Data management functions are handling a much higher volume and diversity of data in response to the increasing scope and demands of global drug development programmes. Real-world data and evidence, electronic clinical outcome assessments, data from smartphones and other mobile devices, social media community data and electronic health/medical records are examples of the new data sources captured during clinical trials. The volume and diversity of data presents integration, compatibility and interoperability challenges that must be proactively addressed in order to improve drug development performance and efficiency.
To better understand current data management practices and experience, the Tufts Center for the Study of Drug Development (Tufts CSDD) – in collaboration with Veeva Systems – conducted a study of nearly 260 pharmaceutical companies and contract research organisations (CROs). The results reveal common challenges in clinical data management across the industry.
A variety of applications but limited data
The findings confirmed that companies use a wide variety of applications to support and execute clinical studies. The top five most commonly used applications include electronic data capture (EDC) with 100% usage, randomisation and trial-supply management applications with 77% usage, both safety/pharmacovigilance and electronic master file (eTMF) systems with 70% usage and clinical trial management systems (CTMS) with 61% usage. Despite the drive toward clinical-data digitalisation, nearly one-third of companies report that they still use paper case-report forms.
The primary EDC system is still largely handling more traditional structured data elements. All respondents use their EDC to manage electronic case-report form (eCRF) data; 60% use it to manage local lab and quality of life (QoL) data; and 57% manage central lab data in their EDC. Elements such as electronic clinical outcome assessment (eCOA), medical-imaging data, mobile health data and genomic data and unstructured data are largely managed outside the primary EDC system, presenting integration and technical challenges.
Database build and loading challenges
The average cycle times to build and lock the study database are long, with no observed improvement over the past decade. The average time to build and release a study database is 68 days and the average time from a study’s last patient, last visit (LPLV) to database lock is 36 days. These average cycle times are consistent with other published and anecdotal reports.
In addition, there is wide variation around the average database-build and database-lock cycle times. This may be due to several factors, including disparity in the scope and complexity of protocols, the phase of the study, the size and operating differences between respondents’ companies and the inconsistent ways that companies implement their data management practices. Nearly half of respondents cite protocol changes as the top cause for database-build delays; 17% cite user acceptance testing and 15% say it is database design functionality.
Companies that cite protocol changes as the top cause of database-build delays report average LPLV-to-database lock times that are more than five days faster than the overall average. On the other hand, companies that cite database-design functionality as the top cause of database-build delays report substantially longer downstream cycle times, and the average LPLV-to-database lock cycle time was 40% longer than the overall average.
Additionally, one-third of companies report that they ‘often’ or ‘always’ release the study-specific EDC application after first patient, first visit (FPFV). This is associated with longer downstream data management cycle times, including time for study staff to enter data after the patient visit and time from LPLV to database lock. Specifically, companies that ‘always’ release the EDC after FPFV have cycle times that are nearly twice as long as those companies that report ‘never’ doing so. These longer cycle times may be the result of several factors, including poor site motivation, lower levels of study staff trust and confidence in a data management system, and ongoing database functionality problems.
The majority (77%) of companies report having challenges loading data into their primary EDC application, with most (66%) of those citing EDC system or integration issues as the main cause. Such challenges risk becoming an even larger problem, as companies utilise more data from more sources. The vast majority (97%) of companies say that they plan to use more clinical data from a wider variety of sources over the next three years and almost three-quarters (70%) plan to use a data source in three years that they are not currently using today.
The clinical data management function must adopt new capabilities and solutions to accommodate, integrate and analyse the rising volume and diversity of data. Life science companies want to collect more data from more sources, in order to make faster, more accurate decisions during clinical trials. But with the majority experiencing challenges in managing the data they have today, they will need to be better prepared to handle data from more diverse and less compatible sources.