True versus apparent malaria infection prevalence: the contribution of a Bayesian approach.

AIMS: To present a new approach for estimating the "true prevalence" of malaria and apply it to datasets from Peru, Vietnam, and Cambodia. METHODS: Bayesian models were developed for estimating both the malaria prevalence using different diagnostic tests (microscopy, PCR & ELISA), without the need of a gold standard, and the tests' characteristics. Several sources of information, i.e. data, expert opinions and other sources of knowledge can be integrated into the model. This approach resulting in an optimal and harmonized estimate of malaria infection prevalence, with no conflict between the different sources of information, was tested on data from Peru, Vietnam and Cambodia. RESULTS: Malaria sero-prevalence was relatively low in all sites, with ELISA showing the highest estimates. The sensitivity of microscopy and ELISA were statistically lower in Vietnam than in the other sites. Similarly, the specificities of microscopy, ELISA and PCR were significantly lower in Vietnam than in the other sites. In Vietnam and Peru, microscopy was closer to the "true" estimate than the other 2 tests while as expected ELISA, with its lower specificity, usually overestimated the prevalence. CONCLUSIONS: Bayesian methods are useful for analyzing prevalence results when no gold standard diagnostic test is available. Though some results are expected, e.g. PCR more sensitive than microscopy, a standardized and context-independent quantification of the diagnostic tests' characteristics (sensitivity and specificity) and the underlying malaria prevalence may be useful for comparing different sites. Indeed, the use of a single diagnostic technique could strongly bias the prevalence estimation. This limitation can be circumvented by using a Bayesian framework taking into account the imperfect characteristics of the currently available diagnostic tests. As discussed in the paper, this approach may further support global malaria burden estimation initiatives.

Application of multiple DNA fingerprinting techniques to study the genetic relationships among three members of the subgenus Trypanozoon (Protozoa: Trypanosomatidae).

Three different DNA fingerprinting techniques, the mobile genetic element (MGE)-PCR, simple sequence repeat (SSR)-PCR and random amplified polymorphic DNA (RAPD)-PCR, were used to define a large set of genetic markers to study genetic similarity within and among Trypanosoma brucei, Trypanosoma equiperdum and Trypanosoma evansi strains (n=18) from China, Africa and South America and to investigate their genetic relationships. Using the three fingerprinting techniques, >890 bands (ranging in size from 0.2 to 2kb) were defined for all 18 strains of Trypanosoma. Within each of the strains, 39-59 bands were defined. The similarity coefficients between strains ranged from approximately 41 to 94%, with a mean of 65%. There was more genetic similarity among strains within T. evansi (mean of approximately 79%) compared with T. equiperdum ( approximately 65%) and T. brucei ( approximately 59%). The similarity coefficient data were used to construct the dendrogram, which revealed that (irrespective of species) the majority of strains from China and South America grouped together to the exclusion of those from Africa. The exceptions were a T. brucei strain from Africa and a T. equiperdum strain of unknown origin. Hence, employing data sets generated using the three different fingerprinting methods, it was not possible to unequivocally distinguish among T. brucei, T. evansi and T. equiperdum, although there was a tendency for T. evansi strains to group together to the exclusion of T. brucei. The findings provide support for the hypothesis that T. evansi originated from a mutated form of T. equiperdum and stimulate further investigations of the genetic make-up and evolution of members of the subgenus Trypanozoon.