Determining Causality: A Review of the Bradford Hill Criteria

Bradford Hill develops several criteria that you shold consider as you try to determine if an association seen in a study is causal or not

[sbasra]

Thanks for putting this video up, I really needed someone to provide examples and explain what each of criteria are.

[wizepower]

thank you very much, actually, i just had an NBME question that I didn't know how to even interpret, after watching this video I can now understand what the choices meant and how to eliminate and reach an answer. grateful

[radykal]

Tanx for this video.Really broadened my understanding after my lectures today.

[peterstrous]

@Terry Shaneyfelt - The interpretation of "Specificity: effect has only one cause" is different to the way I read the BHC. My understanding is that out of all possible causes considered, this particular one has the strongest association, i.e., it is more Specific than any other possible causes considered. I see Specificity and Strength going hand in hand, the Stronger the association, the more Specific the association compared to others, the more likely causality. They are part of the same perspective.

[JasDimacali]

Well explained lesson sir! Helps me understands with clarity.

[ganxtashiz]

thanks for the video! In my uni course, they had a different one which was 'reversibility' meaning if you take away the risk factor you are likely to reduce the amount of disease.

[Jackies]

Specificity on the other hand, IS actually a pretty important criterion... Bradford Hill also acknowledges that the more fine-grained a consideration or correlation is in view of the supposed effect, the higher the likelihood that we have reached the EXACT cause... and this is, in nature, rarely "multifactorial" actually... take HIV as the cause of AIDS, the philadelphia chromosome for leukemia, or h. pylori (wiping away all acid, stress, anxiety, spicy food, etc. hypotheses for peptic ulcer disease...)

here is what Hill actually says: "Indeed I believe that multi-causation is generally more likely than single causation *though possibly if we knew all the answer we might get back to a single factor*." [my emphasis]

Strength of association, however, is, also explicitly according to Bradford Hill, pretty dispensable, just as is the biological gradient (as mild exposure -- highest disease rate is always theoretically possible as well)...

I can't help wondering how that paper can be so often and so much misunderstood...

[atifkatib]

consistency has been explained wrong. simply it is (repeatability) when replication of the findings by different investigators, at different times, in different places, with different methods and the ability to convincingly explain different results.

[Jackies]

Nonsense! There need NOT be a biological plausibility or mechanism for the association. Bradford-Hill clearly states this, too... Having a mechanism is limited by current knowledge! Also it is worth nothing to just have a theoretical idea how something COULD work... Even if there is a mechanism that you can investigate, statements about it will themselves always depend on observation and a statistical evaluation of them or this mechanism in relation to the association in question! Plausibility/Mechanism is completely dispensable in any good philosophy of science that aims to make sense of the world....