Requirements

1. Spotfire DXP version 1.0 or later.

Instructions

1. Right-click on the following link and save the sample data file to your computer: TrialMetrics.csv

   Here is a snippet of this file:
   
   Notice the format of the data: Each row represents a single clinical trial at a single site.

2. Open Spotfire DXP. Drag and drop the TrialMetrics.csv file onto the DXP window. Spotfire shows you the Import Settings dialog box. Click OK.

   Spotfire reads in the data, automatically recognizes all the data fields, and creates an initial scatter plot of SITE_ID vs. TRIAL_ID. Spotfire also creates filters along the right for each data field.

3. We would like to see our trial sites displayed on the scatter plot with respect to screening failure rate and withdrawal rate.

   On the X axis click on the arrow that's part of the "TRIAL_ID" label, and select WITHDRAWAL_RATE.
   On the Y axis click on the arrow that's part of the "SITE_ID" label, and select SCREENING_FAIL_RATE.

   Each square on the scatter plot represents a specific trial at a specific site.

4. Now let’s create a another visualization which will show us the geographic distribution of the trial sites.

   Click on the Bar Chart icon in the toolbar.

   A Bar Chart is created above the Scatter Plot.

   On the Bar Chart Y axis click on the arrow that's part of the "Sum(TRIAL_ID)" label, and select Aggregation->Count.

   The height of each bar now shows the number of trial sites in each therapeutic program.

   On the Bar Chart X axis click on the arrow that's part of the "Program" label, and select Country.

   The Bar Chart automatically responds by showing us the number of trial sites in each country. The US leads with over 500 trial sites, with the remaining countries having roughly equal numbers.

5. Observe how the 2 chart interact with one another.

   Click on the any of the trial sites on the Scatter Plot.

   Notice that not only does the square turn dark green, but so does the bar for the corresponding country in the the Bart Chart.

   Click on any of the bars in the Bar Chart.

   Notice that not only does the bar turn dark green, but so do all the trial sites in that country in the Scatter Plot.

6. Let’s now analyze our data.
   

   We first see that there are two trial sites which have screening failure rates greater than 100%. Since this is impossible, the data management team should be alerted to this. Click on each of these squares, and note the details about the site in the Details-on-Demand window in the lower right corner.
   
   We will now focus in on our top performing sites; sites which have completed at least 10 subjects and have a screen failure and withdrawal rate of less than 30%.
   
   Find the WITHDRAWAL_RATE filter. Move the right slider to 30.
   Find the SCREENING_FAILURE _RATE filter. Move the right slider to 30.
   
   Both the Bar Chart and Scatter Plot respond by showing graphs for the corresponding subset of data.
   Overall the distribution in the Bar Chart looks similar to before, though Sweden looks perhaps a bit better than the other countries.
   
   Find the COMPLETED_TREATMENT filter. Move the left slider to 10.
   We have filtered away most of the trial sites and are left with 150.
   
   The geographic distribution of our top performers now looks different than that of all of our sites. Sweden jumps far ahead of all other countries. The Swedish trial sites significantly outpace the other countries in completing subjects.