Plotting cumulative hazard function using the Nelson Aalen estimator for a time-varing exposure

[blackandwhite]

Hi,

I am trying to create a plot of the cumulative hazard of an outcome over time for a time-varying exposure using the Nelson-Aalen estimator in SAS.

The figure should look something like the Figure 1 attached here. I also need to know the breakdown of the risk set, no events etc. at each time point.

The graph with additional details are in this article:

Munoz-Price LS, Frencken JF, Tarima S, Bonten M. Handling Time-dependent Variables: Antibiotics and Antibiotic Resistance. Clin Infect Dis 2016;62:1558-1563.

Any advice would be really appreciated.

Thanks you.

[Reeza]

---

@blackandwhite wrote:

Hi,

I am trying to create a plot of the cumulative hazard of an outcome over time for a time-varying exposure using the Nelson-Aalen estimator in SAS.

The figure should look something like the Figure 1 attached here. I also need to know the breakdown of the risk set, no events etc. at each time point.

The graph with additional details are in this article:

Munoz-Price LS, Frencken JF, Tarima S, Bonten M. Handling Time-dependent Variables: Antibiotics and Antibiotic Resistance. Clin Infect Dis 2016;62:1558-1563.

Any advice would be really appreciated.

Thanks you.

---

Ok, which part are you having problems with? You haven't provided data or anything so it's a bit hard to comment, but I'm going to guess you want a PROC PHREG to start out with. 

Here's the documentation reference:

https://documentation.sas.com/?docsetId=statug&docsetTarget=statug_phreg_examples06.htm&docsetVersio...

That should get you started.

[Reeza]

After reviewing the paper, actually, this is only PROC LIFETEST that you need, which is much simpler. 

You can specify NELSON on the PROC LIFETEST statement. 

Unfortunately I don't see an easy way to get the plots of those estimates. 

PROC LIFETEST documentation:

https://documentation.sas.com/?docsetId=statug&docsetTarget=statug_lifetest_syntax01.htm&docsetVersi...

Here's a tutorial on getting the estimates, saving them to a data set and plotting it 'manually' using SGPLOT. You'll just have to change the column references. 

https://blogs.sas.com/content/graphicallyspeaking/2018/02/19/survival-plot-twist-using-sgplot-proced...

The full code is at the bottom of the post. I would suggest running the example and then doing yours step by step to convert the code to work with your data. 

EDIT: added reference to PROC LIFETEST

---

@blackandwhite wrote:

Hi,

I am trying to create a plot of the cumulative hazard of an outcome over time for a time-varying exposure using the Nelson-Aalen estimator in SAS.

The figure should look something like the Figure 1 attached here. I also need to know the breakdown of the risk set, no events etc. at each time point.

The graph with additional details are in this article:

Munoz-Price LS, Frencken JF, Tarima S, Bonten M. Handling Time-dependent Variables: Antibiotics and Antibiotic Resistance. Clin Infect Dis 2016;62:1558-1563.

Any advice would be really appreciated.

Thanks you.

---

[blackandwhite]

Hi and thanks for your answer.

Proc lifetest only works for dataset structured with a row per patient ID, not when it is in counting process, which is necessary to accommodate a time-varying exposure.

Perhaps I misunderstood your answer?

[Reeza]

I'm basing this on the methods used in the paper, which seem to be just PROC LIFETEST. 

If you have a different methodology than the paper, then you do need to explain your methodology and provide some reference for the calculation. 

PROC PRHEG will handle time-dependent so that may be the option that works for you.

It's been a few days, so the details are not fresh in my mind anymore.  

[blackandwhite]

Proc phreg won't work to plot a NA curve for a TD variable, neither proc lifetest.

I had already looked into it