On Tuesday, March 31, 2020, CluePoints delivered a webinar centred around the critical importance of early risk detection. Our presenter also shared insights on how to navigate risk management and risk assessment during the COVID-19 crisis. If you missed the webinar, you can watch it on-demand. Watching the webinar before digesting the answers shared in this document will provide context to the questions being answered.
Also, please take a look at our complimentary COVID-19 Risk Assessment package which includes free access to the CluePoints Risk Assessment and Categorization Tool which includes a template to guide you through COVID-19 specific risks that are likely to be relevant to your study.
Below, you will find the questions asked along with answers from our Chief Scientific officer, Steve Young.
At what Time Points/Frequency would you Recommend Running a Study through CluePoints?
It depends on the study, and in particular the expected rate of accumulation of patients and patient data across the study. CluePoints has a “Review Planner” tool that we share with our clients to help them define a schedule for DQA and KRI/QTL reviews, which uses a handful of study planning parameters to recommend the frequency/timing of reviews.
Could you Please Talk a Little bit About Access to Data Needed for the Risk Assessment? e.g. How Does CluePoints have Access to the Audit Data of the ePRO Device? What is Involved in Accessing the Data?
We have found that ePRO/eCOA vendors are easily able to provide the audit trail data as part of their extracts, and as it forms part of the sponsor-owned data, it is readily available once requested by the sponsor or by CluePoints. We have an sFTP facility for receiving data extracts securely. And if needed, we and/or our clients can directly upload files in any number of formats (e.g., SAS, SDTM, csv, etc.).
Does CluePoints Make Recommendations on how to Define QTLs - not what QTLs to Implement, but Rather how to Best Define those that the Study Team will Implement to Ensure the Sensitivity is Optimal?
CluePoints does indeed provide guidance to our clients on optimal design of QTLs, as illustrated during the webinar on how to consider normalization of the QTL parameter.
How Did you get Access to Audit Data for the ePROs you Discussed?
We have found that ePRO/eCOA vendors are easily able to provide the audit trail data as part of their extracts, and as it forms part of the sponsor-owned data, it is readily available once requested by the sponsor or by CluePoints
How Does your System Link to an EDC system, and if we use a CROs EDC, do you Need Some Kind of Data Transfer Agreement in place?
We have connectors already built for commonly used EDC systems: Medidata Rave, Oracle Inform, OmniComm TrialMaster. We don’t need transfer agreements in place with each vendor – instead authorization comes from the sponsor/CRO partner that we are working with. We can easily build connectors to other EDC, eCOA and IRT systems when working in partnership with Sponsors and CROs Besides the connectors, we have an sFTP facility for receiving data extracts securely. And if needed, we and/or our clients can directly upload files in any number of formats (e.g., SAS, SDTM, csv, etc.).
What Information is Needed to Estimate QTL? Like Source Data/SDTM/?
For deriving the QTL metric/parameter, it really depends on the specific QTL. For example, a QTL for patient early termination rate will require source data regarding patient status which is typically captured in EDC.
What is the Percentage of Trials used RBM in Phase 2 Studies?
We don’t have current industry level metrics on % of studies by phase using RBM. However, we are generally observing that organizations implementing RBM are applying it to all study phases including Phase 2. This includes centralized monitoring, which can be applied effectively to detect risk for smaller studies as well as large studies.
What would be your Suggestions for COVID-19 for Studies that are being Paused for Screening and Enrolment?
Generally speaking, we recommend continuing to use central monitoring for risk detection even if enrolment has been paused – assuming that existing patients are continuing to participate in the study. If a study has been put on hold completely, then of course there will be no new activity to monitor until/unless the study is restarted.
Could you Please Discuss Early Phase (Phase 1 or 2) Applications of CluePoints?
The CluePoints RBQM platform is being used increasingly for smaller studies, both phase 1 and 2. And we have developed best practice guidance for our clients for leveraging CluePoints on small studies. We would be happy to share that with you, just contact us.
What are the KPIs you would Suggest for Central Vendor Provided Data? For Example, Imaging data and Biopsy data?
