SPSS Statistics

 View Only
  • 1.  Meta-analyses; Binary data

    Posted Sat March 25, 2023 01:06 PM
    Dear all,
    In the frame of the meta-analyses for binary data menu, SPSS allows data to be handled analysing these possibilities: 
     

     (PS:  Odds ratio are translated as cotes in French..; Rapport = ratio; ). Sorry.

    This implies that I cannot obtain  Forest plots  graphs with absissa as Odds ratio 
     I can obtain them with absissa expressed as  LOG of different ratios  or as risks differences. But not as Odds ratios.

    My problem is that the reviewer of the article  I submitted  my article to, doesn't want LOG Odds ratio but odds ratios.

    Could someone help me in this issue?
    THANKS in advance


    Prof. viviane de maertelaer

    Emeritus and Invited Professor of Biostatistics, ULB, Fac.of Medicine, IRIBHM

    EMA (European Medicine Agency) Expert  &  Agence Fédérale Médicaments Produits de Santé ( AFMPS) Consultant 

    VM-AN Consulting, snc

    Rue du Coq 14, 1380 Ohain, Belgium
    Viviane.De.Maertelaer
    @ulb.be

    Tél : +32 (0)476 420 556

    TVA : 0503 782 762





  • 2.  RE: Meta-analyses; Binary data

    Posted Tue March 28, 2023 08:59 AM

    Hi. See the help for META BINARY, for subcommand /FORESTPLOT:

    "ESFORM The optional keyword controls the plot column and the display of ES and CI when specified by DISPLAY. The process is available when /ESTYPE = LOGOR, LOGPETO, or LOGRR, and is otherwise ignored. FALSE is the default setting, which plots the log odds ratio, Peto's log odds ratio, or log risk ratio, and tabulates the log-transformed statistics. TRUE plots the odds ratio, Peto's odds ratio, or risk ratio, and tabulates the exponential-form statistics."



    ------------------------------
    Rick Marcantonio
    Quality Assurance
    IBM
    ------------------------------



  • 3.  RE: Meta-analyses; Binary data

    Posted Tue March 28, 2023 09:45 AM
    Many thanks.�� 
    My commands hereunder.  

    I indeed selected  ESTYPE=LOGOR.
    I don't see the possibility to chose an ESFORM; Juste an EFORM ..  
    So I just obtain now the forest plots with the logg odds ratios in abscissa, but I want the odds ratios...  due to the reviewer constraints...
    Thanks  for your contribution...
    Prof. viviane de maertelaer

    Emeritus and Invited Professor of Biostatistics, ULB, Fac.of Medicine, IRIBHM

    EMA (European Medicine Agency) Expert  &  Agence Fédérale Médicaments Produits de Santé ( AFMPS) Consultant 

    VM-AN Consulting, snc

    Rue du Coq 14, 1380 Ohain, Belgium
    Viviane.De.Maertelaer
    @ulb.be

    Tél : +32 (0)476 420 556

    TVA : 0503 782 762


    META BINARY
      /DATA TREATMENT=SUCCESS(posCasesMetBC) FAILURE(EchecMnonBC)  CONTROL=SUCCESS(PosCasesRef)
        FAILURE(FailuresRéfBC)   ID=studies  ESTYPE=LOGOR
      /CRITERIA CILEVEL=95 SCOPE=AVAILABLE CLASSMISSING=EXCLUDE MAXITER=100 MAXSTEP=5
        CONVERGENCE=0.000001 ZEROCOUNTS=ONLY ADD=0.5 DROP=TRUE
      /INFERENCE MODEL=RANDOM ESTIMATE=REML ADJUSTSE=NONE
      /FORESTPLOT DISPLAY=ES SE CI WEIGHT POSITION=RIGHT ANNOTATIONS=HOMOGENEITY HETEROGENEITY TEST
        REFLINES=OVERALL NULL EFORM=TRUE.






  • 4.  RE: Meta-analyses; Binary data

    Posted Tue March 28, 2023 10:24 AM
      |   view attached

    Can you tell me, do you see the Odds Ratio in the second analysis below?

    GET FILE=<attached>. 

