Quality Assurance (QA) Assessment for Malaria Rapid Diagnostic Test in Ngoma District, Eastern Province of Rwanda: A cross-sectional prospective study

Currently, malaria rapid diagnostic tests (mRDTs) are increasingly used for diagnosis of malaria, particularly in community where microscopy-based diagnosis is not practical. However, the diagnostic accuracy of mRDTs performed by the community health workers (CHWs) remains unknown. This study was conducted to determine the accuracy of mRDT results performed by CHWs in Ngoma district, eastern province of Rwanda. Method This was a cross sectional prospective study. 420 blood samples of patients self-reported to CHWs for malaria diagnosis were collected and analyzed by CHWs using mRDT and quality control tests were performed by using microscopy as a reference test. The study was conducted from 22nd April to 08th July, 2021. Results Among the 420 patients, 234 (55.71%) were females and 186 (44.29%) were males. Malaria test positivity was 2.62% by using mRDT and 1.67% by using microscopic test. The sensitivity and specicity of mRDT were 85.71% and 98.78% respectively. Negative predictive value, positive predictive value and accuracy of mRDTs were 99.75%, 54.54% and 98.57% respectively. Sensitivity of mRDT was below the WHO recommended sensitivity (>95%) although the specicity (98.78%) was within the WHO recommended specicity (>=90). There was a substantial agreement between mRDT and malaria microscopic test results, k=0.642. mRDTs continue to be an appropriate choice for malaria diagnosis in the absence of microscopy. predictive values rich protein-2 (RHP-2) mRDTs malaria ngoma comparing mRDTs results performed CHWs malaria microscopic results performed qualied laboratory centers or performance


