A randomised public-health trial on automation-assisted screening for cervical cancer in Finland: performance with 470,000 invitations

Our objective was to evaluate automation-assisted screening, in comparison to the conventional method, in a routine population-based cervical cancer-screening programme. Our study is based on an individually randomised design involving approximately 160,000 invitees and 110,000 attendees every year. From 1999 to 2001, 471,297 women were invited to attend and 330,445 smears were screened (attendance rate 70.1%), of which 220,254 were tested conventionally and 110,191 were tested using the automation-assisted method. Cytologic Papanicolaou group II findings were reported slightly more often (RR = 1.04) in the automation-assisted method than in the conventional screening arm. There were 1,291 cases of histologically confirmed dysplasia or carcinoma (0.4% of the screened), one-third of which were severe dysplasia or a more severe finding (CIN3+). The detection rates of histologically verified findings were similar between the 2 screening arms. In Finland, the screening programme has been effective. As the detection rates, particularly of CIN3+, were similar between the screening arms, we will continue the automation-assisted method in the routine screening programme. Further follow-up for interval cancer incidence is required, however, to measure if the effect of screening is the same between the arms. A similar evaluation design is feasible to any other major or competing modification of the screening test or other element in the programme.