Epidemiological data on malaria in Western Kasai, DRC, are limited and inadequate, despite this country is suffering the second-highest burden of malaria globally ; the infrastructure to undertake complex surveillance programmes for a reproducible and efficient surveillance are very poor. This initiative carried out on 306 children produced alarming results: more than 80 % were affected by severe malaria (as suggested by haemoglobin (Hb) measurements performed at the time of testing), and also due to the unexpected species P. vivax. Moreover, and for the first time in DRC, human infections with the zoonotic Babesia microti, which may contribute to anaemia, were detected.
Comparing the present findings with available data, malaria prevalence (82.8 %) evidenced by MD in children of Tshimbulu pointed out a sanitary problem which turned out to be more alarming (43.7 %) than reported by molecular studies on malaria epidemiology in Western Kasai  and, in general, in DRC (33.5 %, 34.1 %) [2, 3]. Furthermore, the presence of P. malariae and P. ovale (with relative prevalence 8.5 and 1.2 %) was confirmed and the additional presence of P. vivax was evidenced. In the studied area, where the lack of the Duffy antigen is widespread, the existence, although at very low prevalence (1.2 %), of this unexpected Plasmodium species should draw attention to the possible clinical and epidemiological consequences. Indeed, it is well known that: i) P. vivax gametocytes are produced earlier than those of P. falciparum , and are transmissible to Anopheles mosquito vectors at lower parasite densities than those of P. falciparum, and more efficiently ; therefore, transmission before diagnosis or treatment may occur ; ii) it is often the ‘last parasite standing’ following P. falciparum elimination [18, 19]; and, iii) as P. falciparum, this species is resistant to chloroquine . That is: all the biological features of P. vivax make it necessary to be aware of its geographical distribution in order to plan appropriate control measures, aside from applying a suitable therapy to infected subjects.
Notwithstanding these relevant findings, the accuracy of diagnoses performed in the limited resource setting of DRC and, then in a research laboratory, was not entirely satisfactory. Indeed, considering that MD was not applied to 106 samples (93 not collected plus 13 not suitable dried spots), and, therefore, in these cases the three methodologies could not be compared, analysis of the results obtained by means of RDT, MA of the blood and MD, suggests the following observations. In respect of overall infection rates, RDT1 and MA1 results were in greater agreement with MD than RDT2 and MA2 findings that, however, were not significantly different from MD (Table 2). On the contrary, when the results are analysed in detail, only MA2 is in agreement with MD: there are four Plasmodium species affecting children in Tshimbulu, and Babesia microti is present, even if all of them are in low percentages, which means in loco performed analyses are satisfactory in detecting infection, but not as satisfactory as identifying the species involved.
However, each diagnostic tool used in loco and then, in a research laboratory during the revision of the pictures of RDT, smears and dried filter papers from Tshimbulu, confirmed the intrinsic bias, which may explain the contrasting results.
Locally applied RDT demonstrated its usefulness for rapid diagnosis of malaria; however, it is useful to note that at least 1.9 % of infections remained undiagnosed, and Pan weak bands were exposed to doubtful interpretation, probably due to partial corruption during transportation (it is known that test line intensity may decrease with either parasite density and exposure to >70 % humidity and/or >30 °C). Therefore, in respect of the performance assured (for P. falciparum: sensitivity 99.7 % and specificity 99.5 %), the employed RDT proved to be less sensitive and specific. Moreover, as expected (data sheet informs that three-band positive sample do not means mixed infection), it did not allow identification of mixed infections and, even less so, the species involved in infections, data very important for possible relapses due to the presence of P. vivax and P. ovale and for possible renal damage induced by P. malariae. Considering the additional well-known criticism about the inability to evaluate parasitic load, which is essential to define the severity of the infection, the immediate measures necessary, and the treatment efficiency, RDT cannot be the only employed method.
Regarding MA, the present findings demonstrate its high performance in detecting all Plasmodium species present only when carried out by well-trained microscopists in a laboratory setting. Otherwise, when a high number of children are awaiting diagnosis, important detail of infections (such as mixed infections) may be overlooked or not recognized. The lower (not significant) sensitivity shown by MA2 (80.0 %) when compared to MD (82.8 %), accepting as a true result the presence of 15 microscopically negative smears (as confirmed by negative RDT) in the first group of children unexamined by MD, becomes less relevant, and totally overlaps the results from children examined by both diagnostics (100 % sensitivity). The same applies for mixed infections: several mixed infections detected by MA2 in the first 100 children may have been lost by the unapplied MD, so that the prevalence of mono-infections with P. falciparum stated by MD is only apparently higher than that identified by MA2, as confirmed by the excellent concordance (K = 0.86) about simple and mixed infections observed on the remaining 164 children proved positive to both diagnostics. However, results obtained by this method sometimes proved to be uncorrected. In detail, co-infections of P. falciparum with P. malariae diagnosed during the revision of the slides proved overestimated, may be due to a misidentification of P. falciparum in cases of severe malaria (the data on corresponding children reported a very low Hb concentration). Indeed, due to P. falciparum cyto-adherence, in these cases peripheral blood included trophozoites developed further than usual little rings: they are very alike P. malariae and infection may be identified as co-infection. This pitfall has been overcome during MA1 because the local team, who frequently identify severe malaria, is probably used to recognizing P. falciparum in these conditions. The second (possibly in the absence of corresponding MD identification) misdiagnosis of MA2 (namely some P. falciparum interpreted as P. vivax/ovale) occurred in cases of macrocytic anaemia and when slides were badly stained, which make it reasonable that there was an absence of evident Schüffner’s granulations: some single infection with P. falciparum may have been read as possible mixed infection. Finally, Babesia microti has been over identified, probably as it is very similar to P. falciparum when just entered in the erythrocyte. In these circumstances, as well as to distinguish the two major forms of P. ovale, parasite morphology proved unsuitable for a systematic analysis of the relationship between the different species and only PCR-based methods proved to be reliable. However, in this study several underestimated mixed infections diagnosed by MD were observed. It could be strictly linked to the difficulties in extracting DNA from the dried blood spots and, consequently, to the possibility of identifying only the most abundant species. Therefore, MD may present weak spots that affect its performance.
As for the comparison of the performance shown by the applied malaria tests, the higher (however in this case not significant) sensitivity of MD in respect to MA and RDT, and the ~90 % specificity of the RDT used, reported in previous studies [3, 20] were confirmed. However, in this experience it is crucial to underline the ability of the local team in detecting positive slides (81.2 % instead of 82.8 %), higher than that reported in the abovementioned paper (22.7 vs 34.1 %) .