Modeling the balance of benefits and harms of cervical cancer screening with cytology and human papillomavirus testing.

Background: Benefits of screening should outweigh its potential harms. We compared various metrics to assess the balance of benefits and harms of cervical cancer screening. Methods: We used a cervical cancer natural history Markov model calibrated to the Canadian context to simulate 100,000 unvaccinated women over a lifetime of screening with either cytology every 3 years or human papillomavirus (HPV) testing every 5 years. We estimated the balance of benefits and harms attributable to screening using various metrics, including colposcopies/life-year gained, and net lifetime quality-adjusted life-years (QALY) gained, a measure integrating women's health preferences. We present the average (minimum-maximum) model predictions. Results: Cytology-based screening led to 1,319,854 screening tests, 30,395 colposcopies, 13,504 life-years gained over a lifetime, 98 screening tests/life-year gained, 2.3 (1.6-3.3) colposcopies/life-year gained, and a net lifetime gain of 10,735 QALY (5,040-17,797). HPV-based screening with cytology triage in the same population would lead to 698,250 screening tests, 73,296 colposcopies, 15,066 life-years gained over a lifetime, 46 screening tests/life-year gained, 4.9 colposcopies/life-year gained (2.9-11.1), and a net lifetime gain of 11,690 QALY (4,409-18,742). HPV-based screening was predicted to prevent more cancers, but also incur more screening harms than cytology-based screening. Conclusions: Metrics using colposcopies as the main harm outcome favored cytology-based screening, whereas metrics based on screening tests and health preferences tended to favor HPV-based screening strategies. Impact: Whether HPV-based screening will improve the balance between benefits and harms of cervical cancer screening depends on how the balance between benefits and harms is assessed.