Risk-Based Clinical Data Anonymization, EMA Policy 0070, and Health Canada’s PRCI: A Comprehensive Overview


In the era of digital health, the importance of clinical data cannot be overstated. However, the need for data privacy and protection is equally paramount. This has led to the development of risk-based clinical data anonymization strategies, regulatory policies like the European Medicines Agency’s (EMA) Clinical Data Publication (CDP) Policy 0070, and initiatives like Health Canada’s Public Release of Clinical Information (PRCI).

Risk-Based Clinical Data Anonymization

Risk-based clinical data anonymization is a strategy that measures the probability of re-identifying individuals (in this case, subjects that have participated in a clinical trial) through indirectly-identifying pieces of information. This probability is then reduced through various data transformations, such as offsetting dates, generalizing disease classifications or demographic values, or removing outlier values. The goal is to balance the need for data utility and the requirement for privacy.

Why EMA and Health Canada Prefer Risk-Based Anonymization

Balance: Risk-based anonymization allows for a more nuanced approach that balances privacy protection with the need for transparency and research. Traditional methods often remove too much data, hindering its usefulness for research purposes.

Minimizing Data Loss: By assessing the risk of re-identification for each data attribute, risk-based approaches can retain more valuable information while still protecting privacy. This allows for more comprehensive analysis and better insights.

Adaptability: The risk of re-identification can vary depending on the context and available information. Risk-based methods can adapt to these changing factors, ensuring appropriate protection in different scenarios.

Compared to other techniques, risk-based anonymization provides a more sophisticated and balanced approach to protecting privacy while enabling valuable research and data sharing in the life sciences industry. This aligns with the goals of regulators like EMA and Health Canada to promote public health and transparency while upholding individual privacy rights.

Both EMA and Health Canada have specific guidelines and regulations outlining their expectations for risk-based anonymization. This ensures consistency and accountability.

Key Considerations of Risk Based Anonymization

Risk Threshold

The risk threshold in the context of clinical data anonymization is defined as the minimum amount of de-identification that must be applied to a dataset for it to be considered de-identified.

In more practical terms, it refers to the probability of correctly assigning an identity to a participant (or clinical trial subject) described in the clinical reports. This is also referred to as the probability of re-identification.

For instance, both the European Medicines Agency (EMA) and Health Canada have set an acceptable probability threshold at 0.09. This means that the likelihood of re-identifying an individual from the anonymized data should be less than 9 in 100 for the data to be considered sufficiently anonymized.

The number of data attributes in the dataset requiring anonymization depends on the dataset’s risk score. Higher risk scores mean more fields must be anonymized. The goal is to ensure that the probability of re-identification is very small, thereby protecting the privacy of individuals while still allowing the data to be useful for research purposes.

The risk threshold in clinical data anonymization is determined based on several factors:

  1. Data Disclosure Precedents and Industry Benchmarks: The risk threshold is often set based on historical data disclosure precedents and industry benchmarks.
  2. Regulatory Guidance: Regulatory authorities such as the European Medicines Agency (EMA) and Health Canada provide guidance on acceptable risk thresholds.
  3. Risk Assessment: Anonymization requires a risk assessment to a predetermined threshold (often 0.09) to determine the probability of re-identification of a clinical trial subject.
  4. Dataset Characteristics: The number of data attributes in the dataset requiring anonymization depends on the dataset’s risk score. Higher risk scores mean more attributes must be anonymized.
  5. Sources of Re-identification Risk: Factors such as the number of participants, whether the trial is in a rare disease, subjective assessment of potential socioeconomic harm to patients if there is re-identification, and the perceived re-identification risk of certain pieces of information (whether they would be knowable by potential adversaries) are considered.

Determining the risk threshold is a complex process that involves considering various factors, including industry standards, regulatory guidance, and the specific characteristics and risks associated with the dataset.

