Women who live in areas with endemic malaria and who are pregnant for the first or second time are more likely to be infected with Plasmodium falciparum malaria than non-pregnant women of a similar age (Brabin 1983). This infection contributes to antenatal anaemia (Brabin 1990; Duffy 2001) and slows fetal growth (Kramer 1987), which may harm the mother and baby.
Drugs have been widely used to prevent infection or its consequences. From the 1980s, prophylaxis to prevent, suppress, or eradicate malaria parasites with a variety of drugs has been tested. From the 1990s, a modification of this called intermittent preventive treatment (IPT) has been used, where women are treated for malaria presumptively at fixed times during the pregnancy, usually drugs with a long half life, such as sulfadoxine-pyrimethamine. There remains debate as to whether the mechanism of IPT actually differs a lot from prophylaxis (White 2005), but the regimens are very different. IPT requires just two or three doses during pregnancy, compared to prophylaxis regimens that may be daily (eg with proguanil) or weekly (eg with chloroquine). Prophylaxis and IPT are in addition to good care during pregnancy, which includes prompt treatment of women when they present clinically with fever or anaemia. Surprisingly, this latter area is remarkably under-researched (Orton 2005).
The research and policy question addressed in this review is whether giving drugs on a regular basis (prophylaxis or IPT) to prevent malaria has additional advantages in terms of health outcomes over prompt, appropriate treatment. The advantages with women taking regular, routine antimalarial drugs are that the drug will treat or suppress parasites in the blood and reduce the chances of illness or anaemia developing. In turn, this may ultimately benefit the fetus and impact on birthweight and long-term survival. The disadvantages of routine use of antimalarial drugs are that many drugs have adverse effects and that widespread use may contribute to the malaria parasite developing resistance to these drugs; and it requires health systems to help ensure their implementation. The question is first addressed in terms of all pregnant women, and then secondly in women of low parity, where the effects of malaria are more marked.
One of the main problems in evaluating the effects of routine antimalarial drugs on the mother and the infant is identifying meaningful outcomes in terms of the direct impact on mother or baby. This review has identified two pragmatic outcome measures:
- Severe antenatal anaemia: there is a strong association between anaemia and poor maternal outcomes, so preventing severe anaemia is likely to be of benefit.
- Perinatal death: preventing death of the fetus and in the first week of life is one of the main reasons for giving malaria prophylaxis.
To assess drugs given to prevent malaria infection and its consequences in pregnant women living in malarial areas. This includes prophylaxis and intermittent preventive treatment (IPT).
Criteria for considering studies for this review
Types of studies
Randomized and quasi-randomized controlled trials.
Types of participants
Pregnant women living in endemic malaria areas.
Types of interventions
Antimalarial drug prophylaxis (eg chloroquine given weekly) or IPT (typically sulfadoxine-pyrimethamine given two to three times during pregnancy).
- No regular or routine antimalarial drugs given; women receive malaria treatments for fever, anaemia, or clinical malaria, according to local protocols.
- Comparator regimens of prophylaxis or IPT.
Types of outcome measures
- Antenatal hospital admission.
- Obstetric complications:
- Caesarean section.
- Complicated labour.
- Severe anaemia*, defined as haemoglobin level less than 8 g/L; or haematocrit equivalent.
- Anaemia (any).
- Mean haematocrit.
- Mean haemoglobin.
- Need for a blood transfusion.
- Malaria infection:
- Antenatal parasitaemia.
- Incidence (number of women infected at least once; number treated for suspected malaria).
- Placental malaria.
- Adverse events leading to discontinuation of the drug or hospitalization.
- Perinatal death* (death of the baby after 22
ndweek of pregnancy and first seven days after birth).
- Neonatal death.
- Infant death.
- Preterm birth.
- Mean birthweight.
- Low birthweight.
- High birthweight.
- Malaria infection.
*Primary outcome measures for the review.
Search methods for identification of studies
We attempted to identify all relevant trials regardless of language or publication status (published, unpublished, in press, and in progress).
