EMA Provides Long-Awaited Guidance Under the Policy on Publication of Clinical Data

The conduct of non-US clinical trials by North American companies continues to grow at an ever-increasing pace. A significant challenge to US-based companies is remaining current on the laws and regulations in countries in which their trials are being conducted. One such important and highly regulated consideration in any country, including the United States, is the ability to provide meaningful and informative publication of clinical trial results while, at the same time, ensuring privacy rights of the participants in the trial. This article summarizes a portion of the guidance recently provided by the European Medicines Agency (“EMA”) on this topic.

In response to growing demands from stakeholders for increased transparency in clinical trial reporting, in 2014, the EMA published the “European Medicines Agency policy on publication of clinical data for medicinal products for human use” (the “Policy”).[1] EMA’s stated purpose for the Policy is “to protect and foster public health,” with transparency being key to the delivery of service to patients and society in general.[2] Through implementation of the Policy, EMA seeks to achieve transparency by proactively making clinical data available to enable both public scrutiny of data and application of new knowledge in future research.[3] With respect to publication of clinical data the Policy is intended to:

  • Avoid duplication of clinical trials, foster innovation, and encourage development of new medicines;
  • Build public trust and confidence in EMA’s decision-making processes; and
  • Help academics and researchers to reassess clinical data.[4]

EMA believes that the Policy will level the playing field by enabling all clinical researchers to benefit from past successes and failures of others and allowing wider use of clinical data by the scientific community in its efforts to gain new knowledge. The stated principles considered by EMA in its development of the Policy are protection of personal data, protection of commercial confidential information, protection of EMA’s and the European Commission’s decision-making processes, and ensuring investment in pharmaceutical research and development in the future. The Policy applies to all clinical reports contained in initial marketing-authorization applications submitted on or after January 1, 2015, and to all applications requesting to vary an existing marketing authorization for an extension of indication or a line extension submitted on or after July 1, 2015. Both researchers who generate initial clinical data and those who create secondary analysis of clinical data are subject to the Policy.


While the Policy provides detailed requirements regarding clinical data publication, it did not include specific implementation guidance as to requirements to ensure company compliance. The long-awaited implementation guidance, known as “External guidance on the implementation of the European Medicines Agency policy on the publication of clinical data for medicinal products for human use,” was published earlier this year in March 2016 (the “Guidance”) and provides specific information regarding:

  • Procedural aspects related to the submission of clinical reports;
  • How to identify and redact commercially confidential information in clinical reports; and
  • Anonymising clinical reports.[5]

This article focuses on the guidance specific to ensuring adequate personal data protection through anonymous reporting.


Chapter 3 of the Guidance addresses anonymisation of clinical reports for the purpose of publication in accordance with EMA Policy 0070 and provides specific direction for achieving anonymous reporting that adequately protects personal data of patients.[6] Anonymisation refers to the process of converting data into a form which does not identify an individual and makes it unlikely that identification of the individual can occur. This chapter of the Guidance is not a mandate to follow a specific method of reporting; rather, it is intended to highlight for stakeholders “the process to be followed to ensure that clinical reports submitted to EMA for publication are rendered anonymous prior to publication.”[7] The stakeholders are referred to by EMA as “Applicant/MAH” which is defined as the persons or organizations submitting clinical reports to EMA in support of applications and the persons or organizations who own intellectual property rights in the clinical reports. Chapter 3 provides:

  • general considerations for achieving anonymisation;
  • advice on application of the general considerations to clinical trial reporting in accordance with the Policy;
  • recommendations on how to best achieve anonymisation; and
  • advice on redaction of personal data of investigators, pharmaceutical company staff, and the Applicant/MAH.


The first general consideration on how to ensure effective anonymisation outlined in Chapter 3 relates to the variance in the level of risk associated with the possibility of re-identification depending on the context of the specific disclosure. Recognizing that in releases of public data to the world at large there are no effective controls that can be implemented, EMA states that the risk of re-identification in those public data releases must be very low. This low risk tolerance by EMA applies to the context of public data release, as the clinical reports published pursuant to the Guidance will be available online and capable of being downloaded by any registered user.

