Risk-based Monitoring vs Traditional Monitoring: the ADAMON Project

Many clinical research professionals are RBM believers, convinced that a risk-based approach to monitoring represents an effective and more efficient alternative to the tradition of frequent on-site visits and 100% SDV.  Many others are cautiously open to the idea, and are waiting for news of successful RBM studies to start filling the industry journals and online forums.  From those same journals and forums, we know that RBM studies have been, and are currently being, conducted.  In collaboration with CluePoints, GSK Vaccines is piloting a central statistical monitoring approach.  Sanofi and Lilly are currently piloting the TransCelerate methodology published in May.  Bayer and other sponsors are using an analysis of EDC data and metadata to steer their on-site monitoring activities.  Many CROs are offering their sponsor clients alternative monitoring solutions; Quintiles has conducted upwards of 60 studies using a number of different RBM implementations, Health Decisions has had years of experience conducting trials using its adaptive monitoring approach, and ICON has just announced its ICONIK Monitoring service after several years of piloting it alongside the traditional approach.  Finally, almost every RBM webinar (RBM-inar?) I’ve attended has mentioned that government and academic clinical trials commonly employ a RBM strategy.

But it’s been quiet.  The big headlines necessary to assuage lingering doubts have not been forthcoming.  Even as we begin to hear about RBM successes, how will we know if RBM is as good as traditional techniques? 

The ADAMON Project
 
The ADAMON project was designed to answer that precise question.  Cited in the FDA’s Risk-based Monitoring Guidance, and run by the German medical research network TMF (Technologie und Methodenplattform für die vernetzte medizinische Forschung), the project’s primary objective is to investigate whether a reduced, risk-based monitoring strategy is equivalent to an extensive “full” monitoring approach.   ADAMON consists of 12 Phase II, III, and IV, non-commercial (academic, government-funded, or investigator-initiated) clinical trials, totaling 3200 patients.  Study sites in these trials have been randomized, and have agreed to receive either traditional or reduced, risk-based monitoring.  As the trials complete, occurrences of serious findings that jeopardized subject safety, subject rights, or data reliability will being assessed and compared.  The ADAMON project is funded through December 2014, but some results are expected as early as this year.
 
Each clinical trial in the ADAMON project underwent a detailed risk analysis to determine the reduced monitoring strategy the randomized sites would receive.  The risk analysis places a trial into 1 of 3 classifications: K1 reflects the highest level of risk; K2 reflects intermediate risk; K3 reflects low risk.  (Why 3 risk levels?  The researchers say that 2 levels separating the highest and lowest risk studies would be insufficient, but 4 or more levels would, “in practice, probably lead to indistinguishable strategies.”)

The ADAMON Risk Analysis
 
ADAMON’s risk analysis begins with assessing the safety risk associated with the therapeutic intervention itself.  Is it comparable to the risk the patient would run if treated outside the study protocol?  Does it present a higher risk than that of standard medical care? Markedly higher?  Other high-level study characteristics, such as study phase, whether the drug is approved, whether new or existing indications are being studied, and whether a combination of therapies is involved, are also considered during this initial safety risk assessment.
 
Next, a series of 15 questions are used to identify additional study-specific risk indicators.  Generally, it only takes the presence of one such indicator (e.g., a vulnerable patient population) to promote a study from its initial classification to the next risk level, (i.e., from K3 to K2, or from K2 to K1).  However, only if the risk associated with an indicator can be mitigated by on-site monitoring, and cannot be mitigated by other means, will the indicator’s presence alter the risk classification of the study.  The questionnaire focuses heavily on subject rights and safety factors.  In doing so, the questionnaire does omit some important risks to data integrity that appear in TransCelerate’s Risk Assessment Categorization Tool, such as level of outsourcing, length of study, or a high volume of data  points, associated, for example, with PK data collection.  Interestingly, the presence of certain study robustness factors (e.g., easily-assessed endpoints) can demote a study’s risk classification from K1 to K2, or K2 to K3.
 
Reduced Monitoring Strategy by Classification
 
A study’s K classification determines the monitoring that sites randomized to receive reduced monitoring will receive.  For example, a K1 study calls for 6+ on-site visits per year, 100% SDV of key data, and SDV of 10% of patients for non-key data.  (Just what data constitutes “key” for a study will vary, but ADAMON suggests at least the following set be considered key for all studies: (1) subject existence, (2) informed consent, (3) SAEs, (4) application and dosage, (5) data used to asses primary endpoints, and (6) data used to assess eligibility.)  A K3 study, in contrast, calls for 1 on-site visit, 100% SDV of subject existence and informed consent data for those enrolled at the time of the visit, at least 20% SDV of other key data, and no SDV on non-key data.  Studies classified as K2 are annually subdivided into 2 groups - those with noticeable problems, and those without, and on-site visit frequency and SDV percentage are adapted accordingly.  A study’s K classification also determines requirements for its pre-study, initiation, and close-out visits.
 
Lastly, the ADAMON risk analysis includes a set of 5 indicators which identify additional site-related requirements: technical, personnel, essential documentation, and IP/sample material handling requirements.  These site-requirement indicators do not affect a study’s K classification, but are taken into the account in defining the task list of the monitoring.
 
For Cause Monitoring Visits
 
ADAMON emphasizes that monitoring must be “embedded in an overall Quality Management (QM) concept which includes central monitoring, as well as prompt and pro-active data management.”  Irregularities detected via QM procedures such as telephone interviews, data validation, query resolution, or statistical monitoring methods can trigger additional for-cause site visits.  Of course, irregularities observed during a scheduled on-site visit can also prompt unplanned, for-cause monitoring visits.
 
Conclusion
 
Other than the ADAMON website and an article published in the Journal for the Society of Clinical Trials, there doesn’t seem to be much written about the ADAMON project.  That’s surprising, given the reference to ADAMON in the finalized FDA RBM guidance and the intense interest in this subject.  Regardless, we can expect some results soon.  How will the adapted monitoring approach stack up against the traditional approach?  Will the results be convincing or tepid?  Will they be easily transferable to a commercial study environment?  Will they dampen the enthusiasm for RBM, or make believers out of the dubious?
 
As always, we’d love to know your thoughts.

By Laurie Meehan