Malaria Infection during Pregnancy: Intrauterine Growth Retardation and Preterm Delivery in Malawi
- Amy D. Sullivan,
- Thomas Nyirenda,
- Timothy Cullinan,
- Terrie Taylor,
- Sioban D. Harlow,
- Sherman A. James and
- Steven R. Meshnick
- Department of Epidemiology, University of Michigan, Ann Arbor, and Department of Internal Medicine, Michigan State University School of Osteopathy, Lansing, Michigan; Ministry of Health and Department of Community Health, University of Malawi College of Medicine, Mangochi, and Malaria Research Project, Queen Elizabeth Central Hospital, Blantyre, Malawi
- Reprints or correspondence: Dr. Amy D. Sullivan, Division of Infectious Diseases, Dept. of Internal Medicine, 3120 Taubman Bldg., Box 0378, University of Michigan School of Medicine, Ann Arbor, MI 48109 (adsulli{at}umich.edu).
Abstract
In sub-Saharan Africa, malaria infection in pregnancy contributes to low birth weight through intrauterine growth retardation (IUGR) and preterm delivery (PTD). It was hypothesized that malaria-associated PTD and IUGR have differing etiologies due to timing of infection. In a prospective cohort of primigravid women enrolled at the antenatal clinic of Mangochi District Hospital in Malawi, the associations were investigated between antenatal or delivery parasitemias and IUGR or PTD. Among 178 singleton deliveries, 35% of infants were preterm or had IUGR. Cord blood parasitemia (odds ratio [OR = 3.34; 95% confidence interval [CI], 1.3–8.8], placental parasitemia (OR = 2.43; 95% CI, 1.2–5.1), and postdelivery maternal peripheral parasitemia (OR = 2.78; 95% CI, 1.3–6.1) were associated with PTD. Parasitemia and/or clinically diagnosed malaria in the antenatal period was associated with IUGR (OR = 5.13; 95% CI, 1.4–19.4). Delivery parasitemias had borderline associations with IUGR. The risk patterns observed suggest that the timing and severity of infection influences the occurrence of IUGR or PTD.
In sub-Saharan Africa, Plasmodium falciparum infections increase the risk for intrauterine growth retardation (IUGR) and preterm delivery (PTD), especially in primigravidae [1–5]. Whether PTD or IUGR is observed may depend on levels of immunity and intensity of transmission [3]: IUGR generally predominates in areas of intense perennial malaria transmission and PTD in areas of seasonal transmission. However, both IUGR and PTD can occur in the same areas [1, 2].
In this study, we examined whether timing of malaria infection during pregnancy is associated with the observation of PTD or IUGR in a prospective cohort of primigravid women attending antenatal services and delivering in Mangochi, Malawi, an area of intense perennial malaria transmission.
Methods
Primigravid women attending the antenatal clinic (ANC) at the Mangochi District Hospital (MDH) from November 1995 through June 1996 were asked to participate in the study. Enrolled women were included if they delivered at MDH before 1 September 1996. Because standard methods of estimating gestational age were unreliable or unavailable in this population, women were enrolled regardless of gestational age. Exclusion criteria included multiple birth, spontaneous abortion or stillbirth, and death or life-threatening illness of the mother after delivery.
We collected demographic, medical, and obstetric information at enrollment. At each antenatal visit and within 24 h after delivery, the mother's self-reported health status was obtained, and fingerprick blood was taken for malaria diagnosis and hematocrit and hemoglobin measurement. Thick smears were Giemsa-stained, and parasite densities were calculated relative to estimated white blood cell count [1]. National guidelines recommend provision of sulfadoxine-pyrimethamine (SP) at ∼18 and 32 weeks of pregnancy. SP doses were supervised by ANC personnel unless women refused treatment.
