| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
 |
ABSTRACT |
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Placental infection is a standard indicator widely used to characterize malaria infection in epidemiologic investigations but, as any single measure, it does not provide information about when in pregnancy a peripheral infection is the most harmful in terms of newborn’s birth weight. Few studies (for example in Malawi8) have related antenatal malaria in pregnant women to LBW. As far as we know, only a single research carried out in central Sudan12 analyzed the relation between the moment of infection in pregnancy and the decrease of birth weight and the risk of LBW. The subject thus remains poorly studied.
In a preceding paper,13 we used data collected during a
randomized trial of malaria prophylaxis in pregnant women in Burkina Faso in 1987–1988 during which repeated measures of peripheral infection were performed during follow-up, leading to an accurate description of the parasitological status of the mother during the overall pregnancy. At delivery, placental infection was assessed on placental smears and the newborns were weighed. We found that placental infection could be related to late malarial infection during pregnancy, but also to early infection (before 4 months of pregnancy). This showed the particular importance of the early pregnancy.
The aim of this study is to assess the consequences of the moment of infection during pregnancy on the newborn’s birth weight. Providing an answer to this question would be of utmost importance in terms of public health, as it may bring valuable information for the optimal duration of malaria prevention policies in pregnant women.
|
MATERIALS AND METHODS |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The details of the study have been described elsewhere.14 Briefly, any pregnant woman living in the area and attending the MCH center for a first prenatal visit between February 1987 and February 1988 was introduced to an investigator who obtained their informed consent and allocated them randomly to a chloroquine or a control group. Women belonging to the chloroquine group received one 300 mg oral dose of chloroquine weekly until delivery; women of the control group did not receive any prophylaxis. Once a week, each of the women was visited at home by an investigator who asked about febrile episodes or drug intake between visits. Peripheral thick blood films were made every 2 weeks and a placental blood smear was taken at delivery to look for malaria infection. In case of the association of a parasitemia with fever > 38°C, women (N = 6) were administered a standard treatment with chloroquine (25 mg/kg over 3 days). A questionnaire was completed to gather information about general characteristics, such as height, occupation of the mother, ethnicity, and district of residence. After delivery, the characteristics of the newborn child were noted (i.e., clinical status and birth weight within 24 hours of birth).
The field study received ethical clearance from the French Comité Consultatif National d’Ethique (CCNE) and from the Ministry of Health (Burkina Faso).
Selection of subjects. From the 1522 women in the initial cohort (after exclusion of 18 women having given birth to twins), 1190 newborns were weighed at delivery. Women with a missing measure of their baby’s birth weight were excluded from the analysis, and compared with selected women for age (continuous variable), parity (5 classes: 1, 2, 3–4, 5–6, > 6), district of residence (8 classes), ethnic group (7 classes), and intake of prophylaxis (yes or no).
Biologic methods. Peripheral and placental thick blood films were stained with Giemsa and sent twice a week to the ORSTOM laboratory in Bobo-Dioulasso, where microscopic examination was carried out for malaria parasites. Over 95% of identified parasites were Plasmodium falciparum.
Statistical analysis.
We compared included and excluded women using 
2 and Wilcoxon tests for qualitative and quantitative variables, respectively.
We used linear or logistic regression according to the status of the dependent variable newborn’s birth weight (continuous or binary defined as LBW < 2500 g). On the basis of the literature, we considered the following adjustment factors: duration of pregnancy (transformed into a four class variables corresponding to the quartiles), parity (primigravid, secundigravid, multigravid), sex of the newborn (male/female), woman’s body mass index (BMI) (weight [kg]/height squared [m2] and transformed into a binary variable—under or over the median), intake of effective prophylaxis before 6 months of pregnancy, and intake of effective prophylaxis after 6 months of pregnancy (one binary variable for each). Effective prophylaxis was defined as to a dose of chloroquine assumed to be effective (three 300 mg tablets) for each period of follow-up, before and after 6 months of pregnancy.
