The results from this analysis show that the current HPV vaccination programme of females ages 12 to 17 in Germany is cost-effective with an ICER of 5,525€/QALY. Excluding the vaccine effect on HPV6/11 diseases was considered, the ICER increased to 10,293€/QALY. While this remains cost-effective using the NICE threshold, it emphasises the added value of protection against HPV6/11 diseases. Increasing vaccine coverage in the recommended population would lead to larger and earlier reductions with only a small impact on the ICER (5,807€/QALY).
At steady state, the model predicted that vaccinating girls aged 12 to 17 could reduce the number of HPV 6/11/16/18-related cervical cancers by 65% and genital warts among women and men by 70% and 48% respectively. The impact on HPV-related disease incidence and costs avoided was found to occur relatively soon after initiating the vaccine programme, with much of the early impact being due to the prevention of HPV6/11-related genital warts. There is consistent and growing evidence from several studies that the effects predicted by this model are being observed in the real world. In Australia, in 2009, 65% of Australian females eligible for free vaccination (i.e. female Australian residents aged between 12 and 26 years old) had been vaccinated
. A study comparing two 12-month periods in 2007–08 and 2010–11 showed a dramatic reduction in the incidence of genital warts from 18.6% to 1.9% and 22.9% to 2.9% in under 21-year-old women and heterosexual men, respectively, attending a sexual health centre in Melbourne 4 years following the start of vaccination implementation
. The same trend was reported in another study of patients attending eight sexual health services in different states of Australia between January 2004 and December 2009
. Since no other relevant health intervention was implemented during this time and the incidence of other sexually-transmitted diseases was unchanged the observed effect is most likely to be attributable to the broad and high vaccine coverage in young women. This is reinforced by similar results reported in New Zealand
 and California.
Our model also predicts a reduction in cervical disease outcomes, which has been reported in two studies, so far. In a case–control study in New York, girls aged 11 to 21 who had received at least one HPV vaccination prior to the first Pap smear were reported to have a statistically significant lower risk of having an abnormal Pap smear result (OR = 0.254; 95%CI: 0.093-0.698; p = 0.008)
. Results from an ecological study suggest an early effect from the HPV vaccination programme on cervical abnormalities in Victoria, Australia
. They reported a decrease in the incidence of high-grade cervical abnormalities of 0.38% (95%CI: 0.61-0.16) in girls under 18 years old less than three years after the introduction of the quadrivalent HPV vaccination programme. This decline was not observed for low-grade cervical abnormalities or in older age groups
Our results are consistent with those predicted using a static model for Germany which found an ICER of 10,530€/QALY for preventing cervical cancer and genital warts
, although the two analyses did not compare the same vaccination cohort and the structures of the models are different.
The limitations of this model have been described in detail elsewhere but here we will briefly discuss some potential limitations specific to this analysis
. Firstly, as no German-specific data on sexual behaviour were available, we used data from the UK. The impact this may have on the predictions is unknown
. Secondly, we used a simple calibration approach, similarly to what was done in the Norwegian and UK adaptation, and we calibrated the model with global incidence but not with age-specific data
[19, 20]. It is clear that the calibration process is one of the critical steps of the model adaptation. The model was initially developed for the US and followed an extensive calibration approach, where age-specific data were considered. As we did not modify any natural history parameters, but only a few parameters such as the proportion of women who do not screen regularly, we probably did not changed the peak of the incidence curves of cervical cancer incidence and mortality, and we think that focusing on global incidence rates was an acceptable approach. Thirdly, we considered only direct costs; in future analyses, indirect costs, such as productivity loss, should be included so that the results would be closer to reality. Another limitation concerns the estimates for the health utilities, which were taken from a study performed in the US. However, as far as we are aware, there are no such estimates available for Germany or for any other European country. We also assumed 100% vaccine adherence with each of the three vaccine doses without making allowance for a drop-out rate after one or two doses which may not reflect the reality, but there is no efficacy data available for less than three doses. Furthermore, we did not consider any potential adverse effects of the vaccine. Although vaccination with the quadrivalent vaccine is generally well tolerated so we would expect that including them into the model would have only a minor impact on the ICER.
Additionally, we did not include potential vaccination impact on non-vaccine types in our model, as the model did not include HPV types non-included in the vaccines. There is uncertainty about the accuracy of the observed cross-protection as it is not always easy to determine the causal HPV type in a lesion
. It is also unknown whether cross-protective efficacy will be long lasting
, recent data suggest that cross-protective effects are short lived and therefore of limited value. The results from a UK modelling analysis comparing the cost-effectiveness of the quadrivalent and bivalent vaccines, showed that additional cross-protection only had a minor contribution in terms of economic benefits (i.e. QALYs gained and costs prevented), especially if compared with benefits for genital warts
. Therefore, we could assume that inclusion of vaccine efficacy against non-vaccine types would only have a limited impact on our results.
Another limitation is that we did not perform probabilistic sensitivity analyses. Nevertheless, results from the sensitivity analysis confirm the findings of many other models and showed that results were sensitive to assumptions about the duration of protection and discount rates.
Lastly, only cervical diseases and genital warts were included in the present model. However, HPV has been found in significant numbers of vulvar, vaginal and anal cancers and the quadrivalent vaccine has demonstrated high efficacy in preventing these lesion
[6, 17, 42]. HPV is also found in a proportion of head and neck squamous cell cancers
. Recently two reviews have concluded that the available data are consistent with the hypothesis that HPV does have a causal role in head and neck cancer
[43, 44]. There are no clinical efficacy data for head and neck cancer because there are no precursor lesions for head and neck cancer and recurrent respiratory papillomatosis (RRP) is very rare. However, it is reasonable to think that the quadrivalent HPV vaccine could prevent these diseases. If these diseases were to be included in the model, the protection by HPV vaccination would be expected to be greater, which would give a lower ICER. HPV-related non-cervical cancers were included in a recent UK modelling study that predicted that, in addition to a median of 700 to 1000 cervical cancers, in the long-term, HPV vaccination could prevent between 620 and 950 cases of these cancers annually in the UK
. Future cost-effectiveness analyses for HPV vaccination should, therefore, include these diseases to provide a more accurate prediction of the cost-effectiveness of HPV vaccination. In light of the recent clinical trial results showing the efficacy of the quadrivalent HPV for the prevention of genital warts and precancerous anal lesions in males it will also be of importance for decision makers to assess the potential impact of vaccination of females and males on female and male HPV-related diseases
[16, 17]. Some countries, e.g. Australia, Canada and the US have recently recommended the inclusion of boys in their vaccination programmes.