The second general consideration addresses the concept of anonymisation to remove personal data. Article 2(a) of Regulation (EC) No 45/2001 of the European Parliament and of the Council of 18 December 2000 defines personal data as “any information relating to an identified or identifiable natural person (‘data subject’); an identifiable person is one who can be identified, directly or indirectly, in particular by reference to an identification number or to one or more factors specific to his/her physical, physiological, mental, economic, cultural or social identity.”[8] Clinical reports contained in applications submitted to EMA may contain individual patient information and, therefore, cannot be considered anonymised. The Policy requires the Applicant/MAH to submit an anonymised report for publication which will be separate and distinct from the scientific evaluation report that accompanies the application.[9] The publishable report must be a copy of the scientific evaluation report from which sufficient information has been transformed or removed so that individuals can no longer be identified. The Applicant/MAH is reminded to keep in mind the impact of removal on the scientific usefulness of the report. In addition, the Applicant/MAH is advised to give careful consideration to the complexity of anonymisation in reports of rare disease and small population trials based on the low number of trial participants.

To qualify as anonymised, the personal data must be processed in a manner that ensures the data cannot be used to identify the person through use of “all means likely reasonably to be used by either the controller or a third party.”[10] The Guidance offers two options. A process that prevents (i) the possibility of singling out an individual, (ii) ability to link records to an individual, and (iii) information being inferred to concern a specific individual, will be considered to meet this threshold and result in the data being anonymous. If all three criteria are not met, the second option is a determination of whether the data has been anonymised based on an evaluation of the identification risks.[11]

The EMA points out that lack of ability to reverse the methodology or technique used to anonymise is also important. Pseudonymisation is described in the Guidance as a security measure; however, it is not considered to be a measure that is adequate to anonymise because indirect identification of the person remains likely.[12] Pseudonymisation must be combined with additional measures, for example, removal and generalization of attributes, deletion of original data, or aggregation of the data at a very high level, in order to effectively anonymise the dataset.

Finally, the Applicant/MAH is instructed to consider future advances in technology that may allow identification. A data controller is required to continuously monitor advances in re-identification techniques and reassess the risk as appropriate.[13]


The goal of the Guidance is to assist the Applicant/MAH in striking the right balance between maximizing scientifically useful information for the benefit of the public and achieving effective anonymisation for the benefit of the individual. The Guidance does not mandate any specific methodology. The choice of methodology and technique is left to the company based on the ultimate purpose of the report and the company’s choice of the two options for anonymisation discussed above. EMA acknowledges that, initially, the Applicant/MAH is likely to use a reactive data anonymisation process to anonymise data retrospectively after submission of the clinical report (e.g., redaction). However, as the Applicant/MAH becomes more experienced, a transition should be made to proactive data anonymisation to maximize the clinical usefulness of published data.

The EMA’s view is that it is unlikely that a clinical report that maximizes usefulness of the published data will be able to meet the three criteria in the first anonymisation option described earlier. This view is based on recognition that the ability to link multiple records of the same trial participant within the report increases the ability to understand the safety and efficacy profile of the trial product. Inference is also important and the EMA advises that the potential impact of inferred data on the trial participants should be emphasized. Therefore, the EMA suggests that the Applicant/MAH will likely need to choose the option to perform a thorough evaluation of the risk of re-identification to achieve anonymisation.

The Guidance discusses several techniques in great detail, each of which has its own strengths and weaknesses. The Applicant/MAH must identify the most effective technique or combination of techniques based on the specific clinical report and the specific clinical trial data. Detailed discussion of the information and analysis of each of the techniques provided in the Guidance is beyond the scope of this article. However, a brief summary of the techniques is provided.

The simplest method is masking, which ensures that redacted information is irreversibly blocked from the report. EMA recommends masking of pre-specified variables, but not entire sections of the report. EMA also recommends randomisation, which is a series of techniques that remove the link between the data and the trial participant. Randomisation techniques recommended by EMA include noise addition (addition of random data to original microdata) and permutation (changing the order of data arranged in a particular order). Further, generalisation is also a family of techniques that dilutes the attributes of the data by modification of the respective scale or order of magnitude. Recommended generalisation techniques include aggregation, in which a value is replaced by a range, and k-anonymity, which groups a trial participant with other trial participants within the same range.

