To illustrate how GEE models can be used to model multiple types of HPV measured across multiple visits, let Yijk represent the infection of HPV type k for the ith person at jth semiannual visit. To assess an overall effect estimate across oncogenic HPV types, a type-specific baseline prevalence is assumed for each type of infection but a common odds ratio is assumed across the types, the infection rate of Yijk,Pijk , is modeled by (1.3)logitPijk=β0+θk+βZij+γWij.