Great video! I did not understand why you needed a separate dataset for the CIF. Would you still need it if you didnt have the log(waitTime) in the model?
Thanks, @skavenos! I’m glad to hear that you enjoyed the video
You need the separate data set so you can specify the values for the model’s covariates for which to plot the CIF, so that data set is different from the data set you used to estimate the model. Yes, you would still need the separate data set for the CIF plots even if you did not include log(waitTime) in the model. In that case, the separate data set would just include the values for the Group covariate you want included in the CIF plot. There is actually an example in the documentation that does this (which was the basis of my example). Apparently, I am unable to include the link to the example or my comment will be deleted. But in the documentation for the PHREG Procedure, the example is called “Analysis of Competing-Risks Data” and it’s Example 92.15 in the latest version of the documentation. So if you search for that (PHREG Procedure Example 92.15 “Analysis of Competing-Risks Data”), then you should be able to find it.
Hello, thank you for the helpful video! I have a question to the "Censored" variable, out of the perspective of a non-mathematican/ statistican. Sadly there are very rare sources in the internet about this subject and I have no help from a statistican. Referring to your leukemia example: If you got the groups "relapse" and "death before relapse", the variable "censored" contains all the patients, who did not get a relapse and did not die during the study-time, is this right? So it is no problem if all those patients get censored out? I am currently performing a Comprisk analysis, where a majority (around 50% of the patients) would fall under the "censored" group, so i dont consider this as a problem? Thank you for your answer.
Hi Max! Censoring is one of the target outcomes, and can be interpreted as ‘no event yet’ (in this case no relapse or death). All patients either relapse or die with probability one as T gets big, but censoring occurs because some patients haven’t experienced one of these two events at the time the analysis is run. This is known as right or ‘end of study’ censoring. The alternative to keeping censored observations in the analysis would be removing them. This would entail a loss of information and could bias the results. If you want to learn more about Survival Analysis we have some excellent classes on this topic. See 2.sas.com/6054M6cLe and 2.sas.com/6055M6cL5 .
Hi Hala, in this video, the narrator is using a dataset called BMT_CR. However we provide in the SASHELP library a dataset called SASHELP.BMT with the data for the same Klein and Moeschberger 2003 study. The SASHELP library should be available to you when you open SAS. If you are using our SAS OnDemand for Academics server, the SASHELP library is automatically defined and available to you for use when you start your SAS Studio session.
Hi, thank you for your video. In my dataset, there is only Time, but not log_wait_time, Should I create a column for log_wait_time, and how can I create it? Thanks!
@@Nw-ec8ke I looked at the predicted cumulative incidence function for each disease group at logWaitTime = 5.2 because that is the median value in the data set and it corresponds to roughly 6 months. But of course you are free to use the values that are of interest to you. As you noted, the times are different for each patient, so one of the goals of fitting the model is to then use it to make these sorts of predictions. Does that make sense? Thanks for watching and thanks for the questions!
Yes, to reproduce my analysis, you need to use the log of the wait time variable. In a DATA step, you can define a new variable with the existing waitTime variable as the input to the log function: logWaitTime=log(WaitTime); To make sure we are on the same page and you are using the same data set, the code to create the data set I use for the demo can be found in the paper “Cause-Specific Analysis of Competing Risks Using the PHREG Procedure” by Changbin Guo and Ying So from SAS Global Forum 2018 (apparently I am unable to provide the direct link to the paper). On page 3, that DATA step code creates the data set I use except I named it BMT_CR instead to avoid confusion with the SASHELP.BMT data set.
Tianshui, Please review the video. At time stamp 5:43 in the video, you'll see that the default confidence interval used in this task is 95%. If you want to change the confidence interval used for the task, you would either change this drop down choice or you would use the ALPHA= option in the PROC PHREG code. Please refer to the PROC PHREG documentation 2.sas.com/6058KSXGI for more information about using the ALPHA= option. We cannot provide detailed Tech Support for your own code here in the RUclips feedback area. Please either post your question in the SAS Community for statistical procedure questions, here: 2.sas.com/6059KSXGL or open a track with Tech Support. To open a track with Tech Support, fill out the form at this link: 2.sas.com/6051KSXGF .
Tianshui, if you have been having issues with your own results getting at 95%CI in your table, we would recommend reaching out in SAS Communities or technical support. See details in our last comment for next steps. If you would like us to create a track for you, you can email us your contact information and include your site number to SAS_Cares@sas.com
Thanks for your question.There are not any macro variables or macro programs used in this video. SAS titles and footnotes are controlled by the TITLE and FOOTNOTE statements in your code. If you want to alter the code in the the task to use a TITLE or FOOTNOTE statement, you would not necessarily need a macro program to do that. Also, you only need to work with SAS templates if you wanted to alter the style characteristics of a title. To open a track with Tech Support, fill out the form at this link: 2.sas.com/6059JiXLd . Your best resource for this question is to either open a track with Tech Support or, since you are using SAS OnDemand for Academics and SAS Studio, you can post your question in the SAS Community for the OnDemand users, here: 2.sas.com/6051JiXL5 (if the question is about SAS Studio) or here in the Statistical Procedures community: 2.sas.com/6052JiXLg (if your question is related to doing survival analysis).
