Safety, efficacy and determinants of effectiveness of antimalarial drugs during pregnancy: implications for prevention programmes in Plasmodium falciparum-endemic sub-Saharan Africa

Authors


Summary

Plasmodium falciparum malaria in pregnancy poses substantial risk to a pregnant woman and her neonate through anaemia and low birth weight (LBW), respectively, and is responsible for up to 35% of preventable LBW in malaria-endemic areas. Chemoprophylaxis or intermittent preventive treatment (IPT) with an effective antimalarial can ameliorate the adverse effects of malaria during pregnancy. Current guidelines from the WHO recommend that women in highly malarious areas receive IPT with an effective antimalarial. Two central considerations in evaluating drugs for use during pregnancy are safety for the mother and her foetus and effectiveness, which is determined by efficacy, cost, availability, deliverability and acceptability of the drug. These factors may be scored and potential drugs or drug combinations ranked in order of potential effectiveness for use in prevention programmes. The seven most promising regimens are all IPT, primarily because they are more easily delivered and less expensive than chemoprophylaxis. Currently, IPT with sulphadoxine–pyrimethamine (SP) is more likely to have the best overall effectiveness in preventing adverse outcomes associated with malaria in pregnancy. Its low cost, wide availability, easy deliverability and acceptability make it the clear choice in countries where efficacy of the drug remains good. For countries where resistance to SP is rising or already high, amodiaquine (alone or in combination with SP or artesunate) artesunate + SP, chlorproguanil–dapsone (with and without artesunate) and artemether–lumefantrine require urgent evaluation for use in pregnancy.

Introduction

Malaria is an enormous global health problem and most of the disease burden affects young children and pregnant women. The adverse impact of malaria in pregnant women is largely caused by Plasmodium falciparum; approximately 90% of P. falciparum clinical cases globally occur in sub-Saharan Africa. This year, there will be approximately 50 million pregnancies in women living in malarious areas, of which approximately half will occur in sub-Saharan Africa (Steketee et al. 2001).

Malarial infection during pregnancy poses substantial risk to the mother, her foetus and the neonate. In areas of low transmission of P. falciparum, women do not acquire substantial antimalarial immunity and are susceptible to episodes of severe malaria, which may result in stillbirths, spontaneous abortions, or maternal death. In areas of high transmission of P. falciparum, where adult women have considerable acquired immunity, women may have asymptomatic infections or be minimally symptomatic, but such infections can contribute to maternal anaemia and cause placental parasitaemia, both of which may subsequently lead to low birth weight (LBW) (Steketee et al. 1996c). Prevalence of parasitaemia is greatest in the second trimester (Brabin 1983), and susceptibility to clinical malaria appears higher in both second and third trimesters (Diagne et al. 2000). Although there are fewer data about the role of P. vivax, there is evidence that it may also lead to anaemia and LBW (Nosten et al. 1999a). LBW is an important contributor to neonatal mortality (McCormick 1985; McDermott et al. 1996). It is estimated that malaria in pregnancy is responsible for 5–12% of all LBW, 35% of LBW that is preventable during pregnancy (Steketee et al. 1996a), and contributes to 75,000–200,000 infant deaths each year (Steketee et al. 2001).

Current guidelines from the WHO Expert Committee on Malaria suggest a package of interventions for pregnant women to combat malaria during pregnancy (WHO 2000a). Those interventions include intermittent preventive treatment (IPT), insecticide-treated nets, malaria case management and treatment of anaemia. Two of these interventions, IPT and case management, require the use of antimalarial drugs. In this paper, we review antimalarial drugs and their potential for programmatic use to prevent the consequences of malaria in pregnancy.

The two major issues for considering antimalarial drug use in a programme for prevention or management of malaria in pregnant women are safety and effectiveness. First, ‘is the drug toxic to the woman or foetus during pregnancy, or to the infant during lactation?’ Safety is, of course, a central tenet of malaria prevention. Secondly, ‘is the drug use strategy and its implementation likely to have its desired effect – reduce the burden of malaria in pregnancy?’ Programme effectiveness is determined by the efficacy of the drug against the parasite and the characteristics of the drug when used programmatically. These characteristics include: affordability and availability of the drug; deliverability (dosing requirements and ease of incorporation into existing antenatal care delivery systems); and acceptability to the population. Although a drug may be safe and efficacious, it may prove a poor choice for prevention programmes because of high cost, poor availability, unfavourable dosing scheme, or unacceptable side effects; such a drug will fail in the transition from efficacy to effectiveness.

Methods

We searched the medical literature from 1966 to the present using both Pubmed and Ovid by several strategies: (1) each drug name as a keyword, (2) ‘antimalarials’ and ‘pregnancy’, and (3) ‘malaria’ and ‘pregnancy’. Relevant articles were then reviewed for inclusion. Where appropriate, articles prior to 1966 and chapters in textbooks were also reviewed. A scale for estimating the programme effectiveness potential for each drug was developed a priori (as outlined below).

Safety

For an antimalarial to be considered for use in pregnancy, it must be safe for the mother, the foetus and later for the breast-feeding infant. However, safety is a relative term and it is difficult to unequivocally prove safety. There are inherent limitations in these drug safety data in pregnancy. For some drugs, there are years of programmatic experience to suggest that a drug is generally safe for use during pregnancy. Sometimes there are no data at all from use in pregnant women. At other times, a drug may be a new combination of component drugs where the individual drugs are considered to be safe in pregnancy, but for which there are no data on the combination. Ultimately, we have widely varying amounts of information on antimalarial drugs that allow us to broadly categorize them as (1) useful for a pregnant woman, as there are extensive data from programmatic use during pregnancy, (2) possibly useful for a pregnant woman, but more data are needed, or (3) not useful for a pregnant woman because they have known adverse events associated with their use in pregnancy, and safe and efficacious alternatives exist. Drugs in this third category are not considered further in the paper, as efficacy and effectiveness are irrelevant in the face of poor or questionable safety. Finally, at the end of this section, we briefly review safety concerning antimalarial use during lactation.

Drugs generally considered safe in pregnancy

A number of antimalarial compounds have received widespread programmatic use during pregnancy, and are thought to be safe, although overdoses and idiosyncratic reactions could lead to detrimental effects on the mother and/or the foetus.

Chloroquine (CQ), perhaps the most widely used antimalarial, is generally considered safe in all trimesters of pregnancy (Phillips-Howard & Wood 1996; WHO 1979). Children born to a cohort of 169 non-immune women who took CQ chemoprophylaxis throughout pregnancy had no more birth defects than 454 births to women who had not received CQ (Wolfe & Cordero 1985). Among >2500 women who received CQ (as IPT or weekly chemoprophylaxis), there was no reported increase in abortions, stillbirths, or congenital abnormalities, although there were frequent non-severe side effects such as itching, dizziness and gastrointestinal complaints (Steketee et al. 1996d). There are reports of increased spontaneous abortions, particularly in patients with systemic lupus erythematosus treated with high doses of CQ over prolonged periods (Levy et al. 1991), and in some settings, CQ has gained a reputation as an abortifacient at higher doses. CQ overdoses have been responsible for numerous deaths (Weniger 1979), but its abortifacient effects appear to be limited to these very high doses that are life-threatening to the mother.

In many countries, quinine remains the principal treatment for severe malaria. It has been associated with teratogenic effects and damage to foetal optic and auditory nerves when taken at very high (abortifacient) doses (West 1938; McKinna 1966; Dannenberg et al. 1983; Briggs et al. 1998), but should be considered safe in pregnancy when taken at normal therapeutic doses (Looareesuwan et al. 1985; Phillips-Howard & Wood 1996). There is, however, a risk of hyperinsulinaemia and subsequent hypoglycaemia in women who take quinine (Looareesuwan et al. 1985). The related compound, quinidine, is also considered safe in pregnancy (Briggs et al. 1998). There are no reports of congenital abnormalities associated with its use during pregnancy, although there have been reports of neonatal thrombocytopenia after maternal use (Briggs et al. 1998).

