Skip to main content

Comparing health services interventions for the prevention of HPV-related cancer

Final Report Summary - COHEAHR (Comparing health services interventions for the prevention of HPV-related cancer)

Executive Summary:
Human papillomavirus (HPV) research has made great progress in the last decade and a growing number of European countries include HPV vaccination and HPV DNA testing in their prevention programs. The CoheaHr project has taken a crucial step in the optimization of the delivery of health services by providing a comprehensive evidence base on the comparative effectiveness and cost-effectiveness of screening and vaccination programs for cervical cancer and other HPV-related diseases. A complete list of publications from the CoheaHr project as well as additional information on the participating institutes can be found at:

The CoheaHr consortium has initiated a number of new studies in the regular screening population to determine the efficacy of HPV testing on self-collected specimens (HPV self-sampling), to assess the safety of life of less frequent screening in women vaccinated at young age, and to measure the acceptability of HPV vaccination in screen-eligible women. Furthermore, the HPV evidence base has been extended with long-term follow-up of European randomized cervical screening studies and systematic reviews.
Key results with regard to vaccination are: i) The acceptability of HPV vaccination at screening ages 25-45 was high across Europe, indicating that HPV vaccination may be extended to older ages. This is particularly interesting in the context of broad-spectrum vaccines that warrant infrequent screening. ii) The frequency of serious adverse events after vaccination did not differ between vaccinated and unvaccinated women, iii) HPV transmission models indicated that including boys in the vaccination program enhances the resilience of the program against a sudden drop in the vaccine uptake, as witnessed in Denmark in 2015. Furthermore, sex-neutral vaccination is very cost-effective at current tender-based vaccine procurement prices. Together, these vaccination results indicate that there is substantial scope for improving the prevention of HPV-related diseases by extending HPV vaccination to older age groups and to males.
Key results with regard to screening are: i) HPV testing on a self-collected specimen is as accurate as HPV testing on a clinician-collected specimen when a clinically validated PCR is used. Similar accuracy for the two collection methods was observed in a meta-analysis and in a trial conducted within the Dutch routine screening population, supporting the use of HPV self-sampling as a primary screening test. ii) The risk of high-grade cervical intraepithelial neoplasia (CIN) was decreased in women that participated in (catch-up) vaccination. iii). Replacing cytology by HPV-based screening does not lead to a substantial increase in overdiagnosis but leads to earlier detection of high-grade CIN iv) An extension of the interval in HPV-based screening programs beyond five years is only warranted in women without a previous positive HPV result, v) Management of HPV-positive women by a single test triage approach offers limited safety for returning women back to routine screening, and vi) A small fraction of lesions may be difficult to detect by cytology, also after multiple rounds of screening, and may give rise to the current significant number of cervical cancers at older age. Together, these screening results suggest that the efficiency of screening programs can be further optimized by implementation of risk-based screening algorithms where the screening interval depends on age, vaccination status and previous screening outcomes.

The results of CoheaHr will support policy makers and other decision makers at the national and EU level to prevent HPV-related disease and to improve the health of European Citizens.

Project Context and Objectives:
Cervical cancer is caused by persistent infection with high-risk human papillomavirus (HPV) and constitutes an important health problem in Europe with about 60,000 new cases per year and 30,000 new deaths per year. Other HPV-related cancers include cancers of the anogenital tract and oropharynx, the incidences of which are increasing. The burden of cervical cancer shows substantial heterogeneity across Europe. The incidence and mortality are particularly high in Central and Eastern European countries where standardized mortality rates of over 10 per 100,000 have been reported.

The heterogeneity in cervical cancer incidence can for a large part be ascribed to variations in cancer prevention efforts in the past. In Finland, organized cytological screening was implemented 40 years ago whereas in most Eastern European countries, cervical screening still relies on opportunistic programmes that typically reach only a subset of the population that is not necessarily at highest risk of disease. In the future, a decrease is expected in the burden of cervical cancer and other HPV-related disease due to the availability of prophylactic HPV vaccines and improved, HPV-based screening methods.


The current organized screening programs in European countries still show considerable variation in design (the starting age of screening, the screening interval) and the coverage rates. In countries with a well-organized screening program and high screening participation, cervical cancer incidence rates have shown a substantial decrease in the last decades of the 20th century and currently, a disproportionally large number of cancer cases are observed in poorly screened and unscreened women. In contrast, in countries that rely on opportunistic screening or that have a program with low screening coverage, cervical cancer incidences have remained high.

The performance of the screening program also depends on the screening test used. Initially, all screening programs were cytology-based, using the PAP-smear for the evaluation of morphological abnormality as indication of a precancerous lesion. HPV DNA testing is more sensitive than cytology for the detection of a precancerous lesion and shows less variability across populations. Besides, a joint analysis of European screening trials (FP7 project PREHDICT, no 242061) showed that HPV testing leads to better prevention of cancer in the subsequent screening round than cytology screening. HPV DNA testing is currently being implemented in several European countries (the Netherlands, Italy, Sweden, UK), but stakeholders may still fear that implementation of HPV screening leads to an unacceptable number of women referred for further gynaecological examination. Therefore, several countries have been reluctant to adopt HPV DNA testing and demand additional evidence on the best follow-up strategy for HPV-positive women.

One of the most important remaining problems of screening is that screening participation is not perfect. To increase participation, “self-sampling” has been proposed as an alternative smear taking method. Self-sampling consists of self-collection of cervico-vaginal specimens for HPV testing at home and is considered as a female-friendly alternative to sampling by the physician. Several self-sampling studies have shown that offering self-sampling attracted 20-40 percent of the screening non-attendees to the program. This indicates that self-sampling may be an effective way to reach marginalized/poorly screened populations and is therefore increasingly used in prevention programs. Furthermore, several studies have indicated that self-sampling leads to less anxiety and pain than clinician-collection.

Another important challenge is that the first vaccinated cohorts will reach screening ages soon. Due to the decrease in background cancer risk in vaccinated cohorts, it is likely that the screening program must undergo changes in order to remain cost-effective. The impact of vaccination on cancer precursor lesions at the first screening round will be observed in the years 2017-2023 depending on the screening program. Vaccination is expected to lead to a decrease in cancer risk in the population and this will lead to a different trade-off between screening efforts and benefits. Coherent prevention strategies that optimally exploit both vaccination and screening in a country-specific setting therefore need to be developed. An important issue is whether less intensive screening should be offered to all women from a vaccinated birth cohort irrespective of vaccine acceptance or to vaccinated women only. The latter approach may be preferable from a cost-effectiveness viewpoint but also places a larger demand on logistics and registry.


Since 2008, 22 European countries have introduced HPV vaccination in the routine immunization program for girls with target ages between ten and fourteen years. Catch-up vaccination programs have also been offered for varying ages but not beyond the age of 26 years, although the vaccines are registered for use in older women as well. The primary target of the first generation vaccines is protection against HPV16 and HPV18 infections, but those vaccines also offer some cross-protection against other oncogenic HPV types. Since 2015, a new vaccine has become available that directly targets seven oncogenic HPV types and two low risk HPV types that are associated with genital warts.

HPV vaccination programs vary within Europe with regard to target age and population (girls only or sex-neutral) and with regard to the reimbursement system and organization. Delivery strategies differ between European countries and the main three delivery strategies are: vaccination at schools, vaccination at health centres, and mass vaccination in sports centres. Furthermore, 3-dose coverage rates vary greatly across European countries and lie between twenty and eighty percent. In Central and Eastern European countries, HPV vaccines have been approved but have only been implemented in a small number of jurisdictions. Two reasons for not having a national HPV immunization program are i) lack of sufficient health resources and ii) negative public perception of vaccination.

HPV vaccination may also be extended beyond the target group of adolescent girls and two new initiatives for HPV vaccination are sex-neutral vaccination and adult female vaccination. Sex-neutral vaccination has been implemented in only a few European countries because of unfavorable cost-effectiveness calculations. However, with HPV vaccine costs declining rapidly because of tender-based procurement, sex-neutral vaccination may be cost-effective and be an interesting preventive strategy for European countries. Nonetheless, many countries await scientific information on the projected direct effect of male vaccination on HPV-related disease among males and the indirect effect on cervical disease before pursuing sex-neutral vaccination.


The variation across European countries in cervical cancer control strategies asks for objective data-supported criteria for future cancer prevention. There are several possibilities for effective prevention of HPV-related cancers which are reflected by the large heterogeneity in prevention policies in Europe. Substantial variation in costs and the burden of HPV-related disease across European countries suggest that the optimal design of preventive policies may differ among countries. A Europe-wide effort to compare the effectiveness of different preventive strategies will enable us to provide a reliable and comparable evidence base that can be used when evaluating the effectiveness and cost-effectiveness of preventive policies implemented under country-specific preventive services conditions. Therefore, the main objective of CoheaHr is to build a reliable resource on the effectiveness and cost-effectiveness of different preventive strategies against HPV-related disease. Our resource contains: i) new data obtained from implementation and feasibility studies, ii) new long-term follow-up data from randomized screening studies, and iii) systematic reviews and meta-analyses of existing HPV studies.

The main objective covers several specific objectives. They can be divided into A. new screening and vaccination studies and B. update of existing screening trials, meta-analyses and mathematical modeling.

