PSI Journal Club: Limitations and Challenges with Mixed Model Repeated Measures (MMRM) Analysis
Moses Mwangi and Florian Lasch present their recent work on MMRM, with discussion lead by Geert Molenberghs.
Every new medicine that is invented must be thoroughly tested to prove that it is not only an effective treatment for a health problem, but that it is safe for patients and does not have any side effects that are too severe.
We do this by conducting clinical trials where patients are administered the experimental new treatment and are compared to a control group of patients administered a placebo or an existing treatment.
Before we go ahead and give patients treatments, we need to make sure that we have robustly designed our trial to give us the best chance of being able to demonstrate that the experimental new treatment is superior to the comparator control treatment.
Their skills in maths and statistics allow them to make very valuable contributions to the clinical trial. Explore below just some of the activities they do in their day to day jobs.
The statistician has some very important jobs to do when designing a clinical trial. To start with, they must calculate how many patients are needed for the trial. We call this the sample size.
It is the statistician’s job to allocate patients to the new treatment or to the placebo or control treatment. Treatments must be allocated at random so that the two groups are similar in every way except for the treatment they are given.
You can do a simple randomisation by tossing a coin and assigning all those who get a head to the control group and all those who get a tail to the new treatment group.
Statisticians have a crucial job to make sure that data collected during the clinical trial is analysed using appropriate methods and is interpreted correctly.
They carry out statistical tests to weigh the evidence held in the data and determine whether the new drug is beneficial for patients.
As well as their statistical qualifications, statisticians need to have good communication skills. They must interpret the results of statistical analyses and explain to non-statisticians, such as the medical team, what the results mean in layman’s terms.
Statisticians are often called upon to give input to decision-making. Their logical thinking and understanding of the whole clinical trial from data collection to analysis results is very valuable.
Statisticians and programmers work very closely together. In fact, in some companies these are not two separate jobs. Both need to be proficient in programming languages such as R and SAS.
Programmers are experts in turning datasets into summary tables and graphs. The statistician and programmer collaborate to decide how the tables and graphs should look and what they should contain, and then the programmer makes these a reality.
Programmers need to be skilled in paying attention to detail and they must carefully document their work so that analysis results are traceable from the raw data. Programmers must also be very organised and able to meet deadlines.