By Rob DiCicco, principal consultant, TransCelerate BioPharma, Inc.
When TransCelerate launched its Common Protocol Template (CPT) initiative in 2014, there were three fundamental challenges:
- The first was to streamline content and improve organization to address key areas of concern for external stakeholders, such as investigators, site staff, ethics committees, and health authorities.
- The second was to harmonize the structure and key elements of content to enable automated reuse of critical protocol information.
- The third was to facilitate adoption across TransCelerate member companies and nonmember companies so that sponsors and stakeholders could recognize gains in both efficiency and quality.
A little more than two years after the initial release of CPT, the template has been downloaded by a variety of study sponsors (TransCelerate members and nonmembers), research organizations, health authorities, and academic institutions, with the total number of downloads approaching 6,000. The majority of TransCelerate member companies are in the process of implementing either the basic Word or the technology-enabled edition (TEE), which permits export of up to 47 variables in standard XML format. This output can then be consumed by other document-based applications, thereby enabling automated reuse. By 2019, TransCelerate envisions that it will be possible to automate the reuse of selected protocol information to support statistical analysis planning and clinical study report authoring. The export function in the TEE also enables the potential for auto-population of clinical trial registries.
To better understand how to leverage machine readability to support downstream automation, TransCelerate conducted a proof-of-concept (PoC) project with IBM Watson Health. The purpose was to demonstrate that constraining or fixing the terminology in the protocol and aligning it to CDISC data standards would permit automated execution of downstream processes during study design and protocol authoring. In this case, the downstream process selected for automation was the setup of a study-specific electronic data capture (EDC). The PoC was also designed to establish traceability from protocol to data collected. While the project was successful in achieving both goals, it also identified three critical weaknesses in the current common practices representative of industry sponsors:
- The implementation of CDISC requirements has been highly variable, with company-specific versions of clinical data standards existing across the industry. This inherent flexibility has resulted in enough inconsistency that achieving automation at scale would be arduous and perhaps cost prohibitive.
- Despite the widespread adoption of CPT, there remains enough variability and ambiguity in the way that study procedures and study design elements are expressed that machine learning and other forms of artificial intelligence require a significant amount of data to correctly interpret what is written.This is an inherent flaw in the current document-based approach to protocol authoring.
- Finally, the biopharma industry lacks a common data model to enable true digital data flow and to permit different disconnected systems to “talk to one another.”
Hence, TransCelerate launched the Digital Data Flow initiative, which aims to deliver the right framework to enable the kind of automation that other industry sectors have been able to achieve, by focusing on study procedures and data elements that are routinely collected with a very high frequency. One could envision that data collected to qualify patients for clinical trials, as well as data collected to demonstrate safety, would be a fertile area for an initial attempt to achieve greater standardization.
Several Apple Research Kit studies have demonstrated the ability to rapidly collect high volumes of data both actively and passively. The ongoing observational study Project Baseline, which is a collaboration between Verily, Stanford, and Duke, is another example of a modern clinical trial enabled by technology. Meaningful value from this type of automation will be recognized when it becomes possible for clinical trial sponsors to utilize this approach without building a new bespoke solution to collect and integrate the data with each new study. To address the gap highlighted by the PoC and to harness the potential of digital technology, projects such as TransCelerate’s Digital Data Flow will attempt to describe the requirements and the operational process changes that would enable technology companies, healthcare providers, and new research partners, such as patients and payers, to accelerate the collection of high-quality data in a way that is repeatable. This holds the promise of generating new insights and expediting the launch of new medicines in the modern era at a cost that is affordable. The demand for this capability becomes more critical as healthcare delivery and reimbursement move from a fee for service model to one that is value-based. With the entry of data-driven technology companies such as Google and Amazon into the healthcare arena, industry-sponsored research organizations will need to embrace practices that permit them to modernize just to keep pace.
Assuming it is possible for clinical trial sponsors, standard setting organizations, and key external stakeholders (health authorities, technology experts, etc.) to agree on fixed terms for commonly collected data elements and to establish an acceptable common data model, TransCelerate expects that the type of automation we recognized in the PoC effort with IBM Watson Health would be repeatable at scale and for projects that more closely approximate the complexity of today’s clinical trials.
Perhaps more important than making yesterday’s and today’s trials more efficient is the ability to plan, design, and deliver future clinical trials that may be more dependent on nontraditional forms of data acquisition. A modern approach to clinical development will likely depend on the need to reliably integrate electronic medical record data, stream data from mobile devices, and share data in collaborative clinical studies, creating a demand for a high degree of interoperability. This would drive productivity and quality across the clinical research enterprise rather than transferring the burden from one segment to another, as was the case with the deployment of EDC where the effort for manual data entry was shifted from sponsors to sites.
The timing of the Digital Data Flow initiative reflects how close we are as an industry to moving from a highly manual document-based approach to study design, study planning, and data collection to one that can rely on currently available technology for the purposes of realizing both efficiency and improving quality. Organizations such as TransCelerate and CDISC have already delivered important building blocks. The CPT has changed the way clinical trial organizations have thought about the value of harmonization over customization and has helped industry study teams learn how to implement change, laying the foundation for adopting new solutions when they become available.
About the Author:
Before joining TransCelerate as principal consultant, Rob DiCicco, Pharm.D., worked with GlaxoSmithKline as vice president of clinical innovation and digital platforms. Rob has more than 25 years of experience in clinical development in a variety of leadership positions in CRO, midsize, and large pharmaceutical companies. He received his Doctor of Pharmacy degree at the University of the Sciences of Philadelphia. Over the course of his career he has had a key role in the development of several successful new medicines. Rob’s area of expertise includes clinical pharmacology and experimental medicine, innovative clinical trial design, project management, and ethics in human research.