Impact of cervical screening on cervical cancer mortality: estimation using stage-specific results from a nested case-control study

Background: It is well established that screening can prevent cervical cancer, but the magnitude of the impact of regular screening on cervical cancer mortality is unknown. Methods: Population-based case-control study using prospectively recorded cervical screening data, England 1988-2013. Case women had cervical cancer diagnosed during April 2007-March 2013 aged 25-79 years (N = 11 619). Two cancer-free controls were individually age matched to each case. We used conditional logistic regression to estimate the odds ratio (OR) of developing stage-specific cancer for women regularly screened or irregularly screened compared with women not screened in the preceding 15 years. Mortality was estimated from excess deaths within 5 years of diagnosis using stage-specific 5-year relative survival from England with adjustment for age within stage based on SEER (Surveillance, Epidemiology and End Results, USA) data. Results: In women aged 35-64 years, regular screening is associated with a 67% (95% confidence interval (CI): 62-73%) reduction in stage 1A cancer and a 95% (95% CI: 94-97%) reduction in stage 3 or worse cervical cancer: the estimated OR comparing regular (<= 5.5yearly) screening to no (or minimal) screening are 0.18 (95% CI: 0.16-0.19) for cancer incidence and 0.08 (95% CI: 0.07-0.09) for mortality. It is estimated that in England screening currently prevents 70% (95% CI: 66-73%) of cervical cancer deaths (all ages); however, if everyone attended screening regularly, 83% (95% CI: 82-84%) could be prevented. Conclusions: The association between cervical cancer screening and incidence is stronger in more advanced stage cancers, and screening is more effective at preventing death from cancer than preventing cancer itself.