Interpretation of Odds Ratios

[rsstats]

I had a basic query regarding interpretation of Odds Ratio. This particular problem was sent to me by one of my colleagues. The risk factor association of different risk factors with pancreatic cancer was studied in a case control study.
The values are:

SITUATION 1:
Risk Factor Odds ratio (95% Confidence limits)
RF-A 2.5 ( 1.1 - 3.1 )
RF-B 1.4 ( 1.3 - 1.7 )

SITUATION 2:
Risk Factor Odds ratio 95% Confidence limits
RF-A 2.5 1.1 - 9.5
RF-B 1.4 1.1 - 1.7

I would be grateful if you could post a reply on what would be the correct
interpretation of the Odds Ratio in situations 1 and 2.

Regards.

[rsstats]

Hi...

Any possible answers?

Is it the OR value that is more important, or the Confidence interval lower limit and range?

[fed1]

Odds ratio is ratio of odds.

Odds( event ) = Pr(event)/Pr(not event).

OR = odds(event given condition one) / odds( event given conditon tow)

For you this means
OR = odds( p.cancer in group A)/ odds( p.cancer in groupB)

Odds ratio has many niftirific property that make scientist like it.

http://en.wikipedia.org/wiki/Odds_ratio

[rsstats]

@fed1
Thanks for your reply. The A and B denote two different Risk Factors, not the case and control groups.

I have modified the original post to better clear this. The query is on the interpretation of the Odds Ratio of the two different risk factors in the two given situations. Thanks.

[giordano]

My two cents: both situation have the same impact but situation 1 is more accurate than 2.

[CowboyBear]

Can you provide a bit more information about the actual problem you're trying to figure out?

The confidence intervals given for the odds ratios are a little bizarre. Their logarithms certainly aren't evenly distributed about the point estimate, which can be the case if a non-conventional estimation techniques such as bootstrapping has been used... but even then, they don't make any sense. E.g. for risk factor A at time 1, the point estimate is much closer to the upper bound of the 95%CI than the lower bound! (usually the opposite, because the OR estimates are the exponent form of the B/CI estimates from the logistic regression).

How did your colleagues calculate these estimates? What is the actual question they're wanting answered? Interpretation depends on context.

(Scratch that, have just noticed the date of the original post!)