Background
Plasmodium falciparum malaria remains a major contributor of the disease burden in sub-Saharan Africa (SSA). Children under five years of age and pregnant women represent the most vulnerable population. During pregnancy, malaria infection leads to parasite sequestration in the maternal placental vascular space, with the consequence of increased risks of abortion, stillbirth, prematurity, intra-uterine growth retardation (IUGR) and maternal anaemia [
1]. Malaria in pregnancy (MIP) is associated with increased risk of low birth weight (LBW) and prenatal, neonatal and infant mortality [
2‐
5]. Infection with
P. falciparum towards the end of gestation increases the likelihood of placental infection [
6]. A previous review by Desai
et al estimated that approximately one in every four pregnant women in malaria endemic areas has evidence of placental infection at the time of delivery [
7].
The diagnostic challenge is that conventional peripheral blood microscopy is unable to detect all infections as parasites can be sequestered in the placenta [
8‐
10]. Such sub-microscopic infections increase the risk of anaemia and poor foetal outcomes in seemingly aparasitaemic pregnant women [
9,
11] thus increasing the risk of maternal and perinatal mortality. Consequently, some clinicians are reluctant to withhold anti-malarials for febrile pregnant women, arguing that microscopy misses placental infections.
However, malaria rapid diagnostic tests (RDTs) may provide a solution as emerging evidence suggests that RDTs are capable of detecting placental malaria better than microscopy and may detect sub-microscopic infections [
12,
13]. RDTs are lateral flow immuno-chromatographic dipstick assays that detect either histidine rich protein-2 (HRP-2) or Plasmodium lactate dehydrogenase (pLDH) produced by infected red blood cells. Although the large majority of RDTs comprise mAbs raised against HRP2 or pLDH, it should be noted here that there are an increasing number of tests that also include a Pan anti-aldolase test line. Whereas good quality microscopy is lacking in many resource-limited settings, as it requires well-trained, competent personnel, infrastructure as well as effective quality control and quality assurance [
13], RDTs are becoming increasingly available and affordable[
14].
With many countries in SSA following World Health Organization (WHO) policy [
15] and scaling up parasite-based diagnosis for malaria through the use of microscopy and RDTs, establishing the effectiveness of RDTs in comparison to microscopy in detecting malaria during pregnancy has become a priority.
Although there is limited evidence on the use of RDTs among pregnant women [
16,
17], RDTs have been found to be comparable in sensitivity to microscopy for malaria [
18] and to improve malaria diagnosis and quality of care at lower level health facilities [
19‐
21]. The few studies that have been done have based the accuracy of RDTs on either placental smear or placental blood polymerase chain reaction (PCR) and not histological examination of the placenta at delivery [
10,
11,
22‐
24].
This study compares HRP-2 RDTs to microscopy for detecting peripheral malaria in feverish pregnant women at time of recruitment in a hospital setting. In addition, at delivery both RDTs and microscopy are compared to placental histopathology performed to detect placental malaria.
Discussion
This study found a malaria prevalence of 38% using microscopy and 54% with RDTs among febrile pregnant women at ANC. When compared with microscopy, RDTs demonstrated acceptable sensitivity (97%) and specificity (74%) for the diagnosis of
P. falciparum malaria among febrile pregnant women attending ANC at a hyper-endemic region in eastern Uganda. The procedures in this study were performed by nurses and midwives at the hospital ANC clinic. The operational nature of this evaluation allowed the investigators to demonstrate that RDTs are reliable for diagnosis of
P. falciparum malaria in pregnancy when performed by midwives. These findings are comparable to previous reports of diagnosed malaria in pregnancy ranging from 9% to 60% in sub-Saharan Africa [
33].