I. Introduction
Internationally, malaria is a major public health problem. In 2014, the World Health Organization (WHO) reported that on average 3.3 billion people were at risk of malaria (1). In 2020, the WHO indicated that 229 million cases of malaria occurred worldwide, and 215 million cases of malaria (94%) occurred in Africa, and globally 409000 deaths occurred due to malaria in 2019 (2). Eastern and southern provinces are the most vulnerable regions to malaria in Rwanda. The  (3). The highest prevalence of malaria in Rwanda was observed in eastern part and estimated to be 11.11% (4). WHO recommend parasitological con rmation through microscopy or malaria rapid diagnostic test (mRDT) for each suspected patient (5), and WHO recommended that all mRDTs sensitivity (compared to microscopy as the "gold standard") should be above 95% and a minimum speci city of at least 90 % for all malaria species(6) (7). mRDTs accounted for 71% of the diagnostic tests performed in sub-Saharan Africa in 2014(8). A study conducted between 2009 and 2015 in 19 sub-Saharan countries revealed that the prevalence of malaria using microscopy test was 24.4% while the prevalence of malaria using mRDT was found to be 30.3% (9). During a study conducted in Nigeria for evaluating and comparing the performance of the microscopy as gold standard test with other four malaria rapid diagnostic test (mRDT) kits (carestart, SB Bioline, LabAcon and Global kits), their respective sensitivity of mRDTs were 83.7%, 86.5%, 84.9% and 86.5%, while their respective speci cities were 96.00%, 95.80%, 95.30% and 95.40% . it was concluded that those mRDT kits could not replace microscopy (10). Another study carried out in 2020 in Nigeria aimed at comparing the sensitivity and speci city of mRDTs (carestart TM) with the microscopy as gold standard, it was found that mRDTs had a sensitivity and speci city of 29% and 89% respectively. These low values of sensitivity were signi cantly associated with parasite density (pv < 0.001) and other mRDTs kit that could detect malaria parasites at low density were recommended (11). Another similar study conducted in Nigeria for assessing the performance of mRDTs in febrile under-ve children at a tertiary health facility level, demonstrated that mRDTs' sensitivity, speci city, positive and negative predictive values were respectively 40.3%, 89.6%, 81.8%, and 56.5%, respectively and it was recommended that febrile children with positive mRDT results would be con rmed as having malaria while those with negative mRDT results would be retested again using microscopy (12). Another study conducted in sierra Leone in 2020 revealed different prevalence of malaria in under-ve children by using mRDTs and microscopy (52.67% and 40.05% respectively) and the overall sensitivity and speci city of mRDTs were 85.52% and 69.23% respectively and these values were lower than the values recommended by WHO (95% and 90% respectively)(13)(6). Other studies assessed the application and accuracy of malaria diagnosis by mRDT among pregnant women in Nigeria. Both sensitivity and speci city of mRDTs were 75% and 25% respectively (14) and therefore they were far lower than recommended values of sensitivity and speci city (95% ,90% ) respectively (6) (7). Another study examined the performance of mRDTs among HIV positive individuals people in Nigeria and the overall sensitivity of mRDT was 58% while the speci city was 97% and it was recommended to monitor quality of mRDTs for malaria diagnosis among HIV co-infected persons (15). Furthermore, a comparative study where they were comparing the performance of three brands of mRDTs namely SD Bioline, Paracheck and Acon, it was found that their sensitivity and speci city were : SD Bioline (86.3%, 99.6%); Paracheck (50%, 97.7%,) and Acon (66.7%, 100%), respectively and they recommended that mRDTs quality should be strongly monitored during transportation and storage process(16). Similarly, a longitudinal study conducted in Tanzania in 2011, in which accuracy and impact of mRDTs were being assessed, sensitivity of mRDT was found to be 88.6% while speci city of mRDTs was 88.2% (17). These sensitivity and speci city values were high, yet they were lower than recommended values of sensitivity and speci city(6) (7). It was argued that the low values of mRDTs (low sensitivity and speci city) were largely in uenced by fever and parasites density. The authors recommended that the use of mRDTs should be coupled with supportive supervision in order to improve treatment of both malaria and non-malaria fevers and prevent the waste of anti-malaria drugs for malaria false positives patients (18).Other studies have shown that false-positive for results for mRDTs occur because the Plasmodium falciparum histidine rich protein 2 (pfHRP2) antigen remains in the bloodstream for some months after an infection is cleared which in turn could lead to overtreatment and the misdiagnosis of the true cause of symptoms (19). False-negative (FN) results of mRDTs were found to be in uenced by either parasite density which is below the mRDTs limit of detection typically in the range of 200 parasites/µL (20) or non -plasmodium falciparum malaria which is not detected by commonly used pfHRP2based mRDTs or adverse storage conditions of mRDTs where they are denatured by heat or humidity (21).
In 2008,the Rwanda Ministry of Health (MOH) piloted and scaled up the mRDTs (by community health workers) as recommended by WHO's Roll Back Malaria program (source) in the most of the regions as a way to expand and strengthen malaria diagnostic capacity throughout the country (22). In Rwanda, Community health workers (CHWs) are the rst point of contact with healthcare services for those with fever or a history of fever. Binome community health workers (a woman and a man) were trained to perform mRDT to all suspected malaria cases and provide anti-malaria drugs to symptomatic individuals in their villages (23). In 2018, a research was carried in kayonza district (Eastern province of Rwanda) and compared the performance of HRP-2 based mRDT and microscopy-based malaria test. This study found that the sensitivity of mRDT was 95.0% while speci city was found to be 59.2% among 264 suspected patients (24) and despite the extensive training of CHWs, the observed value for the speci city was below the acceptable levels. Moreover, little is known about the accuracy and predictive value of mRDT in the diagnosis of malaria infection at community level where asymptomatic people are more likely to have lower parasitaemia than in clinical settings (Vivi Maketa et al. 2013) [1]. Therefore, this study aimed to respond to this knowledge gap through an assessment of the accuracy, sensitivity, speci city and predictive values of histidine rich protein-2 (RHP-2) based mRDTs malaria in ngoma district, eastern province of Rwanda by comparing mRDTs results performed by CHWs with malaria microscopic results performed by quali ed laboratory technicians at health centers (HCs) in order to assure or safeguard the performance of mRDTs.

