The aim of this study was to describe the burden of potentially HPV-related anogenital diseases in 23-25 old women after the introduction of HPV vaccination based on German administrative statutory health insurance claims data for the years 2012 to 2017. The three birth cohorts 1990 to 1992 were included, as they were fully eligible to receive HPV vaccination according to the first STIKO recommendation for HPV in 2007 and further birth cohort 1989, which was partially eligible (girls of birth cohort 1989, who turned 18 years old before March 23, 2007 exceeded the recommended age for vaccination in 2007). We found the highest administrative prevalence for anogenital diseases grade I, followed by genital warts and anogenital diseases grade III among all analyzed birth cohorts. A lower burden of anogenital disease potentially related to HPV was observed in the younger birth cohorts as compared to the older cohort. This was observed especially for genital warts and anogenital disease grade III.
The following section discusses the results in the context of HPV vaccination coverage in Germany. HPV vaccination status of the study population could not be evaluated, as the included birth cohort should have received their HPV vaccination in the years 2007 to 2010, which were not available for this analysis. For the analyzed birth cohorts two publications provide estimates on the HPV vaccination coverage. Hense et al. evaluated initiation of HPV vaccination in 2008 based on data from a statutory health insurance and reported 37% of 16 years old girls with at least one dose, which would overlap with birth cohort 1992. In the same publication rates were 33%, 19% and 18% for 17-, 18-, and 19-year-old women, respectively (corresponding to birth cohorts 1992-1989) . Delere et al. reported HPV vaccination coverage based on self-reported history of approximately 2000 women aged between 18-20 years old in 2010 (cohorts 1990-1992) with 48.5% for three doses and 60.2% for at least one dose . While these figures are to be interpreted with caution as they most likely do not accurately represent the vaccination coverage of our study population in the analyzed time period, it might give three important estimates: 1) coverage rates for HPV vaccination in Germany were generally lower than in countries with vaccination programs in schools . In Australia e.g., the three-dose vaccination coverage for girls turning 15 years of age was 79% in 2015 . Hence, disease burden in birth cohorts eligible for HPV vaccination may still be higher than one might expect with a better vaccination coverage. 2) vaccination coverage increases from birth cohorts that were not entirely eligible for HPV vaccination according to STIKO to birth cohorts that were eligible. Since we found a downward trend in the 3-year APR for selected diagnoses in favor of the younger birth cohorts, it may be speculated that this is due to increasing HPV vaccine coverage in younger cohorts. We found the downward trend in the 3-year APR for genital warts and grade III dysplasia, but not for grade I and II dysplasia. While this might be due to the rather low HPV vaccination coverage in Germany, it might be also correlated with the HPV types that are covered by the vaccines. HPV types causing the majority of genital warts (approximately 90% caused by HPV 6/11) and a number of grade III cervical diseases (approximately 45% caused by HPV 16/18) were covered by HPV vaccines available during the observation period. Grade I disease, however, is caused to a lesser extent by types 6, 11, 16 and 18 . It may be speculated that a vaccine effect might be masked in an analysis without selection for vaccine-HPV type-associated disease. And 3), also for birth cohort 1989, which was not entirely eligible for HPV vaccination according to STIKO recommendation, a coverage of approximately 18% was reported in 2008 . This birth cohort turned 18 during the year 2007 and thus girls may have had the chance to receive the vaccination regularly before their birthday. Also, individual health insurances provided an extended reimbursement for HPV vaccination up to the age of 26, or the vaccine may have been purchased as self-payer. Hence, in none of the analyzed birth cohorts our prevalence rates for HPV diseases do reflect the anogenital disease burden in a population without HPV immunization. Therefore, no assumptions on burden of potentially HPV-related anogenital diseases in unvaccinated women between 23-25 years of age can be made.
This next section discusses the results in the context of previous studies on burden of HPV diseases in Germany. One previous German study examined HPV type prevalence  but has not investigated associated anogenital diseases. Further studies assessing the incidence of anogenital warts before and after the introduction of the HPV vaccination in German statutory health insurance members aged 10-79 years old reported the highest incidence of anogenital warts for the age group 20-24 years [4, 6, 26]. In 2010, incidence was 493 per 100.000 person years in 20-24 year old and 149 per 100.000 person years in 15-19 year old females . In a population-based surveillance study conducted from 2009-2010 in Wolfsburg/Germany to investigate the burden of HPV infections and associated anogenital diseases in women who were born in 1988 or 1989 (and 1983/84), an incidence of 0.72% and a total life risk of 1.4% for genital warts was estimated for the cohort 1988/89 . For the same cohort a prevalence of 0.83% for CIN II and 0.33% for CIN III diseases was detected . The 3-year APR for anogenital warts of 1.30% and for grade III cervical dysplasia of 0.60% for birth cohort 1989 found in our study is generally in line with this dimension. In Australia an observational study based on clinical diagnoses reported a prevalence of genital warts of 18.4% in young women before the introduction of HPV vaccination . On one hand, the remarkably lower burden in our study might be explained by the fact that our analyzes did not include a population in the pre-HPV-vaccine era, and on the other hand the use of administrative claims data in our study may have resulted in an underestimation of diagnoses (see further discussion on limitations below). It is also important to note that our analysis only provides a snapshot of the APR of anogenital diseases in women of a selected, young age (23-25) and are not transferable to the prevalence rates of HPV-related anogenital diseases over the complete lifespan.