Proguanil is generally considered safe during pregnancy. A cohort of pregnant travellers who took proguanil in combination with CQ had no higher rates of spontaneous abortions or congenital anomalies than the expected background rate (Phillips-Howard et al. 1998). A study in Nigeria found no increase in adverse outcomes among pregnant women receiving daily proguanil (100 mg) in addition to weekly CQ (Fleming et al. 1986). Another study in Tanzania found no increase in adverse outcomes among women receiving proguanil alone or in combination with weekly CQ compared with those receiving CQ alone (Mutabingwa et al. 1993).

Sulphonamides are also generally considered safe in the second and third trimesters of pregnancy. Although there is very limited evidence that sulpha drugs may be associated with kernicterus when given to premature neonates (Silverman et al. 1956), this problem has not been noted in studies of IPT where sulphadoxine–pyrimethamine (SP) was administered to the mother (Parise et al. 1998; Verhoeff et al. 1998; Shulman et al. 1999). Studies examining the risk to the foetus from in utero exposure to SP combination have generally not found any increased risk in spontaneous abortions or congenital defects (Anonymous 1983; Parise et al. 1998; Verhoeff et al. 1998). A retrospective study of antifolate drugs given before and during pregnancy found that there was an increased risk of birth defects when such drugs were taken during the first trimester, but not during the second or third trimester (Hernandez-Diaz et al. 2000). When given weekly as prophylaxis, SP has been associated with rare and severe cutaneous reactions such as toxic epidermal necrolysis and Stevens–Johnson syndrome (Miller et al. 1986); there is no evidence that this risk is any greater in pregnant women. In summary, these drugs are considered safe in the second and third trimesters of pregnancy.

Pyrimethamine is usually given in combination with sulphadoxine (see section above). However, studies in which pyrimethamine has been given alone have also found no increase in adverse pregnancy outcomes (Morley et al. 1964).

Dapsone has been used extensively in pregnant women with leprosy, without reported adverse effects (Kahn 1985). Studies in which dapsone was given in combination with pyrimethamine (Maloprim1) for malaria chemoprophylaxis during pregnancy had no increase in adverse pregnancy outcomes (Greenwood et al. 1989, 1994; Menendez et al. 1994). However, a severe hypersensitivity syndrome has been described in a non-pregnant woman receiving weekly Maloprim (Grayson et al. 1988). Additionally, occasional reports have been received of non-pregnant patients developing agranulocytosis following Maloprim chemoprophylaxis, although usually after twice-weekly dosing, which is no longer recommended (Bruce-Chwatt & Hutchinson 1983). A study using dapsone in combination with SP in children found no serious adverse effects (Mutabingwa et al. 2001b). Dapsone is most often used currently with chlorproguanil (as Lapdap, see below).

Clindamycin has been used routinely as an antibiotic in pregnancy without any evidence of adverse effects (Dinsmoor & Gibbs 1988; Zambrano 1991; Briggs et al. 1998). Although the drug does cross the placenta and does accumulate in foetal tissues perhaps to the point of therapeutic levels (Philipson et al. 1973), there is no evidence that this accumulation is harmful. A recent trial in quinine comparing quinine–clindamycin with artesunate for the treatment of falciparum malaria during pregnancy found no serious adverse events, no increase in stillbirths or congenital anomalies above expected levels, and no negative impact on infant development (McGready et al. 2001b).

Drugs with questionable safety during pregnancy or with insufficient data

For a number of other drugs there are some concerns of safety in pregnancy, or there are limited data about their use during pregnancy.

Amodiaquine (AQ), a 4-aminoquinoline related to CQ, falls under the category of antimalarials about which there are insufficient data to be certain about their use in pregnancy. In non-pregnant women taking AQ for chemoprophylaxis, there have been reports of agranulocytosis and granulocytopenia (Hatton et al. 1986; Neftel et al. 1986), hepatitis (Larrey et al. 1986; Bernuau et al. 1988), and increases in serum aspartate aminotransferase (AST) levels (Sturchler et al. 1987). A study comparing AQ with atovaquone + proguanil found no serious adverse events, but found that pruritis, weakness, insomnia and dizziness were more common in the AQ group (Radloff et al. 1996). However, a recent systematic review of studies of AQ for the treatment of malaria found no increase in adverse events when compared with CQ or SP, and found that all adverse events were minor or moderate, and not life-threatening (Olliaro & Mussano 2000). A recently completed three-country trial in Africa found no cases of clinical hepatitis among children >10 years receiving either AQ or AQ + artesunate, but did find that 6% of children developed neutropenia (neutrophil count <1000/μl) and that 60% of children experienced a decline in serial neutrophil counts (Adjuik et al. 2002). In a smaller study involving younger (6–59 month) Tanzanian children, no serious adverse effects of AQ were noted, and no increase in neutropenia was noted when compared with children receiving SP (Schellenberg et al. 2002). In Burma, AQ and quinine were compared for treatment of malaria in pregnant women. A high overall rate of spontaneous abortion was noted, but was not stratified by treatment type (Thet et al. 1988).

Chlorproguanil (Lapudrine), in combination with dapsone (in development as a fixed combination: Lapdap), has only been evaluated in a single published trial among pregnant women (Keuter et al. 1990). However, the study did not specifically address whether any adverse foetal outcomes were observed. There are two studies in children that have demonstrated good safety of Lapdap (Mutabingwa et al. 2001a; Sulo et al. 2002), although one trial did find a higher rate of severe anaemia among children treated with Lapdap than among those treated with SP (Sulo et al. 2002). Given that the component drugs are both considered safe in pregnancy, it is expected that the combination will also be safe for use in pregnancy.

Mefloquine (MQ) is a quinolinemethanol compound (Bruce-Chwatt et al. 1986) that has been used extensively, particularly in Asia for the treatment of P. falciparum. It has also been used extensively for malaria chemoprophylaxis among travellers to areas with CQ-resistant P. falciparum malaria (Croft & Garner 1997).

Safety of MQ during pregnancy has been evaluated through postmarketing surveillance, retrospective and prospective studies. An early dose-finding study in Thailand found no increase in adverse pregnancy outcomes among women who took either 125 or 250 mg/week for prophylaxis during the third trimester (Nosten et al. 1990). Post-marketing data collected by the manufacturers of MQ showed that among 1627 women exposed to the drug during pregnancy (95% for chemoprophylaxis), there was no increase in congenital malformations over the expected background rate (Vanhauwere et al. 1998). In one prospective study of pregnant women, MQ chemoprophylaxis was more frequently associated with stillbirths than SP chemoprophylaxis, (9.1%vs. 2.6%), but this rate did not differ from the background rate of stillbirths among the population studied (7–10%) (Phillips-Howard et al. 1998). Among 451 women in Malawi who took MQ treatment (750 mg) followed by weekly MQ chemoprophylaxis (250 mg), no increase in the incidence of stillbirths or spontaneous abortions was observed (Steketee et al. 1996d). Among US soldiers in Somalia, 72 women used MQ for chemoprophylaxis before learning of their pregnancies, and a greater expected percentage (16.7%) had spontaneous abortions (Smoak et al. 1997). Data from two studies in Thailand have showed no difference in infant development between infants born to mothers given MQ chemoprophylaxis during pregnancy and those given placebo (Nosten et al. 1994), or between women treated with MQ + artesunate and those treated with quinine (McGready et al. 2000).

