Question

Which Relative Risk (RR) and confidence interval (CI) is most likely to be a causal relationship?

A) RR= 3.0 (95% CI 1.9-7.9)

B) RR= 3.0 (95% CI 0.3-12.2)

C) RR= 1.0 (95% CI 0.3-12.2)

D) RR= 1.0 (95% CI 0.3-4.3)

Answer 1

There are few measures to check for causality. Odds ratio and Relative risk are some example. In this question we are given varying measures of Relative risk, henceforth referred to as RR. Now you can take it as a fact that higher the RR, more likely is it that a causal relationship exists. (please do NOTE the usage of the phrase 'more likely').

As the questions requires us to choose the option that is "MOST likely" to result in causality, we can straight away eliminate 2 of the 4 options. We now need to decide between A and B.

Now we reach a point where the question becomes a bit incomplete. We are not given any information as to what is the CI for. What I mean by this is that its unclear whether the CI has been computed for RR or ln(RR) which is normally distributed. Moreover, this information will help us determine the Null for the trial.

However, I assume for the sake of this question that the CI is computed for RR.

Now as a rule of thumb, If the CI is smaller, one can safely (to the limits of statistical significance) assume that the result/ computation is better than those with larger CI's.

( The rationale behind this is that given a smaller CI, 95% of the times (if the trial were to be performed repetitively), the data would be contained within a smaller spread, which means the results are (statistically) significant.)

Hence option A would be the best choice.

I hope this has helped. In case the answer is unclear, please do feel free to leave a comment. I will be more than glad to help you out.