Beecher’s concerns about the premature introduction of novel surgical procedures1
quoted at the start of this paper highlight an ongoing and as yet unresolved controversy regarding the proper way to evaluate novel surgical procedures and ensure adequate oversight of surgical research. The question of when variations in surgical practice or evaluation of outcomes become research and, as such, become subject to the rigors of evaluation by a research ethics board (REB) is not always answered easily. Unlike novel medications, which are subject to Health Canada and US Food and Drug Administration regulations, new surgical procedures have no such agency review before their widespread adaptation into practice. This has raised concerns that “nonvalidated surgical procedures are being smuggled past RCTs and [REBs], our societal checkpoints for innovations.”10
There is a perception that surgeons seek to “have it both ways: to experiment with innovative surgical procedures within the framework of clinical care and without [REB] oversight at the same time as publishing research … in professional journals.”14
Unfortunately, the Canadian standard for human research, the Tri-Council Policy Statement (TCPS) on Ethical Conduct for Research Involving Humans15
provides no specific guidelines for surgical research. The TCPS states that all research involving humans, as well as research involving human remains, cadavers, tissues, biological fluids, embryos or fetuses requires REB review and approval. The TCPS does not specifically define what constitutes research.
Bernstein and Bampoe16
have recommended that REB approval should be sought for all procedures performed in the context of an RCT, for novel procedures scheduled to be performed electively and for major modifications of accepted surgical procedures. Whereas it is often difficult to determine when innovative surgical treatment becomes research, “that boundary is certainly crossed once a report is presented at a scientific meeting, or submitted for scientific review.”17
At this point, all further study should be done in the context of a well-defined research protocol, with particular attention to the informed consent process and submission to local REBs.
Much of what is published in surgical journals involves retrospective case reviews or data culled from prospective databases. Most contend,14,18–20
and we agree, that wherever possible, protocols for such research should be submitted to local REBs. For prospective databases whose sole goal is research publication, efforts should be made to obtain informed consent from patients before enrolment in the database. If there is confusion as to whether a particular activity requires REB approval, we recommend consulting the local REB, because requirements may vary in different jurisdictions. Indeed, a recent study of the requirement for ethical review in human research found that 83% of medical journals required independent ethics committee approval, with most of those requiring disclosure of that approval within the paper.21
Ten years earlier, only 47% of those journals required ethical review.22
Evidence suggests that ethical process for research published in some surgical journals is high for prospective studies (up to 92% are reviewed by an REB) but relatively low for retrospective studies (only 30% are reviewed by an REB).23
Reitsma and Moreno24
found that only 28% of authors publishing results of innovative surgical procedures in 2000–2001 had submitted their protocol to an REB.
A more difficult question to answer is the following: When is further research no longer required; that is, what constitutes sufficient evidence of the efficacy of a surgical intervention? It is one of the ironies of the evidence-based medicine paradigm that, in the end, a majority of expert opinion is still required before a question regarding the efficacy of a surgical procedure is considered “solved,” regardless of whether it has been studied in an RCT.25
In essence, this is when clinical equipoise no longer exists. Pin-pointing when this occurs (clinically and even statistically) is difficult and is often only recognized in retrospect.26