Background: This study aims at obtaining unbiased estimates of the sensitivity and specificity of existing screening tests for Trypanosoma cruzi and at simulating the effectiveness of alternative screening strategies at different prevalence rates.
Study design and methods: A systematic random sample of 400 was taken from 1200 banked serum samples of donors screened between August 1998 and January 1999 in Santa Cruz, Bolivia. Samples were tested with indirect hemagglutination test (IHA), indirect immunofluorescence assay (IFA), and four enzyme-linked immunosorbent assays (ELISAs). Sensitivity and specificity of tests were estimated through latent class analysis.
Results: The sensitivity of individual tests ranged from 96.5 to 100 percent, and their specificity from 87.0 to 98.9 percent. Combinations of two tests used in parallel would, even at 40 percent prevalence, only miss approximately 1 infected unit per 10,000 screened. At 5 percent prevalence, however, they would yield 75 to 120 false-positive units per 1000 units screened. Parallel testing with IHA plus ELISA or with IHA plus IFA is marginally more cost-effective, compared to single IHA testing, than single ELISA or single IFA testing, regardless of the T. cruzi prevalence.
Conclusions: Routine blood donor screening for T. cruzi with a single test results in unacceptable numbers of false-negative samples in highly endemic areas or in at risk population groups. Adding a second test seems mandatory, but which one to choose depends on local cost components and feasibility.