| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 National Jewish Medical and Research Center, and University of Colorado at Denver and Health Sciences Center, Denver, Colorado
Correspondence and requests for reprints should be addressed to E. Rand Sutherland, M.D., M.P.H., National Jewish Medical & Research Center and University of Colorado at Denver and Health Sciences Center, 1400 Jackson Street, J220 Denver, CO 80206. E-mail: sutherlande{at}njc.org
ABSTRACT
Principles of clinical trial design dictate that the placebo control is the most rigorous comparator by which to assess the efficacy of an intervention. In the case of procedural treatments, however, exposure to an inactive (or sham) procedure may be associated with substantial discomfort or risk, a phenomenon not typically encountered with placebo controls employed to assess noninvasive (e.g., orally administered) treatments and one that can influence the decision-making process with regard to choice of control arms. In considering whether or not a sham control is either appropriate or feasible in clinical trials of devices or procedures, the investigator must consider optimal standards of research design and ethical conduct, carefully balancing the risk of the sham control with the benefit that might be obtained from careful evaluation of the procedure itself. This calculation can be challenging when considered in the larger therapeutic context, for while the risk borne by research subjects exposed to sham control may not be insignificant, the aggregate risk to control subjects in a clinical trial may be less that the aggregate risk borne by patients who receive an untested therapy that is ultimately shown not to be beneficial after rigorous assessment in a placebo-controlled trial. This review addresses issues pertaining to the use of sham controls in clinical trials of devices and procedures and provides examples from the literature which demonstrate the importance of and controversy surrounding sham controls.
Key Words: control group • sham • clinical trial
The medical literature contains numerous reports of surgical procedures that were introduced into general practice on the basis of uncontrolled studies and which, through the absence evaluation in clinical trials containing a placebo (or sham) procedural control, failed to account for multiple factors that can lead to inaccurate conclusions regarding efficacy. Notable examples include surgical treatments such as internal mammary artery ligation for angina pectoris (1, 2), cervical glomectomy for the treatment of bronchial asthma (3), and fetal nigral cell transplantation for the treatment of Parkinson's Disease (4), all adopted as procedures thought to be efficacious on the basis of uncontrolled studies and then later shown not to be beneficial after more rigorous experimental evaluation. These examples, which reinforce the importance of a placebo control arm to drawing accurate conclusions in studies of therapeutic efficacy, suggest that inclusion of a sham control is most desirable in studies of medical devices or procedures. However, a number of factors influence decision-making with regard to whether or not a sham procedure is either appropriate or feasible in clinical trials of devices or procedures. These considerations require the investigator to strike a balance between the highest standards of research design and the highest standards of ethical conduct, undertaking a careful assessment of risk versus benefit and a consideration of these factors not just within the context of the clinical trial itself, but within the context of treatment of all patients with the target disorder.
RISK VERSUS BENEFIT
The challenge when studying an intervention based on a procedure or medical device is one of balancing issues of trial design that optimize generalizability of the findings with those issues that optimize the protection of human subjects. This balance is alluded to in criteria for Institutional Review Board approval of research as outlined in the Code of Federal Regulations (45 CFR part 46), which indicates that risk to subjects must be minimized by using procedures that are consistent with sound research design but that do not unnecessarily expose subjects to risk. Furthermore, the risk to subjects should be considered to be reasonable in relationship to anticipated benefits and the importance of the knowledge that may be reasonably expected to result (5). In the example of clinical trials of orally administered agents, balancing safety and efficacy is often relatively straightforward, as a placebo tablet or capsule is typically associated with minimal to no risk when administered to a research subject. In trials of devices or procedures, however, the placebo control is often a sham intervention that differs only minimally from the active or effective procedure and that may expose subjects to a degree of risk similar to that of the active procedure. A recent search of the ClinicalTrials.gov database (6) revealed that studies using a sham control are relatively infrequent, with only 65 of 15,805 actively recruiting protocols (0.4%) involving a sham control.
A sham procedure can be defined as one performed on a control group participant to ensure that he or she experiences the same incidental effects of the operation or procedure as do those participants on whom a true operation is performed (7). Sham control arms, like other placebo control arms in controlled clinical trials, have the potential benefit of reducing the introduction of bias, particularly with regard to three critical areas of experimental design and conduct: treatment allocation, treatment adherence, and the assessment of subjective outcomes modified by treatment. Thus, sham controls are particularly useful for trials of devices or procedures with subjective endpoints (e.g., symptoms) and provide a robust means of controlling for the ancillary effects of a procedure, optimizing the ability of the investigator to evaluate for a placebo or procedural effect in an unbiased fashion. The attractiveness of including a sham control arm increases in cases in which the risk of the sham procedure is particularly low, and studies considering a sham control should carefully consider risk versus benefit.
