A total of 7,769 samples were tested by ELISA (). Of these samples, 26 specimens in 23 sites were found to be positive (data not shown).
Flow chart showing processing of samples.
For PCR confirmation of ELISA-positive specimens we developed a Taqman real-time PCR assay to improve specificity. Using previously collected samples (not from the DRC DHS set), this PCR assay was positive in 33% (10/31) of the subjects with known microscopy-confirmed HAT in peripheral blood samples and none of the uninfected patients (0/17). Thus, this assay had low sensitivity but very high specificity.
All 26 ELISA-positive samples were tested by both trypanolysis and PCR (). Two subjects were positive by trypanolysis with 100% lysis on both LiTat 1.3 and LiTat 1.5 VAT. One of these trypanolysis-positive specimens was also positive by PCR. In addition, one trypanolysis-negative subject was positive by PCR, suggesting a recent infection that had not yet elicited anti-trypanosomal antibodies.
The other 23 ELISA positive samples which were negative by PCR and trypanolysis are likely to be false positives. This false positive rate (23/7766) translates to a very high ELISA test specificity (99.7% with 95% CI: 99.5–99.9). However, given the low prevalence of the disease, the ELISA's positive predictive value is only 11.5% (95% CI: 4.0–28.9).
All 3 trypanolysis and/or PCR-positive subjects were male and were HIV-seronegative. Two were co-infected with P. falciparum 
. These 3 cases were found in two sites, both of which are in known endemic regions ().
The overall prevalence of HAT in the DRC was calculated using standard sampling weights and found to be 29.7 cases/100,000 persons. Assuming a total population of 62.6 million, this leads to an estimated 18,592 people with HAT (95% confidence interval, 4,883–32,302) in the DRC in 2007.
In 2007, the National Trypanosomiasis Control Program reported 8,162 cases of HAT. Our results suggest that 56% of actual HAT cases were not detected and therefore not reported 
. This is very close to estimates of underreporting used by the WHO (65–75%) 
The estimates obtained here are subject to several limitations. First, none of the tests are completely sensitive, so cases of HAT infection could have been missed. Second, HAT is a highly clustered disease, and it is possible that specific small geographic regions with high HAT prevalences were not accurately sampled. Both limitations would have led to understimates of the prevalence of the disease. Nevertheless, our results provide the first nationally representative population-based data on the prevalence of this disease and confirm WHO estimates for under-reporting. This study also confirm that population-based surveys are useful in determining the burden of infectious diseases.