Background
Despite numerous studies conducted over the last decades, malaria in pregnancy (MiP)remains an important public health problem that has proved difficult to tackle. Manystudies from areas with different malaria transmission patterns have investigatedthe consequences of MiP on both maternal health and birth outcomes. While theconsequences of MiP on maternal health are dominated by anaemia, data onmalaria-related maternal mortality are sparse [
1]. For the foetus, the most commonly reported adverse effect of MiP is anincreased risk of low birth weight (LBW) [
2‐
5], which, in turn, is a significant risk factor for both impaireddevelopment [
6‐
8] and infant mortality [
9,
10]. However, most of these studies used only a single measurement point(from cross-sectional surveys or at delivery) to identify MiP and, therefore, do notcapture the multiple factors that play a role over an extended period of time.
While reliable assessment of MiP is critical to elucidating its impact on birthoutcomes and infant health, it is problematic because many factors (some of whichare difficult to fully capture) are relevant to a complete understanding. MiP may beeither continuous or intermittent, depending on a woman’s exposure to vectors,level of immunity and possible co-infections (e.g. other malaria species, HIV orhelminths), and on the efficacy of treatment and prevention interventions availableto her. Tools to measure parasite presence are limited by their sensitivity and byhow often women attend antenatal care services; hence MiP is often only partiallyobserved. To more fully evaluate the impact of MiP on both maternal and infantoutcomes, investigations must consider multiple aspects of malaria infection, suchas timing, frequency, intensity and severity of the infections, as well as thetreatment provided.
Recent studies focused on one or a few features of malaria, such as timing and/orfrequency [
11‐
15], or the effect of a single infection early in pregnancy (when weeklyscreening was routinely provided throughout pregnancy [
13]), and have produced inconsistent results. Several investigations foundthat LBW risk was associated specifically with malaria infections occurring in earlypregnancy [
11,
14,
15]. In contrast, a study conducted in Benin reported a higher risk of LBWassociated with malaria infection after six months of pregnancy [
12], and data from Thailand did not show a significantly lower birth weightin newborns of mothers with a single treated malaria episode in the first trimestercompared to newborns of mothers without malaria infection [
13]. Likewise, conflicting results have also been reported on the associationbetween the number of malaria infections and the risk of LBW [
11,
14‐
16].
MiP is thought to affect birth outcomes through two mechanisms, intrauterine growthrestriction (IUGR) and preterm delivery, which might - at least partially - explainthese discordant findings. It has been estimated that MiP in settings with stablemalaria transmission in Africa is potentially responsible for up to 70% of IUGR and36% of preterm delivery [
4]. The former has been consistently associated with placental infection [
17‐
24], while the latter appears to correlate with systemic manifestations ofmalaria infection in the mother [
25‐
27]. However, accurate determination of gestational age is required todistinguish IUGR from preterm delivery—a determination that is difficult tomake in resource-constrained settings, where tools such as ultrasound are rarelyavailable. As a result, evidence of the relative importance of IUGR versus pretermdelivery due to MiP remains limited [
28].
In recent years control of MiP has relied partly on intermittent preventive treatment(IPT), with WHO currently recommending at least two doses withsulphadoxine-pyrimethamine (SP) [
29]. However, growing resistance of malaria parasites to SP in many regions [
30,
31], combined with the changing epidemiology of malaria, indicate that otherprevention approaches must be strengthened. To help fill the evidence gap regardingthe impact of MiP on delivery outcomes in accurately dated pregnancies, this studyreports on the findings from a prospective cohort of pregnant women with access toweekly antenatal malaria screening and prompt treatment.
Discussion
The novel features of this cohort study are the frequent, intensive malariascreenings (median of 21 screens per pregnancy) and the provision of treatment basedon the presence of parasite in the blood rather than on symptoms—practiceswhich differ markedly from those common in endemic Africa. Another strength of thisstudy is the accurate determination of gestational age for the majority ofpregnancies.
Our results suggest that peripheral malaria infections during pregnancy, includingthose occurring late during gestation, contribute significantly to perinatalmorbidity. Malaria infection at the end of pregnancy and those with fever ratherthan other aspects of malaria exposure, were associated more specifically withmiscarriage or pre-term delivery. A similar association between malaria infectionswith fever and an increased risk of miscarriage has been reported in mothers with asingle malaria episode during the first trimester of pregnancy [
13]. Likewise, increased infant mortality has been reported after symptomaticmalaria infections occurring at the end of the pregnancy [
10,
43]. In low endemic areas, 80% of microscopically detected infections becomesymptomatic if left untreated [
44]. Since women in this cohort were treated if they had a positive bloodsmear, irrespective of whether they showed symptoms, it seems likely that this earlydetection and treatment of asymptomatic infections prevented higher rates ofmiscarriage and pre-term delivery. Current WHO policy calls for “theadministration of at least two doses of SP during the second and third trimesters ofpregnancy” [
45,
46]. More effective protection during late pregnancy is critical inlow-endemic settings such as Mbarara, and addition of an extra (third) SP dose forall pregnant women rather than (as per current WHO policy) only to HIV-infectedwomen, or monthly dosing, could provide more effective protection in allpregnancies.
