Multicentre RCTs evaluating mortality in ICU – Doomed to fail?

Multicentre RCTs evaluating mortality in Intensive Care : Doomed to Fail ? Gustavo A. et al. Crit Care Med, 2008, vol 34, no 4.

The authors performed a systematic search for multicentre RCTs in adult ICU targetting mortality as a primary outcome measure. Perhaps surprisingly they found only 72. Of these, 10 reported a positive effect, 7 negative and 55 neutral.

The authors make some interesting points…

  • of the 10 positive results not all have been implemented, either because of tight inclusion criteria or subsequent neutral/ negative evidence
  • 30% of studies failed to adequately report a power calculation
  • assuming a commonly targeted mortality benefit (10%), they estimate approx 60% of studies would be underpowered at the 90% confidence limit.

So is mortality as unrealistic endpoint in disease states inadequately differentiated (like ARDS or sepsis) ?

Or is the problem a failure of the ICU community to organise sufficiently powered trials (time to learn from the cardiologists) ?

Or do we need to “relax” our ideas about EBM, and begin to take a little more notice of lower grades of evidence?

(PS – this study includes excellent tables of all 72 multicentre RCTS published in ICM, including authors, publication date, pt numbers, mortality outcomes and of course full refs – might be useful for the DICM revision!)

Dan Harvey


2 comments so far

  1. msimmonds on

    I would agree with Dan’s comment about relaxing our ideas about EBM, especially when it comes to our own local efforts.

    The rigors of trying to power a study to show a mortality benefit without enormous resource is virtually impossible. The Chinese are running Clopidogrel trials with well over 10,000 patients involved. This is just not a realistic aim for most western centres, especially ITUs.

    Should we not be focussing our own research efforts on improving process rather than aiming directly at outcome? We all know that simple measures (antibiotics, fluids, basic ventillator management, weaning protocols etc.) improve outcome but all units are guilty of failing to maintain robust mechanisms to ensure consistent care is given within and without the ITU.

    The (proven) mortality, morbidity and length of stay benefit of getting the basics right is far greater than any new drug will ever attain, and yet we still continue to spend time pawing over the latest drug trial (e.g.steroids, APC), swinging our treatments with the latest tide. Surely our time locally would be better spent on our own service improvement, then we may see significant outcome benefit.

  2. djwk on

    I am most impressed by this site Dan.

    I am grateful that you highlighted a paper which would have passed me by.

    I think that this paper highlights one of the problems with ICU research and EBM. The problem is we want to know the answers, but we don’t really know what the question is.

    ICU is the land of syndromes. Diseases like ARDS, sepsis, AKI are at best a spectum of diseases and at worse a group of many distict pathophysiological conditions that just happen to give us respiratory or cardiovascular dysfunction. The assumption that an 80 year old with multiple comorbidities, abdominal sepsis and ARDS is the same as an 18 year old with meningococal sepsis, is at best naive. The belief that you can account for these variances by recruiting 10 000 patients is just brainless.

    I think the first thing we need to do is define our patients more clearly. One way this could be done is to recuit responders only. A simple example is in a vasopressor trial to give everyone a drug like dopamine and only recruit responders to a vasopressin/norad trial and not anyone with multifactorial hypotension.

    I believe responder (or at least better trial candidate selection trials) could still be undertaken in single centre units. This will allow us to recruit quality and not quantity to RCT’s.

    Finally, if we need to recruit 10 000 patients to discover a statistically significant benefit to treatment X then is it clinically relevant?

    Keep up the good work

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: