The fallibility of diagnostic tests for sexually transmitted diseases: the impact of behavioral and epidemiologic studies.

OBJECTIVES: Analysis of sexually transmitted disease (STD) data to identify specific behaviors and risk factors often fails to take into account the misclassification of disease by a less than perfect diagnostic test. The authors consider here how a diagnostic test's performance profile can introduce misclassification and thereby bias measures of association.
METHODS: The authors used hypothetical data relating to diagnostic tests for Chlamydia trachomatis infections and oral contraceptive use to determine odds ratio estimates given a range of sensitivity, specificity, prevalence of infection, and sample size.
RESULTS: Lower specificity in a diagnostic test can result in an underestimation of a risk factor's association with an infection. This bias is particularly severe in low prevalence populations. Use of a diagnostic test with low specificity also will increase the sample size needed to demonstrate the association and, thus, the cost of such surveys.
CONCLUSIONS: Diagnostics tests for sexually transmitted diseases have less than perfect sensitivity and specificity, which affects the validity of analyses of factors associated with sexually transmitted diseases. Analyses done using low prevalence populations and/or small sample sizes may underestimate the magnitude of effect in retrospective studies and clinical trials of behavioral interventions aimed at reducing sexually transmitted disease risk.