I’m grappling with the notion of uncertainty in relation to EBM.
Say a drug has an NNT (based on a placebo controlled trial) of 6. That means 6 patients would need to be treated with the drug for one to benefit. A doctor seeing 6 patients would have little way of knowing which of the 6 would actually benefit.
So it’s fine to say that there is a level of uncertainty here.
However, on another level, is the uncertainty around that uncertainty….
A small scale RCT might suggest an NNT of 4. Subsequent RCTs then give a range of NNTs ranging from 4-10. There is then some uncertainty about what the ‘true’ NNT is. A systematic review is then carried out which gives an NNT of 6 (as mentioned in my example above).
In other words the SR has further defined the level of uncertainty – we can now be fairly confident that the ‘true’ NNT is 6.
Can anyone point me in the direction of any papers that help define this dual level of uncertainty?
Also has any work been carried out examining the impact of this on treatment choices. Say the NNTs from 3 RCTs are 25, 36 and 48 and a SR shows a ‘true’ NNT of 32 – does that extra certainty make much difference when it comes to a clinician’s clinical decision?
I suppose an extension of this is – is there any evidence to suggest that being very certain of the uncertainty makes much odds when it comes to an actual doctor making a clinical decision?
Cheers
jon
-----------------------------------------
Email sent from www.ntlworld.com
Virus-checked using McAfee(R) Software
Visit www.ntlworld.com/security for more information
|