Increased prevalence of malaria in HIV-infected pregnant women and its implications for malaria control

Study site

This study was undertaken between March 1993 and September 1994 in Chikwawa District, Southern Malawi, as part of a project investigating malaria and anaemia in pregnant women and their infants. In this area malaria is perennial with year-round transmission. District health facilities are provided by Chikwawa District Hospital, Montfort Hospital and 13 health centres. The study was conducted in two hospitals located 30 km from each other on the main route from south Malawi to Mozambique.

Enrolment

All women attending the antenatal facilities of the hospitals were enrolled in the study at their first visit after obtaining informed consent. A questionnaire was completed by a trained project nurse which included information on age, previous antimalarial use in pregnancy and obstetric history or illnesses. Literacy status was assessed by asking women to read a simple sentence in the local language. Gestation at enrolment was estimated by calculating the weeks between delivery date minus gestational age estimated postnatally and delivery date minus booking date. A blood sample was collected for a malaria smear. In Chikwawa District Hospital women were given two routine treatment doses of sulphadoxine-pyrimethamine (SP: sulphadoxine 1500 mg; pyrimethamine 75 mg), one at the first antenatal visit after the first trimester and a second dose at 28–34 weeks gestation, conforming to Malawi government policy on antimalarial control during pregnancy. The policy at Montfort Hospital was to screen women at their first antenatal visit with a blood smear and to provide SP treatment if parasitaemic. At later antenatal visits women would receive SP if symptomatic. In both hospitals SP was taken under supervision.

Delivery

At delivery information was only collected from women who were recruited antenatally and who delivered at one of the hospitals. Information on illness during pregnancy and antimalarial use between first antenatal visit and delivery was obtained from the antenatal card. A blood sample was collected for a malaria smear and for HIV testing. A blood smear for malaria was also made from the placenta and cord blood. The newborn baby was examined for gestational age using a modified Ballard method ( Verhoeff et al. 1997 ).

HIV testing

Sera from mothers whose babies were included in an infant follow-up study were HIV-tested after obtaining informed consent. The criterion for recruitment of newborns was based on low birthweight (< 2500 g) and matched normal birthweight babies. Pre- and post-test counselling was offered. Additional pregnancy sera were randomly selected from women whose babies were not in the infant follow-up study; these were tested anonymously.

Laboratory investigations

Malaria slides were stained with Giemsa and read counting asexual Plasmodium parasites against 200 white blood cells (WBC). Malaria parasite density calculations were based on the assumption of 8000 WBC per μl blood. Malaria slides read in Malawi were checked by an experienced microscopist at the Liverpool School of Tropical Medicine. Maternal HIV-infection was defined by the presence of HIV-antibodies determined by ICE*HIV-1.O.2 (Murex; Dartford, UK) and confirmed by the VIDAS HIV-2 new test (bioMérieux; Lyon, France). WHO recommends the use of two different ELISAs for the diagnosis of HIV in asymptomatic patients when the HIV prevalence of the population is above 10% ( WHO 1992).

Analysis

Data were analysed using EPI-info version 6.02 (1994) and SPSS for Windows version 6.1.2 (1995). χ² analysis was used for comparison of proportions or linear trend. The Mantel-Haenszel method and 95% confidence intervals (95%CI) were used for estimating relative risk. A P-value of < 0.05 was considered significant. Ethical approval for the study was granted by the Malawi Health Science and Research Committee.

HIV seroprevalence

Prevalence of HIV infection was 25.6% and all infections were HIV-1. The majority of infected women were multigravidae (77.4%). Antimalarial use before first antenatal visit was reported by 57 women and this was almost doubled in HIV seropositive women (13.2%vs 7.8%, P < 0.05). Illness during pregnancy, excluding malaria, was reported in 38.9% of HIV-infected women compared to 29.2% of HIV-uninfected women (P= 0.02). The most frequently reported symptoms in both groups were abdominal pain (17.2%), respiratory tract infections (14.2%) and vaginal discharge (8.3%).

HIV and P. falciparum malaria

Nearly all infections reported were due to P. falciparum with 1% caused by P. malariae. After excluding women who had taken prior antimalarials, 56.3% of HIV-infected primigravidae were parasitaemic at their first antenatal visit compared to 36.5% in HIV-uninfected primigravidae (P= 0.04) ( Table 3). In multigravidae these prevalence figures were 23.8% and 11.0%, respectively (P= 0.002). The relative risk for parasitaemia by parity group in HIV-infected vs uninfected women is shown in Table 3. At the first antenatal visit the relative risk was significantly increased for three of the five parity groups, including the two of highest-parity. At delivery these risk estimates were increased but did not reach statistical significance, although for all parity classes the lower 95% confidence interval was 0.99. There was a significant decrease in malaria prevalence with maternal age for both HIV-infected and uninfected women, but when stratified by parity this age difference became nonsignificant.

