine residual insecticide house-spraying programs that use alphacypermethrin (coverage rate <60%). P. falciparum isolates are usually resistant to chloroquine and sulfadoxine/pyrimethamine (17), but no chloroquine-resistant P. vivax has been reported. Ethical approval for this study was obtained from the Bolivian Ministry of Health (National Institute of Health Laboratories, La Paz).

Hospital-based Survey
This survey was conducted during December 2002-August 2004 among women giving birth in 2 district hospitals in which >65% of women in the area give birth. Personal history was obtained for all women, including obstetrical antecedents, place of residence, house insecticide spraying, and signed informed consent. After delivery, a placental blood smear was obtained from the maternal side of the placenta. Thick and thin blood films were prepared. Gestational age of neonates was calculated at birth by using the score of Farr et al. (18). Newborns were weighted on a digital scale that was accurate to within 10 g.

Antenatal Survey
In Guayaramerín, all consenting pregnant women receiving antenatal care in 2 rural and 5 urban health centers during May 2003-August 2004 were investigated. During each antenatal visit, we performed physical examinations and blood smear examinations for malaria parasites. Giemsa-stained blood smears were read in each center by a trained malariologist. Women with a malaria infection (P. vivax or P. falciparum) were treated according to the national guidelines at the time (chloroquine or quinine plus clindamycin, respectively). Women were invited to give birth at the district hospital of Guayaramerín and participate in the hospital-based survey.

Laboratory Studies
Hemoglobin levels were determined by using the cyanomethemoglobin method (HemoCue, Cypress, CA, USA). Peripheral and placental smears were stained with Giemsa, and 200 microscopic fields were examined.

Definitions
Neonates were classified as premature if they were <37 weeks gestation at birth. Low birthweight was defined as a body weight <2,500 g. Anemia was defined as a hemoglobin level <11 g/dL, and moderate-to-severe anemia as a hemoglobin level <8 g/dL. Asymptomatic malaria infection was defined as the presence of malaria parasites on blood smears in the absence of fever (axillary temperature >37.5°C) or a history of fever in the preceding 48 hours.

Data Analysis
Twins and stillbirths were excluded from the analysis. P. vivax and P. falciparum infections were dichotomized independently. Categorical variables were compared by using χ 2 or Fisher exact tests, and continuous variables were compared buy using the Mann-Whitney test. We used Stata/MP 11 (StataCorp LP, College Station, TX, USA) for multiple linear or logistic regressions (with backward stepwise elimination) to adjust for potential confounding variables (mother's age, parity, antenatal care attendance, indoor insecticide spraying, site of study, delivery during transmission season, and sex and gestational age of the baby), and to determine the population attributable fraction (PAF), which is also known as the etiologic fraction, or that proportion of all events (severe anemia, low birthweight) associated with the factor of interest (e.g., P. vivax or P. falciparum infection).

