Estimating disease rates from a diagnostic test.

Incidence and remission rates are often estimated from studies that employ a diagnostic test to indicate the presence of disease. The apparent rates that result from a simple study design with one test given at two time points will be substantially different from the true rates, even if the error rates of the test are low. Estimation of the true rates requires extra assumptions or extended designs with more tests or more time points. The authors illustrate their points with the use of two examples, the second of which compares the estimates obtained to actual data from a study of onchocerciasis (river blindness) in Guatemala.