A. New screening and vaccination studies
i) To compare the accuracy and feasibility of HPV testing on self-collected samples (HPV self-sampling) with HPV testing on clinician-collected samples within the setting of routine cervical cancer screening. For this purpose, a diagnostic study was set up in the Dutch screening population where about 14,000 women invited for routine screening were randomized to HPV self-sampling or clinician-based HPV testing. In the 18 months ethical review, patient preference was identified as an important aspect of prevention of HPV-related disease. We have gathered questionnaire information in a subset of about 5500 women with regard to their experiences with self-collection and/or clinician-collection of specimens. The importance of this study is that published studies comparing the accuracy of HPV testing on self-collected versus clinician-collected samples were small or performed in underscreened women, making it difficult to generalize results to the regular screening population. Such evidence is needed to support HPV self-sampling as an efficient and female-friendly primary instrument in routine screening.
ii) To set up and study comparing the effectiveness and cost-effectiveness of different screening policies in women vaccinated at young age. HPV16/18 vaccinated women from Finland were randomized to frequent screening at 22/25 years, less frequent screening at age 25/30 or screening only at age 30. The importance of this study is that vaccination reduces the need for subsequent screening and that it is important to assess the starting age of screening that provides optimal balance between safety and screening-related harms. An early vaccination trial in Finland enabled the CoheaHr consortium to set up a study for obtaining randomized, direct evidence on optimal screening of vaccinated women.
iii) To study screening outcomes in vaccinated women from Sweden. The analysis was conducted in a routine screening setting and information was obtained by linking the vaccination and screening registry.
iv) To identify global and local determinants of acceptability, HPV vaccine compliance, logistics and programmatic issues of vaccinating women 25-45 years attending routine cervical cancer screening. This was a multi-country study where at least 250 women per country were asked to fill out a questionnaire and were offered free vaccination. The importance of this study is that it assesses the feasibility of vaccination at older ages which may further reduce the need for subsequent screening, in particular when broad-spectrum vaccination against multiple HPV types is offered.

B. Updating existing screening trials, meta analyses and mathematical modeling
CoheaHr participants have access to existing randomized screening studies initiated by participants of the CoheaHr consortium. These studies provide a wealth of information to study different screening strategies. A specific objective of CoheaHr is to update data bases to answer questions on the long-term risk of cancer and precancer as a function of baseline and repeat screening outcomes. Furthermore, the management of HPV-positive women by new triage tests (p16/Ki67 staining, viral methylation, host cell methylation, miRNA, HPV genotyping) was evaluated in these screening studies. Scientific evidence from the screening studies was complemented by conducting meta analyses. Meta-analyses were initiated to acquire information on the accuracy of HPV testing in a large number of studies (both clinician-collection and self-collection), the frequency of complications of treatment of precursor lesions and the efficacy and safety of HPV vaccines.
Models were developed to study specific objectives that cannot be studied directly from new clinical data, but involve mathematical modelling. HPV transmission models and Markov disease progression models were built and validated to estimate the expected effectiveness of selected vaccination and screening scenarios in different populations. These include a variety of scenarios including both girls’ only and sex-neutral vaccination. The models were also used to assess the effect of specific effectiveness determinants such as changes in sexual behaviour, herd effects, and cross-protection of vaccination on the predicted impact (i.e. population-level effectiveness) of vaccination. Likewise, the amount of resources were estimated necessary to implement the simulated vaccination and screening scenarios. These modeling objectives could only be achieved by using the output from the new and existing clinical studies and meta analyses as model inputs.
A final objective of CoheaHr was to develop an e-learning course with a CoheaHr module explaining the most important results of the Coheahr project. The course should be available in multiple languages and be virtualized.

Project Results:
This section presents the results of the CoheaHr project. The main results are described first, followed by a more detailed description.



HPV testing at the clinician’s office may be a barrier to attend screening in marginalized populations or settings with limited health resources. A potential alternative to clinician-based HPV testing is HPV testing on a self-collected specimen (HPV self-sampling). In a new diagnostic study, the CoheaHr consortium showed that HPV self-sampling is as accurate as HPV testing on a clinician-collected specimen in the Dutch routine screening population. A similar result was found in a meta-analysis when a clinically validated PCR was used. Together, these studies support the use of HPV self-sampling as a primary screening test and as a test in underscreened populations.

Vaccinated women are expected to require less screening than unvaccinated women. It is therefore important to assess the starting age of screening that provides optimal balance between safety and screening-related harms. An early vaccination trial in Finland enabled the CoheaHr consortium to set up a study for collecting obtain randomized, direct evidence on optimal screening of vaccinated women within a routine screening population. HPV16/18 vaccinated women from Finland were randomized to frequent screening at 22/25 years, less frequent screening at age 25/30 or screening only at age 30. Women with a first screen at age 25 or 30 were screened at age 22, but results were not disclosed unless immediate colposcopy was indicated. The first intermediate screening results at age 22 have been collected and have been submitted for publication, indicating low rates of cervical abnormality and supporting a postponement of the start of screening until age 25 or 30. The latter finding was also confirmed in Swedish real-life data, obtained by linking the vaccination and screening registry.

CoheaHr has provided new evidence on the effectiveness of both cytology and HPV screening. The CoheaHr consortium published a joint analysis of four European screening trials (Swedescreen, POBASCAM, NTCC, Artistic) based on a database created in a previous project (PREHDICT, FP7 242061), that includes a standardised format for registering data and the actual interventions made. This publication (Ronco et al. Lancet 2014) showed that HPV screening better protects against cancer than cytology screening and has been instrumental in health policy making. An update of the Swedescreen and POBASCAM data during the CoheaHr project enabled us to show that HPV screening does not substantially increase the level of overdiagnosis as compared to cytological screening but leads to earlier detection of high-grade CIN. Furthermore, the POBASCAM data indicated that an extension of the interval in HPV-based screening programs beyond five years is only warranted in women without a previous positive HPV result. A model-based evaluation indicated that a small fraction of lesions may be difficult to detect by cytology, also after multiple rounds of screening, and may give rise to a significant number of cervical cancers among older women as obseved in the screening trials.

The CoheaHr consortium also studied several triage strategies for HPV positive, cytology negative women, including genotyping p16/Ki67, viral and host cell methylation and repeat cytology and repeat HPV testing using Swedescreen, POBASCAM Meta analyses revealed that management of HPV-positive women by a single test triage approach offers limited safety for returning women back to routine screening. Pretest-posttest probability plots are useful for visualization and for informing about disease risks.

Meta-analyses were also conducted to study obsterical harms associated with the exicisional treatment of cervical cancer and reveald that the risk of preterm delivery increases with the depth of the excised cone.


The CoheaHr team studied whether an extension of HPV vaccination to older ages is well accepted and whether it is (cost-)effective to include boys in the HPV vaccination program.
In a vaccine acceptability study conducted in eight European countries, it was shown that HPV vaccination at screening ages 25-45 was high across Europe, indicating that HPV vaccination may be extended to older ages. Such an approach is particularly interesting in the context of broad-spectrum vaccines that protect against multiple HPV types and warrant less intensive screening.

The impact of sex-neutral vaccination was studied by mathematical models. The analysis provided two main arguments in favour of sex-neutral vaccination: i. Including boys in the vaccination program enhances the resilience of the program against a sudden drop in the vaccine uptake, as for instance witnessed in Denmark in 2015. ii.Although earlier cost-effectiveness studies indicated that sex-neutral vaccination is not cost-effective at the high vaccine list price, sex-neutral vaccination is very cost-effective when using current tender-based vaccine procurement prices, a two dose vaccination schedule and incorporating health effects on non-cervical HPV-related disease such as anal cancer and oropharyngeal cancer.


More detailed results will be described in this section.


In the period March 2015 till December 2016, about 185,000 Dutch women attending routine screening were invited for participation in a primary HPV self-sampling study. Women that were enrolled were randomized (1:1) between self-collection of cervico-vaginal material (intervention arm) and physician-based collection (control arm). In both arms, HPV DNA testing was conducted on the collected specimen using the validated GP5+/6+ PCR test (Labo Biomedical Products B.V. Rijswijk, The Netherlands). Self-collection was done with the Evalyn brush and clinician-collection with the Cervex Brush (Rovers Medical Devices BV, Oss, The Netherlands). Samples were collected in ThinPrep media (Hologic, Marlborough, MA, USA). Eventually, approximately 14,000 samples were received.

Main reasons for not submitting a sample were that women preferred to participate in the regular programme, only agreed to be randomized to the other sampling method,logistical reasons or difficulty of the sampling procedure. There was no difference in mean age of women not submitting a sample in the self-sampling group and in the clinician-based sampling group.
The HPV-positivity was similar in both arms: 7.4 percent of self-collected and 7.2 percent of clinician-collected samples tested positive for HPV. Of the HPV positive women, 209 women in the self-sampling group and 196 women in the clinician-based sampling group had abnormal cytology and were referred for colposcopy at baseline. 87·6% of the women in the self-sampling group with normal triage cytology at baseline had repeat cytology, showing good compliance with repeat testing. In the clinician-based sampling group, 89·4% of the women with normal cytology at baseline had repeat cervical cytology.

To assess the diagnostic accuracy of a test, both tests should be applied to every woman. Notably, if only one test is performed per woman (unpaired study), a very large sample size is needed to evaluate the accuracy of the test. In our setting, a fully paired design was not feasible for logistical reasons and it was easier to only re-test HPV-positive women on a sample obtained with the other collection method. This design is called a paired-screen positive design (Alonzo and Kittelson, Biometrics 2006) and our sample size of our study was powered on the basis of this design. The main result of the study was that the relative sensitivity of HPV self-sampling as compared to clinician-based sampling was nearly one for both end-point CIN2+ and end-point CIN3+. Furthermore, HPV-self-sampling was non-inferior to HPV testing on clinician-collected samples according to international Meijer consensus criteria for new HPV DNA tests (Meijer, Int J Cancer 2009). The analysis was repeated after excluding women with a poor screening history but gave very similar results. Therefore, we conclude that our study supports the use of HPV self-sampling as an accurate alternative to HPV testing on clinician-collected samples. This result, together with meta-analytic results which showed that HPV self-sampling has good performance in a large number of populations of women with incomplete or limited previous screening (Arbyn et al. Lancet Oncol 2014) opens the way for HPV self-sampling as an alternative primary test in HPV-based screening programs.

