Skip to content

[R-meta] convergence in Mada

6 messages · James Meyer, Philipp Doebler, CHAPPELL Francesca +1 more

James Meyer
# 
How can you tell if convergence has been accomplished using the reitsma() function in mada ? Which package do you recommend using for DTA meta-analysis if you have a small number of studies?

Sincerely
James C Meyer
Sent from Mail<https://go.microsoft.com/fwlink/?LinkId=550986> for Windows 10
Philipp Doebler
# 
Hi James,

I have rarely had convergence problems. There are three potential problems:

(1) Low number of studies (5 model parameters...) but mada throws a warning
if I remember correctly.
(2) Little variance in the false positive rates and/or sensitivities.
Especially relevant if all studies have the same sensitivity of, say, 95%.
This can sometimes lead to a degenerate covariance matrix of the bivarate
random effect.
(3) A perfect between study correlation of logit(sens) and logit(fpr). If
there is any suspicion, calculate the correlation matrix with cov2cor.

Also have a look at all standard errors and confidence intervals. The
recent simulation study of Vogelsang et al. (2018) discusses convergence
issues of mada (and other packages).

Kind wishes,
   Philipp




On Wed, Jun 19, 2019 at 2:32 PM James Meyer <
outlook_655291802F3BAFF3 at outlook.com> wrote:

            

      
      
      
    

        

      
      
    

        

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      CHAPPELL Francesca
    

    

      
      #
    

  


  

      
On a related note, it can also be useful to profile the likelihood of the correlation coefficient - estimation of correlation coefficient can be particularly poor even when convergence is achieved.

Francesca


-----Original Message-----
From: R-sig-meta-analysis [mailto:r-sig-meta-analysis-bounces at r-project.org] On Behalf Of Philipp Doebler
Sent: 19 June 2019 13:59
To: James Meyer <outlook_655291802F3BAFF3 at outlook.com>
Cc: r-sig-meta-analysis at r-project.org
Subject: Re: [R-meta] convergence in Mada

Hi James,

I have rarely had convergence problems. There are three potential problems:

(1) Low number of studies (5 model parameters...) but mada throws a warning if I remember correctly.
(2) Little variance in the false positive rates and/or sensitivities.
Especially relevant if all studies have the same sensitivity of, say, 95%.
This can sometimes lead to a degenerate covariance matrix of the bivarate random effect.
(3) A perfect between study correlation of logit(sens) and logit(fpr). If there is any suspicion, calculate the correlation matrix with cov2cor.

Also have a look at all standard errors and confidence intervals. The recent simulation study of Vogelsang et al. (2018) discusses convergence issues of mada (and other packages).

Kind wishes,
   Philipp

        
On Wed, Jun 19, 2019 at 2:32 PM James Meyer < outlook_655291802F3BAFF3 at outlook.com> wrote:

        

            

      
      
      
    

        

      
      
    

        
--
Prof. Dr. Philipp Doebler
Technische Universit?t Dortmund
Fakult?t Statistik
Vogelpothsweg 87
44227 Dortmund

Tel.: +49 231-755 8259
Fax: +49 231-755 3918
doebler at statistik.tu-dortmund.de
www.statistik.tu-dortmund.de/1261.html


Wichtiger Hinweis: Die Information in dieser E-Mail ist vertraulich. Sie ist ausschlie?lich f?r den Adressaten bestimmt. Sollten Sie nicht der f?r diese E-Mail bestimmte Adressat sein, unterrichten Sie bitte den Absender und vernichten Sie diese Mail. Vielen Dank.
Unbeschadet der Korrespondenz per E-Mail, sind unsere Erkl?rungen ausschlie?lich final rechtsverbindlich, wenn sie in herk?mmlicher Schriftform (mit eigenh?ndiger Unterschrift) oder durch ?bermittlung eines solchen Schriftst?cks per Telefax erfolgen.

Important note: The information included in this e-mail is confidential.
It is solely intended for the recipient. If you are not the intended recipient of this e-mail please contact the sender and delete this message. Thank you.
Without prejudice of e-mail correspondence, our statements are only legally binding when they are made in the conventional written form (with personal signature) or when such documents are sent by fax.


_______________________________________________
R-sig-meta-analysis mailing list
R-sig-meta-analysis at r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-meta-analysis
The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.

  
  


  


      

    

    

      
      

        


  

        

      Gerta Ruecker
    

    

      
      #
    

  


  

      
Dear Philipp,

Could you please give a link or full bibliographic details for the paper 
you cited (Vogelsang et al., 2018)? I couldn't find it.

Best,

Gerta


Am 19.06.2019 um 14:59 schrieb Philipp Doebler:

            

      
      
      
    

        

      
      
    

            

      
      
      
    

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Philipp Doebler
    

    

      
      #
    

  


  

      
Hi Gerta,

the paper appeared in Methods Inf Med:

Vogelgesang F, Schlattmann P, Dewey M (2018).  The Evaluation of Bivariate
Mixed Models in Meta-analyses of Diagnostic Accuracy Studies with SAS,
Stata and R. Methods Inf Med. 2018 May;57(3):111-119. doi:
10.3414/ME17-01-0021.

Kind wishes,
   Philipp




On Wed, Jun 19, 2019 at 3:19 PM Gerta Ruecker <ruecker at imbi.uni-freiburg.de>
wrote:

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Gerta Ruecker
    

    

      
      #
    

  


  

      
Thank you! You had misspelled the first author's name, that's why I 
couldn't find it.

Gerta

Am 19.06.2019 um 15:22 schrieb Philipp Doebler: