Welcome to a series of articles focused on Informatics in Clinical Trial execution; each article contains 500 words or fewer on the topic of choice (not including the précis and the rambling footnotes).

Précis

When a clinical trial is underway, patients are given drugs or placebos by doctors (we call them “investigators” in a trial) for a period of time. The patients visit the investigator’s office (a site) periodically where they are examined to measure their general health, to ensure that they are not having any adverse reactions to the drug, and to see if the drug is effective. The investigator enters this information into a clinical trial management system. In order to support the investigators and ensure that they are getting the data entered in a timely and accurate fashion, “monitors” are sent out to visit the sites. These monitors (frequently called CRAs) will periodically visit each site to ensure that the team are performing their duties correctly.

Risk Based Monitoring Fundamentals

In a trial that is not using RBM, the monitor will visit every site at the same fixed interval (e.g. every 10 weeks) and will effectively perform the same tasks at each site, including verifying the accuracy of 100% of the data entered (this is a called "100% Source Data Verification" or SDV). In this conventional model, the best performing site in a trial gets the same level of attention as every other site, including the worst.

RBM assesses the potential risk areas at the start of the trial and then uses ongoing risk assessments to monitor the performance of individual sites. This means that during the execution of a trial we know which sites are performing well against the specific risk indications. We can then target the monitors’ attention not just on those sites but on specific areas of concern within each site.

Specifically, the key changes that you see in a RBM trial are:

A baseline risk assessment is carried out before the trial starts to evaluate the general study risks, site risks and ongoing operational risks. These are then used as the basis for how to adapt the cadence of visits and the tasks performed onsite. Included in this are areas such as protocol adherence, data collection & quality, subject enrollment & retention, etc.
In a non-RBM trial, the cadence of visits by the monitor is set at the start of the trial, (e.g. they are scheduled to visit each site in the trial every 10 weeks). With RBM, we are able to adapt the levels of intervention based on a site’s performance. In this model, some sites may be visited significantly more frequently and others less so. The RBM informatics system will be continually assessing performance and making recommendations for changes to the frequency. Note that the project team makes the final determination based on the central monitor's recommendation.
Instead of checking 100% of the data entered against the source materials (100% SDV) the monitor will do a smaller percentage of this low-value SDV work and then add a higher-value approach called SDR (Source Data Review). SDR takes a more holistic view of the site’s performance looking at the data on the source documents, verifying adherence to the protocol, gauging investigator involvement, etc. SDR is effective in part because the RBM system is able to give insights into those areas where the monitors attention should be focused.
A new job function is created for RBM called a “central monitor” or an “RBM lead”. Their role is to assess performance against areas of risk and recommend intervention levels. IMHO, this is a role that should be outside of the core CRA team so they can take a disinterested role in assessing and recommending actions to the project team.

All of these steps are shown to improve trial outcomes; specifically the industry reports reductions in critical findings, lower management spending, and significant reductions in missing data.

Notes:

In order to illustrate how this might look in real life, the screenshot below is from Covance’s RBM solution showing data from a real trial. In the first column all sites have a standard intervention level. Over time some sites move to higher levels of intervention and others move to lower levels. The key takeaway is that by the end of the trial we have been able to identify those sites that require very little care and feeding so the monitors can focus their efforts on those few that are still not performing at the right level.

A note on the name RBM. In a highly regulated and risk-averse world like the clinical trials business, the word “risk” invokes a visceral response. Using it without being clear that we are negating, avoiding, obliterating, or otherwise dealing with the negative effects of risk seems foolish. I am thinking that names like Common-sense-based Monitoring, Don’t-waste-time-over-there-based Monitoring, Vulcan Monitoring, may have actually improved the uptake of this excellent application of technology and process.