Mavuto Mukaka: Statistics for medical studies
Medical statisticians help design studies, perform data cleaning and analysis, and interpret findings. Many methods are available, and statisticians help identify and make recommendations. Poorly designed and interpreted studies may lead to wrong conclusions, and statisticians help ensure better findings that can be translated into medical practice and policy change.
My name is Mavuto Mukaka, I am a medical statistician and I work at the Mahidol Oxford Tropical Medicine Research Unit in Thailand. My work involves designing research studies and performing data analysis, as well as the interpretation of research findings. I work in collaboration with medical researchers.
Within MORU, my team and I are statisticians. We are involved in the design of studies, working out the number of individuals that need to be enrolled in a study. We are also involved in statistical analysis of data: we perform data cleaning; we do data aggregation which is combining datasets from several sites. In the end, we make a report which we submit to investigators, and we do interpret the findings of the research. We also take part in manuscript development for publications.
Most recently, our team worked on a study that involved malaria elimination. Our role was to combine datasets, do data cleaning and then perform analysis. The study was done in four countries in Southeast Asia: Cambodia, Vietnam, Laos and Myanmar, with mass drug administration of dihydroartemisinin-piperaquine with a single dose of primaquine.
In our field, there are so many methods that have been developed in the area of medical research, and these include time-to-event analysis methods. It is the responsibility of us as statisticians in my team to identify these methods and recommend them to medical researchers. In the area of medical research, there are traditional methods of analysing data which are using proportions to assess efficiency or effectiveness. But more recently, researchers have recommended that we use survival methods, or time-to-event methods, which are able to capture issues that relate to missing data. These methods are not easily accessible to medical researchers; it takes statisticians to go to the literature, identify these methods and make recommendations to medical researchers. This is where we come in as medical statisticians. In the past, if you use proportions, you would end up losing some data because, if some people did not complete the follow-up period, they may not be part of the analysis. But this new methods allow those individuals that have partial information to be included in research or analysis.
Medical statistics is very important in research. A poorly designed study will not answer the research questions that a medical researcher intended to answer. Unfortunately, you cannot reverse or use any statistics to compensate for a poorly designed study. It means a loss of money. Similarly, if you use wrong methods to analyse your data, you’ll end up with wrong conclusions. Therefore, you need to involve a statistician, right from the design stage and throughout the conduct of the trial, up to reporting and interpretation of findings. This means that we need to make sure that statisticians are funded and included in any activity that involves medical research.
Medical research aims at using research findings to translate into medical practice and policy change. As statisticians, we are part and parcel of that research activity. We are therefore an important component of translational medicine, as statisticians.
This interview was recorded in September 2019