Thread by @MedCrisis

A little tidbit for non-cardiologists who might not've heard of the Smoker's Paradox. WHY INCOMPLETE DATA CAN MISLEAD!

Yrs ago it was observed that although smokers had more heart attacks (MIs), they actually did *better* after having had one, compared to a non-smoker. wtf?

Can you think of any reason why this might be the case?

Forest plot showing adjusted odds ratio for death in smokers following MI. Towards the left & not crossing midline = favouring the Smoker's Paradox.

Reporting this finding alone is what newspapers do with medical trials today. The full picture is that smokers have their heart attacks YOUNGER, and hence have fewer comorbidities & were treated more aggressively by medics.

This is a confounder.

Another is that the effect was only seen in hospitalised patients. Smokers die more often than non-smokers before they reached hospital, so that the ones being recorded in trials are skewed to lower severity.

This is survivor bias.

One can try to correct for confounders, but by their nature, they're not always apparent. Some claimed that even correcting for things like age, the paradox seemed to persist.

However, fewer trials reported it as time passed from the 80s to the 90s & 00s. It 'disappeared'.

Treatment of heart attacks changed during this time. The paradox is only seen in the era of thrombolysis (clot-dissolving medication), which has been replaced by primary coronary angioplasty.

It's thought perhaps the type of clot was different in a smoker than non- so the thrombolysis might've been more effective. When the clot is treated with a balloon and a stent, these subtle pathophysiological changes are clearly not relevant.

So when you read a surprising headline suggesting something odd like vegetarians have more strokes or obesity being 'protective' - there may be some truth to it but always make sure you have the full picture. Which may be easier said than done!