4

Background:
I'm studying people seeking help. Participants described contacts with between 1 and 3 "responders" (e.g., friends, the police) in order- for example, a participant could have contacted just responder 1, or responder 1, then responder 2, then responder 3. I'm trying to predict help-seeking dropout, meaning that, for example, a participant contacted responder 1 but did not go on to contact a second or third responder- that participant would have a dropout at responder 1. So unlike other survival analysis, the observations are responders rather than time points- but they're still ordered in time. The independent variables in my model include characteristics of the people seeking help (e.g., gender) and aspects of their interactions with the responders (e.g., whether they liked the interaction). The data are right-censored for those participants who said that they contacted more than three responders because they could not record more than three responders in the survey. There are two people who only reported on responder 3; those people are left-censored because data are missing for responders 1 and 2.

Data setup:
The data are set up as a person-period dataset such that there is a line for each responder, which means that some participants have multiple lines. Responders are nested within participants. So a participant that contacted two responders would have two lines in the dataset; the participant-level data is the same in both lines and the responder-level data is different.

Here's what the data looks like: https://flic.kr/p/s1fW6k

Variables:
id is the ID number for the participant.
responder represents the responder number in the order that the participant contacted them. Possible values are 1, 2, and 3.
stoppedhelpseeking represents whether the participant stopped seeking help/dropped out after contacting that responder; 0 = no and 1 = yes.
gender is the participant's gender; 1 = woman and 2 = man
likedresponder represents whether the participant liked their interaction with the responder; 0 = no and 1 = yes
censor represents whether the participant did not report a dropout by responder 3.

Here is the code that I have (from the Allison survival analysis/SAS book):
proc phreg data = helpseeking plots=survival;
class id;
model responder*stoppedhelpseeking(0) = gender likedresponder /ties=efron;
run;

My questions:
-If I'm predicting dropout, should the code be stoppedhelpseeking(0) or stoppedhelpseeking(1)?
-How do I account for the right-censoring? I'm concerned that it's not explicitly reflected in my code.
-Do I need to account for the left-censoring, or should I drop those two people from analyses?
-Any other issues with the code that I should know about?

AdamO
  • 52,330
  • 5
  • 104
  • 209
Emily
  • 618
  • 4
  • 16

1 Answers1

1

If I'm predicting dropout, should the code be stoppedhelpseeking(0) or stoppedhelpseeking(1)

The former. Whatever is within the parens in the left-hand side of the model statement is an indicator for censored values, not your event of interest.

How do I account for the right-censoring? I'm concerned that it's not explicitly reflected in my code.

See the answer above. You're expressly taking right-censoring into account with stoppedhelpseeking(0) in your model statement, which is indicating to SAS that those with a value of 0 in that variable is right censored, because they didn't have their event happen, which means it must take place at some future unknown time.

Do I need to account for the left-censoring, or should I drop those two people from analyses?

It may not matter, really, depending on the size of your data set. However, you can set up your data to account for left-censored data using the entry= option discussed here: http://support.sas.com/documentation/cdl/en/statug/63033/HTML/default/viewer.htm#statug_phreg_sect028.htm

Fomite
  • 21,264
  • 10
  • 78
  • 137
  • I think what I'm getting caught up on is that the setup of my data means that the uncensored people have 0s for stoppedhelpseeking (unless they only had one responder) until their last responder. Is it recognizing that some people never have a value of 1 for any responder and thus treating them as censored? – Emily Apr 07 '15 at 19:38
  • Yes. Basically it's recognizing that people with a value of 0 implicitly *don't* have a responder value, and thus have a censored event time. Basically, you're saying "People with 0 didn't have an event" and people without an event implicitly don't have an event time. – Fomite Apr 07 '15 at 21:10
  • The value of 0 is at the level of the responder, though, not the person, since the data is nested. So a person with three responders and a dropout after the third would have three lines in the dataset with 0, 0, and 1, respectively, for the event. You're saying it's recognizing that ONLY people with all 0s didn't have an event? I just want to be totally sure that it's not seeing any line in the dataset with a 0 as a person without an event. – Emily Apr 07 '15 at 21:23
  • Basically, how you've set up your data, each person with a zero will have a censored event time *for that responder*. So for your example, they will be censored, censored and uncensored. Whether you want this or not is a more substantive question - and should probably be split from the current question, which is just on implementation. – Fomite Apr 07 '15 at 21:31
  • OK- thanks. Posted it here http://stats.stackexchange.com/questions/145219/setting-up-dataset-for-cox-regression-with-time-dependent-variables – Emily Apr 07 '15 at 22:56