A United States study of self-collected urine vs clinician-collected cervical sample for HPV and cervical intraepithelial neoplasia grade two or worse detection: Cross-sectional agreement and diagnostic accuracy study.

Objectives: Urine can provide a non-invasive screening method for high-risk human papillomavirus (hrHPV) infections. We aimed to compare detection rates of hrHPV and cervical intraepithelial neoplasia grade two or worse (CIN2+) using first-void urine (FVU) and clinician-collected cervical samples. We assessed whether the time of day of FVU collection impacts hrHPV detection. Methods: We recruited 188 participants, 30-65 years, before their colposcopy or routine screening from September 2020 through February 2022. Each collected FVU twice on the same day: at home and in the clinic before their appointments. We used prevalence-adjusted bias-adjusted kappa (PABAK) scores to compare hrHPV type-specific agreement, followed by McNemar's Exact test. We assessed both absolute and relative sensitivity and specificity to evaluate the accuracy of the collection techniques in predicting CIN2+ by hrHPV type. Results: 95 hrHPV infections were detected in 70/188 (37.2 %) women. The PABAK agreement for hrHPV16 detection between FVU and the speculum was 0.95 (95 % CI: 0.91, 1.0). The agreement for hrHPV detection was almost perfect between home or clinic urine collection and the speculum specimen. The accuracy of CIN2+ detection measured by the sensitivity ratio of the clinic first-void specimen compared to the speculum specimen for hrHPV16 was 1.00 (0.62, 6.20). Most specificity ratios for all other hrHPV types were anchored at 1.0 with narrow confidence intervals. The time of day for FVU gave similar "any hrHPV" detection rates (27.5 % vs 32 %, Conclusion: FVU hrHPV testing offers a potential cervical cancer screening alternative for women who cannot tolerate other collections.