Hospital-based Survey
During December 2002-August 2004, a total of 1,003 women in Guayaramerín and 504 women in Bermejo had singleton births at the 2 district hospitals. In both hospitals, mean parity and proportion of primiparous women were similar (Table 1). However, women were younger (mean ± SD age 23.2 ± 6.4 years vs. 24.2 ± 6.7 years; p = 0.008) and had more antenatal visits (4.7 ± 2.0 visits vs. 3.7 ± 2.1 visits; p<0.001) in Guayaramerín than in Bermejo. The proportion of women without any antenatal visit was 4 times higher (7.8% vs. 1.9%; p<0.001) in Bermejo than in Guayaramerín. Women lived less often in rural settlements (6.8% vs. 22.7%; p<0.001) and had babies more often during the transmission season (60.0% vs. 52.6%; p = 0.006) in Guayaramerín than in Bermejo. Rates of lowbirthweight and moderate-to-severe maternal anemia at birth were similar in both places.
Among 967 women who had babies in Guayaramerín and had a placental smear, 26 (2.7%) were had P. vivax infections in placental blood. In addition, 4 (0.4%) had placental P. falciparum infections. Among 500 women who had babies in Bermejo and had a placental examination, 15 (3.0%) had P. vivax infections in placental blood. Because of these differences, we further adjusted for study area to evaluate the effects of P. vivax infection in pregnant women. We further distinguished infections by P. falciparum or P. vivax for the analysis.
The risk for placental P. vivax infection increased during the transmission season in both places (adjusted odds ratio [OR] 2.7, 95% CI 1.3-5.6, p<0.007). There was no effect of parity, mother's age, antenatal care attendance, or indoor insecticide spraying on placental P. vivax prevalence in both districts.
Women with placental P. vivax infections were more likely than noninfected mothers to have a low-birthweight baby (OR adjusted for study site 3.6, 95% CI 1.4-8.9) (Table 2). These women were also more likely than noninfected women to have moderate-to-severe anemia (adjusted OR 2.5, 95% CI 1.0-6.2).
Factors associated with mean birthweight in a multiple linear regression model are shown in Table 3. Mean birthweight was reduced in premature (-752 g), female (-151 g), first-born (-168 g), and second-born babies (-79 g), as well as in babies born to mothers living in Guayaramerín (-52 g), women who had no antenatal visits (-112 g), and women with placental P. vivax infections (-181 g). Mean birthweight was increased in babies born to women >35 years of age (+181 g) or women 25 35 years of age (+102 g). Logistic regression (Table 3) showed that preterm delivery (p<0.001) and placental P. vivax infections (OR 6.2, 95% CI 2.2-17.6, p<0.001) were associated with an increased risk for low-birthweight babies. The population attributable risk for low-birthweight babies associated with P. vivax infection was 6.1% (95% CI 0.4%-11.4%).
In contrast to placental P. vivax infection, placental P. falciparum infection was more likely to occur in primiparous women than in multiparous women (0.7% vs. 0.1%; p = 0.05). After exclusion of Bermejo and P. vivax infections, the risk for low birthweight increased in premature babies (OR 31.2, 95% CI 15.7-62.1; p<0.001) and in babies born to mothers with placental P. falciparum infections (OR 5.1, 95% CI 1.6-16.6; p = 0.006).
Factors associated with a change in mean birthweight by a multiple linear regression model are shown  in Table 5. Mean birthweight was lower in girls (-135 g), premature babies (-426 g), and first-pregnancy babies (-181 g), as well as in babies born to anemic mothers (-92 g) or to mothers infected with P. vivax during pregnancy (-266 g). Logistic regression showed that preterm delivery (p = 0.001) and P. vivax infection during pregnancy (OR 8.8, 95% CI 2.4-32.5) were associated with low birthweight.

