EURURO Vol. 72 No. 4

E93 656

Letter to the Editor
Reply to Philipp Dahm, Vikram Narayan, and Jae Hung
Jung’s Letter to the Editor re: Richard J. Sylvester,
Steven E. Canfield, Thomas B.L. Lam, et al. Conflict of
Evidence: Resolving Discrepancies When Findings
from Randomized Controlled Trials and Meta-analyses
Disagree. Eur Urol 2017;71:811–9
We would like to thank Drs. Dahm, Narayan, and Jung for
their thoughtful and well-argued comments on our paper
. However, it would appear that they have misinterpreted
the main point of the article, which is to help guideline
developers resolve conflicting findings between large,
multicentre, and robust randomized controlled trials (RCTs)
and those of systematic reviews and meta-analyses (MA).
This issue goes beyond ‘‘conflicting results of different
studies’’ within MAs, especially when the MA largely
consists of small, underpowered, heterogeneous studies
with a high risk of bias and/or poor reporting.
In the example quoted by Dahm et al regarding the
effectiveness of a-blockers as medical expulsive therapy
(MET), a recent MA included 55 RCTs, of which 95% had
fewer than 100 patients in each arm; 75% were not placebo-
controlled; almost 90% either had a high risk of bias or did
not report blinding of outcome assessors; and almost 90%
had three or more domains judged as having either an
unclear or high risk of bias. The MA reported a benefit of a-
blockers on the basis of pooled analysis (risk ratio 1.49, 95%
confidence interval 1.39–1.61). Nevertheless, the authors
acknowledged evidence of clinical, methodological, and
statistical heterogeneity, heterogeneity of stone passage
rates in the control groups, and publication bias. By
contrast, three of the largest, multicentre, prospective,
placebo-controlled, double-blind RCTs published to date
did not find a statistically or clinically significant
benefit of a-blockers for MET for their primary outcome of
stone passage.
In this context, we would argue that MAs are useful in
providing exploratory and hypothesis-generating findings
instead of explanatory ones, primarily because there is a
fundamental flaw in treating all studies as being equal in
relation to the primary outcomes. This flaw cannot be
completely overcome even with the most carefully planned
sensitivity and subgroup analyses. We would contend that
the pooled results for 55 mostly small, heterogeneous
studies are not equal to or better than results from a handful
of large, well-conducted, multicentre RCTs, all of which
show consistently negative findings.
A large, prospective, multicentre, double-blind, placebo-
controlled RCT comparing tamsulosin versus placebo for
ureteric stones <9 mm is currently recruiting (250 patients
in each arm) Should the results prove negative, might
there be a risk that these negative findings will be diluted
amidst a mass of small, heterogeneous studies in future
MAs, thereby perpetuating the vicious cycle of uncertainty?
How many more RCTs do we need to answer this question
once and for all? Is this justifiable ethically, considering the
potential risk of side effects for a treatment with question-
able effectiveness in terms of high-quality evidence?
If anything, one can argue that if another MA is needed
(and this is debatable), an individual patient data MA
incorporating data from homogenous and well-conducted
RCTs is more likely to yield more meaningful findings than a
conventional MA in terms of allowing standardisation of
patient criteria and outcomes, allowing more meaningful
subgroup analyses and exploration of interactions between
different variables.
Conflicts of interest: The authors have nothing to disclose.
References
[1]
.
[2]
.
[3]
.
[4]
[5] Sur RL, Shore N, L’Esperance J, et al. Silodosin to facilitate passage of
ureteral stones: a multi-institutional, randomized, double-blinded,
placebo-controlled trial. Eur Urol 2015;67:959–64
ava ilable at
journal homepage:
DOIs of original articles: , .
0302-2838/# 2017 European Association of Urology. Published by Elsevier B.V. All rights reserved.