On Wednesday, June 24, 2020, CluePoints delivered a webinar to demonstrate how using a combination of powerful analytics and comprehensive data visualization can be employed to ensure that no stone is left unturned in determining what issues are evident in studies and how to course-correct for a successful submission. If you missed the webinar, you can watch it on-demand. Watching the webinar before digesting the answers shared in this post will provide context to the questions being answered.
Below, you will find the questions asked along with answers from our Chief Scientific officer, Steve Young.
Is This Software Linked With eCRF? If Not, How Will Data Be Entered For Analysis In Software?
Yes, the CluePoints platform can link electronically to Medidata Rave, Oracle InForm and Anju TrialMaster so the clinical and operational data is pulled automatically into CluePoints. For other source systems, data can be acquired from files and via SFTP where exports can be delivered from those systems. It’s common for a CluePoints trial to have multiple data sources such as eCRF, ECG, Labs, ePRO/eCOA & CTMS with all the data harmonised for use. It is also possible for users to browse a file of data and upload it manually.
Would You Already Have Flagged Weight As A Critical Data Point (Related to Endpoints)? Or Are You Just Looking For Carelessness/Fraud? In Other Words, Would You Only Do This Measurement For Certain Data Points?
CluePoints gives you the option to identify which data are considered critical, and to run the statistical monitoring (“DQA”) on only the critical data. A current best practice employed by most of our clients is to run the DQA on ALL data, whether critical or not. The value of this is indeed to identify any areas of potential misconduct that is leading to unreliable data. A site that is propagating vital signs is an issue, even if those vital signs are not considered “critical” for the given study. However, even when running the DQA analysis on all data, CluePoints allows users to filter their view of results on only the critical data when they want.
How Do You Address Repeat Outliers That Continue To Show Despite The Initial Communication Of Risk?
Atypical test results (i.e., outliers) – whether from the statistical monitoring (DQA) or from KRIs are formalized into a risk “signal” in CluePoints, to which the test results are linked. Each risk signal will move through a series of statuses as the study team (or owner) follows up and determines whether it’s a real issue, non-issue, etc. Certain statuses are used to identify signals that are “non-issues”, which enables central monitors (and study teams) to bypass further review of the outlier results linked to those signals. Note that any actions taken related to the risk signal are also linked directly to the signal and fully tracked/documented with CluePoints. So the study team always has a complete picture of what has been done with a signal.
The Inability To Reconstruct Full Risk Stories and Risk ID And Risk/Issue Follow Up Resonate. It Seems That Risk Management Is Reactive Rather Than Proactive. Is There A Way To Proactively ID Risk?
Risk management (RBQM) is actually designed to identify and address risks very pro-actively, starting with pre-study risk assessment and risk mitigation planning by the study team. Once a study is underway, the emergence (manifestation) of risks is monitored on an ongoing basis using the central risk monitoring techniques that we discussed during the webinar. CluePoints provides an end-to-end RBQM platform that enables and supports the entire methodology. We are also leading the industry in applying advanced methods that focus on earliest possible detection of emerging risks in your study.
Can You Please Explain The Difference Between RBM and RBQM?
As discussed during the webinar, I’m not aware of any “official” definition of the scope of RBM vs. RBQM. Up until a couple of years ago, RBM was used by many (including myself) to refer to the full methodology starting with pre-study risk planning and continuing through the operational aspects such as central monitoring and adaptive/targeted on-site monitoring. Today, more people are using RBQM to refer to the full paradigm, and RBM to refer the the operational aspects only. Maybe someday we can get an official consensus on these definitions.
Can You Speak Towards What Solutions Would Be Appropriate For Smaller Studies (E.G. Less Than 5 Sites or Fewer Than 30 Subjects) Where The Volume Of Data Is Lower And May Not Support Looking At Trends Across Sites.
These smaller-study scenarios indeed present a challenge for at least the statistical methods to identify potential risks/misconduct at a site level, simply because there is not enough statistical power to identify meaningful anomalies in the data. In these cases, the same statistical tests can be (and are) used to identify anomalous data at a patient level. This is very well supported in CluePoints, including the presentation of such anomalies with the Patient Profile reports that you can create for each study. In addition, KRIs and QTLs with absolute (discrete) risk thresholds are still valuable to identify site-level risks. Finally, as discussed during the webinar, manual exploration of the study data using the advanced visualizations and reports in CluePoints is an additional, pragmatic way to centrally monitor for risks on small studies.
Any Workarounds For Risks/Issues That Are Not At The Study-Level (Program, Portfolio, System, Etc.)?
Some of our clients are currently enabling program- or system-level risk planning within CluePoints by creating a separate “study” that represents a program or system RACT. This is a very effective, pragmatic workaround until we deliver an upcoming version that will more completely support risk planning at these various levels.
We Do A Lot of Investigator-Initiated Studies, And We Do Have A CTMS. Can You Comment On What Our Priorities And Focus Should Be In Terms Of Risk Management?
We expect that the RBQM principles presented in ICH E6 (R2), including risk planning and risk monitoring, apply to all clinical research including IITs. Given the typically smaller size of these studies, it is appropriate to implement a simpler, more streamlined approach to RBQM.