The role of adjustment in age confounding

Want to know how age adjustment works? Watch this short video and become a masterful interpreter of the medical literature!

Franz Wiesbauer, MD MPH
Franz Wiesbauer, MD MPH
16th Aug 2016 • 2m read
Loading...

As we have already learned in one of our previous videos, age confounding is one of the most important phenomena in epidemiology. One powerful tool to account for age confounding is called age adjustment. In this video, you will learn what age adjustment is and how you can use it the next time you evaluate a scientific study.

As always, please leave your thoughts in the comments section below.

Video Transcript

[00:00:00] I just came back finishing another video fresh off the press. I guess you don't press a video but it doesn't matter and it's about age confounding and how you can combat it with age adjustment. Enjoy. In the previous lesson, we learned that comparing stratified rates can help you with confounding. Another method that's often applied when comparing rates, in different populations, is called adjustment. Let's go back to a previous example of death rates in cities A and B. Remember, the overall death rate was higher in city A,

[00:00:30] due to its older age distribution, whereas the age-stratified mortality rates were actually higher in city B. So, the risk of death is really higher in city B. Now, in order to arrive at rates that are free from age confounding, we could apply the age-specific death rates to a standard population. We could either create a fictitious one or just use one of the two cities. Let's choose city A as our standard population. This is the age distribution, these are the numbers of deaths, and this was our death

[00:01:00] rate in city A. Now, how many deaths would we expect, if the age distribution was the same in city B, as in city A? To answer this question, we need the age-specific death rates of city B and apply them to the age strata of our standard population. So, 19 per 100 000, multiplied by 250 000, equals 48 expected deaths, 140 per 100 000, times 250 000, equals

[00:01:30] 350 expected deaths, and 270 per 100 000, times 500 000, equals 1350 expected deaths. So overall, we'd expect 1748 deaths if city B had the same age distribution as city A. And what would be the overall death rate per 100 000 population? That's 1748, divided by 1 000 000, times

[00:02:00] 100 000, so 175. This means that the age-adjusted death rate using city A as the standard population would be 175 per 100 000. Now, we can compare that to the crude death rate of city A of 159. And what we see is that after accounting for age, the death rate of city B is actually higher than in city A. So, the same findings as with the stratified rates. So,

[00:02:30] adjustment is another powerful tool to combat confounding. So, I hope you enjoyed the video. As always, please let me know what you think in the comment section below.