Sham controls are only one of many potential control options in studies of procedures or devices, and alternative control strategies include best medical therapy, therapy with a device or procedure that has already been approved, or a control consisting of subjects as their own controls in a crossover fashion. From the standpoint of experimental design, selection of appropriate control strategies depends upon a number of factors including the indications for and safety of the procedure or device, the extent to which nonplacebo control strategies are able to minimize the introduction of bias, and the nature of the patient population and disease for which a particular treatment is being assessed. From an ethical standpoint (Table 1), Horng and Miller (8) have suggested that the use of a sham control can be ethically justified if (1) there is a valuable question to be answered by the research and the risk of the sham control is justified by the value of the knowledge to be gained, (2) the sham control is methodologically necessary to test the study hypothesis, (3) the risk of the sham control itself has been minimized and does not exceed a threshold of acceptable research risk, and (4) the nature of misleading involved in the administration of a placebo control is adequately disclosed to and accepted by the subject during the informed consent process.
|
One example from the medical literature in which sham surgery has been used as a form of placebo control is Parkinson's Disease, a disorder of motor function with loss of dopaminergic neurons in the substantia nigra. Its clinical course is notable for marked variability in both the magnitude and duration of response to medical therapy, and multiple clinical trials in Parkinson's Disease have shown a substantial placebo effect, with a 20 to 30% improvement in motor scores seen over periods as long as six months (9), with deterioration in clinical status after the discontinuation of placebo and with an apparent cumulative dose–response increase to placebo therapy over time (10). Early open-label trials of surgical transplantation of embryonic dopaminergic tissue transplantation into the brain had suggested that both clinical and radiographic benefits could be observed (10). In these studies, the surgical intervention required stereotactic frame placement, MRI targeting of the transplantation site, general anesthesia, a skin incision, creation of a burr hole and then transplantation of fetal tissue, followed by cyclosporine for 6 months in addition to continued medical therapy. When sham-controlled clinical trials were performed to follow up prior open-label studies, in which control subjects underwent a nearly identical procedure but with only a partial burr hole and no transplantation of fetal tissue, it was determined that the active procedure was associated with no significant improvement in outcomes versus the sham procedure, and the authors of one study concluded that "fetal nigral transplantation currently cannot be recommended as a therapy for Parkinson's disease based on these results" (4). Although ethical issues were raised about the appropriateness of these studies as they were being designed and conducted (10, 11), ultimately the inclusion of a sham control arm in these rigorous studies demonstrated that the procedure had little apparent clinical effect, a result that stood in contrast to prior uncontrolled studies. Although issues of small sample size and questions about the effect on recruitment of public discourse on the ethics of these trials reduce the definitive nature of these results, this example from the literature appears to be one where, had sham-controlled studies not been performed, clinicians might have continued to expose patients to an ineffective surgical therapy.
THE CASE OF ASTHMA
An analogy can be drawn between issues pertinent to clinical trials in Parkinson's Disease to those commonly faced in asthma clinical trials. Like Parkinson's disease, asthma can demonstrate marked variability both in the magnitude and duration of symptoms and response to therapy. Furthermore, clinical trials in asthma have shown the potential for significant placebo effect, including significant improvements in lung function and symptoms (12, 13). There is also a potential target for procedure-based treatment, in that the hyperproliferative and hypercontractile airway smooth muscle observed in patients with asthma (14) has been targeted with bronchial thermoplasty, a procedure in which there is application of controlled radiofrequency energy to the walls of airways between 3 and 10 mm in diameter in an attempt to attenuate airway smooth muscle mass and/or function. Bronchial thermoplasty is one of the more recent additions to a growing list (Table 2) of device- or procedure-based therapies for the treatment of asthma (15–21).
|
Statistically significant improvements were seen in the thermoplasty group with regard to outcomes which included self-reported elements, including mild exacerbations, symptom-free days, symptom scores, asthma quality of life questionnaire scores, and asthma control questionnaire scores. However, of the more objective outcomes only morning peak expiratory flow rates were significantly improved at 12 months, whereas pre-bronchodilator FEV1 and airway hyperresponsiveness were not. While these findings could be interpreted to suggest that the effects of bronchial thermoplasty are primarily on symptom-related domains, in the absence of a sham control one cannot be entirely certain that the observed benefit was not due either to a placebo effect or the introduction of bias into the conclusions regarding efficacy due to the lack of blinding. In their report, the investigators noted that the interpretation of their results was limited by the lack of a sham control comparator and indicated that these data supported the conduct of a placebo-controlled trial involving the use of sham bronchial thermoplasty, a study which is now underway.
SUMMARY
These examples from the clinical trial literature support the conclusion that there is a role for sham controls in clinical trials of devices and procedures, particularly when used either to confirm or refute the results of open-label trials and in circumstances in which outcomes that are partially or entirely subjective are primary. However, while the use of sham controls is likely to provide the highest quality and potentially most generalizable clinical trial data, the use of a sham control must be carefully contemplated in light of its appropriateness and feasibility and applied within a formalized ethical framework. Only through a careful assessment of the risks to research subjects of being exposed to a sham control and of the knowledge that will be gained through direct comparison of a placebo to active device or procedure can investigators decide whether the use of a sham control versus a usual care control is most advantageous.
FOOTNOTES
Conflict of Interest Statement: E.R.S. was a sub-investigator in a study of bronchial thermosplasty sponsored by Asthmatx, Inc. via a grant to National Jewish Medical and Research Center. He received no financial or material support as a result of this grant.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
