Our World Statistics Day conversations have been a great reminder of how much statistics can inform our lives. Do you have an example of how statistics has made a difference in your life? Share your story with the Community!
Choose Language Hide Translation Bar
It’s time to "crossover" to JMP Clinical 4.1

My colleagues are often surprised to hear I went through five majors in college to find my niche in statistics. I started as a physical therapy major, which quickly turned to nuclear medicine, communications, psychology, and finally late in my junior year I became enamored with the beauty of statistics and its real-world applications. I like to think that each period of study helped me grow in a different way. I guess you could say that through my experiences I acted as my own “control” in order to judge what fit me best. I guess you could say that inner yet-discovered statistician in me was already performing an observational crossover study... see where I’m going with this?

When I say it’s time to “crossover” to JMP Clinical 4.1, I don’t just mean that we have an amazing new release out that you should get your hands on. Now, with one of the many new features in the latest JMP Clinical version, we have enabled you to analyze crossover trials with your clinical data! JMP Clinical now automatically detects and performs crossover analysis when a study is added that contains the standard timing variables and treatment period variables necessary for crossover support in the ADSL table per the ADAM Implementation Guide.

Hopefully, your crossover experiment is better designed than my observational college "study." Key characteristics of a well-designed crossover study include balanced patient randomization to sequences of treatments and sufficient wash-out time between treatment periods with the purpose of studying individual treatment effects. Given these principles, statistical analysis can become straightforward; often instead visualization and interpretation of the data is a complicating factor.

Consider a simple two-period AB/BA crossover design. Using mock data from our example Nicardipine study, assume we have a crossover trial for safety analysis with 24 patients. Patients were randomized to receive either Nicardipine (NIC .15) or placebo in the first week and vice versa in the second week. New views and options in the Findings Time Trend Analysis process support crossover visualization to easily see the trends in vital signs (for example) across treatment periods.

With this visualization, it’s easy to see when the crossover occurred and follow the trends of average findings measurements by treatment during each period. The option to pool subjects in trends plots when treatment crossover is detected results in a view that looks much like the standard trend view, except that the subjects that comprise the trend line change depending on the treatment administered during the period the findings test was taken.

Further options to overlay visits across treatment periods (not applicable with this example but a common request when data is collected across cycles) allow for more customization to visualize/compare data with crossover. Trellis plots now showcase individual patient trends across treatment periods.


Findings Time Trends are not the only features of JMP Clinical that support crossover. Medical monitor distribution views for safety domain events, interventions and findings assign the appropriate treatment period and value based on the date/time of the recorded event/intervention/finding measurement. An example of how the AE distribution counts graph can be customized to categorize occurrence of adverse events by treatment period is shown below.

For the more statistically minded, JMP Clinical Incidence Screens support crossover through stratified analysis by patient to compare and report p-values on incidence rates that honor the paired data framework. Results are similar to the typical Incidence Screen volcano plot, where the p-value for testing significant differences in incidence rates (for adverse events for example) is from the Cochran-Mantel-Haenszel (CMH) statistic to account for subject strata. The Breslow-Day test for homogeneity of the odds ratios is also reported in the table.

New features like these are not available in other software tools out there, and I strongly believe that no tool can compare to what JMP Clinical has to offer. Personally, every release is an exciting and rewarding time for me when I can see what our development team was able to create and know the impact it can (and will) make on the industry. It reminds me that I made the right “crossover” to study statistics back in college.

By the way, it is pure coincidence that I began this blog talking about college…and that “Oh the Places You’ll Go,” which was my inspiration for previous posts on Findings Analysis and Drug-Induced Liver Injury, just happens to be the most common gift given to graduates. It really is a coincidence, but hey it was a great excuse to cross-reference those blog posts so you can learn more about new JMP Clinical 4.1 features!

Article Labels

    There are no labels assigned to this post.


AKIO wrote:

Can JMP Clinical 4.1 deal with "3x3" and "4x4" crossover design?

A "4x4" crossover design is typically used for QTc sutdies.


Kelci Miclaus wrote:

Yes it should be able to handle 3x3 and 4x4 crossover as well. For QTc studies, for example, the software would assign the current treatment based on the time of the visit and you could visualize the results through the Findings Time Trends domain for the EG data set. The implementation is intended to be flexible design complexity.