Leprosy Mailing List – July 23, 2018
Ref.: (LML) Single-dose rifampicin chemoprophylaxis
From: Joel Almeida, London and Mumbay
Dear Pieter,
Thanks to Dr. van Brakel for his witty message. We seem to be in agreement on the principle of using normal scientific standards.
The scientific method treats every candidate intervention as being no more effective than placebo (or rain dances) until evidence refutes that view.
Whatever Tiwari et al set out to do, it is clear what they ended up doing: reporting an odds ratio to "contribute in the external validity of chemoprophylaxis results presented in various papers." That self-declared aim and the accompanying claim of 50% reduction in leprosy invites scientific scrutiny. I am truly sorry that the claim is not adequately supported by the facts, using normal scientific standards (or even elementary maths).
I am sorry because the attempt to eliminate leprosy from a small island is worthwhile. But unjustifiable claims invite questioning.
A confidence interval is derived from the standard deviation of a presumed underlying distribution, usually a normal distribution. If a 95% confidence interval does not encompass zero effect, then the observed effect is very unlikely (less than 5% probability) to be due entirely to random effects. Put in another way, if the true magnitude of the effect was zero and if the same procedure were repeated 100 times in 100 separate but materially identical populations, fewer than 5 repetitions would show the observed magnitude of effect. Then we could justifiably claim an effect attributable to the intervention, over and above any random effects.
Dr. van Brakel refers to the 6 year report of the COLEP study. Here's what the 6 year report says, with the figures in brackets showing 95% confidence limits.
"For the total study period of 6 years the incidence rate per 10 000 persons at risk was 18·0 [14·9–21·7] in the placebo arm and 12·9 [10·3–16·1] in the SDR arm."
Therefore, the incidence rate in the placebo arm was very unlikely to be lower than 14.9 (per 10000 person years at risk). While the corresponding rate in the SDR arm was very unlikely to be higher than 16.1.
What is the justifiable claim that emerges from this evidence? That the observed efficacy of SDR could well be zero. Anything else is insufficiently substantiated by the facts.
Think about that. The efficacy of SDR as measured in the COLEP study at 6 years could well be zero. Not at least 10%, as at 4 years (the pre-defined end point of the study). But zero.
In years 3-4 and 5-6, the treated (SDR) group had a HIGHER observed incidence rate of clinical leprosy than the placebo group. Those differences do not attain statistical significance, so we can safely attribute them to random effects. However, they do not support claims about the efficacy of SDR.
I am truly sorry to disagree with Dr. van Brakel, but there is a point beyond which facts cannot be stretched.
I must reiterate how valuable it seems to attempt elimination on a small island. That is the place to try anything and everything, because one can afford to fail without compromising large-scale programs. Meanwhile, large-scale programs rightly prioritise interventions with proven efficacy.
The COLEP trial report authors themselves stated, "After four years' follow-up we cannot yet establish to what extent there is a true prevention of new cases of leprosy by intervention with rifampicin." Evidence from the 6 year COLEP report shows that the true effect of SDR could well be zero. An intervention with zero or near-zero efficacy is almost infinitely cost-INeffective.
Why did the 22 individuals in the guideline development group fail to see all this? They were "either representatives of academia, civil society, associations of persons affected by leprosy or national/regional leprosy programme managers". Perhaps they were mostly experts in areas other than scientific inference. More importantly, their review took place in secret, shielded from the kind of open scientific scrutiny that is possible with published systematic reviews. Whatever the explanation, their review is mistaken. We all make mistakes, so that is understandable.
We can still correct our mistakes now by applying normal scientific standards. Evidence from the 6 year COLEP report shows that the true effect of SDR could well be zero.
M. leprae deserve to be confronted with measures that are firmly grounded in robust scientific standards. We owe this to the vulnerable people who are most at risk of leprosy and its disfigurements. They trust us. Let's deserve their trust by holding ourselves to normal scientific standards.
Joel Almeida
LML - S Deepak, B Naafs, S Noto and P Schreuder
LML blog link: http://leprosymailinglist.blogspot.it/
Contact: Dr Pieter Schreuder << editorlml@gmail.com
No comments:
Post a Comment