Among a subset of about 5000 participating women, three different questionnaires were sent out to gain insight into the experiences with the HPV self-sampling method: HPV-negative women from the self-sampling group were asked about their experiences with self-sampling; HPV-negative women in the clinician-collection group were asked about their experiences with clinician-based sampling; and HPV-positive women from both study groups were asked about their experiences with both self-sampling and clinician-based sampling. HPV-negative women in the self-sampling group reported lower levels of shame, nervosity, discomfort and pain during sampling than HPV-negative women in the clinician-based sampling group. However, trust in correct sampling was significantly higher among HPV-negative women in the clinician-based sampling group than in the self-sampling group. Similar results were obtained among women of group who performed both self- and clinician-based sampling. Furthermore, the majority (>75%) of HPV-positive women that performed self-sampling and clinician-based sampling preferred self-sampling to clinician-based sampling in future screening.

Together, the diagnostic study and the attitude study show that HPV self-sampling is the preferred method for the majority of women invited for routine screening. The relatively low level of trust in the self-collection procedure indicates that reassuring women that HPV self-sampling is as accurate as clinician-based HPV testing is important when implementing HPV self-sampling in organized programs.


Community randomized trial of vaccinated women in Finland
HPV16/18 vaccinated women were randomized in a community-randomized trial (CRT) by the end of 2014 into three different arms. They were all invited for screening at ages 22, 25 and 30. In the frequently screened arm A1, the participants, vaccinated at age 12-15, received information on cytological findings at ages 22, 25 and 30. In the infrequently screened arm A2, the participants, vaccinated at age 12-15, received only information indicative of colposcopy (ASC-H, HSIL) on cytological findings at ages 22 and 25. In the safety arm A3 the participants, who had been vaccinated at age 18, received only information indicative of colposcopy on cytological findings at age 22. At the end of the trial at which time point women have reached 30 years of age, they will be informed about the HPV DNA results. In 2014-2017, all women were invited for a first screening round at the age of 22 years (frequent screening Arm 1, 1783 participants, infrequent screening Arm 2, 1830 participants, and safety Arm 3, 1211 participants). The first birth cohorts were also invited for the second screening round at age 25 in 2017 and 2018.

Baseline characteristics of the trial arms showed homogenous abnormal cytology and high-risk HPV findings, albeit low HPV16/18 prevalence/incidence in the infrequently and frequently screened arms (arm 1 and 2) where women were vaccinated as early adolescents. Safety of infrequent screening was confirmed in 2018 by equal occurrence of CIN3 findings in the infrequently screened, cross-vaccinated participants and the frequently screened participants in the second screening round. The trial will deliver data on the non-inferior accuracy of infrequent screening of HPV vaccinated females in 2025.

In an evaluation on the impact of the vaccine on HPV type prevalence, the CRT showed that the cohort members were protected against infection by non-vaccine HPV types by herd effects, and this effect was especially strong for cross-protected HPV types 31/33/35. Furthermore, cross-vaccination at the end of the CRT during years 2010-2014 formed a sizeable safety arm for the screening trial. Wide-spread cross-protection probably has kept type-replacement at bay in the cohorts followed since 2010.

Registry-based study in Sweden
In Sweden, vaccination started in the 1980’s. The first vaccinated women are now entering the screening program. Differences in findings between vaccinated and unvaccinated women are of interest for future screening programs. HPV vaccination coverage increased substantially in Swedish female birth cohorts from below 10% (1985-1989) to 64% (1993-1999). Also unvaccinated women are expected to benefit from the protective effect of vaccination in the population. The protective effect is expected to increase with increasing vaccination coverage in the population.

The first vaccinated birth cohorts have been invited for their first cytology screen at age 23. After an abnormal cytology, a woman was referred to colposcopy. At the first screen at age 23, the detection rates for pre-cursor lesions (CIN2+) were lower in vaccinated women than in unvaccinated women. In birth cohorts 1989-1992, the percentage of women with a CIN2+ lesion was 0.6% of the vaccinated women and 1.4% of the unvaccinated women. This figure is obtained when only the high-risk women with HSIL+ outcome at cytology were referred to colposcopy. If all women with LSIL+ outcome were included in the calculation for the referral to colposcopy rate, more lesions were found, while the difference between vaccinated and unvaccinated women remained (detection rate 1.3% in vaccinated women and 2.4% in unvaccinated women).

The probability that a CIN2+ lesion was found at colposcopy in women with an abnormal cytology (the positive predictive value) also differed between vaccinated and unvaccinated women, both when all women with LSIL+ outcome were referred to colposcopy and when only those women with HSIL+ were referred to colposcopy. The positive predictive values (PPVs) of LSIL cytology for CIN2+ in birth cohorts 1989-1992 were 19.4%, 13.0% and 9.7% for women unvaccinated, initiating vaccination at age 17-22, and initiating vaccination before age 17, respectively. The PPVs of HSIL cytology for CIN2+ in birth cohorts 1989-1992 were 69.8%, 62.7% and 55.0% for women unvaccinated, initiating vaccination at age 17-22 and initiating vaccination before age 17, respectively.

Already in the 1989-1992 birth cohorts with a low to moderate vaccination coverage, there was a significant difference in findings between vaccinated and unvaccinated women. This effect is likely to be present or even stronger in later birth cohorts since similar decreasing PPV trends were observed from the birth cohort 1993 onwards for the CIN2+ detection rates after cytological LSIL and HSIL findings.


Vaccination trials have documented that HPV vaccines efficacy among adult women up to age 45/55 is excellent provided they are HPV negative at the time of vaccination. However, HPV vaccination coverage of adult women in Europe is low to very low. Therefore, a multinational study was undertaken to identify global and local determinants of acceptability, HPV vaccine compliance, logistics and programmatic issues of vaccinating women 25-45 years attending routine cervical cancer screening.

In total, eight countries within the CoheaHr consortium recruited at least 250 women in the age 25 to 45 per country. In Spain, under-screened women (n=150 planned) were additionally targeted. Participating women self-completed a questionnaire on sociodemographic characteristics, medical history, attitude towards screening, knowledge and opinion on HPV vaccine and vaccine acceptability. Those who accepted, were offered free HPV vaccination and were followed-up for compliance and safety data collection. The study HPV vaccines were provided at no cost by both HPV manufacturers,.

In the UK, the HPV vaccine could not be administered to study participants due to changes in the call-recall system of the cervical cancer screening program, which was planned to be used for recruiting potential participants. Instead, a questionnaire-only study was conducted to assess the potential uptake of HPV vaccination and its reasons in British women.

The recruitment strategies differed by country:
Sweden and Denmark used a population-based approach. Based on age and place of residency, potentially eligible women were identified using screening or census registries and invited to participate in the study via postal letters that included the questionnaire. Therefore, the randomly selected population sample included women (i) already vaccinated or (ii) pregnant, breastfeeding or planning to get pregnant in the near future. Eligible women were asked to return the filled questionnaire and to phone if they accepted vaccination to arrange for a visit.

Spain, Finland, United Kingdom, Belgium, Slovenia, France and Germany used convenient recruitment; only women attending healthcare or screening services were invited to participate using brochures, posters, HCP criteria during the clinical visit and/or phone call to recently screened women. Germany only invited HPV screened women part of the Wolfsburg HPV Epidemiological Study (Petry et al. 2013), a prospective population-based surveillance study on the prevalence and incidence of HPV infections and associated diseases.

Recruitment took place between April 2016 and May 2018, with recruitment periods between 4 and 22 months, depending on the country and site. Spain, Finland, Belgium and Slovenia recruited participants in at least two independent locations. Potential participants were also offered vaccination irrespective of questionnaire completion.

Participation and response rates by country
The different strategies used for study invitation did not allow accurate estimation of the response rate in countries with convenient sampling due to the number of invited women being unknown. In Denmark (n=1,932) and Sweden (n=2,962), 34.4% and 17.3% of invitations were answered. Potential answers included reasons for not participating or completing questionnaire.

4132 women replied to the study invitation; 325 refused to participate and 161 women completed the questionnaire despite being already vaccinated and therefore not eligible to participate. Therefore, a total number of 3,646 eligible women participated in the study; 10 of which were vaccinated but did not returned the completed questionnaire (considered as vaccine acceptors for reporting purposes).

In France, 63 women participated in the study. The targeted number of recruited patients was not achieved due to the site routine clinic workload and an overestimated recruitment rate. The number of participants in the other countries ranged from 307 in Belgium to 693 in Spain.

Vaccine acceptance, uptake and compliance
For each country, among the study participants, we estimated the proportion of:
- Women that accepted vaccination (vaccine acceptance)
- Women that got at least one dose of HPV vaccine (vaccine uptake)
- Women who completed the 3-doses schedule (vaccine compliance).

Acceptance rates were higher in countries with convenient sampling (>80%), except Slovenia (49.8%), than those countries with population-based studies (around 50%). Results in Slovenia are consistent with the acceptance rates observed in the routine immunization program of 12-years old girls (46.4% in 2016/17 (2018).

