Skip to content

[R-meta] Accounting for baseline differences in meta-analysis

3 messages · Gladys Barragan-Jason, Michael Dewey, Gerta Ruecker

#
Dear Gerta,
Thanks for your response. I also realized that most of the time papers do
not report the pre-treatment and/or the pre-control values. Sometimes they
just say that the values are not different but do not report thoses, but
more often they do not measure it at all.  Is there any way to take this
into account?...I am not sure about how to deal with presence vs. absence
of such values in the meta-analysis or whether this just needs to be
discussed as a potential limitation.
Thanks a lot for your response.
Gladys

Le jeu. 28 janv. 2021 ? 12:35, Dr. Gerta R?cker <
ruecker at imbi.uni-freiburg.de> a ?crit :

  
    
#
Dear Gladys

If I understand this correctly in a randomised study you can analyse the 
post treatment values alone or take into account any baseline value you 
wish. You may get a more precise result if you adjust for baseline.

Michael
On 02/02/2021 08:57, Gladys Barragan-Jason wrote:

  
    
  
#
Dear Michael,

That's exactly what I wrote in my earlier post below. The new problem 
Gladys describes is that the baseline value are often missing, thus 
accounting for them would reduce sample size and thus precision (even if 
some missing data approaches would be used).

Best,

Gerta

Am 02.02.2021 um 11:15 schrieb Michael Dewey: