Amarachi Umez-Eronini, MPH, Biostatistician I / Statistical Programmer III, PROMETRIKA, LLC
The Data Standards Series: Part 2
Data Standards Developed by the Clinical Data Interchange Consortium
To streamline the exchange of clinical and nonclinical research data and ensure uniformity across studies submitted to the U.S. Food and Drug Administration (FDA), the Clinical Data Interchange Standards Consortium (CDISC), has developed data exchange standards to streamline clinical research and enable connections to healthcare. As stated on the CDISC website introduction, “Unlocking cures is our life’s work. At CDISC, we enable clinical research to work smarter by allowing data to speak the same language.”
Within the CDISC organization, there are several data exchange standards. The most commonly used exchange standards at PROMETRIKA include: Clinical Data Acquisition Standards Harmonization (CDASH), Study Data Tabulation Model (SDTM), and Analysis Data Model (ADaM). These standards allow sponsors to submit study data in electronic format to the FDA or other regulatory agencies. As of December 17, 2016, with the exception of CDASH (though strongly recommended), the following data formats, supported by the FDA and listed in the FDA’s Data Standards Catalog, are required when submitting study data.
- CDASH — Establishes a standard method for collecting data across studies so that there is clear traceability between data collection and submission. This traceability allows for more transparency with regulatory agencies when reviewing the data. CDASH model version 1.0 has been used at PROMETRIKA. Although not currently required, the FDA strongly recommends sponsors design their trials using CDISC-defined data elements, which allow for much easier SDTM domain creation and study data traceability to the collected source data (such as is possible with use of CDASH).
- SDTM — Provides a standard method for structuring data that simplifies the collecting, managing, analyzing, and reporting of results. SDTM is built around the concept that the observations from clinical study subjects can be described by a series of variables corresponding to a row in a data set or table, with each row being classified according to its role (identifier, topic, timing, and qualifier variables). The SDTM model also supports the Standard for Exchange of Nonclinical Data (SEND), medical devices and pharmacogenomics/genetics studies. SDTM model versions 1.1, 1.2, 1.3, and 1.4; and SEND model version 1.5 are currently supported by the FDA.
- ADaM — Defines data standards that support efficient creation, replication and review of clinical trial statistical analyses. This allows for traceability between analysis results, analysis data, and data represented in the SDTM. ADaM model version 2.1 is currently supported by the FDA.
The Argument for CDASH and SDTM
CDASH and SDTM are often the subject of debate as to whether sponsors and Clinical Research Organizations (CROs) need to use both. Some believe CDASH is unnecessary due to its similarities to SDTM. Although they have much in common, the few differences often result in confusion and extra work. In a CDISC education article “SDTM vs CDASH: Why You Need Both!”, it mentions that 87% of CDASH v2.0 maps directly with mappings included in the SDTM model. However, while the two have certain things in common, they address different problems and each provides unique value to a clinical trial.
CDASH is concerned with how the data is captured and the quality of data. In the framework of CDASH, it’s more important that questions are collected in a manner that capture the true intent for the study and is easy for clinicians to understand or site users to enter study data. Meanwhile, SDTM is concerned with how the data is structured in order to streamline the conduct of analyses and reporting results. The format in CDASH would not be optimal for the creation of SDTM and vice-versa, but the quality of the data collection by using CDASH, facilitates consistent, well defined data across studies for analyses and ultimately submission.
One example that highlights CDASH and SDTM differences is the collection of date/time variables. In SDTM, we end up with a complete date field (year, month, day) and time (hour, minutes, seconds) combined for the purpose of analysis and reporting. In CDASH, this method of complete date-time collection is not helpful or intuitive to a user entering data. Instead, it’s clearer to have a user report date and time as separate components so that missing date-time parts are more easily identified and queried.
|Optimized for data capture, investigator site activities, and data quality.
||Optimized for tabulation, analysis dataset creation, and data submission.
|“Absence of evidence is not evidence of absence”. When capturing data, if data is missing, it must be checked and confirmed that it is truly missing (e.g., data query).
|| “Show me the data, not lack of data”. If data is missing, it is assumed that there is no record or nothing happened.
CDASH and SDTM: Better Together
In practice, the use of CDASH and SDTM together has been very successful at PROMETRIKA. Within our company, we have established a cross-functional standards team which continues to explore and refine how CDASH and SDTM can be used together in projects that will better serve the study team as a whole. With the goal of consistent and quality data across studies, our cross-functional team consisting of members from each department, collaborate on an ongoing basis from the start to end of study. For example, during the early stage design of the CRF, by implementing CDASH guidelines and recommendations, data management personnel work with biostatisticians and programmers to ensure that information needed for the study is captured in a way that streamlines the work needed to convert study data to SDTM format. Feedback from clinical operations and pharmacovigilance ensures that CRF questions are asked in a manner that is comprehendible and enhances the user’s experience. As a result of strategic data planning, this cross-functional collaboration has saved time and cost related to data conversion and mapping. The need for rework has been reduced and the standards implemented have been reproducible across studies.
From the benefits described above, it is recommended to use both CDASH and SDTM. As stated on the CDISC website by Jonathan Chainey (Global Head, Data Standards, Roche), “Having good industry standards for our clinical data is so important because it creates order among the relative chaos of our clinical trial data. When we start to really do that, we can focus our energies on the really innovative and differentiating aspects of the clinical trials we run and stop reinventing the wheel.” Used together, CDASH and SDTM standards positively impact data capture needs, quality, usability, repurposing, and traceability of clinical study data. The goal is to be confident that the data represents the entire picture.
Stay tuned next month for the final installment in our three-part Data Standards series, where we will explore how Data Standards and a Global Library benefit different cross-functional groups within PROMETRIKA.