Does the use of cardiac output monitoring make a difference? NICE seem to think so, the Cochrane group are less certain. What we needed was a large, well conducted trial and this is what we got...
| optimise.pdf |
The study compared two groups undergoing surgery of the gut. The control group did not have a cardiac output monitor, the intervention group had stroke volume guided fluid therapy and low dose Dopexamine intra-op and 6 hours postoperatively. The patients were 'high risk'. The primary outcome was composite. To cut a long story short this trial was NEGATIVE - there were no differences in primary or secondary outcomes between the groups. It's worth taking a closer look however...
First off the addition of Dopexamine may seem odd. To explain this some background may be useful. Rupert Pearse, the first author, is a fan of Dopexamine. He has received honoraria from the company that makes it* and in 2009 he published in Critical Care Medicine a meta-analysis reporting a halving of mortality with low dose Dopexamine (CI 0.3-0.9). This sounds almost too good to be true? Well maybe it was, at the same time another group were doing the same meta-analysis looking at the same papers and they reported no benefit to Dopexamine! Their paper can be found here, (you may recognise the name of one of the authors). So how can this be explained, is meta-analysis not the gold standard for EBM? Well no, it's not, it depends on the quality of the papers in the first place. The reasons for the differing results are explained in this editorial, and Pearse's reply to the paper is here. So whether Dopexamine makes a difference is still unclear, but Pearse it would seem thinks so. The addition of Dopexamine also makes the study original, and would (if you agree with Pearse) increase the likelihood of a positive result.
The outcome measure was a composite. I think we've discussed on a previous post how composite outcomes may be more relevant to the patient (e.g. a stroke, MI or loosing a limb are all bad and I don't want any of them), account for outcomes with low frequencies etc. but that they also may combine to a false negative (Outcome A occurs more frequently with intervention, outcome B less frequently, the composite of A&B is no difference, or worse benefit). A secondary outcome was all-cause mortality at 180 days, on the basis that post-operative complications, even minor ones, seem to affect long term outcome - my own bias however is that this is an association rather than causation (which we've also previously discussed). Do you agree?
The latter bit of the paper was interesting. By adding their (negative) results to a meta-analysis, the intervention becomes more positive. I admit I haven't looked at this in detail, but I wonder if this is a demonstration of the Simpson Paradox which is (honestly) an interesting bit of statistics.
I won't go on; this is another paper that I could talk about for ages, but I just want to highlight some other points:
Colloid was the bolus fluid in the intervention group - the patients didn't all die of renal failure but the dose wasn't massive.
'Dynamic CVP targeting' was recommended in the control group - I find it hard to put into words how I feel about this but I'm not quite as angered by it as I am by the accompanying editorial that claims "Another factor that may have lowered the occurrence of the primary outcome rate in the control group was the protocol recommendation that patients in the usual care group receive dynamic CVP guided fluid administration"!!! For anyone who wants a reference for the lack of usefulness of CVP, there are many out there but here's a meta-analysis.
There is no mention in the discussion about the conflicting pre-existing evidence re Dopexamine.
I'm amazed how many of these patients went to critical care, particularly as level 3 - Hawthorne effect?
If you want to hear what the author has to say about his (negative) results click here. Do you agree with his interpretation?
I look forward to reading what you think........
*Which is fine, but I commend you to always look at people's conflicts of interest
First off the addition of Dopexamine may seem odd. To explain this some background may be useful. Rupert Pearse, the first author, is a fan of Dopexamine. He has received honoraria from the company that makes it* and in 2009 he published in Critical Care Medicine a meta-analysis reporting a halving of mortality with low dose Dopexamine (CI 0.3-0.9). This sounds almost too good to be true? Well maybe it was, at the same time another group were doing the same meta-analysis looking at the same papers and they reported no benefit to Dopexamine! Their paper can be found here, (you may recognise the name of one of the authors). So how can this be explained, is meta-analysis not the gold standard for EBM? Well no, it's not, it depends on the quality of the papers in the first place. The reasons for the differing results are explained in this editorial, and Pearse's reply to the paper is here. So whether Dopexamine makes a difference is still unclear, but Pearse it would seem thinks so. The addition of Dopexamine also makes the study original, and would (if you agree with Pearse) increase the likelihood of a positive result.
The outcome measure was a composite. I think we've discussed on a previous post how composite outcomes may be more relevant to the patient (e.g. a stroke, MI or loosing a limb are all bad and I don't want any of them), account for outcomes with low frequencies etc. but that they also may combine to a false negative (Outcome A occurs more frequently with intervention, outcome B less frequently, the composite of A&B is no difference, or worse benefit). A secondary outcome was all-cause mortality at 180 days, on the basis that post-operative complications, even minor ones, seem to affect long term outcome - my own bias however is that this is an association rather than causation (which we've also previously discussed). Do you agree?
The latter bit of the paper was interesting. By adding their (negative) results to a meta-analysis, the intervention becomes more positive. I admit I haven't looked at this in detail, but I wonder if this is a demonstration of the Simpson Paradox which is (honestly) an interesting bit of statistics.
I won't go on; this is another paper that I could talk about for ages, but I just want to highlight some other points:
Colloid was the bolus fluid in the intervention group - the patients didn't all die of renal failure but the dose wasn't massive.
'Dynamic CVP targeting' was recommended in the control group - I find it hard to put into words how I feel about this but I'm not quite as angered by it as I am by the accompanying editorial that claims "Another factor that may have lowered the occurrence of the primary outcome rate in the control group was the protocol recommendation that patients in the usual care group receive dynamic CVP guided fluid administration"!!! For anyone who wants a reference for the lack of usefulness of CVP, there are many out there but here's a meta-analysis.
There is no mention in the discussion about the conflicting pre-existing evidence re Dopexamine.
I'm amazed how many of these patients went to critical care, particularly as level 3 - Hawthorne effect?
If you want to hear what the author has to say about his (negative) results click here. Do you agree with his interpretation?
I look forward to reading what you think........
*Which is fine, but I commend you to always look at people's conflicts of interest
RSS Feed