In this video, the narrator is using a dataset called BMT_CR. However, we provide in the SASHELP library a dataset called SASHELP.BMT with the data for the same Klein and Moeschberger 2003 study. The SASHELP library should be available to you when you open SAS. If you are using our SAS OnDemand for Academics server, the SASHELP library is automatically defined and available to you for use when you start your SAS Studio session. We hope this helps!
A great resource for this type of question is our SAS Communities 2.sas.com/6050bUF4Q. These communities are a service hosted by SAS to facilitate the exchange of technical information pertaining to SAS products, services and best practices. For assistance with getting answers fast, click here 2.sas.com/6051bUF4v to see some helpful tips on how to post a great question! If you are not able to find the assistance you need, please feel free to speak with Technical Support by calling 800-727-0025, option 2.
@@SASUsers Dear .. can you please guide me where can I ask questions regarding SAS Academics .. please All I want to upload data file to a library and I can not do this
Checout out timestamp 3:52,-the narrator is using a dataset called BMT_CR. However we provide in the SASHELP library a dataset called SASHELP.BMT with the data for the same Klein and Moeschberger 2003 study. There is not any code that is available for this video because the intent of the video was for each person to use the SASHELP.BMT file with SAS Studio or their own data to generate the survival output. The narrator starts the program creation at time stamp 3:44 and continues to 4:48, when you see that the initial code which was generated was only 4 lines of code. By following the rest of the video steps in SAS Studio, the students can learn how to use SAS Studio to develop their own code for survival analysis. Then, each student can save the generated code locally so that they have the code for future reference. Wishing you #SASsuccess
Great video! I did not understand why you needed a separate dataset for the CIF. Would you still need it if you didnt have the log(waitTime) in the model?
We will get right back with you on that question!
Thanks, @skavenos! I’m glad to hear that you enjoyed the video
You need the separate data set so you can specify the values for the model’s covariates for which to plot the CIF, so that data set is different from the data set you used to estimate the model. Yes, you would still need the separate data set for the CIF plots even if you did not include log(waitTime) in the model. In that case, the separate data set would just include the values for the Group covariate you want included in the CIF plot.
There is actually an example in the documentation that does this (which was the basis of my example). Apparently, I am unable to include the link to the example or my comment will be deleted. But in the documentation for the PHREG Procedure, the example is called “Analysis of Competing-Risks Data” and it’s Example 92.15 in the latest version of the documentation. So if you search for that (PHREG Procedure Example 92.15 “Analysis of Competing-Risks Data”), then you should be able to find it.
Does that make sense?
Thanks for watching and thanks for the question!
Fantastic! Thank you!
Hello, how can you test the proprotionnal hypothesis for fine and gray in SAS ?
We'd recommend posting to SAS Communities where experts are on hand to help with programming/content related questions. See here: 2.sas.com/6053txeTD
Hello, thank you for the helpful video! I have a question to the "Censored" variable, out of the perspective of a non-mathematican/ statistican. Sadly there are very rare sources in the internet about this subject and I have no help from a statistican.
Referring to your leukemia example: If you got the groups "relapse" and "death before relapse", the variable "censored" contains all the patients, who did not get a relapse and did not die during the study-time, is this right? So it is no problem if all those patients get censored out? I am currently performing a Comprisk analysis, where a majority (around 50% of the patients) would fall under the "censored" group, so i dont consider this as a problem?
Thank you for your answer.
Thank you for your inquiry! We are checking on this for you!
Hi Max!
Censoring is one of the target outcomes, and can be interpreted as ‘no event yet’ (in this case no relapse or death). All patients either relapse or die with probability one as T gets big, but censoring occurs because some patients haven’t experienced one of these two events at the time the analysis is run. This is known as right or ‘end of study’ censoring. The alternative to keeping censored observations in the analysis would be removing them. This would entail a loss of information and could bias the results.
If you want to learn more about Survival Analysis we have some excellent classes on this topic. See 2.sas.com/6054M6cLe and 2.sas.com/6055M6cL5 .
excellent, can I get the data file please
Thank you for your inquiry! We are checking on this for you!
Hi Hala, in this video, the narrator is using a dataset called BMT_CR. However we provide in the SASHELP library a dataset called SASHELP.BMT with the data for the same Klein and Moeschberger 2003 study. The SASHELP library should be available to you when you open SAS. If you are using our SAS OnDemand for Academics server, the SASHELP library is automatically defined and available to you for use when you start your SAS Studio session.
Hi, thank you for your video. In my dataset, there is only Time, but not log_wait_time, Should I create a column for log_wait_time, and how can I create it? Thanks!
and also, since the time is different for each patient, I wonder why we use log_wait_time = 5.2 here? Thanks again.
