OBJECTIVE: To estimate the optimal relative light unit ratio, as a measure of viral load, of the Hybrid Capture II human papillomavirus (HPV) test in self-collected specimens for detecting cervical intraepithelial neoplasia (CIN). METHODS: Two hundred women referred for colposcopy with abnormal cytologic, self-collected vaginal and vulvar swabs and urine for HPV testing. The receiver operating characteristic (ROC) curve method was used to estimate optimal cutoffs for the Hybrid Capture II test. The reference standard was colposcopy, with directed biopsy as required. RESULTS: The estimated optimal cutoffs of the relative light unit ratio for detecting CIN 2 or higher for urine, vulvar, and vaginal samples gave sensitivities of 72.4%, 82.8%, and 89.0% and specificities of 57.0%, 52.1%, and 55.9%, respectively. At the manufacturer's recommended 1.0 cutoff, sensitivities were 44.8%, 62.1%, and 86.2% for urine, vulvar, and vaginal samples, with specificities of 69.7%, 62.7%, and 53.5%, respectively. The likelihood ratios (likelihood of being truly positive after a positive test result) were similar for the optimal and the 1.0 cutoff. CONCLUSIONS: The ROC curve methods did not improve the overall diagnostic accuracy of the Hybrid Capture II test compared with the 1.0 relative light unit ratio cutoff.
OBJECTIVE: To estimate the optimal relative light unit ratio, as a measure of viral load, of the Hybrid Capture II human papillomavirus (HPV) test in self-collected specimens for detecting cervical intraepithelial neoplasia (CIN). METHODS: Two hundred women referred for colposcopy with abnormal cytologic, self-collected vaginal and vulvar swabs and urine for HPV testing. The receiver operating characteristic (ROC) curve method was used to estimate optimal cutoffs for the Hybrid Capture II test. The reference standard was colposcopy, with directed biopsy as required. RESULTS: The estimated optimal cutoffs of the relative light unit ratio for detecting CIN 2 or higher for urine, vulvar, and vaginal samples gave sensitivities of 72.4%, 82.8%, and 89.0% and specificities of 57.0%, 52.1%, and 55.9%, respectively. At the manufacturer's recommended 1.0 cutoff, sensitivities were 44.8%, 62.1%, and 86.2% for urine, vulvar, and vaginal samples, with specificities of 69.7%, 62.7%, and 53.5%, respectively. The likelihood ratios (likelihood of being truly positive after a positive test result) were similar for the optimal and the 1.0 cutoff. CONCLUSIONS: The ROC curve methods did not improve the overall diagnostic accuracy of the Hybrid Capture II test compared with the 1.0 relative light unit ratio cutoff.