This is archived information for Stat 445 Sect 201 (Spring, 2005).
BHAT Sample Report
This sample clinical trial report (link given below) was produced by the Statistical Data Analysis Center (SDAC) at the University of Wisconsin–Madison. It's based on real data from a clinical trial of the drug propanolol. The point of the trial was to determine if regular use of this drug in patients who had suffered at least one heart attack would lead to increased survival.
In real clinical trials, there is usually an independent board of statisticians and clinicians that monitors the trial. They are prepared to stop the trial if it becomes apparent that the drug is doing more harm than good (which happens often), if the drug is hopelessly ineffective and there's no point in continuing the trial (which also happens quite often), or if it is so obvious the drug is effective that it would be unethical to deny placebo patients the drug any longer (which happens once in a blue moon). Reports like this are used to communicate large amounts of information about the ongoing trial to this monitoring board, so they can make an informed decision. Also, after the trial ends, a final report like this will probably become part of the final submission by the drug company to an appropriate governmental agency, like Health Canada or the U.S. Federal Drug Administration.
I would not suggest trying to copy the layout or format of this sample report, and I obviously don't expect anyone to hand in a 62 page project. I'm just showing you this sample to give you some ideas about what kinds of plots and statistical tests are used to present clinical trial data in the "real world" so you can get an idea of where to start for Project #2.
As you can see, the statistical analysis in this sample report is quite basic: there are boxplots and barplots of various factors by treatment group over time, and these are backed up by simple Wilcoxon rank-sum and Pearson chi-square (though Fisher exact can also often be used) p-values. In addition to these sorts of analyses, I hope that you will find at least one or two more "exciting" analyses to try in your own projects, based on some of the estimators and bootstrapping and permutation test techniques we have studied in class.
- BHAT Sample Report (PDF file, 212k)
This is archived information for Stat 445 Sect 201 (Spring, 2005).