Data from a recent retrospective study in Thailand suggest a significantly increased risk of stillbirth among women exposed to treatment doses of MQ during pregnancy compared with those exposed to quinine [odds ratio (OR) 4.72], other treatments (OR 5.10) or women who had no malaria (OR 3.50) (Nosten et al. 1999b). However, no general patterns of physical abnormalities or specific defects were observed in the stillbirths from MQ-using women.

Although most of the data from a variety of sources suggest that MQ is safe for use in pregnancy, these recent data from Thailand highlight the need for continued vigilance in monitoring adverse events of women treated with MQ in pregnancy.

Artemisinins, a group of related compounds (sesquiterpene lactones) also known by their Chinese name, Qinghaosu, are derived from the medicinal herb Artemisia annua (L.), which is also known as annual or sweet wormwood (Hien & White 1993). Its antimalarial properties appear to have been known in ancient times, but were rediscovered in China in the early 1970s (Hien & White 1993). Artemisinins are available in a variety of oral (artemisinin, artesunate, artemether and dihydroartemisinin), parenteral (artemether, arteether and artesunate), and rectal (artesunate and dihydroartemisinin) formulations WHO (1998). These compounds have received widespread attention in recent years in the treatment of severe malaria, and in the treatment of multidrug-resistant falciparum malaria, particularly in Southeast Asia.

There are limited data on the use of artemisinins during pregnancy. A small case series from China found that among six pregnant women treated for malaria (P. falciparum or P. vivax) at a mean of 21.7 weeks of gestation, none had adverse outcomes (Wang 1989). In another trial from China, seven children exposed in utero between 17 and 27 weeks of gestation were tracked after birth from 3 to 10 years; no adverse outcomes were found (Li 1990). In another clinical trial in China, 21 pregnant women were included and were given a variety of artemisinins; no adverse outcomes were recorded (Li et al. 1994). A study in 83 women in Thailand treated with either artesunate or artemether found no increase in adverse outcomes (4% spontaneous abortion and 3% stillbirth – all of which were explainable by other events). Sixteen of the women in this study were accidentally exposed to artemisinins during the first trimester. Follow-up of the live born children from this cohort found no developmental delay (McGready et al. 1998). Further work in Thailand has shown that women treated with MQ + artesunate (n = 66) (McGready et al. 2000) or 7 days of artesunate alone (McGready et al. 2001b) had no increase in adverse effects or adverse birth outcomes, and no negative developmental impact when compared with women treated with quinine. A further 461 women in Thailand treated with either artesunate (n = 528 treatment courses) or artemether (n = 11 treatment courses) for P. falciparum malaria had no increase in rates of abortion, stillbirth, congenital abnormality, or mean gestation at delivery (McGready et al. 2001a). A total of 287 pregnant women in the Gambia were exposed to artesunate in combination with SP during a mass drug administration; no difference was noted in the rates of abortions, stillbirths, or infant deaths among those exposed or not exposed to the drugs (Deen et al. 2001).

A recent consultation held at WHO has recommended that: (1) because of the limited experience with the artemisinins in pregnancy they should only be used when other treatments are considered unsuitable, (2) presently, they cannot be recommended for treatment of malaria in the first trimester. They should not, however, be withheld if they are considered lifesaving for the mother, (3) to further document the safety of artemisinin compounds in pregnancy, careful follow-up is required, with documentation of pregnancy outcome and the subsequent development of the child whenever possible, and (4) to guide further development of policies on the use of artemisinin derivatives during pregnancy, alone or in combination, there is an urgent need for further research/documentation of their efficacy and safety for use as therapy for malaria and for IPT (B. Nahlen, personal communication).

Malarone, a fixed combination of atovaquone (250 mg) and proguanil (100 mg), is currently being evaluated in pregnant women. (See earlier section on proguanil). In clinical trials among non-pregnant persons, Malarone appears to have an excellent safety profile when used at the dosage recommended for prophylaxis of P. falciparum. The most commonly reported adverse effects include gastrointestinal disturbances (Shanks et al. 1998; Sukwa et al. 1999; van der Berg et al. 1999; Hogh et al. 2000; Overbosch et al. 2001) and headache (Sukwa et al. 1999; van der Berg et al. 1999; Overbosch et al. 2001). There were no significant differences in moderate to severe adverse effects in the randomized, placebo-controlled, double-blind studies (Lell et al. 1998; Shanks et al. 1998; Sukwa et al. 1999).

Generally, atovaquone/proguanil is as well or better tolerated than most drugs at doses necessary for treatment of malaria (Radloff et al. 1996; de Alencar et al. 1997; Anabwani et al. 1999; Bustos et al. 1999; Looareesuwan et al. 1999b; Mulenga et al. 1999), although there are not yet any published data on the use of Malarone for treatment of pregnant women with malaria. Treatment limiting adverse events occur in <1% of patients receiving treatment doses, and serious adverse effects attributable to treatment doses are rare (Looareesuwan et al. 1999a). The most common adverse effects at treatment doses include vomiting, nausea and abdominal pain (Radloff et al. 1996; Sabchareon et al. 1998; Anabwani et al. 1999; Bustos et al. 1999; Looareesuwan et al. 1999b,c; Mulenga et al. 1999; Bouchaud et al. 2000). Elevated alanine aminotransferase (ALT) and AST occurred in patients treated with atovaquone-proguanil and at a greater frequency than patients treated with MQ; however, the differences were not significant and values returned to normal by day 28 in most patients (Looareesuwan et al. 1999b). Although the clinical significance of these elevations is unknown, studies have not shown liver enzyme elevations to be treatment-limiting (Looareesuwan et al. 1999b).

Azithromycin is another compound for which there are limited data about safety in pregnancy. There are several trials in which azithromycin was used in the treatment of sexually transmitted diseases and other genital infections during pregnancy; no adverse neonatal outcomes were noted (Adair et al. 1998; Ogasawara & Goodwin 1999). Azithromycin does not readily cross the placenta (Heikkinen et al. 2000), and the drug appears to be generally safe for use in pregnancy (Duff 1997; Donders 2000). There are no published data evaluating the safety of azithromycin for the treatment or prevention of malaria during pregnancy.

Lumefantrine, formerly known as benflumetol, is a novel aryl amino alcohol antimalarial related to quinine, mefloquine and halofantrine. For clinical use, it has been combined with artemether in a fixed combination pill of 120 mg lumefantrine and 20 mg of artemether (Riamet, Coartem). Although an early trial in African children suggested a possible prolongation of the QT interval after therapy (von Seidlein et al. 1997), other evaluations have found no evidence of QT prolongation or other cardiotoxicity (van Vugt et al. 1999a; Bindschedler et al. 2002). Administration of quinine following co-artemether (Riamet) resulted in a greater risk of QTc prolongation than among patients administered only quinine (Lefevre et al. 2002). In general, the safety of artemether–lumefantrine has been good in both children (von Seidlein et al. 1997) and adults (van Vugt et al. 1999b), but the drug has not been evaluated in pregnant women.

Drugs generally considered contraindicated during pregnancy

Certain antimalarials are generally considered contraindicated in pregnancy because of their effects on the foetus. Two antibacterials, tetracycline and doxycycline, fall into this category. Tetracycline easily crosses the placenta, and can lead to disturbances of skeletal growth, permanent discoloration of teeth, corneas and lenses (Cohlan et al. 1961, 1963). Additionally, tetracycline has been associated with an increased hepatotoxicity among pregnant women, especially in the last trimester (Dowling & Lepper 1964; Whalley et al. 1964; Kunelis et al. 1965). Doxycycline, a closely related compound, has been presumed to be capable of causing similar tooth discoloration, although recent data suggest that perhaps because of its lower binding affinity for calcium than tetracycline, doxycycline does not cause clinically significant staining of permanent teeth following use in childhood (Lochary et al. 1998; Purvis & Edwards 2000).