Adjusting for gestational age at birth, we found that peripheral malaria infectionduring pregnancy was associated with lower birth weight, and that this associationwas consistently seen in both the full dataset and the subset of mother withultrasound examination. Furthermore, more severe birth weight impairment wasobserved after multiple malaria infections and in malaria infections with highparasitaemia, even when IPT and bed net use was reported. These findings underscorethe importance of implementing efficacious prevention, prompt diagnosis and highlyeffective anti-malarial treatment during pregnancy [
47].
In addition to primigravidity, a well-known risk factor for MiP [
1,
4], it was found that low education level and rural residence wereindependently associated with malaria during pregnancy. These findings furthersupport the notion that it is essential to scale up malaria prevention efforts inmore isolated and deprived communities as recently highlighted in a meta-analysis ofdatasets from 25 African countries [
48]. A low number of antenatal visits was also associated with reduced birthweight. The emphasis on at least four antenatal visits is required for improvedcontrol of malaria in pregnancy [
45].
A limitation of this study was that documentation of malaria infection began onlyafter the first trimester of pregnancy, resulting in left censored data. Women withmultiple infections were more likely to have been enrolled later during theirgestation and, therefore, early infections might have been missed. This couldexplain the absence of association in this analysis between MiP early duringgestation and low birth weight, in contrast to results from other studies [
11,
14,
15]. Alternatively, effective treatment of a single infection may allowrecovery from infection and catch-up growth
in utero.
Placental malaria has been shown to be a key intermediate factor in the pathologicalpathway of malaria [
2,
4,
18,
19]. However in our study, the proportion of placental infections (asdetermined from placental smears) was low, which most likely reflects ourstudy’s intensive detection and rapid treatment of malaria infections, aspreviously observed [
20]. However, for placental malaria diagnosis the sensitivity of parasite(rather than pigment) on placenta smear is low, so the actual proportion ofplacental infection might also have been underestimated [
49]. In a cohort study of women actively screened and tested in the Gambia,the presence of pigment was reported to better reflect past infection with malaria [
50], a finding which may explain why an association between low birth weightand the presence of pigment but not with the presence of parasite in placental smearwas found in univariate analysis. Nevertheless, as the placenta cannot be examineduntil delivery, hence until after the adverse effect has already occurred, itsutility for clinical diagnosis and prevention remains limited.
On the other hand, peripheral parasitaemia, which was associated with impaireddelivery outcomes in this cohort, can be detected by frequent screening, so thatprompt treatment can be given and adverse effects of the infection reduced. Sincepreventive efforts (IPT with SP and insecticide-treated bed net) still leave a largeproportion of women with parasitaemia, taking the opportunity to screen women whenthey present to antenatal care is a strategy that should be considered. However,diagnostics for MiP remain problematic, since pregnant women often have low levelsof parasitaemia and require diagnostic tools with greater sensitivity thanmicroscopy (and good specificity)—for example, the Loop-mediated isothermalamplification (LAMP) [
49,
51]. As malaria prevalence decreases, the risk-to-benefit ratio for providingIPT also reduces. Hence efforts to determine the optimal number of screenings forwomen in malaria endemic areas are also required.
In conclusion, this study shows that the timing, parasitaemia, symptoms and number ofperipherally detected malaria infections observed during pregnancy are associatedwith adverse outcomes. Prompt detection and treatment with an effectiveanti-malarial should be offered, irrespective of symptoms and use of otherpreventive measures in pregnancy. While frequent screening was associated withimproved birth outcome, reaching mothers living in remote areas to prevent lateattendance and low number of visits at antenatal care is essential, as they are morelikely to suffer from poor outcomes.
Competing interest
The authors declare that they have no competing interests.
Authors’ contributions
PDB conducted the statistical analysis and wrote the paper. RM and PP designed thestudy, participated in the statistical analysis and manuscript drafting. ETparticipated in data collection, statistical analysis, and manuscript drafting. LWparticipated to the statistical analysis and manuscript review. CN and BTparticipated in the data collection, and manuscript review. YB participated in datacollection, data analysis, and manuscript review. AM performed the histologicalanalysis, interpreted the results and reviewed the manuscript. PG participated tothe study design and manuscript review. All authors read and approved the finalmanuscript.