Hi,
RevMan and R, both can be used to conduct meta-analysis. Can anyone tell me what are the advantages of conducting a meta analysis in RevMan, over R?
Similarly it would be great if anyone can enlist advantages of R over RevMan?
I am sure about RevMan's graphical interface and ease of use have advantages over R's learning curve. But apart from that, couldn't figure out anything.
Thanks in advance.
-: Ram
RevMan is about more than just the meta-analysis, it's an all-inclusive data entry / report-writing tool for SLRs. It links in with the Cochrane collaboration.
But for the meta-analysis itself, you will find R (or Stata, which I personally prefer) to be more flexible (in terms of making the forest plots look how you want them) and more powerful (in terms of additional analyses not available in RevMan). There are also some specialist meta-analysis packages, e.g. Comprehensive meta-analysis, but R is free and both R and Stata can do more. And I think their graphs can look a lot nicer than the RevMan graphs.
If you're new to meta-analysis then you might feel more confident sticking with RevMan. If you want to have more flexibility to change how things look, or do some more powerful stats, then take the time to learn R or Stata.
Thanks Darren for the reply. Yes, I am learning Meta-analysis and wanted to have knowledge of common tools used in the field. I am a bit comfortable with R and thus wanted to compare it with RevMan.
Like you correctly pointed out, RevMan is itself a package for SLR writing. I will surely adhere to your recommendations and will explore RevMan for better insight.
As Darren correctly pointed out - if you are new to meta-analysis, RevMan will be simpler to handle. If you are doing a basic meta-analysis, you can use RevMan easily. However, things get complicated when you try to do NMA or ITC. Another software worth checking out is MedCalc (it has a free 15-day trial).
Darren made a good answer.
This is a good question to make directly to Guido. https://www.researchgate.net/profile/Guido_Schwarzer
He is the author of an R package directed to meta-analysis and, as far as I concern, he is also familiar with RevMan as he is or was in a cochrane group.
There are many step by step guide for meta-analysis in R around the internet. Just goolge it.
The meta package is one of the many available in R but it is quite handy and pretty automated to do lots of additional analysis, mainly the graphics output. I used it many times myself. It is worth to take a look at its examples and documentation. At this package there is a function to import datasets from the RevMan format.
There is also the RisMed package to help to do some stuff that RevMan does in organizing bibliography. Take a look at http://www.edii.uclm.es/~useR-2013/Tutorials/kovalchik/kovalchik_meta_tutorial.pdf
I tried to use RevMan once, but is so directly related to Cochrane group, that I found it is tricky for independent reviewers. And I was familiar with R at the time. So my experience is basically with R only.
Smeet, thanks for the insight. You are correct, NMA or ITC is not possible in RevMan.
Pedro, thanks for the comprehensive answer. Guido will be better able to answer the question here. Somewhere I feel, I am also in the same boot as you as I am also comfortable with R. I know the basics and have used it earlier for other purposes. I am currently learning basics of meta-analysis and thus exploring the fields.
Based on all answers, I feel that having experience of both will be really helpful.
If you're looking for an easy way to run a meta-analysis in R you might check out a package I cowrote on CRAN. Check out MAVIS and if you have questions feel free to pester me
Online Demo: http://kylehamilton.net/shiny/MAVIS/
Hi William,
Thanks for the link. Sure, the website looks impressive and i believe will be really helpful. I will check it over the weekend.
Anyhow, developing and maintaining a package a package is a great contribution towards the community. I salute you for that.
-:Ram
How can I download a full version of stata? from which website?
How can I make the risk of bias plot in R? conducting a meta-analysis in R is very easy and flexible, but I would like to create plots similar than performed by revman but using a R. Thank you
From my experience, I found the following:
Advantages of RevMan over R (metafor package)
1. I think, for a beginner, RevMan (like as usual windows software) is much easier than R (based on codes).
2. Can easily adjust meta-analysis structure based on epidemiological data and randomized-controlled trial data, in case of R you need different scripts (different for generating OR/RR and SMD/MD).
3. Automatically generates funnel plot upon keying in data.
4. Can work directly from RevMan in terms of data, no need to manage another file for data (like in R, you need to have your data in excel or other types of files from where you need to call from the R software, can not key in data in R directly).
Advantages of R (metafor package) over RevMan
1. You can play with the appearance of the plots including forest and funnel plots.
2. There are many sorts of funnel plots you can generate (i.e., standard, trim & fill and Contour-enhanced funnel plot).
3. For the quantitative analysis of risk of publication bias, you can generate both Begg's and Egger's tests.
4. L'Abbé plot is an exclusive plot R can generate which graphically represents heterogeneity.
5. In case of the random-effects model, RevMan can not generate a funnel plot, however, R can generate.
6. If your data has 'zero' in any of the columns (event/total), RevMan will state "Non-estimable data" and will not come up with any forest line. Interestingly, there is a specific code in R, by which you can get results (adding 0.5 to cells containing zero) and can generate the forest line for that particular study without hampering the overall results.
Therefore, it would be best if you can combine both the software to get the best outcome.
Good luck
Regards
Asiful
Having used both R and Stata for MA, I find R more powerful. It took me time to learn R but the cost is paying off now. Please grab R now and keeping growing stronger in using it. Luckily, it's free, absolutely free, and it has great free packages. However, I haven't abandoned Stata because certain analyses are done easily in Stata than R, and vice-versa.
Coming back to this question of 2015. Which R package do you suggest for systematic review/meta-analysis? Which one is the best and why?
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis