Risk-based monitoring: Visualizing the risk to your clinical trials

We’ve talked about the data. We’ve defined our thresholds for risk. Now it’s time to talk about how you can visualize the safety and quality signals from your ongoing clinical trials.

If you want to minimize the impact of quality issues to the data or quickly address any safety concerns at a clinical site, time is of the essence. You can't be burdened with reviewing stacks of tables or listings trying to identify where things went wrong. That’s the old way of doing business. In a modern review environment, the signals need to be front and center. They need to grab our attention and help us easily answer questions about our data. JMP Clinical makes that possible.

See risks at a glance
We have already introduced the site-level risk table shown in Figure 1. Variables for which signal thresholds were defined will be colored in green, yellow or red to clearly indicate any immediate problem areas. Or we can view histograms and box plots to examine the distributions of our risk indicators (Figure 2). Here, we can easily identify any statistical outliers colored based on the indicator selected in the Risk Indicators column switcher (seen to the left in Figure 1). Further, we can change figures to a log-scale for highly skewed variables, or examine tables of quantiles and other important summary statistics.

Figure 1. Site-level risk indicators

Figure 2. Histograms and box-plots of site-level risk indicators

View risks geographically
Perhaps environmental differences at certain locations lend themselves to increased safety issues. Sites within certain areas may have received training from a particular vendor. Different countries may have unique regulatory obligations. Monitors may be responsible for different sites based on their location. We can view these risks geographically (Figure 3). Sites are colored based on the currently selected variable in the Risk Indicators column switcher, so it is straightforward to cycle through all indicators to identify any patterns. Some country-level maps are also available (Figure 4). Further, using the local Data Filter, we can subset indicators based on site-level characteristics, such as trial monitor.

Figure 3. Global map of site-level risk indicators

Figure 4. Site-level risk indicators in Great Britain

Of course, all of the above visualizations are available for the country-level risk table. However, maps differ in that risk is defined for the entire country (Figure 5). This can further identify any problems that may be localized to particular regions, potentially suggesting areas that can benefit from additional training. Country-level risks can also be changed using the Risk Indicators column switcher.

Figure 5. Global map of country-level risk indicators

Cluster sites with similar risk
This highlights some of the analysis and review techniques that will be forthcoming in JMP Clinical 5.0. Of course, there is plenty of additional analysis that’s available; JMP and SAS are at your fingertips, after all! For example, Figure 6 shows a hierarchical clustering based on the site-level risk indicator table. The analysis identified five clusters displayed using differing symbols. The coloring corresponds to the overall risk indicator. In most cases, this more statistical approach corresponds pretty well with our overall risk indicator. This can identify sites with similar problem areas and potentially help rank sites for earlier review.

Figure 6. Hierarchical clustering of sites

tags: Clinical, Clinical Trials, Data Visualization, JMP 11, JMP Clinical, Life Sciences, Risk-Based Monitoring

2 Comments

  1. Mike Clayton
    Posted November 6, 2013 at 10:23 pm | Permalink

    Thanks. Merger of Healthcare and Technology is finally starting to make sense, and progress, just wish FDA terminology was aligned with rest of scientific methodology. Guessing you can really help in that regard. Why not a book mapping FDA terminology and methods to Industrial DOE and Risk Assessment?

    • Richard Zink Richard Zink
      Posted November 7, 2013 at 11:26 am | Permalink

      Hmm... a book is a great idea!

Post a Comment

Your email is never published nor shared. Required fields are marked *

*
*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <p> <pre lang="" line="" escaped=""> <q cite=""> <strike> <strong>