Background: Errors in the classification of male circumcision status could bias studies linking infection to lack of circumcision.
Goal: To determine the frequency and factors associated with the reproducibility of reporting circumcision status.
Study design: Secondary analysis of data using logistic regression modeling from a multicenter randomized controlled trial was performed.
Results: At follow-up assessment, 15.6% of clinician reports on circumcision status disagreed with baseline reports. Disagreement was more common if both clinicians were women than if both were men (odds ratio [OR], 2.8; 95% CI, 1.9-4.1). As compared with whites reported as circumcised (4%, 19/532 visits), the highest disagreement involved uncircumcised Hispanic (OR, 3.3; 95% CI, 1.7-6.3), white (OR, 12.2; 95% CI, 5.8-25.6), or black (OR, 17.1; 95% CI, 10.4-27.9) men.
Conclusions: This is one study among a small number of studies examining the reproducibility of clinician-reported circumcision status by comparing multiple clinical examinations of the same patient. The magnitude of the misclassification discovered could bias results and indicates the need for greater accuracy in reporting circumcision status in future studies.