Primaquine (PQ), an 8-aminoquinolone used primarily for the radical cure of P. vivax and P. ovale infections, but also used as chemoprophylaxis against P. falciparum, is also generally considered contraindicated in pregnancy. Persons who have a deficiency of glucose-6-phosphate dehydrogenase (G6PD) are at increased risk of acute hemolytic events associated with the administration of PQ (Beutler 1991). There is a theoretical increased risk of haemolysis and subsequent jaundice among G6PD deficient infants whose mothers have received PQ. This risk would be greatest in Africa, where the gene responsible for G6PD deficiency is most common. Because of the theoretical risk, the recommendation for laboratory screening for G6PD deficiency, the ability to delay radical cure with PQ until after pregnancy, and the availability of other antimalarials, PQ is not recommended for use in pregnancy.

Tafenoquine (also known as WR 238605) is, like PQ, an 8-aminoquinolone. It has been shown to be effective prophylaxis against P. falciparum infections in semi-immune teenagers and young adults (Lell et al. 2000; Shanks et al. 2001), and effective in preventing relapse with P. vivax (Walsh et al. 1999). There are currently no published data about its safety in pregnancy, as trials to date have specifically excluded pregnant or lactating women. However, it is more likely to have properties similar to other 8-aminoquinolone, and therefore pose a theoretical risk to women and neonates with G6PD deficiency, and therefore be inappropriate for use during pregnancy.

Other antimalarials have been associated with serious side effects in non-pregnant adults and should therefore also be used with caution in pregnant women. Halofantrine has been associated with lengthening of the QT interval, and with fatal arrhythmias in some persons (Nosten et al. 1993; Monlun et al. 1995; Touze et al. 1996). Current recommendations suggest that this drug be used for treatment only in those who have a documented normal electrocardiogram, which makes its use in many developing world settings impractical (Monlun et al. 1995; Anonymous 1997).

The questionable safety of these drugs (tetracycline, doxycycline, PQ, tafenoquine and halofantrine) in pregnant women does not imply that they are completely contraindicated. In the face of serious illness and in settings where a limited number of drugs are available, it is necessary to balance the risk to the life of the mother (whose death could also lead to foetal demise), with hypothetical risks to the foetus.

Antimalarial safety during lactation

A number of antimalarials are considered safe during lactation, including: CQ, quinine, SP, pyrimethamine, dapsone, clindamycin, mefloquine, doxycycline and tetracycline (American Academy of Pediatrics (AAP) Committee on Drugs 1994; Briggs et al. 1998). Six of these (CQ, quinine, pyrimethamine, dapsone, clindamycin and tetracycline) are specifically mentioned as compatible with breastfeeding by the AAP (American Academy of Pediatrics (AAP) Committee on Drugs 1994). There is some evidence that caution should be exercised when using sulphonamides in infants who are sick, premature, or have G6PD deficiency; because these drugs are present in breast milk, lactation under these conditions should take place with caution (American Academy of Pediatrics (AAP) Committee on Drugs 1994). For other drugs, there are no data about secretion into breast milk or use during lactation. However, as proguanil is considered safe for use in pregnancy, and AQ, Lapdap, azithromycin and Coartem are considered safe for use in infancy, it is more likely that these drugs would not pose serious harm to the infant through lactation.

Efficacy

The efficacy of individual or combination antimalarials varies widely by region and changes over time. There are often few data about the efficacy of a specific antimalarial in pregnancy; in the absence of these data, data on efficacy in older children and adolescents (age 5–14 years) may be indicative of likely efficacy in pregnant women. Therapeutic efficacy testing in Malawi has demonstrated that resistance rates of P. falciparum to CQ are similar between pregnant women and children 5–10 years of age, and that these levels are approximately half of resistance levels identified in children <5 years of age (CDC 1995). In addition to achieving clinical cure, an antimalarial in pregnancy should also clear placental parasitaemia, which has been associated with LBW (Steketee et al. 1988, 1996b; Parise et al. 1998). We have therefore elected to review the overall efficacy of antimalarials that might be used programmatically in pregnancy, with a focus on pregnancy-specific data where available.

The most widely used drug for the treatment of malaria worldwide is CQ. Resistance to CQ is widespread and has been described in all countries with P. falciparum malaria except for the island of Hispaniola (Haiti and the Dominican Republic), countries in Central America north-west of the Panama Canal and a few areas of the Middle East (Bloland & Ettling 1999). However, CQ may still be relatively efficacious in many countries in West Africa with P. falciparum transmission.

Many studies have examined the use of CQ chemoprophylaxis during pregnancy and effect on placental parasitaemia, anaemia, and LBW. A randomized controlled trial in Cameroon found that in spite of moderate resistance of P. falciparum to CQ, directly observed chemoprophylaxis of primigravidae with weekly CQ (300 mg) was associated with a decrease in placental parasitaemia and LBW, and an increase in mean birth weight (Cot et al. 1995). In Cameroon and Burkina Faso, women who received CQ chemoprophylaxis during pregnancy had significantly higher haematocrit values than women who had not received chemoprophylaxis (Cot et al. 1998). Women in Zaire who reported having received CQ chemoprophylaxis during pregnancy (duration unknown) were only 39% as likely as women who reported no CQ chemoprophylaxis to deliver a LBW neonate (Nyirjesy et al. 1993). An evaluation of weekly CQ (directly observed), in combination with daily proguanil (self-dosed at home) in Mali, showed a 55% reduction in moderate-to-severe anaemia (Bouvier et al. 1997b), and an increase of 429 g in the birth weight of neonates born to primigravidae and secundigravidae (when taken for 20 weeks or more) (Bouvier et al. 1997a).

Although there are data regarding the use of AQ for treatment during pregnancy (Thet et al. 1988), there are no completed studies (McDermott et al. 1988) regarding its use for malaria prevention during pregnancy. In some settings with CQ resistance, AQ has been found to be more efficacious than CQ among non-pregnant individuals (Olliaro et al. 1996; Brasseur et al. 1999). Given these findings, it is more likely that AQ would perform similarly to CQ, with perhaps slightly better efficacy in areas with increasing CQ-resistant P. falciparum.

Given rising rates of SP resistance, there has been recent interest in using AQ in combination with SP (McIntosh & Greenwood 1998), in part as an inexpensive way to delay the generation of resistance and to improve treatment efficacy. This combination has not been tried in pregnancy, but a recent trial in young children conducted in Tanzania demonstrated better efficacy (fewer treatment failures) in children treated with AQ + SP than in children treated with AQ or SP alone (Schellenberg et al. 2002).

Sulphadoxine–pyrimethamine has had broad use in the treatment of P. falciparum malaria in both pregnant and non-pregnant women worldwide. A trial in Malawi showed that placental parasitaemia rates were lower in women who received IPT as two courses of SP (once in the second and once in the third trimester) (9%), than in women who received SP treatment followed by weekly CQ chemoprophylaxis (26%), or women who received CQ treatment followed by weekly CQ chemoprophylaxis (32%) (Schultz et al. 1994b). Among women in Kenya who received an IPT regimen of two courses of SP, the rate of placental parasitaemia was lower among women who were treated with SP for febrile episodes (case management) (12%vs. 27%) (Parise et al. 1998). A significant reduction in severe anaemia (39% protective efficacy) was observed among a different group of Kenyan women treated with IPT (one to three treatment courses of SP) when compared with controls, as was a reduction, although not statistically significant, in neonatal deaths among those in the SP group (Shulman et al. 1999). Among Malawian babies born to primigravidae treated with SP during pregnancy, a significant reduction in LBW was found when compared with those not treated with SP (Verhoeff et al. 1998). Two or three courses were more efficacious than one course in reducing the incidence of LBW. Anaemia reduction was more pronounced among multigravidae treated with SP than among primigravidae, although this may have been due in part to micronutrient supplementation (Verhoeff et al. 1998). In Malawi, where two-course IPT with SP has been national policy since 1993, prescription of SP was associated with a decrease in placental parasitaemia (33% no prescription vs. 23%≥2 courses SP), and a decrease in LBW (23% no prescription vs. 10%≥2 courses SP), among women delivering at a large urban hospital (Rogerson et al. 2000).