We searched the following databases using the search terms and strategy described in Appendix 1: Cochrane Infectious Diseases Group Specialized Register (March 2006); Central Register of Controlled Trials (CENTRAL), published in The Cochrane Library (2006, Issue 1); MEDLINE (1966 to March 2006); EMBASE (1974 to March 2006); and LILACS (1982 to March 2006).
We contacted researchers working in the field for unpublished data, confidential reports, and raw data of published trials.
Data collection and analysis
Selection of studies
Over the last 10 years, PG has periodically assessed titles and abstracts of articles identified by the literature search. Potentially relevant studies (describing the use of antimalarial drugs in pregnancy) were retrieved and examined. If they were potentially a trial, both authors applied the inclusion criteria and then compared their decisions; differences were resolved by discussion. Trials that appeared relevant but did not meet the inclusion criteria are listed in the 'Characteristics of excluded studies'.
Data extraction and management
Both authors extracted data in most updates; in the 2006 update, PG extracted data and MG cross checked the papers for accuracy. Data were entered into Review Manager 5.
Assessment of risk of bias in included studies
We independently assessed the trials' methodological quality (risk of bias). We judged the method used to generate the allocation sequence to be adequate if it was described and the resulting sequences were unpredictable, unclear if it was stated that the trial is randomized but the method was not described, and inadequate if the sequences could be related to prognosis. We considered the method used to conceal allocation to be adequate if the participants and investigators enrolling participants could not foresee assignment, unclear if the trial was described as randomized but the method was not described, and inadequate if the participants and investigators enrolling participants could foresee the upcoming assignment. Blinding was classified as double (use a placebo or a double dummy technique such that neither the participant or care provider/assessor know which treatment is given), single (the participant or care provider/assessor is aware of the treatment given), or open (all parties are aware of treatment). In our assessment of losses to follow up, we assessed the loss between the initial cohort and those women included in the final analysis of outcomes. Where information appeared to have been collected but was not fully reported, we approached the trial authors for further details.
We used Review Manager 5 for the analysis. We separated trials into those evaluating the intervention in all parity groups (generally the older trials) and those evaluating prophylaxis in low-parity women (parity one and two). Trials were grouped by those that evaluated prophylaxis and those evaluating IPT. We used risk ratios (RR) for dichotomous variables and mean differences (MD) for mean values. In the absence of heterogeneity we used a fixed-effect model with 95% confidence intervals (CI), and where we detected heterogeneity, we used a random-effects model for the meta-analysis. Weighted averages were calculated where required. In trials where standard deviations for birthweight were not available, we used an estimate of 500 g.
Description of studies
Sixteen trials met the inclusion criteria (see 'Characteristics of included studies'); one trial had two arms (Parise 1998i; Parise 1998ii), making 17 separate comparisons. One large trial with 4220 participants from Malawi was intended to be randomized (Steketee 1996), but this did not appear to happen in practice; this and the other excluded studies are detailed in the 'Characteristics of excluded studies'.
One trial was carried out in an area with unstable malaria in Thailand (Nosten 1994). All other trials were carried out in highly endemic areas in countries in Africa: Burkina Faso (Cot 1992); Cameroon (Cot 1995); The Gambia (Greenwood 1989; Menendez 1994); Kenya (Parise 1998i; Parise 1998ii; Shulman 1999); Malawi (Schultz 1994); Mali (Kayentao 2005); Mozambique (Challis 2004); Nigeria (Morley 1964; Fleming 1985; Nahlen 1989); Tanzania (Mutabingwa 1991); and Uganda (Hamilton 1972; Ndyomugyenyi 2000). Drug resistance varied over time and was incompletely reported in the papers.
The review includes 12,638 women (Appendix 2). All women regardless of number of previous pregnancies were included in seven trials (Morley 1964; Hamilton 1972; Greenwood 1989; Nahlen 1989; Mutabingwa 1991; Cot 1992; Nosten 1994); however, two of these trials selected women in their first pregnancy for follow up or reporting (Greenwood 1989; Mutabingwa 1991). Women in their first pregnancy were exclusively recruited in six trials (Fleming 1985; Menendez 1994; Cot 1995; Shulman 1999; Ndyomugyenyi 2000; Challis 2004), and women in their first or second pregnancy were recruited in three trials (Parise 1998i; Parise 1998ii; Schultz 1994; Kayentao 2005).
