Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: RE: Meta-Analysis using --metan-- how to ensure comparison to ES=1 not ES=0


From   Belinda Butcher <bbutcher@writesourcemedical.com.au>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: RE: Meta-Analysis using --metan-- how to ensure comparison to ES=1 not ES=0
Date   Fri, 19 Oct 2012 14:41:50 +1100

Thank you - worked perfectly!

Belinda Butcher BSc(Hons) MBiostat PhD
Director - Biostatistics & Medical Writing
WriteSource Medical Pty Ltd

PO Box 1521
Lane Cove NSW 1595

M: 0418 286 014
E: bbutcher@writesourcemedical.com.au

On 15/10/2012, at 8:47 PM, Trelle Sven wrote:

> Hi Belinda,
> You will need to log-transform your hazard ratios and confidence
> interval (and use the eform option in metan):
> 
> foreach var of varlist pfshr pfsll pfsul {
> 	gen ln_`var' = ln(`var')
> }
> metan ln_pfshr ln_pfsll ln_pfsul, eform effect("Hazard Ratio") fixed
> lcols(study author year treatment) double by(drug) astext(50)
> xlabel(0.2, 0.5, 1,  2, 5) textsize(100) title("RR of PFS") null(1)
> force
> 
> Best
> Sven
> 
> p.s. the null-option only refers to the graph (at least to my knowledge)
> 
> 
> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu
> [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Belinda
> Butcher
> Sent: Montag, 15. Oktober 2012 05:14
> To: statalist@hsphsun2.harvard.edu
> Subject: st: Meta-Analysis using --metan-- how to ensure comparison to
> ES=1 not ES=0
> 
> Hi,
> 
> I use Stata/MP for Mac 11.2, dated 16 May 2012.  All files are up to
> date..
> 
> I am using the --metan-- command (available from SSC:
> http://ideas.repec.org/c/boc/bocode/s456798.html) to combine the results
> of some oncology studies.
> 
> I am combining the hazard ratios for progression free survival, using
> the upper and lower confidence interals).
> 
> I am using the following command:
> 
> metan pfshr pfsll pfsul, effect("Hazard Ratio") fixed lcols(study author
> year treatment) double by(drug) astext(50) xlabel(0.2, 0.5, 1,  2, 5)
> textsize(100) title("RR of PFS") null(1) force
> 
> I need to compare the resultant overall HR to one, rather than zero.  To
> do this, I have included "null(1)" in my command.  However, when I run
> the command, I get a comparison to ES=0 (see output below).
> 
>           Study     |     ES    [95% Conf. Interval]     % Weight
> ---------------------+--------------------------------------------------
> ---------------------+-
>     A
> 1                    |    0.680     0.590     0.780         12.38
> 2                    |    0.758     0.661     0.869         10.33
> Sub-total           |
>  I-V pooled ES      |    0.715     0.645     0.786         22.72
> ---------------------+--------------------------------------------------
> ---------------------+-
>     B
> 3                    |    0.493     0.418     0.581         16.83
> 4                    |    1.190     0.720     1.960          0.29
> 5                    |    1.100     0.570     2.120          0.19
> 6                |    0.720     0.620     0.840          9.24
> Sub-total           |
>  I-V pooled ES      |    0.584     0.519     0.649         26.54
> ---------------------+--------------------------------------------------
> ---------------------+-
>     C
> 7                    |    0.540     0.420     0.710          5.32
> 8                    |    0.680     0.570     0.800          8.45
> 9                    |    0.540     0.440     0.660          9.24
> 10                   |    0.692     0.617     0.776         17.68
>                     |    0.590     0.480     0.720          7.76
> Sub-total           |
>  I-V pooled ES      |    0.628     0.580     0.676         48.45
> ---------------------+--------------------------------------------------
> ---------------------+-
>     D
> 11                   |    1.210     0.820     1.800          0.47
> 12                   |    1.410     0.960     2.070          0.36
> Sub-total           |
>  I-V pooled ES      |    1.298     0.930     1.665          0.83
> ---------------------+--------------------------------------------------
> ---------------------+-
>     E
> 13                   |    0.690     0.490     1.140          1.06
> 14                   |    1.010     0.610     1.660          0.41
> Sub-total           |
>  I-V pooled ES      |    0.779     0.502     1.055          1.46
> ---------------------+--------------------------------------------------
> ---------------------+-
> Overall              |           
>  I-V pooled ES      |    0.644     0.610     0.677        100.00
> ---------------------+--------------------------------------------------
> ---------------------+-
> Heterogeneity calculated by formula
>  Q = SIGMA_i{ (1/variance_i)*(effect_i - effect_pooled)^2 } where
> variance_i = ((upper limit - lower limit)/(2*z))^2 
> 
> Test(s) of heterogeneity:
>               Heterogeneity  degrees of
>                 statistic     freedom      P    I-squared**
> A                     1.18         1      0.278     15.1%
> B                   16.03          3      0.001     81.3%
> C                    7.53          4      0.110     46.9%
> D                    0.28          1      0.596      0.0%
> E                    1.03          1      0.310      3.1%
> Overall             46.85         14      0.000     70.1%
> Overall Test for heterogeneity between sub-groups: 
>                    20.79          4      0.000
> 
> ** I-squared: the variation in ES attributable to heterogeneity)
> 
> Considerable heterogeneity observed (up to 81.3%) in one or more
> sub-groups, Test for heterogeneity between sub-groups likely to be
> invalid
> 
> Significance test(s) of ES=0
> 
> A                     z= 19.99     p = 0.000
> B                     z= 17.64     p = 0.000
> C                     z= 25.62     p = 0.000
> D                     z=  6.92     p = 0.000
> E                     z=  5.52     p = 0.000
> Overall               z= 37.75     p = 0.000
> ------------------------------------------------------------------------
> -
> 
> I have looked in Sterne's book on Meta-Analysis (Sterne J.A.C (2009)
> Meta-Analysis in Stata: An updated collection from the Stata Journal),
> but haven't been able to work this out.
> 
> Your help is appreciated.
> 
> Kind regards,
> 
> Belinda Butcher BSc(Hons) MBiostat PhD
> Director - Biostatistics & Medical Writing WriteSource Medical Pty Ltd
> 
> PO Box 1521
> Lane Cove NSW 1595
> 
> M: 0418 286 014
> E: bbutcher@writesourcemedical.com.au
> 
> 
> *
> *   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/

*
*   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/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index