Simon Finfer is a leading critical care clinical researcher. Hear his candid talk on the reality of research and publication and why it's relevant to you!
4. What will I talk about?
• Research
– Why do it? Gattinoni vs. Finfer
– Testable vs. untestable hypotheses
– “Industrial plant”
– How to do it well – collaborators and industry
– How to mess it up
• Publication
– What do journals want
– Is peer review a good idea
– Could I make it all up and get away with it?
– Things you won’t believe (or at least shouldn’t happen)
– Don’t take “No” for an answer
– Protect yourself
– What’s next …………… ?
9. Why do we do research?
– To understand
– To change behaviour
– To improve patients’
outcomes
10. • How much certainty do we want?
• Few treatments produce dramatic effects
• Most claims for large benefits prove evanescent
11. Strict control of random error
• large numbers of deaths or other relevant outcome
And Bias
• Strict randomisation with allocation concealment
• Appropriate analysis, with no unduly data-dependent
emphasis on specific parts of the overall evidence
13. 13 13
IIT-TC – Eligible but not included
Eligible but not included (16 Trials - 5320 participants)
Other
IIT
trials
(n=16)
Country
Sample
size
UNABLE
provide
data
(n=4)
Ma
2008
China
510
Yu
2005
China
54
He
2007
China
188
Fernandez
2005
USA
207
CollaboraEon
denied
(n=2)
Yang
2009
China
240
Yan
2009
China
356
Withdrew
(n=2)
Van
den
Berghe
2001
Belgium
1,548
Van
den
Berghe
2006
Belgium
1,200
No
response
(n=8)
Grey
2004
USA
61
He
2008
China
122
Taslimi
2009
Iran
129
Lacherade
2008-‐2010
France
440
Benito
2008
Spain
28
De
Azevedo
2007
Brazil
48
Stecher
2006
Slovenia
n/a
Desai
2012
USA
189
29. How to mess it up?
• Launch into it unprepared….
– Without knowing incidence, outcome, realistic effect or
minimal clinically important effect
• Choose an untestable hypothesis
– Dopamine / Frusemide to reverse early AKI
• Work with people seeking to prove something
– “we are not interested in funding you because your
proposed study is not designed to show our product in a
favourable light”
– “can we arrange for these two hospitals to be randomised
to the same group?”
– “can we analyze the data in all three ways and publish the
analysis that shows our product is better?”
• Do your sample size calculation backwards …….
37. What do journals want?
• Advertising revenue
• Advertising revenue
• Advertising revenue …..
38. What do journals want?
• Revenue from reprints
• Revenue from reprints
• Revenue from reprints
39. What do journals want?
• High quality papers or
• Papers that will be bought in the thousands by
company marketing departments or
• Papers that will be widely cited because they will
generate ……
• Advertising revenue ……..
• And publication fees
43. Reviewer 1 – the results appear to be robust
Reviewer 2 – this is a very important study
Reviewer 3 – the authors should be commended on
a well thought out and executed clinical trial
Reviewer 4 – effects of HES on renal function is a
never ending story
Reviewer 5 – “this is not a randomised trial, there
was no allocation concealment, the study was not
blinded, I doubt one can draw meaningful
inferences from these data with any acceptable
degree of certainty.”
50. Too many journals?
• Editor - “We can’t ask them to change it all …..
• Finfer - “Well, it’s the wrong design, the wrong
outcome, the wrong analysis. It’s meaningless. You
can’t publish it.”
• Editor – “It’s better than 90% of the stuff I’m sent,
if I don’t publish this I may as well close the journal
down”
• Finfer – “Sounds like a good idea”
59. Cochrane albumin Review:
Reactions from industry
Manufacturers allocated $2.2 million:
• To prepare literature reviews supporting the use of
albumin, to be sent to leading regulatory agencies
• To prepare and disseminate a Cochrane critique
dossier
• To establish a medical advisory panel to write
articles supporting the use of albumin
62. § Issues?
§ Unblinded
§ Marked baseline imbalance
§ Insufficient data re resuscitation and renal function
§ Stopped for futility but reports “positive” findings
§ Changing P values
§ Senior industry (HES) employee on trial advisory board