In Madagascar, the integrated management of febrile syndromes at the community level remains a priority NMCP strategy for providing impoverished populations with diagnosis results and effective anti-malarial treatment within 24 hours after the febrile individual has access to care. In fact, this strategy relies on a system of CHWs and staff working in health centres, which has already proven to be effective for many years. The operational research project, evaluating the use of ACT and the implementation of RDTs, works through a network of CHWs and existing operations, which enabled us to obtain reliable data on the efficacy of "diagnostic" and "therapeutic" agents. Despite encountering difficulties in establishing a consensus for the type of RDT to be used in Madagascar, the NMCP introduced "combo" RDTs, which detect the HRP2 antigen specific for P. falciparum and the pLDH of four Plasmodium species: P. falciparum, P. vivax, P. malariae and P. ovale [12, 14, 15].
To evaluate the performance of RDTs used by CHWs, their results were compared to those obtained from the two most common methods used to diagnose malaria: PCR, which has the best performance in terms of sensitivity and specificity due to its very low threshold of detection (< 1 parasites/μL); and microscopy, which remains the WHO reference diagnostic despite its higher detection limit (10-50 parasites/μL) . For the 24-month period of the study, the number of febrile episodes in children that consulted with CHWs in the two study areas of high and low transmission was not significantly different. The only notable difference was that the mean age of the febrile children seen by CHWs in the areas of low transmission was higher than that of the children in the high transmission areas (~3 months, P = 0.009).
From an epidemiological standpoint, malaria burdens in the two study areas were assessed through the use of PCR. On average, 51.2% of the children consulting CHWs were carriers of parasites, and this proportion in area of high transmission was nearly twice as high as that in the low-transmission area (56.7% vs. 31.9%, respectively; P < 0.0001). However, this difference was not evident if microscopy, which detected an infection prevalence of 37.2%, was used as the diagnostic method. This finding demonstrates the importance of using molecular methods to study malaria epidemiology [17, 18]. In addition, all children with parasites were infected only with P. falciparum, the predominant species in Madagascar [19, 20]. However, microscopy allowed measuring the extent of parasite densities (48 to 82,000 parasites/μL) in the infected children. Although the difference was not statistically significant, the mean parasite densities were higher in high-transmission areas than in low-transmission areas (2,800 parasites/μL vs 1,500-2,200 parasites/μL). This is consistent with the idea that the "pyrogenic" thresholds of individuals in high-transmission areas are higher than those in low-transmission areas, and that below this threshold, patients do not suffer from malaria "disease" [21, 22].
From a practical standpoint, no major difficulties in the use of RDTs were noticed by CHWs, despite reports to the contrary [23, 24]. Training before the start of the study, the availability of simple data collection tools, and the supervision of CHW activities by physician supervisors facilitated RDT implementation . The simplicity of the RDT procedure has also avoided many errors of manipulation or interpretation. The few difficulties encountered by the CHWs were mainly related to the amount of blood drawn at the finger (especially for very young children) or the reading and interpretation of the results of certain tests, as has been previously noted by Seidahmed et al .
The performance of CHW-administered RDTs was determined by comparing the RDT results to those obtained by PCR and by microscopy. The levels of confirmation were similar for the CHWs in all the study areas. Discrepancies between RDT results and those of microscopy or PCR were mainly related to their different thresholds of detection as has often been observed either at the health centres [12, 15, 16, 27, 28] or at the community level [29, 30]. False positives (HRP2 antigen detection only) generally corresponded to the detection of old malaria infections (and correctly treated) and were related to the persistence of the HRP2 antigen in the bloodstream (up to three weeks after malaria infection) [31, 32]. Overall, the performance of RDTs used by CHWs was similar to that of microscopic analysis performed by an experienced microscopist. Compared to microscopy, RDT sensitivity was excellent (95.9%) both in high-transmission areas (90.2%) and in low-transmission areas (93.7%). However, RDT specificity relative to that of microscopy was lower (87.0%) as previously reported [33–35]. Finally, the RDTs produced an excellent NPV of 95.2%. These encouraging results justify using RDTs to diagnose malaria in areas that are most in need of low-cost diagnostic techniques.
If CHWs had only treated children found RDT-positive, only four patients who were RDT-negative and microscopy-positive, or 2% of the patients seen, did not receive anti-malarial treatment. These four patients who had very low levels of parasitaemia (less than 1,000 parasites/μL) would have been tested again a day or two later by the community worker if clinical signs had persisted. Another encouraging sign for the introduction of RDTs at the community level was the good long-term stability of the RDTs over time, given they were kept in the field conditions by the CHWs. The potential benefits of RDT-negative results (that excluded a malarial infection) could not be acted upon in the study. However, based on the PCR results, 49% of the children with fever who consulted CHWs were not infected with malaria. There were significantly more of these cases in the low-transmission areas (68.1%) than in the high-transmission area (43.3%). These data show that this testing strategy is solely limited to detect malaria infections and malaria is only a fraction of the fever etiologies seen at the community level. It is clear that further studies should be conducted to determine the agents responsible for non-malarial fever and to develop simple algorithms (clinical, biological and therapeutic, if possible) to better manage cases of non-malarial fevers.