Background
Malaria is a public health threat causing morbidity and mortality in Yemen, with
Plasmodium falciparum being the predominant species responsible for almost 99 % of cases. It is estimated that 43 % of the population are at high risk and a total of 63,484 microscopy-confirmed and 39,294 rapid diagnostic test (RDT)-confirmed cases were reported in 2013 [
1]. Yemen is in the control phase, and the adopted malaria control strategies include distributing insecticide-treated nets, indoor residual spraying, prompt diagnosis and treatment with artemisinin-based combination therapy [
2].
Light microscopy (LM) is still the cornerstone of malaria diagnosis in Yemen, especially in hospitals. However, LM has low sensitivity for detection of low parasite densities, is time-consuming and requires skilled technicians and good reagents [
2,
3]. Therefore, it may not reflect the submicroscopic infectious reservoir in Yemen, which is still neglected and needs to be estimated if malaria elimination in the country is to be achieved [
4]. RDTs have been introduced as an alternative to LM, especially when good LM practice cannot be maintained or is not available. RDTs that target
P. falciparum histidine-rich protein-2 (
PfHRP-2) have the highest and most consistent detection rate [
5]. In contrast,
Plasmodium lactate dehydrogenase (pLDH) detects all
Plasmodium species and is usually combined with
PfHRP-2 for malaria screening in areas endemic with multiple species [
6,
7]. The National Malaria Control Programme (NMCP) has been using RDTs for malaria diagnosis and field surveys since 2007 [
2,
8]. Although the World Health Organization (WHO) has provided comparative data on the performance of RDTs that can be used for procurement decision, it is well recognized that clinical sensitivity of RDTs depends on the epidemiology of malaria in the target population [
5], which imposes field evaluation of such tests. In Yemen, malaria is unstable, seasonal and affected by topography and rainfall. The country has been stratified with respect to malaria endemicity into four strata that are different in altitude, intensity, length, and season of transmission and even in the predominant vector species [
2]. This heterogeneous epidemiology of malaria may affect the performance of RDTs, necessitating the need for their evaluation in the four strata. In Yemen, only two previous studies evaluated the performance of
PfHRP-2-based RDTs against LM as the ‘gold standard’ [
9,
10]. It is, however, noteworthy that false-negativity of LM limits its accuracy as reference method. Polymerase chain reaction (PCR) is more sensitive than LM and RDTs for detecting malaria in epidemiological studies assessing asymptomatic carriers in low endemicity settings [
11‐
13]. This is the first community-based survey to evaluate the performance of LM and a
PfHRP-2/pLDH RDT for malaria diagnosis against PCR as the reference method during the transmission season in a malaria-endemic area in Taiz governorate.
Discussion
Prompt malaria diagnosis is a key component of the national malaria control strategy in Yemen, which relies on the use of LM and RDTs. This study was designed to evaluate the diagnostic accuracy of these two methods against nested PCR in Mawza District, Taiz Governorate during the peak seasonal transmission. In the present study, the
PfHRP-2/pLDH RDT and LM showed a fair level of agreement in their performance to detect
P. falciparum in the field, despite approaching 80 %. However, a substantial agreement was observed between RDT and LM for the detection of
P. falciparum among febrile patients. This is consistent with a recent study [
10] that reported a very good level of agreement between LM and CareStart™ HRP-2 RDT results among febrile patients. In the present study, RDT proved effective in detecting all LM-positive cases and in detecting a large proportion of LM-negative cases. The investigated RDT showed higher sensitivity than LM compared with nested PCR (96.0 vs 37.6 %), which is also higher than the sensitivity recommended by the WHO [
5].
The good performance of the
PfHRP-2/pLDH RDT in the field is evidently shown by its ability to detect
P. falciparum in all different degrees of microscopic parasite densities. Moreover, its sensitivity exceeds that of LM for parasite detection (93.9 vs 8.5 %) among afebrile participants, indicating its utility in active case detection. This could help in strategies for reducing malaria transmission by identifying asymptomatic carriers and their subsequent treatment. Similar findings of higher RDT sensitivity have been reported previously [
21]. The WHO has recently demonstrated a good level of sensitivity of RDTs in low parasitaemia [
5]. However, one should consider that not all RDT-positive cases correlated with those obtained by PCR. This in turn indicates that despite the better performance of RDT compared with LM, false positivity of RDT could not be ruled out. However, its performance is still superior to that of LM. In this respect, a moderate agreement (about 80 %) exists between RDT and PCR in detecting falciparum malaria among Yemeni patients in the field compared with a fair agreement (about 62 %) between LM and PCR. Most importantly, the
PfHRP-2/pLDH RDT showed a higher NPV than LM (90.4 vs 51.3 %) during the peak seasonal transmission of malaria. This is advantageous for the definite exclusion of malaria among patients, and the avoiding of unnecessary presumptive treatments. Given that the NPV is 100.0 % for the RDT and 40.0 % for LM among febrile patients, RDT-negative results for patients experiencing fever will be straightforward and will rationalize the prescription of anti-malarial drugs. Similarly, a very recent study [
22] recommends the use of RDTs for diagnosis of suspected malaria among symptomatic pregnant women but not for asymptomatic cases in Papua New Guinea. A limitation of the present study is that it included only a subset of the negative samples for molecular analysis. However, the large difference between RDT-positive and LM-positive samples, which is still of suspected positivity and could be due to persistent antigenaemia (108 samples of the RDT-positive ones), helps to avoid or, at least, reduce any possible bias.
