I am a rising third-year PhD student (in the group of SOT member Rita Strakovsky, PhD, RD) researching the impact of exposure in pregnancy to endocrine-disrupting chemicals (EDCs) on birth outcomes and maternal health. I was thrilled to be awarded the SOT Supplemental Training for Education Program (STEP) award.
Using the funds from this award, I had the opportunity to complete a rigorous, one-week longitudinal analysis course offered by the University of Michigan. This course was important to my advancement as a toxicologist because longitudinal analysis is a noteworthy component of conducting environmental epidemiologic research since it allows us to investigate chemical exposure at multiple timepoints throughout pregnancy to understand their impacts on birth outcomes and maternal health. Furthermore, knowing this method is important to my career development, as my goal is to study maternal and fetal health.
I selected the University of Michigan Summer Session in Epidemiology “Analysis of Longitudinal Data from Epidemiologic Studies” course because my institution does not offer a course dedicated to this method of analysis. I also felt this was a necessary step in my development as a researcher and toxicologist because humans are continuously exposed to chemicals, and concentrations in the body fluctuate over time. The course was instructed by Dr. Ananda Sen, a biostatistician at the University of Michigan. I was excited because the course would be taught in person, which would provide me with the opportunity to be more engaged with the material, and the classroom setting would eliminate distractions. The goal of the course was to explore different statistical methods for conducting longitudinal analyses by understanding the underlying theory behind the equations necessary, incorporating the element of time, and accommodating correlated data.
During the five days, I became comfortable with the base equations and interpretations required for conducting longitudinal analyses, the difference between random and fixed effects, and the importance of covariance structure. We conducted analyses using cleaned data sets in SAS Studio to practice implementing the provided code and understanding each of the resulting tables and figures. These instructor-provided code and instructor-led lab sessions were an important aspect of the course because I learned where each variable fits into the equation and how changing parameters and covariance structures can alter the resulting model. Furthermore, each day of class consisted of review poll questions to determine the information retained from the previous day and what content needed to be revisited—this was an excellent component of the program as it ensured students were understanding the material. Finally, three homework assignments were given to encourage students to practice the instructor-provided class code on new datasets and practice interpretation of results.
I am grateful for the opportunity to take this course with the help of the STEP award. It was a great opportunity to dive deeper into statistical methods that I need to develop as a researcher and that I need to understand in order to conduct analyses when looking at chemical exposure across various timepoints. Furthermore, I have a better understanding of model selection, building, and comparison. Finally, I am more confident in my ability to interpret and share the resulting estimates in the model, which is critical for evaluating the effects of chemicals on health in the population from an epidemiologic standpoint.
Editor’s Note: All SOT Student members are eligible for the STEP award, which can be used to pursue opportunities like the course described in this blog. The next STEP application deadline is October 15, 2023.