Uptake was only slightly lower in countries with convenient sampling in settings where vaccination was scheduled for a date different than when they were invited. In Sweden and Denmark, 69 and 19 of participating women accepted the vaccine in the questionnaire but did not call the vaccination site to schedule the vaccine administration. Of note, 171 women were not eligible for vaccination, mainly due to current or planned pregnancy in the following 12 months (n=155).

Reasons for vaccine acceptance and refusal
Women who accepted or refused vaccination were asked to tick, among the suggested answers, all of those that were relevant to their decision.

A total of 2521 and 660 women who accepted or refused the vaccine provided the reasons associated with their decision. Reasons for vaccine acceptance were not collected in Sweden (n=182), whereas in the remaining countries, 26 (1%) versus 245 (27.1%, range), respectively, chose not to reply.

Potential reasons for accepting vaccination provided in the questionnaire included the vaccine safety and efficacy, its free provision in the study, recommendation to get vaccinated and cervical cancer as a serious disease worth preventing. In all countries, most cited reasons to accept vaccination were the vaccine efficacy (85-97%) and the seriousness of cervical cancer (89-100%). Of note, only 49% and 55% of Danish and German women chose the vaccine safety as one of the reason to get vaccinated.

Potential reasons for refusing vaccination provided in the questionnaire included doubts about the vaccine safety and efficacy, dislike of vaccines in general, lack of time, no need to get vaccinated, recommended not to get vaccinated (others advice), need for more information or consultation.

Most cited reason for not getting the vaccine was safety concerns (range: 30-59%). No reply was <15% in most countries except Spain (23%), Sweden (37%) and Slovenia (43%). Need for more information was especially relevant (around 70%) in the UK and France, although based on low numbers. Eleven out of 13 French women and 27% of Belgian women needed a consultation before getting vaccinated.

Based on the results of the trial, the following can be concluded:
- Acceptance rates were high despite its strong dependence on the recruitment method used and the doctor’s recommendation, especially when the vaccine is offered for free in clinical or screening setting.
- More information and safety reassurance seem to be relevant for informed decision making. Nevertheless, the study results suggest that there are no major barriers to vaccination of adult women with respect to the acceptability.
- In countries with convenient sampling, we cannot ensure the representativeness of the study population. Furthermore, although low (<5%), some women participated in the study following referral by friends and/or family (snowball effect).


The transmission of carcinogenic infections caused by High-Risk (HR) Human Papillomavirus (HPV) types and their progression towards cervical cancer have been reproduced using separate sets of mathematical models. HPV infection transmission and cervical cancer natural histories were modeled separately to keep uncertainties of each biological phase separate and independent. This approach has allowed us to focus separately on the impact of HPV vaccination and cervical cancer screening (in both unvaccinated and vaccinated birth cohorts). Changes in the HR-HPV prevalence are the earliest to manifest and offer the opportunity to monitor the impact of HPV vaccination programs, whereas variations in the occurrence of precancerous lesions and cervical cancers take longer to be observed. All models were also designed to account for the demographic characteristics of the simulated populations. Modeling decisions were regularly discussed among CoheaHr partners at bi-annual consortium meetings and ad-hoc technical workshops.

Both HPV infection transmission and cervical cancer progression have been modeled using compartmental and population-based approaches; both mathematical and biostatistical methods were used. Details about model developments have been published.

To parameterize and calibrate the models, we used data from countries contributing to CoheaHr. In particular, useful information was extracted from large EU screening (i.e. NTCC, Swedescreen, POBASCAM, ARTISTIC) and vaccination (i.e. Finnish Community-Randomized) trials. Data from routine surveillance activities, such as Chlamydia and Cervical Cancer screening programs, were also used to assess HR-HPV prevalence and cervical cancer screening performances, respectively.
Following the parameterization, calibration, and validation of the models, we estimated the expected effectiveness of selected vaccination and screening scenarios in different populations. We also used the models to assess the effect of specific effectiveness determinants such as changes in sexual behavior, herd-effect, and cross-protection of vaccination, on the predicted impact (i.e. population-level effectiveness) of vaccination. Likewise, we estimated the amount of resources necessary to implement the simulated vaccination and screening scenarios. Concurrently, we also conducted surveys, reviews, and epidemiological analyses of existing datasets to complement the model-based findings and to independently validate model-based outputs.

Model development and validation
Besides vaccine efficacy and vaccination coverage, the key determinants of the impact of HPV vaccination are a) the sexual behavior of a population, b) the duration of an HPV infection and c) the patterns of naturally acquired immune protection. Sexual behaviors are regulated by socio-cultural norms and changes over time, HPV infection duration is variable between types, and patterns of naturally acquired immune protection are not well characterized. We have investigated the influence of each of the above-mentioned determinants of the impact of HPV vaccination.

We have reviewed the assumptions about naturally acquired protection incorporated in published transmission models and assessed their validity against available data from vaccine trials. The available data suggest that approximately 15% young unvaccinated women in the PATRICIA trial has some protection from HPV16 infection because of naturally acquired antibodies; this proportion may be different in older women. Naturally acquired immunity has not been demonstrated in men. We have incorporated these findings in our model-based projections. The validity and comparability of our HPV transmission model outputs have been assessed by comparing the predictions of population-level impact, herd immunity, and elimination after HPV vaccination across 16 transmission-dynamic models published in the literature.

Furthermore, we have assessed the consequences on the impact of HPV vaccination of changing sexual behavior in a population. The model-based results suggest that HPV infection is significantly easier to control in a population with a traditional sexual behavior (i.e. populations in which genders display different age-specific sexual activity patterns and a wide gap in ages of spouses or cohabitating sexual partners) than in a population with a gender-similar sexual behavior (i.e. populations in which genders have similar age-specific sexual activity rates and a narrow gap in partner’s age). Therefore, the transition from traditional to gender-similar sexual behavior, as observed in many settings across Europe, which is accompanied by an increased risk of HPV infection, could be attenuated by early implementation of HPV vaccination. We have also shown that international variations in the prevalence of HPV infection also derive from differences in sexual behaviors. Such differences have a direct effect on the levels of herd protection and affect the impact of vaccination programs. Consequently, similar vaccination programs may generate different levels of impact depending upon the population’s pre-vaccination HPV prevalence. Furthermore, we have shown, analyzing the IARC multicentric case-control study, that in women with a large gap in partner’s age (strongly associated also to a younger woman's age at first sexual intercourse) were at a higher risk of developing cervical cancer8. All these findings suggest that a particularly favorable, time-limited window currently exists for the introduction of HPV vaccination in countries with a substantial fraction of the population displaying a traditional sexual behavior.

We assessed the consequences of type-specific HPV infection duration and patterns of naturally acquired immune protection on the impact of HPV vaccination. The model-based results show that the characteristics of individual HR-HPV types strongly influence herd protection and determine the level of coverage and cross-protection of bivalent vaccines required to reduce or eliminate the infection through HPV vaccination. HPV16 infection and related cancers are the most difficult to eliminate. Our model-based findings are strongly supported by the results reported from the Finnish community randomized trial, in which low to moderate coverage with a high efficacy bivalent HPV vaccine against HPV16/18, has shown (as predicted with the model) strong herd effect and cross-protection against HPV18 and HPV31/33/3510.

We have also developed an age-structured model of the progression of HR-HPV infections to invasive cervical cancer through a pre-cancerous lesion stage. The population of women has been subdivided into different health statuses, i.e. HR-HPV infected, with CIN (cervical intraepithelial neoplasia) grade 2/3 lesions (further subdivided in CIN2/3 screened and treated and CIN2/3 undetected by screening), recovered from lesions, and with invasive cervical cancer. The validity of model-based estimates of the impact of cervical cancer screening depends on an accurate representation of HPV progression towards cervical cancer. To optimize the accuracy of our model we a) informed the model’s inputs with high quality data (i.e. HPV prevalence and cervical cancer screening performances) from Italy and Sweden, b) calibrated model-based outputs to age-specific CIN2/3 detection rates and cervical cancer incidence from the two countries, c) simulated a screening trial using the calibrated progression model, and d) compared the model-based outputs with the findings obtained from four large EU screening trials (i.e. NTCC, Swedescreen, POBASCAM, ARTISTIC).

Recently observed data from the EU trials suggest that precancerous cervical lesions are not homogeneously detected by cytology testing and a small fraction of lesions is very difficult to detect even with repeated rounds of screening. This small fraction of lesions may give rise to a significant number of cervical cancers among older women. We have used our model to test this hypothesis and simulated a screening trial assuming alternatively heterogeneity and homogeneity of the lesions with respect to cytology testing and compared the model-based results with EU trials findings (see points c and d in the previous paragraph). The results of our simulations support the hypothesis that a small fraction of lesions is very difficult to detect even with repeated rounds of screening. This finding may have a key implication in the quantification of the delay between the occurrence of an HR-HPV infection and the possible occurrence of invasive cervical cancer, and in defining safe screening round intervals in an unvaccinated population.

Overview of existing vaccination and screening strategies in Europe
An overview of existing vaccination strategies in Europe was provided. This questionnaire-based study aimed at characterizing both the organization of HPV vaccination efforts in Europe as well as the corresponding monitoring systems. This study sets a baseline for a long-term follow-up of the vaccine efficacy and population-level vaccination effectiveness and for the evaluation of integration of vaccination and screening policies. The results of this European-wide study show that while many countries have organized vaccination programs, the level of detail with which these countries can monitor and evaluate their programs is highly heterogeneous. Of note, a novel method was developed to assess the effectiveness of HPV vaccination in absence of HPV prevalence baseline data and of precise information on HPV vaccination coverage. The method has been piloted in Sweden, where findings suggest that the overall effectiveness among first vaccinated birth cohorts ranged between 40% and 50%.