Research Design and Target Population
This was a cross-sectional prospective study. All patients self-reported to CHWs for malaria diagnosis from 22 nd April to 08 th July, 2021 in Ngoma district who were willing to participate in this study were included. Patients who had recovered from malaria within the past three weeks were not included in this study because it was revealed that malaria antigen can be found in blood for 3weeks after completion of treatment which could increase false positive results (17).
Patients who were taking anti malaria drug were not included in this study.

Sample size
Sample size was calculated by using the statistical formula of Cochran (1963:75) for calculating sample size for in nite population was used. It says that for large population for which the variability in proportion is unknown, it is assumed that the maximum variability is equal to 50% (p =0.5) and at 95% con dence level with ±5% precision, the determination of needed sample size is as follows: where critical value of desired con dence level z =1.96 the desired level of precision e=0.05, proportion p= 0.5 (25). Therefore, the desired sample size . By adding the rate of no respondent of 10% (38), the needed sample size, n, becomes 422 malaria suspected patients. In the end only 420 malaria suspected patients were considered in this study.

Study site
The study was conducted in Ngoma district , one of the six high-malaria burden districts in eastern province of Rwanda which has the highest malaria

Materials and data collection procedures
A total of 109 binomes (CHWs) were randomly selected to take a blood sample and performed mRDT. A sample for thick blood smear for malaria microscopic test was also taken by the CHWs and the slides were taken to health center (HC) for processing by quali ed microscopists. At the HC, the thick blood smears were stained using a solution of 10% Giemsa and left for 30 minutes and then washed in buffered water of pH 7.2. The blood smears were examined using 100x high power elds (26) and malaria parasites density was counted by using the formula developed by WHO, where: parasites/μL blood =( No of parasites counted x 8000 white blood cells/μL) / No of white blood cells counted (27). Results of malaria microscopic tests and mRDT results were collected using a data collection form created using Epi info. Laboratory technicians at HCs were trained by Kibungo referral hospital on how to diagnose malaria using microscopy and how to calculate malaria parasites density before the study and similarly 15 data collectors were trained on how to collect data by using a data collection form created in Epi info. Smart phones were used to collect all data.

Data analysis procedure
A mRDT positive case was de ned by the presence of P. falciparum or a pan positive test line while a mRDT negative case was de ned by the absence of P. falciparum and a pan test line. Results of mRDT were compared with results of malaria microscopic test and accuracy, sensitivity, speci city positive and negative predictive value of mRDT (RHP-2 based) were presented in terms percentages. Cohen's Kappa test was performed in order to determine level of agreement between mRDT and malaria microscopic results. Sensitivity and speci city of mRDT (RHP-2 based) kits were compared to WHO recommended value which is greater 95% for sensitivity and value which is greater or equal to 90% for speci city.

Results
3.1. Socio-demographic characteristics of the studied population.
Among 420 patients who were considered in this research 234 (55.71 %) were females and 186 (44.29 %) were males. Minimum age was 18 years while maximum age was 91 years. Mean age was 34.32 years with standard deviation of 14.01 as shown in Table 1 below.

Comparison of mRDT results with malaria microscopic results
Malaria positivity was found to be 2.62% by using mRDT and 1.67% by using microscopy as shown in Table 2. No cases of pan-positive were found during this study, all positive cases were due to P.falciparum. The average parasite density obtained by microscopy was 451.8 with a range of 40-1200 parasites/uL In this study, the malaria microscopic test was used as reference test. The frequencies of mRDT's false positive and negative results are summarized in Table  3. In this study mRDT had a sensitivity of 85.71% , 98.78% speci city, negative predictive value was 99.75%, positive predictive value was 54.54 %, and Cohen's Kappa test K was found to be 0.642. One false negative mRDT was observed with a microscopic malaria parasite density of 1200 parasites/ μl. During this study, by using logistic regression as shown in table 4& 5, it was found that having malaria parasite density which was equal or greater than 100/ μl and was signi cantly associated with mRDT's sensitivity > 95% with P-V= 0.006 at 95% CI (2.6397 342.6845 ) and OR of 30.07. This means that malaria parasites density which was equal or greater than 100/ μl was 30.07 times more associated with having sensitivity > 95% than malaria parasites density less than 100/ μl. However, speci city was not statistically associated with malaria parasites density.

