Estimating effectiveness of HPV vaccination against HPV infection from post-vaccination data in the absence of baseline data

•Effectiveness of vaccination needs to be estimated even if baseline data is missing.•Baseline HPV prevalences can be reconstructed using non-vaccine types as reference.•Effectiveness can be estimated from the reconstructed baseline.•The estimation methods were validated using real-life data from Sweden. HPV vaccination programs have been introduced in large parts of the world, but monitoring of effectiveness is not routinely performed. Many countries introduced vaccination programs without establishing the baseline of HPV prevalences. We developed and validated methods to estimate protective effectiveness (PE) of vaccination from the post-vaccination data alone using references, which are invariant under HPV vaccination. Type-specific HPV prevalence data for 15–39 year-old women were collected from the pre- and post-vaccination era in a region in southern Sweden. In a region in middle Sweden, where no baseline data had been collected, only post-vaccination data was collected. The age-specific baseline prevalence of vaccine HPV types (vtHPV, HPV 6, 11, 16, 18) were reconstructed as Beta distributions from post-vaccination data by applying the reference odds ratios between the target HPV type and non-vaccine-type HPV (nvtHPV) prevalences. Older non-vaccinated age cohorts and the southern Sweden region were used as the references. The methods for baseline reconstructions were validated by computing the Bhattacharyya coefficient (BC), a measure for divergence, between reconstructed and actual observed prevalences for vaccine HPV types in Southern Sweden, and in addition, for non-vaccine types in both regions. The PE estimates among 18–21 year-old women were validated by comparing the PE estimates that were based on the reconstructed baseline prevalences against the PE estimates based on the actual baseline prevalences. In Southern Sweden the PEs against vtHPV were 52.2% (95% CI: 44.9–58.5) using the reconstructed baseline and 49.6% (43.2–55.5) using the actual baseline, with high BC 82.7% between the reconstructed and actual baseline. In the middle Sweden region where baseline data was missing, the PE was estimated at 40.5% (31.6–48.5). Protective effectiveness of HPV vaccination can be estimated from post-vaccination data alone via reconstructing the baseline using non-vaccine HPV type data.