Similarly, we have provided an overview of present cervical cancer screening strategies in Europe. This questionnaire-based study aimed at identifying the key components and indicators of cervical screening management and implementation and at assessing their correlation with cervical cancer protection achieved. The results of this European-wide study set a baseline for regular and standardized monitoring in Europe. Again, the results of this study show that organized efforts for quality assurance, monitoring and evaluation of cervical cancer screening are implemented to a different extent across European countries. In particular, key performance indicators, such as coverage and compliance, are not estimated in a comparable manner between most countries. A second European screening report (2017) also confirms a substantial heterogeneity across Europe in the implementation of cervical cancer screening despite an increase of approximately 12% of the proportion of eligible European populations having access to cervical cancer screening.

Model-based projections of the impact of cervical cancer prevention
We assessed the effectiveness and costs of upscaling HPV vaccination in a high-income setting using Italy as an example. In particular, we have compared 3 options to upscale an already started program targeting 11 year-old girls: a) girls-only coverage improvement; b) addition of 11 year-old boys; or c) 1-time catch-up of older girls. Our findings show that increasing vaccination coverage among girls, if feasible, is the most effective option for decreasing HPV16/18. If the coverage of girls cannot be further improved, vaccinating boys is justifiable if initial vaccine cost has at least halved, because this option would almost double the number of vaccinees. Also, 1-time catch-up campaign of a few cohorts of older girls and younger women would move forward by about 3-to-5 years the impact of vaccination among targeted cohorts regardless of the vaccination strategy adopted for adolescent girls and boys. However, due to the higher probability of previous exposure to HPV16/18 infection, additional catch-up of older birth cohorts would follow a diminishing return pattern. Our previous work was expanded to show how integration of expanded catch-up HPV vaccination with HPV-based cervical cancer screening (so-called “FASTER” concept) has the potential to accelerate the control of HPV-related cancers. Currently, the “FASTER” concept is contemplated as a possible option in the initiative for “Cervical Cancer Elimination as a Public Health Problem” by the World Health Organization (WHO).

We also compared the projected impact of alternative HPV vaccination strategies in a setting of decreasing HPV vaccine prices, such as Sweden. In particular, we have compared the existing vaccination program to alternatives, accounting for a 1-time catch-up vaccination (up to age 26 years), with or without routine vaccination of school-age boys. We also assessed the resilience (i.e. the propensity to maintain high effectiveness) of vaccination alternatives to conditions that might undermine vaccination coverage. Our findings show that vaccination strategies based on catch-up vaccination of females and males are effective for accelerating HPV prevalence reduction. Expanding vaccination programs to include additional cohorts and target males as well results in an increased number of vaccine doses used, especially at the start of the program involving extended catch-up vaccination. Vaccinating boys would double the number of vaccine doses. However, this increase in the number of doses of vaccine could be compensated by the decrease of vaccine price and changes in vaccine dosage schedules (from three doses to two doses per vaccinee). A crucial effect of male vaccination is the mitigation of loss of vaccine effectiveness in the face of an unexpected reduction in coverage. The resilience to temporary changes in vaccination coverage of strategies that include male vaccination is likely to be crucial to the safe adoption of HPV-based cancer prevention programs with robust long-term effectiveness. Ensuring a resilient program will be most important in settings where funding, public opinion, and health priorities are changeable. A study reporting systematically collected information on HPV vaccine unit prices in European tender-based settings from 2007 up to and including 2017 shows that, setting-specific variation notwithstanding, strong price reductions have been achieved over time for the HPV vaccines since vaccine introduction in 2007, thereby increasing the economic efficiency of HPV vaccination. An analysis of ten years of HPV vaccination in the Netherlands shows a recent and remarkable decline of vaccine coverage possibly related to safety concerns, limiting the benefits from girls' vaccination and increasing the potential additional benefit of sex-neutral HPV vaccination. The efficiency of vaccinating boys was evaluated relative to increasing vaccine uptake among girls in the Netherlands and the cost-effectiveness of a sex-neutral program of vaccination was assessed. Overall, the above-reported findings show that vaccinating boys along with girls is only modestly less efficient than increasing uptake among girls and highly likely to be cost-effective. Of note, The Swedish Public Health Agency concluded its evaluation of HPV vaccination of males (2017) and has recommended that males should be included in the routine, school-based vaccination program of 10-12 year-olds.

Manuscripts reporting the results of model-based cost-effectiveness projections of different screening scenarios in unvaccinated and vaccinated birth cohorts are in preparation. In particular, using the HPV progression and cervical cancer screening model, we have estimated the impact of several screening scenarios, in vaccinated and unvaccinated birth cohorts. These projections provide useful information to a range of settings with cervical cancer prevention at different stages of implementation. Estimates were drawn simulating cervical cancer screening in unvaccinated and vaccinated birth cohorts both in Italy and Sweden, where high-quality data are available. The differences in HR-HPV prevalence and cervical cancer screening coverage observed in the two countries encompass a large range of variability as observed across Europe.

We have used cumulative incidence of cervical cancer at years 75 of age as an outcome measure and vaccination coverage as well as lifetime number of screening rounds as measures of resources devoted to cervical cancer prevention. Both vaccination coverage and lifetime number of screening rounds are more flexible indicators than monetary costs and can be used to assess the amount of human and logistic resources as well as financial costs, necessary to implement cervical cancer control and prevention in different settings.

According to our estimates, the introduction of HPV-based screening in the absence of vaccination would systematically improve the current situation even with extended screening intervals. Overall, the impact of screening is expected to increase as a function of lifetime number of screening rounds. Similar results were obtained for birth cohorts vaccinated with a routine vaccination program and with a catch-up program targeted to older ages. The definition of an optimal combination of screening age-range and screening round intervals (i.e. lifetime number of screening rounds) may vary across different settings and will depend upon HPV vaccination coverage, the definition of an acceptable threshold of lifetime cumulative cervical cancer incidence, and available resources to support the costs of the cervical cancer screening program.

Finally, we have shown how modeling can be very useful to design and evaluate health programs but cannot overcome uncertainties in knowledge about the relevant variables, thus, integration with routine surveillance and field studies is needed. We also have clarified how modeling can contribute to assessing national progress in cancer control: access to open-source models and high-quality data are key in enabling countries to plan and effectively monitor health problems generally, and in tailoring preventive actions that one day will lead to the global elimination of cervical cancer as a major public health problem.


There is an urgent need for building a tool for the comparison of different screening policies in unvaccinated and vaccinated women and combinations of screening and vaccination. This enables us to define the best prevention policies by providing direct estimates of positive and undesired effects. Furthermore, it facilitates the design of future intervention studies and mathematical policy models by estimating parameters of the natural history of HPV infection and pre-cancer/cancer.

We have already established a pooled database of European randomised controlled trials of HPV-based cervical screening within a previous FP7 project (FP7 no 242061, PREHDICT), that includes a standardised format for registering data and the actual interventions made. This pooled database has enabled us to show that HPV screening leads to better prevention of cancer than cytology screening. We conducted the analyses during the PREHDICT project and finally published the analysis at the start of the CoheaHr project in the Lancet main journal. This publication, which demonstrated that HPV screening provides better protection against future cervical cancer than cytology screening, has been instrumental in health policy making and since its publication in early 2014, it has been cited about 500 times.

In the CoheaHr project, the data warehouse has been extended to include the longer term follow-up results of the Dutch, Swedish and Italian screening studies (the POBASCAM study, the Swedescreen and the NTCC study). Furthermore, using the trial samples, information has been collected collected on specific biomarkers (e.g. DNA methylation, p16/Ki67 dual staining).

The follow-up of Swedescreen has been udated to 13 years after recruitment, representing four screening rounds plus one year to account for delays in attendance and any diagnostic follow-up. Follow-up was conducted using registry based information from 1997. The long-term risk for development of CIN2+ and CIN3+ have been published (Elfstrom BMJ 2014, In J Cancer 2015) as well as genotype-specific risks (Smelov Int J Cancer 2015). The POBASCAM database was updated to 14 years of follow-up. Follow-up cytology and histology information was obtained by linking the POBASCAM study database to the Dutch, nationwide network and registry for cyto- and histopathology (PALGA). Long-term CIN2+, CIN3+ and cancer risks were analysed and published in 2016 (Dijkstra BMJ 2016). Important insights from the analyses of the updated data sets were: 1.the increased detection of HPV testing over cytology testing against CIN2+ decreased over time, indicating that HPV testing does not lead to substantial overdiagnosis but leads to earlier detection of CIN2+, 2. most cervical abnormalities in HPV-positive women were found in the first rounds and not in subsequent screening rounds, 3. a screening interval beyond five years may be considered in an HPV-based programme, but only for women that are HPV-negative at baseline. Women that are HPV-positive at baseline but have a negative triage testing result require surveillance or retesting after 3-5 years.