    META BINARY
     /DATA
      TREATMENT=SUCCESS(npost) FAILURE(nnegt) CONTROL=SUCCESS(nposc) FAILURE(nnegc) STUDY=@_meta_studylabel
      ESTYPE=LOGOR
     /CRITERIA CILEVEL=95 SCOPE=AVAILABLE CLASSMISSING=EXCLUDE MAXITER=100 MAXSTEP=5
      CONVERGENCE=0.000001 ZEROCOUNTS=ONLY ADD=0.5 DROP=TRUE
     /INFERENCE MODEL=RANDOM ESTIMATE=REML ADJUSTSE=NONE
     /FORESTPLOT DISPLAY=ES SE CI WEIGHT POSITION=RIGHT ANNOTATIONS=HOMOGENEITY HETEROGENEITY TEST
      REFLINES=OVERALL NULL EFORM=FALSE.

    META BINARY
     /DATA
      TREATMENT=SUCCESS(npost) FAILURE(nnegt) CONTROL=SUCCESS(nposc) FAILURE(nnegc) STUDY=@_meta_studylabel
      ESTYPE=LOGOR 
     /CRITERIA CILEVEL=95 SCOPE=AVAILABLE CLASSMISSING=EXCLUDE MAXITER=100 MAXSTEP=5
      CONVERGENCE=0.000001 ZEROCOUNTS=ONLY ADD=0.5 DROP=TRUE
     /INFERENCE MODEL=RANDOM ESTIMATE=REML ADJUSTSE=NONE
     /FORESTPLOT DISPLAY=ES SE CI WEIGHT POSITION=RIGHT ANNOTATIONS=HOMOGENEITY HETEROGENEITY TEST
      REFLINES=OVERALL NULL EFORM=TRUE.



    ------------------------------
    Rick Marcantonio
    Quality Assurance
    IBM
    ------------------------------

    Attachment(s)

    sav
    bcgset.sav   6 KB 1 version


  • 5.  RE: Meta-analyses; Binary data

    Posted Tue March 28, 2023 10:51 AM
    With the syntax;
    META BINARY
    DATA TREATMENT=SUCCESS(posCasesMetBC) FAILURE(EchecMnonBC)  CONTROL=SUCCESS(PosCasesRef)
        FAILURE(FailuresRéfBC)   ID=studies  ESTYPE=LOGOR
     /CRITERIA CILEVEL=95 SCOPE=AVAILABLE CLASSMISSING=EXCLUDE MAXITER=100 MAXSTEP=5
      CONVERGENCE=0.000001 ZEROCOUNTS=ONLY ADD=0.5 DROP=TRUE
     /INFERENCE MODEL=RANDOM ESTIMATE=REML ADJUSTSE=NONE
     /FORESTPLOT DISPLAY=ES SE CI WEIGHT POSITION=RIGHT ANNOTATIONS=HOMOGENEITY HETEROGENEITY TEST
      REFLINES=OVERALL NULL EFORM=TRUE.

    I get 

    Warnings
    This command should specify a valid subcommand at the beginning.
    Execution of this command stops.









    Can you tell me, do you see the Odds Ratio in the second analysis below? GET FILE=<attached>.  META BINARY  /DATA   TREATMENT=SUCCESS(npost)...
    IBM Community

    SPSS Statistics

    Post New Message
    Re: Meta-analyses; Binary data
    Reply to Group Reply to Sender
    Rick Marcantonio
    Mar 28, 2023 10:24 AM  |  attachment  view attached
    Rick Marcantonio

    Can you tell me, do you see the Odds Ratio in the second analysis below?

    GET FILE=<attached>. 

    META BINARY
     /DATA
      TREATMENT=SUCCESS(npost) FAILURE(nnegt) CONTROL=SUCCESS(nposc) FAILURE(nnegc) STUDY=@_meta_studylabel
      ESTYPE=LOGOR
     /CRITERIA CILEVEL=95 SCOPE=AVAILABLE CLASSMISSING=EXCLUDE MAXITER=100 MAXSTEP=5
      CONVERGENCE=0.000001 ZEROCOUNTS=ONLY ADD=0.5 DROP=TRUE
     /INFERENCE MODEL=RANDOM ESTIMATE=REML ADJUSTSE=NONE
     /FORESTPLOT DISPLAY=ES SE CI WEIGHT POSITION=RIGHT ANNOTATIONS=HOMOGENEITY HETEROGENEITY TEST
      REFLINES=OVERALL NULL EFORM=FALSE.

