Azad - longitudinal studies

So longitudinal studies are really important. They allow us to get at temporality. So because you’re starting in the case of the CHILD study, we start from before birth. But even if it’s an adult cohort study, whenever you start, you’re following forward in time. 

So you can see what came first. You know, sometimes in a cross-sectional study it’s difficult to know. So I’ve done some research on artificial sweeteners for example. If you just do a cross section and you ask a bunch of people, Do you consume artificial sweeteners and you measure their body composition, you may find that people who consume more artificial sweeteners tend to have more obesity. But what came first? Is it that they were obese and they’re trying to lose weight so they’re using these products? Or is it that they’ve always consumed these products and that changed their microbiome or changed their metabolism and contributed to their obesity? 

That’s really hard to know in a cross-sectional study, and you can ask people to say, how long have you been doing this? And recall it? But recall’s not always perfect. When you do a longitudinal study, you can start from the beginning and ask people how many sweeteners are you consuming a day? 

And then follow up with them later on and say, okay, now how much weight gain or loss have you experienced? So you can really know what came first. And that’s important for determining causality. In the case of a birth cohort, which is a special case. I think these are so valuable because we can capture data from that critical period of pregnancy and early life and then follow these children forward in time to understand how those early life experiences have affected their development later on. 

And so, you know, ideally, if money was no object and time was no object, we’d only do longitudinal studies. But of course, they take a long time because you’ve got to follow people over time and that costs money. 

And you come into challenges with, for example, retention. So some people lose interest in the study or you can’t find them and then you have missing data. So of course there are pros and cons to every study design, and the key is to understand what is your research question and then what’s the ideal design to do for that purpose?