matched case control data analysis help

[Doc Ash]

hello,

i'm doing a 1:1 matched case control study, gettin confused on how to go about with data analysis, particularly with regards to risk factors. some of the risk factors are binary (yes/no), some have multiple categories (eg; age <20yrs, 20 - 29 yrs, 30 - 39 yrs). how do i calculate the odds ratios for each category??
am confused

[Bruce Weaver]

Doc Ash wrote

hello,

i'm doing a 1:1 matched case control study, gettin confused on how to go about with data analysis, particularly with regards to risk factors. some of the risk factors are binary (yes/no), some have multiple categories (eg; age <20yrs, 20 - 29 yrs, 30 - 39 yrs). how do i calculate the odds ratios for each category??
am confused

With 1:1 matching, you can perform conditional logistic regression via NOMREG.  See David Garson's notes on it here:

  http://faculty.chass.ncsu.edu/garson/PA765/logistic.htm#matchedconditional

[Marta Garcia-Granero]

Hi Bruce (and Doc):

I have always found that using a stratified Cox regression model is much
simpler than David Garson's method using NOMREG. I posted time ago a
solution using COXREG:

http://www.spsstools.net/Syntax/RegressionRepeatedMeasure/ConditionalLogisticRegression.txt

HTH,
Marta GG

Bruce Weaver wrote:

> Doc Ash wrote:
>
>> hello,
>>
>> i'm doing a 1:1 matched case control study, gettin confused on how to go
>> about with data analysis, particularly with regards to risk factors. some
>> of the risk factors are binary (yes/no), some have multiple categories
>> (eg; age <20yrs, 20 - 29 yrs, 30 - 39 yrs). how do i calculate the odds
>> ratios for each category??
>> am confused
>>
>>
>
> With 1:1 matching, you can perform conditional logistic regression via
> NOMREG.  See David Garson's notes on it here:
>
>   http://faculty.chass.ncsu.edu/garson/PA765/logistic.htm#matchedconditional
>
>
>
>
>

--
For miscellaneous SPSS related statistical stuff, visit:
http://gjyp.nl/marta/

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Kelly Vander Ley]

I will be out of the office from May 17-May 26, returning on the 27th. I will be checking e-mail on a limited basis. If you need immediate assistance please contact me at 503/539-5606. Thank you. Kelly

 

[Bruce Weaver]

Marta García-Granero-2 wrote

Hi Bruce (and Doc):

I have always found that using a stratified Cox regression model is much
simpler than David Garson's method using NOMREG. I posted time ago a
solution using COXREG:

http://www.spsstools.net/Syntax/RegressionRepeatedMeasure/ConditionalLogisticRegression.txt

HTH,
Marta GG

Another option, of course, is to use generalized estimating equations (GEE) via GENLIN.  I posted something to the SPSS newsgroup after using that method to reanalyze some old data that I'd analyzed with Stata the first time around.  The GENLIN results from SPSS were very similar to the conditional logistic regression results from Stata.  Unfortunately, I can't find that old discussion right now to give the link.  But here is a comparison of the results for 4 different outcome variables.  View in Courier or some other fixed font (e.g., copy and paste into a text editor).

----------------------------------------------------------
Outcome      | Odds Ratio   [95% Conf. Interval]   METHOD
----------------------------------------------------------
Outcome 1 gp |   1.599      1.064       2.403         1
             |   1.606      1.072       2.406         2
----------------------------------------------------------
Outcome 2 gp |   1.626      1.126       2.346         1
             |   1.631      1.128       2.358         2
----------------------------------------------------------
Outcome 3 gp |   0.755      0.536       1.065         1
             |   0.839      0.616       1.142         2
----------------------------------------------------------
Outcome 4 gp |   1.579      0.734       3.340         1
             |   1.585      0.735       3.419         2
----------------------------------------------------------

Method 1:  Conditional logistic regression (Stata)
Method 2:  GEE (via GENLIN in SPSS)

[Marta Garcia-Granero]

Hi Bruce:

Don't you remember my series of posts concerning the same topic? I
finally found out that GENLIN did not work for conditional logistic
regression (although it worked for conditional Poisson or log-binomial
models), because the outcome can't be matched, only the exposition.
Therefore, for matched case-control studies, GENLIN can't be used,
although it is perfect for matched cohort-studies. Since Doc Ash
metnions a matched case-control study, he can't use GENLIN, and should
use conditional logistic regression, either using NOMREG, or using COXREG.