    META BINARY
     /DATA
      TREATMENT=SUCCESS(npost) FAILURE(nnegt) CONTROL=SUCCESS(nposc) FAILURE(nnegc) STUDY=@_meta_studylabel
      ESTYPE=LOGOR 
     /CRITERIA CILEVEL=95 SCOPE=AVAILABLE CLASSMISSING=EXCLUDE MAXITER=100 MAXSTEP=5
      CONVERGENCE=0.000001 ZEROCOUNTS=ONLY ADD=0.5 DROP=TRUE
     /INFERENCE MODEL=RANDOM ESTIMATE=REML ADJUSTSE=NONE
     /FORESTPLOT DISPLAY=ES SE CI WEIGHT POSITION=RIGHT ANNOTATIONS=HOMOGENEITY HETEROGENEITY TEST
      REFLINES=OVERALL NULL EFORM=TRUE.



    ------------------------------
    Rick Marcantonio
    Quality Assurance
    IBM
    ------------------------------
      Reply to Group Online   View Thread   Recommend   Forward  




     
    You are subscribed to "SPSS Statistics" as viviane.de.maertelaer@ulb.be. To change your subscriptions, go to My Subscriptions. To unsubscribe from this community discussion, go to Unsubscribe.



    Original Message:
    Sent: 3/28/2023 10:24:00 AM
    From: Rick Marcantonio
    Subject: RE: Meta-analyses; Binary data

    Can you tell me, do you see the Odds Ratio in the second analysis below?

    GET FILE=<attached>. 

    META BINARY
     /DATA
      TREATMENT=SUCCESS(npost) FAILURE(nnegt) CONTROL=SUCCESS(nposc) FAILURE(nnegc) STUDY=@_meta_studylabel
      ESTYPE=LOGOR
     /CRITERIA CILEVEL=95 SCOPE=AVAILABLE CLASSMISSING=EXCLUDE MAXITER=100 MAXSTEP=5
      CONVERGENCE=0.000001 ZEROCOUNTS=ONLY ADD=0.5 DROP=TRUE
     /INFERENCE MODEL=RANDOM ESTIMATE=REML ADJUSTSE=NONE
     /FORESTPLOT DISPLAY=ES SE CI WEIGHT POSITION=RIGHT ANNOTATIONS=HOMOGENEITY HETEROGENEITY TEST
      REFLINES=OVERALL NULL EFORM=FALSE.

    META BINARY
     /DATA
      TREATMENT=SUCCESS(npost) FAILURE(nnegt) CONTROL=SUCCESS(nposc) FAILURE(nnegc) STUDY=@_meta_studylabel
      ESTYPE=LOGOR 
     /CRITERIA CILEVEL=95 SCOPE=AVAILABLE CLASSMISSING=EXCLUDE MAXITER=100 MAXSTEP=5
      CONVERGENCE=0.000001 ZEROCOUNTS=ONLY ADD=0.5 DROP=TRUE
     /INFERENCE MODEL=RANDOM ESTIMATE=REML ADJUSTSE=NONE
     /FORESTPLOT DISPLAY=ES SE CI WEIGHT POSITION=RIGHT ANNOTATIONS=HOMOGENEITY HETEROGENEITY TEST
      REFLINES=OVERALL NULL EFORM=TRUE.



    ------------------------------
    Rick Marcantonio
    Quality Assurance
    IBM
    ------------------------------

    Original Message:
    Sent: Tue March 28, 2023 09:44 AM
    From: viviane de maertelaer
    Subject: Meta-analyses; Binary data

    Many thanks.�� 
    My commands hereunder.  