Clinical Data Utility

Data utility in the context of clinical data anonymization refers to the usefulness of the data after it has been anonymized. The goal of risk-based anonymization is to protect the privacy of individuals in a quantifiable manner, but it’s equally important to ensure that the anonymized data remains useful for research purposes.

Preserving data utility during the anonymization process involves quantitative measurements at the document/data level and a well-defined and precise implementation of the selected rules to prevent over-redaction or over-anonymization.

For instance, pseudonymization, which replaces identifiers with a pseudonym, retains more data utility than anonymization, which may involve redacting or masking identifiers. This is because pseudonymization allows for meaningful secondary analyses and follow-on research while maintaining patient confidentiality.

In summary, data utility is a critical aspect of data anonymization. It ensures that the anonymized data can still provide valuable insights and contribute to scientific research, public health, and other secondary purposes.

Methods of Risk Based Clinical Data Anonymization

Clinical data anonymization involves various techniques to ensure the privacy of individuals while maintaining the utility of the data for research purposes. Here are some commonly used methods:

  1. Generalization: Specific values are categorized into groups or ranges. For example, exact ages might be replaced with age groups, and countries might be grouped into continents.
  2. Suppression or Redaction: This involves removing or redacting sensitive attributes entirely.
  3. Masking: Parts of the data are replaced with symbols like (*, $, #).
  4. Date Offsetting: This involves altering an identifiable date related to an individual and applying an alternative or random date throughout the data or document(s). To maintain usefulness of the data, offset dates maintain the same duration between events as compared to the original dates.
  5. Recoding: Categories of a variable are recoded into broader categories.
  6. Local Suppression: Specific values of a variable are suppressed.

These techniques can be used individually or in combination, depending on the specific requirements of the data set and the level of anonymization required.

Evaluating and Selecting a Specialized Risk Based Anonymization Partner

Choosing the right company for risk-based anonymization of clinical data is crucial, as it requires balancing utility with robust privacy protection. These are the principles Real Life Sciences is built upon. Here are some key considerations:

Expertise and experience:

  • Specific experience with clinical data: Look for companies with proven experience in handling sensitive clinical information and understanding its complexities. Familiarity with relevant regulations such as CDP/Policy 0070 and PRCI are essential.
  • Track record of anonymization methods: Evaluate their expertise in applying various anonymization techniques and their suitability for your specific data and goals.
  • Understanding of risk assessment: Ensure they have a solid understanding of risk-based approaches and can tailor the anonymization strategy to your specific risk tolerance and data utility needs.

Technology and infrastructure:

  • Security and compliance: Verify their security measures meet industry standards and regulatory requirements. Look for certifications like ISO 27001 for information security management and a QMS for comprehensive quality processes.
  • Data anonymization tools and algorithms: Assess the robustness and effectiveness of their chosen anonymization tools and algorithms.  
  • Data handling and processing: Understand their data storage, access controls, and destruction procedures.Ensure they align with your security and privacy policies.

Company reputation and ethics:

  • Customer testimonials and references: Seek feedback from past customers, especially those in the life sciences industry about their experience and satisfaction.
  • Transparency and communication: Evaluate their willingness to discuss their approach, answer questions, and address your concerns openly and honestly.
  • Ethical considerations: Confirm their commitment to ethical data handling practices and alignment with data privacy principles.


Risk-based clinical data anonymization, EMA Policy 0070, and Health Canada’s PRCI are all significant strides towards a future where clinical data is both accessible and secure. These initiatives not only foster transparency and trust but also pave the way for innovation and advancement in clinical research.

While these initiatives are a step in the right direction, it is crucial to continue refining these strategies to ensure the balance between data accessibility and privacy is maintained. As we move forward, the focus should be on developing robust, scalable, and efficient methods for data anonymization and public release, keeping in mind the ever-evolving landscape of digital health and data privacy regulations.

When implementing a risk based anonymization approach, engage with experts, like Real Life Sciences for assistance. This will accelerate your project and increase your probability of a high quality and on time project.


crossmenu linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram