Balancing Scientific Value and Identification Risk in Anonymization
By Montse Cuadrado, senior medical writer, Anna Ramírez-Soriano, program manager, medical writing, and Sarah Richardson, principal medical writer.
The European Medicines Agency (EMA) recently added two policies to its current regulatory law: Policy/0043 and Policy/0070. These new policies aim to make data more readily accessible to the public. In Europe, data protection is considered a fundamental right of citizens. However, as data accessibility expands, protecting data, particularly in clinical reports, becomes a more challenging task. Given that, with these two new policies comes increased demand for better anonymization of personal data in clinical reports.
As data accessibility expands, protecting data in clinical reports/ becomes more challenging.
According to a recent report entitled “Anonymisation: A New Challenge for Medical Writers” (authored by PPD employees Montserrat Cuadrado Lafoz, Anna Ramírez‑Soriano and Sarah Richardson) and published in the European Medical Writers Association journal Medical Writing, redaction is the method most frequently used to secure personal data in reports submitted to EMA. Typically, redaction is performed after the clinical reports are submitted to the competent authorities. The use of other anonymization methods that preserve clinically relevant data—such as transformation and generalization—is still limited. These other methods present a low risk of reidentification and achieve data utility (i.e., retain a maximum of scientifically useful information), while redaction limits data utility. However, since redaction is the most common technique to anonymize reports to the EMA, the crux of the data anonymization plan focuses on increasing the efficiency of redaction and using other anonymization methods that retain data utility.
A strong cross-functional strategy can reduce the extent of anonymization required for submissions.
The report concludes that, a strong cross-functional strategy can reduce the extent of the anonymization required for submission of the document at hand by proactively limiting the use of unnecessary identifiers in the initial report drafts, and consequently, contributes to diminish the risk of reidentification. Therefore, the challenge for medical writers is to become involved in the anonymization strategy and in the creation of a data anonymization plan for clinical documents from the initial stages. The creation of a data anonymization plan should focus on the balance between scientific value and risk of reidentification when it comes to the techniques used to anonymize identifiers.
An anonymization plan should focus on the balance between scientific value and risk of identification.
The application of these recommendations, along with the early development of a cross‑functionally generated data anonymization plan that takes into consideration data characteristics, associated risks of re-identification, data utility and the context of release, is helping PPD to bend the time cost curve for our clients. As we move forward, PPD is committed to finding the most innovative and efficient anonymization techniques, helping us to continue to deliver life-changing therapies.