Ten prophylaxis trials used a placebo or no antimalarial drug as the control (Appendix 3) . Different intervention drugs were used: chloroquine (Hamilton 1972; Cot 1992; Cot 1995; Ndyomugyenyi 2000); pyrimethamine (Morley 1964; Nahlen 1989); proguanil (Fleming 1985); pyrimethamine-dapsone (Greenwood 1989; Menendez 1994); and mefloquine (Nosten 1994). One trial compared chloroquine prophylaxis with proguanil (Mutabingwa 1991).
Three trials evaluated IPT with sulfadoxine-pyrimethamine against placebo (Parise 1998; Shulman 1999; Challis 2004), and two evaluated IPT against chloroquine prophylaxis (Schultz 1994; Kayentao 2005).
Three trial reports describe access to treatment in the event of a malarial illness (Morley 1964; Nosten 1994; Shulman 1999), while the other trials did not describe the services available to the trial participants.
Oral iron and folic acid supplements were used in seven trials in various combinations.
Risk of bias in included studies
As shown in Appendix 4, three trials used an adequate method to generate the allocation sequence (Fleming 1985; Shulman 1999; Ndyomugyenyi 2000), seven used an inadequate method (alternate allocation or allocation by day of attendance at clinic), and the method was unclear in six trials. Only two trials concealed allocation (Nosten 1994; Shulman 1999), while 10 trials did not conceal allocation or used inadequate methods to do this, and the methods used were unclear in four trials. For blinding, three trials reported both that health staff and women were blinded from their intervention group (Nosten 1994; Shulman 1999; Challis 2004). One trial was described as double blind (Ndyomugyenyi 2000) and another blinded the patients but was not clear about health staff (Fleming 1985); the other trials did not use blinding.
The percentage of randomized participants included in the analyses varied from 24% to 100% for maternal outcomes, and 62% to 100% for fetal outcomes (Appendix 5).
The two trials from The Gambia trials were randomized by cluster (Greenwood 1989; Menendez 1994), but the analysis did not take the cluster-design effects into consideration. However, pregnancy is a relatively infrequent event and the compounds (which were the units of randomization) generally have only a small number of houses. We therefore used the data as they were presented in the paper and ignored any possible design effects.
Effects of interventions
We examined trials allocating women to prophylaxis or IPT versus a control (placebo or no routine drug), first in trials that recruited and reported women of all parity groups, and then those recruiting and reporting women of low parity. We then examined head-to-head comparisons.
1. Any antimalarial drug versus no drug
1.1. Women of all parity groups
Six trials tested the overall benefit to women of routine drugs given during pregnancy: five were carried out in Africa (Morley 1964; Hamilton 1972; Greenwood 1989; Nahlen 1989; Cot 1992); and one was carried out in an area with unstable malaria in Thailand (Nosten 1994).
No statistically significant difference between the two groups was reported for women with any anaemia (Nosten 1994, 311 participants, Analysis 1.4), mean values of haematocrit (Greenwood 1989, 276 participants, Analysis 1.5), or the need for a blood transfusion (Nosten 1994, 339 participants, Analysis 1.6).
Fewer women in the group that received antimalarial drugs had antenatal parasitaemia (RR 0.53, 95% CI 0.33 to 0.86; 328 participants, 2 trials, Analysis 1.7), at least one malaria infection (RR 0.14, 95% CI 0.06 to 0.34; 337 participants, 1 trial, Analysis 1.8), placental malaria (RR 0.34, 95% CI 0.26 to 0.45; 1236 participants, 3 trials, Analysis 1.9), and episodes of fever (RR 0.42, 95% CI 0.27 to 0.66; 227 participants, 1 trial, Analysis 1.10).
