Calcite | Level 5

## Survival Analysis

Hello friends,

I am in need of help - trying to figure out if Proc Lifetest is the right way to go with the analysis I want to perform.

I am working on a "psychoactive substance use" dataset. Trying to estimate the cumulative probabilities of transitioning from simple use to a "use disorder" (event).To my understanding examining transition is still under the umbrella of survival analysis, right?

The thing is that each observation (over 35,000) has a different start time (age started using the substance).

I understand there is an extension for basic survival analyses in SAS that allows different start times for each observation.

Any help would be much appreciated.

Thanks!

3 REPLIES 3
Super User

## Re: Survival Analysis

The different start times don't matter for survival analysis.

You essentially scale it to have time at start be 0 for everything and then time increments. Unless you believe the pattern is changing over time.

@Psychiatrist wrote:

Hello friends,

I am in need of help - trying to figure out if Proc Lifetest is the right way to go with the analysis I want to perform.

I am working on a "psychoactive substance use" dataset. Trying to estimate the cumulative probabilities of transitioning from simple use to a "use disorder" (event).To my understanding examining transition is still under the umbrella of survival analysis, right?

The thing is that each observation (over 35,000) has a different start time (age started using the substance).

I understand there is an extension for basic survival analyses in SAS that allows different start times for each observation.

Any help would be much appreciated.

Thanks!

Calcite | Level 5

## Re: Survival Analysis

Hi,

Unfortunately I am still in need of assistance.

I am moderately experienced with SAS and been trying to use SAS lifetest for an analysis. Reading about this procedure and survival analysis in general It is still unclear to me if this is the type of analysis I need.
I have a retrospective dataset. Each observation (over 30K) has a discrete age of onset of drug use, which should be considered the start time ( I guess..). The event is the age of transition to drug use disorder. I want to create a plot (and tables) of the estimated projections of the cumulative probability of transitioning from use to disorder within the first year of substance use onset, the first decade after substance use onset, and lifetime in the general population and by two more dichotomous variables. A previous study performed such an analysis using the standard actuarial method (Machin et al., 2006) as implemented in PROC LIFETEST in SAS. They further used the log-rank test to determine whether survival curves differed statistically across certain variables.

I would like to perform a similar analysis and need help with code.

Could anyone perhaps assist or direct me in how to receive more specific tutorial advice..

THANK YOU!

Super User

## Re: Survival Analysis

Have you read through the documentation and worked through the examples? That's probably the recommended starting point.

https://documentation.sas.com/?cdcId=pgmsascdc&cdcVersion=9.4_3.4&docsetId=statug&docsetTarget=statu...

Then if you need details on a specific topic you can post back with a more detailed question, or search on lexjansen.com for survival tutorials.

There's also a book on the topic that's pretty good. If you're affiliated with an academic institution you should have digital access through your library as well.

https://www.amazon.ca/Survival-Analysis-Using-SAS-Practical/dp/1599946408

Discussion stats
• 3 replies
• 1028 views
• 0 likes
• 2 in conversation