Within 16 h of delivery, infants were weighed on an electronic balance (Mettler Toledo, Columbus, OH) or spring scale, depending on electrical supply. Gestational age was evaluated by Dubowitz test performed by trained personnel [1, 6, 7]. Scores did not vary by assessor (p = .50). Infants delivered before 37 weeks of gestation were considered preterm. Percentile weight-for-age was assessed based on Aberdeen standards, controlling for infant's gestational age and sex and mother's parity [8, 9]. Infants in the lowest tenth percentile of weight-for-age were considered IUGR. Five infants with no Dubowitz scores were categorized by weight: ⩾3100 g, full-term normal weight-for-age, 1 infant; ⩽2000 g, preterm, 3 infants; 2001–3099 g, not categorized, 1.
Placental weights and thick smears were taken within 18 h of delivery. Three placentas were discarded before examination; 11 were not weighed. Relative placental overweight is associated with anemia [10, 11]. To evaluate this parameter, infants and placentas were grouped by standardized weight-for-age: less than −2, −2 to −1, −1 to 1, 1 to 2, or >2 [8, 10]. If the placenta was in a higher group than the infant (e.g., placenta, −1 to 1 and infant, −2 to −1), it was considered relatively overweight.
Because antenatal parasitemia could be correlated with number of ANC visits, we constructed a fifth malaria variable: having a positive thick smear and/or clinical malaria. The definition of clinical malaria was documentation of presumptive malaria diagnosis between 10 weeks of gestation and delivery. Documentation included admission slip from MDH, notation on the ANC card, or prescription slip for SP but, as per standard procedures, did not require laboratory confirmation. Clinical malaria was independent of number of ANC visits (P = .12).
Antenatal anemia was defined as ever having a packed cell volume (or hematocrit) <25% [1] or a hemoglobin level <8.3 g/dL. The rainy season in Mangochi occurs from December through April; the postrainy season is from May through August. Education was defined as ever attending school and/or literacy.
Analyses were done with SAS software (SAS Institute, Cary, NC). Pearson's χ2 test was used to assess significant differences in proportions. Unadjusted and adjusted logit odds with 95% confidence intervals (CIs) were calculated. Variables were added sequentially to the logistic models, starting with the malaria parameters. Because of the strong associations among malaria measures at delivery, only one such parameter was included in any given model. Parasitemia was dichotomized—presence versus absence of parasites—for most analyses.
We compared the following variables: all PTD or IUGR (including infants in each group who were both PTD and IUGR) to full-term normal birth weight-for-age (FNBW) infants and PTD only or IUGR only to FNBW infants. The results of the two analyses had similar patterns, so only the analyses with the greater power (e.g., those with all PTD or all IUGR compared with FNBW infants) are presented.
Results
Of 585 women enrolled, 471 could have delivered by the end of data collection; 193 (41%) of these delivered at MDH between 31 December 1995 and 31 August 1996. Among the 193 deliveries, 178 singleton deliveries had sufficient information to include in these analyses. The characteristics of the study population are summarized in table 1.
Table 1
Characteristics of 178 women delivering at Mangochi District Hospital, Malawi.
When we compared all PTD infants to all FNBW infants, each delivery parasitemia measure (cord blood, placenta, post-delivery peripheral parasitemia) was significantly associated with PTD. Women whose infants had cord blood parasitemias generally had the highest maternal postdelivery and placental parasite densities (P <.001, t tests of mean placental or peripheral parasite density in mother-infant pairs with or without cord blood parasitemias). Neither of the two antenatal malaria measures was associated with PTD (table 2). Antenatal SP use was associated with neither PTD nor delivery parasitemia.
Table 2
Bivariate and multivariate associations for preterm delivery and intrauterine growth retardation (IUGR) versus full-term normal weight-for-age (FNBW) infants.
Mothers delivering in the rainy season were >4 times more likely to deliver before term than were women delivering after the rainy season (table 2). Mothers delivering during the rainy season also were more likely to have placental (odds ratio [OR] = 2.71; 95% CI, 1.4–5.3) or postdelivery parasitemia (OR = 2.96; 95% CI, 1.4–6.3) than women delivering after the rainy season (cord blood parasitemia: OR = 1.91; 95% CI, 0.8–4.6). Neither anemia in the antenatal period nor relative placental overweight were associated with PTD (table 2).
