The choice of statistical method(s) must address several aspects of HPV natural history that are not often encountered elsewhere in infectious disease and cancer research. For example, while co-infection by more than one HPV type is common(3,6,7,8,9), prior studies suggest that they infect separate cells, causing separate foci of infection(8). Therefore, the prevalence, incidence, clearance/persistence, progression, of each HPV type-specific infection can be examined as distinct events/outcomes. On the other hand, because of shared risk factors, these separate HPV infections may be correlated. The probability of detecting any given HPV type, for example, is greater among individuals who are currently positive for at least one other HPV type(3,6,7,8,9). Another type of correlation in HPV natural history data relates to repeated (serial) testing of the same women over time. If these two kinds of correlations are ignored, it can affect p-values and confidence intervals, leading to an incorrect statistical inference. On the other hand, statistical models that do not use all the available data (e.g., across all the visits and HPV types examined) may be inefficient.