In most settings with CQ-resistant P. falciparum, MQ continues to be an effective drug for both treatment and chemoprophylaxis. Among pregnant women, studies have also found that the drug has excellent efficacy in clearance of parasitaemia and reduction of malaria-associated adverse outcomes. Among Thai women treated with 500 mg of MQ, followed by chemoprophylaxis with 250 mg of MQ weekly for 4 weeks, and then 125 mg weekly until term, there was 86% protection against P. falciparum infection, but no effect on birth weight or the incidence of LBW among delivered infants (Nosten et al. 1994). Malawian women who received a treatment dose of MQ (750 mg) followed by weekly MQ chemoprophylaxis (250 mg) were less likely to have persistent or breakthrough parasitaemia than those women treated with any of three CQ regimens in comparison groups (Steketee et al. 1996e). That same study also found that women in the MQ group had significantly fewer LBW infants than women treated with CQ-containing regimens (Steketee et al. 1996b). Although there is evidence that pregnant women need larger doses of MQ to achieve comparable blood levels (Na Bangchang et al. 1994), there is also evidence that pregnant women living in areas with little or no MQ resistance can clear P. falciparum parasitaemia with a single reduced dose (12.5 mg/kg) of MQ (Okoyeh et al. 1996).

Maloprim, a fixed combination of pyrimethamine 12.5 mg and dapsone 100 mg, has also been evaluated as a potential chemoprophylaxis drug for use during pregnancy. One study in the Gambia found that primigravidae who received fortnightly Maloprim given by traditional birth attendants (TBAs) had higher mean packed cell volume (30.1 vs. 26.6), delivered infants with higher mean birth weights (mean: 159 g), and fewer LBW infants (6%vs. 22%) than those not receiving Maloprim (Greenwood et al. 1989). Among Gambian women treated with weekly Maloprim by TBAs starting at an average of 24 weeks’ gestation there was a significant decrease in placental parasitaemia, and a significant increase in the birth weight of their infants (mean 153 g increase); however, there was no correlation between placental infection and birth weight (Menendez et al. 1994). When several of the studies in the Gambia were evaluated together, it was estimated that the chemoprophylaxis programme might be expected to reduce infant mortality by up to 18% (Greenwood et al. 1992).

Another potential drug for use in pregnant women is chlorproguanil–dapsone (Lapdap). This combination has been shown to be efficacious in treating non-pregnant persons in a small study in Kenya (100% cure on day 7) (Watkins et al. 1988), but less efficacious in the treatment of acute uncomplicated falciparum malaria in Thailand (10% cure rate). Another study in Kenya found that although clearance of parasitaemia was good following both one- and three-course chlorproguanil–dapsone treatment (93.4% and 98.0%), reinfection rates were just as rapid as those seen in community surveillance (Amukoye et al. 1997). In Kenya and Malawi, research has shown that children treated with chlorproguanil–dapsone did not have higher retreatment rates than children treated with SP, and that treatment failures were less common (Sulo et al. 2002). In Tanzania, chlorproguanil–dapsone was more efficacious than SP at treating drug-resistant falciparum malaria in children <5 years of age (Mutabingwa et al. 2001a). A trial in pregnant women in Kenya comparing a single treatment course with SP, CQ, or chlorproguanil (1.2 mg/kg) and dapsone (2.4 mg/kg) given as a single dose found that chlorproguanil–dapsone was as effective as SP in clearing initial parasitaemia by day 7, but less effective in maintaining parasite clearance by day 28 (67% parasitemic for chlorproguanil–dapsone vs. 19% for SP) (Keuter et al. 1990). A rise in haemoglobin concentrations was observed in all the three groups, but was sustained until day 42 only among those women who remained free of malaria parasites. The study did not examine placental parasitaemia or birth weight. The short half-life of chlorproguanil–dapsone, compared with SP, appears to result in poorer long-term freedom from peripheral parasitaemia, which one might hypothesize would result in poorer prevention of placental parasitaemia. However, there is evidence that the shorter half-life of chlorproguanil–dapsone may exert less selective pressure for drug resistance than longer-acting drugs like SP (Nzila et al. 2000).

There are very few studies that show the efficacy of monotherapy with artemisinins for the treatment of malaria in pregnant women. In a descriptive, non-randomized study along the Thai–Burmese border, only 16% of pregnant women failed treatment with 12 mg/kg of oral artesunate over 7 days (recrudescence within 42 days), which compared favourably with quinine and MQ in that setting (McGready et al. 1998). However, this study did not examine maternal anaemia, placental parasitaemia, or birth weight. Among 287 pregnant women in the Gambia exposed to artesunate in combination with SP during a mass drug administration, the mean birth weight was 0.48 kg higher among the exposed than among the unexposed group (Deen et al. 2001). To date, no studies evaluating artemisinins for IPT during pregnancy have been undertaken, although an evaluation of combination therapy with an artemisinin and SP during pregnancy is planned for Tanzania (Peter Bloland, CDC, personal communication).

Amodiaquine + artesunate is another possible combination for use in pregnant women, although no trials have yet been conducted. However, a recent multicenter African trial in children found improved efficacy in treating falciparum malaria in two of three countries among those treated with AQ + artesunate when compared with AQ alone (Adjuik et al. 2002).

There are no published studies evaluating the efficacy of Malarone in treating or preventing malaria in pregnant women. However, early studies with this drug indicate that it is highly efficacious in the treatment of P. falciparum malaria in men and non-pregnant women (Radloff et al. 1996; Bustos et al. 1999; Bouchaud et al. 2000). Multiple studies have demonstrated high efficacy of Malarone when used as chemoprophylaxis in both non-immune travelers (Hogh et al. 2000; Overbosch et al. 2001), semi-immune adults (Shanks et al. 1998; Sukwa et al. 1999) and semi-immune children (Lell et al. 1998).

There are no studies evaluating artemether–lumefantrine in pregnant women. However, 100% of Gambian children with uncomplicated P. falciparum malaria treated with 3 days of artemether–lumefantrine cleared asexual parasites within 72 h (von Seidlein et al. 1998). Second episodes of malaria by 4 weeks were more common in those treated with artemether–lumefantrine than SP in this trial, although most of these were determined by genotyping to be new infections (von Seidlein et al. 1998). Efficacy in the treatment of adults with multidrug-resistant P. falciparum has also been good; six doses over 5 days appears more effective than four doses over 3 days (van Vugt et al. 1999b; Lefevre et al. 2000). A Cochrane review of artemether–lumefantrine concluded that the combination was more effective than CQ, but less effective than mefloquine or mefloquine–artesunate in treating uncomplicated falciparum malaria, and that no conclusion could be reached regarding comparison with SP (Omari et al. 2002).

There are no studies that have evaluated the efficacy of azithromycin in treating or preventing malaria in pregnant women. A trial conducted among semi-immune non-pregnant adults in Kenya found a protective efficacy against P. falciparum infection of 83% for daily azithromycin (250 mg) and 64% for weekly azithromycin (1000 mg), as compared with 93% for daily doxycycline (Andersen et al. 1998). A trial in Indonesia among adults with limited immunity taking daily azithromycin (250 mg) found fair protection against P. falciparum (72%), and excellent protection against P. vivax (99%) (Taylor et al. 1999).

