USE OF THE POSITIVE PREDICTIVE VALUE TO CORRECT FOR DISEASE MISCLASSIFICATION IN EPIDEMIOLOGIC STUDIES

Misclassification problems of the disease status often arise in large epidemiologic cohort studies in which the outcome is classified on the basis of record linkage with routinely collected error-prone data sou rces, such as cancer registries or mortality statistics. If the miscla ssification is nondifferential, i.e., independent of the exposure stat us, this leads to bias toward the null in estimates of relative risk. A variety of methods have been proposed to correct for this bias. Most approaches are based on estimates of the sensitivity and specificity of disease classification from validation studies, which typically req uire invasive and time-consuming diagnostic procedures. For ethical an d practical reasons, such procedures may often not be applied on indiv iduals classified as not having the disease, in which case estimates o f sensitivity and specificity cannot be obtained. In this paper, an al ternative correction method is proposed based on estimates of the posi tive predictive value, which requires validation of the diagnosis amon g samples of individuals classified as having the disease only. The me thod is applicable in situations with either differential or nondiffer ential specificity of disease classification as long as the sensitivit y is nondifferential. Point estimates and large-sample interval estima tes of the corrected relative risk are algebraically derived. The perf ormance of the method is assessed by extensive simulations and found t o be satisfactory even for small sample sizes.