The high sensitivity of RDTs among symptomatic pregnant mothers is comparable to results from a previous operational study among outpatients in a similar setting [
34]. The high sensitivity gives confidence to clinicians that RDTs are unlikely to miss malaria infections in pregnancy. The PPV was low and this could be due to false positive results attributable to the persistent nature of HRP-2 antigenemia already documented by previous studies [
34‐
36]. The false positive may also have been due to better detection of malaria by HRP2 test in respect to the microscopy used as reference [
37]. Using PCR to validate the false positives did not alter this result [
37]. This was unexpected as the sensitivity of PCR has been documented as significantly higher than RDTs [
17].
This study found a relatively low specificity but a high negative predictive value of 97.5%, almost certainly excluding malaria infection. This NPV result gives confidence that if RDTs are used to diagnose malaria in pregnancy very few infected women will be missed. Moreover, false negatives were linked to very low parasite densities that are unlikely to be the cause of illness in such an endemic setting. The consequences of a false negative in an endemic area, particularly when related to low parasitaemia, mean that the patient is unlikely to die. The infection may either clear without illness, as people living in endemic areas have partial immunity, or clinical symptoms will recur and they would seek care again.
For clinical application, sensitivity and specificity of the RDTs can be combined into a single measure called likelihood ratio (LR). LR provides a summary of how many times more (or less) likely patients with malaria are to have a positive (or negative) result than women without malaria. This study found a LR+ of two which means that a pregnant woman with malaria is about two times more likely to have a positive RDT test than a woman who has not got malaria in pregnancy. It also found a LR- of 0.025 which means that the probability of having a negative test for pregnant women with malaria is 0.025 times of that of those without malaria. Therefore, women without malaria are about 40 times more likely to have negative test than women with malaria. These results imply that RDTs were better at 'ruling out malaria infection' than in confirming active malaria infection. The benefit of likelihood ratios is that they can be used to help the clinician adapt the sensitivity and specificity to tests of individual patients [
38,
39]. In summary, the high sensitivity, specificity, NPV and the LR of RDTs in symptomatic pregnant women, demonstrated in this study, gives confidence to the midwife (or physician) to treat all RDT-positive and not treat women with a negative RDT. However, to make sure that all pregnant women receive the doses of SP for IPT that will hopefully clear any infection and placental infection if present, the midwife should investigate for non-malarial causes of fever when the RDT is negative.
This study found that there was an association between RDT result and clinical or demographic factors. During routine care, the clinician takes a history and examines the patient to estimate the chance (probability) of malaria prior to performing a test. A patient's probability of having malaria after the test result is known (post-test probability) is what the clinicians are most interested in as it can help in deciding where to confirm a diagnosis or rule it out. The actions taken after receiving the test results are weighed in light of the history and examination obtained prior to testing. The factors that influenced post-test probability in this study are consistent with known risk factors associated with malaria infection, and include parity, low maternal haemoglobin, sleeping under a mosquito net and advanced gestational age. Previous studies have shown that primigravidae are at increased risk for malaria in pregnancy and most cases of malaria in pregnancy happen towards the end of gestation [
1,
6,
40,
41]. The findings affirm the notion that HRP2 RDTs are most useful when the clinical and social demographic context is known because of varying risk levels for malaria during pregnancy [
42].
