The Algorithmic FDA Medical Query (aFMQ) Risk Report uses FMQs with associated algorithms to calculate risk measurements for each query. This report requires an Excel file with the FMQs. The download link and instructions for saving this file can be found here.
Note: this video was recorded in JMP Clinical 18.0 (with the Review Subject Filter portion of the report hidden). If you are using JMP Clinical 17.2, there is no option to separate sex-specific aFMQs. Please be aware that running or re-running the report is considerably more time consuming in 17.2 compared to 18.0.
Transcript:
This video demonstrates the use of the Algorithmic FDA Medical Query Risk report, or aFMQ Risk report, using the Nicardipine data set that’s installed with JMP Clinical. This report takes advantage of both the FDA Medical Query mapping of preferred or dictionary derived terms for adverse events, as well as the mapping of the algorithmic components associated with these queries.
The aFMQ report contains two components for comparing the treatment groups: a risk plot at the top of the report, and a table of values below.
Let’s start by looking at the risk plots. In this example data set, we’re comparing two treatment groups, Nicardipine and placebo, but the report also supports data sets with more than two groups.
The percent occurrence for each query is shown in the dot plot on the left, with color-coded markers for each group. You can see the actual percent occurrences by hovering your cursor over any marker.
The forest plot on the right shows the risk difference for each query, and includes the 95% confidence interval on the difference. In this plot, too, you can hover your cursor over a marker to view the value, as well as the lower and upper values for the 95% confidence interval.
The forest plot of the risk difference also includes a reference line at zero. Risk difference measures the risk of an outcome for subjects in the treatment arm compared to subjects in the control arm – in this example, Nicardipine versus placebo. The farther the value is from zero, the greater the risk in one group over the other. Positive values indicate greater risk in the treatment group, while negative values indicate greater risk in the control group.
However, if the 95% confidence interval on the risk difference includes zero, then we can’t conclude that the difference isn’t zero, based on this level of confidence. So, in the Nicardipine study, none of the algorithmic medical queries would be considered an elevated risk.
All of these numbers are also given in the tables below the plots.
In the Options panel on the left side of the report tab, I’ll expand the Data section by clicking the disclosure icon. This section lets you specify which data are included in the report, or how the results are calculated.
For example, you can choose which types of adverse events to include, and whether to ignore treatment emergent flags, specify the risk measurement, select a treatment value to sort by if there are multiple treatments, and change how the baseline is calculated, which might impact how some of the algorithmic components of the aFMQs are determined.
Making changes to any of these options will cause the report to be re-run and updated.
The Display section of the Options panel contains various data filters for the report. You could choose to just display the results for Hyperglycemia, or, for the numeric filters, drag the sliders, or manually enter lower or upper values. If there are criteria in your data not included in the default filter list that you’d like to use, you can click the And button and choose from the available columns.
You can see that selections in these filters update the view of the data, but don’t re-run the report, because nothing is being recalculated.
In summary, the aFMQ risk report provides graphical and tabular output to expedite finding possible hidden safety signals within the clinical trial study data by using the algorithmic components of the FDA Medical Queries to go beyond a simple mapping of medical terms.
For more information about this report in JMP Clinical, please click the question mark icon in the upper right corner of the report tab, or contact support@jmp.com.