Table 3. Parasite prevalence (%) in peripheral blood by parity and HIV status in women who have not taken antimalarials during pregnancy

Parasite prevalence at first antenatal visit in relation to gestational age and HIV status is shown in Table 4. Peak parasite prevalence occurred earlier in gestation in HIV-infected compared to HIV-uninfected primigravidae (16–19 weeks vs 20–23 weeks). Parasitaemia in peripheral and placental blood at delivery was significantly increased in HIV-infected multigravidae (P < 0.05), but not in primigravidae. Cord parasitaemia did not differ with HIV status (HIV+ , 3.8%; HIV−, 6.1%). No significant difference in geometric mean parasite densities (GMPD) was observed with HIV status, although parasite densities were generally higher in HIV-infected women at enrolment (primigravidae: 1.634 parasites/µl if HIV + and 1.048 parasites/μl if HIV−, P= 0.17; in multigravidae: 882 parasites/μl if HIV+ and 729 parasites/μl if HIV−, P= 0.43).

Table 4. P. falciparum prevalence (%) in relation to gestational age and HIV positivity

Season affected parasite prevalence with the highest prevalence for both primigravidae and multigravidae in the postrainy season (March – June). In HIV-infected women 60.0% of primigravidae and 28.9% of multigravidae were parasitaemic at this time, compared to 46.8% (P= 0.37) and 13.0% (P= 0.02), respectively, for HIV-uninfected women.

Table 5 shows that malaria prevalence at delivery following two routine treatment doses of SP during pregnancy was higher in HIV-infected compared to uninfected women although these differences did not reach statistical significance. Three of five HIV-infected primigravidae showed placental parasitaemia and over a quarter of HIV-infected multigravidae had peripheral parasitaemia despite prior treatment. HIV-uninfected women who had received two prior treatment doses of SP also had high parasite prevalence at delivery but this was generally lower than in HIV-infected women. These results are compared in Table 5 with parasite prevalence at delivery in those women who did not receive any SP treatment before delivery.

Table 5. P. falciparum prevalence (%) at delivery in women who received two routine treatment doses of sulphadoxine-pyrimethamine (SP) and who did not receive SP during pregnancy

Increased malaria prevalence in HIV-infected women

Our results show that HIV infection is associated with a significant increase in malaria prevalence in pregnant women from early in gestation. This effect is apparent in primigravidae and multigravidae and independent of maternal age. Previous studies in nonpregnant adults and in children have not shown an increased malaria prevalence in HIV-infected individuals ( Nguyen-Dinh et al. 1987 ; Simooya et al. 1988 ; Greenberg et al. 1991 ; Niyongabo et al. 1994 ). Allen et al. (1991) found no association between HIV and the presence or degree of malaria parasitaemia in a sample of women, an unknown proportion of whom were pregnant. Steketee et al. (1996) reported an increased prevalence of malaria parasitaemia in HIV-infected Malawian pregnant women with a risk ratio of 1.31 (CI; 1.12–1.52), which was lower than our weighted relative risk of 1.74 ( Table 2). This and previous studies ( Schultz et al. 1994 ) did not control for previous antimalarial use which is common in malarious areas and in our study was found to be higher in HIV-infected women. This could be a confounder in malaria prevalence studies. Our study has shown that the prevalence of parasitaemia in HIV-infected women who did not take prior antimalarials is increased from very early in pregnancy and that at delivery this higher prevalence remains even in those who received two scheduled treatment doses of SP during pregnancy. Some caution is needed in interpreting these findings, as the numbers of enrolled subjects in the gravida and gestational age groups are not large. Some of these associations may have been due to chance, and multiplying subgroup analyses in the context of sample fluctuations will increase the probability of identifying a statistically significant result. Nevertheless, 10 of the 12 subgroup comparisons in Table 3 showed higher malaria prevalence in HIV-infected women.

An important confounder was maternal sample selection which was based on HIV screening of mothers whose babies were recruited into an infant cohort study. As both low and normal birthweight infants were recruited for this study, low birthweight selection could confound both HIV and malaria risk in the mother. The subgroup analysis in Table 2 controlled for birthweight and HIV infection status. This showed that malaria prevalence was consistently higher in HIV infected women and that these differences reached significance for both the low and normal birthweight groups at first antenatal visit. The HIV effect on malaria prevalence is therefore greater than that due to possible bias because of low birthweight selection. The prevalence of malaria at first antenatal visit was also not significantly greater in the HIV-tested compared to the HIV-nontested group ( Table 1, 22%vs 18.8%), although this did reach significance at delivery. There is therefore some bias for malaria risk in the HIV-tested group.

HIV infection and parity-specific immunity

Under holo-endemic conditions malaria parasite prevalence falls with increasing gestational age ( Brabin 1983), relating to the development of malaria specific immunity during pregnancy ( Brabin & Perrin 1985; Fried et al. 1998 ). This results in a lowering of malaria prevalence in subsequent pregnancies with the development of parity-specific immunity ( McGregor 1984; Brabin & Brabin 1992), i.e. induction of specific malaria immunity during an initial pregnancy which results in an improved recovery rate from malaria infection in a subsequent pregnancy. The beneficial effects of this are seen as a reduction in low birthweight and maternal anaemia in multigravidae ( Jackson et al. 1991 ). HIV infection appears to impair this process by limiting the acquisition of malaria immunity during pregnancy. What is not clear is the mechanism of this association. No laboratory data were available on the degree of immunosuppression; however, none of these women had AIDS. Selected P.falciparum-specific immune responses were maintained in AIDS-infected adults in Burkino Faso, although cell proliferation and of IFN-α and IL-2 production decreased in response to a P.falciparum in vitro culture ( Migot et al. 1996 ).

