conditional logistic regression

[C_visser]

Dear everyone,

I am trying to use a conditional logistic regression in SPSS. I am using
SPSS version 25. I have looked at a lot of youtube videos explaining the
cox-regression method, but something is going wrong as it seems it is
censoring too many cases. I think this is due to our study design and
studied exposures, but I am not entirely sure (as I am just a beginner). Our
study design is the following:

We have participants with an certain exposure and want to compare their
measurements of certain values before the exposure and afterwards. In this
study they will receive this exposure two times. So we want to compare a
certain variable at the time before the exposure, the time between the first
and second exposure and after the second exposure. As we are measuring these
variables in the same subjects, we think we should use a cross-over design
and paired statics tests. Right now, I want to see if the subjects after the
first and second exposure have a higher risk of a certain complication
compared to before the exposure. An epidemiologist suggested using a
conditional logistic regression.

We have tried to do this test in spss by using the cox regression method. We
have coded a casecontrol variable (before exposure = 2, after exposure = 1)
as time, the depent variable as a categorical variable with 0 = no
complication, 1 = complication, timing as covariate (0 = before exposure; 1
= after exposure) and paired the measurement before and after exposure in
the same subject using an identical identifier. However, when I compare the
outcome with the mcnemar test it is not the same. Also it seems like spss is
running the test on a smaller number of subjects than I expected.

Can somebody help me with this problem?

Kind regards,

Chantal



--
Sent from: http://spssx-discussion.1045642.n5.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

[Bruce Weaver]

Hello Chantal.  Conditional logistic regression is certainly one option when
you have paired observations and a dichotomous outcome.  However,
generalized estimating equations provide another way of dealing with the
correlated nature of the paired observations.  And GENLIN allows you do use
that approach directly, with no need to cajole some other command (e.g.,
COXREG) into estimating the model you want.  You might find this old thread
in comp.soft-sys.stat.spss useful.

https://groups.google.com/g/comp.soft-sys.stat.spss/c/zkxR016mZxM

There are also plenty of articles you can use to defend use of GEE (if
necessary).  E.g., here is one I found useful:

Hanley, J. A., Negassa, A., Edwardes, M. D. D., & Forrester, J. E. (2003).
Statistical analysis of correlated data using generalized estimating
equations: an orientation. American journal of epidemiology, 157(4),
364-375.  https://academic.oup.com/aje/article/157/4/364/78911


HTH.



C_visser wrote

> Dear everyone,
>
> I am trying to use a conditional logistic regression in SPSS. I am using
> SPSS version 25. I have looked at a lot of youtube videos explaining the
> cox-regression method, but something is going wrong as it seems it is
> censoring too many cases. I think this is due to our study design and
> studied exposures, but I am not entirely sure (as I am just a beginner).
> Our
> study design is the following:
>
> We have participants with an certain exposure and want to compare their
> measurements of certain values before the exposure and afterwards. In this
> study they will receive this exposure two times. So we want to compare a
> certain variable at the time before the exposure, the time between the
> first
> and second exposure and after the second exposure. As we are measuring
> these
> variables in the same subjects, we think we should use a cross-over design
> and paired statics tests. Right now, I want to see if the subjects after
> the
> first and second exposure have a higher risk of a certain complication
> compared to before the exposure. An epidemiologist suggested using a
> conditional logistic regression.
>
> We have tried to do this test in spss by using the cox regression method.
> We
> have coded a casecontrol variable (before exposure = 2, after exposure =
> 1)
> as time, the depent variable as a categorical variable with 0 = no
> complication, 1 = complication, timing as covariate (0 = before exposure;
> 1
> = after exposure) and paired the measurement before and after exposure in
> the same subject using an identical identifier. However, when I compare
> the
> outcome with the mcnemar test it is not the same. Also it seems like spss
> is
> running the test on a smaller number of subjects than I expected.
>
> Can somebody help me with this problem?
>
> Kind regards,
>
> Chantal
>
>
>
> --
> Sent from: http://spssx-discussion.1045642.n5.nabble.com/
>
> =====================
> To manage your subscription to SPSSX-L, send a message to

> LISTSERV@.UGA

>  (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
[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.

--
Sent from: http://spssx-discussion.1045642.n5.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

[Rich Ulrich]

Either I don't understand the description, or you are

unnecessarily confusing in what you describe.  I don't see

anything that suggests a crossover design between two

different treatments (beyond that blunt mention of Xover). 

I don't see what sets up a conditional logistic.

The best I can figure out is that you have three data points

for "complication" (no, yes), which are at Times identified as

(Before, After1 and After2).  The "wide" layout should make it

easy to see how many cases are in an analysis - "Missing"

at any Time will drop a case for some analyses.

That layout does suggest simple tests using McNemar's test

for changes, between the sequential conditions. Or (maybe)

between Before and each After period.

I don't see a statement of hypotheses, or a mention of why

the design wanted two test periods.

I've read the posting at least three times after figuring out

my questions -- If I'm missing something, please let me know.

--

Rich Ulrich

---

From: SPSSX(r) Discussion <[hidden email]> on behalf of C_visser <[hidden email]>
Sent: Wednesday, March 31, 2021 4:55 AM
To: [hidden email] <[hidden email]>
Subject: conditional logistic regression

 

Dear everyone,

I am trying to use a conditional logistic regression in SPSS. I am using
SPSS version 25. I have looked at a lot of youtube videos explaining the
cox-regression method, but something is going wrong as it seems it is
censoring too many cases. I think this is due to our study design and
studied exposures, but I am not entirely sure (as I am just a beginner). Our
study design is the following:

We have participants with an certain exposure and want to compare their
measurements of certain values before the exposure and afterwards. In this
study they will receive this exposure two times. So we want to compare a
certain variable at the time before the exposure, the time between the first
and second exposure and after the second exposure. As we are measuring these
variables in the same subjects, we think we should use a cross-over design
and paired statics tests. Right now, I want to see if the subjects after the
first and second exposure have a higher risk of a certain complication
compared to before the exposure. An epidemiologist suggested using a
conditional logistic regression.

We have tried to do this test in spss by using the cox regression method. We
have coded a casecontrol variable (before exposure = 2, after exposure = 1)
as time, the depent variable as a categorical variable with 0 = no
complication, 1 = complication, timing as covariate (0 = before exposure; 1
= after exposure) and paired the measurement before and after exposure in
the same subject using an identical identifier. However, when I compare the
outcome with the mcnemar test it is not the same. Also it seems like spss is
running the test on a smaller number of subjects than I expected.

Can somebody help me with this problem?

Kind regards,

Chantal



--
Sent from: http://spssx-discussion.1045642.n5.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

===================== 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