1. Risk Assessment Tools
– Risk Based
Monitoring
Implementation at GSK
Neill Barron
Clinical Data Strategy
GSK
26th Annual
EuroMeeting
25-27 March 2014
ACV, Vienna
Austria
2. Disclaimer
The views and opinions expressed in the following PowerPoint
slides are those of the individual presenter and should not be
attributed to Drug Information Association, Inc. (“DIA”), its directors,
officers, employees, volunteers, members, chapters, councils,
Special Interest Area Communities or affiliates, or any organization
with which the presenter is employed or affiliated.
These PowerPoint slides are the intellectual property of the
individual presenter and are protected under the copyright laws of
the United States of America and other countries. Used by
permission. All rights reserved. Drug Information Association, DIA
and DIA logo are registered trademarks or trademarks of Drug
Information Association Inc. All other trademarks are the property of
their respective owners.
3. Objectives
• Introduce GSKs approach to implementing
an RBM “data-driven” strategy
• Review use of RBM technology to drive
risk based actions on real studies
• Benefits & Challenges
3
4. A “Data-Driven” approach to monitoring......
• The RBM Technology enables a
single consolidated “helicopter”
view of risk across sites
• Ability to drill into the detail
behind any or all of the risk
signals
• Drives focussed and targeted
monitoring intervention, where
and when needed
• Enables tactical Central Study
Team oversight of monitoring
strategy
eDC CTMS External
Data
Consolidated
RBM Database
Core RBM
Indicators x 11
Study Specific
Indicators (2-
4)
Monitoring
Activity Plan
Oversight
“Helicopter
view” Visuals
CRA Action Central
Oversight
Study
Specific
Data
5. 5
What Do We Mean By “Risks”?
1
Site & Study
Performance
Recruitment rates
Withdrawal rates
Screening failures
Data
completeness
Data currency
Site Activities
Site staff issues
Study quality
issues
Data quality at site
Frequency of site
visits
Overdue activities
Clinical Data
Driven
Data quality
Safety
trends/outliers
Data variability
Study Specific
Identified by the
study team
Supplement
generic indicators
based on needs of
protocol
Typically key
efficacy or safety
Key Risk Indicators
The right data to drive a risk based approach......
7. Site003:
Risk score
Number of patients
Visit Periodicity
Intervention Required
Site031:
Risk score
Number of patients
Visit Periodicity
Intervention Required
CompositeRiskScore
Site Visit Frequency (Visits per month)
Interpreting the data to drive action
Data entered Late
Protocol Deviations
High AE Rate
3 examples of 18 indicators to drill through
DR
10. Example 1 - High Recruiting site exhibiting High
Risk
10
Monitoring Activity
Plan
Adobe Acrobat
Document
11. 1100 Month 0000Presentation title in
footer
Site exhibiting
Increase in
Overall Risk Score
Between October
and December
Example 2 – Early Detection of increase in Risk Score
14. Implementation – Benefits & Challenges
Benefits
• Quality: Enhanced study &
data quality via earlier
detection of emerging risk
• Efficiency: Focus
monitoring intervention on
areas of highest need
• Oversight: Provides a “one-
stop shop” complete data-
driven view of risk across
sites to streamline decision
making
Challenges
• Behavioural: Using risk to
drive action is not intuitive
• Understanding:
Interpreting the risk - “What
does it mean and what do I
need to do?”
• Risk Identification:
Development of the right
study specific indicators to
complement the core
indicators is essential
14
15. Next Steps to Implementation
15
• “Mini-Pilot” deployment of RBM technology to date has
validated sensitivity and accuracy of the tool in assessing
risk at site
• Further “Mini-Pilots” during 2014 will enhance
understanding of risk signals and how they drive
monitoring intervention
• 2H-2014 for full deployment of RBM technology and
process on 3 large early adopter studies
• Scale up across late phase studies to commence Q1-
2015
Editor's Notes
The “Holistic” view of risk across all sites, rather than individual CRA’s view of their own sites, enables powerful central decision making on monitoring frequency
This visual represent a birds eye view of risk across sites – it is driven and fed by individual core and study specific risk indicators
We would expect our higher risk sites to be distributed towards top right of the and green towards bottom left – i.e. Risk is driving Monitoring Visit frequency
The “high risk sites” are a minority which indicates that there is significant opportunity in deploying Risk Based Monitoring
Looking at this visual in the RBM world, a DQL would identify sites like Site003 where there is a high risk score in a site with a relatively large number of patients (deduced from size of circle)
This site could be clicked on and it will allow investigation into the individual risk areas for that site (examples on following slides) so that DQL and CRA can have targeted discussions to address root causes of risk
Also, the DQL will identify sites just like the amber site to the far right – here, the monitoring visit frequency (average) is 2.6 times per month which does not correlate with level of risk and is clearly an outlier...DQL would drive RBM by having targeted conversations with CRA/CSM to understand this lack of correlation, which may result in a change in behaviours, or even perhaps we are missing a risk from the tool that we can use to refine further.
Thresholds are based on 95% and 99.8% control limits, corresponding to approximately 2 and 3 standard deviations respectively, the latter being the classic Shewhart limits.
Here you can see where site003 lies on the funnel plot (highlighted on funnel plot in black)
By clicking on this site the detail on demand in bottom pane and right hand populate so end user can see detail. This detail automatically populates the site report for monitors generated from heatmap (see previous slide) - along with “recommended actions”
In upper part of this chart we are highlighting sites that have a statistically higher number of protocol deviations than the mean – in this example, the details of types of deviation are displayed in bottom pane.
Again these details would be automatically part of the reports sent to monitors from the heatmap, so they could focus efforts on addressing root causes
This was what was referred to as the “Jupiter” site. It was a high enroller displaying highest risk driven by a number of red signals across a number of indicators.
The site exhibited late data entry, high AE rates, issues with IP compliance and well as red signal for disease progression. As a high enrolling site, this site stood out as the one to tackle first
Team had suspicions about this site but couldnt quite put their finger on it. The risk signal drove conversations with the CRA/CSM on the root causes and it drove an onsite visit
We readily identify sites that jump up in risk score – early detection is important
This site’s risk score increase after the addition of study specific indicators relating to (1) Disease Progression (2) IP compliance
Before study specific indicators were added to the tool, this site exhibited a low risk score – demonstrates the importance of good study specific indicators
This drove specific patient retention conversations with site as well as a retraining on IP – detecting this early was important to the study
Driving the right RBM behaviours
Low Risk Site exhibiting a high MV frequency
Validated that low risk at this site was accurate
Site has been remotely monitored since then with average MV frequency decreasing for this site (from 1.1 MVs per month to 0.85 MVs per month)
Commentary/Notes
Identifying this outlier in October identified need for a MV
After MV, site’s risk score improved as outstanding items were addressed on site