That’s how the interaction model looks - as you predicted, significant interaction between cMyC & creatinine:

```
> model1 <- lrm(AMI ~ MyC_0h + creatinine + gender + age_y + dm_base + chol_base + htn_base + previousMI + smoke_tot, data=dat)
> model2 <- lrm(AMI ~ MyC_0h * creatinine + gender + age_y + dm_base + chol_base + htn_base + previousMI + smoke_tot, data=dat)
> model1
Frequencies of Missing Values Due to Each Variable
AMI MyC_0h creatinine gender age_y dm_base chol_base htn_base previousMI
0 0 84 0 0 0 0 0 0
smoke_tot
102
Logistic Regression Model
lrm(formula = AMI ~ MyC_0h + creatinine + gender + age_y + dm_base +
chol_base + htn_base + previousMI + smoke_tot, data = dat)
Model Likelihood Discrimination Rank Discrim.
Ratio Test Indexes Indexes
Obs 605 LR chi2 122.73 R2 0.282 C 0.792
0 472 d.f. 9 g 1.427 Dxy 0.583
1 133 Pr(> chi2) <0.0001 gr 4.168 gamma 0.583
max |deriv| 1e-06 gp 0.177 tau-a 0.200
Brier 0.128
Coef S.E. Wald Z Pr(>|Z|)
Intercept -3.6535 0.7315 -4.99 <0.0001
MyC_0h 0.0022 0.0004 5.11 <0.0001
creatinine -0.0071 0.0038 -1.87 0.0610
gender=male 1.0364 0.2693 3.85 0.0001
age_y 0.0203 0.0094 2.16 0.0310
dm_base=Yes -0.6064 0.3396 -1.79 0.0742
chol_base=Yes 0.2359 0.2977 0.79 0.4281
htn_base=Yes 0.1179 0.2443 0.48 0.6294
previousMI=Yes -0.0451 0.2574 -0.18 0.8610
smoke_tot=Yes 0.5989 0.2664 2.25 0.0246
> model2
Frequencies of Missing Values Due to Each Variable
AMI MyC_0h creatinine gender age_y dm_base chol_base htn_base previousMI
0 0 84 0 0 0 0 0 0
smoke_tot
102
Logistic Regression Model
lrm(formula = AMI ~ MyC_0h * creatinine + gender + age_y + dm_base +
chol_base + htn_base + previousMI + smoke_tot, data = dat)
Model Likelihood Discrimination Rank Discrim.
Ratio Test Indexes Indexes
Obs 605 LR chi2 160.92 R2 0.359 C 0.813
0 472 d.f. 10 g 2.780 Dxy 0.627
1 133 Pr(> chi2) <0.0001 gr 16.125 gamma 0.627
max |deriv| 0.0001 gp 0.195 tau-a 0.215
Brier 0.119
Coef S.E. Wald Z Pr(>|Z|)
Intercept -4.3255 0.7745 -5.59 <0.0001
MyC_0h 0.0123 0.0029 4.28 <0.0001
creatinine 0.0027 0.0049 0.56 0.5784
gender=male 0.9947 0.2899 3.43 0.0006
age_y 0.0158 0.0100 1.59 0.1122
dm_base=Yes -0.4160 0.3351 -1.24 0.2145
chol_base=Yes 0.1780 0.3139 0.57 0.5707
htn_base=Yes 0.1187 0.2552 0.47 0.6419
previousMI=Yes -0.0423 0.2677 -0.16 0.8745
smoke_tot=Yes 0.6103 0.2780 2.20 0.0281
MyC_0h * creatinine -0.0001 0.0000 -4.11 <0.0001
> lrtest(model1, model2)
Model 1: AMI ~ MyC_0h + creatinine + gender + age_y + dm_base + chol_base +
htn_base + previousMI + smoke_tot
Model 2: AMI ~ MyC_0h * creatinine + gender + age_y + dm_base + chol_base +
htn_base + previousMI + smoke_tot
L.R. Chisq d.f. P
3.819232e+01 1.000000e+00 6.410432e-10
```

What I haven’t figured out yet is how to incorporate the interaction in a model using restricted cubic splines - as the argument rcs(MyC_0h, 4) * creatinine as used before when using splines does not compute:

```
singular information matrix in lrm.fit (rank= 13 ). Offending variable(s):
MyC_0h'' * creatinine MyC_0h''
Error in lrm(AMI ~ rcs(MyC_0h, 4) * creatinine + gender + age_y + dm_base + :
Unable to fit model using “lrm.fit”
```