Cervical cancer patterns with automation-assisted and conventional cytological screening: a randomized study

The purpose was to evaluate alternative cytological screening methods in population-based screening for cervical cancer up to cancer incidence and mortality outcome. Automation-assisted screening was compared to conventional cytological screening in a randomized design. The study was based on follow-up of 503,391 women invited in the Finnish cervical cancer screening program during 1999-2003. The endpoints were incident cervical cancer, severe intraepithelial neoplasia and deaths from cervical cancer. One third of the women had been randomly allocated to automation-assisted screening and two thirds to conventional cytology. Information on cervical cancer and severe neoplasia were obtained through 1999-2007 from a linkage between screening and cancer registry files. There were altogether 3.2 million woman-years at risk, and the average follow-up time was 6.3 years. There was no difference in the risk of cervical cancer between the automation-assisted and conventional screening methods; the relative risk (RR) of cervical cancer between the study and control arm was 1.00 (95% confidence interval [Cl] = 0.76-1.29) among all invited and 1.08(95% Cl = 0.76-1.51) among women who were test negative at entry. Comparing women who were test negative with nonscreened, RR of cervical cancer incidence was 0.26, 95% Cl = 0.19-0.36 and of mortality 0.24 (0.13-0.43). Both methods were valid for screening. Because cervical cancer is rare in our country, we cannot rule out small differences between methods. Evidence on alternative methods for cervical cancer screening is increasing and it is thus feasible to evaluate new methods in large-scale population-based screening programs up to cancer outcome.