Statistical Methods to Account for Data Errors Discovered from an Audit

• Activity Number: 380
• Type: Contributed
• Date/Time: Tuesday, August 4, 2009 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Bryan E. Shepherd*+ and Chang Yu
• Companies: Vanderbilt University and Vanderbilt University School of Medicine
• Keywords: measurement error ; HIV ; global health ; mixture distribution ; estimating equations ; data audits

Abstract

A data coordinating team performed on-site audits and discovered discrepancies between the data sent to the coordinating center and that recorded at sites. We present statistical methods for incorporating audit results into analyses. This can be thought of as a measurement error problem, where the distribution of errors is a mixture with a point mass at 0. We show that even if the mean of the discrepancy between the reported and correct values of a predictor is 0, if the error rate is non-zero, estimates of the association between two continuous variables will be biased. We show how to incorporate the error rate and magnitude, estimated from a random subset (the audited data), to compute unbiased estimates of association and proper confidence intervals. We illustrate our methods using data from 5152 HIV-infected patients in Latin America, of whom 184 had their data audited.