In view of possible type replacement upon introduction of human papillomavirus (HPV) vaccination, we aimed to explore patterns of type-specific clustering across populations with various background infection risks. A total of 3,874 women from 3 cross-sectional studies in the Netherlands (in 2007-2009) provided vaginal self-samples, which were tested for 25 HPV genotypes by a sensitive molecular assay (SPF10 line probe assay, DDL Diagnostic Laboratory, Voorburg, the Netherlands). The number of concurrent HPV infections per woman was studied by Poisson regression. Associations between HPV types were investigated by generalized estimating equation analyses. The prevalence of any HPV type was 14% in a population-based study, 54% in a chlamydia screening intervention study, and 73% in a study among attendees of sexually transmitted infection clinics. Overall, multiple HPV infections were detected in 26% of the women. The number of concurrent HPV infections conformed to an overdispersed Poisson distribution, even after correction for known risk factors. Types differed significantly in their tendencies to be involved in coinfections, but no evidence for particular type-type interactions was found. Moreover, the strongest associations were observed in the lowest-risk population and vice versa.We found no indications of pairwise interactions, but our findings do suggest that clustering differs among HPV types and varies across risk groups.
Keywords: antagonism; coinfection; human papillomavirus; monitoring; multilevel analysis; synergism; type replacement; vaccination.