How to interpret the odds ratio in proc glimmix with multi distribution and interaction?

SAS-questioner — Thu, 18 Apr 2024 21:47:49 GMT

Hi, I am conducting hierarchical generalized linear model with a dichotomous (1/0) and polytomous (large/median/small) outcome variables, using the code below separately:

proc glimmix data= one method=laplace noclprint;
class id group;
model dich_outcome (event='1') = group|time / dist=binary link=logit solution oddsratio;
random intercept/ subject=ID cl type=vc solution cl;
run;

proc glimmix data= two method=laplace noclprint;
class id group;
model poly_outcome (event='large') = group time / dist=multi link=clogit solution oddsratio;
random intercept/ subject=ID cl type=vc solution cl;
run;

Since the interaction for dichotomous outcome is significant, I keep it in the model. And interaction for polytomous outcome is not significant, I removed it from the model.

Below output is for dichotomous:

                                          Solutions for Fixed Effects
                                                      Standard
        Effect                Group      Estimate       Error       DF    t Value    Pr > |t|
         Intercept                          1.6820     0.6270      78        2.68     0.0089
         Group                 0           -2.1304     0.7764      256      -2.74     0.0065
        group                  1               0           .        .        .         .
        time                              -0.6136      0.2006      256      -3.06      0.0025
        time*group             0           0.5303      0.2399      256       2.21      0.0279
        time*group            1               0           .        .        .         .
                                        Odds Ratio Estimates
                                                                          95% Confidence
         Group        time    group       _time    Estimate       DF           Limits

           0              2     0             2       0.311      256       0.088       1.094
           0              3     0              2       0.920      256       0.708       1.196
           1              3     0             2       0.541      256       0.365       0.804

Below output is for polytomous outcome:

                                               Solutions for Fixed Effects

                                                          Standard
    Effect           UPCR         Group      Estimate       Error       DF    t Value    Pr > |t|

    Intercept        large                     -1.9201      0.7731       71      -2.48      0.0154
    Intercept        median                     2.4002      0.7877       71       3.05      0.0032
    Group                           0           1.2784      0.8877      136       1.44      0.1521
    Group                           1               0           .        .        .         .
    time                                       -0.7128      0.1790      136      -3.98      0.0001


                                        Odds Ratio Estimates

                                                                        95% Confidence
         Group        time    Group       _time    Estimate       DF           Limits

          0            1.5    1             1.5       3.591      136       0.621      20.780
                       2.5                  1.5       0.490      136       0.344       0.698

My questions are:

1. I understand how to interpret the odds ratio for dichotomous outcome, I am not sure, why there is no odds ratios for 'time' alone?

2. In the solutions for fix effect, there is a p value for group, time, and interaction, but in the odds ratio, we only have two odds ratio for time and group interaction: 0.920[0.708, 1.196],0.541[0.365,0.804] , if I need to report p value and odds ratio, should I put the p value of interaction for both odds ratio, or the p value for time is one of them?

3. Since the polytomous outcome is a three-level categorical outcome, why in the odds ratio output, there is only one odds ratio for group and time each? In this case, how to interpret the 'median', and 'low' using odds ratio? For example, comparing group 1, group 0 was 3 times more likely be in the 'large'. But how to find out the odds ratio for 'median'/'low'?

Re: How to interpret the odds ratio in proc glimmix with multi distribution and interaction?

sbxkoenk — Fri, 19 Apr 2024 07:56:42 GMT

@SAS-questioner wrote:
3. Since the polytomous outcome is a three-level categorical outcome, why in the odds ratio output, there is only one odds ratio for group and time each? In this case, how to interpret the 'median', and 'low' using odds ratio? For example, comparing group 1, group 0 was 3 times more likely be in the 'large'. But how to find out the odds ratio for 'median'/'low'?

Home > Analytics > Stat Procs >
Obtaining odds ratio estimates for each level of ordinal variable in PROC GLIMMIX
https://communities.sas.com/t5/Statistical-Procedures/Obtaining-odds-ratio-estimates-for-each-level-of-ordinal/td-p/725370

BR,
Koen

topic How to interpret the odds ratio in proc glimmix with multi distribution and interaction? in Statistical Procedures

How to interpret the odds ratio in proc glimmix with multi distribution and interaction?

Re: How to interpret the odds ratio in proc glimmix with multi distribution and interaction?