Statistics from Altmetric.com
Is this sufficient evidence to change practice?
In the March issue of BJO1 a randomised controlled trial was reported in which topical versus sub-Tenon’s local anaesthesia for routine cataract surgery is compared. The trial is well designed and simple, using a validated outcome measure with the outcome observers masked to the intervention. It shows a clear preference for sub-Tenon’s block by patients undergoing routine cataract surgery.
Does this prove it? Is this sufficient evidence to change practice? In making such a decision, readers will want to critically appraise this evidence. Those who disagree with its findings will try hard to find flaws in the study while those that agree with it will tend to take it as face value and feel vindicated. Both sides will have to consider the probabilities that the results of the study could be erroneous or point to the truth.
What kinds of errors occur in trials? After bias and confounding have been dealt with (and by design, these seem to have been well controlled), what about the possibility that chance alone could have produced the result? The p values are “significant” and the confidence limits of the estimate of effect do not include unity so that by definition, it is improbable that the estimate of effectiveness could have occurred by chance (though they still could have).
The other sort of error, failing to find an effect when in fact there was one, would not seem to be relevant since an effect was shown. The key to this question is the power of the study based on the number of people randomised and the size of effect worth detecting. Discerning readers will note that there is no such power statement in this report.
When I peer reviewed it, I spotted this omission and asked whether the reason for this was that the study was in fact a pilot study. It is now clearly stated in the methods that this study was designed as a pilot since there was no a priori estimate of effect to guide the sample size estimation (this was not stated in the original version).
This is an important point; the study is small and small studies are prone to both kind of errors—alpha type 1, showing an effect when it is actually just chance, and beta type 2, failing to find an effect when there is one. The weakness in this trial is its small size and its potential for both kinds of errors because chance plays a greater part in smaller experiments. The purpose of a pilot study, among other things, is to get a feeling for a likely effect size and to refine the methods of capturing reliable outcome data. Finding an effect in a pilot study is not a reason not to conduct the definitive experiment. The trialists may need to reappraise the study design and the size of effect that is clinically relevant, then perhaps apply more stringent criteria for the control of an alpha error. It will be the duty of the research ethics committees to make a decision about continuing to withhold the treatment found to be advantageous in the pilot study.
This study, despite being small, is of value and it is right that it should be published, since all results from properly conducted randomised trials add to the body of evidence (even if only in a small way) and should be available in the public domain for all to see. This may soon become a statutory requirement in the United States if legislation to prospectively register all randomised controlled trials, and make their results available to all, becomes law.2
This will be a major step forward for evidence based medicine since it will lock the pharmaceutical industry, as well as everybody else, into making the results of all their trials, whether positive or negative in outcome, publicly available.
The body of evidence grows as experiments are repeated. Another properly powered trial with a sample size calculation and with similar findings has been published in the British Journal ofAnaesthesia3 since Rüschen et al’s paper was accepted. These studies should now be summarised in a properly conducted systematic review. One such review has been published and would appear to support the findings of this study.4 Another Cochrane review is under way which will include the findings of this trial and the other recently published one. Surgeons wanting to make decisions about changing their practice may decide there is sufficient evidence to do so but others may wish to await a larger body of evidence. It may be difficult to justify another trial comparing sub-Tenon’s with topical anaesthesia alone but some will argue that intracameral lignocaine would make a major difference to the comparison. Thus, further trials are needed to address this question.
Note in Proof
Is this sufficient evidence to change practice?