Alessandro Curto, Katelijne van de Vooren, Livio Garattini
Centre for Health Economics. 'Mario Negri' Institute for Pharmacological Research, Italy

Following the Authors’ response,
First of all we are pleased they addressed our commentaries, like it should happen in any debate on a scientific journal, whether open access or not.
Now, we still wonder why the Authors claimed in the Introduction that “in March 2008, a national HPV immunisation programme that uses the quadrivalent vaccine and targets eleven year-old girls was initiated”, being the bivalent vaccine used in Italy too.
This is a clear mistake hard to justify since i) all the five Authors are Italians, ii) all of them recently co-signed an article on Medical Care supported by the quadrivalent manufacturer, [1] and iii) one of them was a manager of that company until three years ago.
Therefore, we are afraid the Authors themselves raised the concern of “a spurious relationship between the objectives of the research and the quadrivalent vaccine”. Accordingly, we think they should change that sentence in the article, by introducing an errata corrige, in order to cancel any doubt on this sensitive issue.
Here below, we react point by point to the specific commentaries addressed by the Authors, following their same numbered list.

1. Lacking Italian epidemiological data, it is always arguable to produce a specific estimate, in particular using data from a very different country like the USA. Moreover, it is pretty obvious to use Italian unit costs in order to monetize them, otherwise it would have been a real nonsense. Then, it is worth noting that the abstract is misleading on this issue in Methods, claiming that secondary Italian sources were used for all the nine conditions (“For each of the nine conditions, we used available Italian secondary data to estimate the lifetime cost per case”), which is not the case for RRP, one disease that makes up a relevant part of the costs.

2. We just underlined that the authors mainly used old data from a WHO/ICO report that refers to 1998-2002 for seven out of the nine diseases, so we still wonder if these data can be considered updated for the Italian setting.

3. We acknowledge the Authors’ explanation on why the number mentioned in reference 26 doubled in the text. However, we are afraid the Authors did not understand our criticism on the over-estimation of the incident number of head and neck cancer cases (and consequently the annual cost) attributable to HPV. We think there are two major problems in their computation.
• The evidence of an association with HPV exists only with oral and pharyngeal cancers since HPV does not increase the risk of other head and neck cancers, including laryngeal, and nasopharyngeal cancer. [2] Therefore, we think that, among head and neck cancers, only oral and pharyngeal cancers should be taken into account in this analysis.
• It is arguable to consider the “HPV prevalence among incident cases” as the “fraction of cases attributable to HPV”. This assumption may be accepted as an approximation for cervical cancer, but seems unacceptable for head and neck cancers. Actually, the HPV detection in a certain proportion of incident cases of nasopharyngeal or laryngeal cancers does not necessarily imply the causal role of HPV in these cancers.

4. Although we recognize the two sentences are not inconsistent, here we meant that tariffs hardly "represent the actual direct costs" since the Italian NHS has always covered the real public hospital expenses, even though real costs are higher than each hospital "turnover" estimated by applying tariffs. Moreover, the tariffs uneven revision, acknowledged by the Authors too (see point 7), is one more reason to consider tariffs just a “proxy” of the NHS real costs.

5. We acknowledge the Authors’ explanation on the viewpoint adopted, although this was not written down in the text.

6. We acknowledge the Authors’ explanation, we did not want to be ironic on this issue. We were just surprised to find out so many sentences on this pretty obvious issue - e.g., in addition to the two sentences quoted by the Authors in their answer, there is one more in Conclusions, where the discounting leads to a very negligible impact on the final figure (around forty euros apparently),so it is hard to understand why this was done (“The DRG 410 sets the cost of one cycle of chemotherapy delivered in day hospital at €465.2 (2011 inflation-adjusted value). Hence, the overall cost for three cycles discounted at a rate of 3% would be equal to €1,353.7 per patient”). We think these frequent explanations are probably justified by the very old data used for many estimates, which obliged the Authors to adjust them for inflation before discounting future costs.

7. Although we acknowledge the Authors’ explanation, we should remind them that authors (and not reviewers) take responsibility of what is written in an article. This arguable choice led to overestimate the costs as a matter of fact, confirming at the same time that tariffs hardly reflect the NHS real costs (see point 4).

8. Here we referred to the two columns before the one labelled b, which should be the sum of them. Actually, this is not the case in the rows mentioned by the Authors, so we will try to explain better our point. The columns ‘HPV 16, 18 fraction’ and ‘HPV 6, 11 fraction’ are not filled in for row q, r, s and only partially for row u, v, z, y, while only in row e ‘0.0%’ is filled in. Row q, column b shows a sum of 64.3% attributable to HPV 6,11,16,18 fraction, although this percentage includes also genotypes 31/33 according to the reference (26), thus leading to an overestimate of the costs apparently. Row u, column b shows a sum of 51,9%, while the ‘HPV 16, 18 fraction’ is 50% and no number is filled in for the ‘HPV 6, 11 fraction’.

9. We still think the (here below) formula, which regards only epidemiological data (and not costs), does not add any value, projecting data from 2011 to 2006 .Moreover, it is unclear why the Authors did not use the 2006 Italian population to estimate the new cases in 2006 for sake of homogeneity.

∑ number new cases 2006 = crude incidence rate/100,000 x resident Italian population 2011/100,000

10. We acknowledge the “independent oncologist” was mentioned in Methods too, although the two quotations do not coincide in the description of the role she played in the study (see below). Moreover, we still wonder how only one, anonymous, so called “independent oncologist” should be considered a “warrant” for reviewing a lot of inputs of this study. If she played an important role in the methodology, why not qualifying and including her as a further Author?
a) “Treatment paradigms for non cervical cancers (anal, head and neck, vaginal, vulvar, and penile) were based on the recommendations of generally accepted Italian Medical guidelines (16–18) and were reviewed by an independent Italian Medical Oncologist.” (Methods, Life Time Cost per case, page 6)
b) “All the inputs that were chosen to inform the cost analysis were reviewed by an independent Medical Oncologist.” (Results, Non Cervical Malignancies, page 12)
To conclude, according to all the evidenced concerns, we still think the article is mainly based on old data and arguable assumptions, so it can hardly provide plausible information on the present burden of these illnesses in the Italian setting.

To conclude, according to all the evidenced concerns, we still think the article is mainly based on old data and arguable assumptions, so it can hardly provide plausible information on the present burden of these illnesses in the Italian setting.


References.
[1] Favato G, Baio G, Capone A, Marcellusi A, Costa S, Garganese G, Picardo M, Drummond M, Jonsson B, Scambia G, Zweifel P, and Mennini F. Novel Health Economic Evaluation of a Vaccination Strategy to Prevent HPV-related Diseases.The BEST Study. Medical Care, Volume 50, Number 12, December 2012
[2] Joseph AW, D'Souza G. Epidemiology of human papillomavirus-related head and neck cancer. Otolaryngol Clin North Am. August;45(4):739-64. 2012