44-48

Boundaries should be more precisely defined. What are the researches that fall outside the boundaries of informed consent? Examples should be provided.  Are the boundaries depending on what the patients are signing or are they “global”, i.e.: with an objective of advancement of science or public health? More comments on this point below (182-183)

67-71

I am not sure that those who generate CT data in the first place (i.e. pharma industry) can be considered as the highest standard of transparency! Otherwise, all their analysis would already be in the public domain… Ok for transparency but saying that the public requesters should be as transparent as pharma industry might be laughable.

77

Prospective: It can be understood that it would be a huge work to make this procedure retroactive. However, some clinical study reports (CSR) or other documents from the Common Technical document (CTD) have already been requested and sent to requesters. These CSR and CTD parts are thus already in an appropriate format for public domain. They should be made available as it is planned for category 2 documents (see lines 150-154). This could show that EMA has recently become more “transparency-minded” and which are the products experts are making researches on.

179

How will the requester be assured that his/her name will not be known from a third party outside EMA? What process will be put in place?

182-183

What is in the interest of public health and what is not? Is searching for a bias in the interest of public health? Is improving knowledge on patients’ evolution in the interest of public health? Details on what is not in the interest of public health should be proposed. The decision that an analysis is in the interest of public health should be based on objective points previously described in the procedure.

Spirit of informed consent: This is not acceptable as such. If, for example, the following sentences are put in the informed consent: “to guarantee confidentiality, your personal data will only be available to the sponsor, medical team in charge of you and health authorities” or “your personal data will only be used for the purpose of the study”, this would mean that data should not be made public, even with controlled access. Either a mandatory sentence should be put in any informed consent stating that de-identified data can be used for public health interest research by people not involved in the study or it should be clearly stated in the procedure that informed consent cannot limit access to the data.

185-187

What does “exhaustive and detailed” mean? It is an open door to refusal of the data for subjective reasons.

Requesters should only have to tick a case: for example meta-analysis, patient description, subgroup analysis, re-analysis, potential bias evaluation, other exploratory analysis, other (+ written precision). Ticking one of these cases could be mandatory but should not limit data availability.

191-192

Not acceptable (as above 182-183). If, for example, the following sentences are put in the informed consent: “to guarantee confidentiality, your personal data will only be available to the sponsor, medical team in charge of you and health authorities” or “your personal data will only be used for the purpose of the study”, this would mean that data should not be made public, even with controlled access. Either a mandatory sentence should be put in any informed consent stating that de-identified data can be used for public health interest research by people not involved in the study or it should be clearly stated in the procedure that informed consent cannot limit access to the data.

194

“Not share, in any way or format”: Some evaluations might need the description of few patients: for example description of specific subtypes of patients, lost for follow up, very good or very bad responders or any other specific subgroups of patients… Some publications are providing readers with a list of patients with some characteristics. Does this sentence forbid requesters to provide a list of patients in their publication? It should not. A sentence such as “no more than 30 patients and no more than 10 items could be published” could be proposed.

198

Under which circumstances is approval from ethics committee needed? What about approval from organism such as CNIL in France? A global approval from CNIL or other agencies dealing with e-databases should be obtained for data de-identified by EMA. This approval is to be obtained for any clinical trial prior start. Is this approval valid for requesters?

201

The agreement paper should contain the limitations described from line 222 to 231

205

CT data access should be destroyed after a delay following the publication to allow the requester to answer to any discussion about the publication. Once the publication is made and the information put on the EMA website, a 6-month delay could be reasonable before data are destroyed.

226

In case of combined analysis of more than one CT data, the one year delay should start after the last CT data are obtained (ex: meta-analysis)

229

Why? By definition, EMA has the same set of data and can perform the analysis to solve any public health issue or need, doesn’t it? Of course EMA can ask the requester to know if he/she can provide information but it is not a reason to disclose the requester’s name. EMA can ask the requester to disclose his/her name but the name of the requester should not be made public by EMA for this reason.

242-247

The CT data should be available in a format which is readable with non-proprietary software.