The collection of further follow-up data was more challenging for the NTCC study than for Swedescreen or POBASCAM. As in the first two rounds of the NTCC trial, data for follow-up were obtained from the local screening registries, supplemented by data on diagnoses of high-grade CIN and invasive cervical cancer obtained from cancer registries. These represent the diagnoses resulting from opportunistic screening activity and from clinical activity due to symptoms. As there is no nationwide Cancer Registry nor Screening Registry data need to be retrieved from local registries, so that an update is more difficult than an update of the Swedescreen and POBASCAM databases. We were able to retrieve data from 7 of the 9 centres as no follow-up was possible in Trento and Imola. This means that the updated database contains data of 81279 women out of 94370 originally enrolled. The database was updated to 9 years of follow-up.

Analyses of trials’ data.

1.Age effect
Pooled analyses of the age effect on the cumulative incidence of high-grade CIN and of cancer after a negative entry cytology or HPV test have been conducted. We showed that the risk of both CIN2/3 and invasive cancer were significantly lower after a negative HPV test than after a negative cytology with no evidence of an effect modification by age (Dijkstra BMJ 2016). However, while the risk of CIN2/3 significantly decreased with increasing age, the risk of invasive cancer was significantly higher in women aged 50+ years at enrolment than in younger women. This occurred after a normal cytology but also after a negative HPV test.
As such results were somewhat unexpected, extensive work has been done in order to provide an explanation. For this purpose there has been intensive interaction with WP5 (modelling, see above). In particular, different variants of the mathematical model corresponding to different possible explanations were run in order to evaluate which ones provided predictions consistent with observed data).

A similar pattern was predicted by models assuming that the sensitivity of cytology is very low for a subgroup of pre-cancers while in its absence models predicted a decrease of cancer risk. The mechanism would be mediated by an accumulation of pre-cancers repeatedly missed by cytology. This hypothesis could also explain the observed results after HPV negative tests because most study women enrolled in the HPV arm were previously screened by cytology and had just 1-2 screening rounds by HPV testing. Cancer risk in older women is expected to strongly decrease after repeated HPV screening rounds. The results have been presented at international conferences and a paper is in preparation.

2. Management of HPV positive women
An analysis of the effect of variable criteria of cytology interpretation for triaging HPV positive women in routine activity in Italy was published (Ronco et al. Eur. J.Cancer 2016). Variations in the proportion of women judged to have abnormal cytology among those HPV positive had little effect on the overall referral to colposcopy. This is plausibly true for any triage test. More sensitive triage tests should be exploited by prolonging intervals before repeat testing.

Cross sectional and longitudinal accuracy of combinations of cytology, genotyping and p16 overexpression have been estimated on NTCC data. Data were presented at the 2015 International papillomavirus conference. Accuracy of combinations of cytology and genotyping after two rounds of HPV screening have been estimated on POBASCAM data. Data were presented at the 2016 Eurogin conference (Veldhuijzen Int J Cancer 2017; Polman Int J Cancer 2018). Results suggest that high sensitivity can be achieved by combining cytology, p16 and genotyping. This could allow a long interval before repeat testing resulting in an overall referral rate to colposcopy similar to the current one. The positive predictive value of cytology in HPV-positive women was similar after one or two rounds of HPV-based screening whereas the positive predictive value of genotyping was lower after two rounds of HPV-based screening.

The possible use of viral methylation as a triage test for HPV positive women has been investigated within the NTCC cohort by studying its association with CIN present at the moment of testing of developing later. For this purpose consensus primers to allow simultaneous pyro-sequencing analysis for different HPV genotypes have been developed (published Gillio-Tos et al. PLOS One 2017). Preliminary analyses suggest a particularly strong association with the future occurrence of high-grade CIN. The role of microRNA (miRNA) present in cervical cellular scrapes for triaging HPV positive women was also investigated. Although a few miRNAs were found to be associated with high-grade CIN in a “discovery” phase this was not confirmed on a larger “confirmatory” sample. The role of host cell methylation was studied in the POBASCAM trial. A longitudinal analysis indicated that the long-term cancer risk after a negative methylation test was lower than the long-term cancer risk after negative cytology whereas the risks after a positive test were similar (de Strooper, Int J Cancer 2018).

3.Type-specific incidence, clearance and progression
A pooled analysis of the incidence of type-specific HPV infection based on the POBASCAM and NTCC data has been conducted and published (Veldhuijzen NJ et al. Cancer Epidemiol Biomarkers Prev. 2015). The Incidence of screen-detected infections was similar in POBASCAM (2.54%) and NTCC (2.74%). It decreased with age for all genotypes but continued to be detected at all ages. The age profile, changed by genotype suggesting that it also depends on biological factors.

An analysis of HPV infection clearance and progression to CIN was conducted on NTCC data and presented at the 2015 International Papillomavirus conference. In contrast with previous results, data suggested lower infection clearance at younger age. However at younger age apparent persistence could in fact be re-infection by the same types.

Meta-analyses and systematic reviews
An aim of WP6 was to update existing systematic reviews and to conduct new meta-analysis on clinical questions related to primary and secondary prevention of cervical cancer.
The following activities resulted in 29 peer-reviewed papers containing the results of the meta-analytical work, 5 Cochrane reviews, 2 PhD theses and 85 papers with derived results of COHEAHR related work.:

Update of previous and conduct of new meta-analyses on:
- triage of women with a positive screen test result (equivocal or mildly abnormal cervical cytology or high-risk HPV positive);
- evaluation of HPV tests that fulfil requirements for primary screening in cervical cancer screening;
- accuracy of HPV tests on vaginal self-specimens;
- efficacy of strategies where self-sampling kits are offered to under- or non-screened women;
- treatment of cervical precancer: prediction of therapeutic failure;
- reproductive health harm associated with treatment of cervical precancer;
- efficacy and safety of HPV vaccines:
- enhancement of European capacity in meta-analysis & systematic review
- development of new statistical procedures for meta-analysis.
The conclusions of the meta-analyses conducted are summarised below.

Triage of women with equivocal or mildly abnormal cytology.
High-risk HPV DNA is more accurate (more sensitive and similarly specific) compared to repeat cytology to detect underlying precancer (CIN2+ or CIN3+) in triage of women with atypical squamous cells of undetermined significance (ASC-US). However, in women with LSIL (low-grade squamous intraepithelial lesions), hrHPV DNA testing shows very poor specificity. Triage can be made more specific by testing for RNA of E6/E7 genes of 14 hrHPV types without significant loss in sensitivity. Overexpression of p16 is even more specific for excluding cervical precancer but is accompanied by a loss in sensitivity when compared to hrHPV DNA testing in both women with minor abnormal cervical cytology. When using p16/Ki67 immunocytochemistry, better specificity and no loss in sensitivity of CIN3+ is noted compared to hrHPV DNA testing in both women with ASC-US or LSIL. Testing for HPV 16/18 to triage women with minor abnormal cytology is poorly sensitive but may be useful as a second triage test after hrHPV testing, with direct referral if the woman is positive for HPV 16/18.

hrHPV testing in women with a cytological result of ASCH (atypical squamous cells where high-grade lesions cannot be excluded) is limited. The usual recommendation for referring women with ASC-H for colposcopy is not altered by a positive triage test, whatever test is used.

Given the high risk of underlying CIN2+/AIS+, women with atypical glandular cytology (AGC) should be referred directly to diagnostic work-up. However, hrHPV test results in combination with the age, may improve the diagnostic process by distinguishing the risk of cervical (pre-)cancer or adenocarcinoma in situ or worse.

Triage of hrHPV positive women
A review published in 2015 revealed that reflex cytology for hrHPV-positive women followed by another cervical cytology examination if reflex-cytology was negative and referral to colposcopy for women who are cytology-positive at any triage step resulted in a good compromise of sensitivity and specificity. This conclusion was based on the findings from trials conducted in the Netherlands and is adopted in the new Dutch screening programme. Another algorithm recommended in the Supplements to the 2nd Edition of the EU guidelines for QA in cervical cancer screening, consists to repeat hrHPV testing twelve months after negative reflex cytology and is applied in several Italian regions that have started HPV-based screening. Many alternative options of triage are being examined including partial or extended genotyping, p16/Ki67 immunocytochemistry, methylation of viral or host genes and other markers. No single step strategy provides sufficient safety to allow triage-negative to be released to routine screening. Strategies involving multiple visits often are accompanied with a considerable loss to follow-up. Baseline genotyping for HPV16/18 enables identifying more women at increased risk of CIN3+ but are accompanied with an increased burden of colposcopy referral. Pretest-posttest-probability plots are interesting tools to visualize useful triage strategies.

HPV testing on vaginal self-samples
Two meta-analyses on HPV testing on self-samples were conducted and updated. The first one demonstrated that high-risk HPV DNA tests on vaginal self-samples are similarly sensitive and specific for cervical intraepithelial neoplasia than on samples taken by a clinician under the condition that a clinically validated PCR is used for hrHPV DNA assay is used. hrHPV DNA assays based on a principle of signal amplification are less accurate on self-samples. mRNA HPV testing is less sensitive on self-samples than on clinician-samples. The second meta-analysis demonstrated that offering self-sampling kits is more effective that invitation/reminder letters to reach women who do not participate in the regular screening programme. Mail-to-all scenarios are more effective than opt-in scenarios.

Updated lists of hrHPV DNA assays validated for cervical cancer screening
Two hrHPV DNA assays are validated for primary screening through randomised trials demonstrating improved protection against cervical cancer compared to good quality cytology-based screening: HC2 and GP5+/6+ PCR-EIA. Six tests are validated through multiple cross-sectional validation studies: Abbott RT hrHPV test (3 studies); Cobas 4800 (3 studies); BD Onclarity (3 studies); PapillocCheck (2 studies); Anyplex II HPV hr (3 studies); Risk HPV assay (2 studies). Three assays were validated in just one study: Xpert HPV; genotyping for 13 hrHPV types with Linear Array, cobas-6800. One test was validated after adjustment of the test-positivity criterion: genotyping for 13 hrHPV types with EUROArray. Two tests are partially validated: RIATOL rtPCR (no inter-laboratory-reproducibility documented) and CERVISTA (conflicting results regarding specificity). Most of the validation studies followed the VALGENT protocol developed during COHEAHR.

