cox regression (where observations are not independent)

[Paul Mcgeoghan]

Hi,

I have a customer who has carried out the following experiment.
He has data as follows:

     Pt_ID Reg_Seq Regimen           Gender Risk_Factor   Ethnicity Employ
Duration Disc_Stat
         1             1                51                           1         3
      1        1          141.57               2
         2             1                51                           1         3
      1        2          211.57               2
         3             1                47                           1         3
      1        1          103.71               1
         3             2                49                           1         3
      1        1           98.57               2
         5             1               127                           1         3
      1        1           21.57               1
         5             2               129                           1         3
      1        1            4.43               1
         5             3               148                           1         3
      1        1           17.43               1

He is looking at Duration until patient gets the symptom which is Disc_Stat=1

Reg_Sequence represents the ordered number of regimen sequence.
Regimen Sequence is a combination of drugs given to the patient.
Patients in the sample could change drug regimen composition and each change
marked an
increase in the number of Reg_Sequence.

The duration will reset to zero with each change of regimen sequence.

I think there could be a problem with the design of the study (the possibility
of a carryover effect from previous Regimen sequences)

What type of analysis can be done on the above as the assumptions of Cox
Regression are violated?
Can the customer just take the last Regimen sequence and analyse that using Cox
Regression?

Paul


==================
Paul McGeoghan,
Application support specialist (Statistics and Databases),
University Infrastructure Group (UIG),
Information Services,
Cardiff University.
Tel. 02920 (875035).

[Marta García-Granero]

Hi Paul

I think your customer should use Cox regression with a covariate (drug
regimen) dependent on time. You can find a nice explanation in chapter 7 of
Hosmer&Lemeshow book (Applied Survival Analysis - Regression Modelling
of Time to Event Data; John Wiley&Sons Eds)

Tuesday, August 22, 2006, 11:26:11 AM, You wrote:

PM> I have a customer who has carried out the following experiment.
PM> He has data as follows:

PM>      Pt_ID Reg_Seq Regimen           Gender Risk_Factor   Ethnicity Employ
PM> Duration Disc_Stat
PM>          1             1                51                    1         3
PM>       1        1          141.57               2
PM>          2             1                51                    1         3
PM>       1        2          211.57               2
PM>          3             1                47                    1         3
PM>       1        1          103.71               1
PM>          3             2                49                    1         3
PM>       1        1           98.57               2
PM>          5             1               127                    1         3
PM>       1        1           21.57               1
PM>          5             2               129                    1         3
PM>       1        1            4.43               1
PM>          5             3               148                    1         3
PM>       1        1           17.43               1

PM> He is looking at Duration until patient gets the symptom which is Disc_Stat=1

PM> Reg_Sequence represents the ordered number of regimen sequence.
PM> Regimen Sequence is a combination of drugs given to the patient.
PM> Patients in the sample could change drug regimen composition and each change
PM> marked an
PM> increase in the number of Reg_Sequence.

PM> The duration will reset to zero with each change of regimen sequence.

PM> I think there could be a problem with the design of the study (the possibility
PM> of a carryover effect from previous Regimen sequences)

PM> What type of analysis can be done on the above as the assumptions of Cox
PM> Regression are violated?
PM> Can the customer just take the last Regimen sequence and analyse that using Cox
PM> Regression?



--
Regards,
Dr. Marta García-Granero,PhD           mailto:[hidden email]
Statistician

---
"It is unwise to use a statistical procedure whose use one does
not understand. SPSS syntax guide cannot supply this knowledge, and it
is certainly no substitute for the basic understanding of statistics
and statistical thinking that is essential for the wise choice of
methods and the correct interpretation of their results".

(Adapted from WinPepi manual - I'm sure Joe Abrahmson will not mind)