While an “ideal” Statistical Computing Environment (SCE) is highly sought for in the pharmaceutical industry, one is yet to be developed. Hoping to nudge tech business leaders to develop such a platform, experts from top global pharma companies have come together and compiled a list of their recommendations for the requirements of an ideal SCE. These requirements were published in December 2021 in a white paper titled “Statistical Computing Environment” (Bynens et al., 2021).
The purpose of this article is to summarize the published white paper and highlight key features that we, at Beaconcure, think are vital for the development of new statistical platforms.
Current practices and software are old and are not scalable in face of technological advancements, especially with new emerging data structures and AI-driven algorithms. There are currently no one-platform solutions that address the full set of pharma end-user requirements.
The SCE is envisioned as a single platform that receives and accesses clinical data at different stages of the drug development process and allows for analysis, tabulation, validation, and submission of deliverables in line with regulatory requirements. The following sections dive deeper into specific components that authors of the paper believe the SCE needs to have.
Breakdown of the SCE
User Interface (UI) and User Experience (UX)
The SCE should be efficient, user-friendly, and tailored specifically to the user group needs by following a user-centered design. Users should not be limited in doing things in a particular way – the system should allow for some flexibility in the user experience.
It is important for the pharma industry that the SCE is scalable to accommodate different data structures, fluctuations in use, as well as a high performance to allow the analysis of heavy statistical models. Therefore, the SCE should be deployed using cloud computing with high performance and scalable computational engines.
An SCE should be language agnostic and format agnostic, providing users the freedom to develop programs in their preferred language and formats. Clinical and statistical programming languages supported by an SCE in addition to Statistical Analysis System (SAS) can include Python, R, MATLAB, Perl, S+, Julia, shell scripts and more. Standardized structures should be developed that are optimized for all the stages of the drug development process. These structures can be enforced through templates and naming conventions.
Programming Workflow and Processes
Changes and edits to files should be tracked through version control and audit trails capabilities. With version control, multiple users can work on the same project and ensure changes are tracked and can be reverted when needed. Audit trail should “record” who accessed the SCE, when, and what operations/changes were performed. The SCE should also have traceability capabilities that allow users to trace final tables, listings and graphs back to the collected raw data.
The SCE must be able to interact with other systems to enable data to be imported or exported seamlessly. The SCE should have an area accessible to outside vendors (e.g. CROs) so that they can access and submit relevant data directly.
Outsourcing and Collaboration
When deliverables are outsourced to CROs, the SCE should have the functionality to easily import or transfer those deliverables back to the sponsor SCE and vice versa. Internal collaboration can also occur within an organization where it should be easy to work together with Medical Writing, Medical Affairs, Drug Safety, Clinical, Research, etc.
Project Management & Metrics
The SCE should integrate with a planning and tracking tool with the relevant metrics that are suitable for all levels of users. Items can be checked off by team members when certain steps/tasks/deliverables are completed.
It should be possible in the SCE to display different reports or dashboards regarding the metrics collected where these metrics can be compared against a company or industry baseline. Reports should be customizable and standardized, when needed.
The SCE should have a robust audit trail, access controls, traceability, version control combined with end-to-end documentation, qualification of personnel and electronic signatures.
The SCE itself is a major producer of metadata needed to manage statistical processes and workflow so it must incorporate metadata management capabilities within its own environment. The SCE should use metadata and existinting standards to enhance productivity and introduce automation to statistical analysis.
Conclusions and Beaconcure’s Role
Many of the points outlined in the white paper align with Beaconcure’s vision. In 2018 Beaconcure created a platform called Verify, that facilitates the validation of statistical clinical analysis data using automation tools. Verify processed to date more than 100 studies. By taking a user-centric approach, we created a scalable and format agnostic platform that emphasizes the end-user needs for seamless validation processes. Beaconcure is committed to continuing to work towards improving the workflow of clinical data processing and will contribute to the standardization efforts that the life science industry is currently working towards.
Bynens et al., (2021). Statistical Computing Environment [White Paper]. Phuse US Connect. https://phuse.s3.eu-central-1.amazonaws.com/Events/2021/US+Connect+2021/SCE_White_Paper_Final_15Dec2021.pdf