Prediction of outcome after treatment of cervical precancer
Women treated for cervical precancer by local surgery are at increased risk of recurrent disease if the excision margins are positive for CIN. However, the margin status is not an accurate predictor of treatment failure. Post-treatment HPV testing predicts treatment failure with significantly higher sensitivity and similar specificity.

Obstetrical harm associated with excisional treatment of cervical precancer
Updated meta-analyses and preliminary results of an ongoing individual-patient data meta-analysis confirm that women who become pregnant after treatment of cervical precancer have an increased risk of preterm delivery. This risk increases with the depth of the excised cone.

Safety and efficacy of HPV vaccines
A Cochrane review confirmed high-certainty evidence that HPV vaccines protect against cervical precancer in adolescent girls and young women aged 15 to 26. The effect is higher for lesions associated with HPV16/18 than for lesions irrespective of HPV type. The effect is greater in those who are negative for hrHPV or HPV16/18 DNA at enrolment than those unselected for HPV DNA status. There is moderate-certainty evidence that HPV vaccines reduce CIN2+ in older women who are HPV16/18 negative, but not when they are unselected by HPV DNA status. The frequency of serious adverse events did not differ between vaccine and control arms. In women older than 25 years, more deaths were observed in the vaccine arms, but the difference was interpreted as not causally related to HPV vaccination. Increased risk of adverse pregnancy outcomes after HPV vaccination cannot be excluded, although the risk of miscarriage and termination are similar between trial arms. Long-term of follow-up is needed to monitor the impact on cervical cancer, occurrence of rare harms and pregnancy outcomes.

Increasing capacity in systematic review work; development of new statistical procedures for meta-analyses
Conclusions of the meta-analyses were used in modelling of cost-effectiveness analyses of combinations of HPV screening and vaccination. The capacity and skills in systematic reviewing were shared with other partners (other COHEAHR partners, Public Health/Cancer Institutes of NL, LU, DK, FR, IE, PO, CANCON, European Federation of Colposcopy, European Society of Gynaecological Oncology, WHO, CDC, NCI) and extended towards prevention and treatment of other HPV-related cancers.

Partner 6 developed the metaPROP macro which is a statistical procedure for pooling of binomial data in Stata and a framework for network meta-analysis of diagnostic test accuracy data. The PPP (pretest-posttest-probability) plot is a tool that allows visualise the utility of the application of a test in a certain field. It connects the pretest prevalence of disease with the post-test risk according to the test result. A test or combination of tests is useful when the results stratifies risk and determines further management which is visualised by coloured user-defined risk zones.

The Institut Català d’Oncologia (ICO) developed an online educational platform to facilitate healthcare professionals and service providers on cancer prevention, diagnosis and treatment updates and education needs. The platform is named E-Oncologia available at The website includes a large offer of online courses specialized in cancer, including a cervical cancer prevention program.
As part of the CoheaHr project, the previous edition of this cervical cancer course has been updated:
- New available data has been incorporated on HPV related burden, HPV vaccines (population impact, reduced number of doses and the newly marketed 9-valent vaccine) and new screening technologies.
- The course, already available in English, Spanish, and French, has been additionally translated into Portuguese (Brazil), Japanese, Russian and Chinese. Greek and Italian translations are currently underway at his final stage.
- The course has also been reformatted into a more user-friendly design.
Additionally, a new CoheaHr project supplementary module has been developed and incorporated into the course. This module contains a summarised rationale of the project and, for each research topic in the study, a selection of publications have been virtualised using the abstract and the most relevant figures and tables.

The updated course is available at The access is free but requires of registration so that the system keeps track of the student progress and can be provided with the certificate of attendance.
To ensure the dissemination of the project results, the supplemental module can also be accessed through the WHO/ICO HPV information center.

Potential Impact:
The CoheaHr project has taken a crucial step in the optimization of the delivery of health services by providing a comprehensive evidence base on the comparative effectiveness and cost-effectiveness of screening and vaccination programs for cervical cancer and other HPV-related diseases.

The results of CoheaHr will support policy makers and other decision makers at the national and EU level to prevent HPV-related disease and to improve the health of European Citizens. CoheaHr results have been disseminated by international peer-reviewed publications, international conferences and have been used to develop an e-learning module. The potential impact will be described in more detail below:

HPV self-sampling

A Dutch trial showed that in routine screening, HPV testing on self-collected specimens (HPV self-sampling) is as accurate as HPV testing on clinician-collected samples in routine screening when using a PCR-based HPV test and a brush collection device (Polman, Lancet Oncology 2019). The Dutch trial result corroborates with a CoheaHr meta-analysis (Arbyn, Lancet Oncology 2014) that indicates that HPV self-collection is appropriate when used in combination with HPV PCR testing. Women participating in the HPV self-sampling trial also indicated that they experienced less anxiety and pain and most of them preferred HPV self-sampling in the future.

Therefore, HPV self-sampling is 1. A promising method for reaching women in areas with low screening coverage or women that refuse conventional screening at the clinician’s office, and 2.An attractive, female-friendly alternative as a primary test in organized screening. Self-collection of specimen is also expected to be less costly than clinician-collection: the number of visits at the general practitioner is expected to decrease by about 90 percent because only HPV-positive women need to visit the general practitioner for triage testing.

In the Netherlands, HPV self-sampling is already being offered to women that do not respond within four months after being invited. The current results may support Dutch health authorities in taking a decision on how to collect cervical specimens in the near future. A new possibility is offer women a choice between HPV self-sampling and HPV testing on a clinician-collected specimen. Health authorities in other countries may use the Coheahr study results as well for their own setting because our clinical accuracy results seem generalizable to other populations. We think that the results are particularly useful for countries where the screening coverage is low.

The CoheaHr researchers have had regular meetings with national health decision makers to discuss their study results and its implications. Both the HPV self-sampling trial results as well as the meta-analyses have been disseminated at international conferences (IPV and Eurogin meetings) as well as in medical journals (Lancet Oncol 2014, 2019, BMJ 2018).

Screening after vaccination

Different countries use different vaccination strategies with respect to vaccination age, vaccine used and target group (girls-only or sex-neutral). While HPV vaccinated women are well (efficacy >90%) protected against cervical cancer, unvaccinated women can benefit from herd protection. By incorporating data from the CoheaHr vaccination studies into the HPV transmission and screening models, we found that a main benefit of sex-neutral vaccination is that it protects against HPV infections via herd effects. This means that HPV vaccination has an indirect effect on females who are not vaccinated and therefore screening may be reduced in unvaccinated women as well.
In some countries vaccinated women have already entered the screening program, while in other countries vaccinated cohorts will enter the screening program in the coming years. Policy makers will have to answer the question on how to screen vaccinated populations. Too frequent screening of HPV vaccinated women is costly, ineffective and, if anything, increases the adverse effects of screening (e.g. premature births). The Finnish randomized trial performed within CoheaHr on infrequent versus frequent screening is based on birth cohorts 1992-1995 vaccinated in 2007-2009, and will have complete follow-up of three screening rounds (at the ages of 22, 25 and 28) by 2023. This, together with a parallel observational study among females participating in different phases of the Swedish HPV vaccination and screening program, brings the urgently needed evidence for stopping repeated screening of HPV vaccinated women within the next five years. This is at least 10 years earlier as compared to a decision solely based on health registry data.
Evidence-based infrequent cervical screening will highly likely improve the quality of life of young adult women. The findings are generalizable to other countries or regions with an effective HPV vaccination program and an organized cervical screening program.

Vaccination of screen-eligible women

The results of the feasibility trial showed that there are no major barriers with respect to the acceptance of vaccination of adult women (25-45 years old). The vaccine acceptability results have been obtained in Europe, where the burden of cervical cancer is lower than in developing countries (lower perceived risk) due to the established cervical cancer screening programmes (alternative prevention strategies) and where there are strong anti-vaccine movements (negative publicity). Therefore, our results suggest that HPV vaccination will likely have the same or higher acceptability rates among adult women in other settings.

The vaccination of adult women, when combined with HPV screening, is a highly attractive cervical cancer prevention strategy (HPV-FASTER, Bosch Nature Review Clin Oncol 2015) in under-screened women. In theory, HPV screening of adult women in combination with an adequate diagnosis, treatment and follow-up of positive subjects will reduce the immediate future burden of disease. HPV vaccination helps to prevent incident infections and their potential associated disease burden over a longer-term timeframe. Therefore, the combination of both strategies would result in a higher negative predictive value of future disease, thus allowing for an extended interval between screening rounds if subsequent screening would still be required.

In population with a low screening coverage, either due to lack of compliance or lack of a suitable screening invitation system (i.e. many developing countries), this screen and vaccinate strategy offers maximum protection by means of a single intervention. Furthermore, in settings where women with a positive HPV test result are unlikely to comply with follow-up of abnormal results, vaccination with screen-and-treat strategies can also be explored.

This study partially explored two of the HPV-FASTER strategy uncertainties; i.e. logistical challenges (how to best integrate vaccine administration in already established screening efforts) and societal challenges (whether HPV vaccination is an accepted cervical cancer prevention strategy when offered to adult women).

