[ccm-l] SAFE TBI in today's NEJM

[Ian Seppelt]

That's not right, Farhad, it's a prospective 24 month followup of a
subgroup of the original SAFE cohort. It was certainly not a planned
analysis when SAFE was originally designed [you will recall that the
primary end point was 28 day mortality] but when the difference in
outcomes in the (prospectively determined) trauma group became apparent
in SAFE at 28 days this was planned. 24 months was chosen, as the first
patients in SAFE were approaching 24 months at the time the decision was
made to do this. So it is still a prospective outcome evaluation of two
groups in an RCT.

The SAFE investigators are currently doing a genuine post hoc analysis
of these patients in more detail, to try to work out the 'WHY' - in
particular to be sure there are not important difference between groups
that were not detected in the initial SAFE dataset [remember that SAFE
was stratified for trauma, but not for TBI per se].

Even if you accept that some as yet unrecognised bias led to an
overstatement of the difference between the groups, I think it is
implausible that accounting for that bias will totally neutralise such a
large difference.

SAFE was designed as a mortality study. The TBI followup looked at both
mortality, and functional neurological outcomes (GOSE, done properly,
prospectively, by a blinded trained assessor). And yes, it obviously DID
have the power, as it detected a significant difference with p = 0.007

I don't think I have overstated the case.

Cheers, Ian

Ian Seppelt FANZCA FJFICM
Senior Staff Specialist
Dept of Intensive Care Medicine
The Nepean Hospital, PO Box 63 Penrith NSW 2751
Director of Clinical Research, Sydney West AHS
Clinical Lecturer, University of Sydney

>>> "Farhad N Kapadia" <fnkapadia at gmail.com> 30/08/2007 1:33pm >>>
 Today's NEJM has the 2 year followup of the traumatic brain injury
> cohort from SAFE which clearly shows worse outcomes in the albumin
> group. Based on this it is reasonable to say that albumin is
absolutely
> contraindicated in TBI,



Ian,

Isn't it possible that you are overstating the case.

Your statement is based on a finding of a subgroup post hoc analysis
which
almost certainly does not have to power to evaluate mortality.


farhad
-- 
Dr. Farhad Kapadia MD FRCP
Consultant Physician & Intensivist
Mumbai, India