Effect of Type-Specific Human Papillomavirus Incidence on Screening Performance and Cost

Abstract

Implementation of human papillomavirus (HPV) vaccination is expected to change the epidemiology of HPV infection. Using age-dependent (inhomogeneous) Markov chains, we tested the effect of type-specific (ie, HPV-16/HPV-18 and other high-risk HPV types) HPV incidence on 3 screening strategies in a cohort of 100,000 women, starting screening at the age of 25 years and continuing until the age of 34 years. All the strategies started with an HPV test; if the result was positive, the next step was triage with cytology (strategy 1), cytology and colposcopy together (strategy 2), or colposcopy only (strategy 3). Published background data were used for the models. Strategy 2 had the highest sensitivity; the absolute number of missed cervical intraepithelial neoplasia (CIN)3+ cases was associated with HPV incidence in all the strategies, but their relative sensitivity remained unaffected. Strategy 2 was cheaper per diagnosed CIN3+ for very low HPV incidence rates, but this changed for higher incidence rates. For any given pair of HPV-16/HPV-18 and other high-risk-type incidence, the cost of the triage and the total cost of screening was highest in screening 2 and lowest in screening 1. For each screening strategy, the cost per diagnosed CIN3+ was mainly determined by the incidence of HPV-16/HPV-18, and the cost of the triage and the total screening cost by the incidence of other high-risk types. Type-specific HPV incidence affects the absolute number of missed CIN3+ cases and the cost of screening in a mathematically describable way and can be used for prediction of changes in these outcomes with changing HPV epidemiology.