Finally, although quinine is generally a safe antimalarial for use in pregnancy and is effective in clearing parasitaemia among infected women in most settings, its short half-life, bitter taste and frequent side effects make it generally unsuitable for consideration as a drug for prevention programmes for malaria in pregnancy. It is usually reserved for the treatment of severe malaria or drug-resistant uncomplicated malaria in both pregnant and non-pregnant persons (WHO 2000b).

Determinants of effectiveness

For a drug to be programmatically effective in the prevention of malaria during pregnancy, in addition to being locally efficacious, it must also be available, affordable, deliverable and acceptable to the population of women who will be taking the drug. Unless a drug is sufficiently available for distribution to pregnant women in rural settings, a prevention programme cannot be successful. Affordability is another key component of effectiveness because an expensive drug may not be purchased by many countries, or may be purchased in insufficient quantities to supply all needs. In addition, some countries have adopted cost-recovery schemes that pass at least some of the treatment cost onto the consumer. In these settings, prescriptions for more expensive drugs may be more likely to go unfilled if the consumer must pay, and therefore contribute to poor compliance (WHO 1994). Even if the drug is free to pregnant women, a more expensive drug may have a greater likelihood of being resold than a less expensive one. Deliverability measures the frequency of required drug dosing, and whether all treatment doses can be given under direct observation during antenatal care (ANC), or must be taken partly unsupervised. Drugs that require more frequent dosing, and more unsupervised doses, are more likely to result in poorer compliance, and thus be less effective. Finally, acceptability is also an important consideration. A bitter drug (in some cultures bitterness is associated with abortifacients) that has common side effects is more likely to be associated with poorer compliance and a consequent decrease in effectiveness.

Measuring and comparing effectiveness

An attempt to compare the potential effectiveness of antimalarial drugs in prevention programmes for malaria in pregnancy is presented in Table 1. Drugs that are currently considered unsafe during pregnancy are not included in the table. We present a scoring system to summarize the evaluation of effectiveness and to provide a means of comparing alternative drugs and drug strategies for malaria prevention in pregnant women. A lower score is better for each of the individual categories, and therefore also for the overall score. Each drug has been scored on efficacy (1–5) and on four determinants of effectiveness: cost (1–5), local availability (1–3), deliverability (1–5) and acceptability (1–3). A specific summary number can be determined for each drug and regimen, however, the values may vary among countries depending on specific situations.

Table 1.  Comparison of effectiveness of antimalarial drug regimens for use in malaria prevention in pregnant women
Drug nameRegimen* Efficacy (1–5) Cost (1–5)Local availability (1–3) Deliverability (1–5) Acceptability (1–3)Total score† (ideal efficacy = 1)Total score† (poor efficacy = 4)
  1. * Regimen: IPT, intermittent preventive treatment; Cpx, chemoprophylaxis.

  2. See text for methods of scoring for categories of efficacy, cost, availability, deliverability and acceptability.

  3. † A low overall score is indicative of ‘good’ overall effectiveness for the regimen.

  4. Scoring system is as follows: Efficacy: 1, >95% ACPR; 2, 85–94% ACPR; 3, 75–84% ACPR; 4, 60–74% ACPR; 5, <60% ACPR; cost: 1, $0.01–$0.25; 2, $0.26–0.50; 3, $0.51–$1.50; 4, $1.51–$5.00; 5, >$5.00; local availability: 1, on national formulary, widely available; 2, on national formulary, not widely available; 3, not on national formulary; deliverability: one, two- or three-course IPT with a single treatment dose that can be given under observation; 2, two- or three-course IPT with multiple treatment doses, some of which must be taken unsupervised; 3, monthly or biweekly dosing for IPT/chemoprophylaxis; 4, weekly chemoprophylaxis; 5, daily chemoprophylaxis; acceptability: 1, generally acceptable; one point can then be added for each of the following categories, bitter taste/pregnancy-related taboos, other adverse reactions (such as itching or dizziness), up to a maximum of three points. Scores are best used for comparative purposes between drugs and regimens. Note that drugs which scores 5 for efficacy [<60% adequate clinical and parasitilogical response (ACPR)] should not be given an overall score and should be excluded from programmatic consideration. Worst-case efficacy for scoring purposes is therefore efficacy, 4 (60–74% ACPR).

  5. ‡ Lapdap is currently in commercial development, but its components (chlorproguanil and dapsone) are available.

Sulfadoxine–pyrimethamine (SP)IPT1–5111158
ChloroquineIPT1–51123811
AmodiaquineIPT1–52222912
Amodiaquine + SPIPT1–52222912
Chlorproguanil–dapsone (Lapdap)‡IPT1–52321912
Sulfadoxine–pyrimethamine +  artesunateIPT1–542211013
Artemether–lumefantrine (Riamet; Coartem)IPT1–542211013
Pyrimethamine–dapsone (Maloprim)Cpx1–532311013
Amodiaquine + artesunateIPT1–542221114
Chlorproguanil–dapsone + artesunateIPT1–543211114
ChloroquineCpx1–521431114
Atovaquone–proguanil (Malarone)IPT1–553211215
MefloquineIPT1–543231316
Mefloquine + artesunateIPT1–553231417
Atovaquone–proguanil (Malarone)Cpx1–553511518
AzithromycinCpx1–553511518
Proguanil + chloroquineCpx1–542531518
MefloquineCpx1–553431619

One possible scoring system for drug efficacy would be based on the latest WHO recommendations for monitoring antimalarial drug-resistance (WHO 2001a), and would be based on adequate clinical and parasitological response (ACPR), defined as the absence of parasitaemia on day 14 irrespective of axillary temperature without previously meeting any of the criteria of early treatment failure or late parasitological failure. The proposed five-point scoring system is: 1 = >95% ACPR; 2 = 85–94% ACPR; 3 = 75–84% ACPR; 4 = 60–74% ACPR; 5 = <60% ACPR. In Table 1, there are two summary scores, one assuming ideal efficacy (ACPR >95%, score = 1), and one assuming poor efficacy (ACPR 60–74%, score = 4). Drugs with an ACPR <60% (score = 5) should not be considered for programmatic use, and a summary score should not be assigned.

One limitation of this approach is that drug efficacy in children with clinical malaria is not equivalent to the clinical efficacy in clearing placental parasitaemia in asymptomatic pregnant women. It is possible that efficacy in asymptomatic women with low levels of parasitaemia may be better than in highly parasitemic sick infants (see discussion at beginning of efficacy section). Unfortunately, there is no standardized methodology for determining true antimalarial efficacy for pregnant women, and relatively few published data are available. In light of this constraint, we suggest using WHO standardized methodology for determining drug efficacy in infants as a substitute until other tools are developed and available.

To score cost, we calculated the average cost of a regimen that would cover the second and third trimesters of pregnancy, assuming that all women began taking medication at the start of the second trimester. Therefore, if the regimen were weekly CQ at 300 mg, then the total cost would be 26 weeks multiplied by the unit cost of a 300 mg dose. If the regimen were two-course SP, then the cost is twice the unit cost of a treatment course of SP. We divided the possible costs of regimens into quintiles, and then assigned scores as follows (1 = $0.01–0.25; 2 = $0.26–0.50; 3 = $0.51–1.50; 4 = $1.51–5; 5 = >$5). Drug costs were derived in general from the recent consultation on antimalarial drugs at WHO (2001b), or other sources if needed (McFayden 1999). Where cost is not yet established, (as is the case with Lapdap), we chose estimates based on component costs. As with efficacy, these costs may vary by region or country depending upon various factors such as tariffs, shipping costs and the availability of local production.

For local availability, scores were assigned as follows: 1, on national formulary, widely available; 2, on national formulary, not widely available; 3, not on national formulary. For the example table, we took the following approach. For certain drugs, such as CQ, we assigned 1, as it is nearly universally available. For drugs that are still in development, and for drugs that have almost no distribution outside of private pharmacies in large cities, we assigned 3, as they are not likely to be on national formularies. For all other drugs we assigned 2.

