Elsevier

World Neurosurgery

Volume 118, October 2018, Pages e99-e104
World Neurosurgery

Original Article
The Introduction of Innovations in Neurovascular Care: Patient Selection and Randomized Allocation

https://doi.org/10.1016/j.wneu.2018.06.127Get rights and content

Highlights

  • RCT methods can protect patients when clinicians are uncertain about the best management option.

  • From a patient perspective, 1:1 randomization is the best way to introduce new devices.

  • Selecting optimal patients for RCTs decreases generalizability of study results.

Most neurovascular innovations have been introduced by using case series followed by observational studies. A better approach would be a pragmatic randomized trial. Two important aspects of trial design, patient selection and randomized allocation, remain poorly understood. We discuss the role trial methodology can play in the protection of patients being offered innovative treatments.

Introduction

The endovascular management of various neurologic problems has been empowered by the introduction of new devices and interventions. Sometimes the advancement brought about by the new technology is so evident that our enthusiasm may be difficult to contain—for example, the first time we saw how flow diversion could perfectly reconstruct a carotid artery previously affected by a giant aneurysm. There was no need for a trial to understand that something important had been uncovered. Such a spectacular result could be enough to make you question whether you should proceed with the planned next case of the day, a carotid occlusion for a patient with a symptomatic cavernous aneurysm, who passed the test occlusion because of an excellent circle of Willis. Parent vessel occlusion is a treatment you have trusted for decades, but it involves the permanent loss of a major cerebral artery. Thus, many interventionists would be tempted to convert their next case to flow diversion. However, the patient on whom you subsequently performed flow diversion for a condition that is rarely life-threatening to save her carotid artery is admitted a few weeks later with a massive intracranial hemorrhage at a distance from the aneurysm, an infrequent complication of flow diversion that you had never encountered with carotid occlusion. You wonder then how you should manage the next patient with a similar problem.

The various ways innovations can be introduced and studied in clinical practice, including case series, epidemiologic studies, explanatory trials, and pragmatic trials, are summarized in Table 1. Enthusiasm, faith, and a large dose of positive energy are necessary to conceive, develop, and introduce innovations in medical care. Industry, regulatory agencies, national institutes, and leaders of our field have the knowledge, power, and financial resources to design trials that could provide a prudent way to introduce innovative treatments and devices in the care of neurovascular patients at the same time as a critical evaluation of whether they improve patient outcomes. However, all endovascular innovations of the last 3 decades, including flow diversion, have been introduced by using case series of 100 or so carefully selected patients, without valid comparisons with existing treatment options.1, 2, 3, 4 Although most clinicians recognize that randomized controlled trials (RCTs) would give the best evidence for decision making, it is often thought that to protect individuals from potential harm, the safety and efficacy of new devices should be first assessed in a small series of carefully selected patients. Our goal in this article is to examine this idea and to propose an alternative. Safety and efficacy do not mean much without the “as compared with” qualifier, which should be verified in practice, and results obtained from a few selected patients cannot provide the reassurance we are looking for; confidence intervals are too wide, and selection hampers generalizability. Finally, we examine the dangers of treating patients using the innovation outside a carefully designed randomized trial.

The choices we have once innovations are adopted in the aforementioned fashion are 1) to experiment without proper methods on our patients, haphazardly modifying indications and techniques according to expert opinions voiced at meetings or ongoing personal experiences (often followed by multiple case series of expanded indications), or 2) to design the needed trials ourselves, after regulatory approval, but without power or financial support. This is how the Flow Diversion in Aneurysms trial was designed.5 However, we are then confronted with other obstacles related to the general lack of understanding of pragmatic trials and of the role they can play in practicing under uncertainty.3

The foregoing clinical anecdotes ended with the question: How to manage the next patient with a similar problem? This question can artificially be split into a question about knowledge (what kind of study is needed to fill the gap in knowledge?) and a question that concerns the ethics of care (how should I care for this next patient?).

Regarding the gap in knowledge, the clinician already believed that flow diversion could work for giant aneurysms; he also believed that, everything else being equal, it would theoretically be better to not sacrifice the carotid artery. Thus, he did not really need an explanatory trial, which is a study designed to answer the question whether treatment work in optimal circumstances, because he had already seen that the treatment could work. What the clinician did not know is whether “everything else” was indeed “equal” (the relative frequency of complications, for example, including the unexpected ones, and their severity). For that, he needs the results of a pragmatic trial, a trial that has verified whether the treatment actually improves patient outcomes in practice as compared with previously available options. What should the clinician do if no such trial results are available? Our aim here is to explain why such a patient should be offered participation in a care trial.6