We have shown that the risk estimate for malaria parasitaemia in HIV-infected grand multigravidae (parity > 4) remains significantly increased at first antenatal visit, suggesting that this effect persists throughout several pregnancies.

Malaria control in pregnancy and HIV infection

Malaria control in pregnancy is problematic even when HIV infection is uncommon. Because of the adverse effects of malaria during pregnancy for mother and child, WHO has recommended initial treatment followed by chemoprophylaxis for pregnant women in malarious regions ( WHO 1986). Chloroquine is the most frequently used drug for this purpose, but increasing drug resistance has greatly reduced its efficacy ( McDermott et al. 1988 ; Mutabingwa et al. 1993 ). This paper indicates that malaria control strategies will now have to take into account prevalence of HIV infection. In countries where this is high, there is a clear need to reassess antimalarial drug policy, in view of the added problem of impaired malaria immunity in higher parities due to HIV. Malawi was the first country in Africa to abandon chloroquine for antimalarial control in pregnancy and to use SP given routinely as two treatment doses in mid-and late pregnancy ( Schultz et al. 1994 ). This strategy fails to clear parasitaemia by delivery in about one in every two HIV-infected primigravidae and one in four HIV-infected multigravidae ( Table 5). Its efficacy is also low in HIV-uninfected women, especially primigravidae and there was no significant difference in parasite prevalence at delivery between those who did and those who did not receive SP during pregnancy. This result is not due to drug resistance to SP, as 90% sensitivity has been shown in 28 day-in vivo tests in children and is as high as 100% in pregnant women in this area of Malawi ( Howorth 1996; Verhoeff et al. 1996b ). The gestational age when the second treatment dose is received (28–34 weeks) could be important in relation to parasite clearance by the time of delivery. Our results do not show a significant reduction in parasite clearance by delivery in HIV-infected women, although the poorer clearance is consistent with the higher prevalence in HIV-infected women. Antimalarial drug efficacy does relate to the host's malaria immunity, as reduced treatment doses of antimalarials are adequate to clear parasitaemia in semi-immune adults ( Targett 1984). Parasite clearance at delivery may also be substantially different with early prescription of the second treatment dose (28 weeks vs 34 weeks gestation) allowing sufficient time for re-infections before delivery. In this context the two-dose SP regimen may significantly reduce maternal anaemia and improved foetal growth despite the observation that what may be late pregnancy re-infections result in high parasite prevalence at delivery ( Verhoeff et al. 1998 ).

The difference between our results and those of Schultz et al. (1994) reporting a study conducted in another area of Malawi in 1992 are striking. An important difference between the two studies is that Schultz et al. (1994) did not exclude women with prior use of antimalarials at recruit-ment and some women who may have had prior treatment appear to have been recruited up to 32 weeks gestation. This combined effect may have led to their reported lower parasite prevalence at delivery (9% placental parasita- emia).

Multigravidae in holo-endemic areas have, compared to primigravidae, a reduced prevalence of anaemia ( Jackson et al. 1991 ) and low birthweight ( Brabin 1991b), as they experience less malaria. With concurrent HIV infection this situation changes. Primigravidae in malarious areas are already at high risk of developing severe anaemia and delivering low-birthweight babies ( Brabin 1991a). This means that the overall impact of HIV infection will be to increase the population-attributable risk of severe maternal anaemia and low birthweight to higher levels. This is likely to have important implications for child survival ( Verhoeff et al. 1998 ).

Increased postnatal mortality related to placental malaria in HIV-infected women has been reported in Malawian women, although this was not proven to be due to increased transmission of HIV from mother to infant ( Bloland et al. 1995 ). Follow-up of infants born to mothers in our study found that infant mortality was 3 times higher for babies of HIV infected women and 6 times higher if these mothers had malaria at booking ( Verhoeff et al. 1996a ). This increased risk occurred despite evidence that vertical transmission of HIV was not increased with maternal malaria. As malaria in pregnancy is associated with significantly increased infant mortality in HIV infected mothers, increased efforts are required to develop and improve malaria control activities to pregnant women in many parts of Africa. The mechanisms which lead to poor survival require investigation, particularly as they relate to maternal anaemia and low birthweight.

Restricting antimalarial treatment in pregnancy to primigravidae, as has been suggested by Greenwood et al. (1994) , would not be appropriate in these circumstances. These results suggest that there is a necessity to provide antimalarials for all parities from as early in pregnancy as can safely be achieved. It would seem important to evaluate interventions with more frequent doses of SP, given perhaps even monthly, with careful monitoring of side-effects, and a requirement to monitor antimalarial in vivo drug efficacy in HIV-infected and HIV-uninfected pregnant women. There is also a need to undertake a longitudinal study of HIV-infected and uninfected pregnant women to assess incidence and severity of malaria episodes.