Deliverability was scored as follows: 1, two- or three-course IPT with a single treatment dose that can be given under observation; 2, two- or three-course IPT with multiple treatment doses, some of which must be taken unsupervised; 3, monthly or biweekly dosing for IPT/chemoprophylaxis; 4, weekly chemoprophylaxis; 5, daily chemoprophylaxis. Likely dosing regimens for each of the drugs or drug combinations are outlined in Table 2.

Table 2.  Possible regimens for treatment and prophylaxis for antimalarial drugs for use in pregnancy
Drug nameIntermittent preventive treatment (IPT) regimen*Chemoprophylaxis regimen
  1. * Frequency of repeating the regimen during pregnancy may vary.

Sulfadoxine (25 mg)–pyrimethamine (500 mg) (SP)Three tablets as a single doseNot recommended
Chloroquine25 mg base/kg total in divided doses over 3 days300 mg base weekly
Amodiaquine30 mg base/kg total in divided doses over 3 daysNot recommended
Amodiaquine + SPAmodiaquine as above + SP as aboveNot recommended
Chlorproguanil (80 mg) + dapsone (100 mg) (Lapdap)Chlorproguanil 2 mg/kg daily for 3 days +  dapsone 2.5 mg/kg daily for 3 daysNo available data
SP + artesunateSP as above + artesunate 4 mg/kg daily for 3 days total in divided doses over 3 days (safety in pregnancy debated)Not recommended
Artemether (20 mg)–lumefantrine (120 mg) (Riamet, Coartem)Four tablets at 0, 8, 24 and 48 h or four tablets at 0, 8, 24, 36, 48 and 60 hNot recommended
Chlorproguanil + dapsone + artesunateDose not yet determinedNot recommended
Amodiaquine + artesunateAmodiaquine as above + artesunate as aboveNot recommended
Pyrimethamine (12.5 mg)–dapsone (100 mg) (Maloprim)Not recommendedOne tablet weekly
Atovaquone (250 mg)–proguanil (100 mg) (Malarone)Four tablets daily for 3 days (safety in pregnancy not established)One tablet daily (safety in pregnancy not established)
Mefloquine15 mg base/kg as a single dose, or 25 mg/kg divided as two doses 6–8 h apart5 mg base/kg weekly (usually 250 mg or one tablet)
Mefloquine + artesunateMefloquine as above + artesunate as above (safety in pregnancy debated)Not recommended
AzithromycinNot recommended250 mg daily (safety for this use not established)
Proguanil + chloroquineNot recommendedProguanil 200 mg daily + chloroquine prophylaxis as above

Acceptability was scored in the following manner: 1, generally acceptable; one point can then be added for each of the following categories, bitter taste/pregnancy-related taboos, other adverse reactions (such as itching or dizziness), up to a maximum of three points. Pregnancy-related taboos may differ from country-to-country; we have elected to summarize available information across countries for the purposes of this example table.

It should be recognized the values assigned in this table should not be viewed as fixed. As new information becomes available, the overall effectiveness score in a specific setting may change. For example, at the moment in west Africa, SP is inexpensive, widely available and generally has good efficacy whereas artemether–lumefantrine is not widely available, and is far more expensive than SP (total score: SP, 5; artemether–lumefantrine, 10). However, if SP efficacy were to fall to 70% ACPR, artemether–lumefantrine were to become widely available (from 2 to 1), and its price were to drop (from 4 to 2), then the relative score of artemether–lumefantrine becomes better than SP (SP, 8; artemether-lumefantrine, 7).

As can be noted in the table, SP generally has the most favourable effectiveness profile (score range: 5–8) of currently and soon to be available antimalarials. SP is very inexpensive ($0.16 for a 2-course IPT regimen), widely available, easily delivered (it can be given as directly observed treatment as a single dose during an ANC visit), and has a good acceptability profile. Some of these were factors identified in a Malawi study that projected that two-course IPT with SP would be more cost-effective than CQ-containing prophylactic regimens in preventing infant deaths (Schultz et al. 1996). SP also retains fair-to-good efficacy across much of Africa, although there is increasing resistance in east Africa. In settings with increasing resistance, SP may not be the best choice.

Chloroquine, if given as two-course IPT regimen (for which there are not yet any published data), also has an attractive effectiveness profile (score: 8–11). CQ is very inexpensive ($0.14 for a 2-course regimen), nearly universally available and relatively easy to deliver, although not as easy as SP because some of the IPT treatment doses must be taken at home. However, CQ has a number of factors that limit its acceptability. Its bitter taste is unacceptable to some people (MacCormick & Lwihula 1983), although new formulations are available in capsule form that mask its bitter taste (Ndesendo et al. 1996), and have been found to increase compliance compared with uncoated formulations (Helitzer-Allen et al. 1993). It is likely, however, that this formulation would be more expensive. Some women believe that CQ, as a bitter medicine, is dangerous in pregnancy (Kaseje et al. 1987; Schultz et al. 1994a). For other women, one of the ‘mild’ adverse reactions associated with CQ, pruritis, is an impediment to usage, and has been reported as a frequent reason for not taking CQ chemoprophylaxis in Kenya (Kaseje et al. 1987) and Tanzania (MacCormick & Lwihula 1983). The principal problem with CQ is that it is not an efficacious antimalarial in most of sub-Saharan Africa. The spreading CQ-resistance will continue to diminish its efficacy and therefore its effectiveness as an antimalarial for pregnant women.

If CQ is given as weekly chemoprophylaxis in regions where the resistance to the drug is low enough to make it a viable option, the overall effectiveness score of CQ drops (score: 11–14) because of a slight increase in price and a large drop in deliverability. While weekly chemoprophylaxis dosing may be reasonable under research conditions, it is more difficult to implement at a programmatic level.

Amodiaquine has a summary effectiveness score that is similar to CQ (score range: 9–12), although the individual scores differ in several important ways. AQ is more expensive than CQ ($0.30) for a 2-course IPT regimen. Its availability is patchy, as it has been taken off the formulary of some countries because of concerns over its safety. Its deliverability as a two-course IPT regimen is fairly good (3 days for each course), and its acceptability is fairly good (not as bitter as CQ, but some patients experience dizziness). Chemoprophylaxis with AQ has been associated with more serious adverse reactions than treatment (Olliaro et al. 1996) and we have therefore not included chemoprophylaxis with AQ as an acceptable regimen.

Amodiaquine in combination with SP is a potentially attractive combination therapy for use as IPT if safety and efficacy can be demonstrated. The cost of this combination ($0.46 for two-dose IPT) is moderate and its combined effectiveness score would be good (9–12).

Although Lapdap is not yet available commercially, it is a potentially attractive option for use in pregnancy (score: 9–12). Given the low cost of its components, which are both available, the cost of Lapdap is likely to be fairly low. It is estimated that the cost of Lapdap will be <$0.50 per 3-day treatment course, for an adult, in the public sector (Peter Winstanley, University of Liverpool, personal communication). As with all new drugs, its availability will initially be poor, but could be expected to improve over time. As a two-course IPT regimen it has good deliverability (although it requires a 3-day treatment regimen) and should be fairly acceptable. The efficacy of a chlorproguanil–dapsone combination in pregnant women has yet to be conclusively shown, but general efficacy appears promising. The required frequency of the IPT regimen for chlorproguanil–dapsone will need to be established because of the short half-life of the component drugs.