Discussion
This study provided information concerning diagnostic accuracy of mRDT performed by CHWs in Ngoma district. In this study, malaria positivity by using mRDT was slightly higher than malaria positivity by using microscopy. This nding agreed with several studies conducted by Ekom et al, Oliver et al and Mohamed et al(28)(9)(13) which reported higher malaria prevalence of 75% , 30.3% and 52.67% by using mRDTs compared to malaria prevalence of 60%, 24.4% and 40.05% reported by using malaria microscopic test respectively.
In this study, mRDT had a sensitivity of 85.71% which was below the sensitivity recommended by WHO (> 95%) and a speci city of 98.78% which is within the recommended WHO value (6  (12). This satisfactory speci city was not associated with malaria parasite density and is very important because it helps to differentiate other malaria-like illness from malaria and health care provider will be able to administer the right medications.
In this present study, positive predictive value and negative predictive value were 54.54% and 99.75% respectively, comparing them to another study conducted in Nigeria, positive predictive value and negative predictive value were 81.8%, and 56.5%, respectively, Negatives predictive values seem to be nearly equal but positive predictive values are slightly different (12). This very low positive predictive value (54.54%) reported in this present study was due to high false positive results of mRDT. False positive results of mRDT were found to be associated with high level of rheumatoid factors in patients (36) and the persistence of HRP-2 antigens in blood for up to 61 days after completion of treatment (19) which is another limitation of mRDTs. However, in the bigger scheme of things, treatment of false positive cases is less detrimental compared to lack of treatment of the false negatives although it contributes to overuse of antimalarials and appropriate treatment of missed non-malaria cases. Therefore, a con rmatory test is needed in case of doubtful positive result of mRDT.
Overall, the accuracy of mRDTs in this study was 98.57% implying that mRDT kits used in this study gave correct results either positive or negative at the level of 98.57% and a Cohen's Kappa test k is equal to 0.642, meaning that there was a good concordance with microscopy. Given that majority of malaria cases are diagnosed at the community lever by CHWs using mRDTs, these results demonstrates that the use of mRDTs by CHWs is appropriate and help increase the trust of population toward health care service provided by CHWs. However, continued supportive supervision and quality assurance of mRDTs is important to ascertain the quality of case management of malaria.

Conclusion
Overall, the accuracy of mRDTs in this study was acceptable and results were well correlated with microscopy results. This implies that the use of mRDTs by CHWs in Ngoma district is appropriate. The continued supportive supervision and quality assurance of mRDTs is important to ascertain the quality of case management of malaria at the community especially given that majority of malaria cases are managed at the community.

Declarations
Ethics approval and consent to participate Before starting data collection, ethical clearances were sought from Institute Review Board of University of Rwanda/ College of Medicine and Health Science and Kibungo referral hospital independent ethic committee. Before patients provided blood samples, they signed consent forms voluntarily. In order to ensure con dentiality of patients who provided samples, names of patients were removed from data set and every patient was represented by his/ her unique code auto generated by Epi info. In order to ensure the security of data set, it was protected with a password and stored in a laptop protected with a password.

Consent for publication
Not applicable

Competing interests
The authors declare that they have no competing interests Availability of data and materials The dataset used and analyzed during the current study is available from the corresponding author on reasonable request.

Funding
This study was fully funded by African Field Epidemiology Network (AFENET). The funding body had no role in the study design, collection, data analysis, data interpretation, or drafting of the manuscript.