Interpretation of Relative risk and confidence interval?

#1
There were fewer primary events in patients assigned to warfarin than in those assigned to aspirin (24, 1·8% per year vs 48, 3·8% per year, RR 0·48, 95% CI: 0·28–0·80). p = 0.0027

Primary event: fatal and non-fatal disabling stroke, intracranial haemorrhage and arterial embolism.

Control group: aspirin
Experimental group: warfarin

How do I interpret RR 0.48% with regards to the above information?

I'll have a go at this, please correct me if I have misunderstood:
The risk of having primary events with patients taking warfarin is 48% compared to the risk having primary events with patients taking aspirin (however could we conclude that there were fewer primary events by knowing Relative Risk only? :confused:).
The range of CI is small enough and indicates that this result statistically significant.
 
#2
Hi nana0709,

Do you have the 2x2 table? Your interpretation is correct, I just wanted to make sure the RR = .48 was too. And it's .48 or 48% not .48%.

Also, your comment about the CI being small is a reflection of sample size. The RR is significant if the CI does not contain 1 - which in this case, does not.
 
#3
hi lumhearts,

Thank you very much for your help.

Sorry I shouldn't have written 0.48%. RR = 0.48 instead.

What's a 2x2 table?

I have some questions regarding to Outcome in the PICO table and Absolute risk reduction.

Firstly, I interpreted Outcome as something that we want the treatment to prevent, in this case, stroke.

However, I read in the paper that says --> Primary outcome: first occurence of fatal and non-fatal disabling stroke (all types), intracranial haemorrhage and arterial embolism.

Isn't it contradicting to have primary outcome written as whether or not stroke occured? since the study is comparing which is better (warfarin or aspirin) at PREVENTING OR REDUCE RISK OF STROKE?
Certainly we do not want stroke to occur? which is the objective of the study? In this case i'm thinking objective of the study is the same as the primary outcome of the study..I'm not sure if I have misunderstood...

Secondly, how do i actually interpret this sentence?
"Yearly risk of extracranial haemorrhage was 1.4% (warfarin) versus 1.6% (aspirin) (relative risk 0.87, 0.43 - 1.73; ARR 0.2%, -0.7 to 1.2)

With a CI of 0.43 - 1.73, does that mean RR is not significant since it contain 1? so that means this result is not significant enough (can disregard it) and play of chance is hugely possible?

What do I say if somebody ask me "What does ARR of 0.2%" mean?
I don't actually understand what ARR tells us even though I had read the definition of ARR like 100 times =/

Truly appreciate your time and effort. Thanks in advance.
 

Link

Ninja say what!?!
#4
Hi,

I hope you don't mind me asking. Are you planning to use these results professionally, or just for your own understanding? I've seen A LOT of people mess up when trying to do statistics without the appropriate training.
 
#5
Hi Link,

I have to do a critical appraisal presentation on a paper =S.

Have to talk about what the results are and what do they all mean.

Does that count as own understanding? =/
 
#6
I think I understand now that you do not have the raw data here. You want help with interpretation, right? A 2x2 table is what you would construct to help calculate and understand the RR - if you had the raw data.

For your first question - it is correct (or okay anyway) to examine if stroke DID occur when the exposure (warafin) is thought to be protective. In this case, you would want (or hope) the RR was below 1 (which it is). When the RR is below 1 it implies decreased risk. So - in your first post - the warafin group had 48% the risk of stroke as compared to aspirin group. You should realize that 48% risk is nearly half the risk. Which lines up with your logic "we do not want stroke to occur". This confuses a lot of people.

For your second question - the RR is the ratio of two risks. So, .014/.016 = .875. When you understand that, then it is easy to see that when the RR = 1 that the treatment and control groups are the same. And then it is easy to see that when the CI contains 1, there is no significant relationship.

Does that help?
 
#7
Wow thanks a lot lumhearts. That certainly saved my life.

So in conclusion, I can say that: The results of the risk of extracranial haemorrhage comparing warfarin to aspirin is not significant, meaning there is a possible play of chance.
Therefore, warfarin might still have a greater risk of bleeding compared to aspirin.

That's how I sum it up. I hope my understanding towards what all these statistics mean is on the right track.

Once again thank you very, very much in helping me out. =)
 
#9
Just an added note to help that you can also say that you reduce the risk of a stroke by 52% for warfarin compared to aspirin, (1-0.48) which may give you an easier message