Accuracy of malaria rapid diagnostic tests in community studies and their impact on treatment of malaria in an area with declining malaria burden in north-eastern Tanzania

Abstract

Background: Despite some problems related to accuracy and applicability of malaria rapid diagnostic tests (RDTs), they are currently the best option in areas with limited laboratory services for improving case management through parasitological diagnosis and reducing over-treatment. This study was conducted in areas with declining malaria burden to assess; 1) the accuracy of RDTs when used at different community settings, 2) the impact of using RDTs on anti-malarial dispensing by community-owned resource persons (CORPs) and 3) adherence of CORPs to treatment guidelines by providing treatment based on RDT results.

Methods: Data were obtained from: 1) a longitudinal study of passive case detection of fevers using CORPs in six villages in Korogwe; and 2) cross-sectional surveys (CSS) in six villages of Korogwe and Muheza districts, north-eastern, Tanzania. Performance of RDTs was compared with microscopy as a gold standard, and factors affecting their accuracy were explored using a multivariate logistic regression model.

Results: Overall sensitivity and specificity of RDTs in the longitudinal study (of 23,793 febrile cases; 18,154 with microscopy and RDTs results) were 88.6% and 88.2%, respectively. In the CSS, the sensitivity was significantly lower (63.4%; χ2=367.7, p<0.001), while the specificity was significantly higher (94.3%; χ2=143.1, p<0.001) when compared to the longitudinal study. As determinants of sensitivity of RDTs in both studies, parasite density of<200 asexual parasites/μl was significantly associated with high risk of false negative RDTs (OR≥16.60, p<0.001), while the risk of false negative test was significantly lower among cases with fever (axillary temperature ≥37.5 °C) (OR≤0.63, p≤0.027). The risk of false positive RDT (as a determinant of specificity) was significantly higher in cases with fever compared to afebrile cases (OR≥2.40, p<0.001). Using RDTs reduced anti-malarials dispensing from 98.9% to 32.1% in cases aged ≥5 years.

Conclusion: Although RDTs had low sensitivity and specificity, which varied widely depending on fever and parasite density, using RDTs reduced over-treatment with anti-malarials significantly. Thus, with declining malaria prevalence, RDTs will potentially identify majority of febrile cases with parasites and lead to improved management of malaria and non-malaria fevers.

Figure 1

Overall sensitivity of RDTs by parasite density for blood smear positive samples in the longitudinal study and CSS in Korogwe and Muheza districts. Bars represent proportion of cases with positive blood smear results by different categories of parasite density, asexual parasites/μl (black bars = longitudinal study, n = 3793; and grey bars = CSS, n = 1045); Solid line = sensitivity of RDTs in the longitudinal study and dotted line = sensitivity of RDTs in the cross-sectional surveys (CSS)

Figure 2

Sensitivity of RDTs by malaria parasite prevalence stratified by the year of study. Bars represent proportion of cases with positive blood smear results by microscopy by year of study from 2007 to 2010 (black bars = longitudinal study and grey bars = cross-sectional surveys - CSS); Solid line = sensitivity of RDTs in the longitudinal study and dotted line = sensitivity of RDTs in the cross-sectional surveys (CSS)

Figure 3

Specificity of RDTs by malaria parasite prevalence stratified by the year of study. Bars represent proportion of cases with positive blood smear results by microscopy by year of study from 2007 to 2010 (black bars = longitudinal study and grey bars = cross-sectional surveys - CSS); Solid line = specificity of RDTs in the longitudinal study and dotted line = specificity of RDTs in the cross-sectional surveys (CSS)