To account for the duration and position of the pregnancy with respect to the malaria transmission period (mid-June to mid-December15), we created a three-level categorical variable (Figure 1
). For instance, the first level corresponds to the last 0 to 6 months of gestation out of the transmission period, whereas the third level corresponds to the last 3 to 6 months of gestation during the transmission period.
|
collection: less than 4 months, 4–6 months, and 6 months or more. Positivity was defined as the presence of parasites in at least one blood sample, and negativity as the absence of parasites in all samples during each gestational age period. Peripheral parasitemia was not available in two main cases: either the woman had not entered the study at a given moment of the pregnancy, or (much less frequently) she was not home when the investigator visited her. For these two reasons, many data were unavailable or missing (66.2% in the first period of pregnancy, 25.3% in the second period of pregnancy, and 7.2% in the third period of pregnancy). To account for this, we used the MICE method (multiple imputation by chained equation) in the same way as in our preceding study.13 This is a multiple imputation method described by Van Buuren16 based on the Gibbs algorithm, implemented in the S+ language. We carried out 20 imputed data sets with 20 iterations for each, and checked the convergence by verifying that the results were unchanged by carrying out 20 more iterations.
Both univariate and multivariate linear and logistic regressions were carried out. In the univariate analysis birth weight or LBW were regressed against the three peripheral parasitemia variables only. In the multivariate analysis, the effect of these variables was adjusted for all of the cofactors cited earlier in this article.
The same analyses were carried out by replacing the three peripheral parasitemia variables by placental infection at delivery. Placental infection status was unknown for 276 women (of 1190), and MICE method was used to deal with this.
To analyze the impact of the number of detected peripheral infections on the occurrence of LBW, the subsample of women having had at least three collected blood samples during the follow-up was used. A variable with three categories was defined: no infection, one or two detected infections, and more than two infections. Then the relation between this variable and LBW was studied through univariate and multivariate logistic regression with adjustment for the number of blood samples collected during the follow-up. To test for a trend of increasing risk of LBW according to the number of peripheral infection detected throughout the pregnancy, a semi-quantitative variable was introduced in the multivariate model. (This variable took three values corresponding to the mean of the number of detected infections within the three categories defined earlier: 0, 1.2, and 3.7 respectively.) Two-sided confidence intervals and P values were computed and are reported throughout this article. All analyses were done using SAS 8.2 (SAS Institute Inc., Cary, NC) and S + 2000 (Statistical Sciences, Inc., Seattle, WA).
|
RESULTS |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
However, the proportion of ethnic groups differed, P < 0.001, Mossi and Lobi-Senoufo were over-represented and
Peuls were under-represented in the analyzed group, as well as intake of prophylaxis, P = 0.007, with a higher proportion of the excluded women not having received any prophylaxis (54.5% versus 46.2% in the analyzed group).
The study sample included 1190 women. General characteristics of the women are presented in Table 1.
|
The median birth weight was 2960 g (interquartile range: [2700, 3220]). The mean age of the pregnant women was rather high, probably reflecting the fact that the women deliver until a relatively old age in Burkina Faso, with high parities.
When birth weight was analyzed in relation to malaria infection, unadjusted estimates showed a highly significant association between infection during all of the three periods of pregnancy and decrease in mean birth weight (Table 2). Univariate analysis showed that mean birth weight was significantly lower among infants born from mothers infected compared with those from mothers who were not infected in each period. After adjustment, the peripheral infection at the end of pregnancy remained significantly related to birth weight, and there was a decreasing trend in mean birth weight when peripheral infection occurred at the beginning of pregnancy. When the mother was infected in the middle of pregnancy, the decrease of mean birth weight disappeared after adjustment.
|
Placental infection was significantly related to a decrease of mean birth weight in the univariate analysis (–120 g, 95% confidence interval (CI): [–213, –27]; P = 0.01), but this was no longer the case after adjustment for cofactors (–40 g, 95% CI: [–128, 48]; P = 0.37).
The crude analysis showed that the peripheral infection is significantly correlated with the newborn’s LBW in each of the three periods of pregnancy. After adjustment for potential confounders, only the peripheral infection at the end of pregnancy was significantly associated with LBW with a twofold increase in the odds (Table 3).
|
When placental infection was introduced into the model instead of the three peripheral infection variables, the corresponding odds-ratio was significant by univariate analysis (odds-ratio [OR] = 2.1; 95% CI: [1.3, 3.3]; P = 0.002), but not after adjustment for cofactors (Adjusted Odds-ratio (AOR) = 1.4; 95% CI: [0.8, 2.4]; P = 0.23).
