The present study analyzed changes in the incidence of AGWs following the HPV vaccine recommendation in Germany and particularly the effects of herd protection among males in a low vaccine coverage setting. We found a u-shaped decrease among the 14- to 24-year-olds with a maximum reduction up to 60% observed for the 16- to 20-year-old females and a similar, but slightly less pronounced reduction for 16- and 18-year-old males.
In our analysis, the decrease of incidence occurred early after the vaccine recommendation in 2007 and stabilized after reaching a lower level in the first quarter of 2009. This reflects the finding that those who are going to be vaccinated do so in a relatively short time window. The decrease in incidence in this respective group reflects the increasing proportion of those getting vaccinated. This appears also plausible, e.g., for females aged 16, 17, or 18 years, as they had only one to 2 years in the recommended age range of vaccination and are the age groups with the most pronounced reduction in our analysis.
The u-shaped incidence reduction within the age groups of 14 to 24 years can be explained by the fact that HPV vaccination coverage increases with age within the recommended age interval from 12 to 17 years . In addition, the study population originated from an SHI which reimburses the HPV vaccination in females up to the age of 26 years, explaining the reduction of incidence up to that age. Still, the uptake of vaccination closer to the upper end of this age interval is lower explaining why the effect diminishes. In the first year after recommendation, HPV vaccine uptake was about 30% for 12- to 17-year-olds, 12% for 18- to 26-year-olds, and only about 2% in 26-year-olds) . One reason for low uptake or coverage rates in the older age groups (18 to 26) might be that neither the vaccination was officially recommended for these age groups nor was it stated in the vaccination schedule. Therefore, women might not have been aware that the reimbursement is possible. The maximum reduction of 60% is broadly consistent with the observed vaccination uptake based on administrative data [20, 21] or with the reported vaccination coverage based on data from several surveys [26,27,28], or from private insurance companies  in Germany. The most prominent decrease of incidence among females 16 to 20 years of age corresponds to the highest uptake in 14- to 17-year-old females of about 37% in 2008 . Also the reduction in younger age groups (13 to 15 years of age) of 20 - 30% as well as in older age groups (18- to 26-year-olds) corresponds well with the reported HPV vaccination uptake in 2008 .
Despite the fact that the recommendation for HPV vaccination only includes females in Germany, we observed parallel effects among males of nearly the same age. Here, the most pronounced decrease was seen in 16- to 18-year-olds which is plausible as females have, on average, similarly aged or one to 2 years older sexual partners during adolescence. Other studies reported decreasing incidences of AGWs in males of the same age as vaccinated females, in a high vaccine coverage setting of about 80% or higher [10, 11, 14, 15, 29]. But also some studies with low age-specific vaccine coverage of about 50% or lower reported a decrease in AGW incidence or prevalence  for males younger than 20  or 25  years of age. This is in line with the incidence reduction of AGWs in our study for 16- to- 20-year-old females and less pronounced for 16- and 18-year-old males. A prior review including some of the above mentioned studies concluded that herd immunity effects were demonstrated only in populations with vaccine coverage of 50% or higher . However, there might be differences with respect to homogeneity of vaccination coverage, for example in countries with an immunization program against HPV a percentage of girls of a certain age is vaccinated. In contrast, in Germany vaccination in general is recommended but voluntary and HPV vaccination can occur at any age between 12 to 17 years. In consequence, coverage is low for the full age-group of 12- to 17-year-olds, but in females aged 18 years vaccine coverage approaches 50%. Such heterogeneity might also explain why effects in males can be observed despite overall low coverage. While the relative changes were similar in both sexes, changes in the absolute incidence among males between pre-vaccination and post-vaccination period were less prominent. However, this might rather be a diagnostic bias than a true epidemiological difference, resulting from the fact, that females usually consult gynaecologists routinely about once a year, while males are not enrolled in a comparable system and are diagnosed only by actively consulting a physician or via an incidental finding. Thus, a higher fraction of the true incidence might be reported for females than for males, resulting in a higher incidence. In some of the age groups among males the incidence was so low, that relative reduction did not achieve significance.