There was no statistically significant difference in the number of perinatal deaths in the meta-analysis of the four trials that reported this (2890 participants, 4 trials, Analysis 2.1), with no significant heterogeneity. Nosten 1994 had a trend of more perinatal deaths in the prophylaxis group that bordered on significance (RR 3.51, 95% CI 1.00 to 12.32; 311 participants). The analyses for stillbirth (1493 participants, 2 trials, Analysis 2.2), neonatal death (419 participants, 1 trial, Analysis 2.3), and infant death (288 participants, 2 trials, Analysis 2.4) were underpowered and infrequently reported, with no significant differences demonstrated.
Four trials reported mean birthweight (2671 participants, Analysis 2.6). There was no significant difference in number with low birthweight (1438 participants, Analysis 2.7) or high birthweight (287 participants, Analysis 2.8, although there was significant heterogeneity between the trials (P = 0.010).
1.2. Women having their first or second baby
Seven trials reported on women in their first or second pregnancy (Fleming 1985; Greenwood 1989; Menendez 1994; Cot 1995; Shulman 1999; Ndyomugyenyi 2000; Challis 2004). In the analyses, we have stratified trials by whether women were given prophylaxis or IPT.
The two trials that reported this outcome measure both used prophylaxis (Greenwood 1989; Ndyomugyenyi 2000). They had too few participants to detect an effect on maternal death (772 participants, 2 trials, Analysis 3.1).
Two small trials that used prophylaxis measured the number of women having a Caesarean section (Fleming 1985; Cot 1995) and the meta-analysis showed no significant difference (294 participants, Analysis 3.2).
The group of women who received antimalarial drugs had fewer instances of severe antenatal anaemia (RR 0.62, 95% CI 0.50 to 0.78; 2809 participants, 1 prophylaxis trial, 2 IPT trials, Analysis 3.3) and any anaemia (RR 0.67, 95% CI 0.47 to 0.95; 566 participants, 1 prophylaxis trial, Analysis 3.4. This was also the case for mean haematocrit (MD 2.63, 95% CI 1.36 to 3.90; 118 participants, 2 prophylaxis trials, Analysis 3.5) and mean haemoglobin (MD 0.40, 95% CI 0.23 to 0.56; 1677 participants, 1 prophylaxis trial and 1 IPT trial, Analysis 3.6.
Routine drug administration was associated with fewer women with parasites in the antenatal period (RR 0.27, 95% CI 0.17 to 0.44, random-effects model; 2906 participants, 3 prophylaxis trials and 3 IPT trials, Analysis 3.7); there was heterogeneity between the trials, but the direction of effect was consistent. The antimalarial drugs also had an effect on reducing the incidence of women having at least one malaria infection (RR 0.12, 95% 0.02 to 0.93; 83 participants, 1 prophylaxis trial, Analysis 2.8), the need for treatment for suspected malaria (RR 0.41, 95% CI 0.18 to 0.91; 133 participants, 1 prophylaxis trial, Analysis 3.9, and the number of women with placental malaria in the three prophylaxis trials (RR 0.71, 95% CI 0.60 to 0.85; 573 participants) and the two IPT trials (RR 0.35, 95% CI 0.27 to 0.47; 1232 participants; Analysis 3.10).
A meta-analysis of two prophylaxis trials and one IPT trial found that the use of antimalarial drugs is associated with fewer perinatal deaths (RR 0.73, 95% CI 0.53 to 0.99; 1986 participants, 2 prophylaxis and 1 IPT trial, Analysis 4.1). A sensitivity analysis of the one adequately concealed trial showed that the effect on perinatal deaths was smaller than in the other two trials (RR 0.78, 95% CI 0.52 to 1.17; 1237 participants, Shulman 1999, Analysis 4.1). For stillbirth, no impact was detected (3454 participants, Analysis 4.2. In the three trials reporting neonatal deaths, the trend was towards protection, but again this was not statistically significant (2505 participants, Analysis 4.3). There was also no significant difference in the number of infant deaths in one small trial (349 participants, Analysis 4.4).
One prophylaxis trial reported no statistically significant difference in the number of preterm births (1051 participants, Analysis 4.5).
