Venn Diagram to Display the Distribution of the Adverse Events

Visualizing the clinical trial data is becoming more common and various plots can be drawn to visualize the data. In previous posts, we discussed various types of plots that can be used in describing the clinical trial data. 

• Graphic Presentation: Kaplan-Meier Plot, Q-Q plot, Box Plot, Funnel Plot, Swimmer Plot, Forest Plot, Spaghetti Plot, Bubble Plot
• Waterfall plot(s) to display the results in oncology and non-oncology clinical trial

Recently, we are discussing the use of the 'Venn diagram' for displaying the distribution of the adverse events - the number and percentage of overlapping AEs. See an example of a four-way Venn diagram below:

Wikipedia introduced the Venn diagram as the following: 

A Venn diagram is a widely-used diagram style that shows the logical relation between sets, popularized by John Venn in the 1880s. The diagrams are used to teach elementary set theory, and to illustrate simple set relationships in probability, logic, statistics, linguistics and computer science. A Venn diagram uses simple closed curves drawn on a plane to represent sets. Very often, these curves are circles or ellipses.

According to the paper "V is for Venn Diagrams!": 

Venn diagrams where introduced in 1883 by John Venn (1834-1923), the Hull born philosopher and mathematician. They are a great way to visualize elements that are unique to only one group and simultaneously visualize elements that intersect with other groups. They are symmetrical by nature and the number of groups in a Venn diagram = 2n (including the group outside the diagram).

In clinical trials or the pharmacovigilance field, Venn Diagram can be used to virtualize the distribution of the adverse events, especially to display the distribution and relationships of the frequent adverse events. 

In NIH's "Guidance on Reviewing and Reporting Unanticipated ProblemsInvolving Risks to Subjects or Others and Adverse Events", the Venn diagram was mentioned for summarizing the general relationship between adverse events and unanticipated problems. 

In a paper by Gattepaille et al"Prospective Evaluation of Adverse Event Recognition Systems in Twitter: Results from the Web‑RADR Project", the Venn diagram was used to summarize the relationship of the recall performance results of the first two components, the relevance filter, and the NER module. 

In a paper by Xie et al "Differential Adverse Event Profiles Associated with BCG as a Preventive Tuberculosis Vaccine or Therapeutic Bladder Cancer Vaccine Identified by Comparative Ontology-Based VAERS and Literature Meta-Analysis", Venn diagram was used to compare four groups of the AEs associated with BCG TB vaccine or bladder cancer vaccine using VAERS and literature resources.

Additional References:

• Driving Clinical Safety Reviews with Data Standards
• SAS JMP can create the Venn diagram
• Cardiovascular Events Associated with Chimeric Antigen Receptor T Cell Therapy: Cross-Sectional FDA Adverse Events Reporting System Analysis
• Major Adverse Kidney Events in Pediatric Sepsis

Are data listings for clinical study reports still needed in the era of CDISC standard data sets (SDTM)?

Traditionally, all data fields that are collected in clinical trials, will be listed in so-called 'data listings'. The data listings are the basis for the summary and analysis tables or figures, and all together, tables, listings, and figures (TLFs) form the basis for the clinical study report (CSR). According to the ICH E3 "Structure and Content of Clinical Study Reports", these data listings are included in the CSR section 16.2 "Patient data listings". 

Now that the CDISC standard has become a mandate for regulatory submission of the clinical trial data, the data sets submitted to the regulatory agencies will follow the same standards (data structure, data set name, variable names,...). This will enable the regulatory reviewers to use the software (the data visualization software such as JMP-Clinical) to visualize and review the data and perform the data mining activities. Logical thinking is that in the era of CDISC standard data sets (SDTM - study data tabulation model), the data listings become obsolete and redundant. Some sponsors argue that there is no need to generate the data listings if the submitted data sets are in SDTM format. 

In a presentation "Alternatives to Static Data Listings for Clinical Study Reports" in a CDISC virtual conference, the presenters argued:

"Sponsors often create voluminous static listings for Clinical Study Reports (CSRs) and submissions, and possibly for internal use to review safety information.  This is likely due to the perception that they are required and/or lack of knowledge of various alternatives.   However, there are other ways of viewing clinical study safety data which can provide an improved user experience, and are made possible by standard structures for such data, such as the Study Data Tabulation Model (SDTM). The purpose of this paper is to explore some alternatives to providing a complete set of static listings and make a case for sponsors to begin considering these alternatives."

On Pinnacle 21's website, there was a blog article "FDA Final Guidance Webinar Q&A":

28. With the advent of a new "misc folder" instead of listings do you think that FDA is getting away from generating listings for each study? It seems that the data sets (SEND, SDTM, and ADaM) would stand alone to support any listings.

Answer: Once the standards requirements are in effect, the idea would be that listings would be replaced by the SDTM tabulations data, so yes.

It seems that FDA is ok not to receive the data listings since the CDISC-compliant data sets are submitted. However, it is still the sponsor's risk to take if data listings are not generated and not included in the CSR. The vast majority of the sponsors are still generating the data listings for the CSR and including the data listings in the submission even in the era of CDISC standard data sets. 

In deciding if the data listings need to be generated for CSR and submission, the sponsor needs to consider: 

• The industry trend. Is it still the industry standard to generate the data listings even after CDISC compliant data sets are submitted?
• Is there an official guideline indicating that the data listings should not be generated? 
• How to handle CSR section 16.2 according to ICH E3 regarding patient data listings if data listings are not generated? 
• Data listings may be used for internal reviews (medical review, medical writing, Quality Control review,…) even though FDA reviewers may not need the data listings because they have specific tools to review the SDTM data sets
• Data listings may be needed if the submissions are needed for other regulatory agencies beyond the FDA.
• Not all reviewers have the tools or can effectively use the tools to review the electronic data sets. Some reviewers still rely on the data listings to review the individual subject-level data. 

In the era of CDISC standards, more effort was spent on SDTM and ADaM data set programming. Once the SDTM and ADaM data sets are generated and the raw data sets are mapped to the standardized SDTM data sets, generating data listings is not so complicated.