As the introductory anecdote illustrates, awesome or catastrophic events may occur as we explore unknown territories when trying innovations in practice. This is why scientific methods are needed, not only to gain knowledge but also to avoid falling prey to various reactions (enthusiasm, disappointment, fear) that may influence our clinical decisions as we are willing to offer promising innovations in practice. More importantly, trial methods should be understood as norms of conduct that limit the amount of risk patients take when they trust a physician who is tempted to try an innovative device or approach.6 Thus, trials differ not only in the sort of questions they address, as we have seen, but also in their capacity to protect the safety of patients being offered the innovation. The ethical role trial methods can play in the safe introduction of innovations in practice remains poorly understood.3

This misunderstanding of trial methods has had momentous consequences in medical care and research. It is as if clinicians, devoted to patients, focusing on care delivery by making daily case-by-case decisions under uncertainty, had delegated the evolution of their science and ethics to nonclinicians.

On the side of ethics, the sad result is that since the Belmont report of the 1970s and according to current regulation, comparing patient outcomes is not a natural concern of care but a subject of research—worse, research cleanly separated from care; obstructed by bureaucracy; and in need of distinct expertise, personnel, and funding.7, 8 This may partly explain the temptation to introduce innovations as if they were standard treatments without randomized allocation, telling each patient that the treatment is selected in his or her best medical interest, in effect performing research disguised as care. The regulation demarcating research and care was meant to protect vulnerable patients from being used as research subjects for the benefit of others, but it has paradoxical results: It ends up encouraging the use of the innovation as if it were normal care, while also constraining the clinical trials that could be designed to protect patients from the potential harm related to the use of an experimental treatment—which would be in their best medical interest. In effect, we end up using patients in the context of care to obtain the data necessary to gain device approval.7, 8

On the side of science, the methods of inquiry that should be so influential to guide our practices seem to have been conceived not by clinicians caring for patients but by scientists foreign to clinical medicine. We are provided with either 1) case series or explanatory trials of narrowly selected patients, designed according to preclinical laboratory principles, to explore some causal mechanism (explanatory trials)9 or 2) epidemiologic methods such as observational studies, as if we had no control over our actions and patients were naturally exposed to our interventions the way populations are exposed to poverty or environmental toxins. Both of these methods, respectively inspired from preclinical or epidemiologic studies, are ethically and scientifically misguided. Between these 2 misguided approaches, our duty is to construct a way whereby physicians can practice innovations that 1) balance the risks for patients who might (or might not) benefit from their use; 2) reintroduce individual patient outcomes as a central concern of clinical care; and 3) use truth-tracking scientific methods that can protect patients from enthusiasm, conflicts of interests, wishful thinking, and pervasive human errors.6 The way to go is to proceed with properly designed RCTs, launched at the time the innovation is being introduced.10

Randomized trials are frequently qualified as “ideal, but unfeasible.”3, 11 The reasoning that “RCTs are difficult; therefore let's use an alternative” is recurrent but absurd—like giving up on truth because being honest is simply too difficult. Once it is accepted that RCTs are necessary to protect individual patients when using innovations in clinical practice, the question is no longer “whether to do an RCT” but “how can an RCT be done?” The task is daunting, but we do not have to start from scratch. In 1967, Schwartz and Lellouch9 paved the way to integrate pertinent clinical research within the realm of clinical care by introducing the distinction between explanatory and pragmatic trials. They emphasized that “explanatory work must be done on animals, therapeutic trials on human subjects being limited to pragmatic experiments” (emphasis ours).12

A detailed discussion of trial design is beyond the scope of this article. However, 2 crucial aspects are addressed that are the object of common misconceptions: heterogeneity (and selection criteria) and randomized allocation.

Section snippets

Heterogeneity and Selection Criteria

Let us examine the problem with the selection of patients. Many interventionists seem to think that there is a proper chronology to evaluate innovative treatments, using a hierarchy of methods based on patient selection: First start with small series of carefully selected patients who are likely to have good outcomes to gain approval from regulatory authorities. Once approval is secured, progressively expand indications and publish a case series or study clinical results on a wider range of

Benefits of Randomized Allocation

A trial offers more than an observational registry. The question of interest to clinical care is does the innovation improve patient outcomes? To that question, a registry cannot provide an answer, any more than other observational studies, because there are no clinical outcomes from standard treatment options with which to compare flow diversion in the same patients. When an innovation becomes available, we are quick to judge patients as “untreatable by other means.” When patients were not

Conclusions: A Randomized Trial Is the Proper Way to Introduce a New Device

On reflection, can physicians select patients? Of course not, they select treatments. Physicians select the best treatment for their patients (including conservative management, when treatments have yet to be shown beneficial). The best treatment is the one that can improve patient outcomes. Until the innovative treatment has been proven to lead to better outcomes than standard options, at which time it will be appropriate to prescribe it, the best treatment is either the time-honored, standard

References (20)

  • J. Raymond et al.