In this study, direct examination of the placental tissue showed that 21% of women had placental infection of which 12% were active infections on the day of delivery. The study demonstrated a modest level of accuracy (80.9% sensitivity, 87.5% specificity, 98% NPV) of RDTs in detecting placental malaria using peripheral blood at time of delivery. The study allowed the investigators to characterize the burden of placental infection among pregnant women with history of febrile illness during the third trimester. The prevalence of placental malaria is similar to that reported from Nigeria [
43]. However active placental malaria infection at the time of delivery in this study was lower than the reports from Cameroon at 60% [
9] and 34% in a study in Ghana [
11]. The possible reason for the difference with previous studies is that those studies used placental microscopy smears rather than histology to detect placental infection. In addition, it is possible that this study presents lower rates of placental malaria because of the high use of IPT-1 at 70% and ITN use at 84%. Use of IPT and ITNs reduces the risk of malaria and placental malaria [
44]. The current study highlighted that as parity and use of ITNs increase, there is a significant reduction in the risk of placental malaria infection. There was no significant difference between RDTs and microscopy in detecting placental malaria. The combined sensitivity and specificity of both peripheral microscopy and RDTs detecting placental malaria was higher than using individual tests and this difference was statistically significant [
32]. This finding is similar to that reported in a study in Cameroon where a combination of microscopy and RDTs detected more placental infection [
10]. This implies that RDTs could complement peripheral microscopy in excluding placental malaria. The clinical relevance of this is that women with negative tests for both microscopy and RDTs should not be treated with anti-malarial drugs as chances of having placental malaria are minimal. However, intermittent presumptive treatment needs to be advocated for and promoted to minimize the risks of placental infection and the associated complications to the mother and foetus [
3,
11,
45]. Malaria in pregnancy and acute placental infection increase the risk of clinical malaria in the mother, severe anaemia, intra uterine growth retardation and death in infancy [
1]. In RDT negative pregnant women, IPT with SP should be given. Although some studies have reported good efficacy of SP, its future usefulness is questionable [
46] given the current prevalence of mutant parasites [
47,
48], and yet alternative effective and safe replacement remains unclear. The sensitivity of the RDT in this study was acceptably high and any clinician who does not treat the patient with a positive RDT would be negligent, but should also explore other reasons for fever, which could in any case coexist with malaria. Some mothers had false positive RDTs without placental or peripheral infection detected. The consequence of false positive result is that someone may be treated for malaria when they are not infected. Based on these study results, health workers will over-treat fewer mothers than if they treated all symptomatic women. It is also likely that some of these malaria infections were identified early before migration of infection to the placenta as shown by positive peripheral smear readings [
43].
This study was integrated into the existing health system and done by the human resources who would be available for eventual scale-up and in close collaboration with National Malaria Control Programme. The study built on previous work published by others by using an implementation/prospective approach and providing an appropriate context in which malaria is diagnosed. This approach in study conduct was important to validate effectiveness of RDTs given this can be context specific [
42]. It has been argued that the effectiveness of RDTs is dependent on malaria transmission intensity and the operational context.
The main limitations of this study included the low hospital delivery rate of 47%, which made it impossible to analyse the deliveries out of the hospital. However the hospital delivery rate was higher than the national rate of 30% and the authors attribute this to the study mothers being incentivized to deliver in the hospital. Some samples were not analysed due to missing data, an expected challenge when using data from a routine ANC care setting. These limitations are to be expected in most operational studies but this approach allows the team to learn through implementation. Another limitation is that the RDTs (Diagnosticks) selected for the study displayed a very low sensitivity (59%) against low parasitaemia blood samples in an independent evaluation conducted by FIND/WHO [
49]. Additionally, it has been noted that this RDT also has quite a high false positivity rate of 8% (8 out of 100 negative blood samples tested positive for
P. falciparum ). The inherently higher rate of false positives recorded for this test may explain the poor PPV observed in the study and also account for lower specificity percentages. It is recommended by the WHO that an RDT should have at least 95% sensitivity at 100 p/µl [
50]. It is possible that another HRP-2 RDT would perform better in terms of sensitivity in this context. In order to provide a more comprehensive and thorough assessment of performance of an RDT when compared to microscopy or any other experimental method, future studies should make use of the WHO interactive online guide for selection of RDTs. Studies should include well characterized RDTs with much higher sensitivities and specificities and more than one test evaluated to provide a more robust assessment of their performance in this context.
Competing interests
The author declares that they have no competing interests.
Authors' contributions
DJK participated in the conception, study design, protocol development, training, and was responsible for project management, data analysis and interpretation, and developed the draft manuscript. NM, CR, DN, and MM participated in the development of tools, interpretation of results and writing of the manuscript. LKT and MN participated in the training of health workers and acquisition of data. HC, JP, SM and JKT participated in the conception study, interpretation of results, provided technical over sight and leadership in writing of the manuscript. All authors read and approved the final manuscript.