The only slightly lower herd immunity effects in males than the direct protection effects in females of the same age suggest strongly assortative mating by age. The relatively short time period since the recommendation of vaccination most likely covers a single sexual relationship, which reportedly lasts for a median time of 26 months for males (interquartile range (IQR), 8.5–42) and 32 months (IQR, 9–49) for females in the relevant age groups . HPV infectivity is high and in heterosexual serially monogamous partnerships, the protection of males by having a vaccinated female partner will decrease with change of partners, particularly as the reported median gap between different partners of 14 to 24 days is shorter than an HPV infection period . Consequently, studies covering longer study periods particularly in low vaccination coverage settings would possibly observe diminishing protection among males while direct effects among females would remain constant. On the other hand, in high coverage settings, overall herd immunity effects may reduce the cumulative effects of change of female partners among males. In accordance with findings of other studies [2, 8, 9, 11,12,13,14,15,16], the incidence of AGWs among older age groups did not display a decreasing trend. Potential herd immunity effects in older age groups are based on sexual mixing with partners from vaccinated age groups. In a low coverage setting, the influence of such indirect protection might be too small to be detected.
In our study, the incidence of AGWs started to decrease among the 16- to 26-year-olds approximately 3 months after the vaccine recommendation by STIKO in Germany in March 2007. The short delay with which the effects of vaccination on AGW incidence were observed is consistent with the biology of infection. As HPV immunity is reported already after one vaccine dose  and AGWs develop after a medium incubation time of about 3 months [6,7,8], the corresponding time lag of decreasing incidence of one to two quarters of a year after the vaccine recommendation for females is plausible. A decreasing incidence in the target vaccination group soon after vaccine introduction has been also reported from several other ecologic studies conducted in Australia, Europe, and the US [2, 8, 9, 11,12,13,14,15,16].
We could not find herd immunity effects in older age groups (> 24 years of age) which could be explained by a limited sample size and less exclusive sexual mixing patterns of older age groups with much younger partners.
Limitations of our study are mainly related to the underlying administrative data used for analyses. For example, it was not possible to validate the diagnoses of AGWs, because we had to rely on what was used for reimbursement. Also, not all relevant variables were available in the desired detail. In particular, exact diagnosis dates in the outpatient sector are not available in the database, making the estimation of incidence less precise. It was not possible to determine the vaccine coverage of non-vaccinated persons per age cohort within the database as the vaccination status before the study period would have to be known, but was not available in our data. However, HPV vaccine uptake has been estimated in 2008  and recently confirmed in a newer study in 2014 . Both sources point towards a coverage of about 40–50%, or even higher if also single vaccine dose is considered. Due to data protection reasons, only the birth year but not the exact date of birth is contained in the database, which might lead to some non-differential misclassification of persons by age as some persons may already be 1 year older according to the definition of age. For the analysis of trends over time, the age of one patient is kept the same throughout all four quarters of the year, which leads to an average aging cohort from the 1st to the 4th quarter of a year resulting in higher incidences in the 4th quarter than in the 1st quarter of each year most prominently for 16- to 20-year-olds. Another limitation is that only patients who see a physician for AGWs could be identified in the database. AGWs do not necessarily cause discomfort or pain, and some patients might be embarrassed and not consult a physician. This would result in a possible underestimation of incidences of AGWs. Furthermore, AGWs may be coded as an unspecific disease, for example, as viral warts or as a sexually transmitted infection not otherwise specified, which would also result in an underestimation of AGW incidence. Finally, because of practical reasons related to data availability and permissions to use, we were unfortunately limited to analyze only data up to 2010. Particularly assessing longer trends would be of interest and should be conducted in the future.
Strengths of the study are the large population size which was not restricted to specific regions or settings in Germany and the almost complete lack of selection effects among participants, as a routine care situation is reflected in the data.