Babies born to women using antimalarial drugs had a higher mean birthweight (MD 126.70 g, 95% CI 88.64 to 164.75; 2648 participants, 8 trials, Analysis 4.6 and were less likely to have a low birthweight (RR 0.57, 95% CI 0.46 to 0.72; 2350 participants, 6 trials, Analysis 4.7. One small trial reported on high birthweight, but there was no significant difference between the groups (121 participants, Analysis 4.8).
Two trials reported on newborn malaria infection; one suggesting an effect and the other not showing a significant difference (639 participants, Analysis 4.9).
2. Head-to-head comparisons
2.1. Proguanil versus chloroquine
One trial carried out in Tanzania in women of all parities examined this comparison (Mutabingwa 1991). Compared with chloroquine, proguanil was associated with fewer fever episodes (RR 0.69, 95% CI 0.59 to 0.81; 223 participants, Analysis 5.3) and fewer women with antenatal parasitaemia (RR 0.80, 95% CI 0.71 to 0.91; 223 participants, Analysis 5.4), but there was no difference for haemoglobin (200 participants, Analysis 5.2). The difference in mean birthweight tended towards favouring proguanil, but this was not significant (197 participants, Analysis 6.3).
2.2. Sulfadoxine-pyrimethamine versus chloroquine
Two trials examined this comparison; both were conducted in Africa in low-parity women (Schultz 1994; Kayentao 2005). There was no significant difference between the two interventions for the number of women with severe anaemia (717 participants, 1 trial, Analysis 5.1), but fewer women in the sulfadoxine-pyrimethamine group had antenatal parasitaemia (RR 0.71, 95% CI 0.56 to 0.91; 848 participants, 2 trials, Analysis 5.4) and placental malaria (RR 0.75, 95% CI 0.60 to 0.95; 831 participants, 2 trials Analysis 5.5).
There was no significant difference for the number of neonatal deaths (696 participants, 1 trial, Analysis 6.1), preterm births (828 participants, 2 trials, Analysis 6.2), or mean birthweight (719 participants, 1 trial, Analysis 6.3); however, low birthweight was less common with sulfadoxine-pyrimethamine (RR 0.77, 95% CI 0.61 to 0.97; 828 participants, 2 trials, Analysis 6.4).
Prophylaxis with a variety of antimalarial drugs or IPT with sulfadoxine-pyrimethamine in women having their first or second baby is associated with fewer women having severe anaemia and fewer perinatal deaths. The size of the effect for severe anaemia is considerable, while the size of the effect with perinatal mortality is more modest. There was no obvious effect when prophylaxis or IPT was given to all women regardless of the number of previous pregnancies. Malaria parasitaemia in the blood or the placenta is also less common with prophylaxis or IPT.
For infants, the review shows emerging evidence of an effect on perinatal death in low-parity women. This also corresponds with the effect seen on birthweight, with fewer low birthweight infants in these women. Women with more than two previous pregnancies may accrue small benefits from prophylaxis, but there are insufficient data to demonstrate or refute this.
In Fleming 1985, some women in the intervention group also received iron or folic acid, or both, and in these women growth (height) also increased during pregnancy. The nutritional supplementation (not the prophylaxis) appeared to assist maternal growth in pregnancy and reduce the risk of Caesarean section.
We anticipated that there may be differences in effect between women living in areas where malaria is endemic and those living in epidemic areas. We know that host immunity means that malaria illness in semi-immune women is less severe than in women who have no host immunity. Thus it is reasonable to expect there to be differences in effects of preventive measures for malaria in pregnancy between women living in areas where malaria is highly endemic and women have acquired some host resistance compared with women living in areas where malaria is less common or occurs in epidemics and a malaria infection is likely to be associated with a more severe illness. We were unable to stratify the analysis to explore for these differences because there were too few trials.
Trials on this question are not easy to conduct, particularly when mortality outcomes are sought, and a meta-analysis has something to offer by combining results from the different trials. Apart from the trial from The Gambia (Greenwood 1989), the recording of pregnancy outcome data in the trials was largely dependent on delivery in hospital. The large number of women lost or excluded from the final analyses could potentially mean the more disadvantaged groups were excluded from the results.
