st: RE: Using the metan command

["Khairallah, Carole [khaicar]" <Carole.Khairallah@liverpool.ac.uk>]

Hi Amal,

Your error comes from the fact that you have 1 data line with value 0 as lower and upper CI.
If you remove this observation, your command will work.

To make the forest plot stratified by e_group, just add in the options: by(e_group)

One option:

metan rr lci uci, ///
	nosecsub sgweight ///			
	by(e_group)   ///
	effect(Risk ratio in ... ) ///
	dp(2) boxsca(30) aspect(0.8) astext(80) textsize(140)
	

Good luck


-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Amal Khanolkar
Sent: 09 July 2013 12:02
To: statalist@hsphsun2.harvard.edu
Subject: st: Using the metan command

Hi all,

Some issues with using the metan command. I've read the help file for the same, but I need some help in better understanding how I use this command.

I have a small dataset as follows:

input id rr lci uci e_group

1 1 0 0 1

2 1.12 0.78 1.62 2
3 0.96 0.83 1.30 2
4 0.96 0.85 1.09 2

5 0.84 0.65 1.07 3
6 0.98 0.90 1.07 3
7 0.95 0.88 1.02 3

8 1.11 0.69 1.80 4
9 1.08 0.90 1.30 4
10 0.85 0.73 0.97 4

11 1.76 1.24 2.50 5
12 1.15 0.98 1.34 5
13 0.86 0.75 0.99 5

14 1.57 1.31 1.87 6
15 0.90 0.82 0.97 6
16 1.07 1.01 1.14 6

17 0.86 0.51 1.45 7
18 0.95 0.80 1.14 7
19 1.31 1.16 1.47 7

20 1.30 1.00 1.70 8
21 0.98 0.88 1.10 8
22 0.76 0.70 0.83 8
 

The five variables above are:  ID, risk ratio (rr), lower confidence limit, upper confidence limit, and ethnic group. In effect it is three RRs for three outcomes for each ethnic group obtained from a multinomial logistic regression model, which I would like to show using a forest plot.

Just to get a basic forest plot I typed:

metan rr lci uci

But I get this in response:

Effect size and confidence intervals invalid:
order should be {effect size, lower ci limit, upper ci limit}


So, I'm not sure if I have to provide the weights as part of the above command to get it to work? If so, how do I generate the weights? This is not a true meta-anaylsis that I'm trying to plot, but just relative risks from a multinomial logistic regression model. Ideally I would also like to group the RR according to the grouping variable above.

Thanks!

/Amal 
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/