Risk Based Approach to on site monitoring Do we need to check everything? Roger Newbery, VP Clinical Management EMEA, PPD

Risk Based Approach to on site monitoring Do we need to check everything? Roger Newbery, VP Clinical Management EMEA, PPD

Highlights of FDA Guidance on Risk Based Trials 4 to 8 week monitoring visit intervals with 100% verification of all data is perceived as FDA s preferred method regardless of trial design and complexity FDA recognizes that they must recognize the value of alternative approaches to facilitate change in industry s monitoring practices Adopting risk based approaches to monitoring, such as focusing on the most critical data elements, are more likely to ensure subject protection and overall study quality No single approach to monitoring is appropriate or necessary for every clinical trial. FDA recommends that each sponsor design a monitoring plan that is tailored to the specific human subject protection and data integrity risks of the trial Reference Aug 2011: http://www.fda.gov/downloads/drugs/guidanc ecomplianceregulatoryinformation/guidances/

How widely adopted is Risk Based Monitoring (RBM)? 100% All Studies Approx 40% Consider an RBM Option 10% Adopt RBM Currently 90% of studies either do not consider or reject a RBM option. Reason: Risk of rejection by FDA or other competent authorities and/or cost savings less than anticipated

Current RBM models Most common models are simplistic, eg: Review the first N subjects for a site and if quality is high review 1 in X subjects. Review a set # of pages or data points. Once quality is confirmed high review every Xth page or data point. Review 1 of every N subjects at random, with at least one subject per site. Problems encountered: Frequently there is no clear expectation of source review or quality threshold. Site risk mitigation and escalation is often too vague and left solely to CRA interpretation. Most models save money over conventional monitoring but risk poorer quality. Tracking tools are lacking or missing.

How do study costs play into RBM? Risk based methods allow for us to analyze more data and more subjects without multiplying the monitoring costs. BUT cost reduction is not proportional to % reduction in SDV Investments will need to be made in the short term for longterm gains. Technology needs to be updated to handle risk based approaches (CTMS, EDC, MVRs) Validated data surveillance tools need to be created to trend/assess data not reviewed on site (Mgmt Reports, Metrics) Until guidance is clear and these improvements are in place: Many clients are hesitant to reduce review of source documentation (medical chart review) and only agree to reduce CRF review (complete and accurate transcription) which means very little reduction in time on site. Manual management/oversight/assessment must occur which may increase management costs, despite some saving in time on site.

Selecting the Correct RBM Model for your study Many sponsors adopt one standard model irrespective of trial design, indication or phase. This goes against the FDA guidance and is counter intuitive to risk based models. We see cases where most data are deemed critical yet a RBM strategy is proposed as part of a new company strategy. If most the data collected and reported on the CRF are truly critical the best strategy is 100% monitoring of source and CRF. The following slides review PPD s recommendations for selecting an RBM methodology for a given study

Recommendations for selecting a RBM Strategy 1 1. Define the level of expected source review to ensure site oversight: Source review establishes: source doc quality, PI oversight, protocol compliance, subject completion status, IP accountability, regulatory compliance, and medical care of subjects. Source review strategies : 100%, sampling by subject, sampling by visit Default for now is 100% source review Sampling can be based on a manual sampling plan or via random sampling supplied by programming 2. Select from the following SDV options to compare CRF to source: SDV review establishes: accuracy of site s transcription of data into the CRF from the source SDV strategies: critical vs. non critical, Reduced By Subject, Reduced By Visit, 100% SDV Critical vs. non critical is the preferred method by PPD Any SDV method can be used with any of the source options. Data available for SDV is a subset of the data that was source reviewed, if it wasn t source reviewed, it can not be part of the SDV plan. Sampling can be based on a manual sampling plan or via random sampling supplied by programming.

Recommendations for selecting a RBM Strategy 2 3. Implement and utilize tools to offset the risk: Remote monitoring to supplement or replace on site monitoring for visit/subjects not reviewed in the field Statistical detection programming Error rate calculations at a granular level Site source and site SDV quality assessments GCP deviation, protocol deviation, and query trending 4. Adapt Source Review and SDV based on quality of the site Assess project /region level plan based on project /region level error rates, query trends, deviations, etc. Assess site level plan based on site level error rates, query trends, deviations, etc. Act on source review quality and SDV quality independently

PPD s Site Risk Assessment Tool Developed a point system to identify sites that warrant additional review. We are flagging sites that could be problematic based on an objective set of criteria gathered from Clinical and Data Management. Three tiered approach: Green : No indications of suspected quality issues at site. We continue to closely monitor key data at site and sample the remaining data as agreed to by team. Yellow : Some minor indications of quality issues at site discovered as witnessed by higher than normal index rating. We would increase sampling rate of non critical data and asses the site for potential issues. Red : Significant indications of quality issues discovered at site as witnessed by an excessively high index rating. We would perform 100% SDV on the study for the site and perform retraining as needed.

Timeliness of CRF delivery to PPD Sample Site Risk Assessment Tool F

PPD s Adaptive Differentiation True risk based approach to helping mitigate risks on trials. No one size fit all model for sites. Reward strong performers and contain risk of problem sites. Once problem fixed at site due to proper corrective action site back to sampling model (truly adaptive model). Criteria used to asses site quality are multi functional and objective. Enhanced listings create greater efficiency for CRA at site. Automated production using statistical sampling algorithm. CRA not introducing bias to the sampling approach.

Challenges with RBM Implementation Systems are not designed to accommodate <100% source and CRF review CTMS systems track per subject or visit EDC systems track per page Site quality tracking capabilities are not well established: Quality of source Quality of CRF Site risk factors Other How to resource fluctuating monitoring plans for each site? How to handle queries on data we weren t going to source review or CRF review? Do queries have to be SDV d? Training staff on various approaches and strategies for RBM and overcoming resistance How to build technology to support our needs with different clients specifying different definitions, expectations, strategies and methods? How do you calculate the costs of an adaptive model?

Conclusions Most sponsors elect to continue with 100% source and CRF review Phase IV trials employ more RBM experimentation Variability in RBM models and methods used Cost savings from RBM are frequently over estimated Trial systems frequently do not support RBM Current inspection and audit practice is a barrier to further adoption Greater consensus and clarity from regulators needed