Best regards,
Marta GG


Bruce Weaver wrote:

>
> Another option, of course, is to use generalized estimating equations (GEE)
> via GENLIN.  I posted something to the SPSS newsgroup after using that
> method to reanalyze some old data that I'd analyzed with Stata the first
> time around.  The GENLIN results from SPSS were very similar to the
> conditional logistic regression results from Stata.  Unfortunately, I can't
> find that old discussion right now to give the link.  But here is a
> comparison of the results for 4 different outcome variables.  View in
> Courier or some other fixed font (e.g., copy and paste into a text editor).
>
> ----------------------------------------------------------
> Outcome      | Odds Ratio   [95% Conf. Interval]   METHOD
> ----------------------------------------------------------
> Outcome 1 gp |   1.599      1.064       2.403         1
>              |   1.606      1.072       2.406         2
> ----------------------------------------------------------
> Outcome 2 gp |   1.626      1.126       2.346         1
>              |   1.631      1.128       2.358         2
> ----------------------------------------------------------
> Outcome 3 gp |   0.755      0.536       1.065         1
>              |   0.839      0.616       1.142         2
> ----------------------------------------------------------
> Outcome 4 gp |   1.579      0.734       3.340         1
>              |   1.585      0.735       3.419         2
> ----------------------------------------------------------
>
> Method 1:  Conditional logistic regression (Stata)
> Method 2:  GEE (via GENLIN in SPSS)
>
>
>
> -----
> --
> Bruce Weaver
> [hidden email]
> http://sites.google.com/a/lakeheadu.ca/bweaver/
> "When all else fails, RTFM."
>
> NOTE:  My Hotmail account is not monitored regularly.
> To send me an e-mail, please use the address shown above.
> --
> View this message in context: http://old.nabble.com/matched-case-control-data-analysis-help-tp28568766p28575575.html
> Sent from the SPSSX Discussion mailing list archive at Nabble.com.
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
> [hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD
>
>

--
For miscellaneous SPSS related statistical stuff, visit:
http://gjyp.nl/marta/

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Bruce Weaver]

Hi Marta.  Now that you mention it, I do remember your series of posts, but I'll have to review them.  I will make one comment now, though:  I don't think the GEE approach (with GENLIN) is meant to perform conditional logistic regresssion per se; rather, I think it is considered an alternative to conditional logistic regression.  The following article, for example, suggests it should be fine for analysis of correlated binary outcomes:

   http://aje.oxfordjournals.org/cgi/content/abstract/157/4/364

Cheers,
Bruce

Marta García-Granero-2 wrote

Hi Bruce:

Don't you remember my series of posts concerning the same topic? I
finally found out that GENLIN did not work for conditional logistic
regression (although it worked for conditional Poisson or log-binomial
models), because the outcome can't be matched, only the exposition.
Therefore, for matched case-control studies, GENLIN can't be used,
although it is perfect for matched cohort-studies. Since Doc Ash
metnions a matched case-control study, he can't use GENLIN, and should
use conditional logistic regression, either using NOMREG, or using COXREG.

Best regards,
Marta GG


Bruce Weaver wrote:
>
> Another option, of course, is to use generalized estimating equations (GEE)
> via GENLIN.  I posted something to the SPSS newsgroup after using that
> method to reanalyze some old data that I'd analyzed with Stata the first
> time around.  The GENLIN results from SPSS were very similar to the
> conditional logistic regression results from Stata.  Unfortunately, I can't
> find that old discussion right now to give the link.  But here is a
> comparison of the results for 4 different outcome variables.  View in
> Courier or some other fixed font (e.g., copy and paste into a text editor).
>
> ----------------------------------------------------------
> Outcome      | Odds Ratio   [95% Conf. Interval]   METHOD
> ----------------------------------------------------------
> Outcome 1 gp |   1.599      1.064       2.403         1
>              |   1.606      1.072       2.406         2
> ----------------------------------------------------------
> Outcome 2 gp |   1.626      1.126       2.346         1
>              |   1.631      1.128       2.358         2
> ----------------------------------------------------------
> Outcome 3 gp |   0.755      0.536       1.065         1
>              |   0.839      0.616       1.142         2
> ----------------------------------------------------------
> Outcome 4 gp |   1.579      0.734       3.340         1
>              |   1.585      0.735       3.419         2
> ----------------------------------------------------------
>
> Method 1:  Conditional logistic regression (Stata)
> Method 2:  GEE (via GENLIN in SPSS)
>
>
>
> -----
> --
> Bruce Weaver
> [e-mail]
> http://sites.google.com/a/lakeheadu.ca/bweaver/
> "When all else fails, RTFM."
>
> NOTE:  My Hotmail account is not monitored regularly.
> To send me an e-mail, please use the address shown above.
> --
> View this message in context: http://old.nabble.com/matched-case-control-data-analysis-help-tp28568766p28575575.html
> Sent from the SPSSX Discussion mailing list archive at Nabble.com.
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
> [e-mail] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD
>
>


--
For miscellaneous SPSS related statistical stuff, visit:
http://gjyp.nl/marta/

=====================
To manage your subscription to SPSSX-L, send a message to
[e-mail] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD