RCT class #7 Survival Analysis Flashcards
why not just use end of study methods?
This doesnt reflect the change in survival over the course of the study!!
The whole point of KM analysis so remember to comment on what the curves are doing
What is censoring?
Loss to follow up or study ends (right censoring)
When patients dont all start a trial on the same day this is called
Staggard entry
For surival analysis what do we do with the staggard entry
Convert it to a common starting time
How to compute the exact surviving proportion for each time interval
Number at risk (surviving) = total - (dead + censored) .
survivring proportion = number alive/number at risk
Again how do you report the survialval
get number at risk. Then divide the number still alive by the number at rik
number at risk (suriviing) = total- (dead + censored)
still alive/ number at risk
Cumulative surviving proportion is what?
The product of the surviving proportions acriss all monhts
What is a. Hazard
b. Hazard ration
a. Hazard it the prob of an outcome occuring in a short period given that it hasnt occured to date
b. hazard ratio is the measure of RR for the two survival curves.
a. test of sig for survival curves
b,. How does it work
c. IF one is outside of the CI then the difference is sig/non sid
d. you can get 95% CI’s around the curve.
Log Rank
a weighted combined chi sq
sig
What is the goal of the primary efficay analysis
test the null hypothesis with the best most unbiased etimate of the efficacy of the two interventions. Using the fewest possible assumptions.
Key to primary efficacy anaysis
uses fewest assumptions possible
if have 20 sites and want to do a comparision for a drug vs. placebo. how many strata will you have
40 strate =20x2
CMH test for what?
What is the Ho for CMH?
homogenity across a strata
Ho: the OR is the same across each strata
Where is the text to put into your presentation for the survival analysis stuff
end of lecture 7. He give the exact verbage
