The prevalence of submicroscopic P. falciparum infections was substantial among pregnant women at delivery in our cohort in Malawi. Compared with microscopy, a real-time PCR assay targeting the pfldh gene detected more P. falciparum infections: 11 (2.3%) women were microscopy-positive and 51 (10.7%) were real-time PCR positive. Compared with microscopy, the sensitivity and specificity of our real-time PCR assay were 90.9% and 91.2% respectively. Real-time PCR can be a sensitive tool to detect low-level malaria infections and thus to evaluate the effect of these infections on birth outcomes. Additionally, we found few infections with non-falciparum species in our pregnancy cohort.
This detection of a large number of submicroscopic infectious by real-time PCR is similar to previous studies of malaria in pregnancy [9, 10]. When testing both peripheral and placental blood of pregnant women in Gabon at delivery with an assay targeting Plasmodia rDNA,  a real-time PCR detected three-fold more P. falciparum infections than microscopy (10% by microscopy and 31% by real-time PCR). Similarly, in a Kenyan study , four-fold more parasitaemia were detected at delivery with a real-time PCR assay compared with microscopy (37.9% and 9.4%, respectively). Submicroscopic parasitaemia detected by real-time PCR assays have been detected with similar frequency as ours at earlier gestational ages in studies in both the Sudan  and Mozambique . These data, which indicate that the prevalence of submicroscopic infections is similar both early in pregnancy and at delivery, suggest that the good adherence to IPTp, as was common in our study, may not substantially affect the prevalence of these low-level malaria infections.
Submicroscopic infections in the general population are significant in serving as a reservoir of parasites to drive transmission intensity , but they usually have few consequences for the individual. In this study, parasitaemia at delivery - as detected by either blood smear or the pfldh PCR assay - was not associated with poor birth outcomes compared with aparasitaemic women. Previous studies have produced conflicting data on the effect of these infections on maternal or fetal health outcomes. Two earlier reports found no association between low-birth weight and submicroscopic infections in peripheral blood as detected by either rapid antigen test or PCR assays [20, 21]. In contrast, real-time PCR positivity was correlated with lower birth weight in a cohort of Kenyan women, though it is unclear if this difference was accounted for by the smear-positive or by the submicroscopic infections . Only one study in Gabon  has demonstrated a clear association between submicroscopic parasitaemia at delivery and low-birth weight, and this has yet to be confirmed in subsequent studies. Additionally, maternal anaemia has been associated with submicroscopic infections in some studies [9, 17, 22] but not in others [23, 24]. Taken together, evidence is suggestive that submicroscopic parasitaemia at delivery is likely associated with deleterious maternal and fetal health outcomes, though the overall low prevalence of parasitaemia at delivery likely prevented the detection of deleterious consequences. More definitive characterization of these effects requires larger studies and a simple and reliable tool to detect such parasitaemia.
The pfldh real-time PCR assay we employed is highly sensitive and specific for P. falciparum infections, and offers several diagnostic advantages to conventional microscopy. Drying blood on filter paper is an effective form of DNA storage and samples prepared in this manner are cheap and safe to transport . DNA extraction from dried blood spots is relatively easy and rapid to perform, and automated extraction minimizes the risk of contamination. In general, real-time PCR assays offer specific advantages over other PCR-based chemistries: they are less labor-intensive; they are performed in closed systems which minimize the risk of post-amplification contamination; assay results can be obtained in a relatively short period of time; and the assay design is automated and is well-suited to multipurpose studies [26–28]. Additionally, the pfldh PCR assay can be used to detect parasitaemia from peripheral blood as well as placental or cord blood.
This study has several limitations. First, though the pfldh PCR assay can theoretically quantify the amount of P. falciparum gDNA, slight variations in blood spot size and efficiency of extraction prevent reliable quantitation of parasite burden. Though it is likely that the infections detected only by the assay were, in fact, "submicroscopic", they may also be the result of poorly-sensitive microscopy. Moreover, microscopy can be an imperfect reference standard used for malaria diagnosis, and this study did not have microscopy quality control [29, 30]. Additionally, the pfldh PCR assay, though simple and easy to perform, detects only P. falciparum infections. However, a species-specific assay was employed in a directed fashion to account for this, and the contribution of other species to infections at delivery was low. Plasmodium vivax was not addressed due to its rarity in sub-Saharan Africa . Finally, due to participants who did not deliver at local health facilities, the study population may not be representative of all pregnant women with malaria at the study site, and its generalizability to other geographic areas is difficult to predict.
New diagnostic tools may be required owing to the changing epidemiology of malaria. In 1996, the prevalence of microscopy-positive P. falciparum malaria was 26% at delivery in a study similar to ours in Malawi , and in this cohort it was only 2.3%. This decrease in prevalence, which is temporally associated with increased utilization of prevention measures such as bed nets and IPTp, will impact the predictive value of blood smear results. Specifically, with our pfldh PCR assay as the reference, the sensitivity of microscopy was only 20% (10/51) and the positive predictive value was 91% (10/11). Further decline in prevalence will diminish the positive predictive value, necessitating a more sensitive method for detection to establish the presence of parasitaemia as a possible risk factor for poor birth outcomes.
Microscopy remains the most practical diagnostic tool in malaria-endemic areas, but its use as a 'gold standard' may produce misleading results in clinical trials [26, 29, 33, 34]. Good-quality microscopy requires proper blood film preparation, staining, and reading by technicians with substantial expertise, and the operating characteristics of microscopy vary widely in endemic areas . When compared with molecular diagnostics for malaria, microscopy readings frequently exhibit substantial discordance, producing both false-negative readings (as would be expected for submicroscopic parasitaemia) as well as false-positive readings, which can significantly impact the results of clinical trials . For these reasons, molecular diagnostics may offer a more reliable means to detect malaria parasites in clinical research studies.
The added utility of more sensitive diagnostic tools such as PCR may vary seasonally according to the malaria transmission season. In this study, molecular detection of malaria yielded the greatest diagnostic value in the months at the end of the rainy season, when malaria transmission is waning, and detected few infections during the dry, low-transmission season. Because molecular testing is resource-intensive, a parsimonious approach to its application could focus on this period of declining transmission, when the added diagnostic value is highest.
Malaria control programmes that employ insecticide-treated bed nets and IPTp have significantly reduced the incidence of malaria during pregnancy, but more sensitive parasite detection methods have demonstrated a large pool of submicroscopic infections which may have negative consequences for mother and child. Determining the contribution of these submicroscopic infections to poor birth outcomes is critical. To this end, simple and reliable methods to detect these infections are essential, and we believe the pfldh PCR assay described herein is well-suited to future studies of submicroscopic malaria infections during pregnancy.