The short half-life of artemisinin compounds diminishes their suitability as single agents for IPT; their role as a component of a combination drug IPT regimen remains to be determined (both in terms of safety and efficacy). For settings where the efficacy of SP as a single agent is diminishing, the use of artemisinins in combination with other drugs could prove useful. Artesunate combined with SP (score range: 10–13) would be more costly than SP alone ($4.48 for two-course IPT), would be easily available, would be slightly less easily delivered (requires multiple treatment doses that could not be given as directly observed therapy in the ANC clinic), and should have similar acceptability. Artesunate combined with AQ has an effectiveness score range (11–14) that is not as good as AQ alone (9–12), because it would be much more expensive ($4.62 for two-dose IPT). Clearly, the advantage of this regimen over AQ would be that in areas with AQ resistance, the addition of artesunate to the regimen would likely result in a highly efficacious regimen. Artesunate combined with MQ (score range: 14–17) would be more costly than MQ ($10.76 for two-course IPT), less easily available, similarly deliverable, and no more acceptable (since the acceptability profile of MQ is already poor). Artesunate combined with chlorproguanil–dapsone would have a moderate cost (not currently known, although slightly greater than artesunate–SP) and other scores comparable with an artesunate–SP combination.

The effectiveness of artemether–lumefantrine appears to be fairly good (score 10–13). The drug does not currently have wide availability outside of Southeast Asia, although it is registered for use in many countries. It is more likely to have a moderately high cost if used programmatically (approximately $5), have good deliverability and good acceptability.

The effectiveness profile of Maloprim appears to be fairly good (score range: 10–13). It has a moderate cost and patchy availability, but is generally well tolerated and acceptable. Deliverability is somewhat problematic as the drug has typically been given as biweekly chemoprophylaxis.

Malarone is another potential candidate for use in malaria prevention (IPT score: 12–15, chemoprophylaxis score: 15–18) although its safety and efficacy have yet to be determined in pregnant women. The primary limiting factor for its use in pregnant women is the cost, estimated at $84 for a two-course IPT regimen, nearly 10 times more expensive than the most expensive alternative regimen. Availability is also currently very limited, although deliverability and acceptability should be good. As a chemoprophylaxis regimen, the cost would become even more prohibitive, and the deliverability, as a daily drug, would be poor.

The long half-life of MQ and its efficacy in clearing parasitaemia make MQ and MQ-containing combination therapy theoretically attractive options for chemoprophylaxis or IPT (IPT score range: 13–16, chemoprophylaxis score range: 16–19). However, MQ has several serious limitations. The cost of MQ, estimated at $2.14 per treatment course, is high. MQ is also not widely available in most countries in sub-Saharan Africa. MQ as a drug for IPT has good deliverability, as 25 mg/kg could be given over 2 days. A regimen of weekly chemoprophylaxis would make the regimen less deliverable. MQ has also been associated with a range of side effects that could diminish compliance and therefore effectiveness. A trial in Thailand found increased incidence of dizziness among women after the first prophylactic dose of MQ, although this difference was not seen during the larger second phase of the trial (Nosten et al. 1994). Like CQ, MQ is a bitter medicine, which may lead to diminished acceptability among women in cultures where bitter foods and medicines are proscribed during pregnancy. Finally, the experience in South-east Asia with the rapid rise of MQ-resistance may make the use of MQ as a single agent for pregnant women a poor choice in Africa. The addition of artesunate to MQ for IPT (score: 15–19) would improve efficacy in regions with MQ resistance, but would raise the overall cost of the regimen.

Proguanil, when combined with CQ, also has less ideal effectiveness as an antimalarial for use in pregnancy (score range: 15–18). Although the unit cost of both proguanil and CQ is low, the daily dose of proguanil combined with the cost of weekly CQ would result in a high cost over many weeks of chemoprophylaxis. Availability of proguanil is patchy and its bitterness may limit its acceptability to pregnant women in some countries. The main constraint for use of proguanil, even in an area where it is efficacious, is the need for daily dosing, which gives it a very poor score for deliverability.

Azithromycin could prove to be an effective agent in the prevention of malaria during pregnancy in semi-immune women, as it is generally well tolerated by pregnant women who are being treated for other infections (score: 15–18) (Bush & Rosa 1994; Adair et al. 1998). However, it has a high cost, limited availability and poor deliverability as a daily chemoprophylactic. These features combine to give azithromycin a poor summary score for effectiveness, even assuming ideal efficacy.

Summary recommendations

In summary, optimal antimalarial regimens for the prevention of malaria in pregnancy do exist, but there are few. By examining a variety of determinants of effectiveness, we found that the seven most promising regimens were all IPT as opposed to chemoprophylaxis. Given the enormous logistic complexity and added cost of chemoprophylaxis, this is not surprising.

Presently, IPT with SP is likely to have the best overall effectiveness in preventing the adverse outcomes associated with malaria in pregnancy in countries where resistance remains relatively low. Its low cost, wide availability, easy deliverability and good acceptability make it the clear choice in countries where efficacy of the drug remains good. For the increasing number of countries where resistance to SP is rising or already high, the choice becomes more complicated. In no country should it be acceptable to promote an inefficacious drug, regardless of what its other attractive properties might be. In most of these settings, CQ is not efficacious and therefore its overall effectiveness will be severely reduced. Also, CQ only receives a good score for effectiveness if two-course IPT with CQ proves efficacious in the settings with CQ-sensitive P. falciparum. As a chemoprophylactic regimen, the poor deliverability of CQ makes it and all other chemoprophylactic regimens less effective options.

Lapdap, AQ, AQ + SP, SP + artesunate, AQ + artesunate and artemether–lumefantrine are other options that appear promising for use in control programmes in pregnancy. However, currently there are limited data about the use of these regimens in pregnancy. There is, therefore, a pressing need to evaluate the efficacy and safety of these drugs and drug combinations for IPT during pregnancy.

Many of the other choices for IPT are sharply limited by their high cost. Malarone, MQ and MQ in combination with artesunate would be reasonable options, but only with a drastic reduction in their market price.

Ultimately, drug efficacy is the most important factor in determining antimalarial selection for programmatic use (among drugs that are considered safe for use in pregnancy). Therefore, there is an urgent need to develop a standardized way of assessing antimalarial drug efficacy in semi-immune pregnant women, as drug efficacy in this population likely remains far greater than in young children, in whom the majority of therapeutic efficacy monitoring is undertaken.

For now, SP must be seen as the antimalarial with the best overall effectiveness for prevention of malaria in pregnancy in areas of the world where substantial SP resistance has not yet developed. In areas of the world where resistance to SP may be developing, AQ (alone or in combination with SP or artesunate) and artesunate in combination with SP may be promising alternatives, pending further data about safety of AQ and artemisinins for use during pregnancy. In areas where resistance to SP is already high, alternatives to SP alone or in combination with artemisinin compounds, such as Lapdap and artemether–lumefantrine, because of their favourable effectiveness profiles, require urgent evaluation for use in pregnancy. Given that formal trials of the safety of these drugs in pregnancy may be difficult to conduct, it is essential that post-marketing surveillance of pregnant women who are exposed to new drugs be as vigilant and complete as possible.

Footnotes

  • 1

    Use of trade names is for identification only and does not imply endorsement by the Public Health Service or by the US Department of Health and Human Service.

Authors
Robert D. Newman (corresponding author), Dr Monica E. Parise, Laurence Slutsker and Richard W. Steketee, Malaria Epidemiology Branch, Division of Parasitic Diseases, NCID, Centers for Disease Control and Prevention, 4770 Buford Highway NE, MS F-22, Atlanta, GA 30341, USA. Tel.: +1 770 488 7559; Fax: +1 770 488 4206; E-mail: rnewman@cdc.gov
Dr Bernard Nahlen, Roll Back Malaria, World Health Organization, 20 Avenue Appia. CH-1211 Geneva 27, Switzerland. E-mail: nahlenb@who.int

Ancillary