The study results will be used to continue working on the feasibility of the HPV-FASTER strategy, a concept listed among the 30 innovative ideas to be pursued to address the Sustainable Development Goals in the coming decade (PATH 2015) by significantly improving cervical cancer prevention in the most vulnerable populations and thus reducing the existing social inequalities.

Next steps include the follow-up of vaccinated cohorts at catch-up campaigns in order to determine the subsequent risk of disease of post-vaccination and screening requirements in terms of number of lifetime visits.


The public health implications of the main modelling findings have been presented and discussed in several expert meetings both at national and international level. In particular, the potential implications of extended catch-up and sex-neutral coverage of HPV vaccination in settings with different levels of vaccination coverage and pre-vaccination HPV prevalence were discussed at the "Haut Conseil de la santé publique" of the French Ministry of Health (June 2015), with the Italian Group for cervical screening (May and November 2015), and at the 5th Helsinki HPV Workshop on Vaccinating Women and Men against Premature Death (January 2016). Overall our findings about vaccination program effectiveness and resilience have contributed to inform the current set of evidence on impact of catch-up and gender-neutral vaccination. In the summer of 2017, the Swedish Public Health Agency concluded its evaluation of HPV vaccination of males and has recommended that males should be included in the routine, school-based vaccination program of 10-12 year-olds. The results on the cost-effectiveness of sex-neutral vaccination at low vaccine prices as well as the development of tender-based vaccine prices over the last ten years have been presented at Eurogin 2017 and 2018. In a special session at Eurogin 2018 on prevention efforts in Central and Eastern European countries, cost-effectiveness of integrated vaccination and screening efforts and the development of tender-based prices in Central and Eastern Europe has been presented.

Similarly, the above-mentioned results were presented in 2015 at the World Health Organization (WHO) Headquarters and were reported in 2016 both at the GAVI vaccine alliance and at the meeting of the WHO’s Advisory Committee on Immunization and Vaccines-related Implementation Research in Geneva. Our findings were discussed in the context of workshops dedicated to global HPV vaccine introduction and to the definition of global HPV vaccine research priorities. Our reports contributed to inform the recommendations issued by the Strategic Advisory Group of Expert of the WHO and the subsequent WHO position paper on HPV vaccines published in May 2017.
The published overviews of existing vaccination and screening strategies in Europe conducted in conjunction with the European Union Screening Reports, coordinated by IARC, set a baseline for regular and standardized monitoring in Europe, where organized efforts to vaccinate against HPV and to screen women for cervical cancer have the potential to dramatically reduce the burden of HPV-related disease. The opportunity to implement comparable strategies for organization, monitoring, evaluation, and optimization of programs can significantly affect the health care resource usage and disease burden in Europe. We expect that the results of our modelling of cervical cancer screening among vaccinated and unvaccinated birth cohorts will also contribute to inform future comparable screening policies in several European countries. Manuscripts reporting the results of our modelling studies are in preparation. Preliminary communications have been reported in international workshops and conferences.

To ensure that public health institutions across Europe have access to transparent, validated, and well documented models to quantify the long-term medical, societal, and economic benefits of vaccination and screening of cervical cancer, we are developing open-source versions of the models using the CRAN/R language platform. A working prototype of the HPV transmission model is currently available, whereas the cervical cancer progression and screening model is in preparation.
The results obtained from the overview of existing vaccination and screening strategies in Europe, the development and validation of the models, model-based projections of the impact of cervical cancer prevention have been regularly presented at the International Papillomavirus Society, EUROGIN conferences, and IARC scientific council. Several manuscripts have been published in international peer-reviewed journals.

Trial data

We published the cancer incidences observed in the pooled database of European cervical cancer screening trials, that includes a standardised format for registering data and the actual interventions made. We published the analysis at the start of the CoheaHr project (Ronco, Lancet 2014). This publication, which demonstrated that HPV screening provides better protection against future cervical cancer than cytology screening, has been instrumental in health policy making and since its publication in early 2014, it has been cited about 500 times. Health decision makers in several countries have used this publication to support HPV-based screening as an alternative to cytology based screening.

The updated follow-up of the screening trials has been used to support the choice of screening intervals in HPV-based programs and to support the idea of a screening interval beyond five years in HPV-negative women (5-10 years) in the new Dutch screening program. These results were disseminated in peer-reviewed publications (Elfstrom BMJ 2014, Dijkstra BMJ 2016) and presented on Eurogin and IPV conferences.

Several triage strategies for HPV-positive women were evaluated by means of the Swedescreen, POBASCAM and NTCC data. The results of those evaluations have been used by health decision makers to decide on the management of HPV-positive results.

The next step in HPV-based screening is to replace the one-size-fits-all protocols, as currently used in European countries, by a risk-based screening protocol. In the near future, women with greatly differing disease risks can be identified on the basis of screening history, demographics and vaccination status. This offers the opportunity to optimize screening by defining risk-based screening protocols, in which women are invited for screening based on their current CIN3 and cancer risk.


The systematic reviews conducted in the framework of COHEAHR were pivotal in the development of the Supplements of second edition of the European Guidelines for Quality Assurance for Cervical Cancer Screening (2015) , recommending the switch from cytology to HPV-based primary screening in European member states. Since then several countries in Europe and other continents have started or are planning to introduce cervical cancer screening using a validated high-risk HPV assay as primary screening test. The capacity of the Unit Cancer Epidemiology (Belgian Cancer Centre), built up through CoheaHr, is currently requested from several EU member states involved in the adaptation of cervical cancer prevention guidelines.
The continuously updated list of hrHPV DNA tests that fulfil minimal equivalence requirements for primary cervical cancer screening are currently widely accepted among decision makers and stakeholders who have the responsibility to define new screening policies. The VALGENT protocol developed during CoheaHR for validation of hrHPV assays has become a major framework for HPV tests validation in Europe and elsewhere.

The systematic reviews, summarizing evidence regarding HPV testing on self-samples was influential in the set-up of a large number of new diagnostic test accuracy studies and participation trials (see also section on HPV self-sampling). The availability of self-sampling kits may have a substantial influence on population coverage among hard-to-reach women but may also become a preferred method of sample collection among women that participate already in the screening program.
The meta-analysis on predictors of success or failure of cervical precancer treatment were decisive in adapting quality indicators for colposcopy practice developed by the European Federation of Colposcopy. The updated meta-analyses on obstetrical post-treatment complications and on therapeutic failure yielded recommendation for more conservative and aggressive procedures, respectively. Not even one study up hereto was conducted assessing jointly the obstetrical and oncological safety of cervical precancer treatment. Linkage studies should be set up in countries with good data joining data from treatment centres, cancer/pathology registries and maternity files.

The Cochrane review on safety and efficacy of HPV vaccines was widely covered in the popular press. However, the review also attracted criticism of anti-vaccine groups also claiming their own principles of evidence. This new wave of activism against HPV vaccination and HPV screening may undermine the confidence of the target population resulting in reduced population coverage and hesitation of decision makers towards implementation of improved methods of cervical cancer prevention. Public health authorities and research funders should be aware of the need of high-quality and independent evidence assessment and provide sufficient resources for it. However, mechanisms of private-public collaborations could be considered as complements to the shrinking resources of public funding for implementation research and systematic reviews.

Web-based dissemination activities

The e-learning tool will provide healthcare professionals and service provider worldwide with up to date knowledge on cervical cancer prevention. The course is available in many languages and contains a special CoheaHr module.
First in collaboration with World Health Organization (WHO) and now with International Agency for Research on Cancer (IARC), the Catalan Institute of Oncology (ICO)/IARC Information Centre on HPV and Cancer compiles and processes country-specific information on: cervical cancer and other HPV-related disease burden and associated risk factors, prevention strategies, screening activities, and immunization programmes. The main aim of the centre is to facilitate global, regional and country-specific decisions on current and novel options for the prevention of cervical cancer and other HPV-related diseases.

The HPV Information Centre has become an international benchmark for the provision of data on the epidemiology of HPV and related cancers and their prevention. As part of the activities, the centre prepares systematic reviews and meta-analyses of the relevant literature, presented at country and regional level. The centre has developed a highly efficient web-based communication between researchers, medical communities, public health professionals and decision-makers on a worldwide scale:

The website contains (i) summarised information on HPV prevention, (ii) a user-friendly data query tool that allows to retrieve the most up-to-date results stored in the local server, (iii) access to regional and scientific reports using the generated data and (iv) links to useful HPV-related resources
As part of the CoheaHr project, the website has been modified to include a new section about the project. This section contains the rationale for the study and its objectives, the methodology used (i.e. work packages structure and the planned worked on each of them), the study results reports and the list of the project publications.

Additionally, as part of the project, an online course containing a summary of some selected publications from each research topic was developed. This course has also been uploaded to the website to facilitate the access to the main project results and its adequate dissemination and exploitation.


Results of CoheaHr have been disseminated at national and international meetings and conferences. A special CoheaHr session has been held at Eurogin 2015 (Seville, Spain), Eurogin 2016 (Salzburg, Austria), Eurogin 2017 (Amsterdam, the Netherlands) and Eurogin 2018 (Lisbon, Portugal) with presentations of all key results. Those sessions were attended by many scientists and stakeholders from several European countries. Furthermore, the results were reported to panels of public health experts, and submitted for publication to peer-reviewed journals. A complete list of publications from the CoheaHr consortium, as well as additional information on the project and the consortium partners, can be found at:

List of Websites:
Public website:
Contact details:
Prof J. (Hans) Berkhof