The superiority of RDTs compared with LM could be explained by the fact that sequestered
P. falciparum missed by LM can be detected by RDTs because of the release of
PfHRP-2 by parasites and its circulation in the blood [
23,
24]. Meanwhile, the low sensitivity of LM in the present study could also be attributed to the high proportion of asymptomatic, very low-parasite density malaria cases. RDTs targeting
PfHRP-2 have been suggested as a better alternative to LM in areas of low-density parasitaemia and their false positives compared with LM have been confirmed by PCR to be cases below the threshold detection of LM [
25]. In addition to the diagnostic limitation imposed by microscopist expertise, the poor-quality LM in developing countries contributes to its low sensitivity in detecting low-parasite density infections. In Yemen, poor performance of LM for malaria diagnosis has been ascribed to low quality reagents, laboratory equipment and supplies [
2]. LM of low quality has been reported to influence its sensitivity and specificity for malaria diagnosis [
3].
In contrast, LM had higher overall specificity than RDT (97.6 vs 56 %) compared with the reference method. LM is still the gold standard for species identification and detecting the severity of malaria by quantifying parasitaemia and for differentiation of transmissible stages from those responsible for clinical disease [
26]. Low specificity of the
PfHRP-2/pLDH RDT in the present study is in contrast to the high specificity (96.1 %) recorded for the CareStart™ HRP-2 RDT tested in an earlier study [
10], which compared with microscopy for the detection of falciparum malaria among febrile patients. However, such low specificity is in agreement with a previous study comparing four brands of
PfHRP2-based RDTs for falciparum malaria diagnosis among febrile patients in Malawi, where specificity of 39–68 % was reported [
27]. It is noteworthy that two cases were positive with both LM and
PfHRP-2/pLDH RDT but negative by the PCR reference method. Although these were considered as false-positive results compared to nested PCR as the reference method, unperceived factors contributing to the inhibition of PCR could not be ruled out. Moreover, PCR false-negativity has been documented in the literature compared to LM [
12,
28‐
30]. The false positivity of the RDT in the present study could overestimate the prevalence rate by about 25 % as indicated by the PPV (76.9 %) compared with nested PCR. This is in agreement with the high false-positive rates of
P. falciparum using
PfHRP-2-based RDTs reported from Congo [
31,
32] and Burkina Faso [
33]. The
PfHRP-2-based RDT false positivity and its relatively low PPV could be attributed to the persistence of
PfHRP-2 antigenaemia in the blood circulation for 4–5 weeks after parasite clearance with successful treatment [
31,
33,
34]. The possible impact of persistent antigenaemia on the specificity of RDT investigated in the present could, in part, explain its dropped specificity to 30 % among patients with history of anti-malarial drug intake. Furthermore, malaria survey at the peak seasonal transmission, when prevalence rate is >10 % (Lina et al., unpublished data), may partially account for the low specificity of
PfHRP-2/pLDH RDT in the present study. Previous studies showed a negative correlation between the specificity of RDTs and malaria prevalence [
31,
35‐
37]. False-positive results by
PfHRP-2/pLDH RDTs can lead to overdiagnosis and subsequent overtreatment, which may contribute to the emergence and spread of drug resistance [
38]. Therefore, its combination with a more specific test is recommended. Furthermore, in addition to genus-specific pLDH,
P. falciparum-specific LDH-based RDTs should be evaluated for screening of falciparum malaria in Yemen. This may help avoid the drawback of
PfHRP-2 RDTs resulting from persistent antigenaemia in blood after treatment and cure, minimizing the false positivity rate to reasonable and acceptable levels. However,
PfHRP2 positivity in the absence of
P. falciparum-specific LDH or pan-specific LDH does not necessarily mean a false-positive result due to persistent antigenaemia [
39]. Although PCR is the most sensitive and specific tool for malaria diagnosis [
40], it is not practical for routine use in Yemen due to the limited resources.
Differences in sensitivity and specificity reflect on the estimation of falciparum malaria prevalence in the country, particularly among asymptomatic patients. In the present study, the overall prevalence of
P. falciparum was three times higher when using
PfHRP-2/pLDH RDT compared with LM and 16 times higher among asymptomatic patients. In this context, Mappin et al. [
41] reported a strong, non-linear relationship between malaria prevalence rates derived from the LM and RDTs. Higher RDT-based prevalence rates were also reported from Ethiopia and Tanzania, being two times and three times higher than those by LM, respectively [
42,
43]. Although the sensitivity of RDTs may, to some extent, reflect the true prevalence of symptomatic as well as asymptomatic-treated cases that cannot be detected by LM or PCR in the field over a certain period, the prevalence estimates by RDT and LM need to be standardized if they are to be used for epidemiological purposes, such as mapping [
41]. However, it poses a problem for case management, where unnecessary treatments could contribute to the emergence and spread of drug resistance. The better performance of RDTs over LM in field surveys has also been reported from the Brazilian Amazon [
44] and Angola [
45]. Overall, the findings of the present study suggest RDTs as a promising tool for epidemiological surveys in Yemen, even in low transmission settings and among asymptomatic carriers.