    I indeed selected  ESTYPE=LOGOR.
    I don't see the possibility to chose an ESFORM; Juste an EFORM ..  
    So I just obtain now the forest plots with the logg odds ratios in abscissa, but I want the odds ratios...  due to the reviewer constraints...
    Thanks  for your contribution...
    Prof. viviane de maertelaer

    Emeritus and Invited Professor of Biostatistics, ULB, Fac.of Medicine, IRIBHM

    EMA (European Medicine Agency) Expert  &  Agence Fédérale Médicaments Produits de Santé ( AFMPS) Consultant 

    VM-AN Consulting, snc

    Rue du Coq 14, 1380 Ohain, Belgium
    Viviane.De.Maertelaer
    @ulb.be

    Tél : +32 (0)476 420 556

    TVA : 0503 782 762


    META BINARY
      /DATA TREATMENT=SUCCESS(posCasesMetBC) FAILURE(EchecMnonBC)  CONTROL=SUCCESS(PosCasesRef)
        FAILURE(FailuresRéfBC)   ID=studies  ESTYPE=LOGOR
      /CRITERIA CILEVEL=95 SCOPE=AVAILABLE CLASSMISSING=EXCLUDE MAXITER=100 MAXSTEP=5
        CONVERGENCE=0.000001 ZEROCOUNTS=ONLY ADD=0.5 DROP=TRUE
      /INFERENCE MODEL=RANDOM ESTIMATE=REML ADJUSTSE=NONE
      /FORESTPLOT DISPLAY=ES SE CI WEIGHT POSITION=RIGHT ANNOTATIONS=HOMOGENEITY HETEROGENEITY TEST
        REFLINES=OVERALL NULL EFORM=TRUE.




    Original Message:
    Sent: 3/28/2023 8:59:00 AM
    From: Rick Marcantonio
    Subject: RE: Meta-analyses; Binary data

    Hi. See the help for META BINARY, for subcommand /FORESTPLOT:

    "ESFORM The optional keyword controls the plot column and the display of ES and CI when specified by DISPLAY. The process is available when /ESTYPE = LOGOR, LOGPETO, or LOGRR, and is otherwise ignored. FALSE is the default setting, which plots the log odds ratio, Peto's log odds ratio, or log risk ratio, and tabulates the log-transformed statistics. TRUE plots the odds ratio, Peto's odds ratio, or risk ratio, and tabulates the exponential-form statistics."



    ------------------------------
    Rick Marcantonio
    Quality Assurance
    IBM

    Original Message:
    Sent: Sat March 25, 2023 01:06 PM
    From: viviane de maertelaer
    Subject: Meta-analyses; Binary data

    Dear all,
    In the frame of the meta-analyses for binary data menu, SPSS allows data to be handled analysing these possibilities: 
     

     (PS:  Odds ratio are translated as cotes in French..; Rapport = ratio; ). Sorry.

    This implies that I cannot obtain  Forest plots  graphs with absissa as Odds ratio 
     I can obtain them with absissa expressed as  LOG of different ratios  or as risks differences. But not as Odds ratios.

    My problem is that the reviewer of the article  I submitted  my article to, doesn't want LOG Odds ratio but odds ratios.

    Could someone help me in this issue?
    THANKS in advance


    Prof. viviane de maertelaer

    Emeritus and Invited Professor of Biostatistics, ULB, Fac.of Medicine, IRIBHM

    EMA (European Medicine Agency) Expert  &  Agence Fédérale Médicaments Produits de Santé ( AFMPS) Consultant 

    VM-AN Consulting, snc

    Rue du Coq 14, 1380 Ohain, Belgium
    Viviane.De.Maertelaer
    @ulb.be

    Tél : +32 (0)476 420 556

    TVA : 0503 782 762





  • 6.  RE: Meta-analyses; Binary data

    Posted Wed March 29, 2023 09:57 AM

    Hello, you need a slash ('/') character in front of the DATA TREATMENT= subcommand. So it should look like:

    /DATA TREATMENT=

    Regards,



    ------------------------------
    Curtis Browning
    SPSS Statistics Architect
    ------------------------------



  • 7.  RE: Meta-analyses; Binary data

    Posted Mon April 03, 2023 10:16 AM

    From one of our statisticians:

    "EFORM only controls the column display in the annotations and the plot values, but not the chart x-axis scale. META BINARY does not have a feature to plot the exponential form of the statistics on a linear scale."



    ------------------------------
    Rick Marcantonio
    Quality Assurance
    IBM
    ------------------------------