Hello everyone,

I would appreciate it if someone could help answer my questions regarding a Cox proportional hazards model with time-varying covariates in SPSS.

Here's the case:

I am investigating the effect of two continuous variables (V1 and V2) measured at time points 3, 6, and 24 on mortality.

Additionally, I want to adjust for time-dependent covariates (CV1 and CV2) measured at time points 1, 3, and 24, as well as covariates (CV3 and CV4) measured at time points 6, 12, and 24 hours.

I restructured my data file into 'long format' so that each subject has 5 rows with T_start (patient at risk), T_end (censored or death), and the event (1 = death, 0 = censored).

I realized that the standard COXREG function cannot handle this data format, but the Complex Sample Cox Regression function in SPSS can.

However, it appears that the T_COV*V1 interaction is no longer needed when using the Complex Sample Cox Regression function. When I include Time (T_) as a time-dependent predictor, it is not possible to create an interaction term with V1. Additionally, when I add Time as a covariate along with V1, the SPSS output indicates that the B coefficient of time is set to zero because the parameter is redundant. Is it correct that in this setting it is not required to include the interaction with time (T_COV)? 

Additionally, is this the easiest way to handle the data as described in the case? I already read this section: Multiple Cases per Subject in Complex Samples Cox Regression - IBM Documentation 

Thank you in advance for any assistance.

Best regards,