Each delivery parameter remained significant when education was included in the model. The similarity of the adjusted and unadjusted ORs indicated that malaria at delivery was not confounded by education. Delivery in the rainy season remained significant after controlling for education, but more than one ANC visit was not associated with PTD when controlling for education. When season was entered into the models, the only delivery parasitemia measure remaining significant was cord blood parasitemia (table 2). There was no significant interaction between malaria and either education or season. Antenatal measures of parasitemia, which were not significant in the bivariate analyses, were not significant in the multivariate models.
Women with documented parasitemia or clinical episode of malaria in the antenatal period were more than 3 times more likely to deliver an IUGR infant than were other women (table 2). The association between IUGR and antenatal maternal parasitemia was not statistically significant, although the direction of association was similar. No measures of parasitemia at delivery were significantly associated with IUGR. Incidence of IUGR did not vary between the rainy and postrainy seasons.
Any SP use did not protect against IUGR (table 2), but the incidence of IUGR decreased as SP use increased (23% for no SP, 19% for 1 dose, and 5% for 2 doses; P = .07). Women with no education were 6 times more likely to deliver an IUGR infant than were women with any education (table 2). Education was not associated with parasitemia but was associated with receiving SP in the antenatal period (any education was protective: OR = 0.43; 95% CI, 0.2–1.0). Having only one ANC visit was also a risk factor for IUGR.
IUGR was not associated with anemia in the antenatal period but was associated with relative placental overweight (table 2). Women with relatively overweight placentas had lowermean hematocrit than other women (packed cell volume = 31% vs. 33%, P = .008). Anemia in women with overweight placentas may have been malaria-associated, since malaria at a given ANC visit was associated with a lower hematocrit at that visit (P ⩽ .0001, coefficient of log parasite density).
Malaria parasitemia or clinical malaria in the antenatal period remained a significant predictor of IUGR even when controlling for only one antenatal visit, relative placental overweight, and education (table 2). There were no significant interactions between malaria in the antenatal period and number of visits, relative placental overweight, or education. No other variables were significant in these models.
To evaluate potential loss to follow-up bias, we used enrollment data to compare women delivering at MDH with women delivering elsewhere. Compared with women delivering elsewhere, those delivering at MDH were more likely to have any education (61% vs. 37%, P = .001) and to live in town (26% vs. 15%; P = .001). Women living in town were also more likely to be educated (81% vs. 51% outside town). Women not delivering at MDH were more likely to be parasitemic at enrollment then women delivering at MDH (48% vs. 32%, respectively, P = .001).
In evaluating possible detection bias, we found that women with positive thick smears in the antenatal period had more antenatal visits (4 visits) than women without positive thick smears (3 visits; P < .0001). However, women receiving ⩾2 doses of SP also had a greater number of visits (4 visits) than women who received inadequate SP (3 visits; P = .0001). Although having only one ANC visit was associated with IUGR, women with IUGR infants had the same mean number of ANC visits as women with FNBW infants (i.e., women in both groups had similar numbers of opportunities for parasitemia detection). The number of documented parasitemias in the antenatal period was not associated with IUGR (P = .137). Women delivering preterm had fewer mean ANC visits (3 visits) than women with FNBW infants (4 visits; P = .01). There were no associations between measures of delivery parasitemia and number of ANC visits.
Discussion
In evaluating the hypothesis that timing of infection can affect malaria-associated PTD and IUGR, we investigated associations between antenatal and delivery parasitemias and IUGR or PTD in primigravid women. Parasitemia at delivery, especially cord blood parasitemia, was associated with PTD. Parasitemia or clinically diagnosed malaria in the antenatal period was associated with IUGR. These differences, among others, suggest that timing and severity of infection may be important in determining whether malaria-associated PTD or IUGR occurs.