    Pragmatic trials can be designed as optimal medical care: principles and methods of care trials

    J Clin Epidemiol

    (2014)
  • D. Schwartz et al.

    Explanatory and pragmatic attitudes in therapeutical trials

    J Chronic Dis

    (1967)
  • J. Raymond et al.

    Randomize the first patient

    J Neuroradiol

    (2017)
  • D.F. Kallmes et al.

    Aneurysm Study of Pipeline in an Observational Registry (ASPIRe)

    Interv Neurol

    (2016)
  • T. Becske et al.

    Pipeline for uncoilable or failed aneurysms: results from a multicenter clinical trial

    Radiology

    (2013)
  • D.F. Kallmes et al.

    Letter to the Editor. Flow Diversion in the Treatment of Intracranial Aneurysm Trial

    J Neurosurg

    (2017)
  • M. Eikermann et al.

    Commentary: Europe needs a central, transparent, and evidence based regulation process for devices

    BMJ

    (2013)
  • J. Raymond et al.

    Flow diversion in the treatment of aneurysms: a randomized care trial and registry

    J Neurosurg

    (2017)
  • T. Beauchamp

    The Belmont Report: Ethical Principles and Guidelines for the Protection of Human Subjects of Research

    (1978)
  • T.L. Beauchamp et al.

    The historical foundations of the research-practice distinction in bioethics

    Theor Med Bioeth

    (2012)
There are more references available in the full text version of this article.

Cited by (16)

  • Understanding the choice of control group: A systematic review of vertebroplasty trials for osteoporotic vertebral compression fractures

    2023, Neurochirurgie
    Citation Excerpt :

    In spite of repeated appeals to re-introduce scientific principles into medical ethics and practice, it may already be too late.[34] The necessary randomized trials should instead be designed early on, at the time of the introduction of the innovation.[35,36] The design of such trials can raise several difficulties, and although refinements have been developed over the last decades, this review of the profusion of RCTs on vertebroplasty shows that, for most of us, it remains a work in progress.

  • Understanding the problems with recruitment in surgical randomized trials: A lesson from landmark trials on temporal lobe epilepsy

    2022, Neurochirurgie
    Citation Excerpt :

    This approach is routinely enforced in most countries for new drugs. We have previously argued that in the case of surgical innovations and new devices, constraining the use of such innovations to RCTs may be conceived as an ethical imperative, provided the trial is designed in the best interest of patients [8,9]. While routine to introduce new drugs, this strategy is nearly impossible to implement when the treatment to be tested is an already commonly used surgical intervention.

  • Understanding burden of proof and equipoise in the design of pragmatic clinical trials: An example from a trial on brain arteriovenous malformations

    2022, Neurochirurgie
    Citation Excerpt :

    Similarly, promising innovations should not be used in practice, outside of a properly designed research context. Patients should be offered only a 50% chance of receiving the promising innovation, and at the same time offered a 50% chance of getting the standard, validated treatment, to minimize risks at the individual patient level [25,26]. Trials remain difficult to conduct, particularly in neurosurgery [27].

  • Middle Cerebral Artery Aneurysm Trial (MCAAT): A Randomized Care Trial Comparing Surgical and Endovascular Management of MCA Aneurysm Patients

    2022, World Neurosurgery
    Citation Excerpt :

    Anatomic features of MCA aneurysms, such as a wide neck often including a branch artery origin, render endovascular treatment difficult.3,4 New devices have been and continue to be developed to address these difficulties, including stents,5 flow diverters,6 and intra-saccular flow disruptors (ISFDs),7 but none have been shown to improve endovascular results in a randomized trial.8-10 The International Subarachnoid Aneurysm Trial (ISAT) has shown that small anterior circulation good-grade ruptured aneurysm patients have better 1-year clinical outcomes after being coiled than clipped,11 but those results may not necessarily apply to patients with MCA aneurysms.

  • Unruptured aneurysms: Why observational studies fall short no matter how “Big” the Data

    2021, Neurochirurgie
    Citation Excerpt :

    Laboratory methods inspired the methodology of explanatory clinical trials, a design that is still too frequently used today, even though it was recognized as early as 1967 that it cannot properly inform clinical practice [26]. Perhaps, the most frequent misconception is to believe that patients must be carefully selected to show meaningful results, which in turn can only come from rigid experimental protocols applied in expert centres [27–29]. Basic sciences and clinical research require different methods for they have different aims.

View all citing articles on Scopus

Conflict of interest statement: The authors declare that the article content was composed in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

View full text