design of a vaccine study

[Clara Chu]

I am trying to design a vaccine study. The objective of the study is to
determine if vaccine exposure contributes to an increased risk in
developing a certain type of cardiac event.

The study population consists of patients who were vaccinated. I have 2
groups defined based on 2 time intervals: 1 year prior to the date the
vaccine was given and 1 year after the date the vaccine was given. The
endpoint is time to cardiac event, and we only consider the first event.
If I look at the time intervals, 1 year prior to a person's vaccination
date and 1 year after the vaccination date, then it is possible to define
time to event for each individual who had an event ([event date - vax
date] for those who had an event postvax and [event date - 1 yr prior to
vax date] for those with an event pre-vax). Each individual then has one
measurement for time to event. For those who never experienced an event
(censored), then it's possible to define 2 time intervals, 1 year prior to
vax date and 1 year after vax date. Then, anyone who is censored for this
2 year period would contribute time to both the post vax and pre-vax
groups. However, don't the groups need to be mutually exclusive to
calculate hazard ratios using a Cox regression? The hazard ratio would be
defined with the pre-vax group as the reference.

Can anyone verify? Thanks!

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Maguin, Eugene]

Clara,

I think the key issue in your design is that you have a single group
repeated measures design in which you want to use the year prior to
vaccination as a control for the year post. If you didn't take time to event
in to account, you'd have a test of correlated proportions. I'd guess that
the hazard rates would be correlated. I don't know how that design would be
analyzed but I'd bet that others on the list do.

Gene Maguin


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Clara Chu
Sent: Wednesday, April 13, 2011 9:07 AM
To: [hidden email]
Subject: design of a vaccine study

I am trying to design a vaccine study. The objective of the study is to
determine if vaccine exposure contributes to an increased risk in
developing a certain type of cardiac event.

The study population consists of patients who were vaccinated. I have 2
groups defined based on 2 time intervals: 1 year prior to the date the
vaccine was given and 1 year after the date the vaccine was given. The
endpoint is time to cardiac event, and we only consider the first event.
If I look at the time intervals, 1 year prior to a person's vaccination
date and 1 year after the vaccination date, then it is possible to define
time to event for each individual who had an event ([event date - vax
date] for those who had an event postvax and [event date - 1 yr prior to
vax date] for those with an event pre-vax). Each individual then has one
measurement for time to event. For those who never experienced an event
(censored), then it's possible to define 2 time intervals, 1 year prior to
vax date and 1 year after vax date. Then, anyone who is censored for this
2 year period would contribute time to both the post vax and pre-vax
groups. However, don't the groups need to be mutually exclusive to
calculate hazard ratios using a Cox regression? The hazard ratio would be
defined with the pre-vax group as the reference.

Can anyone verify? Thanks!

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD