Creating funnel plot from Odds Ratios and C.I.

[Anna Roberto]

Hi all,

I am doing meta-analysis where data available from original studies are: odds ratios and their confidence intervals. I was able to create forest plot using the command:

"metan logrr loglow_95 logupper_95, label(namevar=Study) eform"

However, I am not sure how to create funnel plot and tests for publication bias (I have installed "metafunnel" command).

Thanks in advance

[Marcos Almeida]

Hello Anna,

Welcome to the Stata Forum!

Please try this:

. use http://fmwww.bc.edu/repec/bocode/m/metan_example_data
. metafunnel OR ORlci ORuci, eform

Warning: varlist has 3 variables but option 'ci' not specified; 'ci' assumed.

Warning: ci option assumes that ratio measures are being used

As you can see, there is no need to type the CI option, provided you include only 3 items in the varlist.

Hopefully that helps!

Best,

Marcos

[Marcos Almeida]

Now, for the sake of further clarifying the matter: instead of OR and CIs, we will select the log of OR and its SE:

. gen logor = ln( (tdeath*cnodeath)/(tnodeath*cdeath) )

. gen selogor = sqrt( (1/tdeath) + (1/tnodeath) + (1/cdeath) + (1/cnodeath) )

. metafunnel logor selogor

Note: default data input format (theta, se_theta) assumed.

Now, you may see the graph and compare with the one we previously shared in #2

Best,

Marcos

[Anna Roberto]

Hi Marcos,

Thank you so much. That was so clear and of great help.

Best regards,

[Anna Roberto]

Just a note..

For the graph in #3: as I have only OR and CI, I calculated the SE using the equation from:

http://handbook.cochrane.org/chapter...ervals_and.htm

gen logor = ln(OR)
gen loglow_95 = ln(ORlci)
gen logupper_95 = ln(ORuci)
gen selogor = (logupper_95-loglow_95) / 3.92