Can I perform a meta analysis of a prospective observational cohort without a control? If so, can I do it on RevMan and how? Also, is it possible to analyse both prospective and retrospective studies in the same meta analysis?
Thank you
Hi,
1. Yes, you can perform meta-analysis of descriptive cohort (without control). We have to use the proportion and its SE. Meta-analysis is done by the generic inverse variance method.
2. Not sure if this can be done in RevMan. But one can use Stata. If you are familiar with Stata, I can send you the do file for this.
Best
Jeeva
Hi Jeeva
Yes, I have used STATA for meta analyses as well. Would be very grateful to you if you could send the file.
Thanks
Hi, Try the following in STATA:
gen p=.
gen se=.
forv i= 1(1)11 {
cii total[`i'] cases[`i']
qui replace p= r(mean) in `i'
qui replace se= r(se) in `i'
}
gen cons =1
vwls p cons, sd(se)
metan p se, random nooverall nowarning lcols(study weeks samplesize)
Generic inverse variance Data Type is used when the data is available only in Mean Difference form (sometimes log ratio). The possible Effect Measures in Review Manager are Odds Ratio, Risk Ratio, Risk Difference, Mean Difference and Std. Mean Difference with IV Statistical Method.
Estimates and their SEs may be entered directly into RevMan under the ‘Generic inverse variance’ outcome. For ratio measures of intervention effect, the data should be entered as natural logarithms (for example as a log odds ratio and the standard error of the log odds ratio).
You can do both continuous as well as dichotomous Data Type.
Good Luck
https://www.researchgate.net/post/can_we_do_meta-analysis_of_single_arm_studies_observational_in_revman-for_continuous_as_well_as_dichotomous_data/1
Hi all.
I am doing a meta-analysis to understand the burden of respiratory viruses in childhood pneumonia where I have only one arm with the total of children with pneumonia and the number became positive for respiratory viruses. I have been asked to do the analysis by RevMan by my supervisors. Just gone through the discussion and found the issues around having one arm in RevMan had been previously discussed. There are researchers here that suggested STATA over RevMan for such kind of analysis. I am very familiar with STATA. WOuld anybody kindly send me do file ([email protected]) for doing meta-analysis in STATA. It will be an immensely beneficial to me. thanks.
Can share my stata codes ... There are some nice R packages as well.
In addition to R and STATA, OpenMeta[Analyst] is very easy software to perform meta-analysis of single arm studies.
Hi Jeeva Sankar,
I'm currently planning to run some meta-analysis analysis on cross-sectional studies on STATA. Do you mind sharing the do file for the generic inverse variance method you've recommended? My email is [email protected]. Thank you in advance for your help. =)
Use OpenMeta[Analyst] free available
Link: http://www.cebm.brown.edu/openmeta/download.html
It uses correction factor of 0.5 for '0' cells.
Yes! it can be done, but make sure that the studies are comparable and that you've covered heterogeneity issues. OpenMeta[Analyst] does a great job honestly. But there are plenty of options if you use R's "metaprop", it needs a background in programming (which is totally educable but needs time and patience). Secondly, I've heard from an Oncologist that published a lot of MAs that it's not that dangerous to include prospective and retrospective together. But it's the randomization status that you need to worry about mixing.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis