I want to say like if smoking and Ca bronchus has OR of 2.3, then what does it actually mean? Is it 2.3 times risk? Or it is only positive association! During writing the discussion can I write smoking has 2.3 times risk or it has positive risk only ? What to write when OR is
Hi Casper and Rajat,
ex: odd ratio= [p11/p21](1-p21/1-p11), while RR = p11/p21.
They point in the same direction but are not the same as you can see. They use diferent formulas of calculation. But they are related (file attached before).
ex: OR = 2,3 means that smokers have 2,3 times more opportunity of having cancer than no smokers.
When OR = 2.3, then smokers have 2.3 times more risk of getting Ca Bronchus than non-smokers. Or, equivalently, people with Ca Bronchus are 2.3 times more often a smoker. It is indeed positive association, not causation.
When OR < 1, then - in your example - smokers would have *less* Ca Bronchus than non-smokers.
Hi Casper and Rajat,
ex: odd ratio= [p11/p21](1-p21/1-p11), while RR = p11/p21.
They point in the same direction but are not the same as you can see. They use diferent formulas of calculation. But they are related (file attached before).
ex: OR = 2,3 means that smokers have 2,3 times more opportunity of having cancer than no smokers.
Hi Rajat.
The difference between OR and RR is the order in which the exposition and the outcome are studied, in Epidemiology you can control this in the design of your research.
If you follow a cohort of health people controlling who were exposed and who were not exposed you can make a probabilistic relation with the outcome (Relative Risk or RR), as Helena shows in her answer, is a probability division, then you can conclude which is the probability to develop some disease (outcome) when you are exposed to a risk factor.
In other hand, Odds Ratio (OR) is an association measure between the disease and the exposition; in this order you study a group with the disease (and a healthy control group) and look for the exposition. Then the OR can be interpreted once a disease is diagnosticated, how many times you have the chance that this was for the exposition, you can see that OR is a relation measure. For that reason OR can be estimated (cautiously) in prevalence studies, but RR no.
Best regards,
OR is a function of RR and absolute risk (not a pretty one, BTW) ; equivalently, RR is function of OR and absolute risk. For *rare* events, however, OR is an acceptable approximation of RR (and vice-versa). Federico is right to point out that OR and RR have different domains of "ease of use". but both express the same supposed (direct or indirect) link between two observed events.
the Wikipedia-page on odds ratios is very good. I suggest consulting that for your question.
https://en.wikipedia.org/wiki/Odds_ratio
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics