Hello,
I came across this study in circulation (attached) that is reporting an annualized event rate (AER) across the studies included. I'm trying to replicate this. Essentially I have two groups with data from 9 studies on event rate. what I did is divide the event rate by the mean or median of follow up for each individual study. But then how to combine all these event rates to see the difference between those two groups?
The attached paper describes a method but I was not able to replicate it based on the provided description (it's an appendix by the end of the paper).
"To compare the differences in annualized event rate between the Scar+ and Scar - groups, a generalized inverse variance meta-analysis of the difference in annualized event rates was performed using RevMan 5.2. The effect-size for each study was determined as AER Scar+ - AER Scar -. The variance for the difference in AER was calculated using the method described by Normand (2). The 2 parameter for the random effects model was determined using the method of moments."
Thank you
The author simply states that he did a random effects meta-analysis, measured Heterogenity. You can use a specialized meta analysis software as CMA to replicate the findings.
It is not really about the random effect, it is about how they combined event rate from different studies in Revman, since I couldnt replicate that.
Unfortunately I dont have CMA.
To be more specific, how they came up with Figure 3 from table 3 is what I dont understand (I used the exact same numbers from table 3 and I had slightly different results than what in Figure 3).
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis