What's the optimal statistical technique in a variety of health settings?
Michael Rotondi’s research program develops advanced statistical methods for practical problems in health research. His expertise as a biostatistician puts him in high demand with collaborators who are exploring an assortment of timely health issues, from Alzheimer’s disease to diabetes.
Many smart young people are told that they should be a doctor. I was always very strong in math and statistics, and I’ve been able to combine them with aspects of human health. I was attracted to biostatistics instead of economics or business because of this potential to contribute to health research. I enjoy it even more so than I expected.
I like the idea that, at some level, my statistical research contributes to improving people’s health.
As a biostatistician, the techniques I develop are used in a variety of health projects. I’ve been contacted by researchers from all over the world who have made use of my techniques in their own work. I develop a method, but I don’t know where it’s going to end up – people can modify it, change it, and apply it to all sorts of studies.
For example, I’m currently part of a team conducting a randomized controlled trial to test the effectiveness of an iPhone-based diabetes management system for teenagers. The study will tell us whether or not the teenagers who are using the iPhones have better management of their diabetes then those who didn’t use them. I’m the statistical lead on a team of engineers and clinicians out of SickKids, which means I lead all aspects of the statistical analyses and ensure the randomization scheme and allocation is all appropriate and valid. At the end, I’m going to be the one who’s figuring out whether this thing worked.
In my doctoral work I developed a new statistical method, and used it to show that the prevalence of vitamin A deficiency is highly linked to the effectiveness of newborn vitamin A supplementation in the developing world. We would expect people who are more deficient to have a bigger benefit, but until I developed my new statistical model, there was a lot of controversy surrounding this idea since some studies showed a strong effect and others did not. My method provided a rationale for one of the potential explanations for what was going on, which in turn helps policy makers decide on how to approach vitamin A supplementation.
I also developed a new sample size estimation technique and software to assist researchers in determining how many subjects they need for reliability studies. The idea here is that it’s almost like a Goldilocks technique – you don’t want to recruit too many people or it’s a waste of time and money, and you don’t want to recruit too few or the study won’t be valid. Based on what the investigator wants to do, the software will help them find the “just right” sample size.
I’m currently working collaboratively with researchers at St. Michael’s Hospital on factors linked to diabetes and hypertension in the Aboriginal community, with a team of physicians and engineers at Sick Kids on an app to help teenagers manage their diabetes, as well as researchers with the Toronto Central Community Care Access Centre who are looking at the impact of caregiver support on the health of older adults. The diversity of these connections has provided me with a range of interesting projects. That’s one of the perks of being a biostatistician – I typically get to work with a variety of people on exciting projects.
I’m also currently working on developing and testing my new statistical technique for my CIHR operating grant. So we’ll soon be able to apply the technique to look at factors that are linked to diabetes and hypertension in the Aboriginal community.