Analysis of repeated measures data with proc mixed

[Yasu]

Dear all,

Now I'm trying to analyze repeated measures data with proc mixed.

The sas code is as follows and X denotes the variable of laboratory data such as blood pressure.

  PROC MIXED DATA=DATA;

        CLASS ID TIME;

        MODEL X=TIME / DDFM=KR;

        REPEATED  TIME / SUBJECT=ID TYPE=UN ;

        LSMEANS TIME/ DIFF=CONTROL('1')

        ADJUST=DUNNETT;

  RUN;

Though the code seems to work correctly with  DDFM=KR and TYPE=UN,

the warning message “Warning: Stopped because of infinite likelihood” occurs  with DDFM=KR and TYPE=CS.

I know such a message can occur when there are some records with the same ID and same time value in the dataset.

But the data I want to analyze does't have such records.

Why does the message occur?

Should I set  DDFM=KR and TYPE=UN rather than  DDFM=KR and TYPE=CS?

I'd really appreciate it if someone would help me.

Thanks in advance.

Yasu

[SteveDenham]

This is a difficult question.  Could you share the covariance estimates under the TYPE=UN specification?  I am thinking that there may be some pathological values.

Other things to try:  Try a variety of other types.  The first that comes to mind is CSH.  If the timepoints are evenly spaced, or very nearly evenly spaced, try AR(1) and ARH(1).  If all of these lead to the infinite likelihood, and UN does not, then there is something very unusual about the data.

Try using PROC GLIMMIX with a different optimization method using the NLOPTIONS statement.

Using the results from the TYPE=UN fit, see if the estimates can be put into a form that would give starting values under the TYPE=CS specification.  Use the median of the diagonals and off-diagonals as starting parameters in the PARMS statement.  This probably would only work if the infinite likelihood occurs on the first iteration--if it shows up after the procedure has done several iterations then it is almost surely data pathology.

Steve Denham

[Yasu]

Dear Steve,

Thank you for your continued sopport.

the covariance estimates under the TYPE=UN specification as follows:

Estimated R Matrix for SUBJID 0001-001

                                                      Row        Col1        Col2        Col3        Col4        Col5

                                                        1      331.08      301.34      278.17      269.69      260.96
                                                        2      301.34      330.16      302.41      285.32      270.93
                                                        3      278.17      302.41      419.88      356.43      283.12

                                                        4      269.69      285.32      356.43      473.49      320.51

                                                        5      260.96      270.93      283.12      320.51      400.03

  Covariance Parameter Estimates                                                                

                                                                   Cov Parm    Subject    Estimate

                                                                     UN(1,1)     SUBJID       331.08
                                                                     UN(2,1)     SUBJID       301.34
                                                                     UN(2,2)     SUBJID       330.16
                                                                     UN(3,1)     SUBJID       278.17
                                                                     UN(3,2)     SUBJID       302.41
                                                                     UN(3,3)     SUBJID       419.88
                                                                     UN(4,1)     SUBJID       269.69
                                                                     UN(4,2)     SUBJID       285.32
                                                                     UN(4,3)     SUBJID       356.43
                                                                     UN(4,4)     SUBJID       473.49
                                                                     UN(5,1)     SUBJID       260.96
                                                                     UN(5,2)     SUBJID       270.93
                                                                     UN(5,3)     SUBJID       283.12
                                                                     UN(5,4)     SUBJID       320.51
                                                                     UN(5,5)     SUBJID       400.03
                                                                     UN(6,1)     SUBJID       255.04
                                                                     UN(6,2)     SUBJID       264.57
                                                                     UN(6,3)     SUBJID       276.21
                                                                     UN(6,4)     SUBJID       294.30
                                                                     UN(6,5)     SUBJID       323.57
                                                                     UN(6,6)     SUBJID       373.06
                                                                     UN(7,1)     SUBJID       248.73
                                                                     UN(7,2)     SUBJID       256.85
                                                                     UN(7,3)     SUBJID       274.29
                                                                     UN(7,4)     SUBJID       284.68
                                                                     UN(7,5)     SUBJID       295.96
                                                                     UN(7,6)     SUBJID       310.99
                                                                     UN(7,7)     SUBJID       351.47
                                                                     UN(8,1)     SUBJID       249.61
                                                                     UN(8,2)     SUBJID       257.29
                                                                     UN(8,3)     SUBJID       275.29
                                                                     UN(8,4)     SUBJID       285.32
                                                                     UN(8,5)     SUBJID       296.04
                                                                     UN(8,6)     SUBJID       298.96
                                                                     UN(8,7)     SUBJID       329.12
                                                                     UN(8,8)     SUBJID       360.03

Looking at the covariance estimates under the TYPE=UN specification, the assumption of compound symmetry is no longer valid.

I' d really appreciate it if you would give me your  inputs.

Thanks,

Yasu

[SteveDenham]

There looks like a relatively constant correlation from time(i) to time(i+1), which implies to me that an autoregressive error structure may be appropriate.  Given that the diagonal entries seem relatively constant, consider type=AR(1) if your time points are equally or very nearly equally spaced, or type=sp(pow)(time1) if they are not evenly spaced.  You will need to construct time1 as a continuous variable in previous data step (time1=time), since time is specified as a categorical variable in the class statement.

I still fear that there may be some data pathology that is causing the infinite likelihood.

See how this works.

Steve Denham

[Yasu]

The time points are equally spaced for each subject, so I tried type=AR(1) and type=sp(pow)(time1).

But the "infinite likelihood" error message occured with both type specification.

I'n not sure what is causing the infinite liklihood error yet...

[SteveDenham]

Yuck.  I had hoped that would work.

OK--when does the infinite likelihood error occur?  Is it at the initial iteration, or does it occur after several iterations?  If it is the first case, it is almost certainly a problem with a duplicate record for one of the subjects at one of the timepoints.  If it is the second, what is going on in the iteration history?  Does it look like there is a relatively smooth history for the objective function up until something happens and it jumps off the tracks?  Or is the history erratic?

Can't say I have an answer yet, but knowing the answers to these questions might help?

Also, it might be time to open a ticket with tech support, especially if you can share the data with them.

Steve Denham

[Yasu]

Dear Steve,

I really appreciate your continued support.

Looking at the log, it seems to occur at the initial iteration. And I looked into the data once again, but there are no such a record.

In addition, I have another question. If there are some duplicate record, why does not the infinite likelihood occur with type=UN specificstions?

Thanks,

Yasu

[SteveDenham]

I jerryrigged up some data and fit it with type=un.  Duplicates also lead to the infinite likelihood with the message about a nonpositive definite R matrix. If I try GLIMMIX, the warning is that it failed to obtain mivque starting values. For HPMIXED, I get a specific error message that duplicate measures have been detected.

So, that is not the problem.  Have you tried PROC HPMIXED?  Something like:

PROC HPMIXED DATA=DATA; 

        CLASS ID TIME;

        MODEL X=TIME ;

          test time;

        REPEATED  TIME / SUBJECT=ID TYPE=AR(1) ;

        LSMEANS TIME/ DIFF=CONTROL('1'):

        *ADJUST=DUNNETT;

  RUN;

Adjustments are not available in HPMIXED (at least according to my documentation), thus the commenting out of adjust=dunnett.  So if you need them, then it would probably be necessary to output using the ODS statement, and then post-process using PROC MULTTEST.  And then only if HPMIXED actually works.

I am thinking that you REALLY, REALLY need to open a ticket with tech support.

Good luck.

Steve Denham

[Yasu]

Dear Steve,

I really appreciate your continued support !

I'll check with tech support about this.

Once agan, thank you  for your kind reply.

Yasu