We are using Multivariate analysis for identifying determinant factors for different years (Studies). We want to get a pooled effect (Combined Odds Ratio) of the significant variables from Multivariate Analysis.
Hi Shyam
Michael's answer sounds relevant to me if you do not have individual information. i'd perform the same technic
Best regards
I recomment to apply following approach for this meta-analysis:
fixed-effect meta-analysis of hazard ratio according to the method of Peto [Yusuf S, Peto R, Lewis JA, et al.: Beta blockade during and after myocardial infarction: an overview of the randomized trials. Progress in Cardiovas Dis 1985, 27: 335-71].
Firstly, the natural logarithm (ln) of the hazard ratio (HR) from each study/paper should be calculated. Standard errors for the ln(HR) you could calculate from CIs (if available). Then you will calculate weighted mean of ln(HR) together with weighted standard error of mean HR estimation between different studies as well as the mean of the standard errors from each study. Weights could be a) numbers of patients included into each study; b) if you have standard errors from all included studies you can prefferably use weight = 1/(standard error)^2.
Standard errors between-studies and mean standard errors from involved studies should be combined according to Peto’s method.
If it seems for you as too difficult to program it don't worry - there are plenty of softwares doing it, e.g.,
NCSS software http://ncss.wpengine.netdna-cdn.com/wp-content/themes/ncss/pdf/Procedures/NCSS/Meta-Analysis_of_Hazard_Ratios.pdf
... I think you can start free trial to analyse one data https://www.ncss.com/software/ncss/meta-analysis-in-ncss/
MEDCALC software https://www.medcalc.org/manual/meta-analysis-oddsratio.php
R software https://www.biometricsociety.org/wp-content/uploads/2015/01/talk1-intro-handout-2x2.pdf
https://www.jichi.ac.jp/saitama-sct/SaitamaHP.files/functions.html
I did it personally in SAS.
BR
Lada
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics