Abstract

This study illustrates the bias associated with the use of an estimation approach called the patient infected status algorithm (PISA), which has been recently introduced and is increasingly used to produce sensitivity, and specificity estimates for Chlamydia trachomatis sand Neisseria gonorrhoea screening tests. PISA-based estimates have been published in the medical and microbiological literature and have been included in FDA approved package inserts of nucleic acid amplification tests for detecting Chlamydia trachomatis. In this study, we show that the PISA is an estimation procedure that can produce biased estimates of sensitivity, specificity and prevalence parameters. In a series of simulated scenarios we considered, none of the 95% CIs for PISA-based estimates of sensitivity and prevalence contained the true values. In addition, we show that the PISA-based estimates of sensitivity and specificity change markedly as the true prevalence changes. Thus, like earlier estimates such as discrepant analysis based estimates and unadjusted culture-based estimates of sensitivity and specificity, PISA based estimates are also biased.