Attributable Risk Ques

[hartbot]

Can someone please explain to me the diff between attributable risk and absolute risk reduction?

---

Members don't see this ad.

 

[knuckles]

It's only semantics specific to what you're researching.

If the study is investigating risk factors, eg unprotected sex and AIDS, then you would determine attributable risk. Then 1/AR would be the number needed to harm, with this risk.

If you're studying a treatment, eg an antiretroviral drug, then you would determine attributable risk reduction. Then 1/ARR would be the number needed to treat with the drug.

 

[aashkab]

i hope they don't go into semantics for crap like this because i just know how to solve problems using what is now some kind of mathematical common sense, but i have no set formulas, semantics, or anything. i really hope it doesn't screw me

 

[knuckles]

Come on now, biostats isn't bad at all. If you can work through UWorld's problems, and understand how to do each one, you'll be more than fine. Based on the recent feedback, the biostats questions being asked aren't much more than plug and play.

 

[PokerDoc]

I cant even think of a scenario where youd have to decide whether you are calculating attributable risk and absolute risk reduction.

its either one or the other, and the formula is exactly the same, if you dont know what youre calculating u can still get it right because the formula is the same.

same applies for NNH or NNT

 

[doc29]

same applies for NNH or NNT

Can someone please explain to me what scenerio do we use NNH and NNT.
My problem is i read the question, and i dont know what formula i should use(especially those ARR,NNH etc). any help would be appreciated!

 

[knuckles]

See my post. If that doesn't help, then learn english.

 

[doc29]

See my post. If that doesn't help, then learn english.

sorry knuckles..i didnt see your post earlier. Anyway, thanks

 

[strongboy2005]

According to Wikipedia, absolute risk reduction is experimental event rate minus control event rate (EER-CER).
Attributable risk is total experimental events/total events minus total experimental non-events/total non-events [EE/(EE+CE)]-[EN/(EN+CN)].
These come out to different values, and I still do not understand what information they are trying to give.

For instance, lets say the study is looking at two groups of 100 people each. The experimental group was exposed to asbestos, the control group was not exposed to asbestos. 25/100 in the experimental group (asbestos-exposed) got cancer (EER=.25), while only 10/100 in the control group (asbestos non-exposed) got cancer (CER=.10).

Calculations:
Absolute risk increase = .25-.10 = .15 = 15%
Relative risk increase = (.25-.10)/.10 = 1.5 = 150%
Number needed to harm = 1/(.25-.10) = 6.67
Relative risk = .25/.10 = 2.5 = 250%

These calculations all seem right so far. Then Wikipedia gives this equation for attributable risk: AR=[EE/(EE+CE)]-[EN/(EN+CN)]
So, using my example above, EE=25, CE=10, EN=75, CN=90.

Attributable risk=(25/(25+10))-(75/(75+90))=.2597=25.97%

Is this equation wrong? Is the attributable risk really just the absolute risk increase or is it something else entirely?

 

[clivuslives]

I think this all in terms of calculating the NNT (numbers needed to treat) or NNH (numbers needed to harm). The absolute risk reduction (ARR) and the attributable risk (AR) are basically the same thing, just taking the absolute value of the result of one to get the other:

NNT = 1/ARR; where the ARR = Control Event Rate (CER) - Experimental Event Rate (EER)

NNH = 1/AR; where AR = EER - ARR

Basically, if you have a negative NNT, then you are going to harm people. If you have a negative NNH, you are going to help people.