Hi,
I've used the stmixed command by M. Crowther to model clustered survival data which has time-dependent effects. The help file is very useful however i'm uncertain about the interpretation of the value given for the higher level effect (unit).
Unfortunately I can't share the data and output in the standard way but have attached a screenshot of the output.
Treatment is a binary variable. I'm attempting to show the effect of treatment on survival, while accounting for time-dependent effects and clustering by unit.
The coefficients provided in the output are log hazard-ratios.
My understanding is the overall treatment effect is exp(0.6925). HR (2.00, 95% CI: 1.64-1.43), indicating an increased risk of death for the treatment group. This model can be used to demonstrate how the effect changes over time.
How should the unit effect be interpreted and presented? In the help file it discusses how this is the estimated frailty standard deviation, and demonstrates the level of heterogeneity in the baseline hazard function. I'm not familiar with this output compared to standard mixed models.
I've used the AIC and likelihood ratio test to show the model is improved when including this additional level.
Many thanks for your help
I've used the stmixed command by M. Crowther to model clustered survival data which has time-dependent effects. The help file is very useful however i'm uncertain about the interpretation of the value given for the higher level effect (unit).
Unfortunately I can't share the data and output in the standard way but have attached a screenshot of the output.
Treatment is a binary variable. I'm attempting to show the effect of treatment on survival, while accounting for time-dependent effects and clustering by unit.
The coefficients provided in the output are log hazard-ratios.
My understanding is the overall treatment effect is exp(0.6925). HR (2.00, 95% CI: 1.64-1.43), indicating an increased risk of death for the treatment group. This model can be used to demonstrate how the effect changes over time.
How should the unit effect be interpreted and presented? In the help file it discusses how this is the estimated frailty standard deviation, and demonstrates the level of heterogeneity in the baseline hazard function. I'm not familiar with this output compared to standard mixed models.
I've used the AIC and likelihood ratio test to show the model is improved when including this additional level.
Many thanks for your help
Attached Files
