Offering self-sampling for human papillomavirus testing to non-attendees of the cervical screening programme: Characteristics of the responders

Background: Self-sampling for high-risk human papillomavirus (hrHPV) testing is accepted by up to 30% of non-attendees to the regular cervical screening programme. Here, the yield of cervical intraepithelial neoplasia (CIN)2 or worse (≥CIN2) and CIN3 or worse (≥CIN3) of 15, 274 HPV self-sampling responders amongst non-attendees were compared to that of 176, 027 women participating in regular screening in the same period and in the same region. We also analysed which subpopulations amongst non-attendees are targeted by HPV self-sampling, and which characteristics relate to hrHPV prevalence and yield of ≥CIN2/≥CIN3. Method: Data from two consecutive self-sampling studies were pooled. ≥CIN2/≥CIN3 yields, screening history, age and ethnic status were retrieved from centralised pathology and screening databases, respectively. A logistic regression model was fitted to analyse method of invitation, ethnicity, age group, and screening history as predictors for response rate, hrHPV presence and ≥CIN2/≥CIN3 in non-attendees. For screening history analyses, women <34 years were excluded since it was the first screening round in their life. Findings: ≥CIN2/≥CIN3 yields of HPV self-sampling responders were higher than those of screening participants (≥CIN2: relative risk (RR) = 1.6, 95% confidence interval = 1.4-1.9; ≥CIN3: RR = 1.8, 95% CI = 1.5-2.1 with relative risk values increasing with age (test of homogeneity: ≥CIN2: p = 0.04; ≥CIN3: p = 0.03). Native Dutch non-attendees responded better than immigrants (32% versus 22%, p < 0·001) and those screened in the previous round revealed a higher response than underscreened (i.e. previous smear taken >7 years ago) or never screened (34% versus 25%, p < 0·001) women. Strikingly, amongst under- and never screened women aged ≥39 years, never screened women responded better (25% versus 23%, p < 0·001). ≥CIN2 rates were higher amongst responding native Dutch women than immigrants (p < 0·01), and higher in under-/never screened women than in women screened in the previous round (p < 0·01). Interpretation: Offering hrHPV self-sampling increases the efficacy of the screening programme by targeting a substantial portion of non-attendees of all ethnic groups who have not regularly been screened and are at highest risk of ≥CIN2.