# The importance of age confounding in the medical literature

Age is usually the most important confounder in clinical studies. This short video will explain what age confounding is and how to spot it if it’s present.

Franz Wiesbauer, MD MPH
4th Aug 2016 • 3m read

When comparing mortality rates between two or more different groups, it’s wise to check their respective age distributions. Otherwise, you might draw the wrong conclusions from your comparisons. This is called age confounding and is one of the most important concepts in epidemiology. There are a couple of simple strategies that will help you get a quick grasp of whether age confounding is present or not. One of these strategies is to look at age-stratified mortality rates. This video from our Epidemiology Essentials course is a primer on age confounding checking to see whether or not it is present.

## Video Transcript

[00:00:00] Hey, everyone. I'm heading over to my studio to record a video for you and that video is about age confounding. That's a super important concept in epidemiology and absolutely crucial if you want to understand the scientific literature. So, enjoy. We often use mortality rates in order to compare their development over time or between communities. This is important when evaluating the effectiveness of an intervention, a before-after comparison so to speak and for resource

[00:00:30] allocation. If we have to decide if one community should get the resources over another one. One problem with comparing crude mortality rates between time periods, communities or whatever groups you're comparing is that these groups usually have a different age distribution and age is the strongest predictor of death. Let's compare two populations again. City A has a population of 1 000 000 inhabitants and so does city B. in city A, 1588 people died, whereas in

[00:01:00] city B 995 people died. Let's calculate the death rates per 100 000 population. 1588, divided by 1 000 000, times 100 000, equals a death per 100 000 of 159. And 995, divided by 1 000 000, times 100 000, equals a rounded death rate of 100 per 100 000. So, it looks like the mortality rate in city A is much higher than in city B. In other words,

[00:01:30] the risk of death seems to be much higher in city A but we said that this calculation does not factor in the age distribution of both cities. So, let's look at the age-specific death rates in both of them. Now, you see a very different picture. The age-specific death rates are actually higher in city B for all age strata. nineteen versus fifteen for the age group of 20 to 39-year-olds, 140 versus 120 for the 40 to 59-year-olds, and 270 versus

[00:02:00] 250 for the age group of 60 years and above. So, why in the world is the overall mortality rate higher in city A, whereas all age-specific death rates are higher in city B. Well, because of the age distribution. As you can see, city A is way older. 50% are 60 or above, one quarter is 40 to 59, and one quarter is 20 to 39. In city B, on the other hand, only one fifth is 60 or above, one quarter is 40 to 59, and over half are 20 to

[00:02:30] 39. So, actually, the risk of death is higher in city B but because of the older age distribution in city A, more people died in city A, which is reflected in the overall mortality rate. So, the key learning of this lesson is whenever you want to get a quick grasp of the effect of age on your rates, compare the age-stratified rates. In fact, you could do this for any potential confounder. If you thought that gender was confounding your overall rates, compare the gender-stratified

[00:03:00] or gender-specific rates. If you thought that income confounded your rates, look at the income stratified rates and so forth. Okay, so I hope you liked the video. I'm off to grab a coffee and if you like the video please leave a comment. If you didn't like the video, also leave a comment and I will talk to you soon. Take care.