Longitudinal performance of self-collected human papilloma virus testing and other mainstream strategies in detecting prevalent and incident cervical precancer: a 15-year cohort study in china

Comprehensive long-term evaluation for self-collected human papilloma virus (HPV) testing is not well established. We aimed to assess its predictive capacity from a long-term perspective compared with other cervical cancer mainstream screening methods. 1997 women aged 35–45 years from Shanxi, China were screened by self-collected HPV testing, doctor-collected HPV testing, liquid based cytology, and visual inspection with acetic acid in 1999. The women were followed up in 2005, 2010, and 2014 with doctor-collected HPV testing, liquid based cytology, and visual inspection with acetic acid (except in 2014). On the basis of baseline screening results, we calculated cross-sectional and prospective clinical performance and 15-year cumulative incidence rates (CIRs) of cervical intraepithelial neoplasia grade 2 or higher (CIN2+). Incident sensitivities for CIN2+ of all methods decreased with time, whereas incident specificity remained stable. Self-collected HPV testing predicted 84·2%, 53·2%, and 43·4% CIN2+ incident cases in 2005, 2010, and 2014, respectively, showing no significant difference with doctor-collected HPV testing and cytology at cutoff atypical squamous cells of undetermined significance (ASC-US), but higher than cytology (p=0·008, 0·004 and 0·008) and visual inspection with acetic acid (p=0·004, 0·003, 0·029) at cutoff low-grade squamous intraepithelial lesions at follow-up in 2005, 2010, and 2014, respectively. The difference of incident sensitivity between self-collected HPV testing and cotest was comparable until at 15-year follow-up (43·4% vs 56·6%, p=0·006). Women with baseline-positive self-collected HPV testing showed comparable CIN2+ CIRs with those women who were baseline-positive by doctor-collected HPV testing during 15-year follow-up, but higher than cytology at cutoff ASC-US (p=0·004), cotest (p