The following section discusses the results more specifically in the context of previous studies on HPV vaccination impact in Germany. Single previous reports focused on HPV vaccine impact in Germany and are in line with our results. Deleré et al. found a significant lower HPV 16/18 prevalence in vaccinated women suggesting first effects of the vaccination. Data from the German Pharmacoepidemiological Research Database (GePaRD) also demonstrated a decline of anogenital warts among males and females at the age of 14-24 comparing the timeframe prior to vaccination with the timeframe after vaccination. While the largest decrease (by 60%) was observed for women in the age group 16-20, the incidence ratio in women aged between 21-26 years was reduced by 10-20% . In our study, genital warts have been reduced by 28% (from 1.30% in birth cohort 1989 to 0.94% in birth cohort 1992), and HPV related anogenital diseases grade III have been reduced by 35% (from 1.09% in birth cohort 1989 to 0.71 in birth cohort 1992).
This last section discusses further potential limitations of the study in detail, most inherent with the use of health insurance claims data. Claims data are primarily collected for reimbursement purposes. Therefore, only patients who seek physician treatment and cause reimbursement for the health insurance could be identified in the database. Patients without symptoms might not seek medical advice, individuals who do not participate in screening examinations might not be identified, and patients who treat their disease (e.g. genital warts) by themselves or ignore the condition are not recorded in the database. Therefore, this study only presents the administrative prevalence based on reimbursement data and not the clinical prevalence of potentially HPV-related anogenital diseases on a population level. Furthermore, Germany had an opportunistic cervical cancer screening beginning at the age of 20 and thus, most documented diagnoses of intraepithelial neoplasia most likely have been identified during screening and further work-up (cytologically or histologically). The annual and biannual cervical cancer screening participation rate in women 25-29 years old was reported with approximately 60% and 70% in 2011, respectively . As not all women attend the screening program for cervical cancer it is likely that some of the diseases might not have been diagnosed and recorded by a physician. Therefore, our results might underestimate the true clinical prevalence of the assessed anogenital diseases in Germany for women between 23-25 years.
For the identification of potentially HPV related anogenital diseases we used ICD-10-GM codes, which is the official classification for the encoding of diagnoses in inpatient and outpatient medical care in Germany since 2000 . Clinicians in the outpatient setting are required to add one of the following specifications to the ICD-10-GM codes: “suspected diagnosis”, “diagnosis ruled out”, “status post”, or “verified diagnosis”. For instance, “suspected” may be coded, if the physician is not certain about the presence of the coded disease and a confirming laboratory analysis is still pending. To ensure the accuracy of diagnoses only “verified” diagnoses in the outpatient and primary and secondary diagnoses in the inpatient sector were used. With this approach however, we excluded women who might be suspected to have one of the investigated anogenital diseases or women who have been cured from of the anogenital diseases (e.g. genital warts) and see their physician for a follow-up visit. This could have led to an underestimation of the APR. Additionally, we decided to only consider very specific ICD-10-GM codes potentially associated with HPV-related anogenital diseases, but physicians may use less specific codes. This might also have contributed to the underestimation of the true rate of clinical diagnoses.
Cervical cancer screening in Germany is Pap-based for women in their twenties. Documented diagnoses in our analysis are most likely cytologically or histologically defined. No laboratory data was available for this study to demonstrate an HPV association of anogenital diseases. Therefore, it is not possible to distinguish between anogenital diseases, which were caused by an HPV infection and those not caused by an HPV infection. However, it is expected that high-risk HPV infections cause almost all cervical cancers and precancers, approximately 90% of high-grade anal, vulvar and vaginal intraepithelial neoplasia and 30%, 70% and 90% of vulvar, vaginal and anal cancers, respectively  and therefore, a very high rate of HPV association is expected for our captured diagnoses.
Finally, the instrument of ARP is subject to limitations that might influence the results of our study. Data on outcomes might be collected in different ways over time for the different study cohorts. Migration of populations affecting the study cohort might influence the differences between the groups. Seasonal or cyclical variations might result in fluctuations that affect the outcome trend. These limitations are assumed to be negligible in our study as we expect no major changes in the recording behavior of physicians during the study period and the impact of migration will not have affected the SHI, as medical services are paid by other payers . Seasonal or cyclical variations in HPV types have not been reported in the literature so far.