The WHO has recently recommended that treatment with anti-malarial drugs should be given to patients with malaria parasites confirmed by laboratory tests or RDTs in areas lacking the capacity for detection of malaria parasite by microscopy . The Tanzanian Ministry of Health and Social Welfare, through the NMCP plans to scale-up malaria diagnosis by introducing RDTs to improve case management and reduce inappropriate dispensing of ACT, which was introduced in the country in early 2007 . However, RDTs have, as also confirmed in the present study some technical and operational challenges including their accuracy and implementation. The accuracy of RDTs which is commonly measured by their sensitivity and specificity (when compared to microscopy as a gold standard) is very critical to avoid denying anti-malarial drugs to patients with malaria due to false negative results and unnecessary dispensing of drugs by treating patients with false positive RDTs. Furthermore, as recently shown [18, 30, 31], syndromic treatment due to lack of diagnostic facilities or non-adherence to RDT results, reduces the motivation for health-care providers to search and treat alternative causes of fever, and thus perpetuating high drug pressure which might eventually lead to parasite tolerance/resistance to ACT.
The findings of this study showed that both RDT brands (Paracheck and ParaHIT) had low sensitivity but relatively high specificity depending on fever status, parasite density, malaria parasite prevalence and transmission intensity. The overall sensitivity of RDTs was significantly higher in the longitudinal study compared to CSS and decreased with declining malaria parasite prevalence in both studies corresponding to the declining burden of malaria in the study areas as recently shown by Mmbando et al.  and Ishengoma et al (unpublished data). The low sensitivity observed in the CSS might mainly be due to large number of cases with low level parasitaemia, which was below the detection limits of RDTs giving rise to high rate of false negative results. The low sensitivity of RDTs observed in the current studies was similar to what has been reported in other field studies conducted in South-eastern Tanzania [21, 32] and other parts with similarly low malaria endemicity [33–35]. However, it was lower than the sensitivity reported in Kilombero and north-eastern Tanzania before the recent decline in the burden of malaria [31, 36] and also other parts of Africa (including East Africa, Kenya and Uganda [37–39], and others [15, 40, 41]).
Despite the differences in parasite density in the two studies (longitudinal study and CSS) which increased with decreasing parasite prevalence, the sensitivity of RDTs was higher in febrile cases, and this was similar to what has been reported in studies conducted elsewhere in Africa [19, 37, 42]. Increasing parasite density with declining malaria burden observed in the study villages could be due small sample size in the recent years which led to inflated geometric mean parasite density as recently shown by other studies . However, the geometric mean parasite density observed in both studies was lower than those reported in other studies using similar RDTs based on HRP-2 detection  and this could be the cause of relatively lower sensitivity observed in these community studies. Although the sensitivity of RDTs was higher in under-fives compared to those aged ≥5 years, the difference was not statistically significant after adjusting for other covariates. This is different from the findings reported from Kilombero, where the sensitivity of RDTs was significantly lower among older patients .
Despite significantly higher specificity in the CSS compared to the longitudinal study (which increased with declining parasite prevalence), the specificity observed in both studies was higher than what has been recently reported with similar RDTs (based on HRP-2) in Malawi . False positive RDT results which lead to low specificity are commonly attributed to persistence of HRP-2 antigens mainly due to continued exposure to low level infections leading to sub-patent parasitaemia, gametocytaemia or delayed clearance of HRP-2 after treatment . HRP-2 antigens remain in blood for over 30 days after clearance of the parasites and persistence of HRP-2 has been shown to depend mainly on parasite density at the initiation of treatment . High specificity observed in the CSS indicate that most of the cases without malaria parasites residing in communities with low level of parasite prevalence will most likely be correctly diagnosed by the RDTs. However, the low specificity of RDTs among cases with fever in the longitudinal study which resulted from large number of false positive RDTs indicates that most of such cases were treated with anti-malarials despite negative results by microscopy leading to over-treatment and wastage of drugs.
