[R-meta] [netmeta] Error message in network meta-analysis: "Problem with multi-arm studies: Studies with inconsistent treatment estimates"

[Dr. Gerta Rücker]

Dear Rebecca,


The code you used "doesn't know" whether a study is three-arm. You still 
calculate SMD for each two arms, ignoring the third. Though you have 
already a netmeta-compatible structure, the structure of your data makes 
it more difficult in this case because of the problems with the SMDs. 
Better to start with a different structure and to use pairwise().


Function pairwise() expects

- either a format with one row per study and three (say) blocks of 
columns (studlab, TR1, TR2, TR3, n1, n2, n3, mean1, mean2, mean3, sd1, 
sd2, sd3, ...);� for an example see help(dietaryfat),

- or an arm-based format with one row per study arm like this (for an 
example of the second kind, see help(Woods2010)):


studlab TR��� n� mean��� sd

107���� MBI� 68 46.90 18.20

107���� PE�� 68 50.60 18.50

107���� PsEd 66 55.20 15.30

...���� ...� .. ..... .....


Hope this helps.


Best,

Gerta


Am 08.02.2020 um 14:38 schrieb Rogasch, Rebecca Manuela Madeleine:
>
> Thank you very much for the quick Reply and your advice!
>
>
> That Explanation makes a lot of sense.
>
> I have now tried using pairwise() to calculate the SMDs:
>
>
> paired<- pairwise(list(TR1, TR2),
> n = list(n_TR1, n_TR2),
> mean = list(Mt2TR1, Mt2TR2),
> sd = list(SDt2TR1, SDt2TR2),
> data = DATA1, studlab = studlab, sm= "SMD")
>
>
>
> This is an exerpt from the data Frame;
>
> 3-arm study in Purple:
>
>
> studlab treat1 treat2 n1 mean1�� sd1 n2 mean2�� sd2 TE����� seTE�� age 
> females��� severity����� income
> 1 1�� EMDR���� WL 15 22.87 20.27 14 54.21 16.26� -1.65116388 0.4387386 
> 36.54��� 0.76��� clinical����� lowmid
> 44 100��� NET���� WL 19 55.07 27.01 19 76.86 17.14 -0.94310177 
> 0.3439769 29.40��� 0.31��� clinical high income
> 45 104��� NET�� PsEd 17 19.10 11.70 12 21.20� 9.40 -0.18860625 
> 0.3779770 32.85��� 0.62��� clinical lowmid
> 48 107��� MBI�� PsEd 68 46.90 18.20 66 55.20 15.30 -0.49022797 
> 0.1754460 47.04��� 0.17��� clinical high income
> 47 107���� PE�� PsEd 68 50.60 18.50 66 55.20 15.30 -0.26905282 
> 0.1735963 47.04��� 0.17��� clinical high income
> 46 107���� PE��� MBI 68 50.60 18.50 68 46.90 18.20 0.20049752 
> 0.1719418 47.04��� 0.17��� clinical high income
> 78 11��� MBI���� WL NA��� NA��� NA NA��� NA��� NA -0.48920000 
> 0.2800000 51.80��� 0.99������� <NA> high income
>
> sample arms������� self_exp n_total������� tr_size analysis
>
> 1 refugees��� 2����� self-rated����� 29���� very small completers
> 44 refugees��� 2 clinician-rated����� 38���� very small������ �ITT
> 45 refugees��� 2����� self-rated����� 29���� very small������� ITT
> 48 vetmil��� 3����� self-rated���� 134 large������� ITT
> 47 vetmil��� 3����� self-rated���� 134 large������� ITT
> 46 vetmil��� 3����� self-rated���� 136 large������� ITT
> 78 civilian��� 2����� self-rated����� 71 small-moderate������� ITT
>
>
> However, I still get the same error messagein the Network meta-Analysis.
>
> A more detailed copy of the Output is available if needed.
>
>
> Any advice is very much appreciated!
>
> Best regards,
>
> Rebecca
>
> ------------------------------------------------------------------------
> *Von:* Guido Schwarzer <sc using imbi.uni-freiburg.de>
> *Gesendet:* Freitag, 7. Februar 2020 17:49:50
> *An:* Rogasch, Rebecca Manuela Madeleine; 
> r-sig-meta-analysis using r-project.org
> *Betreff:* Re: [R-meta] [netmeta] Error message in network 
> meta-analysis: "Problem with multi-arm studies: Studies with 
> inconsistent treatment estimates"
> Hi Rebecca,
>
> I assume the problem is related to calculating SMDs in multi-arm
> studies. Each pairwise comparison (in a multi-arm study) uses a
> different pooled standard deviation to calculate the SMD which leads to
> inconsistent effects.
>
> You could / should use pairwise() to calculate the SMDs which considers
> the same pooled standard deviation to calculate the SMD in multi-arm
> studies.
>
> Best wishes, Guido
>
> -- 
> Dr. Guido Schwarzer
> Institute of Medical Biometry and Statistics,
> Faculty of Medicine and Medical Center - University of Freiburg
>
> Postal address: Stefan-Meier-Str. 26, D-79104 Freiburg
>
> Phone: +49/761/203-6668
> Mail: sc using imbi.uni-freiburg.de
> Homepage: http://www.imbi.uni-freiburg.de
>
> ORCID iD: https://orcid.org/0000-0001-6214-9087
> R-book: https://www.springer.com/gp/book/9783319214153
>
>
> _______________________________________________
> R-sig-meta-analysis mailing list
> R-sig-meta-analysis using r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-meta-analysis

	[[alternative HTML version deleted]]