The analysis of the relation between the number of detected infections and LBW is shown in Table 4. There were 1027 women who had more than two collected blood samples. There is a clear trend of increasing risk of LBW with the increase in the number of malaria infections detected. The test for trend was significant (AOR corresponding to the semi-quantitative variable is 1.3; 95% CI: [1.1, 1.6]; P = 0.009).
|
|
DISCUSSION |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The main reason for excluding women from the study (due to a lack of measure of birth weight) was delivery out of the Province hospital (at home or in another clinic). Distance to the hospital does not seem to explain this situation because the district of residence was not related to the missing measure of birth weight. Women of the prophylaxis group (over-represented in the analyzed group) may have been more inclined for a complete follow-up of their pregnancy in the same structure, and thus to deliver at the Province hospital. Though there was a difference in the overall parity between groups, it should be noted that it affected only the grand multipara (> 6). First and second pregnancies, more prone to have an influence on birth weight, were similar in both groups. No obvious reason appears to justify the higher rate of in-hospital deliveries among the Mossi and Lobi-Senoufo ethnic groups, especially as this difference is not repeated
among districts of residence, which are ethnically homogenous. Moreover, it was further verified that ethnic group was not related to birth weight, and other factors (parity and intake of prophylaxis) have been taken into account in multivariate statistical analysis. Thus, it is unlikely that these small differences have led to the wrong conclusions.
In our study, gestational age was derived from the date of the last menstruation cycle as declared by the women. Other methods have led to more accurate estimates (such as ultrasound examination) but they were not available in our study. Declared gestational age is potentially subject to misclassification, but it is unlikely to be related to peripheral infection or any other risk factor, and such a misclassification should not affect the overall risk of LBW. Moreover, the mean gestational age obtained with this variable was quite plausible, and this variable was strongly related to the birth weight.
Because 66% women were included in our study after 4 months of pregnancy there was a large amount of unavailable data concerning malarial infection in early pregnancy. This is not surprising, as most studies performed in sub-Saharan Africa have shown that women generally attend their first antenatal visit only in the second half of pregnancy.17 In such settings, it is thus inevitable to deal with large amounts of missing data. Nevertheless, peripheral infection status during the first period of pregnancy was identified for 403 women—the most extensive information collected until now on malaria infection in early pregnancy. Statistical studies have shown that limiting the analysis to those observations with complete data drastically reduces the sample size (and thus the power of statistical tests) and can also lead to selection bias.18 We then used the sophisticated multiple imputation method (MICE) we already used in our last work.
Univariate analysis confirmed the usual relation between placental infection and LBW,19 but the relation was no longer significant after adjustment for cofactors, even if the adjusted OR was > 1. Others studies found similar results, with a non significant relation between placental infection and decrease of birth weight.20,21 This result does not imply that placental infection by itself cannot influence birth weight, but that the follow-up of peripheral parasitemia during the entire pregnancy can tell us more about the risk of LBW than placental infection, which is the standard indicator. However, as placental infection is easier to detect than peripheral parasitemia during pregnancy, it still remains the most useful indicator for public health studies.
This analysis has led to interesting results concerning the moment in pregnancy when the occurrence of malarial infection is the most liable to negatively affect the newborn’s birth weight. Our data suggest a differential effect, depending on the period of the pregnancy when infection occurred. Infection during the end of the pregnancy was significantly related to both the risk of LBW and the decrease of mean birth weight, whereas a trend (P < 0.1) was found between an early infection in pregnancy and a decrease in mean birth weight, probably due to a lack of power in the statistical tests, as there was a large amount of unknown data during this period of pregnancy. All this is consistent with our preceding work, which showed that these early infections, as well as late infections, were significantly related to an increased probability of placental infection at delivery13 and also with other studies in Malawi22 or in Kenya,23 suggesting that the beginning of pregnancy could be a period of particular importance with possible harmful consequences at delivery. However, our data have not shown that such early infections in pregnancy lead to an increased risk of LBW.
As longitudinal studies involving a sufficient number of subjects are scarce, there is little information on the parasitological status of the mother during pregnancy. Only one study carried out in central Sudan during 1989 and 1990 in close to 1100 pregnant women12 aimed to relate the newborn’s birth weight and malarial infection at different periods of pregnancy. Results showed that the risk of LBW was highest when malarial infection occurred in the first and, to a lesser extent, the second trimester, and not at the end of pregnancy. Our data suggest a relation between an early maternal infection (
4 months of pregnancy) and a decrease in mean birth weight (even if not statistically significant), but also a much stronger effect at the end of pregnancy. Such a discrepancy between the two studies may be attributed to distinct transmission patterns (low and occasionally unstable malaria in Sudan, stable seasonal malaria in Burkina Faso) or differences in the exposed populations. As our data were collected prospectively in a randomized trial, whereas the data from Sudan were retrospective, based on a questionnaire at delivery, the former should be less subject to misclassification. Like the Sudanese study, a clear trend of increasing risk of LBW with a greater number of infections during pregnancy was found, even when after adjusting for potential confounders.
Our results concerning the effect of the infection period on birth weight are of importance in terms of public health and prevention strategies. As stated earlier in this article, the first strategy to limit the consequences of infection by malaria in pregnant women was a chloroquine-based prophylaxis and this proved at first to be useful in West Africa. However, due to the rapid spread of parasite resistance against chloroquine, associated with poor compliance in pregnant women in most African settings, the strategy did not live up to its expectations. WHO has recently reconsidered its strategy and has proposed an intermittent treatment (IPTp). This consists of two doses of sulfadoxine-pyrimethamine at antenatal visits, which cannot be given in the first trimester, because of potential teratogenic effects of the folic acid antagonists.24 It is recommended to start the treatment after the first active fetal movements have appeared (at approximately 20 weeks of pregnancy).
It is generally believed that women rarely attend antenatal clinics (ANC) in the beginning of their pregnancies. However, as mentioned earlier, in our study, one third of the included women were seen in ANC visits during the first period of pregnancy, without any particular coaxing from the trial’s promoters.
IPTp, as it is presently offered in most African countries, is certainly an effective way of reducing the risk of LBW by eliminating late pregnancy parasitemias, which seem to have the most severe consequences. However, it cannot prevent early pregnancy infections, which may also influence birth outcomes. Because we have shown that a higher number of infections during the entire course of pregnancy raises the probability of delivering a LBW baby, early prevention of malaria in pregnant women is likely to reduce the length of exposure to infection and hence the risk of delivering a LBW baby. Possible solutions are the systematic use of preventive measures such as insecticide-treated nets at the very beginning of pregnancy and the possible replacement of Sulfadoxine-Pyrimethamine with a drug that would not be contra-indicated during the first trimester of pregnancy, if available in the next future.
Received September 13, 2006. Accepted for publication October 24, 2006.
Acknowledgements: The authors thank Dr. Hugo Pilkington, for comments regarding the manuscript and the English, Dr. Alzuma Yada, former head of the Banfora Health District, and all the midwives who participated in the study.
Financial support: This study was supported by a grant from the INSERM (Institut National de la Santé et de la Recherche Médicale): réseau Nord-Sud n°486 NS2.
* Address correspondence to Gilles Cottrell, Faculté de Pharmacie Laboratoire de Parasitologie, 4 Avenue de l’Observatoire, 75270 Paris Cedex 06. E-mail: cottrell{at}ird.fr 
Authors’ addresses: Gilles Cottrell and Michel Cot, Faculté de Pharmacie, Laboratoire de Parasitologie, 4 Avenue de l’Observatoire 75270 Paris Cedex 06, France. Jean-Yves Mary, INSERM U717, Université Paris 7, DBIM, Hôpital St Louis 1 av Claude Vellefaux, 75010 Paris, France. Drissa Barro, Service d’Anesthésie-Réanimation au Centre Hospitalier Universitaire Souro Sanou de Bobo Dioulasso, B.P 676.
Reprint requests: Gilles Cottrell, Faculté de Pharmacie Laboratoire de Parasitologie, 4 Avenue de l’Observatoire 75270 Paris Cedex 06. Telephone: (33) 1 53 73 15 06, E-mail: cottrell{at}ird.fr.
|
REFERENCES |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
represents a delivery within the considered interval.