@@Nw-ec8ke I looked at the predicted cumulative incidence function for each disease group at logWaitTime = 5.2 because that is the median value in the data set and it corresponds to roughly 6 months. But of course you are free to use the values that are of interest to you. As you noted, the times are different for each patient, so one of the goals of fitting the model is to then use it to make these sorts of predictions.
Does that make sense?
Thanks for watching and thanks for the questions!
Yes, to reproduce my analysis, you need to use the log of the wait time variable. In a DATA step, you can define a new variable with the existing waitTime variable as the input to the log function: logWaitTime=log(WaitTime);
To make sure we are on the same page and you are using the same data set, the code to create the data set I use for the demo can be found in the paper “Cause-Specific Analysis of Competing Risks Using the PHREG Procedure” by Changbin Guo and Ying So from SAS Global Forum 2018 (apparently I am unable to provide the direct link to the paper).
On page 3, that DATA step code creates the data set I use except I named it BMT_CR instead to avoid confusion with the SASHELP.BMT data set.
Hi Brian, could you tell me how to get 95%CI in the table? Thank you!
Tianshui! Thank you for our inquiry! We are checking on this for you!
Tianshui, Please review the video. At time stamp 5:43 in the video, you'll see that the default confidence interval used in this task is 95%. If you want to change the confidence interval used for the task, you would either change this drop down choice or you would use the ALPHA= option in the PROC PHREG code. Please refer to the PROC PHREG documentation 2.sas.com/6058KSXGI for more information about using the ALPHA= option. We cannot provide detailed Tech Support for your own code here in the RUclips feedback area. Please either post your question in the SAS Community for statistical procedure questions, here: 2.sas.com/6059KSXGL or open a track with Tech Support. To open a track with Tech Support, fill out the form at this link: 2.sas.com/6051KSXGF .
@@SASUsers Thank you so much for your reply. But I still can't find the results of 95%CI in the output table. What should I do to get it?
Tianshui, if you have been having issues with your own results getting at 95%CI in your table, we would recommend reaching out in SAS Communities or technical support. See details in our last comment for next steps. If you would like us to create a track for you, you can email us your contact information and include your site number to SAS_Cares@sas.com
@@SASUsers OK, thanks for your help!
Brian, I am using the On-demand SAS studio. Where should I go to access the macros housing the templates for custom titles, etc.?
Checking on this for you!
@@SASUsers thank you
Thanks for your question.There are not any macro variables or macro programs used in this video. SAS titles and footnotes are controlled by the TITLE and FOOTNOTE statements in your code. If you want to alter the code in the the task to use a TITLE or FOOTNOTE statement, you would not necessarily need a macro program to do that. Also, you only need to work with SAS templates if you wanted to alter the style characteristics of a title.
To open a track with Tech Support, fill out the form at this link: 2.sas.com/6059JiXLd . Your best resource for this question is to either open a track with Tech Support or, since you are using SAS OnDemand for Academics and SAS Studio, you can post your question in the SAS Community for the OnDemand users, here: 2.sas.com/6051JiXL5 (if the question is about SAS Studio) or here in the Statistical Procedures community: 2.sas.com/6052JiXLg (if your question is related to doing survival analysis).
how can I download the data file
We are checking to see if this available for you.
In this video, the narrator is using a dataset called BMT_CR. However, we provide in the SASHELP library a dataset called SASHELP.BMT with the data for the same Klein and Moeschberger 2003 study. The SASHELP library should be available to you when you open SAS. If you are using our SAS OnDemand for Academics server, the SASHELP library is automatically defined and available to you for use when you start your SAS Studio session. We hope this helps!
@@SASUsers I need help .. I can not upload my saved SPSS file to a library that I can you in SAS.. please help me
A great resource for this type of question is our SAS Communities 2.sas.com/6050bUF4Q. These communities are a service hosted by SAS to facilitate the exchange of technical information pertaining to SAS products, services and best practices. For assistance with getting answers fast, click here 2.sas.com/6051bUF4v to see some helpful tips on how to post a great question! If you are not able to find the assistance you need, please feel free to speak with Technical Support by calling 800-727-0025, option 2.
@@SASUsers Dear .. can you please guide me where can I ask questions regarding SAS Academics .. please
All I want to upload data file to a library and I can not do this
Hi, do you have the sas codes?
Checking for you!
Checout out timestamp 3:52,-the narrator is using a dataset called BMT_CR. However we provide in the SASHELP library a dataset called SASHELP.BMT with the data for the same Klein and Moeschberger 2003 study. There is not any code that is available for this video because the intent of the video was for each person to use the SASHELP.BMT file with SAS Studio or their own data to generate the survival output. The narrator starts the program creation at time stamp 3:44 and continues to 4:48, when you see that the initial code which was generated was only 4 lines of code. By following the rest of the video steps in SAS Studio, the students can learn how to use SAS Studio to develop their own code for survival analysis. Then, each student can save the generated code locally so that they have the code for future reference. Wishing you #SASsuccess