James Gaiser, Associate Director, Statistical Programming, PROMETRIKA, LLC.
Quick – What is BIMO?
Is it a flying superhero? Is it an energy drink? Is it the next reality TV show? (Well, it is reality.) Here is a brief overview of BIMO and how it came to be.
Back in 1977, the FDA was granted the right to audit clinical research sites. To meet this new responsibility, the FDA established the Bioresearch Monitoring (BIMO) Program, which created guidelines for agency inspections of clinical trial sites. Gathering site-level data such as details about the investigator, IRB, subject data, and the site itself, to determine which locations to audit, is a time-consuming task for agency reviewers. As you can imagine, this slows the audit planning and completion process, which, ultimately, slows the review and approval of a new treatment. To improve efficiency, the FDA began to look to sponsors for review aids.
In 2011, the FDA published “Specifications for Preparing and Submitting Summary Level Clinical Site Data for CDER’s Inspection Planning,” which detailed the data elements and formats that agency reviewers need for site selection and for use during an audit. The following spring, I joined the Pharmaceutical Users Software Exchange (PhUSE) Computational Sciences Working Group, which was tasked with a detailed evaluation of each proposed data element in the FDA’s guidelines. In May of 2016, the group released our recommendations in “Inspection Site Selection Standard Data Elements: Gap Analysis of FDA Requirements and Existing CDISC Standards”.
The formatted data requested by the FDA fall into three general categories: general study-related and investigator-related information; individual subject data listings, by site; and a dataset summarizing 39 different elements, by site. Upon FDA request, these outputs are provided by the sponsor for each pivotal trial in the submission.
General and investigator information is a tabular compilation of such things as site location, subject disposition, CRO information, and location of source data. It also may include the annotated CRF and the Protocol and Amendments. This compilation cannot be finalized by the sponsor until the FDA makes a specific request following New Drug or Biologic Licensing Application (NDA or BLA) submission, outlining what they wish to see.
Individual subject data listings include screen failures and the reasons for failure. For treated subjects, data should include treatment assignment; reasons for discontinuation; protocol violations; adverse events (AEs), serious adverse events (SAEs), and all relevant safety information; primary and secondary efficacy endpoint results; concomitant medications; and laboratory data.
Perhaps the most complex element in the specifications for data formatting is the summary dataset. In February 2018, FDA published an updated BIMO Technical Conformance Guide, outlining the 39 variables that make up the dataset. These elements include administrative information, such as IND/NDA number and study/protocol number, and site-level safety and efficacy information. Some of the most important of the 39 elements are subject screening/screen failure data; descriptions of the treatment arms; AE/SAE data; and financial disclosure information.
Following on the success of the PhUSE working group, I have enlisted with industry colleagues on a special CDISC ADaM BIMO sub-committee working with the FDA to standardize the site-level dataset conventions by applying CDISC guidelines for data format and coding. Successfully establishing these conventions will improve sponsors’ efficiency in providing the FDA with the requested BIMO materials for new drug and biologic applications.
Although not a requirement at present, my experience leads me to strongly recommend that sponsors also prepare a BIMO Reviewers Guide for all submitted material. The guide would specify any term definitions or derivations used in preparing the dataset, and may also include such information as which sites screened, but did not enroll, subjects and, therefore, are not represented in the dataset. In short, any information that aids FDA reviewers and makes the review process efficient. With these data and tools in hand, the FDA can more quickly identify sites to be audited prior to approval of the application.
In recent years, there has been a notable increase in FDA requests for BIMO-formatted data. In my day-to-day work, I’ve seen an increase in sponsor requests for BIMO formatting, even before they have an FDA request. The complex nature of the data means that a good deal of time is needed to develop these reviewer’s aids. Sponsors should take note and consider development of BIMO elements in the planning of NDA/BLA submission activities from the very beginning. I believe this proactive approach will ultimately help shorten new product approvals, bringing innovative treatments to the market more efficiently.