Hi Jacob,
I think a conclusion of 'No more trials' would be justified if a systematic review of appropriately-sized and well-conducted randomised trials have provided an estimate of effect with a 95% confidence interval that is sufficiently tight for us to be able to say that further research (eg more trials) is unlikely to change our confidence in that estimate of effect. The included studies have no more than minor limitations, their results are consistent, they measure pretty much the same important outcome and there's no evidence of publication bias. More trial data might tweak the estimate a bit but we wouldn't expect much of a change. If the primary studies involved in a systematic review are of low quality then it is highly unlikely that a reviewer would (or could) conclude that no more trials are needed because the estimate of effect that the reviewers could provide is likely to have a wide confidence interval.
You are right that a statement saying no more trials are needed will make ethical approval of further studies difficult but that, I think, is only right and proper if the totality of (high-quality) evidence suggests that we have pretty much nailed down the effect of the intervention. Basically, there's no more uncertainty about the intervention, so why deny it (if effective) to, or force it upon (if ineffective) more trial participants? Of course, if the new trial involved a new population (older or younger participants, say) then perhaps it would be justified, depending on who was involved in the existing trials.
There's some useful information on this topic presented in the recent series of BMJ papers on GRADE, the first of which is BMJ 2008;336;924-926.
Regards,
Shaun
Dr Shaun Treweek
Research Fellow
Tayside Centre for General Practice
University of Dundee, UK
Researcher
Norwegian Knowledge Centre for the Health Services, Norway
Contact details:
Tayside Centre for General Practice
University of Dundee
Kirsty Semple Way
Dundee DD2 4BF
UK
Email: [log in to unmask]
Web: http://homepage.mac.com/streweek/start.html
Skype: streweek
Mob: +44 777 901 6955
Tel: +44 1382 420046
Fax: +44 1382 420010
http://www.sapc.ac.uk/09/
The University of Dundee is a Scottish Registered Charity, No. SC015096.
________________________________________
From: Evidence based health (EBH) [[log in to unmask]] On Behalf Of Puliyel [[log in to unmask]]
Sent: 28 September 2008 19:12
To: [log in to unmask]
Subject: Evidence-based prohibition of further research
Dear All
It is not unusual for authors of systematic reviews to write in their conclusion that: 'Further RCTs are no longer warranted'(1). It is like the authors want to be the last word in the matter.
My question is: Is there an evidence-based-threshold at which authors of systematic reviews can make this statement.
There are two aspects of this issue that are important.
1. A meta analysis looks at many parameters. Take the Cohrane meta analysis done in 1994 of pneumococcal conjugate vaccine which showed no benefit against clinical pneumonia.
Later in 2005, Cutts et al published a study from The Gambia. Re-meta analysis adding this study to the previous studies shows there is a significant reduction in clinical pneumonia.
Thus meta-analysis is an evolving process and the conclusions can change with newer studies. It is perhaps not right to say; 'no further studies should be done', unless the benefits/harms are so glaring. What is the measure of what is 'glaring'.
2. The second aspect is, that such a statement will make ethical approval of subsequent studies difficult. If there is no evidence-based-threshold at which to make such a statement, then such statements should not be made in an evidence based paper.
The BMJ recently published a paper on the Edmonston Zagreb measles vaccination in infants (2).
In 1992 the World Health Organization (WHO) working group on Expanded Programme on Immunization reviewing studies on the vaccine concluded that immunization programmes should no longer use the Edmonston Zagreb measles vaccine and that no more field trials of this vaccine should occur (3).
This thus throws up unnecessary ethical questions.
Sincerely
Jacob Puliyel
Head of Division of Pediatrics
St Stephens Hospital
Tis Hazari
Delhi
[log in to unmask]<mailto:[log in to unmask]>
References
1. Prophylactic Administration of Natural Surfactant Extract). Soll RF. Prophylactic synthetic surfactant for preventing morbidity and mortality in preterm infants. Cochrane Database of Systematic Reviews 1998, Issue 2. Art. No.: CD001079. DOI: 10.1002/14651858.CD001079.
2. Martins CL, Garly ML, Bale C, Rodrigues A, Ravn H, Whittle HC, Lisse IM, Aaby P. Protective efficacy of standard Edmonston Zagreb measles vaccination in infants aged 4.5 months: interim analysis of a randomized clinical trial. BMJ 2008;337:a661
3. Expanded programme on immunization. Safety of high titre measles vaccines. Wkly Epidemiol Rec. 1992;67:357-61
___________________________
Jacob M. Puliyel MD MRCP MPhil
eFax 00 44 7092-124285
Phone 00 91 11 23946388
00 91 9868035091
|