Veeva CDB
n recent years, the data workbench has emerged as a powerful tool for aggregating and reviewing data from multiple study sources. The data workbench—often called the clinical data platform—reduces cleaning cycle times and speeds the preparation of analysis-ready data. With 87% of sponsors and CROs deploying decentralized clinical trials and the volume and variety of non-EDC data expanding, the need for consolidated data management and review will only intensify.
Siloed Systems Slow Data Review
Disconnected technologies and serial, stepwise workflows create inefficiencies and delays from the first data upload to database lock. Aggregating and reconciling data from disparate sources requires multiple handoffs. Data consolidation is manual, and the data is no longer current when reconciliation reports are ready.
Data review is equally cumbersome. Data managers scour multiple “offline” data listings from SAS or a similar tool, hoping not to overlook any missing or inconsistent data. Then they repeat the process in the following review cycle, with no easy way to differentiate outstanding and complete records.
Managing queries is even more complicated. Best case: Data managers log into the EDC system, then navigate to the correct study, site, subject, form, and field. For data captured from non-EDC sources, issues are tracked in spreadsheets, and queries are handled via email. Gleaning operational insights across all data sources takes a significant manual effort to merge, map, and transform data for downstream needs.
Data Workbenches Unify Data Review
A data workbench combines all study data and review activities in a single system, and can transform how an organization manages and ensures the quality of trial data with capabilities that include:
- Aggregation and harmonization of data from all study sources, including EDC systems, labs, safety, eCOA, ePRO, wearables, and more
- Pre-built and programmable listings and reports to review, clean, and reconcile data
- Centralized issue/query management, including integration with EDC systems
- Dashboards and visualizations for operational and clinical insights
- Communication between the study team, vendors, and clinical sites
- Transformation and export of study data for downstream users.
With a data workbench, data managers spend less time on the process and more time reviewing the data. Data managers can review listings and reconciliation reports, be confident that the data are current, zero in on flagged and open issues, and drill down into specific records for contextual information. The status of data review across all study sources is readily available, avoiding the need for tracking sheets and issue logs.
A data workbench makes it easy for data managers to collaborate with other study team members. Programmers and data managers can iterate on reports and data listings before the study kickoff, while data managers can discuss and resolve data issues with internal and external staff without switching to email or the EDC system during study execution.
Adopting a Data Workbench
Implementing a data workbench is not as simple as swapping one electronic trial master file (eTMF) system for another. The organization needs to define new processes, data listings, reports, and metrics, and preparing study teams for these changes is crucial.
1. Start Small
Selecting the right first study for a new data workbench is critical to early success. Start small and consider factors like duration, complexity, visibility, and therapeutic area. A low-profile, two-month study offers a less pressure-filled environment for rapid experimentation and learning than a multiyear study for the next blockbuster drug.
Similarly, don’t attempt to completely transform all aspects of clinical data review. Address the most pressing business requirement first, whether it’s reconciling data from multiple sources, streamlining data review, or managing queries more efficiently. Look for opportunities to make meaningful improvements within the project’s first three to six months and include subject matter experts from multiple domains in this prioritization.
2. Revisit Requirements
Begin with a blank slate. Don’t default to specifications for existing listings and reports, which reflect the limitations of siloed systems and workflows. Instead, take a step back and focus on the core requirements for each data output. Who will review it? What are they looking for? What action will they take? One large, multifunction report can often be broken into several smaller, purposeful outputs that are easier to both program and review.
Work closely with data reviewers in different domains to understand their job activities and goals. For example, when a major biotechnology company adopted a data workbench, they held “a day in the life of a data reviewer” workshops to uncover user needs and pain points. Reviewers also participated in report creation. Collaboration with data reviewers at all stages, from ideation to optimization, helped.
3. Let Go of Legacy (Processes)
Your organization has probably compiled dozens of SOPs, work instructions, templates, and job aids for data review that, like your data listings and reports, were all designed for a specific system (e.g., SAS) and the “old way of doing things.” To allow for change, authorize the study team to deviate from existing SOPs.
A data workbench creates opportunities for process improvement through automation, collaboration, and concurrent review. Programmers no longer aggregate data manually. Data managers don’t have to send extracts from a multipurpose report to study team members. Reviewers can look at the same data in the same tool and manage issues without spreadsheets and email. All these changes (and more) make existing operating procedures obsolete.
The experience at one European affiliate of a major biopharmaceutical company provides a good example. “When we adopted a data workbench, we initially tried to maintain our existing processes,” explains one of its principal clinical data managers. “After a lot of effort, we realized that our old processes couldn’t be adapted to the new technology. My advice to others implementing a data workbench is to be open to change.”
As your data workbench journey begins, encourage your team to develop new processes that leverage the system’s capabilities. Explore ways to eliminate handoffs, redundant reports, manual tracking, and switching between systems while remaining flexible. Don’t try to establish a permanent process during your first set of studies.
4. Use Your Internal Experts
Nominate initial users who are experts in their domain, because they can inform data workbench uses and best practices moving forward. Recruit study team members from data management, medical monitoring, pharmacovigilance, clinical operations, programming, and biostatistics.
It’s vital that subject matter experts feel empowered to make decisions, mistakes, adjustments, and tradeoffs; otherwise, they may revert to old processes, and you won’t realize the full benefits of a data workbench. If the team feels ownership of the implementation, they will experiment, innovate, and have a meaningful impact on cycle times and data quality. Their early wins will create a buzz in the organization, promoting broader adoption of the new technology.
5. Apply Additional Resources
Recognize that the first studies with a data workbench will take longer and require more data management resources than an equivalent study with traditional tools. The study team will need time to create new listings and reports and learn a new system and data cleaning process. It’s also a good idea to allocate time for initial users to review and refine how work is best accomplished in the new system.
The previously mentioned principal clinical data manager recommends involving multiple data managers in the implementation project. Tapping into the experience and perspective of various people will resolve issues faster and result in a better data management solution.
6. Stay Agile
Take an iterative, incremental approach to data workbench adoption. After completing the first study, share and discuss learnings before embarking on the second study—and another round of exploration. “Recognize that early studies are a period of trial and error with unexpected considerations and complications,” this data manager advises. “Allow processes to evolve over time as users grow more comfortable with the new system.”
Apply a similar strategy to developing data outputs and visualizations. For instance, if your organization wants to create 30 data listings, don’t attempt all 30 simultaneously. Instead, begin with five to 10 high-priority listings and take them from specification, through development and review, to outputting live data in the production environment. You’ll learn faster and be able to use that knowledge when building the next batch of reports.
7. Modify Metrics
New processes require new metrics. Key performance indicators (KPIs) used with past review processes aren’t the best measures when data, reporting, and cleaning activities are combined in a single system.
For example, legacy KPIs for report configuration might include: How long did it take for clinical programming to prepare the report and for data management to approve it? How many review cycles were needed, and how many change requests occurred after production?
A data workbench might change those KPIs: What was the cumulative effort for the study to prepare a draft report? And once in production, what was the cumulative effort to generate and review the report for the duration of the study?
Remember that processes will be in flux during your initial studies. Don’t attempt to hard-code metrics while the first study teams are still working out the best way to use the new system. Let your organization settle on a process, then define the appropriate performance benchmarks.
Clean Data, Faster
With the number and variety of data sources skyrocketing, an outdated data review process can slow study timelines and regulatory submissions. A data workbench modernizes and accelerates cleaning and reconciling data from multiple study sources. Organizations should prepare to evolve their data review practices to take full advantage of a data workbench solution. An agile approach to adoption, with iterative learning and experimentation by multiple data managers, ensures a strong start.