Much of the research emphasis has been on drugs for prevention, rather than prompt and appropriate treatment of illness episodes, including anaemia. Pregnant women need access to services that can provide treatment for illness episodes and anaemia. It may be that prophylaxis is an appropriate approach where such services are not available because the health system is insufficient to provide clinical care for illness episodes. An additional advantage of IPT could be in terms of adherence.
Implications for practice
Routine antimalarial drugs have been shown to reduce antenatal parasitaemia and fever in pregnant women living in areas with endemic malaria. For women in their first or second pregnancy, this intervention reduces the instances of severe antenatal anaemia, antenatal parasitaemia, and perinatal deaths, and it has a positive effect on birthweight.
Even if quite modest, these effects may well be worth pursuing. The sulfadoxine-pyrimethamine regimen is feasible and practical to implement. There are public health consequences to giving drugs regularly to women during pregnancy in terms of increasing the risk of drug resistance developing.
There remains a question over what policies to adopt when sulfadoxine-pyrimethamine resistance rises. As impregnated mosquito nets have been shown to be effective in pregnancy (Gamble 2006), then IPT may be used in combination with them. Nets, if used in combination with IPT or prophylaxis, may assist in preventing drug resistant strains of the parasite from spreading.
Implications for research
The confidence intervals for the meta-analysis for perinatal death in women in their first or second pregnancy just reaches statistical significance, but the point estimate indicates that the potential impact of this intervention on death could be high. Until the appropriate research is done, we cannot be sure how effective the intervention is in reducing perinatal deaths.
Large simple trials implemented through routine health services measuring mortality outcomes are widely used in areas outside malaria. Such a trial could compare prophylaxis or IPT with prompt regular treatment of morbidity in the mother (usually fever or anaemia). The outcomes should examine the effects on pregnancy outcome and death in the neonate and infant. Establishing whether there is an impact of routine prophylaxis or IPT is likely to re-emerge as an issue if sulfadoxine-pyrimethamine becomes ineffective through drug resistance, and policy makers are then faced with the potential of changing to effective drugs.
Iron deficiency is an important cause of anaemia in areas with malaria, so researchers contemplating a trial should consider a factorial design to examine also the impact of routine iron supplementation in malaria areas.
On earlier versions of this review, we thank Iain Chalmers and Bernard Brabin for their helpful comments; and François Nosten, Alice Greenwood, TK Mutabingwa, Michel Cot, Caroline Shulman, and Monica Parise for contributing unpublished data; and funding from the European Commission (Directorate General XII) in Belgium and the World Health Organization.
This document is an output from a project funded by the UK Department for International Development (DFID) for the benefit of developing countries. The views expressed are not necessarily those of DFID.
Data and analyses
- Top of page
- Authors' conclusions
- Data and analyses
- What's new
- Contributions of authors
- Declarations of interest
- Sources of support
- Index terms
Appendix 1. Search methods: detailed search strategies
Appendix 2. Trial participants: number of previous pregnancies
Appendix 3. Comparisons evaluated in the trials
Appendix 4. Methodological quality (risk of bias)
Appendix 5. Percentage of randomized participants included in the analyses
Last assessed as up-to-date: 19 August 2006.
Protocol first published: Issue 1, 1995
Review first published: Issue 1, 1995
Contributions of authors
Paul Garner wrote the first version of this review in 1995. Metin Gülmezoglu carried out the first update, and Paul Garner has been updating the review since then in consultation with Metin Gülmezoglu.
Declarations of interest
Sources of support
- Liverpool School of Tropical Medicine, UK.
- HRP-UNDP/UNFPA/WHO/World Bank Special Programme in Human Reproduction, Geneva, Switzerland.
- Department for International Development (DFID), UK.
Medical Subject Headings (MeSH)
Antimalarials [*therapeutic use]; Infant, Newborn; Malaria [drug therapy; *prevention & control]; Mosquito Control; Pregnancy Complications, Parasitic [drug therapy; *prevention & control]; Randomized Controlled Trials as Topic
MeSH check words
Female; Humans; Pregnancy
* Indicates the major publication for the study