1/ I said I won’t be spending too much time on Twitter this week but I feel I need to point something out at this time given the headlines about covid mortality in different occupational groups. It’s not as simple as it seems. Here’s a quick

2/ Occupational groups may differ in their characteristics for various cultural, social and system reasons. For example, in a certain country there may be more people on average from ethnic minorities in the occupational group of doctors compared to teachers.

3/ Now if we want to look at how job type affects health or mortality then we must account for differences in characteristics if they are linked with *both* the exposure of interest (job type)& outcome of interest. They are called ‘confounders’ bec they confound the relationship.

4/ For example again, if we already know that death from covid is linked to ethnicity, having diabetes or obesity, and we know that these factors also tend to be higher or lower on average in teachers compared to doctors then..

4/..if we don’t account for these factors, we may wrongly attribute higher or lower deaths to the job type itself while they just happen in that occupational group because its members have on average different characteristics to the comparison group or the whole population.

5/ so we then say having a certain job type is *associated* with higher or lower mortality but not necessarily causing it because we need to account for all the confounders that are *unequally* distributed between the comparison groups in order to say that.

6/ age & sex are one of the most common factors that are usually taken into account in such comparisons which is right but that doesn’t fix the problem. It’s actually not a problem if more understand the issue of confounding & interpret such stats as associations not causes. End.

*Age & sex are two obviously not one .. and I really I wish I can do one thread without typos before I die

There’s even a typo in the typo correction. I give up