Central data should generally be included in DQA statistical analyses along with site-generated data, and sometimes included in KRIs depending on study-specific risk assessment. Site-level and/or systemic
In one of the case studies presented in the webinar, there were a lot of Errors in Site 105. How About Other Outstanding Sites Such as Site 101, 102?
Yes, there were other sites that showed high degrees of atypical data as you saw in the bubble plot. On that particular trial those additional sites had their own problems, but they were problems that were managed by the study team at the time and didn’t escalate into anything nearly as problematic as the site we focused on. It’s just not possible to change some of the data that’s collected, or the metadata around it like when it was collected. In this respect the data you have will always tell a story which is why understanding what those stories are before submission is so important, as regulators like the FDA (who use CluePoints to help identify Sites for audit) can analyse the data and see the same trends and results independently. It’s a great to have found all such patterns in the data and have justifications and corrective actions at hand, be they good or bad, before the data is submitted as it avoids delay in the submission process responding to the regulators. I recently heard that a different customer had seen a site be flagged by the system because of a significantly higher number of adverse events reported. During the subsequent investigation, where data was being reviewed and checks done to make sure that medical histories weren’t being mistakenly reported as AEs etc, it turned out that the site was actually an oncology site. It wasn’t an oncology trial, but there was nothing precluding oncology patients participating in the trial and it just so happened that one site was an oncology site. The system rightly detected that the site was different, but the subsequent investigation work concluded that it was understandably different and not a concern. That’s all good information to have documented prior to submission and it can also be referenced in the CSR so that the regulator knows all about it if they notice the same pattern in the data when they review it.
What will be your Recommendation if Key Data Cannot be Remotely SDVed (e.g. ICFs, Medical Records) - in EU it is not Permitted for Example?
The good news around SDV is that many of the problems that could be detected by having eyes on the source at site can also be detected remotely, other than the check to see if the data has been transcribed properly. The addition of statistical monitoring will test all the data that you can receive electronically to destruction and identify atypical patterns that show procedural issues at the site or data tampering/fabrication (it’s very hard to cheat well – as we’ve proven many times). It’s detecting these other issues via the process of SDV that I think still drives trials to have a lot of SDV implemented, when these higher value problems in the data can still be detected by other means. When I sat through the DIA webinar last week, presented by local research professionals in China, they certainly called out statistical monitoring as being a valuable tool to taking up the strain of losing the on-site activity, but what wasn’t clear was whether it had been an existing quality control technique already in use that then became increasingly important, or whether study teams have implemented statistical monitoring as a new technique in response to losing their eyes on-site. As well as statistical monitoring, remote patient data review can also be valuable, especially if it’s through visualization techniques such as Patient Profiles, where it’s easy to see patient specific data from multiple datasets on a single Gantt chart view. We’re having conversations with a number of existing customers that hadn’t chosen the Patient Profile component of our platform originally, but now want to implement that onto their existing trials to facilitate better remote patient review. The regulators seem to all be saying it’s fine to leverage appropriate and proportionate alternatives, be they centralised and or remote as you feel necessary, and have those changes documented.
Are the % Progress Milestones Based on Clendar Dates (i.e. from Study Start-Up Date to Study Closeout Date), or Based on Data Availability within the Clinical Database (i.e. 5%, 10%, 20%, etc. of Data Available in Database)?
Each progress milestone represented the date by which that particular % of patient visits had been conducted at the key problem site.
What do CROs think About your Tools? Against their Business? In Favour?
The CRO dilemma has been an interesting one over the years. At the advent of RBM, CROs were uncertain about replacing any on-site monitoring efforts in favour of this sort of remote monitoring technology. However, as time has gone on, CROs realise that the paradigm is changing, and the astute players have tried to get out in front of the change to ensure that they are the innovators in the field. We are now enjoying some excellent partnerships with CRO partners and the adoption is growing rapidly within the CRO community. RBQM isn’t going to go away and the COVID-19 crisis is making it a priority. Any CRO who is still in the dark ages is going to have to get their act together pretty quickly now, I think. And of course, sponsors can (and do!) insist on the use of Central Statistical Monitoring by their CROs to ensure their trials are better, faster and cheaper.