The findings in this study are consistent with prior work in Mangochi [1]. Both studies found a strong association between cord blood parasitemia and preterm delivery. Our study had a relatively high incidence of PTD, likely because our funding restricted collection of delivery data to the rainy and postrainy seasons, compared with data for a full 3 years in the prior study. The magnitude of the association between placental parasitemia and IUGR was similar in both studies, although the association was not significant in our study (where the power to detect risks for IUGR was limited by the small number of IUGR births). On the basis of malaria at enrollment only, the previous study observed no association between antenatal malaria and IUGR. In our study, only 60% of women who were ultimately parasitemic in the antenatal period were parasitemic at enrollment, indicating that our analytical approach better characterized the antenatal period as a whole.
As the average number of ANC visits did not vary by IUGR status, it is unlikely that the association between antenatal malaria and IUGR was due to detection bias. The shorter duration of pregnancy may explain why women delivering before term had relatively fewer ANC visits; however, the possible under-detection of antenatal malaria in these women may have lead us to underestimate the association between antenatal malaria and PTD. If loss to follow-up was due to competing risk(s) associated with both malaria and PTD/IUGR, the resulting bias could have affected the associations observed. For example, the better-educated women of Mangochi were overrepresented in our study. These women had better access to malaria treatment and ANC care, and they had a relatively low risk of delivering IUGR or preterm infants. Considering the role of education in follow-up, SP use, and birth outcome, we may have underestimated the associations between malaria and IUGR or PTD. Consistency between this study and the previous study, which had somewhat different study populations, argues against substantial bias. While parasitemia at delivery appears causal for PTD, it is possible that an unmeasured third factor could be involved (e.g., human immunodeficiency virus infection, a documented risk factor for higher parasitemias) [12, 13].
Malaria infection in pregnancy contributes to low birth weight through both IUGR and PTD; however, the timing and severity of infection (including fetal involvement [14]) appears to influence the occurrence of IUGR or PTD. IUGR appears to result from placental insufficiency in the antenatal period. This mechanism has been suggested previously [3] and is supported by the associations among anemia, parasitemia, and placental overweight seen in this study. PTD may be caused by mechanisms more commonly seen in other infectious diseases (i.e., acute induction of prostaglandins and arachidonate lipogenase metabolites in late pregnancy) [15]. Understanding possible differences in malaria-associated low birth weight will help us adapt existing malaria prevention programs, reassessing the needs of women very late in pregnancy with respect to malaria prevention and treatment.
Acknowledgments
We thank James Akimu, Fineness Banda, Janet Bwanado, Edward Chikadza, Joyce Gondwe, Evelyn Kalamosi, Enipher Mtika, Daniel Nasambo, M. Pangani, Lucy Kasamale, and the MDH staff for data collection; the Malaria Research Project, Blantyre, and the Malawian Community Health Sciences Unit for valuable assistance; and M. Anthony Schork, Teija Kulmala, Malcolm Molyneux, Richard Steketee, Francine VerHoeff, and Fiona Yodell for suggestions and comments.
Footnotes
-
The study protocol was approved by the Malawi Health Science Research Committee and the University of Michigan Health Sciences Internal Review Board. Informed consent was obtained from all participants.
-
Grant support: Conservation, Food, and Health Foundation; International Rotary Foundation.
- Received July 24, 1998.
- Revision received December 28, 1998.
- © 1999 by the Infectious Diseases Society of America
References
This Article
- AbstractFree
- » Full Text (HTML)Free
- Full Text (PDF)Free
Classifications
Services
Citing Articles
Google Scholar
PubMed
Related Content
Share
- Email this article
-
CiteULike
Connotea
Delicious
Facebook
Google+
Mendeley
Twitter
Published on behalf of
IDSA Members: For your free access to this journal, log in via the IDSA members area.
The Journal
Impact Factor: 6.288
5-Yr impact factor: 6.103
For The Media
For Authors
- Services for authors
- Self-archiving policy
- IDSA Conflict of Interest policy
- Instructions for Authors
- ICMJE Form
- Submit Now!
Open access options for authors visit Oxford Open
This journal enables compliance with the NIH Public Access Policy
Corporate Services
Alerting Services
Related IDSA Publications
