Meta-analysis of agreement/concordance statistics in studies comparing self- vs clinician-collected samples for HPV testing in cervical cancer screening.

ABSTRACT

We conducted a meta-analysis of test agreement/concordance between human papillomavirus (HPV) testing in self-collected vs clinician-collected samples in 26 studies (10 071 participants) updating a previous meta-analysis on accuracy for cervical precancer. Pooled overall agreement was 88.7% (95% CI 86.3%-90.9%), positive agreement was 84.6% (95% CI 79.9%-88.7%), negative agreement was 91.7% (95% CI 89.1%-94.0%) and kappa was 0.72 (95% CI 0.66-0.78). Subgroup meta-analyses suggested higher overall agreement for target amplification-based DNA assays (90.4%) compared to signal amplification-based DNA assays (86.7%; P = .175) or RNA assays (82.3%; P < .001). HPV test agreement/concordance targets may provide criteria to extend existing validations toward alternative sampling approaches and devices/storage media.