Before introduction of RDTs, almost all cases attended by CORPs were treated with SP based on presenting symptoms leading to a significant level of over-treatment, since only 20.1% of all cases treated with SP during this period had malaria parasites confirmed by microscopy. However, this approach was cost-effective due to low cost of SP and was also safe since most of the febrile cases (regardless of parasite infection status) were promptly treated with anti-malarials thus reducing the risks associated with severe malaria, which often occurs when a patient is left without treatment. Despite the observed low accuracy of RDTs, which was lower than the sensitivity and specificity of ≥95% recommended by WHO , deployment of RDTs in the longitudinal study reduced anti-malarial dispensing by CORPs from 99% in the pre-RDT to 32% (among cases aged ≥5 years of age) in the post-RDT period. Similar studies conducted in other parts of Tanzania have recently shown that correct use of RDTs and adherence to test results by health workers reduced dispensing of anti-malarial drugs by > 60% . Furthermore, majority of the cases treated with ALu during the post-RDTs period were confirmed to have malaria parasites by microscopy since only 28.6% of the cases treated had negative results by microscopy while during the pre-RDTs, 78.5% of the cases treated with SP had negative results.
Adherence to treatment guidelines by CORPs as stipulated in the SOPs was high (> 95%) in cases of all age groups. The level of adherence to treatment guidelines shown in this study was higher than what was reported by previous studies conducted in Zanzibar  and Burkina Faso [42, 46] possibly due to the training provided to CORPs and weekly supportive supervision conducted by an experienced team. However, some cases confirmed to have malaria by microscopy (during both periods) were not treated with anti-malarials and were also not referred to the nearest health facilities for further treatment as required by the guidelines. These were relatively few (< 0.5%) and could possibly return to CORPs for treatment in case their medical conditions worsened because the services were readily available within their communities. The level and consistency of supervision has been shown to influence the proper implementation of RDTs and adherence to test results by service providers under routine health facility settings [21, 32]. However, such level of supervision can hardly be provided and maintained by the health authorities when RDTs are implemented and widely used under routine clinical settings. Thus, during deployment of RDTs, health authorities will need to design and implement a sustainable scheme of supportive supervision (relevant to their local settings) to ensure the accuracy of RDTs is maintained and service providers adherence to National treatment policy and guidelines. Such strategies will also help to maintain health workers confidence on RDT results, reduce the need for syndromic treatment and target management of other causes of fevers in non-malaria cases.
The current study utilized two brands of HRP-2 based RDTs (Paracheck and ParaHIT) which showed variable results regarding their accuracy, whereby both brands had similar sensitivity in the longitudinal study while Paracheck showed higher sensitivity in the CSS. In both studies, ParaHIT had higher specificity compared to Paracheck. Although these differences could be a true reflection of their performance under field conditions in an area where malaria has remarkably declined , this study was not designed to address such question. The tests were used based on their availability from the NMCP and thus the study did not make any attempt to compare their performance.
Furthermore, the performance of RDTs was compared to expert microscopy whose quality cannot be attained under health facility settings as recently shown by studies conducted in Tanzania  and other malaria endemic areas . The tests were also used for parasite detection in CSS in order to identify cases that were targeted for in-vivo efficacy study of anti-malarials and in-vitro monitoring of drug resistance (which were stopped due to lack of enough samples). Unlike in health facilities and community studies where febrile patients are commonly attended and RDTs are intended to detect patients with parasites, the tests cannot be similarly used in CSS due to low number of patients with fever, low parasite density and therefore low sensitivity as shown by the findings of the CSS. However, these findings provide important information which will potentially guide future applicability of RDTs when fully introduced in many endemic countries for malaria diagnosis in the health facilities and communities; emphasizing that the tests have more relevance for case management than for disease surveillance.
In the longitudinal study, some cases with malaria parasites (confirmed by microscopy) were neither treated with anti-malarials (SP or ALu during the pre-RDT and post-RDT periods, respectively) nor given a referral to be attended at nearby health facilities as stipulated in the guidelines given to CORPs. While it would intuitively be important to know what happened to such cases, these issues are beyond the scope of this study since only passive case detection of fever episodes was performed, and follow-up of cases after treatment or those with negative RDT results who were not given anti-malarials was not done. However, other studies conducted in northern Tanzania showed that only 1% of children with negative RDT results developed parasitaemia during the 28 days of follow-up and only 1 out of 816 children developed a severe malaria condition demanding hospitalisation (Reyburn et al. personal communication). Based on the setting of the longitudinal study where medical services were made available in the community through CORPs, cases not given anti-malarials had an opportunity to get treatment whenever they felt unwell, thus making this strategy safe and rational for testing the implementation of malaria case management based on RDT results.