Fixed sized samples for type-specific surveillance of human papillomavirus in genital warts

Surveillance data suggest that human papillomavirus (HPV) vaccination in Australia is reducing the incidence of genital warts. However, existing surveillance measures do not assess the proportion of the remaining cases of warts that are caused by HPV types other than 6 or 11, against which the vaccine has no demonstrated effectiveness. Using computer simulation rather than sample size formulae, we established that genotyping at least 60 warts can accurately test whether the proportion of warts due to HPV types not targeted by the vaccine has increased (Type I error probability ≤ 0.05, Type II error probability <0.07). Standard formulae for calculating sample size, in contrast, suggest that a sample size of more than 130 would be required for this task, but using these formulae entails making several strong assumptions. Our methods require fewer assumptions and demonstrate that a smaller sample size than anticipated could be used to address the question of what proportion of post-vaccination cases of warts are due to nonvaccine types. In conjunction with indications of incidence and prevalence provided by existing surveillance measures, this could indicate the number of cases of post-vaccination warts due to nonvaccine types and hence whether type replacement is occurring.