Impact of variations in triage cytology interpretation on human papillomavirus–based cervical screening and implications for screening algorithms

Abstract Background Women positive to human papillomavirus (HPV+) testing at cervical screening need triage, typically cytology and immediate colposcopy in case of atypical squamous cells of undetermined significance (ASCUS) or worse (ASCUS+) or, in cytology-normal HPV+ women, HPV test repeat after 1 year and colposcopy referral if still HPV+. Our hypothesis was that substantial variations in triage positivity and sensitivity may produce little variation in overall referral to colposcopy and on sensitivity of the entire screening process. Methods Centre- and age-aggregated data from 72,869 women aged 35–64 years were derived from 10 organised screening programmes which have piloted HPV screening in Italy since 2012. Overall colposcopy referral was evaluated as a function of immediate colposcopy referral and overall CIN2+ detection as a function of the proportion of all CIN2+ detected by immediate referral (a proxy of cytology's sensitivity). We fitted additive regression models, adjusted for centre, age, compliance to HPV retesting and to colposcopy, by generalised estimation equations. Results The proportion of HPV+ women directly referred to colposcopy varied across programmes (20–57%; average 37%) and so did CIN2+ detection (49–94%; average 77%). Overall, 63% (range 41–75%) of HPV+ were referred to colposcopy either immediately or at HPV repeat. An absolute 10% increase in immediate colposcopy referral resulted in 4.2% (95% CI: 3.3–5.1%) increase in overall referral. An absolute 10% increase in cytology's sensitivity resulted in a 1.1% (95% CI: 0.1–2.0%) increase in overall CIN2+ detection. Conclusions Repeat HPV testing limits the effect of subjectivity of cytology interpretation on overall referral and sensitivity. These will change only slightly when replacing cytology with another test if the interval to HPV repeat remains unchanged.