Applicants/MAHs are reminded that scientific review of the data and marketing authorisation assessment must be completed before anonymisation of the data. The Applicant/MAH’s primary focus should be to ensure that “the risk of re-identification is acceptably low and in line with requirements for public disclosure and that the data transformation resulting from the applied anonymisation techniques will not lead to a different interpretation of the study results.”[14]

EMA recommends an anonymisation process that includes the following steps:

  • Identify and remove or protect both direct identifiers and quasi(indirect) identifiers;
  • Identify possible adversaries and likely attackers on the data and evaluate the re-identification risk associated with each;
  • Ensure the anonymised results will produce analysis by third parties similar to the analysis based on the original clinical report prior to anonymisation (the goal is a balance between an acceptable low risk of re-identification and high level of data utility);
  • Determine the risk of re-identification threshold and evaluate the actual risk of re-identification;
  • Identify the most appropriate technique and describe how the risk of re-identification is reduced by this technique; and
  • Document both the process chosen and implemented and the rationale for the choice; specify the option chosen by the Applicant/MAH to achieve anonymity.[15]


The identity of the sponsor and the coordinating investigator of the clinical study must be published, with their contact information and signature being redacted. Personal data of other clinical trial staff will not be published. Further, research site number, name, and the investigator at each research site are not published based on the risk that publication of this information may facilitate identification of trial participants.


Through issuance of the Guidance, EMA is providing instruction on the advancement of the operational implementation of the Policy. The Guidance informs companies of what is required moving forward with respect to publication of clinical trial data. As US-based companies become more active in clinical trials governed by the EMA, knowledge of and compliance with both the Policy and Guidance are crucial. As stated above, an analysis of the entire ninety-one page Guidance document is beyond the scope of one article. Therefore, it is important that any company involved in or considering participation in clinical trial conduct or participation in Europe should ensure its knowledge and understanding of the provisions of the Policy and Guidance.

[1] European Medicines Agency, Policy on publication of clinical data for medicinal products for human use (2014), available at http://www.ema.europa.eu/docs/en_GB/document_library/Other/2014/10/WC500174796.pdf.

[2] Id. at 1.

[3] Id. at 4.

[4] European Medicines Agency, Clinical data publication, available at http://www.ema.europa.eu/ema/index.jsp?curl=pages/special_topics/general/general_content_000555.jsp&mid=WC0b01ac05809f363e.

[5] European Medicines Agency, External guidance on the implementation of the European Medicines Agency policy on the publication of clinical data for medicinal products for human use, 31-43 (2016), available at http://www.ema.europa.eu/docs/en_GB/document_library/Regulatory_and_procedural_guideline/2016/03/WC500202621.pdf; European Medicines Agency, Clinical data publication.

[6] European Medicines Agency, External guidance, at 31-42.

[7] Id. at 32.

[8] Official Journal of the European Communities, “Regulation (EC) No. 45/2001 of the European Parliament and of the Council of 18 December 2000 on the protection of individuals with regard to the processing of personal data by the Community institutions and bodies and on the free movement of such data,” Article 2(a), at L 8/4 (2001), available at http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2001:008:0001:0022:en:PDF.

[9] European Medicines Agency, Policy on publication, at 4-7.

[10] Official Journal of the European Communities, “Regulation (EC) No. 45/2001, at L8/4; Official Journal of the European Communities “Directive 95/46/EC of the European Parliament and of the Council of 24 October 1995 on the protection of individuals with regard to the processing of personal data and on the free movement of such data”, L 281, § 26 (1995), available at http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=CELEX:31995L0046:en:HTM.

[11] Article 29 Data Protection Working Party: The Working Party on the Protection of Individuals with Regard to the Processing of Personal Data, Opinion 05/2014 on Anonymisation Techniques (2014), available at http://ec.europa.eu/justice/dataprotection/article-29/documentation/opinion-recommendation/files/2014/wp216_en.pdf.

[12] Pseudonymisation refers to replacement of one attribute in a record by another.

[13] Opinion 05/2014.

[14] European Medicines Agency, External guidance, at 38.

[15] European Medicines Agency, External guidance, at 41.