I am working on a systematic review of a procedure, of which outcome is pre and post-procedure data, for example, oxygen saturation level, diameter, etc. What would you recommend to report this data on a pooled analysis or meta-analysis? I have heard of median difference or such. Please give me your opinion. Thanks!
You are looking for statistical analysis of paired data, essentially. Given the examples that you stated which are continuous outcomes, a meta-analysis/pooling, assuming that a meta-analysis is really appropriate here, based on paired/dependent samples t-test might be an option.
The following resources might be of help in your decision-making.
https://heart.bmj.com/content/96/19/1525.long
https://eprints.whiterose.ac.uk/88476/3/WRRO_88476.pdf
Article Adding non-randomised studies to a Cochrane review brings co...
"Wilcoxon signed rank test" may be applied to compare median two dependent(paired) samples. This tutorial link may be useful
Wilcoxon signed rank test for two dependent (paired) variables:
http://countbio.com/web_pages/left_object/R_for_biology/R_biostatistics_part-1/wilcoxon_signed_rank_test_two_dependent_variables.html
Download Prometa3 software via this link "https://idostatistics.com/prometa3/". In this software, you should select the data entry type as "Means, independent groups, longitudinal data". Using the java based-software a bit hard, but you can deal with it with a little effort.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis