To access data from AstraZeneca Sponsored clinical studies, researchers should use the Vivli platform to submit a request. Data approved for sharing will subsequently be anonymized and be shared through the Vivli research platform in a secure research environment. A Data Use Agreement needs to be signed before the approved data request will be shared.

AstraZeneca guidelines for anonymizing patient level data are derived from the industry best practices provided by organizations such as PhUSE, TransCelerate, and regulatory guidance from EMA and Health Canada.

Anonymizing datasets enables us to fulfil our legal and ethical obligation to protect our patients whilst at the same time, utilizing patient’s data in progressing medical research to its full potential in the best interests of public health.

The following data protection procedures for personally identifiable information are performed by AZ prior to data sharing:
• Standard model: SDTM
• Patient identifiers (subject number): scrambled
• Other IDs: scrambled
• Height/Weight/Body Surface Area/Waist Circumference: remove
• Site ID: remove
• Dates: replaced by a relative study day
• Rare events: x-out
• Sensitive terms: remove
• Free text: remove

Based on the risk simulation results we keep/remove/aggregate below variables (ordered by rank):
1. AGE
2. SEX
3. RACE (in case study contains HCG data – always drop)
4. BMI
5. ETHNICTY
6. COUNTRY

In some cases, we may be able to carry out additional work to customize the anonymization strategy to address particular research needs; this will require justification and additional time to prepare and provision the data sets.

For additional details on AstraZeneca’s data sharing policy, please refer to our disclosure commitments. Note, there are some studies listed under an AstraZeneca subsidiary sponsor name: Alexion, MedImmune LLC, Pearl Therapeutics Inc., ZS Pharm Inc. and University of Zurich in line with how the study sponsorship is recorded on the Clinicaltrials.gov registry. Requests for these studies will be processed by AstraZeneca.

• Studies where data sharing is prohibited by the consent, legal, regulatory, risk to AZ IP rights, or contractual constraints on data sharing
• Studies where the data requested is not appropriate for the research proposal and/or does not contain the data necessary to answer the scientific question proposed by the research.
• Studies where there is a risk of subjects being re-identified (for example, single center studies, rare indications with a small patient population or data types that lack anonymization standards).
• Studies where the data or documents are not in English, not in a format that enable sharing and/or completed prior to 2009.
• Studies where a significant proportion of subjects must be removed due to national regulations.
• Studies that are ongoing and/or do not have results posted are out of scope of sharing.
• Requests with the aim to create High or Very High Risk AI as defined under the AstraZeneca AI Global Standard. As part of responsible data sharing, AstraZeneca reserves the right to decline requests for data where proposed research will involve Artificial Intelligence (AI) solutions deemed High or Very High risk level. At AstraZeneca we define high or very high risk level as development of AI solutions that would have direct impact on people’s wellbeing and may or may not involve autonomous decisions. The ethical standards and obligations around using Individual Patient Data to develop AI techniques are in early stages of development across our industry as AI regulations and public perception are still forming. At this point in time AstraZeneca cannot maintain adequate governance and oversight for High/Very High AI risk level external data request. AstraZeneca’s principles for ethical development of Artificial Intelligence