Discussion
Long after the first studies on pregnancy-associated malaria conducted in Africa, most studies in Latin America during the past decade, mainly case-series studies, reported numerous adverse conditions, such as a high frequency of maternal anemia, miscarriage, stillbirth, preterm delivery, and low birthweight (6,(19)(20)(21)(22)(23), related to malaria infections with P. falciparum or P. vivax during pregnancy. As observed in other malaria-endemic areas, a cohort study in Peru and a cross-sectional study in Brazil reported a 2.5fold increase in susceptibility to P. falciparum malaria among pregnant women than among nonpregnant women (7,8). Neither study demonstrated a similar higher frequency of P. vivax infection in pregnant women.
In the current study, P. vivax infection was detected in 7.9% of pregnant women attending antenatal visits. This proportion is similar to rates in other settings, such as in Thailand (6.4%-8.5%) and Honduras (9.1%) (5,9,13). These findings suggest a constant proportion of P. vivax infections during pregnancy in different malaria transmission patterns. In Thailand, 23% of all P. vivax malaria episodes were symptomatic (13), but this rate reached 42% in Bolivia. In addition to possible differences in background immunity resulting from more unstable transmission in Bolivia, this difference might also be caused by prompt diagnosis and treatment on a weekly basis in the study in Thailand, which enabled parasite detection and cure before onset of symptoms. In our study, diagnosis and treatment were performed monthly, which is the approximate interval between 2 antenatal visits, which enabled a longer time for symptoms to develop.
Among pregnant women, primiparae women are most vulnerable to P. falciparum infections, and the difference between primiparae and multiparae women is more pronounced in areas of stable than unstable malaria transmission (11,12,24). We observed similar differences, despite a limited number of P. falciparum-infected women. In contrast and consistent with previous reports (10), P. vivax infection was observed in a similar proportion of women of all parities. However, 1 study reported an increased risk for P. vivax infection in primiparae than in multiparae (13).
A high proportion of pregnant women in both study sites in Bolivia had anemia, and the proportion of women with moderate-to-severe anemia increased with parity. As observed in unstable malaria transmission settings, the risk for maternal anemia was more pronounced in multiparae than in primiparae women (9,10,12). In our study, P. vivax infection was associated with a reduction of 0.7 g/dL in the hemoglobin level of infected pregnant women than that of noninfected women. A similar difference (0.8 g/dL) was observed in Honduras (5) between P. vivax-infected and noninfected women. Logistic regression showed that the risk for maternal anemia was associated with P. vivax infection at delivery, multiparity, and the study district in northern Bolivia. In our antenatal cohort study, P. vivax infection acquired during pregnancy remained independently associated with the risk for moderate-to-severe anemia. A similar relationship was observed in Thailand (13). Other studies also reported the effect of infection with P. falciparum or P. vivax during pregnancy on the risk for maternal anemia, but confounding malaria species could have led to classification bias (9,10,12,19). Babies born to P. vivax-infected mothers showed a major mean birthweight reduction of 181 g when compared with babies born to noninfected women, which is consistent with observations in Thailand and Honduras (107 and 198 g, respectively) (5,13). Mean birthweight was also highly reduced in case of preterm delivery, of poor antenatal clinic attendance, and in babies born to first-and second-pregnancy. These factors were consistently identified in studies performed in unstable malaria settings (10,13). In our study, placental P. vivax infection was associated with a 6-fold higher risk for low birthweight, which is ≈4 times higher than the risk estimated in Thailand (13). However, in Madagascar, in areas of unstable malaria transmission, the risk for low birthweight associated with P. falciparum infection was 2.5 times that in areas with stable transmission (11). As suggested by the higher proportion of symptomatic infections in our study in Bolivia, the index of stability may be lower in Bolivia than in Asia if one takes into account a higher risk for low birthweight associated with P. vivax infection. In our study, the risk for low birthweight increased with the number of P. vivax infections that occurred during pregnancy (by test for trend). These data are consistent with a similar dose-dependent effect in a study in Thailand, which reported a greater reduction in birthweight in mothers infected >5 times than in mothers infected only 1 time (9).
In Bolivia, the PAF of moderate-to-severe anemia associated with P. vivax malaria was 3.5%, and the PAF for low birthweight was 6.1% for P. vivax. Our estimation is consistent with the 2%-15% estimation of the PAF for severe anemia related to P. falciparum in malaria-endemic areas (25). In contrast, P. vivax seemed to have less of an effect on the risk for low birthweight than P. falciparum in malaria-endemic areas in Africa (risk 8%-14% estimated by Steketee et al.) (25).
Our cross-sectional survey has limitations, including selection biases (if most women do not attend selected structures or because they give birth at home because private and nongovernment organization sectors predominate in the public sector) and representativeness. In the 2 districts we studied, private and nongovernment organization sectors are negligible and most births are in public sector facilities. However, ≈25%-33% of births were at home in the regions studied. This factor is a possible limitation because we did not assess deliveries at home. This limitation is similar for prenatal visits, but the number of pregnant women who receive prenatal care in Bolivia is high (>80%). We conducted the prenatal follow-up study in 7 health centers to ensure representativeness. To avoid missing the transmission season, we conducted the study in >1 calendar year.
Although P. vivax infections are clearly associated with serious adverse outcomes during pregnancy, accumulation of P. vivax in the placenta has not been reported. P. vivaxinfected erythrocytes can bind chondroitin sulfate A, the placental binding receptor (26), but at a 10-fold lower level than P. falciparum-infected erythrocytes (27), and at a similar level in isolates from pregnant women or nonpregnant persons (28). The paucity of P. vivax in the placenta has been reported (29,30), and P. vivax has been inconsistently associated with the presence of malaria pigment in the placenta, but not associated with placental pathologic changes (14,31). High circulating levels of inflammatory cytokines during the paroxysms of P. vivax malaria (32) may be sufficient to impair fetal growth and cause maternal anemia, as hypothesized by Nosten et al. (2). Moreover, rosette formation is a frequent cytoadhesive phenotype in P. vivax infections and has been associated with an increased risk for anemia (28). Nevertheless, phenomena involved in pathologic mechanisms specific for P. vivax infection during pregnancy remain to be elucidated.
P. falciparum and P. vivax infections during early pregnancy have been shown to result in impaired fetal growth (33), which emphasizes the need to include early pregnancy in the prevention strategies of pregnancyassociated malaria. In addition, almost half of P. vivax infections were asymptomatic, suggesting that women should be screened for malaria at every antenatal clinic visit, and treated if test results were positive. Although the effects of P. vivax infection during pregnancy have become increasingly documented, health personnel in malaria-endemic areas of Latin America still largely ignore recommendations for diagnosis and treatment of malaria in pregnant women (34). Efforts should be undertaken to increase staff training to limit the effect of malaria during pregnancy.