Clinical trials in medicine have come under public scrutiny as never before in the
last decade with scandal after scandal about misrepresentation or misinterpretation
of results leading to erroneous clinical conclusions. High-profile campaigns have
been started by the British Medical Journal and others to allow access to original trial data, even down to the original case
record forms, so that trials can be reanalyzed where foul play is suspected. In some
cases, this has altered the primary outcome of the trial[1]; however, more subtle changes to trial analysis can occur during the study process.
A common issue is that it is harder to publish negative trials than positive ones,
which shouldn't be the case if the original trial was correctly designed to answer
the clinical question with sufficient statistical precision.
Nevertheless, it is not uncommon to see a secondary outcome or a per protocol analysis
(rather than the more rigorous intention-to-treat analysis) that is statistically
significant rise to prominence in the study report as the key finding, despite the
trial being powered for a different primary outcome measure. The lack of correction
for multiple testing of the data is another common omission, sometimes justified by
describing such analysis as exploratory or hypothesis generating. While acceptable,
clinical inference cannot be drawn for such uncorrected statistical presentation,
nor should these results be presented as main trial outcomes.
Endoscopic trials are certainly not immune from these phenomena, although the level
of commercial input and drive has been less than seen with the millions of dollars
that pharmaceutical companies have riding on a successful new drug. Endoscopic trials
have often been single-center and often single-operator, with or without minimal funding,
save for the time and enthusiasm of a committed endoscopic innovator. Such trials
are often of very limited generalizability and essentially simply represent phase
one safety studies.
More recently much larger endoscopic studies have been developed that are multicenter,
multi-operator and often international. These are phase two safety and efficacy studies,
with some phase three regulatory studies now occurring and subsequent phase four post-marketing
registries now also seen. This represents a maturing of endoscopic research approaches
with grant funding for some of these larger trials, which entails formal peer review.
Detailed peer review by grant funding bodies is to be commended, as it usually strengthens
studies by making them more conservative in their assumptions, especially around recruitment,
which makes completion of recruitment (and hence delivery of planned statistical power)
more likely, requires patient involvement, and ensures the statistical analysis plan
is rigorous and predefined, with a clear primary outcome measure.
In many areas of the world there is an expectation that to secure grant funding, a
clinical trials unit will be involved as part of the study team and help with trial
design and ensure trial logistics can be delivered. Industry-delivered studies increasingly
involve clinical trials units for endoscopic studies as well a clinical research organizations
to ensure delivery on the substantial investment that is required. As medical devices
come under increasing scrutiny, and companies wish to make regulatory claims about
devices, this is likely to become a bigger part of endoscopic research [2].
Transparency has already become part of publishing endoscopic studies with all major
gastrointestinal and endoscopic journals requiring pre-recruitment trial registration
to publish studies [3]. The purpose of this registration is mainly to publicly disclose study rationale
and ethical considerations before recruiting subjects into the trial. Moreover, these
registries minimize publication bias, since even high-quality studies may end up unpublished
due to negative or unwanted results. This raises the question whether we actually
need publications of endoscopy trials protocols and what we can learn from them.
Publishing endoscopy study protocols can help improve the quality of endoscopy research
in multiple ways [4]. First, it allows researchers to obtain feedback on study protocols through peer
review. Although in studies that have received ethics approval or a grant from a major
funding body, feedback on draft protocols is preexisting, there is no doubt that even
with the most elegantly designed study, it may be difficult to defend controversial
elements in the protocol. Moreover, protocol publication enables readers to compare
what was originally intended with what was actually executed. That is of great importance,
preventing the manipulation or dredging of data. It also minimizes post-hoc revision
of study aims and endpoints ([Table 1]).
Table 1
Key reasons to publish trial protocols.
|
|
|
|
|
|
|
|
|
|
Furthermore, publishing studies protocols allows for researchers and funders to understand
what studies are under way, thus minimizing unnecessary duplication of research effort.
In addition, it enables systemic reviewers to find trials, which may in turn reduce
distortion from publication bias. Finally, publishing endoscopy study protocols enables
patients to see what studies are under way for which they may consider volunteering.
The publication of the ADENOMA trial study protocol for colonoscopy with or without
endocuff with adenoma detection rate as the primary outcome measure is the first study
protocol to be published in Endoscopy International Open [5]. It is a large multicenter study with public and industry funding, and is supported
by a clinical trials unit. As such, it represents the leading edge of a new wave of
high-quality endoscopic studies which are potentially widely practice changing, hence
there is a need for special rigor and oversight of them. Is publication of the trial
protocol worth the paper (electronic or otherwise) it is written on? To some extent,
that judgment can only be made when the final study report is published, as the protocol
allows us to establish what the study team originally proposed, and compare it to
what was delivered and reported. Importantly, the level of methodological detail in
a protocol is far greater than the few hundred words usually devoted to methodology
in a standard publication and allows us to drill down into the detail of the claims
made.
Examples here include details of masking and allocation concealment: The study is
not blinded, since by nature of the intervention the allocation is known to the endoscopist
and allocation concealment cannot be offered to the personnel who will register the
outcomes. A degree of bias favoring the intervention arm cannot be excluded when identifying
or purposely “ignoring” polyps (especially since colonoscopy is not video recorded
or evaluated by a second endoscopist), or when recording time to insertion to cecum.
The same applies to endoscopy nurses when recording patient satisfaction and discomfort
score during colonoscopy. How withdraws and dropouts are handled is also important.
During a colonoscopy procedure, the endoscopist can remove the endocuff if there is
an acute angulation in a fixed sigmoid, preventing further scope advancement. It is
important that these patients are not excluded from the final analysis so as to ensure
that the Intention-to-Treat (ITT) analysis is preserved. Some of these more subtle
aspects of the study design can easily be lost in a brief methodological summary.
Ultimately we believe that publication of trial protocols for endoscopic studies improves
the transparency and accountability of our research, making it the best possible standard
upon which to strive to improve care for our patients. We have seen the movement to
study protocol publication in other major clinical specialties, notably cardiology
and oncology. It is time for endoscopy to step up to the same level of research excellence
to drive rigorous widespread practice changes based on the highest quality evidence
[2].
