ehealth digital library

Digital library of
the Tanzania

Evaluation of Two Malaria Rapid Diagnostic Tests Quality Assurance (mRDT’s QA) Methods in Peripheral Health Facilities, Rural Tanzania.

Masanja, I. , Maganga, M. , Sumari, D. , Lucchi, N. , Udhayakumar, V., McMorrow, M. and Kachur, P. (2012) Evaluation of Two Malaria Rapid Diagnostic Tests Quality Assurance (mRDT’s QA) Methods in Peripheral Health Facilities, Rural Tanzania. Other. UNSPECIFIED. (Submitted)

Available under License Creative Commons Attribution Non-commercial.

Download (624kB)


WHO recommends confirming suspected malaria cases before initiation of treatment. Due to the imited availability of quality microscopy services, this recommendation has been followed with increased use of antigen-detecting malaria rapid diagnostic tests (mRDTs) in many malaria endemic countries. With the increased use of mRDTs, the need for a thorough mRDT quality assurance (RDT QA) method has become more apparent. One of the WHO recommendations for RDT QA is to monitor the tests in field use monthly, by comparing mRDT results to reference microscopy. This study was carried out to monitor mRDT performance in selected health facilities using two quality assurance methods; first based on ference microscopy and second based on detection of parasite DNA by real time quantitative PCR (qPCR) on dried blood spots (DBS); as well as assessing the cost and timeliness of the two QA methods. Blood samples were collected from patients undergoing a rapid test for malaria for two to three consecutive days per month, for five months, in 12 health facilities in Iringa rural and Mufindi districts. The health workers were instructed to label RDT cassettes, blood smear slides, and filter papers for DBS with matching unique ID stickers. A sticker was also placed in the log book where RDT results were recorded. Blood smears (BS) were first read at the district hospital (BS1) and then transported to Bagamoyo for a reference reading at the IHI- Bagamoyo laboratory (BS2). A third BS reader (BS3) was consulted from Muhimbili University of Health and Allied Sciences (MUHAS) in case of discordant results between BS1 and BS2. Molecular analysis involved extraction of parasite DNA from DBS samples using a QIAamp DNA Mini Kit. Sample DNA aliquots were compared against standard solutions with parasite DNA diluted 10-fold to give a parasitemia ranging from 200,000/μL to 20/μL. About 20% of the study DNA aliquots were sent to the CDC laboratory in Atlanta in order to validate qPCR results performed at the Bagamoyo laboratory. Data were entered in Microsoft Access (Microsoft Corporation, 2006) and analyzed in STATA 10 (StataCorp, Texas USA). Because of the known limitations of mRDTs to detect parasitemia below 200 parasites/μL, BS and PCR results greater than or equal to 5 parasites/200 WBC or 200 parasites/μL were considered positive in comparisons with mRDT performance. In the univariate analysis, proportions of positive tests were compared among the three types of tests: mRDT, microscopy and qPCR. Microscopy readings were categorized into 3 groups; BS1, BS2 and /or BS summary which is an average of BS1 and BS2. In case of discordant results between BS1 and BS2, a third reader- BS3 was consulted. Chi-squared test was done to assess differences in proportion of positive tests per district; whereas McNemar’s test was Malaria RDT QA Final Report, March 2012 5 used to assess the difference in test positivity by type of test. Kappa statistic was used to quantify the strength of the agreement between tests results. In addition, we examined health workers performance of the testing procedure when attending patients at a health facility, using a predefined checklist. Towards the end of the study, an evaluation of health worker acceptability was carried out to assess preferences between the two RDT QA methods. We received 2369 samples and 2324 (98%) had complete information. mRDTs had the highest positivity rate (6.5%). The proportion of positive tests by all types of tests was slightly higher in Iringa DC, but only qPCR and BS2 showed statistically significant differences in positivity rate between the two districts, where Iringa DC had more positive tests than Mufindi DC (p<0.05). When qPCR was a gold standard, mRDTs had higher sensitivity (68.6%, 95%CI: 55.0-79.7) than microscopy (53.7%, 95%CI: 38.7- 68.0) but highest mRDT sensitivity was achieved with comparison to microscopy (85.3%, 95%CI: 70.0- 93.6). All tests had higher inter-observer agreement than would be expected by chance. Substantial high inter-observer agreement (kappa =0.75; p<0.001) was seen amongst the microscopists i.e. district’s quality assurance officers and the reference microscopy readings. Assessment of the time needed to process BS at the district level revealed that, smears at district level took on average 8 days (min 2 to max 33) to be processed and provide feedback; but up to an average of 44 days (min 19 to max 98) to get a second reading. Many health workers were aware that the use of mRDTs was due to changes in treatment policy (11/30), and patients who qualify for the test are those suspected to have malaria. Majority (16/30) related assessment of control line as a measure of test accuracy and suggested the use of microscopy for quality control of mRDT results (15/30). Their major concerns were mRDTs’ inability to give parasite count, stock-out of the tests kits in their working areas and the frequency of negative results. This evaluation encountered several challenges, among them were 1. Poor quality of blood smears made at health facilities, especially dispensaries, which do not have laboratory services. 2. About 3.5% of BS1 slides could not be processed for BS2 because they were damaged during transportation and/or poor quality of smears. This accounts for the small difference in the numbers of BS assessed between two readers. 3. We were not able to prepare standard concentration solutions for qPCR analysis in the country. 4. Problems with PCR machine and inability to repair it that necessitated shipment of the machine, to and from, the manufacturers in Europe (Germany). Malaria RDT QA Final Report, March 2012 6 Due to these challenges, qPCR results were not available until after specimen collection had ended. In this study malaria positivity was higher with mRDTs than microscopy and qPCR for the 200 parasites/μL lower boundary of positivity threshold. This could either be due to the strict lower cut-off point for microscopy and qPCR parasite density or higher false positivity of mRDTs due to persistent antigen in blood, errors in mRDTs performance or other patient’s characteristics. When qPCR was taken as gold standard, mRDTs showed better sensitivity than microscopy, but when microscopy was regarded as a gold standard, mRDTs showed higher sensitivity than with qPCR. However, results of qPCR demonstrated a better correlation (inter-observer agreement) with those of microscopy than with mRDTs. The challenges of performing qPCR, as observed in this evaluation, make it unsuitable for quality assurance of mRDTs in routine care, Tanzania. The high inter-observer agreement between districts’ and reference microscopists (K=0.75) and higher tests performances of BS1 when BS2 was a comparator, demonstrates the competence shown by district’s technicians/ technologists to suffice their involvement as reference microscopists for quality assurance of mRDTs in their respective districts. This is also complimented by a fact that, both BS1 and BS2 had more similar performance when qPCR was taken as a gold standard. In this setting, a microscopy-based quality assurance system to assess mRDT performance in routine use may be a practical and suitable method. However, long distance transportation of smears should be avoided.

Item Type: Report (Other)
Keywords: malaria, diagnostic tests, Quality assurance, Health facilities, Tanzania
Subjects: Malaria > Surveillance, monitoring, evaluation
Malaria > Diagnosis & treatment
Divisions: Ifakara Health Institute > Impact Evaluation
Depositing User: Mr Joseph Madata
Date Deposited: 05 Dec 2012 09:26
Last Modified: 05 Dec 2012 09:26

Actions (login required)

Edit Item Edit Item


Downloads per month over past year

View more statistics