Skip to main content

Table 4 Overview of the different patient level data access models and their pros and cons from the Researcher’s and Data Holder’s perspectives

From: EFSPI/PSI working group on data sharing: accessing and working with pharmaceutical clinical trial patient level datasets – a primer for academic researchers

Pros Cons
OPEN access:
 Immediate access for researchers. No pre-requisites needed regarding qualifications or documentation of the research objectives and analysis plan.
 Data Holder:
 No need for a research request and review process
No guarantee of direct access to the Data Holder if they encounter difficulties in navigating the data or if they have questions regarding the study conduct.
No knowledge of who else is accessing the data. Potential for overlapping or repeated research questions arising from the same dataset leading to increased chances of errors or increased type 1 errors.
Data Holder:
High risk to patient confidentiality as the data could be combined with other datasets.
No traceability regarding who has accessed the data, whether they are qualified in statistical analysis and how they have consequently used the data.
As pre-specification of analysis is not needed nor monitored, there is a risk for data dredging and over-interpretation of findings.
High internal costs if all trials are required to be anonymized and posted prospectively some of which may never be accessed.
Direct Sharing
 No limitation on the statistical software that can be used.
Responsible for the security of any information held on their systems.
 Easier to merge and combine data from a variety of sources. Potential impact on research credibility if collaboration by Researcher is seen as not truly independent from the Data Holder.
 Increased opportunity to address data and analysis questions with the Data Holders’ study personnel.
 Data Holder:
 Potential for identifying synergies where the research is in keeping with research interests of the organization. Potential opportunity to collaborate with the Researcher and address any questions they have during their research work.
Data Holder:
Security of the datasets is reliant on the security of the requesters systems.
Reliant on the Researcher adhering to the terms of the DSA relating to not sharing data outside the research group and only using the data for research activities that have been approved. Without a DSA there is a risk of data being misused.
Level of interaction between research and Data Holder project team could impact on internal resources and timelines for other activities
Medium resource-intensive
Controlled Access
 Ability to access data from multiple Data Holders in a defined process
Required to use only the analysis software supplied within the secure website. May not be the software that they usually use, or with which they are familiar leading to inconvenience and increased chance of erroneous analysis.
 Data Holder:
 Datasets are supplied in a secure environment so the risk of patient identification via merging with other datasets is reduced
 The named researchers are the only ones who can interact with the data.
 Ability to compare pre-specified analyses versus published findings.
 Able to share data with a wide range of researchers with less impact on internal resourcing (compared to Open and Direct Sharing approaches)
Commitments to publish/share their results with the Data Holder may be required.
Concern that the work held on the system could be viewed by the Data Holder.
Long term access may be tricky requiring archiving procedures.
May not be able to use data made accessible within a controlled environment with data provided directly due to different Data Holder sharing strategies.
Cannot be combined with controlled access data from other sources.
Data Holder:
Data are used for purposes beyond that outlined in the original access request (controlled by data sharing agreement).
Cost to the sponsor for the website.
Relatively low resource intensity
Third Party Analysis
 Researcher does not require statistical analysis expertise in the team as this will be provided by the third party.
Lack of direct interaction with the data could be frustrating. Reliant on a positive collaboration with the third party analysts.
 Able to focus his/her time on interpreting the analyses rather than in data manipulation and programming. Analysis work may be convoluted as there will need to be a lot of interaction between the third party and the researcher.
 Data Holder:
 The analyses performed on the data is in keeping with the proposal provided by the researcher. Reduced potential for unplanned, additional data explorations.
Data Holder:
Cost implications. Not practical for small organisations. Increased chance of mis-understanding of data structures and thus possible quality issues if the Data Holder is not involved.
Analysis not considered to be independent as Data Holder owns the contract with the third party.
Relatively low resource intensity but cost could be HIGH